Benefactives and Malefactives
Typological Studies in Language (TSL) A companion series to the journal Studies in Language. Volumes in this series are functionally and typologically oriented, covering specific topics in language by collecting together data from a wide variety of languages and language typologies.
Editor Spike Gildea
University of Oregon
Editorial Board Balthasar Bickel
John Haiman
Marianne Mithun
Bernard Comrie
Martin Haspelmath
Doris L. Payne
Denis Creissels
Bernd Heine
Franz Plank
William Croft
Paul J. Hopper
Anna Siewierska
Nicholas Evans
Andrej A. Kibrik
Dan I. Slobin
Carol Genetti
František Lichtenberk
Sandra A. Thompson
Leipzig
Leipzig / Santa Barbara Lyon
Albuquerque Canberra
Santa Barbara
St Paul
Leipzig Köln
Pittsburgh Moscow
Auckland
Santa Barbara Eugene, OR Konstanz
Lancaster Berkeley
Santa Barbara
Volume 92 Benefactives and Malefactives. Typological perspectives and case studies Edited by Fernando Zúñiga and Seppo Kittilä
Benefactives and Malefactives Typological perspectives and case studies Edited by
Fernando Zúñiga University of Zurich
Seppo Kittilä University of Helsinki
John Benjamins Publishing Company Amsterdamâ•›/â•›Philadelphia
8
TM
The paper used in this publication meets the minimum requirements of American National Standard for Information Sciences – Permanence of Paper for Printed Library Materials, ansi z39.48-1984.
Library of Congress Cataloging-in-Publication Data Benefactives and malefactives : typological perspectives and case studies / edited by Fernando Zúñiga, Seppo Kittilä. p. cm. (Typological Studies in Language, issn 0167-7373 ; v. 92) Includes bibliographical references and index. 1. Grammar, Comparative and general--Benefactive constructions. 2. Grammar, Comparative and general--Case. 3. Case grammar. I. Zúñiga, Fernando. II. Kittilä, Seppo. P291.27.B46â•… 2010 415--dc22 2010000735 isbn 978 90 272 0673 2 (Hb ; alk. paper) isbn 978 90 272 8831 8 (Eb)
© 2010 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Co. · P.O. Box 36224 · 1020 me Amsterdam · The Netherlands John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa
Table of contents Preface List of contributors Introduction. Benefaction and malefaction from a cross-linguistic perspective Seppo Kittilä & Fernando Zúñiga Benefactive applicative periphrases: A typological approach Denis Creissels Cross-linguistic categorization of benefactives by event structure: A preliminary framework for benefactive typology Tomoko Yamashita Smith
vii ix 1 29
71
An areal and cross-linguistic study of benefactive and malefactive constructions Paula Radetzky & Tomoko Smith
97
The role of benefactives and related notions in the typology of purpose clauses Karsten Schmidtke-Bode
121
Benefactive and malefactive uses of Salish applicatives Kaoru Kiyosawa & Donna B. Gerdts
147
Beneficiaries and recipients in Toba (Guaycurú) Marisa Censabella
185
Benefactive and malefactive applicativization in Mapudungun Fernando Zúñiga
203
The benefactive semantic potential of ‘caused reception’ constructions: A case study of English, German, French, and Dutch Timothy Colleman
219
Beneficiary coding in Finnish Seppo Kittilä
245
Benefactives in Laz René Lacroix
271
Benefactive and malefactive verb extensions in the Koalib verb system Nicolas Quint
295
 Benefactives and Malefactives
Benefactives and malefactives in Gumer (Gurage) Sascha Völlmin A “reflexive benefactive” in Chamba-Daka (Adamawa branch, Niger-Congo family) Raymond Boyd
317
331
Beneficiary and other roles of the dative in Tashelhiyt Christian J. Rapold
351
Benefactive strategies in Thai Mathias Jenny
377
Korean benefactive particles and their meanings Jae Jung Song
393
Malefactivity in Japanese Eijiro Tsuboi
419
Index
437
Preface This article collection originated in the Workshop on the Typology of Benefactives and Malefactives we organized and held October 25–26, 2007 in Zurich. We would like to thank the Hochschulstiftung of the University of Zurich for providing financial support to the workshop. After learning from the late Mickey Noonan that he was interested in such an article collection for the Benjamins series Typological Studies in Language, we received the papers from the contributors and embarked on the editorial process, which is invariably time-consuming but sometimes—as in this case—very rewarding. We were fortunate enough to be able to interest Masayoshi Shibatani in this project, and he assisted us as guest editor despite his many commitments and extensive traveling during the period. Not only did we benefit from his experience and knowledge, but his expertise in benefactives also made him an invaluable collaborator. We are greatly indebted to Matt for his help, particularly considering the fact that some of the articles even take issue with him on specific analytical decisions. Matt’s assistance notwithstanding, it is we who are responsible for any remaining shortcomings of the present volume, of course.
Fernando Zúñiga Seppo Kittilä Eugene and Helsinki, 2009
List of contributors Raymond Boyd Centre National de la Recherche Scientifique UMR 8135 : Langage, Langues et Cultures d’Afrique Noire 7, rue Guy Môquet 94801 VILLEJUIF Cedex France
[email protected] Marisa Censabella Universidad Nacional del Nordeste, Argentina Instituto de Investigaciones Geohistóricas — CONICET López y Planes 636 (3500) Resistencia Provincia del Chaco Argentina
[email protected] [email protected] Timothy Colleman Ghent University Dutch Linguistics Department Blandijnberg 2 B-9000 Gent Belgium
[email protected] Denis Creissels DDL-ISH 14 avenue Berthelot 69363 Lyon cedex O7 France
[email protected] Donna Gerdts 2563 W. 7th Ave. Vancouver, BC V6K 1Y8 Canada
[email protected]
Mathias Jenny Seminar für allgemeine Sprachwissenschaft Plattenstrasse 54 8032 Zürich Switzerland
[email protected] Seppo Kittilä General Linguistics P.O.Box 24 00014 University of Helsinki Finland
[email protected] Kaoru Kiyosawa 7-2-5 Toshima, Kita-ku Tokyo 114-0003 Japan
[email protected] René Lacroix 328, rue de Belleville 75020 Paris France
[email protected] Nicolas Quint LLACAN-CNRS 7 rue Guy Môquet 94-801 Villejuif-Cédex France
[email protected] Paula Radetzky National Tsing Hua University Department of Linguistics 101, Section 2 Kuang-Fu Road, Hsinchu, 30013 Taiwan
[email protected]

Benefactives and Malefactives Christian Rapold Max Planck Institute for Psycholinguistics PO Box 310 6500 AH Nijmegen The Netherlands
[email protected] Karsten Schmidtke-Bode Friedrich Schiller University Jena Department of English and American Studies English Linguistics: Language and Cognition (Chair: Prof. Dr. Holger Diessel) Ernst-Abbe-Platz 8 07743 Jena Germany
[email protected] Tomoko Yamashita Smith Advisement Office for International Students Graduate School of Engineering Osaka University 2-1 Yamada-oka, Suita Osaka 565-0871 Japan
[email protected] Jae Jung Song Linguistics Programme University of Otago, P.O. Box 56 Dunedin 9054 New Zealand
[email protected]
Eijiro Tsuboi Department of Language and Information Sciences Graduate School of Arts and Sciences University of Tokyo, Komaba 3-8-1 Meguro, Tokyo 153-8902 Japan
[email protected] Sascha Völlmin Universität Zürich Seminar für Allgemeine Sprachwissenschaft Plattenstrasse 54 CH-8032 Zürich Switzerland
[email protected] Fernando Zúñiga Seminar für allgemeine Sprachwissenschaft Plattenstrasse 54 8032 Zürich Switzerland
[email protected]
Introduction Benefaction and malefaction from a cross-linguistic perspective* Seppo Kittilä & Fernando Zúñiga
University of Helsinki/University of Zurich The present volume consists of articles on the expression of benefaction and malefaction, including both a number of detailed studies across a selection of languages from Eurasia, Africa and the Americas, and some cross-linguistic/typological studies dealing with comparative issues. In this introduction we summarize the relevant theoretical presuppositions and suggest generalizations that can be made on the basis of comparison of the individual studies. Section 1 deals with the linguistic notion of benefaction and malefaction, and Section€ 2 describes several formal mechanisms used for the coding of beneficiaries and maleficiaries. Then Section 3 addresses the semantically determined variation in the coding of benefaction and malefaction. Section 4 gives an overview of polysemous/ multifunctional marking strategies for the notions under investigation. Section€5 gives a short preview of each of the following chapters. Finally, in Section 6 we suggest a number of fruitful lines for further research.
1.â•… Defining benefaction and malefaction Due to the considerable attention that the notion of benefaction has received in the linguistic literature, it is not surprising that there are a number of different definitions of beneficiary, either as a purportedly pre-theoretic notion or as a Case/thematic/ semantic role. Both the literature on case and studies of ditransitive clauses have usually used a broad and basically circular definition; the one found in Lehmann et al. *We are indebted to Matt Shibatani for his valuable comments on an earlier version of this introduction. He does not necessarily agree with each and every one of our terminological and/or analytical choices, however, and any remaining errors and misconceptions are our own responsibility. Seppo Kittilä would also like to thank Academy of Finland (grant number 1127724) for providing funding for this study.
Seppo Kittilä & Fernando Zúñiga
(2000: 68; our translation) is a case in point: “[a] benefactive situation is one that occurs to the benefit of a participant”. The goal of this introduction is neither to provide a comprehensive survey of earlier studies of benefaction nor to propose a definition that would make all earlier accounts of the beneficiary role unnecessary. Rather, our objective is to discuss the notion of benefaction from a perspective that makes fruitful comparison of different instances of benefaction possible, both in individual languages and cross-linguistically. Instead of claiming that defining the beneficiary in the terms proposed below is the most appropriate way to conceptualize it, we simply observe that such a framework makes meaningful cross-linguistic analysis possible and advance the following, admittedly simplified, working definition of this role here:1 The beneficiary is a participant that is advantageously affected by an event without being its obligatory participant (either agent or primary target, i.e. patient). Since normally only animate participants are capable of making use of the benefit bestowed upon them, beneficiaries are typically animate.2
English examples include cases as heterogeneous as the dentist painted the house for/ instead of me, the dentist baked me a cake and the children ate olives for their father. In the first case, the beneficiary benefits from the denoted event in not having to carry out the denoted action him/herself. In the second case, the beneficiary is also a recipient, and reception is usually (albeit not necessarily) deemed beneficial. In the third case, benefaction is more indirect in nature, and may mean, for example, that the father receives a promotion resulting from his children’s action (e.g. because the father’s boss
1.â•… The two single most important cross-linguistic studies of benefaction for the present article collection are an article by Matt Shibatani (1996) and a monograph by Carmen Conti (2008). The former proposes a cognitive account of some interesting properties of constructions that treat beneficiaries as core syntactic arguments in a dozen languages, while the latter uses an RRG framework to explore in great detail the morphosyntactic expression of beneficiaries and recipients in a 100-language sample controlled for genetic and areal bias. See Zúñiga (i.p.) for a detailed discussion of some issues related to the definition barely mentioned here. 2.â•… The adverbs normally and typically in this definition naturally raise two related but different issues. First, they imply a judgment on frequency values that reflects our impression after browsing through grammars and specialized studies. It goes without saying that it is data to be analyzed by quantitative-typological studies still to be conducted that can either substantiate or contradict such an impression. Second, they make the wording sound as if the definition were formulated in terms of Prototype Theory, which is not far from the truth: we do not want to exclude “peripheral” beneficiaries from the definition altogether—but see further down in the main body of text concerning the notion of prototype.
Introduction
had brought the olives as a gift and was expecting a positive reaction on the part of her hosts—an arguably rather marked context that is nonetheless possible). The above definition raises a number of important questions that we have deliberately decided not to address in formal terms here. For instance, the beneficiary has been thought of by some scholars as a role with both prototypical and peripheral instantiations, e.g.: “[t]he situation is prototypically controlled, i.e. includes an Actor— the benefactor. A prototypically benefactive situation also includes an Undergoer, which is created (i.e. effected) or affected as benefactum for the benefit of the beneficiary” (Lehmann et al. 2000: 68, our translation).3 In such a view, examples like he bought her flowers are prototypical while those like a puff of wind opened the door for me and the patient coughed for the physician are not. This in turn leads to several related questions we merely limit ourselves to mentioning here. How shall one best treat unspecified/implicit undergoers (sing [a song], eat [something], etc.), which are often not overtly realized in a clause but are arguably included in the semantic representation of the predicate and its arguments? Do canonical benefactive constructions give rise to non-prototypical readings in addition to the default (i.e. prototypical) but unavailable reading when applied to non-canonical predicate-argument scenarios? In other words, what is the status of the non-prototypical meanings: can they be conventionally or conversationally triggered “nuances” in some contexts but meanings firmly associated with particular constructions in others? Are these non-prototypical meanings best thought of as different roles from the beneficiary sensu stricto? In he bought her flowers, under which conditions is the fact that she was relieved from buying flowers herself conveyed by the construction in addition to the fact that he bought the items in order to give them to her? In order to make cross-linguistic comparison as easy as possible, however, but without claiming that such a choice is necessarily the best one, we have decided (a)€not to provide too strict a definition of the beneficiary, and (b) not to articulate the definition in terms of a prototype that includes an agent and a theme. As a consequence, different beneficial effects are granted equal status for descriptive purposes, and instead of postulating additional semantic roles we have preferred to treat different participants covered by our definition as somewhat loose subtypes of the beneficiary role that can, but need not, have specialized morphosyntactic strategies associated to them. This is part of a working definition, of course, whose adequacy remains to be further investigated.
3.â•… Shibatani’s (1996) account centering on the semantics of transfer constructions is also a case in point.
Seppo Kittilä & Fernando Zúñiga
We elaborate in what follows on the central explicit aspects of the working definition given above, viz. the non-obligatoriness of the participant, the fact that beneficiaries are neither agents nor patients, the beneficial quality of the effect, and the beneficiary’s animacy.
1.1â•… Beneficiaries are typically optional Beneficiaries have features in common with recipients, and the two semantic roles are coded alike or at least similarly in a number of languages (e.g. Finnish, German, Lokono Dian, Warlpiri, etc., see also 4.1 below). However, recipients are often obligatory arguments of verbs, most notably with verbs of giving (see Kittilä 2006: 585f), while the non-obligatory nature of beneficiaries is manifested in the fact that they can often be omitted. In other words, recipients can be semantic arguments of non-derived verbs while beneficiaries are peripheral. The more obligatory nature of recipients is also reflected in their more core-like morphosyntactic coding; recipients bear core-like case marking in many languages, while beneficiaries are more often coded by non-core cases or adpositions with non-applicativized verbs, like for in some of the English examples given above. Nevertheless, beneficiaries may be obligatory and marked as core syntactic arguments with applicativized verbs. By the same token, in some languages, such as Kham, beneficiaries are obligatory participants with some verbs (Watters 2002: 250).
1.2â•… Beneficiaries are neither agents nor primary targets of events Events are often intended to benefit, damage, or simply to affect a particular participant in a neutral way. In many cases the agent him/herself is a beneficiary as well (e.g. I brushed my teeth, I baked a cake (for me to eat it)), but in the case of typical beneficiaries—i.e. where the arguments are coded as such: I baked a cake for my daughter—, the agent and the beneficiary are not coreferential. While some languages simply ban agents from being beneficiaries in the same clause, others feature a specialized construction used in these cases, viz. a self-benefactive or autobenefactive construction (see Smith, this volume). Similarly, beneficiaries are not primary targets of the events they benefit from. Consequently, they are usually coded differently from patients and often take oblique marking, e.g. she opened the door for me. Beneficiaries are often affected by the result of an event (this is particularly evident in the case of beneficiaries that are also recipients), while patients are integral parts of the event that results in the affectedness of the beneficiary. For example, in the event ‘he painted the house for me’, the agent targets his/her action primarily at the house, while the beneficiary may be affected by the result of the event in question, e.g. in having a painted house after the event has occurred. Moreover, the effect on the patient appears to be more relevant to the lexical semantics of the denoted event; different lexemes are used for coding events with
Introduction
different kinds of patients, while neither the kind of beneficiary nor the nature of the benefaction usually yield lexical differences in the predicates involved.
1.3â•… The effect is beneficial This feature distinguishes the beneficiary from two related but different roles: the maleficiary and the “affectee” (which can be conceived of as a kind of macro-role comprising both beneficiaries and maleficiaries). The maleficiary constitutes the opposite of the beneficiary in that malefactive events affect the relevant participant adversely. Interestingly enough, malefaction has received considerably less attention than benefaction in the linguistic literature, with the notable exception of detailed studies of the so-called adversative passive constructions in Japanese and Korean (see Tsuboi and also Smith, this volume). However, the feature [beneficial effect] seems to be rather insignificant crosslinguistically: many languages do not make a formal distinction between beneficiaries and maleficiaries but simply mark affectees, or at least the coding of the latter via case or adpositions is somehow less grammaticalized, systematic and/or elaborated (see Section 4.3 for the polysemy of these roles). Examples below illustrate an explicit formal opposition between two different constructions, viz. a benefactive (a) and a malefactive one (b):
(1) Amharic (Amberber 2002: 58)
a.
Aster-in fәrrәd-ә-ll-at. A.-acc judge.perf-3m.sbj-appl.ben-3f.obj ‘He judged in Aster’s favor.’
b. Aster-in fәrrәd-ә-bb-at. A.-acc judge.perf-3m.sbj-appl.mal-3f.obj ‘He judged to the disadvantage of Aster.’
(2) Finnish (personal knowledge)
a.
Vanhempi piir-si lapse-lle hevose-n. parent.nom draw-3sg.pst child-all horse-acc ‘A parent drew a horse for the child.’
b. Minu-lta hajo-si pyörä. 1sg-abl break-3sg.pst bicycle.nom ‘My bicycle broke down on me.’
Amharic has two distinct applicative(-like) affixes depending on whether the effect is beneficial or adverse, viz. -bb with beneficiaries and -ll with maleficiaries; in contrast, both arguments bear accusative case marking. In Finnish, a similar difference is expressed by case marking; beneficiary bears allative coding, while maleficiary occurs in the ablative case.
Seppo Kittilä & Fernando Zúñiga
The (inherently) beneficial nature of the affectedness can also be claimed to contribute to the formal differences of recipient and beneficiary found, e.g. in Mupun and Finnish:
(3) Mupun (Frajzyngier 1993: 227)
a.
N-sin siwol n-wur (mbә laa fin). 1sg-give money to-3m for school 3m) ‘I gave him money (for the school).’
b. *N-sin brәŋ-nә mbә Audu. 1sg-give horse-def for A. Intended: ‘I gave the horse to Audu.’
(4) Finnish (personal knowledge)
a.
Nainen anto-i poja-lle kirja-n. woman.nom give-3sg.pst boy-all book-acc ‘The woman gave the boy a/the book.’
b. *Nainen anto-i kirja-n poja-n vuoksi. woman.nom give-3sg.pst book-acc boy-gen for Intended: ‘The woman gave the boy a/the book.’
In Mupun and Finnish, recipients cannot be formally treated as beneficiaries, as shown by the ungrammaticality of (3b) and (4b). As noted above (and as will be discussed in more detail below), recipients can often be seen as beneficiaries. However, this is not entailed, and reception may also be highly detrimental, as in a lunatic sent me a package with anthrax. Recipients are therefore only optionally beneficiaries, while beneficiaries are only optionally recipients (in the case of recipient-beneficiaries), which is manifested in their coding.
1.4â•… Beneficiaries are usually animate Beneficiaries usually have an animate referent (see e.g. Blake 1994: 70 and Lehmann€et€al. 2000: 68). This is not surprising, since benefaction usually implies that the beneficiary is capable of using the result of the denoted event for his/her purposes in some way (similarly, recipients are usually animate, while goals are inanimate). Animacy also distinguishes beneficiaries from many other peripheral participants such as instruments and some instances of locatives. Moreover, benefaction is intimately associated with reasons or (indirect) causes, which explains the polysemy of benefaction/reason markers attested in many languages (see below for more discussion). For instance, the Finnish adposition vuoksi ‘for’ encodes benefaction with animate referents and reason with inanimate referents:
(5) Finnish (personal knowledge)
a.
Hän tek-i se-n minu-n vuokse-ni. s/he.nom do-3sg.pst it-acc 1sg-gen for-1sg.psr ‘S/he did it for me.’
Introduction
b. Jäätelö sul-i sähkökatko-n vuoksi. ice.cream melt-3sg.pst power.failure-gen for/because.of ‘The ice cream melted because of the power failure.’
Example (5a) can also be seen as involving reason, which is almost always present with beneficiaries (the denoted event would probably not have occurred without the beneficiary being present), but it also includes benefaction. On the other hand, in (5b) benefaction is absent altogether, and the postposition vuoksi expresses reason only. The relevant difference here is represented by animacy; both animate and inanimate participants are potential causers of events, but only animate participants are capable of genuinely undergoing benefaction.
2.â•… Formal mechanisms of beneficiary/maleficiary coding In this section, we will illustrate the central mechanisms languages employ for the linguistic coding of benefaction and malefaction. The ones illustrated and discussed here include cases, adpositions, serial verb constructions, and applicativization, and any given language may employ more than one of these mechanisms. Since some languages formally distinguish between different instances of beneficiaries and maleficiaries (specific beneficiaries) while others do not (general beneficiaries), we will also briefly discuss this issue below.
2.1â•… Major mechanisms 2.1.1â•… Case Examples of languages that use case morphology for expressing benefaction and/or malefaction are easy to find and include Lokono Dian, German, Finnish, Mordvinian, and Basque. The dative is probably the case most frequently used cross-linguistically to encode beneficiaries (6). Languages employ allative (7), ablative (8) and specific benefactive cases (9–10) for coding benefaction. Typically, these cases are polysemous; the expression of benefaction is only one of its functions (see Section 4 for the polysemy of these markers in general and Rice & Kabata 2007 for the polysemy of the allative in particular).
(6) Czech (Janda 1993: 57)
Ludmila mu uvařila kaši. L.nom him.dat cooked kasha.acc ‘Ludmila cooked kasha for him.’
(7) Finnish (personal knowledge)
Henkilö ava-si ove-n yksilö-lle. person.nom open-3sg.pst door-acc individual-all ‘A person opened the door for an individual.’
Seppo Kittilä & Fernando Zúñiga
(8) Kuuk Thayorre (Alice Gaby, p.c.)
Nhangun may yump ngathanma (ngathun). 3sg.dat food make.imper 1sg.abl (1sg.dat) ‘Make him some food for me (on my behalf).’
(9) Iranian Azari (Denghani 2000: 146ff, 158f)
Män kitab-i ušag-ičin al-di-m. 1sg book-acc child-ben buy-pst-1sg ‘I bought the book for the child.’ (10) Basque (Laka 1996) Amaia-rentzat erosi dut opari-a. A.-sg.ben bought aux.1sg.erg.3sg.abs present-det ‘I have bought the present for Amaia.’
2.1.2â•… Adpositions Adpositional coding of benefaction and malefaction is also frequently attested across languages. Both prepositions and postpositions can be used, as the examples in (11)–(13) show (see also the English translations): (11) Icelandic (example courtesy of Jóhanna Barðdal) Hann lagði bílnum fyrir mig. he.nom park.pst car.acc for 1sg.acc ‘He parked the car for me.’ (12) Sulka (Tharp 1996: 141) Ko-kol a-hor ngang e-pruo. 1sg.pst-get sg-leaf for pn-P. ‘I got a leaf for Pruo.’ (13) Finnish (personal knowledge) Men-i-n kaupunki-in häne-n harmikse-en. go-pst-1sg town-ill 3sg-gen to.the.detriment-3.psr ‘I went to town to his/her detriment.’
In Icelandic and Sulka, prepositions are used for coding benefaction, while in Finnish some postpositions (in addition to cases) express malefaction (and also benefaction).
2.1.3â•… Serial verb constructions In many languages that largely lack case morphology (and in some cases also adpositions), serial verb constructions are a productive way of expressing benefaction and malefaction. Some examples are found in (14)–(16)
Introduction
(14) Thai (a–b Bisang 1992: 366f; c–f Jenny, this volume) a.
Deeŋ jiŋ nóg hâj Sùdaa. D. shoot bird give S. ‘Deng shoots a bird for Sudaa.’
b. Deeŋ paj talàad hâj Sùdaa. D. go market give S. ‘Deng is going to the market for Sudaa.’ c.
Khaˇw pay tәlàat thεεn phш ˆ әn. 3hum go market replace friend ‘He goes to the market in his friend’s stead.’
´ ш khәnoˇm líәŋ luˆuk. d. Mεˆε sш mother buy sweets feed child ‘The mother buys sweets for her children (to eat).’ e.
´ ш naˇŋsш ˇ ш fàak Phfˆf sш luˆuk. father buy book entrust child ‘The father bought a book for his son (as a present).’
f.
´ ш kεεŋ phш ` ә 1εεŋ sш 1am. Daeng buy curry set.aside Dam ‘Daeng bought some curry for Dam (besides buying some for herself).’
(15) Vietnamese (Bisang 1992: 315) Bá đua hính tôi cho ban giùm tôi. you.hon bring photo I give friend help I ‘Would you (honorific) please give this photo to my friend for me.’ (16) Khmer (Bisang 1992: 425–426) a.
Kh\om nшŋ b>t tvì:әr ôaoy. I fut close door give ‘I close the door for him.’
b. Kfәt baәk tvì:әr sfmrap kh\om. 3sg open door use 1sg ‘He opened the door for me (in order for it to remain open and I can make a use of this favor).’
The most typical verb used for coding benefaction seems to be ‘give’ (in addition to the languages above, examples include Mandarin Chinese, Hokkien Chinese and Berbice Dutch Creole, among many others). In many cases, ‘give’ has largely lost its original semantics and is best considered a semantically vacuous verb, which is used for adding a beneficiary/recipient-like participant into the denoted event (the verb may further even develop into an adposition). In addition, less grammaticalized verbs
 Seppo Kittilä & Fernando Zúñiga
like ‘replace’, ‘help’ and ‘use’ occur in serial verb constructions coding benefaction, as shown above.
2.1.4â•… Applicativization The last major mechanism of benefaction/malefaction expression is applicativization, which is also common across languages. Some examples are given in (17)–(20): (17) Southeastern Tepehuan (Willett 1991: 76f, 182f) Ma’n-ap jiñ-som-dya-’ gu cutun. one-2sg 1sg-sew-appl-fut art shirt ‘Please sew a shirt for me.’ (18) Kinyarwanda (Kimenyi 1980: 32) Umugóre a-rá-kor-er-a umugabo. woman she.sbj-pres-work-ben-asp man ‘The woman is working for the man.’ (19) Kunuz Nubian (Abdel-Hafiz 1988: 114) a.
E:n kade:-g sukki-de:s-s-u. woman dress-acc wash-ben-pst-3sg ‘The woman washed the dresses for me.’
b. Ay-gi ir:-g noddi-de:s-s-a. 1sg-acc rope-acc cut-ben-pst-3pl ‘They cut the rope (to my detriment).’ (20) Ika (Frank 1990: 69) Kafé Pablo-se% k-ngei%-na-rua ni. coffee P.-loc peri-sell-distpst-1sg cert ‘I sold coffee to Pablo’
The examples in (17)–(19) illustrate rather typical instances of applicatives employed for expressing benefaction. The verb takes an applicative affix, and the beneficiary or maleficiary introduced to the clause takes direct object morphology and occupies its slot. In the construction found in Ika (20), by contrast, the beneficiary appears as a non-core argument, and the construction does not constitute a case of applicativization proper but of something similar that Thomas Payne has called “nucleative” (Payne 2000). Observe that in Kunuz Nubian, the same applicative morpheme can introduce both beneficiaries (19a) and maleficiaries (19b).
2.2â•… Specific vs. general beneficiary markers In addition to the formal variation in their coding, beneficiary markers differ with respect to their semantic specificity. Benefaction can either be quite specific (and
Introduction
therefore express only one subtype of benefaction) or not distinguish between subtypes (or even between beneficiaries and maleficiaries). These cases are examined below.
2.2.1â•… Specific beneficiary markers Specific beneficiary markers are those that mark a specific type of benefaction only (such as plain benefaction, see below), and whose semantics is constant across different predicate classes. Examples are found in (21)–(23): (21) Finnish (personal knowledge) Henkilö men-i kaupunki-in puolesta-ni. person.nom go-3sg.pst town-ill instead.of-1sg.psr ‘A person went to town instead of me.’ (22) Khmer (Bisang 1992: 426) Kfәt baәk tvì:әr sfmrap kh\om. 3sg open door use 1sg ‘He opened the door for me (in order for it to remain open and for me to make use of this favor).’ (23) Korean (Song, this volume) a.
Kiho-ka Yenghi-eykey mwun-ul yel-e-cwu-ess-ta K.-nom Y.-dat door-acc open-pf-give-pst-ind ‘Keeho opened the door for Yonghee.’
b. *Kiho-ka Yenghi-eykey mwun-ul tat-a-cwu-ess-ta K.-nom Y.-dat door-acc close-pf-give-pst-ind ‘Keeho closed the door for Yonghee.’
All the examples above illustrate specific beneficiary markers, but there are some differences between the individual cases. The Finnish postposition puolesta ‘instead of ’ expresses only deputative benefaction and is semantically constant, i.e. its reading does not vary according to the semantics of the construction it appears in. The Khmer example also illustrates a beneficiary marker with a very specific semantics. The verb sfmrap ‘use’ contrasts with the general verb ôaoy ‘give’ in the expression of benefaction; ôaoy can code a range of beneficiaries, while sfmrap can only express true “engager-beneficiaries” (see Song, this volume, and Example 28 below). The examples from Korean illustrate a different type of specific marker. Similarly to Finnish and Khmer, the beneficiary construction of Korean exemplified in (23) is semantically specific, and it can only encode engager-beneficiaries. In contrast to Finnish, however, the distribution of the construction is rather limited, and it can only appear in cases in which the beneficiary can somehow engage with the theme. This is possible in (23a), in which the beneficiary can use the opened door, for example, for entering a room. On the other hand, a closed door cannot be engaged with in any customarily/culturally

 Seppo Kittilä & Fernando Zúñiga
relevant fashion, which yields (23b) unacceptable not on strictly grammatical or semantic grounds but due to a stringent pragmatic restriction.
2.2.2â•… General beneficiary markers General markers also express benefaction, but their exact meaning varies according to the semantics of the construction they appear in. A very typical example of this is provided by the dative of Germanic and Slavic languages, exemplified for Czech in (24). Other examples are found in (25) and (26): (24) Czech (Janda 1993: 57, 87) a.
Ludmila mu uvařila kaši. L.nom him.dat cooked kasha.acc ‘Ludmila cooked kasha for him.’
b. Zlomil jí ruku. broke her.dat arm.acc ‘He broke her arm.’ (25) Finnish (personal knowledge) a.
Isä raken-si lapse-lle legotalo-n. father.nom build-3sg.pst child-all Lego.house-acc ‘The father built the child a Lego-house.’
b. Henkilö ava-si yksilö-lle ikkuna-n. person.nom open-3sg.pst individual-all window-acc ‘A person opened a window for the individual.’ c.
Vanhempi laula-a lapse-lle. parent.nom sing-3sg.pres child-all ‘The parent is singing to the child.’
d. Ystävä-ni tek-i tämä-n minu-lle. friend-1sg.psr do-3sg.pst this-acc 1sg-all ‘My friend did this to me.’ (Maleficiary) (26) Kham (Watters 2002: 249) a.
No-e ŋa-lai ŋa-ka:h sәih-d-y-ã:-ke-o. 3sg.m-erg 1sg-dat 1sg-dog kill-nfut-ben-1sg-perf-3sg ‘He killed my dog (to my detriment).’
b. ža: no-lai gohr jәi-dә ŋa-e-ke. 1sg 3sg.m-dat plow make-nfut 1sg-ben-perf ‘I made a plow for him.’
The dative in Germanic and Slavic languages can express different subtypes of benefaction (see below for discussion of these) and is therefore semantically rather general, especially bearing in mind that under appropriate conditions it can also code maleficiaries. In other words, the readings of the dative vary according to the semantics of
Introduction 
the construction it occurs in. The Finnish allative is also rather general in nature, and it can express all the three major types of benefaction proposed by Van Valin and LaPolla (see Section 3). In addition, the allative can also express malefaction, but its use in this function is very limited and highly context-dependent. The limited nature of the allative is manifest e.g. in (25d): if tämän ‘this.acc’ refers to something concrete that is created by the denoted process, it codes a recipient-beneficiary, in which case the reading would be something like ‘he made the house for me’. A malefaction reading, in turn, is the only possible one if tämän refers to something that the agent has done to the detriment of the maleficiary, i.e. it cannot refer to anything concrete (in which case the reading would be, for example, ‘he killed my dog to my detriment’). In Kham, the same applicative marker can introduce both beneficiaries and maleficiaries, as shown in (26) above. A somewhat different case is attested in Shipibo-Konibo. In this language, the applicative suffix -xon can introduce both beneficiaries and maleficiaries to the clause: (27) Shipibo-Konibo (Valenzuela 1997: 117, 127) a.
Pexe Piko-n-ra ea-ø ka-xon-ke. P. P.-erg-ass 1-abs go-ben-compl ‘Pexe Piko left (and this affects me either positively or negatively).’
b. Ea-ø bimi-ø pota-xon-we. I-abs fruit-abs throw-ben-inc ‘Throw me some fruits.’ (Beneficiary only) c.
Ja-n-ra ea-ø ochíti-ø bo-on(aan)-ke. 3-erg-ass 1-abs dog-abs carry-mal-compl ‘He took the dog from me (to my detriment).’ (Maleficiary only)
However, the kind of variation illustrated in (27a) only occurs in intransitive clauses. In transitive clauses, as in (27b), the introduced argument can only bear the role of beneficiary. Shipibo-Konibo also has a specific maleficiary marker, illustrated in (27c). It is interesting to note here that the maleficiary applicative marker can appear only in transitive clauses, in very much the same way the marker -xon can only have a beneficiary reading in transitive clauses. Transitivity is therefore the conditioning factor of (non-)specificity here.
3.â•…Semantically determined variation in the coding of beneficiaries and maleficiaries The cross-linguistic variation in the coding of beneficiaries examined in the preceding section is attested for all functional domains, and benefaction is no exception. In this section, we will proceed to discussing features that affect the coding of beneficiaries
 Seppo Kittilä & Fernando Zúñiga
within languages. It is not the goal of this section to illustrate semantically determined variation in beneficiary coding exhaustively, but rather to give a short introduction to the relevant features. Detailed studies of this variation in individual languages are found in other chapters of this volume.
3.1â•… Nature of benefaction The linguistic literature has acknowledged that there may be different subtypes of benefaction that need to be distinguished in a given language. In other words, beneficiaries benefit from events in different ways, which has consequences for their linguistic coding. For example, Van Valin and LaPolla (1997: 383f) mention three subtypes of what we have called beneficiary here, which they label “plain benefactive”, “deputative-benefactive” and “benefactive-recipient” (but note that Van Valin and LaPolla use the label beneficiary only for the first two subtypes). In the first case, benefaction consists in amusing/entertaining the beneficiary, as in the mother is singing for the children. In the second case, someone is substituting the beneficiary as the agent of the denoted event, e.g. in John painted the house instead of me. In the third case, the beneficiary is also a recipient, as in the father baked his daughter a cake. Along similar lines, Kittilä (2005) makes a distinction between (pure) beneficiaries and recipient-beneficiaries. The former notion more or less includes Van Valin and LaPolla’s plain and deputative-benefactives, while the latter notion is very similar to their benefactive-recipient. Song (this volume) makes a similar distinction between different subtypes of benefaction, but he speaks of “engager-beneficiaries” instead of recipient-beneficiaries. The reason for this is that a participant does not need to be a genuine recipient for being coded as a recipient-benefactive, but it suffices that s/he somehow engages with the theme. Consider (28): (28) Korean (Song, this volume) a.
Kiho-ka Yenghi-eykey ipwul-ul kkal-a-cwu-ess-ta. K.-nom Y.-dat futon-acc lay.out-pf-give-pst-ind ‘Keeho laid out the futon for Yonghee.’
b. *Kiho-ka Yenghi-eykey ipwul-ul kay-e-cwu-ess-ta. K.-nom Y.-dat futon-acc fold.up.put.away-pf-give-pst-ind (‘Keeho folded up the futon and put it away for Yonghee.’)
Example (28a) does not necessarily mean that the futon that has been laid out enters the beneficiary’s domain of possession in the strict sense of the term, and therefore Song does not regard Yonghee as a genuine recipient in (28a). The reason for the difference in acceptability between (28a) and (28b) lies in the nature of the denoted event: the beneficiary can engage with a laid out futon, for example by lying on it. On the other hand, a futon that is folded up cannot be engaged with in a similar fashion,
Introduction 
which yields (28b) ungrammatical. The label Song uses for the type of beneficiary in (28a) is, then, arguably more accurate than the label recipient-beneficiary—not only for Korean but also for Finnish—, because the difference in acceptability between (28a) and (28b) is not determined by the presence or lack of reception but by whether the beneficiary can engage with the theme in some plausible way.4 Since engagement is possible only if there is a theme to engage with, constructions with an engagerbeneficiary are usually possible only with transitive clauses.
3.2â•… Other features relevant to beneficiary coding In addition to the nature of benefaction itself, other features also contribute to the linguistic coding of benefaction. Relevant in this regard are at least the following features. The list is not exhaustive, and future studies of benefaction may show that some revisions are necessary: (i) animacy of agent and beneficiary, (ii) volitionality/willingness of agent and beneficiary, (iii) consciousness of the participants of the beneficial affectedness, and (iv) capacity on the part of the beneficiary to perform the action him/herself. First, as noted above, the beneficiary is usually an animate participant. Remember that in the case of engager-benefaction it is essential that the beneficiary is capable of (actively) engaging with the theme, which is possible only if the beneficiary is animate. For example, constructions with an inanimate beneficiary are often deemed ungrammatical in Finnish, or at least less felicitous than constructions with an animate beneficiary: (29) Finnish (personal knowledge) a.
Vanhempi leipo-i kaku-n lapse-lle. parent.nom bake-3sg.pst cake-acc child-all ‘The parent baked a cake for the child.’
b. */??Vanhempi leipo-i kaku-n talo-lle. parent.nom bake-3sg.pst cake-acc house-all Intended: ‘The parent baked a cake for the house.’
4.â•… Spike Gildea (p.c.) rightly points out that Song’s approach raises a number of significant questions. The most relevant here bears relation to the fact that there is no recipient in the strict sense of the term in an English sentence like I handed him my glasses to hold for me while I adjusted my seat either—if a recipient sensu stricto means a permanent, rather than a temporary, intended possessor (some languages of the Caucasus, e.g. code these two differently; we are indebted to Michael Daniel, p.c., for this observation). In our opinion, Song’s proposal is analytically interesting and worth taking into consideration in a detailed discussion of how to best account for the variation found in the linguistic expression of benefaction.
 Seppo Kittilä & Fernando Zúñiga
The only difference between (29a) and (29b) resides in the animacy of the beneficiary. (29a) illustrates the paradigmatic case of benefaction, and the construction is thus naturally grammatical, while inanimate participants cannot bear the role of recipientbeneficiary, which renders (29b) ungrammatical (for more examples and discussion see Kittilä, this volume). The animacy of the agent may also be relevant to the linguistic coding of benefaction (see also Lehmann et al. 2000: 68): (30) Finnish (personal knowledge) a.
Tuuli tuho-si vaja-n puolesta-ni. wind.nom destroy-3sg.pst shed-acc instead.of-1sg.psr ‘the wind destroyed the shed instead of me’
b. Tuuli tuhosi vajan *takia-ni/*vuokse-ni/*minu-a varten/*minu-lle. ╛↜for-1sg.psr/for-1sg.psr/1sg-part for/1sg-all Intended: ‘The wind destroyed the shed for me.’
Example (30a) is perhaps somewhat marginal, but it is possible in case I have intended to destroy a shed, but instead a strong wind does this instead of me. The animacy of the agent is thus not relevant with the postposition puolesta: inanimate agents are also allowed. On the other hand, all the mechanisms exemplified in (30b) are ungrammatical: their semantics implies consciousness from the agent, which is lacking in the case of inanimates (see Kittilä, this volume for more details). The (in)animacy of agent and beneficiary is closely related to their willingness and volitionality. Finnish is a language in which the coding of beneficiary is, among other things, influenced by these features. Consider the following examples: (31) Finnish (personal knowledge) a.
Minä pesi-n auto-n puolesta-si. 1sg.nom wash.pst-1sg car-acc instead-2sg.psr ‘I washed the car instead of you.’
b. Minä pesi-n auto-n takia-si. 1sg.nom wash.pst-1sg car-acc because-2sg.psr ‘I washed the car because of you.’
Both (31a) and (31b) can be regarded as instances of deputative benefaction, even though (31b) is also intimately related to reason or (indirect) cause (see Kittilä, this volume for more details). The difference is that in (31a) the agent washes the car willingly instead of the beneficiary, while in (31b) s/he is more reluctant to do so. Consciousness on the part of the beneficiary of his/her beneficial affectedness has consequences for the linguistic coding of the relevant participant, for example, in Thai, as shown in (32):
Introduction 
(32) Thai (Jenny, this volume) a.
Mεˆε rfˆfŋ phleeŋ hây lûuk. mother sing song give child ‘The mother is singing a song to her child.’
b. Mεˆε rfˆfŋ phleeŋ phш ˆ ә lûuk. mother sing song for child ‘The mother is singing a song for her child.’
The child needs to be aware of the beneficial act done for him/her in (32a), while this is not necessary in (32b). As Jenny puts it: “While it sounds odd to add ‘… but the child doesn’t know it’ to (32a), it is perfectly natural to do so in (32b)”. Beneficiaries differ according to how probable or expected their participation is in the event performed for them. There are cases in which the beneficiary could have been the agent in the denoted event him/herself, and cases in which the beneficiary was not an expected agent. Finnish is a language in which this difference has consequences for beneficiary coding: (33) Finnish (personal knowledge) a.
isä ava-si ikkuna-n lapse-n puolesta father.nom open-3sg.pst window-acc child-gen instead.of ‘The father opened the window instead of the child.’
b. lapsi ava-si ikkuna-n koira-lle-en child.nom open-3sg.pst window-acc dog-all-3.psr ‘The child opened the window for his/her dog.’ c.
isä vaihto-i vauva-lle vaipa-n father.nom change.pst-3sg baby-all diaper-acc ‘The father changed the baby’s diaper.’
d. ??lapsi ava-si ikkuna-n koira-n-sa puolesta child.nom open-3sg.pst window-acc dog-gen-3.psr instead.of Intended: ‘The child opened the window instead of his/her dog.’ e. ??/*isä vaihto-i vaipa-n vauva-n puolesta father.nom change.pst-3sg diaper-acc baby-gen instead.of Intended: ‘The father changed the diaper instead of the baby.’
In Finnish, the postposition puolesta ‘instead of ’ and the allative case can both be used to denote deputative-benefaction (for more details, see Kittilä, this volume). The two mechanisms are, however, not in free variation, but the formal variation has a semantic basis. The postposition puolesta is used whenever the beneficiary him/herself would have been a possible agent in the given event. The allative, in turn, is more felicitous if the beneficiary is not an expected agent in the event done in his/her stead. This difference yields (33d) and (33e) marginal at best; the use of puolesta implies that the beneficiary could have performed the action him/her/itself, which is a relatively unlikely reading in these cases.
 Seppo Kittilä & Fernando Zúñiga
4.â•… On the polysemy of benefactive markers Across languages, beneficiaries share common formal features with other arguments, such as recipients, (external) possessors, maleficiaries, experiencers, reasons/(indirect) causes, goals and causees. In this section, we will discuss the semantics of the beneficiary from the viewpoint of these attested instances of polysemy.5 All the cases discussed below (perhaps with the exception of causees) also have a semantic basis. For example, recipients are often beneficiaries, which makes the polysemy of recipient and beneficiary not only natural, but also expected.
4.1â•… Recipient and beneficiary The polysemy of recipient and beneficiary is rather common across languages; consider: (34) Warlpiri (Simpson 1991: 206, 380f) a.
Karnta-ngku ka-rla kurdu-ku miyi yi-nyi… woman-erg pres-3.dat baby-dat food give-npst ‘The woman is giving food to the baby…’
b. Nyampu ka-rna-la kurlada wurduju-ma-ni. this pres-1sg-3.dat spear good-caus-npst ‘I am making this spear for him.’ c.
Karnta ka-rla kurdu-ku parnka-mi. woman pres-3.dat child-dat run-npst ‘The woman is running for the sake of the child.’
(35) Lolovoli Ambae (Hyslop 2001: 147f) a.
Mo bete lawe na tangaloi haro. real give dat acc people random ‘S/he gives to just anybody.’
b. Na=ni solo na gawu-mu lawe=go. 1sg.sbj=irr wash acc clothes-2sg.psr dat/ben=2sg.obj ‘I’ll wash your clothes for you.’ c.
Go=ni mas uli na leta lawe=eu. 2sg.sbj=irr must write acc letter dat/ben=1sg.obj ‘You must write a letter to/for me.’
In Warlpiri, the dative case can encode recipient (34a), recipient-benefaction (34b) and also deputative-kind of benefaction, as in (34c). In the Lolovoli dialect of Ambae, 5.â•… There are also interesting cases of polysemy of, e.g. locative/ablative markers used in order to encode malefaction, but we have not addressed this issue here.
Introduction 
some sentences are even ambiguous between a reception and a deputative benefaction reading, as in (35c). As noted above, the frequently attested polysemy of recipient and beneficiary is rather unsurprising. Recipients can be seen as beneficiaries as well, since reception in general is usually understood as beneficial. For example, the recipient in an event such as ‘s/he gave me a present’ is usually seen as a beneficiary, since the transfer is deemed beneficial, or at least it is meant to be beneficial. It should be stressed, however, that this “dual nature” of recipients is not entailed in the constructions and that reception can also be highly detrimental, as in the maniac sent me a letter bomb. Nevertheless, the relation between reception and benefaction is intimate enough to give rise to the polysemy illustrated in (34) and (35) and attested in numerous other languages as well.
4.2â•… Beneficiary and (external) possessor The polysemy of beneficiaries and possessors is also common. Examples (36) and (37) are taken from Oceanic languages (see also Lichtenberk 2002 for a more detailed discussion) and illustrate cases in which canonical possession and benefaction are coded formally alike. The polysemy of the marker is especially evident in Saliba, since (37) can have two readings. (36) Paamese (Crowley 1982: 212f, as cited in Lichtenberk 2002: 444) a.
oai ma-k water poss.cnt-1sg.psr ‘my water’
b. Kaik ki-muni ma-k. 2sg 2sg.dis-drink.3.obj poss.cnt-1sg.psr ‘Drink it for me.’ (37) Saliba (Margetts 1999: 324) Yo-gu puwaka ku-unui-Ø. cl1-1sg.psr pig 2sg-kill/catch-3sg.obj ‘Catch my pig/catch a pig for me.’
In addition to this kind of polysemy, there are many languages in which beneficiaries and external possessors receive the same formal treatment. Consider: (38) Finnish (personal knowledge) a.
Vanhempi kampa-si lapse-lta tuka-n. parent.nom comb-3sg.pst child-abl hair-acc ‘A parent combed the child’s hair.’
b. Sairaanhoitaja vaihto-i potilaa-lta lakana-t. nurse.nom change-3sg.pst patient-abl sheet-pl(acc) ‘A nurse changed the patient’s sheets.’
Seppo Kittilä & Fernando Zúñiga
(39) German (personal knowledge) a.
Er wusch dem Kind die Hände. he washed art.sg.dat child art.pl.acc hand.pl ‘He washed the child’s hands.’
b. Er baute seinem Kind ein Haus. he built his.dat child art.sg.acc house ‘He built his child a house.’
In Finnish, the ablative case can code both external possession (38a) and benefaction (38b). (The use of the ablative for benefaction coding is limited, however; see Kittilä, this volume.) In German (as in many other languages, especially in Europe), the dative is used to encode both external possession and benefaction. Note that in Finnish and German, the polysemy of benefaction and possession is restricted to external possession (unlike in Oceanic languages). The formal identity of (external) possession and benefaction can be accounted for as follows. Actions targeted at entities in my possession affect me in a more direct fashion, and usually in more ways, than actions that are only carried out instead of me. This is especially evident in the case of external possession, where the effect is often beneficial, which in turn explains the occurrence of cases like Finnish, German and Oceanic languages, in which the coding of possession and benefaction (partially) overlaps.
4.3â•… Beneficiary and maleficiary Even though beneficiaries and maleficiaries are distinguished from each other based on the nature of affectedness, there are numerous languages in which these roles receive identical coding (examples repeated and re-numbered here), cf. (40) Kunuz Nubian (Abdel-Hafiz 1988: 114) a.
E:n kade:-g sukki-de:s-s-u. woman dress-acc wash-appl-pst-3sg ‘The woman washed the dressed for me.’
b. Ay-gi ir:-g noddi-de:s-s-a. 1sg-acc rope-acc cut-appl-pst-3sg ‘They cut the rope (to my detriment).’ (41) Kham (Watters 2002: 249) a.
No-e ŋa-lai ŋa-ka:h sәih-d-y-ã:-ke-o. 3sg.m-erg 1sg-dat 1sg-dog kill-nfut-ben-1sg-perf-3sg ‘He killed my dog (to my detriment).’
b. ža: no-lai gohr jәi-dә ŋa-e-ke. 1sg 3sg.m-dat plow make-nfut 1sg-ben-perf ‘I made a plow for him.’
Introduction 
Such a polysemy (cf. also the Indo-European “dative of interest” in this context) may appear somewhat odd at first, but it is possibly due to the fact that benefaction and malefaction are subjective notions: a given event can be seen as either beneficial or detrimental for an indirectly affected participant depending on the context and the speaker’s judgment. It is usually not necessary to distinguish between these notions explicitly, since contextual information (or, in many cases, verbal semantics) frequently suffices for disambiguation. Beneficiaries and maleficiaries are both participants affected by events in an indirect fashion, which also makes them semantically similar enough to receive identical formal treatment as the underspecified “affectee” mentioned in 1.3 above. This underspecified category is the notion of ‘participant with respect to which the action holds’ or ‘participant for which the action is valid’ known from traditional Indo-European studies.
4.4â•… Beneficiary and experiencer The same device may mark both experiencers and beneficiaries, as in German mir ist kalt ‘I am cold’ and er hat mir ein Haus gebaut ‘he built a house for me’ (in this case, the device is the dative pronoun mir ‘me’), but also in Yurakaré and Dolakha Newar. Consider (42) Yurakaré (Van Gijn 2006: 155, 163) a.
Ti-n-ewe-ø. 1sg-io-sweep-3 ‘He swept my house for me.’
b. Ti-n-kukku. 1sg-io-nice ‘I like it.’ (43) Dolakha Newar (Genetti 2007: 335, 318) a.
Janta lukhā khoŋ-an bi-sin. 1sg.dat door open-partic give-imp ‘Open the door for me.’
b. Janta pyātāwāt-a. 1sg.dat hungry-3sg.pst ‘I am hungry.’
In Yurakaré, the same applicative affix may introduce both beneficiaries, as in (42a), and experiencers (42b) into clauses. In Dolakha Newar, both participant roles may be coded by the dative (which is a rather frequently attested instance of polysemy across languages). The connection between beneficiary and experiencer is perhaps best explained based on the rather low degree of affectedness displayed by both participants. The
Seppo Kittilä & Fernando Zúñiga
overall affectedness of beneficiaries is arguably rather low. A typical experience event, in turn, involves an experiencer and a stimulus. As has been noted, for example, by Næss (2007: 103), stimuli of experience are relatively unaffected by the event they partake in. These events do, however, have some effect on the experiencer, but the effect is less drastic than in the case of canonical patients. Moreover, the effect does not result from an act directly targeted at the beneficiary or at the experiencer. Experiencers are, unlike beneficiaries and maleficiaries, integral parts of events, and thus necessarily affected by them, but all these participants share the feature of lower, and also non-salient, affectedness. This distinguishes them from patients, and explains their identical coding (distinct from the marking of patients) attested in many languages.
4.5â•… Beneficiary and reason/(indirect) cause Another rather frequently attested instance of polysemy is illustrated by the polysemy of benefaction and reason or (indirect) cause.6 In Finnish, the postposition vuoksi and also the postposition takia (both meaning ‘for’ and ‘because of ’) can express both notions: (44) Finnish a.
Henkilö opettel-i suome-a yksilö-n vuoksi. person.nom learn-3sg.pst Finnish-part individual-gen for ‘A person learnt Finnish for an individual.’
b. Jäätelö sul-i sähkökatko-n vuoksi. ice.cream.nom melt-3sg.pst power.failure-gen because.of ‘The ice cream melted because of the power failure.’
In Tashelhiyt, the dative can mark beneficiaries, as in (45a) and also unintentional causers (“involuntary agents”, see Kittilä, to appear), as in (45b): (45) Tashelhiyt (Rapold, this volume) a.
Arki> a-sn t-ssnu-t mad ššta-n; … until dat-3pers 2-cook-2sg who/what eat.ipv-3pers ‘Until you cook something to eat for them; …’
6.â•… The term indirect causation is used in the literature as a cover label for at least two subtypes of causative meanings: there is an intermediary (Hindi, Jinghpaw) or speech (Yimas, Hungarian) involved between causer and causee. By contrast, with direct causation no intermediary is implied and/or there is some sort of physical manipulation of the causee on the part of the causer. In Mary made me laugh, thus, Mary can be either an indirect causer (e.g. she told a joke to me) or a direct causer (e.g. she tickled me). We use the somewhat cumbersome label reason/(indirect) cause here for participants that instigate states of affairs even less directly than via indirect causation (which in turn correlates with other values concerning some properties of the causer and/or the causee), e.g. in I laughed because of Mary (e.g. I remember something funny she did).
Introduction
b. T-dr a-s tngult ti>wr.din. ˙ 3sg.f-fall dat-3sg f.sg.nom.loaf.of.bread behind ‘The load of bread fell down behind to his detriment.’
In Kuku Yalanji (46), the dative-marked argument is animate in both cases, but the notion of benefaction is lacking in (46b), since the participant in question cannot benefit from the crying in any way: (46) Kuku Yalanji (Patz 2002: 132f) a.
Nyulu Cooktown dunga-y yaba-nka. 3sg.nom C. go-npst elder.brother-dat ‘He goes to Cooktown for his elder brother.’
b. Nyulu dingkar ngamu-nku badi-ny. 3sg.nom man.abs mother-dat cry-pst ‘The man cried because of his mother (she had died).’
This kind of polysemy makes sense to the extent that beneficiaries can be seen as (indirect) causes of the events they are involved in. Causation is usually less direct in such cases, since the agent often willingly participates in the denoted event for the beneficiary, but the event may not have occurred without the beneficiary being somehow involved in it. The beneficiary may even be fully unaware of causing an event to occur. In the case of indirect cause, in turn, the participation of the causing entity is usually less volitional, as seen in (44) and (45). In Finnish, the notion of indirect cause is strong both in (44a) and (44b), but it is clearly more prominent in (44b), since, due to the inanimacy of the causing entity, there cannot be benefaction. Since the difference is inferable from animacy of the argument followed by vuoksi/takia, the language does not need to have formally distinct mechanisms for expressing this difference. The polysemy of the dative is explainable largely on the same basis in Kuku Yalanji, since a deceased person is arguably a poor choice for the role of beneficiary, and the sense of causation is strong in (46b). In Tashelhiyt, the causer is not acting volitionally in (45b).
4.6â•… Beneficiary and goal Some languages code some instances of beneficiary identically to goals. Consider (47) Tashelhiyt Berber (Rapold, this volume) a.
Arki> a-sn t-ssnu-t mad ššta-n; … until dat-3pers 2-cook-2sg who/what eat.ipv-3pers ‘until you cook something to eat for them; …’
b. I-huwd ufrux lli i yan lxla… 3sg-go.down sg.m.nom.boy res.ref dat m.one desert ‘The boy went down to a desert…’
Seppo Kittilä & Fernando Zúñiga
(48) Japanese (Sachiko Sosa, personal communication) a.
Okaasan ga kodomo ni keeki o yakimasita. mother nom child dat cake acc bake.pst ‘The mother baked the child a cake.’
b. Kodomo ga gakkou ni hasitte ikimasita. child nom school dat run.inf go.pst ‘The child ran to the school.’ (49) Finnish (personal knowledge) a.
Vanhempi leipo-i kaku-n lapse-lle. parent.nom bake-3sg.pst cake-acc child-all ‘The parent baked the child a cake.’
b. Foneetikko juok-si koulu-lle. phonetician.nom run-3sg.pst school-all ‘The phonetician ran to the school.’
In Tashelhiyt and Japanese, dative case can code both beneficiaries (a-examples) and goals (b-examples). In Finnish, the allative case may code both of these roles. It is in order to note, however, that in Tashelhiyt the dative can code goals only with a very limited set of verbs, and it is usually the allative that codes goals. The semantic motivation of the polysemy illustrated in (47)–(49) is perhaps not as evident as in the case of beneficiaries and recipients. This polysemy can, however, also be explained semantically if we consider the different types of benefaction discussed in Section 3.1. In Japanese, only recipient-beneficiaries and plain beneficiaries may receive dative coding. The same applies largely to the Finnish allative, even though the allative can also code deputative benefaction in some cases (for more details see Kittilä, this volume). This means that beneficiaries may be coded as goals only when they share common features with recipients, or when the action is directly targeted at them, as in the mother is singing for her child. If the feature of direct targeting is lacking, the use of the dative or the allative is not possible. In Tasheyhlit, the situation is somewhat different, but the dative is nonetheless a polysemous case (see Rapold, this volume), which probably contributes to the identical coding of goal and beneficiary as well.
4.7â•… Beneficiary and causee The last instance of polysemy of beneficiary markers discussed here is the polysemy of beneficiary and causee found, for example, in Japanese and Turkish: (50) Japanese (Sachiko Sosa, Nobufumi Inaba, personal communication) a.
Okaasan ga kodomo ni keeki o yakimasita. mother nom child dat cake acc bake.pst ‘The mother baked the child a cake.’
Introduction
b. Pia ga Anssi ni baa e ik-ase-ta. P. nom A. dat bar to go-caus-pst ‘Pia let Anssi go to a bar.’ c.
Pia ga Anssi o baa e ik-ase-ta. P. nom A. acc bar to go-caus-pst ‘Pia made Anssi go to a bar.’
(51) Turkish (Jaklin Kornfilt, personal communication; Comrie 1975: 6) a.
Anne-m ban-a bir pasta pişir-di. mother-1sg I-dat a cake bake-pst ‘My mother baked a cake for me.’
b. Dişçi mektub-u müdür-e imzala-t-tı. dentist letter-acc director-dat sign-caus-pst ‘The dentist made the director sign the letter.’ c.
Ali Hasan-ı öl-dür-dü. A. H.-acc die-caus-past ‘Ali killed Hasan.’
In Japanese, both beneficiaries and causees may be followed by the dative particle ni, and both participants typically appear in the dative case in Turkish. These languages differ from each other, however, with respect to how the dative coding of the causee is motivated. In Japanese, in clauses with transitive predicates like yaku ‘bake’ it is not possible for both the causee and the patient to take accusative case; the causee has to appear in the dative (50a). With intransitive verbs like iku ‘go’ in baa e iku ‘go to a bar’, on the other hand, the assignment of dative case has a semantic basis: the causee occurs in the dative in the case of permissive causation, as also shown by the free translation of (50b); in the case of coercive causation, the causee bears accusative coding (50c). By contrast, the marking of the causee is solely determined by transitivity in Turkish: the causee occurs in the dative if the causativized clause is originally transitive, as in (51b); in the case of ditransitive and intransitive clauses (51c), dative coding of the causee is not possible. Observe that the two languages also differ as to the motivation of the polysemy. In Japanese, the identical coding of (some instances of) causees and beneficiaries can be given a semantic explanation based on the lower degree of affectedness of the datively coded participants. Highly affected participants, i.e. patients and causees of more direct causation, in turn, appear in the accusative. On the other hand, a semantic explanation is not valid for Turkish, since the identical coding of causees and beneficiaries is formally determined in that language, viz. by the transitivity of the original clause. The causee appears in the dative only in case the original clause is transitive, and thus the first free slot in the hierarchy [subject > direct object > indirect object > oblique] is that of the indirect object, which puts the causee in the dative (see Comrie 1975 for a more detailed discussion of this).
Seppo Kittilä & Fernando Zúñiga
5.â•… Contents of this volume The present volume includes four cross-linguistic/typological studies and thirteen articles that deal with individual languages. Creissels’ paper is a comprehensive study of benefactive applicative periphrases, i.e. biverbal constructions that introduce beneficiaries and related participants to a clause. Two of its main findings are the significant role played by ‘give’-verbs both in serialization and complex predicates worldwide and the comparatively limited role played by ‘take’-verbs in autobenefactive constructions in Eurasia. Smith’s study of the semantics of benefactives proposes two parameters along which benefactives are classified: (i) whether there is an agentive participant or not, and (ii) restrictions placed on the beneficiary’s reference. In addition, Smith proposes to include the English get-to-construction when discussing the notion of benefaction cross-linguistically. Radetzky and Smith propose a classification of languages according to whether benefactives and malefactives are subsumed under one unspecified expression (a pattern found in Europe, Africa, and the Americas) or the two notions correspond to separate morphosyntactic strategies (a pattern found mainly in India, Southeast Asia, and East Asia). Schmitdke-Bode’s paper is based on the author’s comprehensive typological study of purposive constructions and explores the connections between purposives and benefactives. Three articles deal with languages of the Americas. Kiyosawa and Gerdts provide a survey of the benefactive and malefactive uses of the applicative suffixes found in the Salish languages (Canada and USA). Censabella systematizes the specialized applicatives of Toba (Guaicuruan; Argentina), some of which encode benefaction and malefaction. Ziga presents an analysis of two applicatives found in Mapudungun (isolate; Chile and Argentina) according to which the “malefactive” was an originally neutral transitivizer that became semantically marked in some environments as a true “benefactive” entered the picture. Three articles deal with languages of Europe. Colleman’s is a cross-linguistic study of the English and the Dutch benefactive double-object construction on the one hand and the French and the German benefactive dative construction on the other. Kittil’s is an in-depth study of seven benefactive strategies found in Finnish (Uralic; Finland). Lacroix deals with an applicative derivation and an adpositional strategy used to express bene-/malefaction in Laz (Kartvelian; Turkey). Four papers deal with African languages. Quint describes and discusses the specialized benefactive and malefactive verb derivations in Koalib (Niger-Congo; Sudan). Vllmin’s article addresses the specialized benefactive and malefactive object agreement sets found in Gumer (Semitic; Ethiopia), as well as a biclausal benefactive strategy featuring a ‘say’-verb in the same language. Boyd describes the syntax and the semantics
Introduction
of “reflexive benefactives” in Chamba-Daka (Niger-Congo; Nigeria). Rapold explores the semantic roles covered by the dative in Tashelhiyt (Berber; Morocco). Finally, three papers deal with Asian languages. Jenny’s study of Thai (Tai-Kadai; Thailand) benefactive serial verb constructions suggests that an additional parameter to take into account is the consciousness/awareness on the part of the beneficiary. Song examines the semantics and pragmatics of benefactive complex predicates and the corresponding case markers found in Korean (isolate; Korea), and tsuboi addresses the role played by grammaticalization and pragmatics in the acceptability and the interpretation of the malefactive construction in Japanese (isolate; Japan).
6.â•… Topics for further investigation Much could be said about fruitful lines for further research; we will limit ourselves to mentioning four topics that are especially salient here. First, more detailed studies of benefactive constructions in a substantial number of languages are likely to refine and deepen our understanding of benefaction in ways that are not easy to predict— particularly with respect to specialized strategies that encode “less frequent” nuances of benefaction. Second, much more work needs to be done on malefaction, both for individual languages and cross-linguistically. Third, potentially very illuminating studies will address the division of labor between the different mechanisms in languages with multiple ways of coding benefaction. Finally, the formal/semantic/pragmatic similarities and differences between constructional benefactives and adpositional benefactives is an interesting topic worth pursuing.
References Abdel-Hafiz, Ahmed Sokarno. 1988. A Reference Grammar of Kunuz Nubian. Ph.D. dissertation, SUNY, Buffalo. Amberber, Mengistu. 2002. Verb Classes and Transitivity in Amharic. Munich: Lincom Europa. Bisang, Walter. 1992. Das Verb im Chinesischen, Hmong, Vietnamesischen, Thai und Khmer. Tübingen: Gunter Narr. Blake, Barry. 1994. Case. Cambridge: CUP. Comrie, Bernard. 1975. Causatives and universal grammar. Transactions of the Philological Society 1974: 1–32. Conti, Carmen. 2008. Receptores y beneficiarios. Estudio tipológico de la ditransitividad. Munich: Lincom Europa. Crowley, Terry. 1982. The Paamese Language of Vanuatu. Canberra: Pacific Linguistics. Denghani, Yavar. 2000. A Grammar of Iranian Azari. Munich: Lincom Europa.
Seppo Kittilä & Fernando Zúñiga Frajzyngier, Zygmunt. 1993. A Grammar of Mupun. Berlin: Dietrich Reimer. Frank, Paul. 1990. Ika Syntax. Dallas & Arlington TX: SIL & University of Texas. Genetti, Carol. 2007. A Grammar of Dolakha Newar. Berlin: Mouton de Gruyter. Hyslop, Catriona. 2001. The Lolovoli Dialect of the North-East Ambae Language, Vanuatu [Pacific Linguistics 515]. Canberra: Australian National University. Janda, Laura. 1993. A Geography of Case Semantics. The Czech Dative and the Russian Instrumental. Berlin: Mouton de Gruyter. Kittilä, Seppo. 2005. Recipient prominence vs. beneficiary prominence. Linguistic Typology 9: 269–297. Kittilä, Seppo. 2006. Anomaly of the verb ‘give’ explained by its high (formal and semantic) transitivity. Linguistics 44(3): 569–612. Kimenyi, Alexander. 1980. A Relational Grammar of Kinyarwanda. Berkeley CA: University of California Press. Laka, Itziar. 1996. Euskera, the Basque language.
, 18 September, 2008. Lehmann, Christian, Shin, Yong-Min & Verhoeven, Elisabeth. 2000. Direkte und indirekte partizipation. Zur Typologie der sprachlichen Repräsentation konzeptueller Relationen. Munich: Lincom Europa. Lichtenberk, Frantisek. 2002. The possessive-benefactive connection. Oceanic Linguistics 41(2): 439–474. Margetts, Anna. 1999. Valence and Transitivity in Saliba, an Oceanic Language of Papua New Guinea. Nijmegen: Max Planck Institute for Psycholinguistics. Naess, Åshild. 2007. Prototypical Transitivity [Typological Studies in Language 72]. Amsterdam: John Benjamins. Patz, Elisabeth. 2002. A Grammar of the Kuku Yalanji Language of North Queensland. Pacific Linguistics, Research School of Pacific and Asian Studies, The Australian National University. Payne, Thomas. 2000. Towards a substantive typology of applicative constructions. Ms, University of Oregon and SIL International. Rice, Sally & Kabata, Kaori. 2007. Crosslinguistic grammaticalization patterns of the allative. Linguistic Typology 11(3): 451–514. Shibatani, Masayoshi. 1996. Applicatives and benefactives: A cognitive account. In Grammatical Constructions. Their Form and Meaning, Masayoshi Shibatani & Sandra Thompson (eds.), 157–194. Oxford: OUP. Simpson, Jane. 1991. Warlpiri Morpho-Syntax: A Lexicalist Approach. Dordrecht: Kluwer. Tharp, Doug. 1996. Sulka grammar essentials. In Two Non-Austronesian Grammars from the Islands, John M. Clifton (ed.), 77–161. Ukarumpa, Papua New Guinea: SIL. Valenzuela, Pilar. 1997. Basic Verb Types and Argument Structures in Shipibo-Conibo. MA thesis, University of Oregon. Van Gijn, Rik. 2006. A Grammar of Yurakaré. Ph.D. dissertation, University of Nijmegen. Van Valin, Robert D., Jr. & LaPolla, Randy. 1997. Syntax. Cambridge: CUP. Watters, David. 2002. A Grammar of Kham. Cambridge: CUP. Willett, Thomas. 1991. A Reference Grammar of Southeastern Tepehuan. Dallas & Arlington TX: SIL & University of Texas. Zúñiga, Fernando. In preparation. The Grammar of Benefaction.
Benefactive applicative periphrases A typological approach Denis Creissels
University of Lyon After defining applicative periphrases as constructions in which a verb (the verb-operator) acts as a valency operator licensing the expression of an additional participant fulfilling a given semantic role in the event encoded by another verb (the lexical verb), this paper analyses applicative periphrases licensing beneficiaries. I recognize three formal types with respect to the grammatical nature of the two verb forms involved in the periphrasis and examine their distribution in the languages of the world. I then discuss grammaticalization paths leading to benefactive applicative periphrases, as well as further evolutions of benefactive applicative periphrases. The last two€sections are devoted to autobenefactive applicative periphrases and to the use of verbs other than ‘give’ as verb-operator in benefactive applicative periphrases.
1.â•… Introduction The recognition and analysis of periphrastic constructions functionally equivalent to morphological derivations encoding operations on verbal valency has been a recurrent topic in the study of passive and causative constructions. By contrast, applicative periphrases are largely neglected in the general literature on valency changes, although such constructions have been described in many languages. This paper examines applicative periphrases licensing benefactives in a typological perspective. This paper is organized as follows. In Section 2, after defining the notion of Â�applicative periphrasis, I define benefactive applicative periphrases as a semantic subtype of applicative periphrases, and I examine their general properties. In Section 3, I propose to distinguish three formal types of benefactive applicative periphrases according to the grammatical nature of the verb forms involved, and in Section 4, I examine their geographical distribution. In Section 5, I examine the grammaticalization paths in which benefactive applicative periphrases are involved. Section 6 is devoted to the use of ‘take’ and ‘eat’ in complex predicates expressing
Denis Creissels
autobenefaction. Section 7 examines the use of verbs other than ‘give’ in benefactive applicative periphrases.1
2.â•…Benefactive applicative periphrases: Definition and introductory remarks 2.1â•… Applicative periphrases In applicative constructions, a participant that cannot be treated as a core term of the corresponding non-applicative construction shows morphosyntactic properties identical or similar to those of the patient in the prototypical transitive construction. Applicative constructions may thus promote participants otherwise encoded as adjuncts to the status of core syntactic terms, but there are also obligatory applicatives (particularly common among Bantu languages), i.e. applicative constructions that constitute the only possible way to encode some semantic roles.2 Applicative periphrases are biverbal constructions functionally comparable to monoverbal constructions headed by applicative verb forms. The two verbs they involve can be designated as lexical verb (abbreviated as Vlex) and verb-operator (abbreviated as Vop). The lexical verb determines the type of event encoded by the applicative periphrasis, and the argument structure of the applicative periphrasis is the argument structure of the lexical verb augmented by an additional participant. Vop acts as a valency operator whose contribution to the construction is limited to licensing the expression of an additional participant fulfilling a given semantic role in the event encoded by the lexical verb, without modifying the morphosyntactic treatment of the other participants. The type of semantic role assigned by the verb acting as a valency operator in an applicative periphrasis has a historical connection with one of the roles assigned by the same verb when independently used in predicate function, but semantic evolutions may result in that, synchronically, verbs in valency operator function in applicative periphrases assign roles that sometimes have no direct connection with their argument structure as independent verbs.
1.â•… I am grateful to the following colleagues for their comments on earlier versions of this paper and/or their help in collecting data: Azeb Amha, Isabelle Bril, Injoo Choi-Jonin, Antoine Guillaume, Miyuki Ishibashi, Guillaume Jacques, Mathias Jenny, Renée Lambert-Brétière, Annie Montaut, Christiane Pilot-Raichoor, and Masayoshi Shibatani. I am particularly grateful to Karen Ebert for her help with Asian data. 2.â•… For a more detailed presentation of my own views on applicatives, see Creissels (2006:73–84).
Benefactive applicative periphrases 
2.2â•… Benefactive applicative periphrases A benefactive applicative periphrasis (henceforth abbreviated as BAP) is an applicative periphrasis licensing beneficiaries. BAPs constitute a particularly common type of applicative periphrases. Ex. (1) to (3) illustrate BAPs licensing the three sub-types of beneficiaries (recipient-beneficiaries, deputative beneficiaries, and plain beneficiaries) recognized by Van Valin & LaPolla (1997:383–4) (1)
Yoruba – Rowlands (1969:83) Rà á fún mi. buy 3sg give 1sg ‘Buy it for me.’ (recipient beneficiary)
(2)
Yoruba – Abraham (1962:348) Ó jís» é» fún mi. 3sg go_on_an_errand give 1sg ‘He went on an errand for me.’ (deputative beneficiary)
(3)
Twi – Christaller (1933:566) Owu kyεε me. 3sg.die share 1sg ‘He died for me, for my benefit.’ (plain beneficiary)
2.3â•…Benefactive applicative periphrases and other types of benefactive periphrases The definition formulated in Section 2.2 does not apply to all periphrastic constructions expressing benefactive/malefactive meanings, but only to those in which the verb-operator can be described as adding a beneficiary to the argument structure of the lexical verb in a way comparable to what can be observed with morphological applicatives. For example, the periphrastic passive of Vietnamese, illustrated by Ex. (4), is outside the scope of this study, since it cannot be described as involving a valency change of the applicative type. Rather, a specification is added to the semantic role of the P argument of the lexical verb by means of a control construction in€which the subject of được ‘get’ or bị ‘undergo’ controls the missing object of the lexical verb. (4) Vietnamese – Dauphin (1977:46) a. Học sinh được thầy giáo khen. pupil get teacher praise ‘The pupil was [positively affected by being] praised by the teacher.’ b. Học sinh bị thầy giáo đánh. pupil undergo teacher beat ‘The pupil was [negatively affected by being] beaten by the teacher.’
Similarly, Santali (Munda) has a periphrasis in which jfm- ‘eat’ expresses a malefactive€ meaning, illustrated by Ex. (5). However, this construction does not license a
 Denis Creissels
valency-external participant affected to his/her detriment, and consequently does not fall under the definition formulated above. It is rather a passive construction implying that the patient of a transitive verb (promoted to subject role) is negatively affected by the action of the agent (represented by a dative-marked NP). (5)
Santali – Neukom (2001:17) Uni-t» hεn 6d ˛ i ruhεt’-i\ jfm-akat’a. that(an)-dat much scold-a1sg eat-pfv.act-ind ‘I got scolded badly by him.’ (lit. ‘I ate much scolding from him’)
A similar phenomenon is found in Hindi, which has pairs of light verb constructions such as dhokhā denā ‘give cheating’ > ‘deceive’/dhokhā khānā ‘eat cheating’ > ‘get deceived’, in which the light verb ‘eat’ expresses passive diathesis. Further illustrations of this use of ‘give’ in Hindi are for example mār khānā ‘eat blow’ > ‘be beaten’, gālī khānā ‘eat insult’ > ‘be insulted’, etc. – Montaut (2004:91 & p.c).
2.4â•… Benefactive and malefactive Rather than a specifically benefactive meaning, BAPs may express a more general meaning of affectedness lending itself to malefactive interpretations, depending on the context. Ex. (6) illustrates the malefactive reading of the Yoruba ‘give’ periphrasis, other possible readings of which have been illustrated by Ex. (1) & (2) above. (6)
Yoruba – Abraham (1962:316) Ó purọʹ fún mi. 3sg lie give 1sg ‘He lied to me.’
However, the malefactive use of BAPs seems to be less common than the malefactive use of benefactive constructions involving applicative derivation or adpositions, which suggests that the extension of the use of BAPs to the expression of malefaction tends to occur at a relatively advanced stage of the grammaticalization process. An important observation concerning malefactives is that I have found no mention of applicative periphrases involving verb-operators expressing malefactive meanings only, either in descriptive grammars or in the literature on benefactives and malefactives. The grammaticalized malefactive serial verb constructions illustrated by Ex. (4) & (5) above and analyzed among others by Radetzky & Yamashita Smith (this volume) are not applicative constructions, and consequently fall beyond the scope of this paper. Similarly, a superficial look at the Baule construction illustrated by Ex. (7) could suggest analyzing it as an applicative periphrasis specifically expressing malefaction. However, the use of this construction is limited to expressing the exclusion of a participant, which is not the same thing as malefaction: the excluded participant is negatively affected by the fact of being excluded, not by the event from which (s)he is excluded.
(7)
Benefactive applicative periphrases
Baule (pers. doc.) Bè-dí bè-kpὲ mín. a3pl-eat a3pl-cut 1sg ‘They eat without giving anything to me.’
The absence of dedicated applicative-malefactive periphrases is consistent with the fact that, more generally, dedicated malefactive markers are much less common in the languages of the world than dedicated benefactive markers, and at least some of the cases of benefactive vs. malefactive marking that have been reported seem to boil down to a contextual interpretation of a basically centripetal vs. centrifugal contrast, based on a tendency to associate centripetal with benefactive, and centrifugal with malefactive – see e.g. Salas (2006:121–4) and Zúñiga (this volume) on Mapudungun.
2.5â•… Benefactive and autobenefactive In principle, any type of construction licensing a beneficiary NP can express autobenefaction via reflexivization. However, some of the languages that have a BAP also have a distinct periphrasis expressing autobenefaction. This question is developed in Section 7.
2.6â•… General characteristics of benefactive applicative periphrases BAPs tend to share the following two characteristics: a. The verb in valency operator function, when used independently, almost always expresses the general meaning ‘give’, or denotes a particular type of giving, e.g. ‘share’, as in Ex. (3) above (the use of verbs other than ‘give’ as valency operators in BAPs is examined in Section 8). b. Irrespective of the status of the language in question with respect to constituent order typology, ‘give’ almost always occupies the second position in BAPs.3 Mandarin Chinese constitutes the best-known exception to the generalization according to which ‘give’ occurs in second position in BAPs – Ex. (8).4 (8)
Mandarin Chinese – Li & Thompson (1981:388) Woˇ gěi tā jì le yi fēng xìn. 1sg give 3sg mail pfv one clf letter ‘I mailed a letter for him/her.’
3.â•… Note that this is not a general property of applicative periphrases. For example, ‘take’ acting as a valency operator in instrumental periphrases normally occurs in V1 position. There is an obvious connection between the linear ordering of applicative periphrases and the chronological succession of the phases of complex events. 4.â•… Many speakers of Mandarin, especially towards the south, allow however BAPs with ‘give’ in second position (Masayoshi Shibatani, p.c.).
 Denis Creissels
The other cases of BAPs with ‘give’ in first position I am aware of are Yongning Na (Tibeto-Burman) – Lidz (2006), Abui (Papuan) – Kratochvíl (2007:394–6), and Ecuadorian Highland Spanish – Haboud (1994 & 1998:215–223). We will return to the case of Mandarin, Yongning Na and Abui in Sections 6.3.2–3, since the exceptional order of the BAP in those languages can be viewed as a piece of evidence that the construction is only superficially identifiable as a BAP, and that ‘give’ has been reanalyzed as an adposition. But this analysis is not possible in the case of Ecuadorian Highland Â�Spanish, which will be discussed in Section 4.3.3.
2.7â•… Identifying benefactive applicative periphrases 2.7.1â•… Distinguishing benefactive applicative periphrases from biclausal constructions BAPs may be similar or even identical to biclausal constructions expressing that a giving event encoded by the second clause follows the event encoded by the first clause. Intonational contours can be expected to provide a clue to the distinction between monoclausal and biclausal expressions, but unfortunately, the documentation available on most languages does not allow using intonation as a criterion in a typological study relying mainly on second-hand data. However, in addition to the language-specific formal manifestations of a distinction between a biclausal construction and a complex predicate, the distinction is generally made obvious by the contrast between the role assigning properties of ‘give’ as a monoverbal predicate and its function in BAPs. From this point of view, the only ambiguous case is constituted by sentences describing situations involving recipient-like beneficiaries (such as I cooked a cake for the children/I cooked a cake and gave it to the children), which precisely can be viewed as providing the crucial context for the development of the reanalysis of ‘give’ as an applicative operator with a benefactive function. Deputative beneficiaries or plain beneficiaries cannot be analyzed as receiving their semantic role from ‘give’ in a biclausal construction, and role assignment in such cases results from the interaction of the meaning of the lexical verb and a general benefactive meaning (or an even more abstract meaning) contributed by ‘give’ in valency operator function. Most sources do not comment the radical change in the role assigning properties of ‘give’ involved in a BAP, but some authors insist on the specificity of ‘give’ verbs used as valency operators. For example, in Kokota (Papuan), Palmer (1999:176) describes benefactive/malefactive constructions involving what he calls “the affective verb tufa”, and uses for this verb the special gloss affect, in spite of the fact that tufa “is normally interpreted in isolation with a meaning similar to ‘give’â•›”. 2.7.2â•…Distinguishing benefactive applicative periphrases from adpositional constructions Applicative periphrases are easy to distinguish from adpositional constructions insofar as the verb in valency operator function shows verbal inflection. But if it occurs in
Benefactive applicative periphrases 
an invariable form, an alternative analysis is that, synchronically, the word licensing the expression of a beneficiary is not a verb form, but rather an adposition homonymous with a verb. Unfortunately, it is very difficult if not impossible to consistently solve this question on the basis of universally valid criteria, and no consistency must be expected in the way different descriptive traditions deal with it. Some linguists tend to consider that invariability in a given construction excludes analyzing a word as a verb in the construction in question, even if the same form is found in contexts in which it shows the inflectional variations that characterize verbs. Others tend to consider that, insofar as a word occurring in a construction in which it could be analyzed as an adposition coincides with a semantically related form encountered in other contexts with a clearly verbal status, this word must be uniformly analyzed as a verb, and the construction in which it could be analyzed as a preposition is rather a complex predicate. Others try to find syntactic evidence supporting one of the two competing analyses. On the initial stage of a typological study of BAPs, the only way to ensure the homogeneity of the data examined is to operate with a broad definition, and to leave for further discussion the question of the possible decategorialization of verb forms in valency operator function. Consequently, the notion of BAP must be taken here in a broad sense including any applicative construction in which a benefactive NP is licensed by a word that also occurs with a related meaning in constructions in which it clearly has the status of verb. Some possible criteria for analyzing the categorial status of uninflected verb forms in valency operator function are presented in Section 6.3.
2.8â•…Benefactive applicative periphrases and other grammaticalized uses of ‘give’ verbs ‘Give’ verbs are among the verbs most commonly involved in grammaticalization processes. Their possible functions in applicative periphrases are not limited to licensing beneficiaries. For example, according to Hagemeijer (2000), in SãoTomense (Creole; São Tomé), da ‘give’ in applicative operator function can assign the following semantic roles: benefactive (9a), goal (9b), experiencer (9c),5 recipient (9d), and even source (9e). (9) São-Tomense – Hagemeijer (2000:32/105/106) a. Sela pa n toma zawa pa n ba pya da bo. must for 1sg take urine for 1sg go see give 2sg ‘I must take your urine in order to check it for you.’
5.â•… The categorization of the semantic role exemplified in this sentence as experiencer is somewhat questionable, but Hagemeier (2000) provides no clearer example of ‘give’ licensing an Â�experiencer. I also have maintained the characterization of Ex. (9e) given by Â�Hagemeier, although it might be argued that (9e) rather instantiates cause.
 Denis Creissels
b.
Ola hour so then
ku-e ka bili zanela, that-2sg ipfv open window n ga zuga vunvu se da glentu ke. 1sg ipfv throw bee dem give inside house
‘When he opens the window, I will throw the bees inside the house.’
c.
Fogon ka sa kentchi da non. kitchen ipfv be hot give 1pl ‘The kitchen is getting hot for us.’
d. Fatu se ku men bo mole fika da bo. costume dem that mother 2sg die stay give 2sg ‘that costume that remained for you when your mother died.’ e.
Inen mina se tava ka kole da koblo. 3pl child dem pst ipfv run give snake ‘Those children were running away from the snake.’
In Yoruba, the possible semantic roles licensed by fún ‘give’ in applicative operator function include not only beneficiary, recipient, experiencer, and purpose, but also reason – Ex. (10a); fún can even be used to introduce temporal adjuncts – Ex. (10b). (10) Yoruba – Abraham (1962:226) a. Ó n´ kú lọ fún ebi. 3sg prog die go6 give hunger ‘He is dying of hunger.’ b. Mo sinmi fún wákàtíī kọn. 3sg rest give hour one ‘I rested for an hour.’
Grammaticalized uses of ‘give’ verbs as role assigners in the construction of other verbs are not always limited to licensing more or less peripheral participants, and may extend to argument marking: in Hakka (Sinitic), the use of ‘give’ to introduce recipient NPs has extended to ‘give’ itself, giving rise to the construction illustrated by Ex. (11), currently analyzed as involving a first occurrence of ‘give’ in verbal predicate function and a second occurrence of ‘give’ in dative preposition function. (11)
Hakka – Lai H.-L. (2001:141) Gia ba bun yi kiu tien bun gi. 3sg.gen father give one clf field give 3sg ‘His father gave a piece of field to him.’
Moreover, ‘give’ verbs quite commonly grammaticalize as valency operators, not only in applicative periphrases, but also in passive periphrases, and in causative or permissive 6.â•… In this construction, lọ ‘go away’ expresses continuative aspect.
Benefactive applicative periphrases 
periphrases. As discussed among others by Heine & Kuteva (2002:149–155; 321), they may also grammaticalize as complementizers or TAM markers, and they are often involved in polygrammaticalization processes. For example, in Phnong (Bahnaric, Mon-Khmer), ôan ‘give’ occurs in V2 position in dative/benefactive constructions, but also in V1 position in causative/permissive constructions, and in medial position in triverbal constructions expressing purpose – Ex. (12).7 (12) Phnong – Vogel (2006:108/133/166) a. Gº fp ntum ŋ8j dº ro˜n ôan paŋ. 1sg teach speak Khmer give 3sg ‘I teach him Khmer.’ b. Gº fp ôan paŋ kh8t. 1sg give 3sg die ‘I let him die.’ c.
Cho˜ŋ piaŋ ôaŋ lεô do˜. eat rice give finish imp ‘Eat the rice so that none of it is left.’
This aspect of the question will not be addressed further in this paper, since the main focus is on BAPs.
3.â•… Formal types of benefactive applicative periphrases Three formal types of BAPs can be recognized with respect to the grammatical nature of the two verb forms involved in the periphrasis. Note that, in the definition of formal types of BAPs, marked applied to a verb form must be understood as an abbreviation for ‘showing morphological evidence of a dependent status’.
3.1╅ The serializing type I adopt the definition according to which a serial verb construction (henceforth abbreviated as SVC) is a complex predicate (i.e. a monoclausal construction involving two or more verbs) showing the following two characteristics: a. no linking element is present between the verbs involved in the construction; b. none of the verbs involved in the construction is in a form implying a �nonautonomous status.
7.â•… Mon k# ‘give’, Thai hâj ‘give’ and Burmese pèi ‘give’ show a similar range of grammaticalized functions – see Jenny (2005:214–5). On the relationship between benefactive and causative constructions involving the same ‘give’ verb, see in particular Iwasaki & Yap (2000).
 Denis Creissels
This definition basically coincides with that formulated by Aikhenvald (2003:1). It crucially differs from earlier definitions in that it explicitly excludes covert clause or VP coordination from SVCs. Covert coordination may be the historical source of SVCs, and in some languages the limit between covert coordination and SVCs may be fuzzy, but a notion of SVC that did not include this restriction would be too vague to be useful in the analysis of syntactic structures. For recent discussions of the relationship between SVCs and covert coordination, see among others Ameka (2003), Newmeyer (2004), Paul (2004). Although there is now a relative consensus on this definition, it must be mentioned at this point that the current conception of SVCs has been vigorously criticized by Shibatani (2006). He argues that this conception relies on a confusion between verb forms as pure forms and verb forms as members of a paradigm occupying a given slot in a given construction, and that SVCs as currently identified share so many properties with other formal types of complex predicates that it is not justified to view them as a cross-linguistically valid type of construction. Concerning the distinction between complex predicates of the serial type and the combination of a monoverbal predicate with an adposition, the uncontroversial cases of SVCs are those in which each of the verbs involved in the construction shows at least some of the inflectional variations characteristic of verb forms heading independent clauses. In such cases, it may happen that the verbs involved in a SVC show parallel inflection (if V1 and V2 show identical inflectional marks) or distributed inflection (if some of the inflectional marks characteristic of verbs heading independent clauses attach to V1, and some others, to V2). Ex. (13) from Baule (Kwa) and (14) from Dagaare (Gur) illustrate benefactive SVCs with parallel inflection of the lexical verb and of ‘give’ in valency operator function. (13)
Baule (pers. doc.) Ákísí à-t`fn duô à-màn Kòfí. Akissi prf-cook yam prf-give Kofi ‘Akissi has cooked yam for Kofi.’
(14)
Dagaare – Bodomo & van Oostendorp (1994:23–4) Bayff zo-ro gε-rε wuo-ro la haani Bayor run-ipfv go-ipfv collect-ipfv decl blackberry waa-na k~- rf ma. come-ipfv give-ipfv 1sg
‘Bayor is presently going and collecting some blackberries for me.’
As already commented in Section 2.3.2, the analysis of SVCs involving a fully inflected lexical verb and a verb-operator invariably occurring in bare stem form is much less obvious. Ex. (10) from Yoruba, reproduced here as (15), and Ex. (16) from Kana (Cross-River), illustrate this type of benefactive SVC.
Benefactive applicative periphrases
(15)
Yoruba – Abraham (1962:226) Ó ń kú lọ fún ebi. 3sg prog die go give hunger ‘He is dying of hunger.’
(16)
Kana – Ikoro (1996:254) žwíìkā wēè fʹ b túú nὲ Nūtὲ. Nwiika pst roast three-leave_yam give Nute ‘Nwiika roasted a three-leave yam for Nute.’
3.2â•… The marked-Vop type Ex. (17) from Mankon (Grassfields Bantu) illustrates a type of BAP in which Vlex is inflected like verbs heading monoverbal independent clauses, whereas ‘give’ in valency operator function is in a non-autonomous “sequential” form typically used for verbs heading non-initial clauses in clause chains encoding sequences of events. (17)
Mankon – Leroy (2003:459) Mà m≠і‒ʹ fàôá γ≠á mbó zш´ә´ 1sg fut work seq.give to 3sg.enunc ‘I will work for him.’ (lit. something like ‘I will work and-give him’)
Welmers (1973:366–72) describes benefactive ‘give’ periphrases and other complex predicates found in some Benue-Congo languages of South East Nigeria (e.g. Efik, Igbo), in which the first verb is fully inflected as an independent verb form, and the TAM value it expresses determines the form of the second verb. In some tenses, the verb in V2 position is in the dependent form otherwise used in clause chains encoding sequences of events, whereas in other tenses, the construction looks like a SVC. (18) Efik – Welmers (1973:369–70) a. Nám útóm έmì n`f mî! do work dem give 1sg ‘Do this work for me!’ b. Ánàm útóm fʹ n`f mî. 3sg.prs.do work 3sg.seq.give 1sg ‘He is working for me.’
3.3â•… The marked-Vlex type Beria (Saharan) – Ex. (19) – illustrates the type of BAP in which ‘give’ in valency operator function is inflected like verbs heading monoverbal independent clauses, whereas the lexical verb is in a non-autonomous form typically used, in the languages that have this type of BAP, for verbs heading non-final clauses in clause chains encoding sequences of events. A variety of terms are used to label such forms in descriptive
 Denis Creissels
grammars; in this paper they are uniformly designated as converbs, and their characteristic affixes are uniformly glossed cvb, whatever the terms used in the sources I have consulted. (19)
Beria – Jakobi & Crass (2004:171) A´ská gí-n-έ é-géí! door open-a2sg-cvb p1sg-give.imp ‘Open the door for me!’ (lit. ‘Opening the door give me!’)
In this type of BAP, ‘give’ can be characterized as auxiliary in the sense of function word inflected like a independent verb and combined with a dependent form of the verbal lexeme with which it constitutes a complex predicate. Vector verb is another term found in the literature (in particular on Indo-Aryan languages) to characterize verbs fulfilling a grammaticalized function in this type of compound predicate.
4.â•… The geographical distribution of benefactive applicative periphrases 4.1â•… The distribution of the serializing type 4.1.1â•…Benefactive applicative periphrases in language families or areas in which serialization has been recognized as a common phenomenon BAPs of the serial type are common among the serializing languages of West Africa – Ex. (1), (2), (3), (6), (10), (13), (14), & (16) above, South Eastern Asia – Ex. (20) to (22), and New Guinea – Ex. (23) & (24). (20)
Thai – Lord (2002:220) Kháw thamŋaan hây phîichaay. 3sg work give older_brother ‘He works for his brother.’
(21)
Vietnamese – Dauphin (1977:39) Đê tôi làm cho anh! let 1sg do give older brother ‘Let me do it for you!’
(22)
Yao Samsao (Tibeto-Burman) – Matisoff (1991:428) Yi6 tsiáô nfm daan pun nîn. 1sg weave clf basket give 3sg ‘I wove a basket for him.’
(23)
Kokota (Papuan) – Palmer (1999:176) Fa doli tufa-nau zuta-na! caus be alive give-p1sg lamp-dem ‘Light that lamp for me!’
Benefactive applicative periphrases 
Ð
(24) Dom (Papuan) – Tida (2006:169) ^Flawa nu na ^to-gwe. flour knead8 1exc give-a3sg.ind ‘She kneaded flour for me.’ ^
BAPs of the serial type are also found in those of the pidgin and creole languages that are known for making extensive use of SVCs – Ex. (9) above. Among the language families in which serialization is a widespread phenomenon, Oceanic is the only one in which applicative, and in particular benefactive SVCs are not common: One widely attested development in the world’s languages regarding the grammaticalization of serial verbs is the co-opting of verbs meaning ‘take’ and ‘give’ to function as prepositions expressing instrumental and benefactive meanings respectively. It is worth pointing out at the outset that in Oceanic languages, these verbs are rarely encountered in serial constructions, so prepositions with these particular verbal origins are seldom convincingly attested. Crowley (2002:173)
4.1.2â•… Others In addition to language families or areas in which they are particularly common, BAPs of the serial type are sporadically attested in language families or areas in which serialization does not constitute a widespread phenomenon. Keo (Western Austronesian; Flores, Indonesia) illustrates the case of a BAP of the serial type in a language belonging to a family within which serialization is not widespread, but located not very far from an area in which such constructions are common – Ex. (25). (25)
Keo – Louise Baird, quoted by Margetts & Austin (2007) Ja’o tendo jawa ti’i ’ine. 1sg plant corn give mum ‘I’m planting corn for mum.’
In Africa, BAPs of the serial type are sporadically found outside the area of West Africa characterized by a particular concentration of languages having very productive SVCs, for example in Ngbandi (Ubangian) – Toronzoni (1989), and in the Sara (Central Sudanic) languages Kabba – Moser (2005) – and Sar – Palayer (1989).
8.â•… Note that Tida misleadingly glosses bare verbal stems occurring in V1 position in the SVCs of Dom as ‘INF(initive)’.
 Denis Creissels
BAPs of the serial type with parallel inflection (i.e. with the same inflectional mark on the two verbs) are attested in Old Turkic sources – Anderson (2001). BAPs are very common among Turkic languages, but in modern Turkic languages, they uniformly belong to the marked-Vlex type – Section 4.3. Contrary to most Indo-Aryan languages, Hindi has a BAP in which, in the same way as in other auxiliary (or “vector verb”) constructions, the lexical verb occurs in a form consisting of the bare stem. This form is currently analyzed as a zero-marked converb (or “conjunctive participle” – Montaut (2004:93)), but whatever the justification for such an analysis, the external appearance of the Hindi ‘give’ periphrasis is that of a SVC with verbal inflection concentrated on the verb in V2 position. Among Amerindian languages, benefactive SVCs with parallel inflection are found in the Siouan languages Hidatsa and Mandan – Ex. (26). (26)
Mandan – Mixco (1997:50) Wą´wąrąhku-rą té wa-hræ-ak rútrá-sit deer-top die 1sg-cause-ds rib by heat-roast wa-hræ wa-rį-kųô-rįt-oôš. 1sg-cause 1sg-2-give-2pl-indma
‘I’ve killed a deer and roasted the ribs for you.’
4.2â•… The distribution of the marked-Vop type This type has been illustrated above by Mankon, an atypical (and geographically peripheral) Bantu language that has lost the morphological applicative attested in most Bantu languages and reconstructible at least at Proto-Bantu level, and has compensated this loss by the creation of an applicative periphrasis formally identical to a clause chain encoding the successive phases of a complex event. The only attestations of this type I am aware of come from Benue-Congo languages (Bantu and non-Bantu) spoken in West Cameroon and South-East Nigeria and sharing with Mankon both the loss of the morphological applicative and the retention of overtly inflected sequential verb forms. As already mentioned in Section 3.1 above, many of the languages spoken in this area have complex predicates (including BAPs) with morphological characteristics that make them look like SVCs in some tenses, and constructions of the marked-Vop type in some others.
4.3â•… The distribution of the marked-Vlex type This type is extremely common among the verb-final languages of Asia, from Ainu and Japanese to the East to Turkish to the West and Tamil to the South. Not surprisingly, benefactive ‘give’ compounds and derived benefactive verb forms including an applicative marker originating from a ‘give’ verb, which clearly result from the evolution of BAPs, are also very common among the languages spoken in this area. Outside this area, BAPs of the marked-Vlex type are only sporadic.
Benefactive applicative periphrases 
4.3.1â•… Asian attestations LaPolla (2003:33) observes that, among Tibeto-Burman languages, a commonly found development is the grammaticalization of a benefactive construction. This most commonly takes the form of an auxiliary verb derived from a verb meaning ‘to give’, as in Jinghpaw (-t∫a33), Tamang (pín), Tsangla (bi), Camling (bi), Belhare (-per), and Lahu (pî …). As can be seen from these examples, the verb used in this construction is often the P[roto]-S[ino]-T[ibetan] verb *biy, but the constructions themselves were independently innovated.9
As illustrated by Ex. (27) to (34), BAPs of the marked-Vlex type occur not only among Tibeto-Burman languages (illustrated below by Dolakha Newar), but also in Ainu, Japanese, Korean, and in languages belonging to the Mongolic, Turkic (illustrated below by Ojrot), Indo-Aryan, and Dravidian families. (27)
Dolakha Newar – Genetti (2007:334) Janta lukhā khoŋ-an bi-sin! 1sg.dat door open-cvb give-imp ‘Open the door for me!’
(28)
Ainu – Tamura (2000:181) Néno iki wa en-kore hani? similar do cvb 1sg-give enunc ‘Would you do that for me?’
(29)
Japanese (Miyuki Ishibashi, p.c.) Kodomo-ni kutsu-o kat-te yat-ta. child-dat shoe-acc buy-cvb give-pst ‘I have bought shoes for the child.’
(30)
Korean (Injoo Choi, p.c.) Yumi-ka Sumi-eykey mwun-ul yel-e cwu-ess-ta. Yumi-sbj Sumi-dat door-acc open-cvb give-pst-decl ‘Yumi opened the door for Sumi.’
(31)
Mongolian – Gaunt & Bayarmandakh (2004:29) Ter bidend xool xijž ögnö. 3sg 1pl.dat food make.cvb give.ipfv ‘He will make food for us.’
(32)
Ojrot – Dyrenkova (1940:191) Uulčak bis-ke d’ol ayd-i‒p ber-di. boy 1pl-dat road tell-cvb give-pfv.a3sg ‘The boy showed us the road.’
9.╅ Note that some Tibeto-Burman languages have BAPs of the serial type. This is consistent with the fact that the territory occupied by Tibeto-Burman languages overlaps with the South East Asian linguistic area, characterized by extensive use of SVCs in general, and of �benefactive SVCs in particular.
 Denis Creissels
(33) Pāli – Hendriksen (1944:134), quoted by Butt & Tantos (2004) … assamapadam» ānetvā aggim» katvā adāsi. hermitage.acc lead.cvbfire.acc make.cvb give.iprf.3sg ‘… brought her to his hermitage and made a fire for her.’ (34)
Tamil – Krishnamurti (2003:376) Rājā kumār-ukkuk katav-ai tir ‒. ‒ant-u kozutt-ān Raajaa Kumaar-dat door-acc open-cvb give.pst-3sgm ‘Raajaa opened the door for Kumaar.’
Mamatov & al. (2005) mention a ‘give’ periphrasis of this type in Tajik (Iranian), but it probably constitutes an instance of contact-induced development, since extensive Â�Turkic-Iranian bilingualism is characteristic of the area where Tajik is spoken, and I have been able to find no other attestation of this construction among Iranian languages.
4.3.2â•… African attestations Converbal constructions similar to those of the Asian languages mentioned in the preceding section are common among North East African verb-final languages – Azeb & Dimmendaal (2006), but apart from the Saharan language Beria, already mentioned in Section 3.3, Old Nubian is the only African language in which the descriptions I have been able to consult mention ‘give’ periphrases of the marked-Vlex type, and my only source for this language characterizes the construction as “dative” without mentioning the possibility of a benefactive function: den- ‘to give’ (to me/us) and tr‒- ‘to give’ (to you/him/them) are the so-called ‘dative verbs’, used to specify an indirect object: e.g. St. 3.10-11 ouka pl»¿lïgra- denjisna- ‘he revealed to us’ (lit., ‘revealing, he gave to us’) and M. 7.5-6 tan` eila` outra trsna ‘she placed it in his hand’. Browne (2002:65)
Ijo languages, which constitute an exceptional case of consistently OV languages in West Africa,10 should perhaps be considered as having a BAP of this type. According to Williamson (1965), in the BAP of Ịzọn, as in other semantic types of complex predicates, the verb in V2 position is fully inflected, whereas V1, exactly like non-final verbs in the sequential construction, alternates between the bare stem form if the following word begins with a consonant, and a form characterized by an ending -n(i․`) if the following word begins with a vowel – Ex. (35). (35)
Ịzọn – Williamson & Timitimi (1983:160) Okokodía sọk(ị)-n(ị) ị-pịrí! ˙ ˙ coconut pick out-nỊ 1sg-give.imp ‘Pick the coconut out (of its shell) for me!’
10.â•… Apart from the Ijo languages spoken in the delta of Niger, the Dogon languages spoken in the eastern part of Mali are the only group of West African languages consistently showing OV typology.
Benefactive applicative periphrases 
Consequently, the construction has the appearance of a SVC if the word following Vlex begins with a vowel and of a marked-Vlex construction if the word following Vlex begins with a consonant. Two alternative interpretations can be considered, but the available data do not make it possible to decide which one should be selected: a. -n(ị) is a converbal ending with a phonologically conditioned zero allomorph, and consequently the construction belongs to the marked-Vlex type; b. -n(ị) has the purely morphophonological function of preventing the deletion of the final vowel of the first verb (which in Ịzọn would automatically be deleted in contact with another vowel), and consequently the construction belongs to the serial type.
4.3.3â•… American attestations Haboud (1994, 1998:215–223) describes a BAP in which ‘give’ in Vop function fully inflected like a verb heading a monoverbal independent clause combines with a dependent form of the lexical verb in Ecuadorian Highland Spanish. An exceptional feature of this “dar + gerund” BAP of Ecuadorian Highland Spanish is that, in conformity with Spanish word order, dar ‘give’ in auxiliary function precedes the gerund with which it forms a BAP. Although particularly usual in the imperative, this construction is also used in other TAMs, as illustrated by Ex. (36). Note in particular the possibility to combine dar acting as a benefactive auxiliary with dar in the gerund form expressing its basic meaning ‘give’. (36) Ecuadorian Highland Spanish – Haboud (1994) a. Me dio cocinando. d1sg give.pfv.a3sg cook.ger ‘(S)he cooked for/instead of me.’ b. Él me da haciendo el pan 3sgm d1sg give.prs.a3sg make.ger def.sgm bread
mientras yo lavo. while 1sg wash.prs.a1sg
‘He bakes the bread for/instead of me while I wash.’
c.
Él me dio dando el cuchillo 3sgm d1sg give.pfv.a3sg give.ger def.sgm knife
a la to def.sgf
‘He gave the knife to María instead of me.’
María. María
The first explanation that comes to mind is that this construction might be the result of a transfer from Ecuadorian Quechua, since Quechua has clause chains of the same type as the Asian languages that have BAPs of the marked-Vlex type. However, this construction does not seem to be attested in other varieties of Quechua, and Haboud observes that, in Ecuadorian Quechua, it occurs in direct elicitation with bilingual
 Denis Creissels
Quechua-Spanish speakers, but not in spontaneous productions of speakers having a lower command of Spanish. She tries to explain it as resulting from the transfer of an imperative honorific suffix found in the local variety of Quechua and traceable back to the Quechua applicative suffix -pu-, but this explanation is not very convincing, since€the transfer of syntactic structures typically involves a word-to-word rather than affix-to-word relation. Although the emergence of this construction must certainly be viewed as an instance of contact-induced language change, it is reasonable to assume that it involves a historical scenario more complex than the mere transfer of a Quechua construction. The point is that Ecuadorian Quechua is the result of a relatively recent expansion, since the Inca had conquered the region just one century before the arrival of the Spaniards. At the time of the Spanish conquest, many Indian groups in the Ecuadorian Highlands were still speaking their own languages. Linguistically, the Spanish conquest was followed by an extensive process of Quechuanization, which means that the Ecuadorian varieties of both Quechua and Spanish have evolved in a context favoring the development of complex pidginization/creolization processes. The explanation of the particularities of Ecuadorian Highland Spanish may therefore lie in a complex contact situation involving not only Spanish and Quechua, but also languages that are now extinct, so that the explanation of linguistic particularities of Ecuadorian Highland Spanish that cannot be the result of a straightforward transfer from Quechua is condemned to remain purely speculative.
4.4â•… Correlations When I began this investigation, my idea was that there should be a correlation between the type of BAP that can be found in a language and the type of clause chain used in the same language to encode complex events conceived as a sequence of elementary events, since diachronically, BAPs probably result from the reanalysis of constructions that, originally, express the meaning ‘create/manipulate something and give it to someone’. However, the data just presented suggest a weaker claim. They confirm that, in general, marked-Vlex BAPs are found in languages or language families having sequential constructions in which non-final verbs are in a form overtly marked as non-autonomous, and marked-Vop BAPs are found in languages or language families having sequential constructions in which non-initial verbs are in a form overtly marked as non-autonomous. By contrast, BAPs of the serial type are not limited to languages using covert coordination in clause chains. In particular, Papuan languages typically use clause chains in which the final verb is the only one occurring in an independent form, but many of them also make extensive use of SVCs, including benefactive ones.
Benefactive applicative periphrases 
A possible explanation of this imperfect correlation is that the grammaticalization process converting biclausal constructions into complex predicates may involve �reduction of morphological marking. Another possibility is that multiverbal constructions grammaticalized as complex predicates do not necessarily follow changes occurring in the make-up of multiclausal constructions.11
5.â•… Grammaticalization Similarly to other applicative periphrases, BAPs constitute an intermediary stage in grammaticalization chains, either from verb to adposition-like or case-marker-like items, or from verbs to applicative verb affixes. The initial impetus for the development of applicative periphrases in general, and BAPs in particular, is probably a tendency to limit monoverbal constructions to the expression of arguments, and to make the role of more or less peripheral or optional participants more explicit by using multiverbal constructions. For example, SãoÂ�Tomense uses a double object construction in order to mention the source/maleficiary argument of futa ‘steal’, whereas an SVC is required to express a beneficiary with the same verb – Ex. (37).12 (37) São-Tomense – Hagemeijer (2000:106) a. Ladlon futa mu djelu. thief steal 1sg money ‘The thief stole money from me.’ b. Ladlon futa djelu da mu. thief steal money give 1sg ‘The thief stole money for me.’
5.1â•… From sequential construction to benefactive applicative periphrasis 5.1.1â•… The reanalysis of ‘give’ as a benefactive operator It seems reasonable to assume that the starting point of the grammaticalization chains involving applicative periphrases expressing benefaction is typically a sequential construction in which the second clause describes a giving event constituting the second phase of a complex event whose first phase is described by the first clause, as in English She sewed a dress and gave it to her daughter. 11.â•… On this and next section, see Shibatani & Chung (2007). 12.â•… In some languages, the tendency to avoid sentences with more than two NPs in the construction of each verb affects the expression of arguments too, as observed by Censabella (this volume) for Toba (Guaycuruan).
 Denis Creissels
In such a construction, the recipient of give can also be viewed as the beneficiary of the first event: →
She sewed a dress and gave (it) to her daughter She sewed a dress for the benefit of her daughter
An applicative periphrasis in which ‘give’ acts as a modifying element of the first verb can therefore emerge as the result of the conventionalization of this implicature. Once the reanalysis has been completed, ‘give’ is no longer interpreted as encoding a giving event involving a giver and a gift already involved in an event preceding the giving event, and the NP that originally represented the recipient in a subsequent giving event is interpreted as representing the beneficiary of the first event, without any hint at the precise reason why this participant can be regarded as a beneficiary. It is reasonable to assume that the reanalysis of sequential constructions involving a ‘give’ verb as BAPs develops first with recipient-like beneficiaries (e.g. ‘buy something for someone’) before being extended to the expression of other subtypes of beneficiaries.
5.1.2â•… Evidence of the reanalysis of ‘give’ as a valency operator The possibility to have constructions that are lit. X opened the door gave Y or X having opened the door gave Y, in which it is excluded to interpret Y as being assigned the role of recipient in the same way as in X gave the door to Y, can be used as a diagnostic of the reanalysis of the sequential construction as an applicative periphrasis expressing benefactive. Additional evidence may be provided by the malefactive interpretation of ‘give’ in constructions that are lit. X ate Y gave Z or X having eaten Y gave Z interpreted as ‘X ate Y to the detriment of Z’: here again, the interpretation of Z as a recipient is excluded. Constituent order may also provide evidence of the reanalysis of a sequential construction involving two clauses as a monoclausal construction. For example, in Japanese, when the converb formed with -te occurs in a clause chain, noun phrases belonging to the clause headed by the final verb are inserted between the converb and the final verb, as in Ex. (38a), whereas in a complex predicate in which the final verb is ‘give’ in valency operator function, such phrases cannot be inserted between the converb and the final verb – Ex. (38b). (38) Japanese (Miyuki Ishibashi, p.c.) a. Machi-e it-te, eiga-o mi-ta. town-all go-cvb movie-acc see-pst ‘I went to town and saw a movie.’ b. Yamada-san-wa Tanaka-san-ni tegami-o kai-te yat-ta. Yamada-Mr.-top Tanaka-Mr.-dat letter-acc write-cvb give-pst ‘Mr. Yamada wrote a letter to/for Mr. Tanaka.’
Benefactive applicative periphrases 
In Igbo, the fact that nyé ‘give’ in Vlex role can combine with nyé ‘give’ in Vop role, as in Ex. (39b), provides evidence of the grammaticalization of ‘give’ as a valency operator. (39) Igbo – Onumajuru (1985:173) a. ọʹ-zù.â•›-rù.â•› ánù.â•› nyé Áányí․. a3sg-buy-pfv meat give 1pl ‘S/he bought meat for us.’ b. ó-nyè-rè-m¡ jí Ányé Áù.â•›bá. a3sg-give-pfv-p1sg yam give Uba ‘S/he gave me yams for Uba.’
As in other cases of grammaticalization, ambiguities may however subsist (at least if intonation is not taken into consideration) between BAPs and superficially identical constructions still interpretable as sequential constructions – Ex. (40). (40)
Dolakha Newar – Genetti (2007:336) Lũ=e bo thi-pta hā-en bi-u! gold=gen plate one-clf bring-cvb give-imp a. ‘Bring a plate of gold and give it (to him)!’ b. ‘Bring a plate of gold for him!’
Conversely, BAPs may be bound by semantic restrictions due to the retention of elements of their original signification. For example, as discussed by Shibatani (2003:282–3), in spite of being clearly distinct from a sequential construction, the BAP of Japanese is restricted to situations involving “the transfer of possessive control of a certain entity, whether concrete or abstract”. In Japanese, intransitive verbs, or transitive verbs whose object NP denotes an object that is not normally transferred to a beneficiary cannot occur in a BAP including an overtly expressed beneficiary – Ex. (41a–b); note however that the same verbs can occur in the BAP provided the Â�beneficiary is not overtly expressed – Ex. (41c–d). (41) Japanese – Shibatani (2003:283) a. *Ken-wa hahaoya-ni itiba-e it-te yat-ta. Ken-top mother-dat market-all go-cvb give-pst Intended: ‘Ken went to the market for mother.’ b. *Ken-wa watashi-ni gomi-o sute-te kure-ta. Ken-top 1sg-dat garbage-acc throw away-cvb give-pst Intended: ‘Ken threw away the garbage for me.’ c.
Ken-wa itiba-e it-te yat-ta. Ken-top market-all go-cvb give-pst ‘Ken did (someone) the favor of going to the market.’
 Denis Creissels
d. Ken-wa gomi-o sute-te kure-ta. Ken-top garbage-acc throw away-cvb give-pst ‘Ken threw away the garbage (for me).’13
5.2â•… The grammaticalization of ‘give’ as a benefactive auxiliary In BAPs of the marked-Vlex type, and in some of the BAPs of the serial type, ‘give’ in benefactive operator function bears full verbal inflection. A ‘give’ verb occurring in such constructions may lose the ability to be used in monoverbal constructions with the meaning ‘give’, or take in this construction a form different from the form it has when expressing the meaning ‘give’, giving thus raise to a benefactive auxiliary whose relation to a ‘give’ verb can be recognized in a diachronic perspective only. This evolution seems to be common among Tibeto-Burman languages.
5.3â•… The grammaticalization of ‘give’ as a benefactive adposition/case marker ‘Give’ verbs are widely recognized as a possible source of benefactive adpositions – see among others Heine & Kuteva (2002:149–151). According to Carlson (1991:214), in Tagbana and Jimini (Senufo, Gur), a benefactive marker is developing from the verb kan or kã ‘give’ in constructions in which “it is unclear from the sources whether kã is a serial verb or a postposition”, whereas Karaboro (another Senufo language) has the cognate benefactive postposition k6~ but has lost the corresponding form for ‘give’, replaced by another verb wãr. In the same geographic area, Dzùùngoo (Mande) has a benefactive postposition k`f whose probable origin is a SVC with the verb k`f´ ‘give’ in V2 position. However, in the SVCs of Dzùùngoo, verbal inflection is borne by the verb in V2 position, whereas benefactive k`f is invariable, which shows that it must not be analyzed as the second term of as SVC, but rather as a postposition etymologically related to the verb k`f´ – Solomiac (2007). Like other function words, ‘give’ verbs reanalyzed as benefactive adpositions may become bound forms, and consequently evolve towards the status of benefactive case markers. For example, according to Remijsen (to appear), in Magey Matbat (Austronesian), “In propositions involving a verb of transfer, the recipient semantic role can be expressed by a prepositional phrase … The preposition in question, be, is segmentally
13.â•… Japanese has several ‘give’ verbs, depending on honorificity level and orientation – Kuno (1973). Yaru and kureru are both used in neutral speech register but differ in that kureru implies orientation towards the speaker’s deictic center – Shibatani (2003:279–81); consequently, the beneficiary in this sentence is most likely to be the speaker. Cross-linguistically, ‘give’ verbs including a deictic component in their lexical meaning (either 1 vs. 2/3 or 1/2 vs. 3) are not uncommon – see among others Browne (2002) on Nubian, Asher & Kumari (1997) on Malayalam, Emeneau (1984) on Toda.
Benefactive applicative periphrases 
identical to the verb be21 ‘give’. Unlike the verb, though, prepositional be is prosodically weak, cliticising to its argument.” The problem is to determine at which stage of its evolution a ‘give’ verb engaged in such a grammaticalization path can be recognized as having been converted into an adposition. The analysis of verb forms devoid of overt inflection marks acting as valency operators is particularly problematic in languages in which regular verb inflection includes forms coinciding with the bare verb stem.
5.3.1â•… Evidence from extraction Evidence of an ongoing process of grammaticalization from verb to preposition in a BAP has been discussed by Voorhoeve (1975) for Sranan (Creole). He argues that the existence of two alternative cleft constructions shows that for the Sranan speakers who front the phrase gi NP, as in (42c), gi ‘give’ in the benefactive construction is a preposition, whereas for those who front only the NP following gi, as in (42b), it is a verb. (42) Sranan – Vorhoeve (1975), quoted by Lord (1993:42–3) a. Mi wroko gi en. 1sg work give 3sg ‘I worked for him.’ b. Na [en] mi wroko gi. foc 3sg 1sg work give ‘It’s him I worked for.’ c.
Na [gi en] mi wroko. foc give 3sg 1sg work. ‘It’s for him I worked’
In São-Tomense, the extraction of the complement of a preposition triggers the use of a resumptive pronoun, as in (43a), whereas resumptive pronouns do not occur when the complement of a verb is extracted. According to Hagemeier (2000), the fact that ‘give’ in the BAP behaves like a verb from this point of view, as shown by Ex. (43), provides evidence that it has not fully grammaticalized as an adposition yet. (43) São-Tomense – Hagemeijer (2000:108) a. Ke kwa ku piskado bili vwado ku-e? / which thing comp fisherman open flying fish with-3sg ‘With what did the fisherman open the flying fish?’
*… ku?
b. Ke nge ku Zon tlaba da? / *… d’e? which person comp Zon work give ‘For whom did Zon work?’
5.3.2â•… Evidence from constituent order: The case of Mandarin Chinese Mandarin Chinese has been mentioned as having a BAP in which ‘give’ precedes the lexical verb – Ex. (8), repeated here as (44a). (44b) shows that with some lexical verbs
 Denis Creissels
at least, ‘give’ in valency operator function can occur in second position, but with a different meaning. (44) Mandarin Chinese – Li & Thompson (1981:388) a. Woˇ geˇi tā jì le yi fēng xìn. 1sg give 3sg mail pfv one clf letter ‘I mailed a letter for him/her.’ b. Woˇ jì le yi fēng xìn geˇi tā. 1sg mail pfv one clf letter give 3sg ‘I mailed a letter to him/her.’
It is interesting to observe that ‘give’ in second position assigns the recipient role to the NP that follows it, whereas ‘give’ in first position assigns the beneficiary role. Gei belongs to a class of items commonly termed “coverbs”, which according to Mandarin Chinese grammars function as prepositions but show more or less evidence of a verbal origin. Since phrases headed by coverbs most often precede the verb,14 the position of benefactive gei can be viewed as resulting from alignment with a class of prepositionlike items, contrasting with the retention of the order of the original sequential construction in the case of dative gei.
5.3.3â•… Evidence from constituent order: The case of Abui and Yongning Na Abui (Papuan) is another case in point. Kratochvíl (2007:394–96) states that SVCs with l ‘give’ in first position have the expression of a benefactive or malefactive participant as one of their two possible functions (their other function being the expression of a topical undergoer participant). However, none of the examples provided straightforwardly involves a benefactive meaning, and this construction seems to be best described as having the more abstract meaning of expression of a participant towards whom the action is directed (‘bite someone’, ‘hit someone’ ‘look for someone’, etc.). It seems therefore that, synchronically, this Abui construction is not a BAP properly speaking, in spite of the fact that it includes a role assigner etymologically related to a ‘give’ verb. Whatever the exact function of this construction, since Abui is a verb-final language, the a priori exceptional position of l in valency operator function can be viewed as alignment with the position normally occupied by postpositions, providing thus evidence of reanalysis. The same analysis applies to Yongning Na (Tibeto-Burman), in which ki33 ‘give’ has grammaticalized as a dative-benefactive-allative postposition: Yongnin Na is a verb-final language, and postposition phrases headed by ki33 used as a postposition precede the verb in the same way as other postposition phrases – Lidz (2006). 14.â•… According to Dryer (2003:48–9), this is highly unusual for preposition phrases in a language having SVO order as dominant order at the clause level.
Benefactive applicative periphrases 
5.4â•…From benefactive applicative periphrases to benefactive verbal compounds, and from benefactive verbal compounds to derived benefactive verb forms In BAPs, ‘give’ in valency operator function may be obligatorily contiguous to the lexical verb. This is particularly common in BAPs of the marked-Vlex type. When the two verbs constituting a BAP are obligatorily contiguous, there may be morphological and/or phonological evidence that ‘give’ in valency operator function nevertheless constitutes a separate word. For example, in Japanese, the fact that Vlex shows the same suffix as forms productively used to mark non-final verbs in clause chains (Ex. (37) above) provides evidence against analyzing the BAP as involving compounding, in spite of the impossibility to insert NPs between Vlex and Vop. However, words obligatorily contiguous to each other tend to coalesce into a single word, which may result in constructions in which the relation between the two verbs is best treated in terms of compounding. Subsequent evolutions (in particular, phonological modifications of ‘give’ in second position in a benefactive compound, or the loss of the possibility to use a former ‘give’ verb in constructions other than benefactive compounds) may result in the reanalysis of ‘give’ as an applicative affix.
5.4.1â•…Benefactive ‘give’ compounds and applicative affixes cognate with ‘give’ verbs in Asian languages Korean has BAPs in which a fully inflected form of cwuta ‘give (plain)’ or tulita ‘give (humble)’ is immediately preceded by a converbal form characterized by the suffix -a/e. However, in most sources, no separation is marked in writing between the converb and cwuta, which suggests that this sequence tends to be reinterpreted as a compound. Note however that particles can be inserted, which shows that the construction cannot be straightforwardly described in terms of compounding (Masayoshi Shibatani, p.c.). (45) Korean (Injoo Choi, p.c.) a. Yumi-ka Sumi-eykey chayk-ul cwu-ess-ta. Yumi-sbj Sumi-dat book-acc give-pst-decl ‘Yumi gave Sumi a book.’ b. Yumi-ka Sumi-eykey mwun-ul yel-e Yumi-sbj Sumi-dat door-acc open-cvb
cwu-ess-ta. ~… yel-e-cwu-ess-ta. give-pst-decl
‘Yumi opened the door for Sumi.’
The hypothesis that the Korean BAP is engaged in a process of further grammaticalization is consistent with the fact that the Korean converb formed with -a/e does not seem to be used productively in clause chaining, and seems to be used mainly, if not exclusively, within complex predicates. It is however interesting to observe that, in
 Denis Creissels
spite of being perhaps formally more grammaticalized than its Japanese equivalent, the Korean BAP shows similar restrictions due to the retention of its original meaning, as illustrated by Ex. (46). (46) Korean (Song, this volume) a. *Kiho-ka yenghi-eykey sicang-ey ka-cwu-ess-ta. Keeho-sbj Yonghee-dat market-all go-give-pst-decl Intended: ‘Keeho went to the market for Yonghee.’ b. *Kiho-ka yenghi-eykey mwun-ul tat-a-cwu-ess-ta. Keeho-sbj Yonghee-dat door-acc close-cvb-give-pst-decl Intended: ‘Keeho closed the door for Yonghee.’
According to Tsumagari (2003), Dagur (Mongolic) has a “benefactive mood” imperfective converb + ukw- ‘give’, and this construction also has the synthetic (suffixalized) variant -j-ukw-. In Xakas (Turkic), Anderson (2001) describes a phonologically conditioned zero realization of the converbal suffix in the BAP and other “auxiliary verb constructions”, which can be viewed as evidence of evolution towards compounding. According to Peterson (2007), in Hakha Lai (Tibeto-Burman), the benefactive/ malefactive applicative suffix -piak, seen in (47a), closely resembles the ‘give’ verb seen in (47b), “reflecting a grammaticalization path already well established for this verb”. (47) Hakha Lai – Peterson (2007:131–2) a. Tsewmaŋ=niô door-ôaô ôa-ka-kal-piak. Tsewmang=erg market-all/loc a3sg-p1sg-go-ben ‘Tsewmang went to the market for me.’ b. Tsewmaŋ=niô ôaar-saa ôa-ka-peek. Tsewmang=erg chicken-meat a3sg-p1sg-give ‘Tsewmang gave me chicken meat.’
5.4.2â•…Benefactive ‘give’ compounds and applicative affixes cognate with ‘give’ verbs in the languages of the Pacific Alamblak (Papuan) has benefactive ‘give’ constructions identified by Bruce (1988) as verbal compounds – Ex. (48). (48)
Alamblak – Bruce (1988:39) Na yawyt yimam wikna-ha-më-an-m. 1sg dog people buy-give-r.pst-a1sg-p3pl ‘I bought the dog for the people.’
Quigley (2002:58–62) describes benefactive ‘give’ compounds in Awara (Papuan) and discusses morphological evidence of the distinction between such compounds and biverbal constructions. However, the affix indexing the beneficiary occurs between the two verb roots, which is unusual for compounds.
Benefactive applicative periphrases 
Other descriptions of Papuan languages do not analyze apparently similar constructions in terms of compounding, but rather as involving benefactive applicative affixes cognate with the homonymous ‘give’ verb – see Foley (1991:308–9) on Yimas, Anderson (1995) on Usan, Feldman (1986:48–9) on Awtuw. Unfortunately, it is difficult to see to what extent this choice really reflects differences in the nature of the constructions.
5.4.3â•…Benefactive ‘give’ compounds and applicative affixes cognate with ‘give’ verbs in African languages Benefactive constructions in which ‘give’ is contiguous to the other verb involved in the construction have been described in the Khoisan languages Ju|’hoan and Hoan – see Dickens (2005) on Ju|’hoan, Collins (2003) on Hoan. These constructions, illustrated by Ex. (49) & (50), are currently treated as SVCs, but they could equally be analyzed in terms of compounding, since nothing can be inserted between the two verbs, and this is the analysis reflected by the orthography used by Collins. (49)
Ju|’hoan – Dickens (2005:41) Dshàú n|óá |´àn ha dà´ámá kò ´msì. woman cook give 3sg child prep food ‘The woman cooked food for her child.’
(50)
Hoan – Collins (2003:2)
Gya”m-|a’a ’a-tsaxo-cu ‘am gye ki ||a”e. child-dim.pl prog-cook-give poss1sg mother prep meat ‘The children are cooking meat for my mother.’
In Section 6.1.2, Igbo has been mentioned as having complex predicates with nyé ‘give’ in benefactive operator function, but nyé also occurs as the second formant of compound verbs in which -nyé acts as a valency operator licensing not only benefactive complements, as in Ex. (51), but also dative or allative complements.15
15.╅ According to Lord (1977), a change from SOV to SVO constituent order would have been responsible for the emergence of Igbo compound verbs assuming functions that, in other Benue-Congo languages spoken in the same region, are more commonly assumed by SVCs, but the evidence in favor of this hypothesis is not very convincing. The point is that, when Carol Lord wrote this article, the hypothesis of a shift from proto-Niger-Congo SOV order to the SVO order attested in most Niger-Congo languages was advocated by several specialists. However, subsequent studies have cast serious doubts on the possibility to reconstruct �proto-Niger-Congo constituent order (see in particular Creissels (2005) for a discussion of evidence from West African languages). Moreover, the data I have collected includes ample evidence of compound verbs originating from BAPs in language families in which there is no evidence pointing to a possible relation between the emergence of such compounds and a change in constituent order.
 Denis Creissels
(51)
Igbo – Uchechukwu (2008) Dè-nyé nwókē à ákwù.â•›kwó. íkíké. write-give man dem paper permit ‘Write a permit for this man.’
Moreover, some Igbo dialects show evidence that nyé in such compounds tends to be reanalyzed as an applicative suffix. According to Uchechukwu (2008), the Ìgbúzò. Â�dialect shows a -nyé ~ -nyá alternation governed by the vowel harmony rule characteristic of Igbo affixes. For example, this dialect has gbá-nyá ‘pour in’, zù.â•›-nyá ‘buy for’ vs. bè-nyè ‘cut out for’, kú-nyé ‘scoop out for’ corresponding to Standard Igbo gbá-nyé, zù.â•›-nyé, bè-nyè, and kú-nyé respectively. Old Nubian has been mentioned in Section 4.3.2 as having “dative” periphrases involving the verbs den- ‘give (to me/us)’ and tr‒‒ ‘give (to you/him/them)’. Not surprisingly, modern Nubian languages have benefactive applicative markers resulting from the grammaticalization of these verbs, for example -dèen- and -tir- in Kunuz Nubian – Ex. (52). (52)
Kunuz Nubian – Abdel-Hafiz (1988:231) Id ay-gi baab-ki alle-deen-s-u. man 1sg-acc door-acc repair-ben-pst-a3sg ‘The man repaired the door for me.’
5.4.4â•…Benefactive ‘give’ compounds and applicative affixes cognate with ‘give’ verbs in Amerindian languages Among Amerindian languages, benefactive compounds occur in Kwaza (Amazonian isolate) – Ex. (53). In (53b), the possibility to express ‘give for’ by means of a compound formally constituted by two occurrences of wady ‘give’ can be viewed as evidence that ‘give’ in second position in benefactive compounds should rather be analyzed as having grammaticalized as an applicative suffix. (53) Kwaza – van der Voort (2004:373) a. Kudεrε-’wã mãmãñẽ=wa’dy-da-ki. Canderé-ao sing=give-1sg-decl ‘I sang for Canderé.’ b. Wεra-’wã haru’rai wady=wa’dy-taôy˜-ra. Vera-ao armadillo give=give-p1sg-imp ‘Bring the armadillo (meat) to Vera for me.’
Salas (2006:177–8) describes a productive mechanism of creation of verbal compounds by mere juxtaposition of two verb roots in Mapudungun, which suggests that the derived verbs formed by means of the applicative suffix -(l)el might originate from compound verbs whose second formant was elu- ‘give’. Benefactive applicative affixes cognate with a ‘give’ verb have also been signaled in€Sahaptian-Klamath by Rude (1991), in Iroquian by Mithun (2001), in Slave
Benefactive applicative periphrases 
� (Athabaskan) by Rice (1989), and in Tonkawa (an extinct language of Texas) by Hoijer€(1933).
6.â•… Autobenefactive applicative periphrases In principle, any type of construction licensing a benefactive NP can express autobenefactive via reflexivization. However, some languages have developed an expression of autobenefactive that formally cannot be analyzed as a combination of benefactive marking and reflexive marking. Kartvelian “version” (see among others Lacroix (this volume) on Laz) illustrates the possibility of a morphological distinction between nonreflexive benefactive (“objective version”) and self-benefactive (“subjective version”). In this section, we examine languages having an autobenefactive applicative periphrasis distinct from the periphrasis licensing a beneficiary other than the subject.
6.1â•… Autobenefactive ‘take’ periphrases In almost all cases I am aware of, the autobenefactive periphrasis differs from the periphrasis licensing a beneficiary other than the subject by involving a verb ‘take’ instead of ‘give’, as illustrated in Ex. (54) from Ojrot (Turkic, also known as Altai-kiži), and in Ex. (55) from Hindi. (54) Ojrot – Dyrenkova (1940:191) a. Uulčak bis-ke d’ol ayd-i‒p ber-di. boy 1pl-dat road tell-cvb give-pfv.a3sg ‘The boy showed us the road.’ b. Men bali‒k tud-up al-di‒-m. 1sg fish catch-cvb take-pfv-a1sg ‘I caught (for myself) a fish.’ (55) Hindi – Montaut (2004:125) a. Tum apnā kām jaldī kar lo! 2 refl work quickly do take-imp ‘Do your work quickly!’ ~ b. Maĩ tumhārā kām jaldī kar du‒gā.
1sg your work quickly do give.fut ‘I will do your work quickly.’
Not all languages that have BAPs have developed this expression of autobenefaction: most attestations of autobenefactive ‘take’ periphrases I have been able to find come from an area including the following language families: Mongolic, Turkic, IndoAryan, Dravidian, Tibeto-Burman, and Austroasiatic languages. It has already been established that benefactive ‘give’ periphrases, either of the marked-Vlex type or of the serial type, are particularly widespread among the languages spoken in this area.
 Denis Creissels
Autobenefactive periphrases seem particularly common among Mongolic and Turkic languages, as already noted by Krueger (1964). References on autobenefactive ‘take’ periphrases in Mongolic languages include among others Skribnik (2003:117) on Buryat,16 Nugteren (2003:281) on Shira Yughur,17 Hugjiltu (2003:342) on Bonan,18 and Kim (2003:360) on Santa.19 References on autobenefactive ‘take’ periphrases in Turkic languages include among others Bodrogligeti (2001:287) on Chagatay,20 and Grunina (2005:287) on Turkmen.21 As signaled in Section 4.3.1, Tajik (Iranian) has a ‘give’ periphrases expressing benefaction that probably developed under the influence of Turkic languages; not surprisingly, Tajik also has a ‘take’ periphrasis expressing autobenefaction, as in navišta giriftan ‘write down for oneself ’, lit. ‘take writing’ vs. navišta dodan ‘write down for someone’, lit. ‘give writing’ – Mamatov & al. (2005). Anderson (2001, 2002) analyzes the ‘give’ vs. ‘take’ contrast in Turkic BAPs. Starting from the description of ‘take’ and ‘give’ periphrases in Altai-Sayan Turkic languages (Tofa, Tuvan, Xakas, Ojrot) he observes that similar periphrases are attested in “languages from the farthest reaches of the Turkic-speaking world, both temporally and geographically, with data from such languages as modern Yakut (Saxa), Turkmen, Uyγur, Tatar, XalaJˇ and Čuvaš, and Old Turkic”. He concludes that this feature dates back to Proto-Turkic times.
16.â•… “Other common auxiliaries include … üge- ‘to give’ vs. aba- ‘to take’ [benefactive].” 17.â•… “The combination -j’ ab- (imperfective converb + ‘to take’) indicates that the action is performed for the subject’s benefit … Similarly -j’ ög- (imperfective converb + ‘to give’) indicates that the action is performed for someone else’s benefit, e.g. … ci nanda misgi xala-j’ ög ‘sew a garment for me!’ (ci=2sg, nanda=1sg.dat).” 18.â•… “The Bonan auxiliaries may be divided into three main groups: … (3) those indicating the beneficiary of the action: aw- ‘to take; to do for oneself ’ (<*ab-), oke- ‘to give; to do for somebody else’.” 19.â•… “The two auxiliaries agi- ‘to take’ (irregularly from *ab-) vs. ogi- ‘to give’ (<*ög-) indicate that the action has a beneficiary (‘for’).” 20.â•… “[used as ‘descriptive verbs’ with gerunds in -p] al- ‘to take, to receive’ indicates that the action takes place in the interest of the subject … ber- ‘to give’ signals that the action is carried out in the interest of someone else.” 21.â•… “The verbs al- ‘give’ and ber- ‘take’ can occur as auxiliary verbs with the main verb in the form of the gerund in -Ip. In this case, almak indicates that the action is performed for the subject, in his interest, or is directed towards him, cf. adresini y´azyp aldym ‘I wrote his address (for myself)’. The verb bermek in this function points to an action directed from the subject to someone else or performed in the interest of someone else.”
Benefactive applicative periphrases 
References on autobenefactive ‘take’ periphrases in Indo-Aryan languages include among others Bhatia (1993:326–7) on Punjabi,22 Paul (2003:3) on Bengali,23 and Pradeshi (2001) on Marathi. Among Dravidian languages, an autobenefactive ‘take’ periphrases is signaled by Krishnamurti (2003:381) in Telugu.24 Among Tibeto-Burman languages, Ebert (1994:61) briefly mentions the autobenefactive use of ‘take’ in verb combinations of the serial type in the Kiranti languages Bantawa and Thulung, and Okell & Allott (2001:176) mention the use of the Burmese verb yu ‘take’ as a postverbal marker expressing ‘V and take’, ‘V for oneself ’. Among Austroasiatic languages, an autobenefactive ‘take’ periphrasis is signaled by Jenny (2005:204) in Mon.25
6.2â•… Autobenefactive ‘eat’ periphrases ‘Eat’ verbs semantically depart from the most typical action verbs in that the manipulation exerted by the agent of ‘eat’ on the patient is not the real aim of an eating event: by manipulating the patient, the agent of ‘eat’ aims at satisfying his/her hunger, i.e. at producing an effect on him/herself. Consequently, it is not surprising that ‘eat’ verbs can grammaticalize as operators in autobenefactive periphrases. Mundari and Ho (Munda) attest the grammaticalization of jom- ‘eat’ as an autobenefactive operator – Hook (1991). Munda languages attest other grammaticalized uses of the same root: –â•fi –â•fi
as already mentioned in Section 2.3, Santali uses jfm- ‘eat’ as an operator in passive periphrases; in Kharia, jom- is not used as a full verb anymore, but subsists as an “autoÂ�poesis” marker whose meaning has a clear connection with autobenefaction, since it denotes “that something happened on its own, i.e. there was no outside force
22.â•… “laiNaa ‘to take’ indicates self-benefactive meaning; co-occurring with main verbs such as ‘to cry’, ‘to laugh’, it exhibits an introvert action; deuNaa ‘to give’ expresses benefactive meaning, and the beneficiary is other than the subject of the sentence. With verbs such as ‘to cry’, ‘to laugh’, it denotes an overt action.” 23.â•… “Consider a verbal root Sajano ‘decorate’. It participates in following compound verb constructions: Sajiye deoya ‘decorate for other’s benefit’ … Sajiye neoya ‘decorate and the result is directed towards the actor’.” 24.â•… “In Modern Telugu, the valency changing auxiliaries … are: … (A2.4) Reflexive: Vppl + kon- ‘take’, e.g. cēs- ‘to do’: //cēs-i-kon-// → /cēs-u-kon-/ ‘to do something for oneself ’, wiragagozz-u-kon- ‘to break (a body part) by oneself ’.” 25.â•… “At least since M(iddle) M(on), the use of
‘take’ as postverbal operator implying action for one’s benefit or purpose is attested.”
 Denis Creissels
which caused it to happen. With potentially volitional predicates on the other hand, it denotes that the agent simply performed the action because s/he wanted to and was under no obligation to do so.” – Peterson (2006:233). Among Tibeto-Burman languages, the autobenefactive use of ‘eat’ verbs is attested in South East Kiranti language – see Ebert (1997) on Athpare, Rai (1985) on Bantawa.
7.â•…Benefactive applicative periphrases with verbs other than ‘give’ in valency operator function When discussing the existence of BAPs with verbs other than ‘give’ in valency operator function, true periphrases involving a verb grammaticalized as a valency operator must be distinguished from biverbal constructions in which a benefactive meaning is implied by the inherent semantics of the verb in second position, but in which this verb retains its specific semantics, as discussed for Thai by Jenny (this volume). The Tibeto-Burman language Lahu uses pî ‘give’ for third person beneficiaries only, and uses a particle cognate with là ‘come’ to express that an action is performed for the benefit of a speech act participant: “Lahu is careful to specify for whose benefit the verbal action is performed. This is done by two morphemes, the Vv pî ‘give’ and the Pv lâ (< là ‘come’). The outer-directed pî is used to indicate that the action affects a third person, while the inner-directed lâ shows that the action affects a non-third person, e.g. chf lâ (Vh + Pv) ‘chop for me/us/you’; chf pî (Vh + Vv) ‘chop for him/her/them’.”26
Matisoff (2003:21) BAPs with a verb glossed ‘help’, ‘put’, ‘do/act for’ in valency operator function are Â�sporadically attested. In Cantonese, béi ‘give’ occurs in V2 position in SVCs with a dative rather than benefactive function – Ex. (56a), and typical beneficiaries are introduced by bōng ‘help’ in the construction illustrated by Ex. (56b). (56) Cantonese – Matthews & Yip (1994:201/143) a. Kéuih kàhmmáahn dá-dihnwá béi ngóh. 3sg last night call-phone give 1sg ‘S/he gave me a call last night.’ b. Ngóh bōng léih dá-dihnwá. 1sg help 2sg call-phone ‘I’ll phone for you.’27 26.â•… Vh = ‘main verb’ in a verb concatenation, Vv = post-head ‘versatile’ verb, Pv = verb particle. 27.â•… Note that, as explicitly stated by Matthews & Yip, “although bōng in isolation means ‘help’, the addressee here is not expected to participate actively; rather, the speaker is offering to
Benefactive applicative periphrases 
LaPolla (2003b) mentions a BAP involving a verb glossed ‘help’ in Dulong (Â�Nungish, Tibeto-Burman). In contrast to Cantonese, ‘help’ occurs in second position. As illustrated by Ex. (57), Dulong also has a benefactive construction involving a “benefactive auxiliary” č taking verb inflection and occupying a position that could be identified as the V2 position in a SVC, but apparently devoid of independent verbal uses. The etymology of this “benefactive auxiliary” is not discussed by LaPolla (2003b). (57) Dulong – LaPolla (2003b:678) a. àŋ . –-ŋ 6ˇgf` tεˇ rī s6ˇn6¿ŋ. 3sg wood 1sg erg carry help.1sg ‘I carry wood for him’ (lit. ‘I help carry his wood’) b. 6ˇgf` tεˇ a`ŋ . –-ŋ rī čŋ.
m
m
1sg erg 3sg wood carry ben.1sg ‘I carry wood for him.’
The use of a verb glossed ‘put’ as a valency operator in a BAP has been signaled in Hua (Papuan) – Ex. (58), and in Telugu (Dravidian) – Ex. (59).28 (58)
Hua – Haiman (1980), quoted by Foley (1986:98) Zu ki-na d-te. house build-a3sg p1sg-put.a3sg.decl ‘He built me a house.’
(59)
Telugu – Krishnamurti (2003:381) Mā āwi2a rōjū padimandiki annam wa^2-i pe2utundi. poss1sg wife daily for_ten_persons food cook-cvb put ‘My wife daily cooks food for ten persons.’
The use of a verb glossed ‘do for’ as a benefactive operator is found in Tukang Besi (Austronesian) – Ex. (60). (60)
Tukang Besi – Donohue (1999:187) No-wila kua daoa ako te ina-no. a3pl-go all market do for core mother-3pl ‘They went to the market for their mother.’
The use of verbs glossed ‘say’ licensing the expression of a beneficiary is less easy to analyze. In Amharic, Gumer and other Ethiosemitic languages, the mention of a perform the action single-handedly”. In other words, a possible English equivalent would be I’ll help you by phoning, but NOT I’ll help you to phone. 28.â•… In Telugu, according to Masayoshi Shibatani (p.c.), this meaning can be expressed without the use of ‘put’. Many native speakers find it difficult to distinguish the meaning of this expression with or without ‘put’. Some detect the ‘in preparation for the future action’ meaning when ‘put’ is used.
 Denis Creissels
beneficiary may require the use of a periphrasis involving a converbal form of a verb glossed as ‘say’, as illustrated by Ex. (61) from Amharic. (61)
Amharic – Azeb Amha (p.c.) Arägga lä-Abbäbä bïlo Käbbädä-n gäddälä. Aregga dat-Abbebe say.3sgm.cvb Kebbede-acc kill.3sgm.pst ‘Aregga killed Kebbede for (the sake/benefit of) Abbebe.’ (lit. ‘Aregga killed Kebbede saying “for Abbebe”.’)
In this construction, the converbal form of ‘say’ clearly has an applicative function. However, the role of beneficiary is not encoded by the applicative operator itself (the converbal form of ‘say’), but by the dative case. In other words, this construction is not inherently a BAP, since it dissociates the applicative function proper from the role assignment function.29 In Gumer, according to Völlmin (this volume), beneficiaries occur in the dative case, and applicative ‘say’ is required when another dative-marked NP is present in the construction of the lexical verb – in particular, in the presence of a patient NP requiring dative marking, as in Ex. (62). (62) Gumer (Völlmin, this volume) Ar6gga y6-K6bb6d6 y-Abb6β6 Aregga dat-Kebbede dat-Abbebe t-i-βi‒r k’w6t’t’6r-6-n-i‒m. when-a3sgm-say.ipfv kill.pfv-a3sgm-p3sgm-??? ‘Aregga killed Kebbede for Abbebe.’
In other words, the use of ‘say’ in applicative operator function in Gumer can be viewed as a disambiguating strategy aiming to avoid the presence of two NPs with identical case marking in the construction of the same verb. Here again, in spite of the fact that this construction may be obligatory in order to express benefaction, it would not be correct to identify it as a BAP. BAPs involving verbs glossed ‘show’ are sometimes mentioned, in particular in Akan (Kwa). However, ‘show’ periphrases seem to license recipients or goals rather
29.â•… The applicative use of ‘say’ verbs must be explained in the light of their well-known tendency to develop polysemous meanings including in particular volition. Note also that the use of ‘say’ verbs in light verb constructions in which most languages use ‘do’ verbs is particularly widespread in Ethiosemitic and neighboring languages. A converbal form of a ‘say’ verb may thus grammaticalize as an applicative operator whose presence is required in order to expand the construction of an action verb.
Benefactive applicative periphrases 
than beneficiaries proper, and I have come across no unquestionable case of benefactive ‘show’ periphrasis. Aikhenvald (2003:437) argues that Tariana (Maipurean, Amazonia) has benefactive SVCs formed with -uma ‘seek’, ‘find’ and -ni ‘do’ as their first components. However, judging from the two examples she provides, the analysis of these verbs as possible benefactive operators in Tariana SVCs seems highly dubious: –â•fi
–â•fi
In the example with -uma ‘seek’ (whose relevant part is translated as ‘I look for food for women’), it seems obvious that -uma contributes to the meaning of the serial construction seek eat → look for food with its basic meaning ‘seek’, and there is no evidence that it is involved in the assignment of the beneficiary role to ‘women’. No word glossed as ‘prepare’ figures in the gloss of the example with -ni ‘do’, translated as ‘Prepare manioc for them to eat’, and the gloss suggests that it should better be analyzed as involving a causative SVC with the meaning ‘Make them eat manioc’.
8.â•… Conclusion In this paper, I have surveyed BAPs of different types. The main conclusions can be summarized as follows: a. BAPs using verbs other than ‘give’ in valency operator function, or in which ‘give’ occurs in first position, are exceptional. b. Two of the three formal types of BAPs (the serial type and the marked-Vlex type) are widely attested in the languages of the world. They are particularly common on the Asian continent, where they occupy two distinct but contiÂ� guous areas. Outside Asia, attestations of BAPs of the marked-Vlex type are sporadic, whereas the serial type of BAP is common in all language families or areas known for their overall tendency towards serialization, with the exception of Oceanic. c. In benefactive ‘give’ periphrases, ‘give’ may grammaticalize as a benefactive adposition or an applicative marker. Benefactive verbal compounds constitute an intermediate stage in the conversion of ‘give’ into an applicative marker. d. Autobenefactive ‘take’ periphrases are particularly common among Mongolic and Turkic languages, and are also attested in Indo-Aryan, Dravidian, TibetoBurman, and Austroasiatic languages, but do not seem to be attested outside this area.
 Denis Creissels
Abbreviations 1/2/3 a
1rst/2nd/3rd person person mark referring to the agent of prototypical action verbs acc accusative act active all allative an animate ao animate object ben benefactive caus causative clf classifier comp complementizer core core argument cvb converb d person mark referring to a participant represented by a dative NP dat dative decl declarative def definite dem demonstrative ds different subject enunc enunciative particle erg ergative exc exclusive foc focalization fut future gen genitive
ger imp ind indma inf ipfv iprf loc p
pfv pl poss prep prf prog prs pst refl rel r.pst sbj seq sg sgf sgm top
gerund imperative indicative indicative, male addressee infinitive imperfective imperfect locative person mark referring to the patient of prototypical action verbs perfective plural possessive preposition perfect progressive present past reflexive relativizer recent past subject sequential singular singular feminine singular masculine topic
References Abdel-Hafiz, Ahmed S. 1988. A Reference Grammar of Kunuz Nubian. Ph.D. dissertation, State University of New York at Buffalo. Abraham, Roy C. 1962. Dictionary of Modern Yoruba, 2nd edn. London: Hodder & Stoughton. Aikhenvald, Alexandra Y. 2006a. Serial verb constructions in typological perspective. In Serial Verb Constructions, Alexandra Y. Aikhenvald & Robert M.W. Dixon (eds.), 1–68. Oxford: OUP. Aikhenvald, Alexandra Y. 2006b. A Grammar of Tariana, from Northwest Amazonia. Cambridge: CUP.
Benefactive applicative periphrases 
Ameka, Felix K. 2003. Multiverb constructions in a West African areal typological perspective. Paper presented at the International Workshop on Serial Verb Constructions, La Trobe. Anderson, Gregory D.S. 1995. Ditransitives, possessor raising, copying-to-OBJ: Animacy in morphosyntax. Chicago Linguistic Society 31. 1–17. Anderson, Gregory D.S. 2001. Subject version and object version in Tofa auxiliary verb constructions. Turkic Languages 5(1): 240–269. Anderson, Gregory D.S. 2002. Auxiliary verb constructions in Old Turkic and Altai-Sayan Â�Turkic. Paper presented at the Old Turkic Symposium, Frankfurt. Asher, Ronald E. & Kumari, T.C. 1997. Malayalam. London: Routledge. Azeb, Amha & Dimmendaal, Gerrit 2006. Converbs from an African perspective. In Catching Language: Issues in Grammar Writing, Felix Ameka, Alan Dench & Nicholas Evans (eds.), 393–440. Berlin: Mouton de Gruyter. Bhatia, Tej. 1993. Punjabi. A Cognitive-descriptive Grammar. London: Routledge. Bodomo, Adams & van Oostendorp, Marc. 1993. Serial verb nominalization in Dagaare. Paper presented at the 24th Annual Conference on African Linguistics. Colombus OH. Bodrogligeti, András J.E. 2001. A Grammar of Chagatay. Munich: Lincom Europa. Browne, Gerald M. 2002. Old Nubian Grammar. Munich: Lincom Europa. Bruce, Leslie P. 1984. The Alamblak Language of Papua New Guinea East Sepik. Canberra: Â�Australian National University. Butt, Miriam & Tantos, Alexandros. 2004. Verbal semantics via Petri nets. Paper presented at the LFG04 Conference, University of Canterbury. Carlson, Robert J. 1991. From verb to preposition in Senufo. In Approaches to Grammaticalization [Typological Studies in Language 19:2], Elizabeth Closs Traugott & Bernd Heine (eds.), 201–223. Amsterdam: John Benjamins. Christaller, Johann G. 1933. Dictionary of the Asante and Fante Language Called Tshi (Twi), 2nd edn, revised and enlarged. Basel: The Basel Evangelical Missionary Society. Collins, Chris. 2003. The internal structure of vP in Ju|’hoansi and Hoan. Studia Linguistica 57: 1–25. Creissels, Denis. 2005. S-O-V-X constituent order and constituent order alternations in West African languages. In Proceedings of the 31st Annual Meeting of the Berkeley Linguistics Society: Special Session on Languages of West Africa, Rebecca T. Cover & Yuni Kim (eds.), 37–52. Berkeley CA: Berkeley Linguistics Society. Creissels, Denis. 2006. Syntaxe générale, une introduction typologique, Vol. 2: la phrase. Paris: Hermès. Crowley, Terry. 2002. Serial Verbs in Oceanic, a Descriptive Typology. Oxford: OUP. Dauphin, Antoine. 1977. Cours de Vietnamien. Paris: L’Asiathèque. Dickens, Patrick J. 2005. A Concise Grammar of Ju|’hoan. Cologne: Rüdiger Köppe. Donohue, Mark. 1999. A Grammar of Tukang Besi. Berlin: Mouton de Gruyter. Dryer, Matthew S. 2003. Word order in Sino-Tibetan languages from a typological and geographical perspective. In The Sino-Tibetan Languages, Graham Thurgood & Randy J. LaPolla (eds.), 43–55. London: Routledge. Dyrenkova, Nadežda P. 1940. Grammatika ojrotskogo jazyka. Leningrad: Akademia Nauk SSSR. Ebert, Karen. 1994. The Structure of Kiranti Languages. Zurich: Universität Zürich, Seminär für Allgemeine Sprachwissenschaft. Ebert, Karen. 1997. A Grammar of Athpare. Munich: Lincom Europa. Emeneau, Murray B. 1984. Toda Grammar and Texts. Philadelphia PA: American Philosophical Society.
 Denis Creissels Feldman, Harry. 1986. A Grammar of Awtuw. Canberra: Pacific Linguistics. Foley, William A. 1986. The Papuan Languages of New Guinea. Cambridge: CUP. Foley, William A. 1991. The Yimas Language of New Guinea. Stanford CA: Stanford University Press. Gaunt, John & Bayarmandakh, L. 2004. Modern Mongolian: A Course-Book. London: Routledge. Genetti, Carol. 2007. A Grammar of Dolakha Newar. Berlin: Mouton de Gruyter. Grunina, El’vira A. 2005. Turkmenskij jazyk. Moscow: ‘Vostočnaja literatura’ RAN. Haboud, Marleen. 1994. On language contact and grammaticalization in Ecuadorian Highland Spanish. Ms, University of Oregon. Haboud, Marleen. 1998. Quichua y castellano en los Andes Ecuatorianos, los efectos de un contacto prolongado. Quito: Ediciones Abya-Yala. Hagemeijer, Tjerk. 2000. Serial Verb Constructions in São-Tomense. MA thesis, University of Lisbon. Haiman, John. 1980. Hua, a Papuan Language of the Eastern Highlands of New Guinea [Studies in Language Companion Series 5]. Amsterdam: John Benjamins. Heine, Bernd & Kuteva, Tania. 2002. World Lexicon of Grammaticalization. Cambridge: CUP. Hendriksen, Hans. 1944. Syntax of the Infinite Verb Forms of Pāli. Copenhagen: Einar Munksgaard. Hoijer, Harry. 1933. Tonkawa: An Indian Language of Texas. New York NY: Columbia University. (Extract from Handbook of American Indian languages, Vol. 3). Hook, Peter E. 1991. The compound verb in Munda: An areal and typological overview. Language Sciences 13/2. 181–195. Hugjiltu, Wu. 2003. Bonan. In The Mongolic Languages, Juha Janhunen (ed.), 325–45. London: Routledge. Ikoro, Suanu M. 1996. The Kana Language. Leiden: Research School CNWS. Iwasaki, Shoichi & Yap, Foong-Ha. 2000. Give constructions in Thai and beyond: A cognitive and grammaticalization perspective. Proceedings of the International Conference on Tai Studies, July 29–31, 1998. 371–382. Bangkok: Institute of Language and Culture for Rural Development, Mahidol University. Jakobi, Angelika & Crass, Joachim. 2004. Grammaire du beria (langue saharienne). Köln:Â� Rüdiger Köppe. Jenny, Mathias. 2005. The Verb System of Mon. Zurich: Universität Zürich, Seminär für Allgemeine Sprachwissenschaft. Kim, Stephen S. 2003. Santa. In The Mongolic Languages, Juha Janhunen (ed.), 346–63. London: Routledge. Kratochvíl, František. 2007. A Grammar of Abui. A Papuan Language of Alor. Utrecht: LOT. Krishnamurti, Bhadriraju. 2003. The Dravidian Languages. Cambridge: CUP. Krueger, John R. 1964. The Turco-Mongolian benefactive. Journal of the American Oriental Society 84/4. 410–1. Kuno, Susumu. 1973. The Structure of the Japanese Language. Cambridge MA: The MIT Press. Lai, Huei-Ling. 2001. On Hakka BUN: A case of polygrammaticalization. Language and Linguistics 2(2): 137–153. LaPolla, Randy J. 2003a. Overview of Sino-Tibetan morphosyntax. In The Sino-Tibetan Languages, Graham Thurgood & Randy J. LaPolla (eds.), 22–42. London: Routledge. LaPolla, Randy J. 2003b. Dulong. In The Sino-Tibetan Languages, Graham Thurgood & Randy J. LaPolla (eds.), 674–82. London: Routledge.
Benefactive applicative periphrases 
Leroy, Jacqueline. 2003. Grammaire du mankon. Thèse de doctorat d’état, University of Paris 3. Li, Charles N. & Thompson, Sandra A. 1981. Mandarin Chinese. A Functional Reference Grammar. Berkeley CA: University of California Press. Lidz, Liberty A. 2006. A synopsis of Yongnin Na (Mosuo). Paper presented at the 39th International Conference on Sino-Tibetan Languages and Linguistics. Seattle WA: University of Washington. Lord, Carol. 1977. How Igbo got from SOV serializing to SVO compounding. Studies in African Linguistics Supplement 7: 145–155. Lord, Carol. 1993. Historical Change in Serial Verb Constructions [Typological Studies in Language 26]. Amsterdam: John Benjamins. Lord, Carol, Yap, Foong-Ha & Iwasaki, Shoichi. 2002. Grammaticalization of ‘give’: African and€ Asian perspectives. In New Reflections on Grammaticalization [Typological Studies in Language 49] Ilse Wischer & Gabriele Diewald (eds.), 217–235. Amsterdam: John Benjamins. Mamatov, Jahangir, Harrell, S.J., Kehoe, Kathy & Khodjibaev, Karim. 2005. Tajik-English Dictionary. Springfield VA: Dunwoody Press. Margetts, Anna & Peter K. Austin. 2007. Three-participant events in the languages of the world: Towards a crosslinguistic typology. Linguistics 45/3. 393–450. Matisoff, James A. 1991. Areal and universal dimensions of grammaticalization in Lahu. In Approaches to Grammaticalization [Typological Studies in Language 19:2], Elizabeth Closs Traugott & Bernd Heine (eds.), 383–453. Amsterdam: John Benjamins. Matisoff, James A. 2003. Lahu. In The Sino-Tibetan Languages, Graham Thurgood & Randy J. LaPolla (eds.),208–21. London: Routledge. Matthews, Stephen & Yip, Virginia. 1994. Cantonese. A Comprehensive Grammar. London: Routledge. Mithun, Marianne. 2001. Understanding and explaining applicatives. Chicago Linguistic Society 37(2): 73–97. Mixco, Mauricio. 1997. Mandan. Munich: Lincom Europa. Montaut, Annie. 2004. A Grammar of Hindi. Munich: Lincom Europa. Moser, Rosmarie. 2005. Grammaticalization chains of the verb kàre ‘to give’ in Kabba. In Studies in African Linguistic Typology [Typological Studies in Language 64], Erhard F.K. Voeltz (ed.), 277–301. Amsterdam: John Benjamins. Neukom, Lukas. 2001. Santali. Munich: Lincom Europa. Newmeyer, Frederick J. 2004. Some thoughts on the serial verb construction. Workshop ‘La notion de “construction verbale en série” est-elle opératoire?’ Paris: EHESS. Nugteren, Hans. 2003. Shira Yughur. In The Mongolic Languages, Jahu Janhunen (ed.), 265–85. London: Routledge. Okell, John & Allot, Anna J. 2001. Burmese/Myanmar Dictionary of Grammatical Forms.Â� London: Curzon. Onumajuru, Emeka M. 1985. Système verbal de la langue igbo (parler d’Orlu). Ph.D. dissertation, University of Grenoble. Palayer, Pierre. 1989. La langue sar (sud du Tchad). Thèse de doctorat d’état, University of Tours. Palmer, Bill. 1999. A Grammar of the Kokota Language, Santa Isabel, Solomon Islands. Ph.D. dissertation, University of Sidney. Pardeshi, Prashant. 2001. The explicator compound verb in Marathi: Definitional issues and criteria for identification. Kobe Papers in Linguistics 3: 94–111.
 Denis Creissels Paul, Soma. 2003. Composition of compound verbs in Bangla. In Proceedings of the workshop on Multi-Verb Constructions, Dorothee Beermann & Lars Hellan (eds.), Trondheim: Norwegian University of Science and Technology. Paul, Waltraud. 2004. The ‘Serial Verb Construction’ in Chinese: A Gordian knot. Workshop ‘La notion de ‘construction verbale en série’ est-elle opératoire?’ Paris: EHESS. Peterson, David A. 2007. Applicative Constructions. Oxford: Oxford University Press. Peterson, John. 2006. Kharia, a South Munda language, Vol. 1: Grammatical analysis. Ms, University of Osnabrück. Quigley, Suzan R. 2002. The Awara Verbal System. MA thesis, University of North Dakota. Rai, Novel K. 1985. A Descriptive Study of Bantawa. Ph.D. dissertation, University of Poona. Remijsen, Bert. To appear. Nouns and verbs in Magey Matbat. In Typological and Areal Analyses: Contributions from East Nusantara, Marian Klamer & Michael Ewing (eds.). Leiden: KITLV Press. Rice, Keren. 1989. A Grammar of Slave. Berlin: Mouton de Gruyter. Rowlands, Evan C. 1969. Teach Yourself Yoruba. Lincolnwood IL: Teach Yourself Books. Rude, Noel. 1991. Verbs to promotional suffixes in Sahaptian and Klamath. In Approaches to Grammaticalization [Typological Studies in Language 19:2], Elizabeth Closs Traugott & Bernd Heine (eds.), 183–99. Amsterdam: John Benjamins. Salas, Adalberto. 2006. El mapuche o araucano. Santiago de Chile: Centro de Estudios Públicos. Shibatani, Masayoshi. 2003. Directional verbs in Japanese. In Motion, Direction and Location in Languages [Typological Studies in Language 56], Erin Shay & Uwe Seibert (eds.), 259–286. Amsterdam: John Benjamins. Shibatani, Masayoshi. 2006. On the form of complex predicates. Demystifying/demythologizing serial verbs. Seminario de complejidad sintáctica. Hermosillo, Sonora. Shibatani, Masayoshi & Sung-Yeo Chung. 2007. On the grammaticalization of motion verbs: A€Japanese-Korean perspective. Japanese/Korean Linguistics 15. Skribnik, Elena. 2003. Buryat. In The Mongolic Languages, Juha Janhunen (ed.), 102–28. London: Routledge. Solomiac, Paul. 2007. Phonologie et morphosyntaxe du dzùùngoo de Samogohiri. Ph.D. Â�dissertation, University of Lyon. Tamura, Suzuko. 2000. The Ainu Language. Tokyo: Sanseido. Tida, Syuntarô. 2006. A Grammar of the Dom Language. Ph.D. dissertation, University of Kyoto. Toronzoni, Ngama-Nzombio T.N. 1989. Description du ngbandi. Ph.D. dissertation, Université libre de Bruxelles. Tsumagari, Toshiro. 2003. Dagur. In The Mongolic Languages, Juha Janhunen (ed.), 129–53. London: Routledge. Uchechukwu, Chinedu. 2008. The grammaticalization of prepositional markers in Igbo: The example of the verb root -nyé ‘give’. In Syntax and Semantics of Spatial P [Linguistik Aktuell/Linguistics Today 120], Anna Asbury, Jakub Dotlacil & Berit Gehrke (eds.), 398–407. Amsterdam: John Benjamins. van der Voort, Hein. 2004. A Grammar of Kwaza. Berlin: Mouton De Gruyter. Van Valin, Robert D. & LaPolla, Randy J. 1997. Syntax. Structure, Meaning and Function. Â�Cambridge: CUP. Vogel, Sylvain. 2006. Introduction à la langue et aux dits traditionnels des Phnong de Mondulkiri. Phnom Penh: Editions Funan.
Benefactive applicative periphrases 
Voorhoeve, Jan. 1975. Serial verbs in Creole. Paper presented at the International Conference on Pidgins and Creoles, University of Hawai’i. Welmers, William E. 1973. African Language Structures. Berkeley CA: University of California Press. Werner, Roland. 1987. Grammatik des Nobiin (Nilnubisch). Hamburg: Helmut Buske. Williamson, Kay. 1965. A Grammar of the Kolokuma Dialect of Ịjọ. Cambridge: CUP. Williamson, Kay & Timitimi, A.O. 1983. Short Ịzọn-English Dictionary. Port Harcourt: University of Port Harcourt Press. Zúñiga, Fernando. 2010. ‘Benefactive and malefactive applicativization in Mapudungun’. This volume: 203–218.
Cross-linguistic categorization of benefactives by event structure A preliminary framework for benefactive typology* Tomoko Yamashita Smith Osaka University
The first of its kind, this study classifies the semantics of benefactive constructions from over sixty languages according to event structure types, identifying two basic types. The “agentive benefactive” construction, always includes an agent and expresses that the agent intentionally carries out the act for the sake of the affectee. The “event benefactive” construction, on the other hand, expresses that the affecting event may or may not have an agent, and includes windfall situations. The subtypes of the agentive benefactive are: “unrestricted agentive benefactive” (agent can be anybody), “non-self-benefactive” (agent is not an affectee), “selfbenefactive” (agent and affectee are identical), and “shared-benefit construction” (which expresses that ‘X does something for the benefit of X himself and Y’).
1.â•… Introduction This paper is a cross-linguistic study of constructions that express the notion of benefit, namely benefactive constructions. Prototypically, benefactive constructions
*I would like to express my appreciation to the following people: Mathias Jenny, Seppo Kittilä, Prashant Pardeshi, Paula Radetzky, Masayoshi Shibatani, Tony Smith, Ken VanBik, Naoko Yamashita, Su Yutang, Fernando Zúñiga, and the anonymous reviewers. The abbreviations used in this paper are as follows: 1 2 3 acc ben cp cv dat des
first person second person third person accusative benefactive conjunctive participle converb dative desiderative
f erg fut gen loc m n nsben nrpst
feminine ergative future genitive locative masculine neuter non-self-benefactive near past
p pst.i prs ptcpl pst quo s sben top
plural unmarked past present participle past quotation singular self-benefactive topic
Tomoko Yamashita Smith
express that someone does an action for the sake of another, as in the English sentences Linda made Hana pancakes and Leo played the piano for Mark.1 Comparable benefactive constructions are widely attested in the world’s languages, and it may well be the case that all languages have special constructions to describe this common sort of “benefactive” event. However, this sort of benefactive event is not the only one that we experience in our lives. For example, one regularly does actions just for oneself, or one may encounter a windfall, in which one benefits from an action without an agent2 intentionally performing an action or from some kind of agentless event. These semantically different benefactive events can correspond to different benefactive constructions. For example, Vietnamese has a verbal morpheme to express that the subject referent undergoes a pleasant experience. In Finnish, there are many benefactive postpositions that code beneficiary and their semantics are much more specific (e.g.€‘instead of ’, ‘to the joy of ’) (Kittilä 2006, 2009 this volume). Therefore, it is important to investigate what sort of benefactive constructions exist in the world’s languages and categorize their semantic types. In contrast to earlier typologies of benefaction (such as Van Valin & LaPolla 1997 and Kittilä 2005), the focus in this paper lies not on the nature of benefaction, but on features of beneficiary (Section 2.2) and the nature of the causer/agent (Section 2.3). There are some cross-linguistic/typological studies on benefactive constructions. For instance, Shibatani (1996) provides a cognitive account of cross-linguistic variation between benefactive constructions with an agent. In order to discuss some syntactic and semantic restrictions, he limits true benefactive constructions to those in which the beneficiaries are coded as an argument. Yamada (1996, 2001, 2004) finds that regular give verbs are more likely to be grammaticalized into benefactive markers than receive verbs in languages. As for the semantic types of benefactive constructions, Van Valin & LaPolla (1997: 383–384) discuss three types of benefactives: (1) Recipient benefactive, in which the beneficiary is also the recipient, as in John baked Mary a pie, (2) plain benefactive, in which the act is done for some kind of benefit for the beneficiary, such as amusement and enjoyment, as in Leo played the piano for Linda, and (3) deputative benefactives, in which the agent performs some action in place of the beneficiary, as in Pat stood in line for Kim (384). Moreover, Kittilä (2005) contrasts two types of benefactive roles, (pure) beneficiary and recipient-beneficiary and argues that the two roles are encoded differently in some languages and the traditional term “benefactive” is not always sufficient. In most of the cases, the previous studies have focused on benefactives that always include an agent, such as ditransitive constructions and benefactive uses of the dative
1.â•… Fillmore (1968) points out that the beneficiary requires an agent elsewhere in the sentence. 2.â•… I use the term agent to refer to the doer who performs, initiates, or controls the event.
Cross-linguistic categorization of benefactive constructions
case (e.g. dativus commodi in Latin);3 however, in order to fully understand how we encode the notion of benefit in languages, we must answer questions such as, “Who is the beneficiary: the agent himself or someone else?” and “What kind of benefactive events do human languages really encode?” “Are there only the three semantic types that are discussed in Van Valin & LaPolla (1997)?” Smith (2005) is one of the first studies to tackle such questions, and to introduce and discuss different semantic types of benefactive constructions using cross-linguistic data. Since then, I have examined more data, and new analysis and discussion are included in this study. I take the functionalist approach and examine semantic as well as pragmatic functions of benefactive constructions, and compare equivalent phenomena in different languages. The semantic types of benefactive constructions I discuss in this study are the result of looking at reference grammars of individual languages, past studies on benefactives and eliciting data from native-speaker informants. Unfortunately, the grammar reference books lack full descriptions of benefactive constructions, so I was not able to complete a true typological study. However, working with the means at hand I was nevertheless able to find clear patterns of the event structures of benefactives cross-linguistically.
1.1â•… Definition of benefactive constructions In this study, benefactive constructions explicitly indicate an event as good4 for a participant using some kind of morphosyntactic means. Thus, sentences such as he admires me and we luckily arrived in time, which express benefactive scenes strictly via lexical meaning, are not instances of benefactive constructions. For example, English has at least three types of benefactive constructions.
English
(1) Mark bought Linda a book. (2) Mark bought a book for Linda. (3) I got to visit my grandmother last week.
3.â•… The dative case is also often used to indicate the notion of adversity, and neither the benefactive nor malefactive function can be considered primary; therefore, this kind of construction is called affectedness construction (Smith 2005; Radetzky & Smith, this volume), rather than simply benefactive. I will include some benefactive uses of the affectedness construction in the following discussions, whenever necessary, because the benefactive uses of the affectedness construction are essentially the same as explicit benefactives. For more discussion of affectedness constructions, see Smith (2005) and Radetzky & Smith (this volume). 4.â•… This notion of ‘good’ is essentially the same as the semantic primitive discussed by Wierzbicka (1988: 210–223). I use the term ‘good’ in my definition in order to bring together the different semantic types of benefactives. This study is different from Wierzbicka in that her analysis is limited to benefactives with some type of agent, whereas my analysis includes benefactive constructions without an agent.
 Tomoko Yamashita Smith
Sentence (1) is an example of the ditransitive construction, where the beneficiary is also the recipient of the book. Sentence (2) makes use of the preposition for, and the beneficiary is not necessarily the recipient of the book. There are many constructions in other languages that are similar to these two English constructions, and the notion of benefit can be expressed with some kind of special morphemes (e.g. adpositions, cases, and auxiliary verbs) or with special word orders, which contrast with basic transitive constructions. The third example should also be regarded as a benefactive construction, which I call the get to-construction (Smith 2005).5 This sentence expresses that the subject feels fortunate that s/he has visited the grandmother. All the native speakers of American English whom I have consulted agree that the construction indicates that the event described by the sentence is good for the subject. The full lexical verb get has multiple meanings with the central meaning ‘to obtain or receive’ (Gronemeyer 1999), and it has been extended to express the notion of obligation (have got to) and is used in the get passive construction. Likewise, it has been extended to express the notion of benefit. It should be noted here that although sentences with manage to and succeed in also similarly often indicate the notion of benefit, I consider these uses lexical because the original meanings of the verbs seem to be largely maintained. Japanese, on the other hand, has a variety of benefactive constructions. It uses three types of auxiliary verbs to describe benefactive scenes, as in (4) through (6). These auxiliary verbs are grammaticalized from lexical verbs of giving and receiving, and this kind of grammaticalization is also found in other languages (e.g. Matisoff 1991; Newman 1996; Yamada 1996, 2001, 2004; and Smith 1998b, 2005).
Japanese
(4)
maaku wa rinda ni hon o katte-age-ta Mark top Linda dat book acc buy-give-pst ‘Mark did Linda the favor of buying a book.’ ‘Mark bought Linda a book’
(5) naoko wa kei ni susi o tukutte-morat-ta Naoko top Kei dat sushi acc make-receive-pst ‘Naoko received the benefit of Kei making sushi for her.’ (6) ame ga hutte-kure-ta rain nom fall-give-pst ‘It rained (and I am thankful for that).’
5.â•… I thank Eve Sweetser for first drawing my attention to this construction.
Cross-linguistic categorization of benefactive constructions 
Sentence (4) is equivalent to the English ditransitive sentence (1), where the beneficiary, the dative argument, Linda, is also the recipient of the book. In Sentence (5), the beneficiary is the subject, who typically instigates the event, thus can be considered a kind of causative (Kuno 1987). Yet, its main function is to indicate that the subject benefits from the event. Sentence (6) has no agent, but the sentence describes that the speaker felt fortunate that it rained. The above examples demonstrate that the Japanese benefactive constructions are semantically less restricted than the English benefactives. It is important to note here that because the kureru benefactive construction expresses both agentive and agentless events, we need to include agentless benefactives in the discussion of benefactive constructions. Moreover, what may be unusual is that with all types of auxiliary verbs in Japanese, the beneficiary may also be the patient or the primary target of the event, as in the following examples.
Japanese
(7) kei wa naoko o homete-age-ta Kei top Naoko acc praise-give-pst ‘Kei did Naoko the favor of praising her (Naoko).’ (8) naoko wa kei ni homete-morat-ta Naoko top Kei acc praise-receive-pst ‘Naoko received the benefit of Kei praising her.’
As we can see in the above English and Japanese examples, the semantics of benefactive constructions are quite diverse, and I will classify these semantics based on their event types in the following section. Some of the benefactive constructions I discuss are not typical ones with an agent (e.g. passive sentences in Vietnamese); however, as their function is to denote the notion of benefit, I treat them as kinds of benefactive construction. I think that it is important to include such constructions in the discussion in order to fully understand the semantics/pragmatics of benefactive constructions per se.
2.â•… Different types of benefactive constructions 2.1â•… Two basic types I have found two basic types of benefactive construction in my data. One type always includes an agent, and always expresses that the agent intentionally carries out the act for the beneficiary; that is, the act is always presented as good for the beneficiary. I refer to this type as the “agentive benefactive” construction. The other covers more general benefactive events, and I call this type the “event benefactive” construction.
Tomoko Yamashita Smith
In my data, event benefactives are those that are used to express that the affecting event sometimes include the agent of the affecting event and sometimes does not. Note that I have not identified any single type of benefactive construction that is always agentless, i.e. it is more typical of benefactive events to include an agent.6 The most common semantic type of benefactive construction seems to be the agentive benefactive construction. I further divide this construction type into subtypes, which are different from Van Valin & LaPolla’s (1997) or Kittilä’s (2005). These subtypes also seem to be important with respect to the semantics of benefaction in general, because they represent the basic event types of benefactive situations, and they need to be clearly distinguished and discussed. In the following two sections, 2.2 will discuss the agentive benefactive construction and its subtypes, and 2.3 will present the event benefactive construction.
2.2â•… Agentive benefactive constructions Among the agentive benefactive constructions, there are various configurations in which the agent can be the same as or different from the beneficiary. In some languages, a benefactive construction allows the beneficiary to be anybody, including the agent him/herself, while in other languages the beneficiary must not be co-referential with the agent. Moreover, there are some languages that have a self-benefactive construction, where the agent carries out the action for his/her own benefit. (Note that cases in which the benefit is only pragmatic in nature, such as ‘I bought a car (for myself)’ are not included in the discussion.) Finally, I have found another type of construction that expresses a special kind of benefaction, ‘in addition to performing an action for his/her own benefit, one also performs it for the benefit of someone else’. The following sections illustrate and expound on these four types of constructions.
2.2.1â•… Unrestricted agentive benefactives: ‘X does something for someone (X/Y)’ It appears that the most common agentive benefactive construction type expresses a benefit to the agent him/herself or to another. There is no restriction on the beneÂ� ficiary, i.e. the beneficiary can be any animate entity (inanimate entities are usually dispreferred beneficiaries, see Kittilä & Zúñiga, this volume). I call this construction type unrestricted agentive benefactive. The distinction between self-benefactives and non-self-benefactives (see Sections 2.2.2 and 2.2.3) is not relevant here, and unrestricted benefactive constructions do not distinguish between them formally.
6.â•… I chose term event benefactive to include constructions that indicate either agentive or agentless benefaction. In this construction, there may be an agent who intentionally performs an action for the beneficiary, or there may be only an event somehow positively affects the beneficiary.
Cross-linguistic categorization of benefactive constructions
English has an unrestricted agentive benefactive construction with the preposition for, and many other languages such as French, Polish, Arabic, Thai, and Japanese have equivalent constructions, as shown in the following examples. French uses the preposition pour, and a Thai benefactive construction employs the give verb hây, which has been grammaticalized to introduce the beneficiary.7
English and French
(9) Do it for your family. Fais-le pour ta famille.
(Niedzielski 1981: 44)
(10) He works for himself. Il travaille pour lui-même.
(Niedzielski 1981: 44)
Thai (11) chán b`ffk thāaŋ hây kháo I tell way give he ‘I showed him the way.’ (12) chán rffŋ phleŋ hây tuuaeŋ I sing song give myself ‘I sang a song for myself.’
(Goda 1983)
(Bandhumedha Navavan, p.c.)
Japanese uses a periphrastic construction with a complex postposition to express unrestricted agentive benefactives:8 Japanese (13) kazoku no tame ni hatarai-ta family gen sake dat work-pst ‘(I) worked for my family.’ (14) zibun no tame ni hatarai-ta self gen sake dat work-pst ‘(I) worked for myself.’
7.â•… For more detailed discussions on the syntax and semantics of beneficiaries in some languages, see Niedzielski (1979: Polish and English), Niedzielski (1981: French and English), Jenny (2008, this volume: Thai). 8.â•… Notice the sentences do not have the benefactive auxiliary verbs, as in (13) and (14), and this periphrastic postposition can alone indicate the notion of benefit. In general, the beneficiary is not marked only with the dative, but with different postpositions, depending on the semantics of the verb. See Ôso (1983) for the acceptability of the dative marking in the benefactive constructions in Japanese.
 Tomoko Yamashita Smith
(15) tomodati no tame ni susi o tukutte-age-ta friend gen sake dat sushi acc make-give-pst ‘(I) made sushi for my friend.’
In addition to marking the beneficiary with for, English has another benefactive construction, which is ditransitive, as follows. (16) Mark wrote Linda a letter. (17) Mark wrote himself a letter.
This construction has more restrictions than the for-construction above in that it takes a limited number of verbs (see Goldberg 1995 for the discussion on partial productivity of the ditransitive construction). Compare the following sentences:
(18) (19) (20) (21)
Mark baked Linda a pie. Mark baked a pie for Linda. ??Mark sliced Linda a pie. Mark sliced a pie for Linda.
When both constructions are acceptable, the for-construction seems to be preferred for emphatic or stylistically marked statements (Niedzielski 1981). It should also be mentioned that the above benefactive constructions are often syntactically reflexive when the agent and the beneficiary are identical. In English, taking regular pronouns after for is marginally acceptable as in ?I work for me, and normally reflexive pronouns are used. On the other hand, in Mandarin both regular and reflexive pronouns are grammatical, and self-benefactive sentences with regular pronouns are more emphatic than ones with reflexive pronouns. Mandarin (22) woˇ gěi nıˇ zuò chaˇo fàn I for you make fried rice ‘I’ll make fried rice for you.’
(Li & Thompson 1974: 271)
(23) woˇ gěi zìjıˇ zuò chaˇo fàn I for self make fried rice ‘I’ll make fried rice for myself.’
(Su Yutang, p.c.)
(24) wŏ gĕi wŏ zuò chăo fàn I for I make fried rice ‘I’ll make fried rice for me.’
(Su Yutang, p.c.)
As expected and as the above examples indicate, there are some language-specific restrictions and variations on unrestricted agentive benefactive constructions.
2.2.2â•… Non-self-benefactives: ‘X does something for the benefit of Y’ In addition to the unrestricted benefactives discussed above, some languages have benefactive constructions whose semantics are more restricted and the beneficiary
Cross-linguistic categorization of benefactive constructions
cannot be the agent him/herself; that is the beneficiary needs to be someone other than the agent. I will call this construction type non-self-benefactive as opposed to the self-benefactive discussed in Section 2.2.3.9 For example, Indo-Aryan and Dravidian languages make a formal distinction between self-benefactives and nonself-benefactives (see Masica 1976 and Pardeshi 2001): the former does not include an extra participant as the beneficiary, while the latter includes an additional participant who benefits from the event described by the predicate. This section introduces some examples of the non-self-benefactive constructions in Marathi; the self-benefactive construction found in this language will be discussed in the following section. In Marathi both the give and show verbs have been grammaticalized into the non-self-benefactive auxiliary verbs (Pardeshi 1998). The show verb is used for situations involving audio-visual performance, such as singing a song or performing something for someone, as in (25) and (26); otherwise the give verb is used, as in (27) and€(28): Marathi (25) mī tyā-lā gāηe mhaη-un dākhaw-l-e 1s 3s-dat song.n sing-cp show-pst-n ‘I sang a song for him.’
(Pardeshi 1998: 148)
(26) mī tyā-lā nnakkal kar-un dākhaw-l-ī 1s 3s-dat mimicry.f do-cp show-pst-f ‘I performed mimicry for him.’
(Pardeshi 1998: 149)
(27) rām-ne sitā-lā pustak ghe-un di-l-e Ram-erg Sita-dat book.n take-cp give-pst-n ‘Ram bought Sita a book.’
(Pardeshi 1998: 146)
(28) rām-ne sitā-lā kholī zā2-un di-l-ī Ram-erg Sita-dat room.f sweep-cp give-pst-f ‘Ram swept/cleaned the room for Sita.’
(Pardeshi 1998: 146)
Indonesian employs the suffix -kan in its benefactive construction, where the beneficiary cannot be the agent him/herself and is coded as the primary object as in (29) and€(30).10
9.â•… The terms “self ” and “non-self ” are equivalent to egocentric and allocentric benefactives in Smith (2005). I have changed terms here, because the term egocentric may be misleading for some readers when referring only to the first person. Moreover, egocentric and allocentric are used to mean ‘human oriented’ and ‘non-human oriented’ to discuss special perspective in cognitive psychology (Li & Gleitman 2002). 10.â•… It is interesting that -kan is also used as a marker in causative constructions (Purwo 1997).
 Tomoko Yamashita Smith
Indonesian (29) dia membuat-kan saya kursi itu he make-ben I chair this ‘He made me this chair.’
(Shibatani 1996: 160)
(30) saya membunuh-kan ana lipas I kill-ben Ana centipede ‘I killed the centipede for Ana.’
(Shibatani 1996: 171)
Finally, there are two types of benefactive constructions in Japanese that are used to describe non-self-benefaction.11 The first type makes use of the auxiliary verbs of give, and the agent is coded as the subject, as in (31), while the second type makes use of the auxiliary verbs of receive, and the beneficiary is coded as the subject, as in (32).12 Japanese (31) watasi wa taroo ni hon o yonde-age-ta I top Taro dat book acc read-give-pst ‘I did Taro the favor of reading a book.’ (32) watasi wa taroo ni hon o yonde-morat-ta I top Taro dat book acc read-receive-pst ‘I received the favor of Taro reading a book for me.’
In order to describe self-benefaction, the more complex periphrastic postposition, no tame ni ‘for the sake of ’ is used alone, as in (33) (also in (14) in Section 2.2.1), and the auxiliary verbs cannot be used, with very few exceptions. Compare the following examples. Japanese (33) watasi wa zibun no tame ni hon o yon-da I top self gen sake dat book acc read-pst ‘I read a book for myself.’ (34) ???watasi wa zibun ni hon o yonde-age-ta I top self dat book acc read-give-pst ‘I did myself the favor of reading a book.’ (35) *watasi wa zibun ni hon o yonde-morat-ta I top self dat book acc read-receive-pst ‘I received (appreciated) the favor of me reading a book for myself.’
11.â•… Although there are three types of benefactive constructions in Japanese altogether, one of them belongs to another type, not the non-self-benefactive construction, and it will be discussed in Section 2.3.2. 12.â•… All three types of the benefactive construction have more than one auxiliary verb to serve various sociolinguistic factors, such as honorification and speech levels. See Kuno (1973) for a detailed description of Japanese benefactives.
Cross-linguistic categorization of benefactive constructions 
(36) ?watasi wa zibun o homete-age-tai I top self acc praise-give-des ‘I want to praise myself.’ (37) zibun ni yokuyatta to itte-yari-tai self dat did.a.great.job quo say-give-des ‘I want to say to myself that I did a great job.’ (38) ?watasizisin ni yurusite-morai-tai myself dat forgive-receive-des ‘I want myself to forgive me.’
As in (34) and (35), in general, neither kind of construction is used to express selfbenefaction. However, highly marked statements like (36), (37), and (38) may be possible. For example, (36) and (37) may be uttered when the speaker has a special achievement to praise him/herself for, and these sentences are acceptable to me, a native speaker of Japanese. Sentence (37) has another give verb, yaru, which means to give something to someone inferior, and (37) is more acceptable than (36). Notice that in (36) and (37) the verb takes the desiderative auxiliary tai. If the verb takes the simple past tense instead, then the sentences are less acceptable than (36) and (37). With receive verbs, exceptions such as (38) are harder to find than with the give verbs. This is perhaps because the receive constructions are often used to express “gratitude” to the agent, not just to describe a benefactive event. It seems unnecessary to express gratitude to oneself and perhaps inappropriate to do so in Japanese society.
2.2.3â•… Self-benefactive constructions This section deals with self-benefactive constructions, where the agent carries out an action for the benefit of him/herself. It is the co-reference of agent and beneficiary that distinguishes between self-benefactives and non-self-benefactives. As with nonself-benefactives, self-benefactives seem to be less common than unrestricted agentive benefactives. However, as mentioned earlier, they are quite prevalent among languages in India, such as Hindi, Marathi, and Tamil (Masica 1976; Pardeshi 2001), and also among Mongolic and Turkic languages (Anderson 1998, 2001, 2002, Creissels, this volume). The following are examples of the construction in Hindi and Marathi:13 Hindi (39) us-ne khat likh liyā he-erg letter.m write take.pst.m ‘He wrote the letter (for his own benefit).’
(Pardeshi 2001: 92)
13.â•… These constructions are also used to express the notion of adversity (Pardeshi 2001, Radetzky & Smith, this volume).
 Tomoko Yamashita Smith
Marathi (40) tyaa-na daa dhi kar-un ghet-l-i he-erg beard.f do-cp take-pst-f ‘He shaved his beard/He got his beard shaved.’
(Pardeshi 2001: 92)
It should be noted here that both Hindi and Marathi have non-self-benefactive constructions as well, and recall the Marathi examples given in the previous section. Analogies to these self-benefactive constructions in Indic languages are the uses of Indo-European middle voice expressing self-benefaction (see Kemmer 1993; Smith 2005). However, since middle voice covers a wider range of semantics, it should not be treated just as a benefactive. Similarly, Xakas, a Turkic language of Siberia also uses the verb stem al to indicate self-benefaction (Anderson 1998, 2002). The lexical source of the stem is a verb meaning ‘take’ or ‘get’. Xakas (41) min tayγada ččörčedıp, köp čistek teerıp al-γam I taiga.loc walk.prs.cv a.lot berry gather.cv sben-pst.i.1 While walking in the taiga, I gathered up a lot of berries (for myself)’ (Anderson 1998: 54)
The last examples are from Kewapi, the eastern dialect of the Kewa language in Papua New Guinea. Yarapea (1993) claims that Kewapi has person-number-tense suffixes that are also used to mark benefaction: self- and non-self-benefaction (see also Franklin 1971 for the Kewa language in general). Sentences (42), (44) and (46) are marked as self-benefactive, while (43), (45) and (47) are marked as non-self-benefactive:14 Kawapi (42) ne-me pea-lo 1s-erg do-1s.prs.sben ‘I am doing (something) for myself.’
(Yarapea 1993: 96)
(43) ne-me pea-ato 1s-erg do-1s.prs.nsben ‘I am doing (something) for someone.’
(Yarapea 1993: 96)
(44) naa enda na-ma 1p food eat-1p.nrpst.sben ‘We ate food.’
(Yarapea 1993: 99)
14.â•… Yarapea (1993) uses the terms egocentric and altrocentric benefaction instead of self- and non-self-benefaction.
Cross-linguistic categorization of benefactive constructions 
(45) naa-me enda na-arima 1p-erg food eat-1p.nrpst. nsben ‘We ate food on someone’s behalf.’
(Yarapea 1993: 99)
(46) ni enda abi ne-lua 1s food today eat-1s.fut.sben ‘I will eat food today for myself.’
(Yarapea 1993: 102)
(47) ni-me enda abi na-alua 1s-erg food today eat-1s.fut.nsben ‘I will eat food today on someone’s behalf.’
(Yarapea 1993: 102)
As the above examples show, the person-number-tense suffixes in Kewapi agree in person and number with the subject referent, indicate the tense of the verb, and mark benefaction. Although Yarapea suggests that it is economical to treat these suffixes as portmanteau morphemes, they can be separated into individual categories (Franklin 1971). There are many unclear cases; however, it appears that the self-benefaction is zero-marked and non-self-benefaction is marked with a (e.g. ma vs. a-rima for the near past-1st plural and lua and a-lua for the 1st singular future). If this is the case, then the self-benefactive should be considered as more basic. Interestingly, there is another contrast in the two types of benefaction, namely the subject of the non-self-benefactive is marked as ergative, while the subject of the self-benefactive is unmarked.
2.2.4â•… Shared-benefit construction: tseôm in Lai A further type of agentive benefactive construction is found in Lai, a Tibeto-Burman language, spoken in Western Burma15 (see Section 2.3.1 for a discussion of regular benefactive constructions often used to express non-self-benefaction in Lai; Smith 2005; Peterson 1998, 2007). First, self-benefaction is expressed with the middle marking, as in (46) (Smith 1998a). (45) vanhree niô vok ôa-ka-tsok-piak Vanhree erg pig 3sg-1s-buy-ben ‘Vanhree bought a pig for me.’ (46) vanhree niô vok ôaa-tsok Vanhree erg pig 3sg.middle-buy ‘Vanhree bought a pig for himself.’
15.â•… All of the Lai examples are elicited from Ken VanBik, unless its reference is specified in the example. I truly benefited from Ken’s help.
 Tomoko Yamashita Smith
In addition, Lai has another benefactive construction, and I call it the shared-benefit construction. It expresses ‘X does something for the benefit of X himself and Y’, as in the following examples: (47) tsewmaŋ niô they ôa-ka-loô-tsemô Tsewmang erg fruit 3s-1s-pick-ben ‘Tsewmang picked fruits for me (in addition to picking them for himself).’ (48) tsewmaŋ niô thil ôa-ka-tsook-tsemô Tsewmang erg thing 3s-1s-buy-ben ‘Tsewmang bought things for me (in addition to buying things for himself).’
In this construction, the subject referent is already performing the action denoted by the predicate. In addition to performing that action for his own benefit, he also performs it for the benefit of someone else. Though I have found only one other language that has a grammatical construction of this sort with phY ` ә in Thai (Jenny, this volume), the semantics of this construction seem to be quite clear and practical. This type of benefactive situation occurs frequently in daily life: doing laundry for oneself and for other members of family, fixing breakfast for oneself and for a guest, or making photocopies for oneself and for a classmate.
2.2.5â•… Discussion of agentive benefactive constructions As the data and the discussion above suggest, agentive benefactive constructions are quite common. Many languages possess unrestricted agentive benefactives in which the beneficiary can be either the agent him/herself or another person. There are also non-self- and self-benefactives, as shown in 2.2.2 and 2.2.3. In the former, the beneficiary cannot be the agent him/herself, while in the latter it must be the agent him/herself. Lastly, there are two languages that have a unique benefactive construction, where the agent performs an action not only for him/herself but also for someone else. As expected, languages vary according to the benefactive constructions they have. For example, Japanese has both unrestricted agentive benefactives and non-selfbenefactives, but no self-benefactive constructions, and English has only unrestricted benefactives. I have not found a single language with only self-benefactive constructions, which is not surprising, since we often do things to benefit others. Considering the data, I propose the following hierarchy to rank the likelihood of agentive benefactive construction types appearing in a given language. (49) unrestricted agentive benefactive and/or > self-benefactive > (other types) non-self-benefactive
The hierarchy indicates that if a language has an agentive benefactive construction, then it is most likely to be an unrestricted agentive or non-self-benefactive construction.
Cross-linguistic categorization of benefactive constructions 
Languages with an unrestricted agentive or non-self-benefactive construction may also have an additional one, namely a self-benefactive construction. This means that in order for a language to have a self-benefactive construction, it should also have another benefactive construction that indicates non-self-benefaction. More data is required to draw a firm conclusion concerning the other types. As discussed in the previous section, in addition to the shared benefit construction, Lai has a general benefactive construction that is often used to express non-self-benefaction, and its middle voice expresseses self-benefaction. Therefore, at present, I suggest that if a language has both self- and non-self-benefactive constructions (or some kind of benefactive construction that express both notions), then that language may also have another type of benefactive construction. In many unrestricted benefactive constructions, self-benefaction sentences are reflexive, and they can be regarded as more marked than non-self-benefaction sentences, as in I baked him a cake and I baked myself a cake. Thus, the real contrast here is between self- and non-self-benefaction. I suggest that the reason why nonself-benefactive constructions are more common in languages than self-benefactives is because, with non-self-benefaction, we have to state clearly ‘who does what for whom’. By contrast, with self-benefaction, such a statement is not necessarily required, perhaps because our actions are often done for our own benefit, such as eating and going to school. In many languages, such activities are often expressed without any benefactive marking. Consider English sentences, such as Mark ate dinner and Linda went to school in Mexico. Both sentences describe events that the subject referent benefits from. Moreover, in English, there are some verbs for which the beneficiary is unspecified with self- benefaction, whereas it is specified with non-self-benefaction, as in I bathed vs. I bathed Fluffy.
2.3â•… Event benefactive construction In Section 2.2., I discussed cases in which benefaction follows from an action by the agent. Event benefactive constructions, in turn, express the meaning ‘X benefits from Event Z’, where a certain event is good for the beneficiary. In contract to the agentive benefactives, the event does not have to include an agent.16 The semantics of the construction type thus differs from the constructions discussed in 2.2., where features of the beneficiary were more important. There are two types of event benefactives in my data. One type I refer to collectively as the give construction, in which the beneficiary
16.â•… Kittilä (this volume) discusses animacy of the agent in benefactive constructions in Finnish. Inanimate subjects or events (such as rain, fire and a car’s breaking down) are permitted in some (limited) contexts, and he treats these as inanimate agents (causes of the event). I, on the other hand, would treat such instances as event benefactives.
 Tomoko Yamashita Smith
is coded as direct object. It is used to express non-self-benefaction, but also agentless benefaction, and Lai and Japanese have such a construction. The other type is the get/receive construction, such as that of Vietnamese and English, and it expresses that the subject referent undergoes a pleasant experience.17
2.3.1â•… The Lai piak-construction Lai has a piak-construction that expresses not only non-self-benefaction, as in (50) and (51), but also agentless benefaction, as in (52), (53), and (54). For the latter, I have opted for considering this construction an event benefactive construction. (50) tsewmaŋ tsa-ôuk ka-tsook-piak Tsewmang letter-cover 1s-buy2-ben ‘I bought Tsewmang a book.’ (51) tsewmaŋ door ka-kal-piak Tsewmang market 1s-go-ben ‘I went to market for Tsewmang.’ (52) paŋpaar niô ôan-kan-paar-piak flowers erg 3p-1p-bloom-ben ‘The flowers bloomed for us.’ (53) ti niô ôa-kan-kiô-piak water erg 3s-1p-be cold-ben ‘The water is cold for us.’ (54) ka-leeŋ niô ôa-ka-roô-piak 1s-cart erg 3s-1s-break2-ben18 ‘My cart broke down for me (so that I can buy a new one).’
The lexical give verb in Lai is peek, and it is highly likely that piak has grammaticalized from peek. I suggest that the piak-construction originated as an agentive benefactive
17.â•… Some uses of the dative constructions (“dative of good fortune”, Wierzbicka1988) may convey that the dative referent is positively affected by an event; yet the event is understood as at least partly due to the beneficiary of the event. For example, a Polish sentence Ciasto ładnie mi się upeikło ‘The cake baked itself nicely to me.’ needs to be uttered by the person who baked the cake (212). Interestingly, the malefactive uses of the dative are quite common. See Wierzbicka (1988) for more discussion on the different semantics of the two types of constructions. 18.â•… Numerals in interlinear glosses indicate the transitive-intransitive alternations of the verb stems. The intransitive clause takes verb form1, while the transitive clause takes verb form 2.
Cross-linguistic categorization of benefactive constructions 
construction and later extended to express more general benefaction, namely agentless benefaction.
2.3.2â•… Japanese kureru-construction The Japanese kureru-construction is different from the other two types discussed earlier in that it also expresses agentless benefaction. Recall that the other two express only non-self-benefaction, as in (55)–(56). Sentence (55) is an example with a give verb, ageru, and sentence (56) is an example with a receive verb, morau: (55) watasi wa ken ni hon o yonde-age-ta I top Ken dat book acc read-give-pst ‘I did Ken the favor of reading a book.’ (56) watasi wa ken ni hon o yonde-morat-ta I top Ken dat book acc read-receive-pst ‘I received the favor of Ken reading a book for me.’
Kureru is also a give verb, but it is a counterpart of ageru, since the Japanese give verbs have a deictic opposition: incoming give (toward the speaker and/or his/her in-group) and non-incoming give (either giving that is away from the speaker and/or his/her in-group or giving between people who are neither the speaker nor among those in the speaker’s in-group).19 Thus, kureru expresses incoming give while ageru expresses non-incoming give. With non-incoming give verbs, the speaker (and/or his/her in-group) must be the agent as in (55) above and (57) below, while with the incoming give verbs, the speaker (and/or his/her in-group) must be the beneficiary as in (58). Compare the following examples: (57) watasi wa taroo ni susi o tukutte-age-ta I top Taro dat sushi acc make-give-pst ‘I did Taro the favor of making sushi.’ (58) taroo wa watasi ni susi o tukutte-kure-ta Taro top I dat sushi acc make-give-pst ‘Taro did the favor of making sushi for me.’ (59) *taroo wa watasi ni susi o tukutte-age-ta Taro top I dat sushi acc make-give-pst ‘Taro did the favor of making sushi for me.’
19.â•… There are many studies on the Japanese benefactive constructions (e.g. Kuno 1973; Masuoka 1981; Goda 1983; Shibatani 1996, 2000; Hidaka 2007). See also Kuno (1976) for the discussion of the giving and receiving verbs in terms of empathy perspective, and Shibatani (2003) for general discussion of the directional verbs, including giving and receiving verbs in Japanese.
 Tomoko Yamashita Smith
(60) *watasi wa taroo ni susi o tukutte-kure-ta I top Taro dat sushi acc make-give-pst ‘I did Taro the favor of making sushi.’
In present-day Japanese, there are two in-coming give verbs; kudasaru and kureru. Generally, the former is used when the agent is socially higher than the beneficiary; otherwise, the latter is used. The kudasaru benefactive expresses only non-selfbenefaction. The kureru benefactive, however, can be used to express not only nonself-benefaction, as in (58), but also benefactive events that do not include an agent, as the following examples indicate. These sentences express the gratitude of the speaker. In other words, the speaker is thankful for a certain event, such as ‘the stopping of rain’ in (61) and ‘children going to sleep’ in (61). It should be noted here that the auxiliary kureru may be omitted from these two sentences, and then the sentences are simply statements of the facts, as in (62) and (64). (61) ame ga yande-kure-ta rain nom stop-give-pst ‘It stopped raining (and I am thankful for that).’ (62) ame ga yan-da rain nom stop-pst ‘It stopped raining.’ (63) kodomo ga nete-kure-ta child nom sleep-give-pst ‘The kids fell asleep (finally and I am thankful for that).’ (64) kodomo ga ne-ta child nom sleep-pst ‘The kids fell asleep.’
The kureru agentless benefactive sentences are more restricted than the kureru agentive benefactive sentences in that the beneficiary is always the speaker,20 and the beneficiary cannot be overtly stated in a sentence, as in the following example. (65) *ame ga watasi ni/ no tame ni yande-kure-ta rain nom 1s dat/ gen sake dat stop-give-pst ‘It stopped raining (and I am thankful for that).’
The sentence expresses that the speaker is thankful that an event occurred.21 As in the case of Lai’s piak-construction, it seems that kureru was first grammaticalized to 20.â•… In some limited contexts, the hearer can be the beneficiary, as in a confirmation or question. 21.â•… There seems to be another explanation as to why the affectee must be the speaker. CharacÂ� teristically, the Japanese language does not allow the speaker to report the psychological
Cross-linguistic categorization of benefactive constructions 
express non-self-benefaction and later extended to express agentless benefaction.22 The fact that the lexical meaning of kureru is ‘give’ and its agentless benefactive uses are more restricted than the agentive benefactive uses supports this claim.
2.3.3â•… Vietnamese đu’o’c-construction ˙ Vietnamese has a benefactive construction that expresses that the subject referent undergoes a pleasant experience, using the benefactive verb đuoc (Clark 1974), as in the following examples. (68) ông đó đu’o’c mua sách ˙ gentleman that get buy book ‘He has the good fortune to be able to buy the books.’
(Clark 1974: 81)
(69) ông đó đu’o’c mát ˙ gentleman that get cold ‘That man has the good fortune of being cool.’
(Clark 1974: 82)
(70) Kim đu’o’c John khen ˙ Kim get John compliment ‘Kim was complimented by John’
(Siewierska 1984: 151)
(71) John đu’o’c Kim đánh ˙ John get Kim hit ‘John was hit by Kim (he liked it, it was good for him)’
(Siewierska 1984: 151)
Although the benefactive verb đuoc has maintained the grammatical status of a full lexical verb (Liêm 1969; Clark 1974) with the most basic meanings ‘receive, get, obtain’ (Thompson 1965), its uses have been extended to take verbal complements as in the above examples. Since the sentences (70) and (71) can be analyzed as kinds of passive
state of others in a direct form (e.g. Shibatani 1990; Hasegawa 1998). Compare the following examples. (66)
watasi wa uresii/samui (67) *tanaka-san wa uresii/samui I top glad/cold Tanaka-Mr. top glad/cold ‘I am glad/cold.’ ‘Mr. Tanaka is glad/cold.’
I suggest that the most important function of the kureru agentless benefactive is to express the beneficiary’s gratitude. Since gratitude should be considered as one’s psychological state, the beneficiary is limited to the speaker. Only the speaker can directly state his/her own feeling of gratitude. 22.â•… I am not including kudasaru in this account in order to make the point clear. More strictly speaking, both kureru and kudasaru may have acquired the non-self-benefactive use (distinguishing them on the grounds of social-relations), and further only kureru has acquired the agentless benefactive use.
 Tomoko Yamashita Smith
clauses, the đuoc construction has sometimes been treated as a real passive (Truitner 1972; Le 1976). Clark (1974), on the other hand, rejects such an analysis and suggests that the construction has a verb that requires a sentential complement as the underlying structure, and it undergoes equi-NP deletion. My analysis of the construction is that all four sentences above can simply be regarded as instances of benefactive constructions with the verb đuoc, since it would be difficult to treat sentences like (68) and (69) as passives.23 Moreover, this construction expresses both agentive and agentless benefaction. There is an agent in (70) and (71), who directly performs an action on the beneficiary, which is seen as good for the subject referent, and in (68) there is an agent who performs an action, which affects him/her positively. These are all instances of agentive benefaction. On the other hand, in Sentence (69), the verb is stative, and the subject is a mere experiencer, not an agent. In conclusion, the construction is best treated as an event benefactive construction.
2.3.4â•… English get-to-construction As noted earlier, I suggest that English uses the get-to-construction to express the notion of benefit.24 Kimball (1973) and Gronemeyer (1999) both use the term ‘permission’ for this use. Gronemeyer notes that “I use permission in the standard sense of deontic modality, meaning that the subject is permitted by an external circumstance or force the possibility or opportunity of doing something. Closely related are the senses ‘manage to’ and ‘secure opportunity’ which I take to be contextually determined variants” (p. 7). Her claim seems to be an accurate observation; however, the notion of benefit needs to be included in the semantics of get to. An important part of the semantics of this construction is that the subject is fortunate to experience the event described by the predicate. The events described by this construction sometime indicate that an agent has an opportunity to something s/he wanted/appreciated, as in (72) through (74), while the subject should not be considered as an agent, as in (75) through (78).
(72) (73) (74) (75) (76)
I got to drink anything I wanted at the party. I get to talk with my friends at lunch. We didn’t get to do anything fun lately. She gets to be her own boss. I got to be on TV, by chance.
23.â•… Vietnamese has another construction that has been treated as an adversative passive and denotes the notion of a negative affect on the subject referent (see Radetzky & Smith, this volume). 24.â•… See Smith (2005) for more discussion of this construction using corpus data.
Cross-linguistic categorization of benefactive constructions 
(77) He got to be popular in high school. (78) You get to feel cool, when you drive that car.
Notice that this construction also expresses adversity as sarcasm, as in (79) and (80); however, its primary use is as a benefactive. (79) I got to wash dished last three days. (80) Why do you get to have all the fun while I get to do all the worrying? (Gronemeyer 1999: 31)
Both Vietnamese and English make use of get/receive verbs in the above benefactive constructions. Since this type of grammaticalization has arisen from the lexical uses of get/receive verbs, the beneficiary is always coded as the subject, and the syntactic realization is different from give benefactive constructions. The get/ receive constructions expressing positive experience (‘have/had the opportunity to’) are also found in other Southeast Asian languages, such as Thai, Lahu, Yao (Mien) Samsao, and Khmer (Matisoff 1991), and it is highly likely that these constructions are also used to indicate agentless benefactive events. This linguistic area needs to be Â�further investigated.
3.â•… Discussion Previous sections have shown that there are two basic types of benefactives: the agentive benefactive, where the agent carries out an action for the benefit of the beneficiary, and the event benefactive where an agent may or may not be involved. The agentive benefactive has three basic subtypes: the unrestricted agentive benefactive, non-selfbenefactive, and self-benefactive. In addition, a few languages have shared-benefit constructions that express ‘X does something for the benefit of X and Y’. I summarize the restrictions on the agent and beneficiary in Table 1. As discussed earlier, each construction type expresses different event/situation types, and languages often possess more than one construction type. Table 1.╇ Restrictions on the agent and beneficiary in benefactives25 Type of construction
Is there an agent?
Who is the beneficiary?
Unrestricted agentive benefactive Non-self-benefactive Self-benefactive Shared-benefit construction Event benefactive
yes yes yes yes agentless is also OK
anyone not the agent agent agent and other(s) anyone25
.╅ In the case of the Japanese kureru-construction, it is the speaker only.
Tomoko Yamashita Smith
As has been pointed out, the most common benefactive type is the agentive benefactive, and this reflects the common assumption that good events usually result from intention, planning and the acts of people (Wierzbicka 1988). In addition, some languages have event benefactive constructions, where the agent may not necessarily be involved. Occasionally, one can experience good events without the intentional acts of people. We may recognize this kind of good event as an instance of rare luck. The English word windfall and the Japanese expression, tana kara botamoti (lit. ‘a rice cake falls from a shelf ’, with the meaning windfall), express this idea. Therefore, it might be expected that some languages have grammatical constructions that express good fortune, including agentless benefaction. Note that I have not found a single benefactive construction that is dedicated to expressing only agentless benefaction, and this supports my claim that agentless benefaction is uncommon. In my data, all the languages that have event benefactive constructions also possess one or more agentive benefactive constructions. I propose the following hierarchy to show the likelihood that benefactive construction types appear in a given language. (81) Agentive benefactives > Event benefactives
This hierarchy predicts that if a language has the Event benefactive construction, then it should also possess at least one agentive benefactive construction. For example, English has the get to-construction for the ‘X benefits from Event Z’ construction, so one or more agentive benefactive constructions should also exist, and they do: the for construction and the benefactive ditransitive construction. Japanese has the kureru benefactive, whose semantics include agentless benefaction, in addition to other agentive benefactive constructions. Note that there may be other types of agentless benefactive constructions to be found. Further study should shed light on our understanding of benefactive constructions in general. In addition, I would like to point out two phenomena which should be further investigated in the context of benefactive constructions. First, indicating deictic distinctions is an essential factor of benefactive constructions in some languages, as is the case in Japanese. Roughly speaking, Japanese makes a distinction between first person and non-first person (2nd and 3rd). Abelam, a Ndu language spoken in the Sepik district of New Guinea, has two benefactive markers that clearly contrast first person beneficiary, ty%, and non-first person beneficiary, kway, (Laycock 1965: 54). Lahu has the third person benefactive particle, pî (< ‘give’) and non-third person benefactive particle, lâ (< ‘come’) (Matisoff 1991). These phenomena need to be further studied together with other deictic distinctions of directional verbs in general (Tokunaga 1986; Shibatani 2003). The second phenomenon is that benefactives sometimes indicate the speaker’s attitude towards the event or one of its participants. For example, the Japanese kureru construction is used to indicate the speaker’s gratitude (i.e. the speaker is grateful that
Cross-linguistic categorization of benefactive constructions
someone did a favor for his/her in-group) (Tokunaga 1986; Smith 2005). The uses of the ethical dative are usually limited to the first and second person, and it expresses the speaker’s feeling toward the event, such as ‘surprise’, ‘joy’, and ‘displeasure’ (Janda 1993; Draye 1996; Van Hoecke 1996; Dąbrowska 1997). Moreover, utterances such as he gets to eat all the food on the table convey that the speaker is envious. There is still much to do in finding what sorts of attitudes are indicated in benefactive constructions in languages. Finally, a great deal of work remains to be done before a typological study of benefactives can be completed, yet I believe the system of classification I present in this study can facilitate future empirical work on benefactives and help to bring about a full typological study.
References Anderson, Gregory. 1998. Xakas [Languages of the World/Materials 251]. Munich: Lincom Europa. Anderson, Gregory. 2001. Subject version and object version in Tofa auxiliary verb constructions. Turkic Languages 5(1): 240–269. Anderson, Gregory. 2002. Auxiliary verb constructions in Old Turkic and Altai-Sayan Turkic. Paper presented at the Old Turkic Symposium. Frankfurt. Clark, Marybeth. 1974. Submissive verbs as adversatives in some Asian languages. In South-East Asian Linguistic Studies, Nguyên Đăng Liêm (ed.), 89–110. Canberra: The Australian National University. Dąbrowska, Ewa. 1997. Cognitive Semantics and the Polish Dative. Berlin: Mouton de Gruyter. Draye, Luk. 1996. The German dative. In The Dative, Vol. 1: Descriptive Studies [Case and Grammatical Relation across Language 1], William Van Belle & Willy Van Langendonck (eds.), 155–215. Amsterdam: John Benjamins. Fillmore, Charles. 1968. The case for case. In Universals in Linguistics theory, Emmon Bach & Robert T. Harms (eds.), 1–88. New York NY: Holt, Rinehart and Winston. Franklin, Karl. 1971. A Grammar of Kewa, New Guinea [Pacific linguistics C, 16]. Canberra: Australian National University. Goda, Sumire. 1983. Te-yaru, te-kureru, te-morau to taigo no hyôgen: Hâi no yôhô ni chûmoku shite. Nihongo kyôiku 49: 119–132. Goldberg, Adele. 1995. Constructions: A Construction Grammar Approach to Argument Structure. Chicago IL: University of Chicago Press. Gronemeyer, Claire. 1999. On deriving complex polysemy: The grammaticalization of get. English Language and Linguistics 3(1): 1–39. Hasegawa, Yoko. 1998. Linguistic systems and social models: A case study of Japanese. BLS 24:€117–128. Hidaka, Mizuho. 2007. Juyo dôshi no taishô hôgengaku teki kenkyû. Tokyô: Hitsuji Shôbo. Janda, Laura. 1993. A Geography of Case Semantics: The Czech Dative and the Russian Instrumental. Berlin: Mouton de Gruyter. Kemmer, Suzanne. 1993. The Middle Voice [Typological Studies in Language 23]. Amsterdam: John Benjamins.
 Tomoko Yamashita Smith Kimball, John P. 1973. Get. In Syntax and Semantics, Vol. 2: John Kimball (ed.), 205–215. New York NY: Seminar Press. Kittilä, Seppo. 2005. Recipient-prominence vs. beneficiary-prominence Linguistic Typology 9(2):€269–297. Kittilä, Seppo. 2006. On distinguishing between recipient and beneficiary in Finnish. In Grammar from the Human Perspective: Case, Space and Person in Finnish [Current Issues in Lingusitic Theory 277], Marja-Liisa Helasvuo & Lyle Campbell, 129–152. Amsterdam: John Benjamins. Kuno, Susumu. 1973. The Structure of the Japanese Language. Cambridge MA: The MIT Press. Kuno, Susumu. 1987. Functional Syntax: Anaphora, Discourse and Empathy. Chicago IL: University of Chicago Press. Laycock. 1965. The Ndu Language Family (Sepik District, New Guinea). Canberra: Australian National University. Le, Tam Duy. 1976. Vietnamese passives. CLS 12: 438–449. Li, Charles & Thompson, Sandra A. 1974. Co-verb in Mandarin Chinese: Verbs or prepositions? Journal of Chinese Linguistics 2: 257–278. Li, Peggy & Gleitman, Lila. 2002. Turning the tables: Language and spatial reasoning. Cognition 83: 265–294. Liêm, Nguyên Đăng. 1969. Vietnamese Grammar: A Combined Tagmemic and Transformational Approach [Pacific linguistics, Series C, 4]. Canberra: The Australian National University. Masica, Colin. 1976. Defining a Linguistic Area: South Asia. Chicago IL: University of Chicago Press. Masuoka, Takashi. 1981. Semantics of the benefactive constructions in Japanese. Descriptive and Applied Linguistics 14: 67–78. Matisoff, James. 1991. Areal and universal dimensions of grammatization in Lahu. In Approaches to Grammaticalization [Typological Studies in Language 19: 2], Elizabeth Closs Traugott & Bernd Heine (eds.), 383–453. Amsterdam: John Benjamins. Newman, John. 1996. Give. A Cognitive Linguistic Study. Berlin: Mouton de Gruyter. Niedzielski, Henry. 1979. Lexical realization of benefactive and beneficiary in Polish and English. Papers and Studies in Contrastive Linguistics 9: 165–180. Niedzielski, Henry. 1981. Lexical realizations of benefactive and beneficiary in English and French. Review of Applied Linguistics 53: 37–52. Ôso, Mieko. 1983. Judôbun to ni meishiku. Nihongokenkyû 50: 118–124. Pardeshi, Prashant. 1998. A contrastive study of benefactive constructions in Japanese and Marathi. Sekai no nihongo kyôiku 8: 141–165. Pardeshi, Prashant. 2001. Taking of the veil of the janus-faced ‘take’: The story of ‘take’ as a vector in Indic languages. Proceeding of the 123rd meeting of the Linguistic Society of Japan held at Kyûshû University. 92–97. Peterson, David. 1998. The morphosyntax of transitivization in Lai (Haka Chin). Linguistics of the Tibeto-Burman Area 21(1): 87–153. Peterson, David. 2007. Applicative Constructions. Oxford: OUP. Purwo, Bambang Kaswanti. 1997. The direct object in bi-transitive clauses in Indonesian. In Grammatical Relations: A Functionalist Perspective [Typological Studies in Language 35], T. Givón (ed.), 233–254. Amsterdam: John Benjamins. Siewierska, Anna. 1984. The Passive: A comparative Linguistic Analysis. London: Croom Helm. Shibatani, Masayoshi. 1990. The Languages of Japan. Cambridge: CUP.
Cross-linguistic categorization of benefactive constructions 
Shibatani, Masayoshi. 1996. Applicatives and benefactives: A cognitive account. In Grammatical Constructions: Their Form and Meaning, Masayoshi Shibatani & Sandra A. Thompson, 157–194. Oxford: Clarendon Press. Shibatani, Masayoshi. 2000. Japanese benefactive constructions: Their cognitive bases and autonomy. In Syntax and Functional Explorations: In Honor of Susumu Kuno. Tokyo: Kuroshio Publishers. Shibatani, Masayoshi. 2003. Directional verbs in Japanese. In Motion, Direction and Location in Languages: In Honor of Zygmunt Frajzyngier [Typological Studies in Language 56], Erin Shay & Uwe Seibert (eds.). Amsterdam: John Benjamins. Smith, Tomoko Yamashita. 1998a. The middle voice in Lai. Linguistics of the Tibeto-Burman Area 21(1): 1–52. Smith, Tomoko Yamashita. 1998b. How ‘give’ and ‘receive’ provide structure for more abstract notions: The case of benefactives, adversatives, causatives, and passives. BLS 24: 219–231. Smith, Tomoko Yamashita. 2005. Affectedness Constructions: How Languages Indicate Positive and Negative Events. Ph.D. dissertation, University of California at Berkeley. Tokunaga, Misato. 1986. Affective Deixis in Japanese: A Case Study of Directional Verbs. Ph.D. dissertation, University of Michigan. Truitner, Nga. 1972. Passive sentences in Vietnamese. CLS 8: 368–378. Van Hoecke, Willy. 1996. The Latin dative. In The Dative, Vol. 1: Descriptive Studies [Case and Grammatical Relation across Languages 1], William Van Belle & Willy Van Langendonck, 155–215. Amsterdam: John Benjamins. Van Valin, Robert D., Jr. & LaPolla, Randy J. 1997. Syntax: Structure, Meaning, and Function. Cambridge: CUP. Wierzbicka, Anna. 1988. The Semantics of Grammar [Studies in Language Companion Series 18]. Amsterdam: John Benjamins. Yamada, Toshihiro. 1996. Some universal features of benefactive constructions. Nihongakuhô 15: 27–45. Yamada, Toshihiro. 2001. Nihongo ni okeru benefactive no kijutsuteki kenkyû. Nihongogaku 20.1: 84–93. Yamada, Toshihiro. 2004. Nihongo no benefactive: Teyaru, tekureru, temorau no bunpo. Tokyo: Meiji shoin. Yarapea, Apoi. 1993. Kewapi verbal morphology and semantics. Language and Linguistics in Melanesia 24: 95–110.
An areal and cross-linguistic study of benefactive and malefactive constructions* Paula Radetzky & Tomoko Smith
National Tsing Hua University/Osaka University In this study, we compare languages of Europe with those in Asia, establishing a two-part typology of strategies that these languages use to express notions of benefactivity and malefactivity. Languages in Europe have a generalized and semantically underspecified affectedness construction as the most unmarked construction in their repertoire; this can be used in both benefactive and malefactive situations. On the other hand, in the vast majority of the languages of South Asia, mainland Southeast Asia, and East Asia, benefactive constructions are separate from malefactive constructions, and each one is semantically robust and restricted in its applicability to either fortunate or unfortunate situations. An important question to raise is the interrelatedness between this areal distribution and each area’s societal values and cultural practice.
*We would like to express our appreciation to the following people: Alexander Adelaar, Willem Adelaar, Alexandra Aikhenvald, E. Annamalai, Mathias Jenny, Terrence Kaufman, Andrew Keller, Ngoc Kim, Alan King, Seppo Kittilä, Kyung-Hwan Mo, Johanna Nichols, Weera Ostapirat, Prashant Pardeshi, Bijeta Singh, Udaya Narayana Singh, Iru Su, Karumuri Subbarao, Soteria Svorou, Eve Sweetser, Cheng-Yu Tsai, Wei-Tien Tsai, Ken Vanbik, Bertlinde Voegel, William F. Weigel, Fernando Zúñiga, and the anonymous reviewers. The abbreviations used in this paper are as follows 1 2 3 abl acc agt all aor ben caus compl cp dat decl
first person second person third person ablative accusative agent allative aorist benefactive causative completed action conjunctive dative declarative
erg f fut gen inf irr loc m mal n neg nom nonpst obl
ergative feminine future genitive infinitive irrealis locative masculine malefactive neuter negative nominative non-past oblique
pass pft pfv pl poss pres pst ptl pv s sd sub top
passive perfect perfective plural possessive present past particle verb particle singular sandhi subject topic
Paula Radetzky & Tomoko Smith
1.â•… Introduction As human beings, we can construe a particular event as being fortunate or unfortunate, and we have the option of overtly including this interpretation in an utterance. In this study, we compare languages of Europe with those in Asia, establishing a two-part typology of strategies that languages can use to express these notions of benefactivity and malefactivity. Our examples come from languages belonging to, among others, the Indo-European, Uralic, Nakh-Daghestanian, Dravidian, Austroasiatic, and SinoTibetan families.1 From a semantic perspective, languages can roughly speaking encode benefactives and malefactives in two different ways. In Option 1, benefactive and malefactive meanings are collapsed into one single superconstruction, the affectedness construction, which can include dative constructions and applicative constructions as subtypes;2 This distinction is formally motivated, the former includes a new argument in the dative (or similar) case, while in the latter, verb morphology is modified for making the introduction of a new argument possible. In the affectedness construction, the entity that is affected (the affectee) is underspecified as to whether it has been affected beneficially or maleficially, and it is the discourse context which must provide the information favoring one interpretation over the other.3
1.â•… Benefactive constructions are well-attested in the world’s languages, and there have been studies of them in individual languages as well as from a cross-linguistic perspective (e.g., Niedzielski 1979; Masuoka 1981; Shibatani 1994; Goldberg 1995; Miyake 1996; Newman 1996; Yamada 1996; Van Valin & LaPolla 1997; Pardeshi 1998; Yamada 2001; and Kittilä 2005). Malefactive constructions have not been studied as much, and it is unusual to find them discussed in the literature; exceptions are Menn (1972), Clark (1974), Chappell (1986), Davies (1993), Malchukov (1993), and Davies (1995). 2.â•… “Dative” here is in the sense of the Latin ‘dative of advantage and disadvantage’ (dativus commodi et incommodi). See also Dąbrowska’s “beneficiary dative” (1997: 35) and Janda’s “dative of beneficiary” (1993: 90). 3.â•… The term affectedness has been used not just in discussions of benefactivity and malefactivity but also in discussions of patients and undergoers (e.g., Foley & Van Valin 1984; Klaiman 1988). Likewise, the term affectee has been used by linguists to refer to undergoertype arguments, including non-agentive arguments (e.g., the ‘I’ in I broke my leg; Washio 1993; Takehisa 1999; Takehisa 2000), and patients in causative sentences (e.g., ‘vase’ in The blast caused the vase to break; Wolff, Song & Driscoll 2002). In this paper, however, we consider only the cases where the notions of benefactivity or malefactivity are explicit and encoded grammatically. The term affectedness construction was coined by Smith (2005), while Chung (1993) used the term affected construction. Most linguistic studies have treated this kind of generalized construction as a type of dative construction or applicative construction,
Areal and cross-linguistic study
In Option 2, one construction (of perhaps several in a language) inherently encodes a particular kind of affectedness, be it benefactive or malefactive. In other words, a discourse context is not needed to determine whether the event is benefactive or malefactive. We shall use the term benefactive construction to refer to constructions used in encoding situations as being fortunate (and call their affected arguments the benefactees); and we shall use the term malefactive construction to refer to the constructions used in encoding situations as being unfortunate (and call their affected arguments the malefactees).4 In this option, the benefactive and malefactive constructions differ from each other morphologically and/or syntactically. The benefactive constructions usually employ verbs meaning ‘give’, ‘receive’, or ‘take’, and malefactive constructions often use verbs such as ‘undergo’ or ‘do bad things’. If we look at the world’s languages, some tend toward using Option 1 strategies, and others tend toward using Option 2 strategies. Although languages which use the first option also use the second, in these languages, the Option 1 strategies are much more grammaticalized, while their Option 2 strategies are usually periphrastic and used in more marked situations of information packaging (such as focusing) and pragmatics (such as politeness). For example, Spanish, like other Romance languages, uses Option 1 as its favored strategy. In addition, it has complex prepositions which encode benefactees or malefactees, such as en beneficio de ‘for the benefit of ’ and en perjuicio de ‘to the detriment of ’, used in formal contexts. Many, however, would argue that these expressions are lexical rather than grammatical. Interestingly, among languages which favor Option 2, it is exceedingly unusual to find completely or genuinely underspecified affectee constructions. In other words, while languages which use Option 1 have periphrastic or ungrammaticalized ways of distinguishing between benefactives and malefactives, languages which favor Option 2 rarely have semantically underspecified affectedness constructions. Occasionally, single morphemes in these languages can be used to mark both benefactivity and malefactivity; however, in these instances, the benefactive meaning is primary and the malefactive meaning is only an extension. This is unlike Option 1, where the general affectedness construction is semantically so underspecified as to make neither meaning basic.5
depending on the grammatical characteristics of the language. For a detailed semantic analysis of these constructions, see Wierzbicka (1988: 210–23). 4.â•… The terms malefactive and adversative, as well as some others (including antibenefactive, detrimental, and pejorative) have been used in the literature. We have chosen the term malefactive and use it to include unfortunate (i.e., adversative) consequences generally, not merely those resulting from malice. However, if a particular term is widely established (e.g., the Japanese adversative passive), we will keep that term. 5.â•… English and Finnish are exceptions: they have extremely abstract but separate morphemes for benefactive and malefactive meanings. We discuss this below.
 Paula Radetzky & Tomoko Smith
Looking at the languages in Europe and Asia, those which primarily employ Option 1 are in Europe, while those which primarily employ Option 2 are found in Asia. In our consideration of affectedness, benefactive, and malefactive constructions, we include constructions which can be analyzed as non-argument-adding, i.e., where the affected participant is realized as an agent or a patient. For example, in the Korean malefactive passive construction with tangha ‘be subject to’, the malefactee, who is the patient, is encoded as a subject: (1) pholo-ka henpieng-eykey kwutha-tangha-ess-ta P.O.W.-sub M.P.-agt beat-pass-pst-decl ‘The prisoner of war was beaten (was subject to a beating) by an M.P.’ (Lee 1974: 151)
There are several reasons why it is advantageous to regard non-argument-adding constructions as affectedness, benefactive, and malefactive constructions. First of all, these constructions are semantically benefactive and malefactive, and we see no reason to exclude these constructions based solely on the fact that they are not argument-adding. In a second, related vein, in some languages, including Japanese and Mandarin, verb valency is not as grammaticalized as in other languages (e.g., Ono€& Thompson 1997/2000), so it is more difficult to determine whether a clause has an added argument or not. It seems disadvantageous to exclude these cases, which are semantically benefactive or malefactive but ambiguous as to argument structure. A third important reason for including these cases is related to their historical evolution. It should be pointed out that the Option 2 constructions overwhelmingly come from clause-chaining or verb-chaining constructions (cf. Foley & Olsen 1985; DeLancey 1991; Hopper & Traugott 1993; Lord 1993; among many others). For example, in (2), we see a bi-clausal sentence which contains two events, a ‘make’ event and a ‘give’ event. There is no argument added to either of these verbs, and it is not an instantiation of a benefactive construction. In (3), however, age- is no longer a lexical verb but an auxiliary. (These differences are marked via the distinct glosses of age-ta in each of the sentences.) (2) susi-o tukutte Taroo-ni age-ta sushi-acc make Taroo-dat give-pst ‘I made sushi and gave it to Taroo.’ (2-clause, non-argument-adding) (3) Taroo-ni susi-o tukutte age-ta Taroo-dat sushi-acc make give-pst ‘I made sushi for Taroo.’ (1-clause, argument-adding)
At the earlier stage in (2), there are two verbs, each with its own argument(s). However, as the morpheme meaning ‘give’ becomes grammaticalized—more auxiliary-like and bound to the main verb—the construction becomes argument-adding. In order to
Areal and cross-linguistic study 
capture this insight into the history of these constructions, it is important to include in our consideration the original semantically benefactive and malefactive constructions which are not strictly argument-adding. We have just discussed the case in which a non-argument-adding construction becomes an argument-adding one over time. There is also a reverse path of development – from an affectedness construction that is syntactically argument-adding to a semantically affectedness construction which is not argument-adding. In this scenario, after the benefactive or malefactive argument-adding morpheme develops (such as age- ‘give’ in (3)), its semantics (benefactivity or malefactivity) gain primacy, such that it obscures the original argument-adding requirement of the construction, which is then lost. This is the historical trajectory from sentence (3) to (4), which has no added argument. (4) iti-zikan mo Taroo-o matte age-ta one-hour even Taroo-acc wait give-pst ‘I waited an entire hour for Taroo (for his benefit).’ (1-clause, non-argument-adding)
Again, because of this historical continuum, it is important to include semantically (but not necessarily syntactically) affective constructions in our consideration. Finally, there is a more language-specific reason to include the non-argumentadding constructions under the rubric of affectedness, benefactive, and malefactive constructions. This evidence comes from Japanese, where there is a passive morpheme -(r)are, which can be used in both pure (i.e., non-argument-adding) passives and in adversative (i.e., argument-adding) passives.6 Given that these are homophonous, and given that the pure Japanese passive is used overwhelmingly in malefactive situations (Yoshida 1996), it is artificial to include the adversative passive -(r)are and exclude the pure (but usually semantically adversative) -(r)are. We also cannot ignore the fact that these two morphemes are historically related and that there are most certainly other languages in the world where this same grammaticalization has occurred. The structure of our paper is as follows. In Section 2 below, we examine in detail data from languages which primarily follow the Option 1 pattern, where the benefactive and malefactive meanings are expressed by one semantically underspecified affectedness construction. Our examples here are from Europe. In Section 3, we will do the same for languages which primarily use the Option 2 strategy, taking examples from the languages of India, Sino-Tibetan languages, other languages of Southeast Asia, and languages in East Asia. Section 4 will be our conclusion.
6.â•… There has been a large amount of literature on this. See, among others, Kuroda (1965), Shibatani (1972), Kuno (1973), Howard & Niyekawa-Howard (1976), Shibatani (1978), Teramura (1982), Klaiman (1987), Radetzky & Smith (1997/2000), and Tsuboi (this volume).
 Paula Radetzky & Tomoko Smith
2.â•… Option 1 2.1â•… Indo-European languages Let us first examine two Germanic languages which use a generalized, semantically underspecified affectee construction. In (5) below (‘Heinz repairs the car for me’), the dative encodes a benefactee. In (6), the dative encodes malefactees. German benefactive (5) Heinz repariert mir das Auto Heinz repairs I.dat the car ‘Heinz repairs the car for me.’
(Ogawa 1997: 2)
German malefactive (6) a.
man hat ihm das Haus angezündet they have he.dat the house set.fire.to ‘They set fire to the house on him.’
(Ogawa 1997: 2)
b. das Haus ist ihm abgebrannt the house is he.dat burned.down ‘The house burned down on him.’
(Ogawa 1997: 3)
This is the same case in Icelandic: in (7) and (8), the affectees (benefactees and malefactees) are encoded as datives. Icelandic benefactive (7) a.
þetta tæki auðveldar okkur störfin this.nom tool.nom facilitates us.dat the.jobs.acc ‘This tool makes the jobs easier for us.’
b. þetta sparar fólkinu mikla vinnu this.nom saves the.people.dat much work.acc ‘This saves the people a lot of work.’
(Jónsson 2000: 79)
(Jónsson 2000: 79)
Icelandic malefactive (8) myrkrið torveldaði þeim leitina the.dark.nom made.difficult them.dat the.search.acc ‘The dark made the search difficult for them.’
(Jónsson 2000: 79)
Next, we see that Latin also uses Option 1. For example, the first Latin sentence below (‘he praised my brother’) is neutral, in that there is no benefactee or malefactee that is overtly expressed. In the near-minimal pair in (10), however, an affectee is included in the shape of a first person dative singular pronoun, the resultant sense being ‘out of regard for me’. Latin neutral (9) laudāvit frātrem meum praise.pft.3s brother.s.acc 1s.poss.m.acc ‘He praised my brother.’
(Greenough et al. 1903: 235)
Areal and cross-linguistic study 
Latin benefactive (10) laudāvit mihi frātrem praise.pft.3s 1s.dat brother.s.acc ‘He praised my brother (out of regard for me).’
(Greenough et al. 1903: 234)
As can be surmised, Latin encodes malefactees as datives as well. In (11), the sense is that progress is hindered to the Carthaginians’ disadvantage (‘Carthaginian.pl.dat’): Latin malefactive (11) iter Poenīs vel corporibus suīs obstruere march Carthaginian.pl.dat even body.pl.abl 3pl.abl block.inf ‘to block the march of the Carthaginians even with their own bodies (to [the Carthaginians’] disadvantage)’ (Catullus M. 75, in Greenough et al. 1903: 235)
The same situation as in Latin holds for French. For example, in (12) below, the benefactee appears as the dative pronoun lui. In (13), however, lui encodes a malefactee. French benefactive (12) elle lui tricote un pull she.nom he.dat knits a sweater ‘She is knitting a sweater for him.’
(Melis 1996: 46)
French malefactive (13) il lui a fermé la porte he.nom he.dat pft shut the door ‘He shut the door on him.’
(Melis 1996: 46)
The Slavic languages Polish and Czech primarily employ Option 1 as well: Polish benefactive (14)
bielizna juz mi wyschła washing already to.me dried ‘Washing already to-me got dry.’ (Wierzbicka 1988: 212) ‘The washing has already dried (fortunately, to my benefit).’
Polish malefactive (15) a.
bielizna mi zmokła washing to.me got.wet ‘Washing to-me got wet.’ ‘The washing got wet (unfortunately for me).’
b. matka mi umarła mother to.me died ‘Mother to-me died.’ ‘My mother died on me.’
(Wierzbicka 1988: 211)
(Wierzbicka 1988: 212)
 Paula Radetzky & Tomoko Smith
Czech benefactive (16) Ludmila mu uvařila kaši Ludmila.nom he.dat cooked kasha.acc ‘Ludmila cooked kasha for him.’
(Janda 1993: 57)
Czech malefactive (17) a.
zemřela mu matka died he.dat mother.nom ‘His mother died.’
(Janda 1993: 85)
b. dítě nám v noci pláče child.nom we.dat in night.loc cries ‘The child cries at night (and we are affected by this).’ (Soteria Svorou, p.c.)
In Modern Greek, nouns and pronouns are in the genitive case whether they are benefactees or malefactees: Modern Greek benefactive (18) mi mou ti stenoxorite tin klia (mou) not me.gen her.acc upset.2pl the.acc Clea.acc me.gen ‘For my sake [please], do not upset (my) Clea.’ (Holton et al. 1997: 194) Modern Greek malefactive (19) yati mas xalases tin parea why us.gen spoil the party ‘Why did you spoil the party for us?’
(Holton et al. 1997: 194)
2.2â•… Non-Indo-European languages of Europe The examples we have seen above are all taken from Indo-European languages. However, there are non-Indo-European languages in Europe which also exhibit this type of system. Let us examine a few such examples from Lezgian (Nakh-Daghestanian; Russia), Hungarian, and Basque. In the two examples below, Haspelmath shows us that Lezgian uses the dative case to mark benefactees as well as malefactees. Lezgian benefactive (20) za wa-z ada-n wiri simfoni-jar ja-da I.erg you-dat he-gen all symphony-pl play-fut ‘I will play all his symphonies for you.’ (Haspelmath 1993: 88) Lezgian malefactive (21) čna a q˜eq˜werag suna-di-z wuč-na q’wan? we.erg that poor Suna-obl-dat do.what-aor ptl ‘What did we do to that poor Suna?’ (Haspelmath 1993: 88)
Areal and cross-linguistic study 
Hungarian, too, has a single construction encompassing both benefactive and malefactive semantics. In their grammar of Hungarian, Benkő and Imre state, “… the use of the dative (like the Latin dativus commodi et incommodi) is widespread…” (1972: 105). Hungarian benefactive (22) a.
nek-i hasznos valami dat-3s useful something ‘Something is useful to him.’
b. kellemes valami valaki-nek pleasant something someone-dat ‘Something is pleasant to someone.’
(Benkő & Imre 1972: 105)
(Benkő & Imre 1972: 105)
Hungarian malefactive (23) a.
nek-i rossz valami dat-3s bad something ‘Something is bad to him.’
b. kellemetlen valami valaki-nek unpleasant something someone-dat ‘Something is unpleasant to someone.’
(Benkő & Imre 1972: 105)
(Benkő & Imre 1972: 105)
Basque is an important case among European languages: although, as a language isolate, it is unrelated to its neighbors, it too employs an affectedness construction as its preferred strategy for encoding benefactives and malefactives. Basque neutral (24) a.
ama hil da mother die(d) is ‘The mother has died.’
b. pastela jan du cake eat(en) has ‘S/he has eaten the cake.’
(Alan King, p.c.)
(Alan King, p.c.)
Basque malefactive (25) ama hil zai-o gizonari mother die(d) is.dat-3s man.dat ‘The man’s mother has died on him.’
(Alan King, p.c.)
Basque benefactive or malefactive (26) pastela jan di-t cake eat(en) has.dat-1s ‘S/he has eaten the cake for me.’/‘S/he has eaten the cake on me.’ (Alan King, p.c.)
 Paula Radetzky & Tomoko Smith
The use of the underspecified affectedness construction appears to be an areal phenomenon: we have found that the vast majority of European languages (in addition to those examined here) evince the Option 1 pattern, whatever their genetic affiliation.
3.â•… Option 2 In contrast to the languages seen above, which readily make use of an underspecified affectedness construction, languages of South Asia, mainland Southeast Asia, and East Asia overwhelmingly use separate benefactive and malefactive constructions. One of the probable reasons for this is that they need separate constructions for each meaning is that every construction is semantically much more concrete than those in the languages of Europe. For example, benefactive constructions often take a verb meaning ‘give’ or ‘receive’, while malefactives take verbs meaning ‘pity’, ‘hit’, ‘undergo’, or the like; in other words, there is a substantial difference between the semantics of these two groups of verbs.7 And, although some benefactive constructions can occasionally be used in malefactive situations, there are many restrictions on this kind of usage; that is to say, the benefactive meaning is the primary one, and any malefactive uses the construction has are by extension. This is unlike the European languages we saw above, where the affectedness constructions are devoid of particular benefactive or malefactive semantics and where these kinds of meanings need to be intuited from the discourse context. Below, we examine some specific examples, moving roughly in a west-to-east direction within the part of the world where they predominate: the non-Sino-Tibetan languages of South and then Southeast Asia, then Sino-Tibetan languages, and finally Korean and Japanese.
3.1â•… Languages of India In Hindi, a Central-Zone Indic language, denā ‘give’ is used for benefactive constructions. The verb ‘take’ is used for both self-benefactives and self-malefactives, but its primary use is as a self-benefactive (Prashant Pardeshi, p.c.). Hindi benefactive (27) ashok-ne sarita se [apne liye cāy banā dene ko] kahā Ashok-erg Sarita with for self tea make give to asked ‘Ashok asked Sarita to make tea for him.’ (Subbarao 2004: 8)
7.â•… For a discussion of serial verb constructions in the South Asian linguistic area, see Masica (1976).
Areal and cross-linguistic study 
Hindi neutral (28) us-ne khat likha ˙˙ he-erg letter(m) wrote ‘He wrote the letter.’
(Bijeta Singh, p.c.)
Hindi self-benefactive (29) us-ne khat likh liyā ˙˙ he-erg letter(m) write take.pst.m ‘He wrote the letter (for his own benefit).’
(Pardeshi 2001a: 92)
Hindi self-malefactive (30) us-ne padhpadhkar ānkhe kharāb kar lī ˙ ˙ he-erg excessive.reading eyes(f) bad do take.pst.f ‘He damaged his eyesight by excessive reading.’ (Pardeshi 2001a: 92)
The same situation as Hindi holds for Marathi, a Southern-Zone Indic language: ‘give’ is used for benefactives (Ex. 32, see also Pardeshi 1998) and ‘take’ is used for selfbenefactives (Ex. 33, see also Pardeshi 2001a: 92). (Note that the self-benefactive in (33) is a minimal pair with the neutral in (31).) ‘Take’ can also be used for self-malefactives, as seen in (34), but the benefactive meaning is the primary one (Prashant Pardeshi, p.c.). Marathi neutral (31) tyā-na donhi hātā-ni tond dzhāk-l-a ˙ he-erg both hands(m)-with face(n) cover-pft-n.sg ‘He covered his face with both hands.’
(Pardeshi 2001b: 102)
Marathi benefactive (32) rām-ne shām-lā kholī dzhād-un di-l-i ˙ Ram(m)-erg Sham(m)-dat room(f) sweep-cp give-pft-f.sg ‘Ram cleaned the room for Sham.’ (Prashant Pardeshi, p.c.) Marathi self-benefactive (33) tyā-na donhi hātā-ni tond dzhāk-un ghet-l-a ˙ he-erg both hands(m)-with face(n) cover-cp take-pft-n.sg ‘He covered his face with both hands.’ (Gokhle 1979: 297) Marathi self-malefactive (34) tyā-na abartsabar khā-un pōta bighad-av-un ghet-l-a ˙ ˙ he-erg this.and.that eat-cp stomach(n) upset-caus-cp take-pft-n.sg ‘He upset his stomach by eating this and that.’ (Pardeshi 2001a: 92)
Finally, in Maithili, an Eastern-Zone Indic language, lelak ‘take’ is used for selfbenefactives and self-malefactives; however, the latter is infrequent. ‘Give’, as expected, is used for non-self-benefactives (Prashant Pardeshi, p.c.).
 Paula Radetzky & Tomoko Smith
Maithili self-benefactive (35) rām apan kesh katabā lelak ˙ Ram self ’s hair cut take ‘Ram cut his hair.’
(Pardeshi 2001a: 92)
Maithili self-malefactive (36) rām galatīsã apan aũri kāti lelak ˙ Ram by.mistake self ’s finger cut take ‘Ram cut his finger by mistake.’
(Pardeshi 2001a: 93)
Now that we have seen that languages from three different branches of Indic all use semantically comparable verbs in benefactive, self-benefactive, and self-malefactive constructions, let us look at the Dravidian languages of India. In the Dravidian languages that we have investigated, the situation is the same as in the Indic languages. Below, we provide some examples from Telugu and Tamil. In Telugu and Tamil, ‘give’ is used for benefactives, while ‘buy’ and ‘take’ are used as self-benefactives and occasionally self-malefactives. As expected, however, their primary meanings are self-benefactive (Pardeshi 2001a: 93). Telugu self-benefactive (37) atanu (tana-kōsam) wanta cēsu-kon-tā-du ˙˙ ˙˙ ˙ he (self-for) cooking do-buy-pres-3s.m ‘He will cook food (for his own benefit).’
(Karumuri Subbarao, p.c.)
Telugu self-malefactive (38) pillawādu (porapātu-na) wēlu koruk-konn-ā-du ˙ ˙ ˙ child (mistake-by) finger bite-buy-pst-3s.m ‘The child bit his finger (by mistake).’ (Karumuri Subbarao, p.c.) Tamil benefactive _ (39) rājā kumār-ukku-k katav-ai-t tir_a-ntu kotu-tt-aan ˙ Raja Kumar-dat-sd door-acc-sd open-cp give-pst-3s.m ‘Raja opened the door for Kumar.’ (Heine & Kuteva 2002: 151) Tamil self-benefactive (40) rām (avan) kaikalai-k kazuvi-kontaan ˙˙ Ram self ’s hands-sd wash-take.pst ‘Ram washed his hands.’
(Pardeshi 2001a: 93)
Tamil self-malefactive (41) avan (marattil-iruntu kiizee vizuntu) kālai utaittu-kontaan ˙ ˙˙ he (tree-from down fall) leg break-take.pst ‘He broke his leg (by falling from a tree).’ (Pardeshi 2001a: 93)
Areal and cross-linguistic study 
3.2â•… Languages of Southeast Asia Next, we will discuss the situation in two non-Sino-Tibetan languages of Southeast Asia, namely Thai (Tai-Kadai) and Vietnamese (Austroasiatic). Thai has serial verb constructions in which the verbs hây ‘give’ and dâaj-ráb ‘receive’ are used for benefactives and thùuk (< ‘touch, hit on the point’) is used for malefactives.8 Thai benefactive (42) a.
chaˇn 'f¢fk thaaŋ hây khaˇw I tell way give he ‘I showed him the way.’
b. phoˇm 1ây-ráp chƒƒn I get-receive invite ‘I was invited.’ c.
phoˇm 1ây-ráp t7 ¡ ŋtâŋ pen pr6thaan I get-receive appoint be chairperson ‘I was appointed to be the chairperson.’
(Goda 1983)
(Noss 1964: 124)
(Weera Ostapirat, p.c.)
Thai malefactive (43) a.
sùk thùuk rót chon Suk hit car collide ‘Suk was hit by the car.’
b. sùô1aa thùuk kh6mooy khY ¥ n 'âan Sudaa hit thief enter house ‘Sudaa had her house entered into by a thief.’
(Siewierska 1984: 149)
(Siewierska 1984: 156)
In Vietnamese, benefactive constructions and malefactive constructions also employ serial verbs. Again, ‘receive, get’ (đu’o’c) is the verb used in the benefactive construc˙ tion, and bi ‘undergo’ is the verb used in the malefactive construction. For example: ˙ Vietnamese benefactive (44) a.
Kim đu’o’c John khen ˙ Kim receive John compliment ‘Kim was complimented by John.’
(Siewierska 1984: 151)
b. John đu’o’c moi ngu´o`’i chú ý ˙ ˙ John receive every person look at ‘John was looked at by everyone, it’s a good thing.’ (Wierzbicka 1988: 210)
8.â•… The malefactive use of thùuk is of a formal register (M. Jenny, p.c.). See also Jenny (this volume) for a discussion of Thai benefactives.
 Paula Radetzky & Tomoko Smith
Vietnamese malefactive (45) a.
Nam bi Nga đánh ˙ Nam undergo Nga beat ‘Nam was beaten by Nga.’
b. John bi moi ngu´o`’i chú ý ˙ ˙ John undergo every person look at ‘John was looked at by everyone, it’s a bad thing.’
(Siewierska 1984: 149)
(Wierzbicka 1988: 210)
3.3â•… Sino-Tibetan languages All of the Sino-Tibetan languages that we have examined in various areas (Nepal, Burma, Thailand, and China) have separate benefactive and malefactive constructions. Interestingly, some of these are by serial verb and some of them are by applicative, but they are nonetheless two separate constructions. In what follows below, we examine Lai (Tibeto-Burman; Burma), Lahu (Tibeto-Burman; northern Thailand and adjacent areas), Hakka (Sinitic; China), and Mandarin in turn. In Lai, the morpheme -piak (< peek ‘give’) is mainly used for benefactives, as can be seen in (46) below. The same morpheme can occasionally be used for malefactives, as in (47); however, this is a rare occurrence (K. Vanbik, p.c.). The main malefactive constructions of Lai are formed with -sual (< ‘do bad things’, ‘be bad’) (Smith 2005: 110) and -hnoô, as in (48) and (49). Lai benefactive (46) a.
tsewmaŋ-niô law ôa-ka-thloô-piak Tsewmang-erg field 3s-1s-weed-ben ‘Tsewmang weeded the field for me.’
b. tsewmaŋ door ka-kal-piak Tsewmang market 1s-go-ben ‘I went to the market for Tsewmang.’ c.
paŋpaar-niô ôan-kan-paar-piak flower-erg 3pl-1pl-bloom-ben ‘The flowers bloomed for us.’
(Smith 2005: 77)
(Smith 2005: 59)
(Smith 2005: 59)
Lai malefactive with -piak (47) tsewmaŋ ôa-vok ôan-thaô-piak Tsewmang 3s-pig 1pl-kill-mal ‘They killed Tsewmang’s pig on him.’
(Smith 2005: 77)
Lai malefactive with -sual (48) tsewmaŋ-niô ka-ke ôa-ka-lamô-sual Tsewmang-erg 1s-leg 3s-step.on-mal ‘Tsewmang stepped on my foot (and it is too bad).’
(Smith 2005: 111)
Areal and cross-linguistic study 
Lai malefactive with -hnoô (49) rul-niô ka-ôin-ôaô ôa-ka-luô-hnoô snake-erg 1s.poss-house-loc 3s-1s-enter-mal ‘A snake came into my house on me.’
(Smith 2005: 104)
In Lahu, a Tibeto-Burman language spoken in northern Thailand and adjacent areas, pî ‘give’ is used with third-person benefactees, while lâ ‘come’ is used with non-third-person benefactees. A separate construction with šē is used for malefactives. Lahu benefactive (third person) (50) a.
chf pî
b. phεˆ pî c.
‘chop for him/her/them’ ‘release him/her/them’
ph+¡ô-mā pî ‘confess to him/her/them’
(Matisoff 1991: 396)
Lahu benefactive (non-third person) (51) a.
chf lâ
b. phεˆ lâ c.
‘chop for me/us/you’ ‘release me/us/you’
ph+¡ô-mā lâ ‘confess to me/us/you’
(Matisoff 1991: 396)
Lahu malefactive (Matisoff 1973: 331) (52) a.
chu šē e l7 fat mal pv ptl ‘continuing, alas, to get fat’
b. qhε ` ô šē la tù ve yō be.chipped mal pv irr ptl ptl ‘It’s about to chip off (which is regrettable).’ c.
tê to ve jèô-nεˆô tí phε ` ô šē ò whole body gen earth-wet only become mal compl ‘Her whole body got covered with mud (and isn’t that too bad).’
Sinitic languages also primarily use Option 2: benefactives and malefactives have their separate constructions. For example, Mandarin has a construction with gěi ‘give’ which is almost always benefactive, and it has several other morphemes, such as bèi ‘put on, throw on’, which are used in malefactive contexts. Aside from bèi, malefactives in Mandarin can also be formed with ràng and jiào. Mandarin benefactive woˇ gěi nHˇ zuò chaˇo fàn I for you make fried rice ‘I’ll make fried rice for you.’
(Li & Thompson 1974: 271)
b. woˇ gěi zìjHˇ zuò chaˇo fàn I for self make fried rice ‘I’ll make fried rice for myself.’
(Iru Su, p.c.)
(53) a.
 Paula Radetzky & Tomoko Smith
Mandarin malefactive with gěi (highly restricted) (54) tā jūrán gěi woˇ paˇo-le he unexpectedly gei I run-pfv ‘He ran away on me unexpectedly.’
(Wei-Tien Tsai, p.c.)
Mandarin malefactive with bèi (55) wáng xiaˇo’er bèi guHˇzi daˇ-sHˇ-le Wang Xiao’er bei devil strike-die-pfv ‘Wang Xiao’er was killed by the Japanese.’
(Chappell 1986: 1040)
Hakka, also a Sinitic language, has a morpheme lau, which is generally used as a benefactive. However, Lai (2003) states, “[s]ometimes the benefactive scenario can be extended to … a context where the action performed by the subject can in fact bring some negative effect to the lau phrase” (365). Hakka benefactive (56) ngai qinlian lau ng zodo I try.best lau you do.finish ‘I’ll do my best (for you).’
(Lai 2003: 365)
Hakka malefactive (57) ng mo qincai lau ngai longongfa you neg reckless lau me talk.nonsense ‘Don’t say any nonsense to embarrass me.’
(Lai 2003: 365)
3.4â•… Korean and Japanese Finally, let us look at Korean and Japanese, both non-Sinitic languages of East Asia. They, too, use the Option 2 pattern: benefactives and malefactives are separate constructions. For example, in (58) and (59), we see that Korean cwu ‘give’ and pat ‘receive’ are used in benefactive and benefactive passive constructions, respectively. Korean benefactive with cwu ‘give’ (58) na-nun Hanako-eykey mun-ul yele-cwu-ess-ta I-top Hanako-dat door-acc open-give-pst-decl ‘I opened the door for Hanako.’
(Shibatani 1994: 44)
Korean benefactive passive with pat ‘receive’ (59) ku sensayng-un haksayng-tul-eykey conkyeng-pat-nun-ta the teacher-top student-pl-agt respect-pass-pres-decl ‘The teacher is respected by students.’ (Lee 1974: 151)
Areal and cross-linguistic study 
In the malefactive passive sentences in (60) and (61), we find constructions with the verbs tangha ‘to face, be subject to’ and mac ‘encounter’. Korean malefactive passive with tangha ‘be subject to’ (60) pholo-ka henpieng-eykey kwutha-tangha-ess-ta P.O.W.-sub M.P.-agt beat-pass-pst-decl ‘The prisoner of war was beaten (was subject to a beating) by an M.P.’ (Lee 1974: 151) Korean malefactive passive with mac ‘encounter’ (61) ce yeca-ka namphyen-eykey sopak-mac-ess-ta that woman-sub husband-agt abandon-pass-pst-decl ‘That woman was abandoned by her husband.’
(Lee 1974: 151)
Japanese, like Korean, has different strategies for encoding benefactive and malefactive semantics. Japanese has several verbs meaning ‘give’. Kure- is used for things or events that are given to the first person or his/her perceived in-group (i.e., incoming ‘give’); yar- is for things or events that are given to entities that are of lower status than the speaker (i.e., outgoing-downwards ‘give’, such as when feeding a pet); age- is for things or events that are given to another entity of equal or higher status (i.e., outgoing-level and outgoing-upwards ‘give’). Moraw- ‘ask for and receive’ is often used as a causative-benefactive, with the person doing the asking being the instigator (or causer) (Teramura 1982; Kuno 1987; Ishihara 1991).9
9.â•… Although kure-, yar-, and moraw- can occasionally be used in malefactive situations, they are highly constrained and marked (e.g., indicating sarcasm or, as a derived function of that, when challenging someone; see Yamamoto 2002 and references therein). (Japanese adversatives are standardly expressed with the adversative passive, -(r)are.) For example, moraw- can apparently only be used in sarcasm or in subordinate clauses with one particular main verb meaning ‘[I/We] would be in trouble if…’ (-te ha komaru). In other words, the negative meaning must be expressed explicitly (and is limited to basically one collocation); aside from sarcasm, morawalone cannot be used alone as a main verb with malefactive situations. Yar- is used more frequently than the other two verbs in malefactive situations, but this is because, as mentioned above, the lexical verb is used only for recipients who are of lower status (i.e., more contemptible) than oneself. Its usage in malefactive situations is still highly constrained and marked. The use of Korean cwuta ‘give’ for malefactive situations is even more highly marked than the Japanese use of ‘give’ and ‘receive’ verbs in such contexts. This is perhaps because Korean lacks a grammaticalized or lexical in-group/out-group distinction, and correlatively cwuta is the only verb meaning ‘give’. In other words, there is no separate verb for ‘give down to a person who is not in the speaker’s in-group and, moreover, who is more contemptible than him/her’, which more easily lends itself to malefactive meanings. Native speakers of Korean have an extremely difficult time interpreting sentences with cwuta as anything but benefactive (K.-H. Mo, p.c.).
 Paula Radetzky & Tomoko Smith
Japanese benefactive (62) a.
Taroo-wa watasi-ni susi-o tukutte-kure-ta Taroo-top I-dat sushi-acc make-give-pst ‘Taroo made me sushi.’
b. watasi-wa Taroo-ni hon-o yonde-age-ta I-top Taroo-dat book-acc read-give-pst ‘I read a book to Taroo.’ c.
watasi-wa Taroo-ni hana-o katte-moraw-ta I-top Taroo-dat flowers-acc buy-receive-pst ‘I (asked for and) received the favor of Taroo buying me flowers.’
Japanese adversative with ‘give’ (rarely found) (63) a.
Taroo-o izimete-yar-ta Taroo-acc pick.on-give-pst ‘I picked on Taroo (he was asking for it, he deserved it [in the nasty sense]).’
b. wasurute moraw-tewa komar-u forget receive-if have.difficulty-nonpst ‘It’ll be bad (for me) if you forget.’ Japanese adversative with passive -(r)are (64) a.
kinoo ame-ni hur-are-ta yesterday rain-dat fall-pass-pst ‘[We] got rained on yesterday.’
b. kodomo-ni hitobanzyuu nak-are-ta child-dat all.night cry-pass-pst ‘My child cried on me all night long.’
In this section, we have examined languages belonging to various language families of Asia. We can observe that using specific benefactive and malefactive constructions is a phenomenon that covers a geographically contiguous area from India to Korea and Japan. It is especially striking that the Indo-European languages of the Indian subcontinent also exhibit this pattern and not the pattern found among their European cousins.
4.â•… Conclusion The preceding pages have shown that there are two strategies a language can take to express the notions of benefactivity and malefactivity. Those languages in Europe have a generalized and semantically underspecified affectedness construction as the most unmarked construction in their repertoire; this can be used in both benefactive and malefactive situations. On the other hand, in the vast majority of the languages of South Asia, mainland Southeast Asia, and East Asia, any benefactive constructions are
Areal and cross-linguistic study 
separate from any malefactive constructions, and however many of these constructions a language may have, each one is semantically robust and to a great extent restricted in its applicability to either fortunate or unfortunate situations. These are significant findings which have not been described in the previous literature.10 We mentioned above that English and Finnish are exceptions within the European area, in that they have at least one abstract construction for encoding benefactive situations and a separate abstract construction for encoding malefactive situations. English uses a construction with for to encode benefactees (I sing for her) and a construction with on to encode malefactees (my bicycle broke down on me). Finnish makes the distinction with the allative case versus the ablative case (both of which can also code the opposite role, but the constructions below illustrate unmarked cases): Finnish benefactive (65) minä laula-n häne-lle 1s.nom sing.pres-1s 3s-all ‘I sing for (lit. to) him/her.’
(Seppo Kittilä, this volume)
Finnish malefactive (66) minu-lta hajo-si pyörä 1s-abl break-3s.pst bicycle.nom ‘My bicycle broke down on me.’
(Seppo Kittilä, p.c.)
Are English and Finnish more similar to other languages in the European area, or are they more similar to languages of Asia? In one sense, they are more like the Asian languages we have seen, since they have separate constructions for each of these notions. However, in a more important sense, they are akin to European languages. In Asian languages, as we have seen throughout, benefactive and malefactive constructions come from clause- or verb-chaining constructions. In other words, a main verb meaning ‘give’, ‘buy’, ‘hit’, or the like becomes an auxiliary verb (cf. (2)~(3)). These verbs are contentful and include benefactive or malefactive notions in their semantics. On the other hand, European languages use more abstract morphemes in their affectedness constructions. For example, all the way down from Proto-Indo-European, the dative simply marks a recipient or goal (and, by extension, an affectee), and the clause can be interpreted as being either benefactive or malefactive outside of its discourse context. In this sense, English and Finnish are more European-like: for, on, allative, and ablative are very abstract. These prepositions or cases did not grammaticalize from verbs with concrete semantics (or at least this development is not visible anymore);
10.â•… Indeed, Creissels goes so far as to say, “I have found no mention of languages having malefactive periphrases in which the verb acting as a valency operator would be different from the one found in the benefactive periphrasis” (this volume).
 Paula Radetzky & Tomoko Smith
instead, they already existed in their own right as spatial markers11 and later on became imbued with their benefactive and/or malefactive semantics, for example by metaphorical extension (e.g., something going away from you—ablative—is usually bad). Of course, many issues remain for further research. The first would be to build a more complete areal picture of these phenomena. In this article, we focused on the distribution of Option 1 and Option 2 strategies between Europe versus South Asia, mainland Southeast Asia, and East Asia. What is the situation in Africa, the Americas, Australia, and Oceania? Our preliminary research has revealed that in the last three areas, the great majority of the languages use Option 1. In the Americas, this most likely has to do with the fact there is almost no verb serialization (W. Adelaar, A. Aikhenvald, T. Kaufman, and J. Nichols, p.c.). In Oceania, although there is a great deal of serialization, curiously almost none of the languages use constructions with ‘give’ to express benefactive notions (Crowley 2002: 173). In the first region (i.e., Africa), Bantu languages use the applicative—an Option 1 strategy—to encode beneficiaries and maleficiaries, while non-Bantu languages present a mixed picture, with the co-existence of languages using serial verb constructions (especially in West Africa) and those having more applicative-type constructions. Unfortunately, however, given the paucity of grammatical descriptions, we may never be able to characterize as many of these languages as we would like to. There is often no reference to benefactives or malefactives even in existing grammatical works, and when mention is made, it is usually about the benefactive only; no information is given as to (a) whether the benefactive can also be used malefactively, or (b) whether there does or does not exist a separate malefactive construction in the language. We have seen that benefactive and malefactive constructions can encode finegrained (and overlapping) semantics, specifying to a greater or lesser degree what kind of speaker attitude is being encoded: gratitude, sympathy, regret, blame, disapproval, et cetera. Constructions can also encode whether the event occurs due to someone’s purposeful action (volitionally) or accidentally. A second area of research, then, would be a more detailed investigation of the precise meanings, grammaticalization paths, and semantic extensions behind each construction. A final and important question we would like to pose is about the reasons behind the areal distribution we have found. As mentioned, languages which use Option 2 are in large part restricted to a swath of Asia stretching from the Indian subcontinent to Korea and Japan. Why do we find this particular distribution? We believe that established historical linguistic approaches—grammaticalization theory, comparative linguistics, and language contact studies—may not alone be able to answer this question.
11.â•… The English for is etymologically from a spatio-temporal morpheme meaning ‘before’ (and is related to that word).
Areal and cross-linguistic study 
In addition to these traditional approaches, it is imperative that we examine the data from an ethnosyntactic angle: as N.J. Enfield put it in Ethnosyntax, “Grammar is thick with cultural meaning” (2002: 3). We suggest that a fruitful area for further inquiry would be the relationship between this linguistic typology on the one hand and societal values and cultural practice on the other.
References Benkő, Loránd & Imre, Samu (eds.) 1972. The Hungarian Language. The Hague: Mouton. Chappell, Hilary. 1986. Formal and colloquial adversative passives in Standard Chinese. Linguistics 24: 1025–1052. Chung, Taegoo. 1993. The Affected Construction in Korean and Japanese. In Japanese/Korean Linguistics 3: 154–170. Clark, Marybeth. 1974. Submissive verbs as adversatives in some Asian languages. In Southeast Asian Linguistic Studies, Đăng Liêm Nguyê˜n (ed.), 89–110. Canberra: Pacific Linguistics. Crowley, Terry. 2002. Serial Verbs in Oceanic: A Descriptive Typology. Oxford: OUP. Dąbrowska, Ewa. 1997. Cognitive Semantics and the Polish Dative. Berlin: Mouton. Davies, William D. 1993. Javanese adversatives, the 1-advancement exclusiveness law and mapping theory. BLS 19: 101–111. Davies, William D. 1995. Javanese adversatives, passives and mapping theory. Journal of Linguistics 31: 15–51. DeLancey, Scott. 1991. The origins of verb serialization in Modern Tibetan. Studies in Language 15: 1–23. Enfield, Nick. 2002. Ethnosyntax: Explorations in Grammar and Culture. Oxford: OUP. Foley, William A. & Olsen, M. 1985. Clausehood and verb serialization. In Grammar inside and outside the Clause, Johanna Nichols & Anthony C. Woodbury (eds.), 17–60. Cambridge: CUP. Foley, William A. & Van Valin, Robert D., Jr. 1984. Functional Syntax and Universal Grammar. Cambridge: CUP. Goda, Sumire. 1983. Te-yaru, te-kureru, te-morau to tai-go no hyôgen: Hâi no yôhô ni chûmoku shite (Te-yaru, te-kureru, te-morau in Japanese and their corresponding expressions in the Thai language: Focusing on the usage of hâi). Nihongo Kyôiku (Journal of Japanese Language Teaching) 49: 119–132. Gokhle, Arwind. 1979. Arvinda gokhle yanci kathā: nivdak kathāntsā prātinidhik sangraha ˙ (Stories by Arvind Gokhle: A collection of representative stories), 2nd edn. Pune: Continental Prakashan. Goldberg, Adele E. 1995. Constructions: A Construction Grammar Approach to Argument Structure. Chicago IL: University of Chicago Press. Greenough, James Bradstreet, Kittredge, G.L., Howard, A.A. & D’Ooge, Benjamin L. (eds.). 1903. Allen and Greenough’s New Latin Grammar for Schools and Colleges. Boston MA: The Athenaeum Press. Haspelmath, Martin. 1993. A Grammar of Lezgian. Berlin: Mouton de Gruyter. Heine, Bernd & Kuteva, Tania. 2002. World Lexicon of Grammaticalization. Cambridge: CUP.
 Paula Radetzky & Tomoko Smith Holton, David, Mackridge, Peter & Philippaki-Warburton, Irene. 1997. Greek: A Comprehensive Grammar of the Modern Language. London: Routledge. Hopper, Paul J. & Traugott, Elizabeth Closs. 1993. Grammaticalization. Cambridge: CUP. Howard, Irwin & Niyekawa-Howard, Agnes M. 1976. Passivization. In Syntax and Semantics 5: Japanese Generative Grammar, Masayoshi Shibatani (ed.), 201–37. New York NY: Academic Press. Ishihara, Yoshito. 1991. Hyôgen naiyô o jûshishita bunkei no teishutsu junjo: Te morau hyôgen o megutte (Constructions according to the semantic aspects—Some remarks on the use of te-morau). Nihongo Kyôiku (Journal of Japanese Language Teaching) 74: 86–97. Janda, Laura. 1993. A Geography of Case Semantics: The Czech Dative and the Russian Instrumental. Berlin: Mouton. Jónsson, Jóhannes. 2000. Case and double objects in Icelandic. Leeds Working Papers in Linguistics 8: 71–94. Kittilä, Seppo. 2005. Recipient-prominence vs. beneficiary-prominence. Linguistic Typology 9: 269–297. Klaiman, Miriam H. 1987. Aktionsart, semantics, and function in the Japanese ‘passive’. Studies in Language 11: 401–34. Klaiman, Miriam H. 1988. Affectedness and control: A typology of voice systems. In Passive and Voice [Typological Studies in Language 16], Masayoshi Shibatani (ed.), 25–83. Amsterdam: John Benjamins. Kuno, Susumu. 1973. The Structure of the Japanese Language. Cambridge MA: The MIT Press. Kuno, Susumu. 1987. Functional Syntax: Anaphora, Discourse and Empathy. Chicago IL: University of Chicago Press. Kuroda, Shige-Yuki. 1965. Generative Grammatical Studies in the Japanese Language. Ph.D. dissertation, MIT. Lai, Huei-Ling. 2003. Hakka lau constructions: A constructional approach. Language and Linguistics 4: 353–78. Lee, Chung-Min. 1974. Abstract Syntax and Korean with Reference to English. Seoul: Pan Korea Book Corp. Li, Charles & Thompson, Sandra A. 1974. Co-verbs in Mandarin Chinese: Verbs or prepositions? Journal of Chinese Linguistics 2: 257–78. Lord, Carol. 1993. Historical Change in Serial Verb Constructions [Typological Studies in Language 26]. Amsterdam: John Benjamins. Malchukov, Andrej. 1993. The syntax and semantics of adversative constructions in Even. Gengo Kenkyû 135: 1–36. Masica, Colin P. 1976. Defining a Linguistic Area: South Asia. Chicago IL: University of Chicago Press. Masuoka, Takashi. 1981. Semantics of the benefactive constructions in Japanese. Descriptive and Applied Linguistics 14: 67–78. Matisoff, James. 1973. The Grammar of Lahu. Berkeley CA: University of California Press. Matisoff, James. 1991. Areal and universal dimensions of grammatization in Lahu. In Approaches to Grammaticalization [Typological Studies in Language 19:2], Elizabeth Closs Traugott & Bernd Heine (eds.), 383–453. Amsterdam: John Benjamins. Melis, Ludo. 1996. The dative in Modern French. In The Dative, Vol. 1: Descriptive Studies [Case and Grammatical Relations across Languages 1], William van Belle & Willy van Langendonck (eds.), 39–72. Amsterdam: John Benjamins.
Areal and cross-linguistic study 
Menn, Lise. 1972. On me. Linguistic Inquiry 3: 228–33. Miyake, Tomohiro. 1996. On benefactive constructions in Japanese. Kokugogaku (Studies in the Japanese Language) 186: 104–91. Niedzielski, Henry. 1979. Lexical realization of benefactive and beneficiary in Polish and English. Papers and Studies in Contrastive Linguistics 9: 165–80. Newman, John. 1996. Give: A Cognitive Linguistic Study. Berlin: Mouton de Gruyter. Noss, Richard B. 1964. Thai Reference Grammar. Washington DC: US Foreign Service Institute. Ogawa, Akio. 1997. On the syntax and semantics of the German dative. In The Dative and Related Phenomena, Kazuto Matsumura & Hayasi Tooru (eds.), 1–30. Tokyo: Hituzi Syobô. Ono, Tsuyoshi & Thompson, Sandra A. 1997/2000. Deconstructing ‘zero anaphora’. BLS 23: 481–491. Pardeshi, Prashant. 1998. A contrastive study of benefactive constructions in Japanese and Marathi. Sekai no Nihongo Kyôiku (Japanese-Language Education Around the Globe) 8: 141–65. Pardeshi, Prashant. 2001a. Taking off the veil of the Janus-faced ‘take’: The story of ‘take’ as a vector in Indic languages. Proceedings of the 123rd meeting of the Linguistic Society of Japan, 92–7. Pardeshi, Prashant. 2001b. The compound verb in Marathi: Definitial issues and criteria for identification. Kobe Gengogaku Ronsô (Kobe Papers in Linguistics) 3: 94–111. Radetzky, Paula & Smith, Tomoko. 1997/2000. A unified account of the Japanese causative, moraw- benefactive, and passive constructions. BLS 23: 290–302. Shibatani, Masayoshi. 1972. Remarks on the controversy over the Japanese passive. Papers in Japanese Linguistics 1: 145–66. Shibatani, Masayoshi. 1978. Nihongo no bunseki (The analysis of Japanese). Tokyo: Taishukan Publishing Co. Shibatani, Masayoshi. 1994. Benefactive constructions: A Japanese-Korean comparative perspective. Japanese/Korean Linguistics 4: 39–74. Siewierska, Anna. 1984. The Passive: A Comparative Linguistic Analysis. London: Croom Helm. Smith, Tomoko Y. 2005. Affectedness Constructions: How Languages Indicate Positive and Negative Events. Ph.D. dissertation, University of California at Berkeley. Subbarao, Karumuri. 2004. The role of particles and clitics in disambiguation. Paper presented at the Syntax of the World’s Languages (SWL1) Conference, Leipzig, Germany. Takehisa, Tomokazu. 1999. A preliminary note on the causer/affectee ambiguity. Linguistic Research 16: 85–102. Takehisa, Tomokazu. 2000. On the verb condition for affectee. Linguistic Research 17: 31–45. Teramura, Hideo. 1982. Nihongo no shintakusu to imi I (Japanese syntax and meaning I). Tokyo: Kurosio Publishers. Van Valin, Robert D., Jr. & LaPolla, Randy J. 1997. Syntax: Structure, Meaning, and Function. Cambridge: CUP. Washio, Ryuichi. 1993. When causatives mean passive: A cross-linguistic perspective. Journal of East Asian Linguistics 2: 45–90. Wierzbicka, Anna. 1988. The Semantics of Grammar [Studies in Language Companion Series 18] Amsterdam: John Benjamins. Wolff, Phillip, Song, Grace & Driscoll, David. 2002. Models of causation and causal verbs. CLS 37-1: 607–22. Yamada, Toshihiro. 1996. Some universal features of benefactive constructions. Nihongakuhô (Journal of Japanese Studies) 15: 27–45.
 Paula Radetzky & Tomoko Smith Yamada, Toshihiro. 2001. Nihongo ni okeru benefakutibu no kijutsuteki kenkyû. Nihongogaku 20: 84–93. Yamamoto, Hiroko. 2002. -Te morau no kinô ni tsuite: -tekureru to taihi shite (A study of -te morau: Comparing with -te kureru). Journal of Nagoya Women’s University (Humanities & Social Science) 48: 263–276. Yoshida, Eri. 1996. A Discourse Study of Passives in Japanese Conversation. Ph.D. dissertation, University of California at Los Angeles.
The role of benefactives and related notions in the typology of purpose clauses Karsten Schmidtke-Bode
Friedrich Schiller University Jena Across the world’s languages, adverbial clauses of purpose provide an important constructional environment for the occurrence of benefactive and semantically related markers. Based on a recent cross-linguistic study of purpose clauses (Schmidtke-Bode 2009), the present paper provides a systematic survey of the variegated functions of benefactives in the encoding of purposive relations. It will be argued that the particular distribution of benefactive markers across different types of purpose clauses is well-motivated from an ‘integrative functional’ point of view, which combines cognitive, functional and diachronic explanations for linguistic structure.
1.â•… Introduction It has long been noted in the typological literature that adverbial clauses of purpose provide a common constructional environment for the occurrence of benefactive markers. Thompson and Longacre (1985: 186), for example, observe that dative case marking may occur in various slots of a purpose clause, notably as the very marker that lends the clause its purposive interpretation in the first place. However, the first comprehensive study devoted solely to the typology of purpose clauses (Schmidtke-Bode 2009) reveals that benefactive morphemes have a yet more interesting role to play in the expression of purposive relations across the world’s languages. In fact, they form part of a cluster of closely related semantic functions that turns out to show particular distributions over various structural types of purpose clauses. Benefactives surface as either primary or secondary gestalt features of purposive constructions; they share their territory with allative, recipient and more general dative markers and are historically related to these other functions in intricate ways; benefactive NP-arguments can substitute for entire purpose clauses and thus provide a compact, economical way of coding purposive situations in language; and while benefactives are very common in certain types of purpose clauses, they tend to be entirely
 Karsten Schmidtke-Bode
absent from others. The aim of the present paper is to provide a systematic overview of these phenomena, thereby contributing to our understanding of the diversified roles that benefactives play in grammar or, more precisely, in the typology of specific grammatical constructions. Before we proceed to the actual analysis, a few methodological and theoretical considerations need to be clarified. The paper builds directly on my study on purpose clause constructions in the world’s languages (Schmidtke-Bode 2009). Purpose clauses were defined there as complex sentence constructions which encode that one verbal situation, that of the matrix clause, is performed with the intention of bringing about another situation, that of the purpose clause. A complex sentence, in turn, was identified as a construction that expresses a specific (contingent) relationship between at least two situations in at least two clauses, with clauses being agreed upon as a universally applicable level of linguistic organisation (cf. Thompson & Couper-Kuhlen 2005; Van Valin & LaPolla 1997: 25–27). This functional definition ensured that different structural types of purpose clauses would enter into the sample and thus provide a sufficiently representative picture of the variation space that purpose clauses unfold across languages. With regard to the data presented here, I also draw on the original database, a controlled variety sample of 80 languages.1 Since most of the languages considered exhibit more than one syntactic means to express purposive relations, I extracted all distinct purpose clause constructions found across the sample. In keeping with typologically-informed versions of Construction Grammar (e.g. Croft 2001; Goldberg 2006), individual constructions were taken to be characterised by a specific constellation
1.â•… The original sample was established by the criteria of genealogical independence and areal dispersion of languages, trying to represent the number of languages chosen from each of the well-known macro-areas in proportion to the number of stocks in that area (cf. Nichols 2004). Of utmost importance in the sampling process were the quality of available material and the structural diversity of purpose clauses in the respective language family, resulting in a genuine variety (rather than a strict probability) sample. For a brief overview, I will list the languages of the original sample alphabetically: Abkhaz, Abun, Acehnese, Amele, Babungo, Barasano, Basque, Duuŋidjawu, Epena Pedee, Evenki, (Boumaa) Fijian, Georgian, (Modern) Greek, Hausa, Hdi, Hixkaryana, Hmong Njua, Hungarian, Imonda, Jamul Tiipay, Kana, Kannada, Kayardild, Kewa, Kiowa, Kobon, Kolyma Yukaghir, Korean, Koyra Chiini, Krongo, Lakhota, Lango, Lavukaleve, Lealao Chinantec, Lezgian, Ma’di, Mandarin Chinese, Maori, Mapudungun, Martuthunira, Meithei, (Chalcatongo) Mixtec, Nama (Khoekhoe), Ndyuka, Ngiyambaa, Nishnaabemwin, Nkore-Kiga, Noon, Nunggubuyu, Paumarí, Persian, Pirahã, Punjabi, Qiang, (Huallaga) Quechua, Sanumá, Semelai, Slave, Somali, Supyire, Tetun, Tibetan, Tiwi, Trumai, Tukang Besi, Tümpisa Shoshone, Turkish, Tzutujil, Ungarinjin, Ute, Wambaya, Wardaman, Warao, Wari’, West ↜ Greenlandic, Wolaytta, Yagua, Yaqui, Yidi\, Yimas.
Benefactives in purpose clauses 
of so-called ‘gestalt features’ that ensures the unique categorisation of the construction in the flow of speech. The resulting more specific corpus comprised a total of 218 purposive constructions, which will form the baseline for all quantitative statements made in this paper. In addition, however, the present study will also discuss material from languages and language families that were not included in the original sample. With regard to theoretical commitments, finally, the explanations to be offered for the specifics of benefactive marking in purpose clauses are couched in Croft’s (1995) framework of ‘integrative functionalism’. A basic assumption of this approach is that language structure emerges from the way linguistic symbols are used in discursive interaction (cf. also Hopper 1987), and that various external factors exert selection pressures, as it were, on the shape and combinatorial possibilities of these symbols. Such pressures arise, for example, from the need of mutual intelligibility in discourse (often resulting in a trade-off between economical and iconic motivations), from the human processing system (e.g. the influence of processing on the linear order of elements in constructions, e.g. Diessel 2001, or on the differential ease of relativization and extraction, e.g. Hawkins 2004), or from rhetorical considerations of informationpackaging in discourse. Furthermore, cognitively-oriented typologists emphasise that grammatical structure reflects the conceptualization of the experiences we wish to convey (e.g. Croft & Cruse 2004; Langacker 1999). On this view, fundamental cognitivepsychological processes such as categorisation, selection and adjustment of attention, schematization, and association are operative in the acquisition, mental representation and actual use of linguistic knowledge. Perhaps the most prominent example of these cognitive processes is the ability to conceptualise an abstract domain of experience in terms of a more concrete, well-developed knowledge structure. Such cross-domain mappings, known as conceptual metaphors in cognitive linguistics (e.g. Lakoff & Johnson 1980), typically arise when two domains of experience regularly co-occur, such that a “permanent neural connection” (Lakoff & Johnson 1999: 46) is established between them. This is reflected in the way we import the language associated with the source domain in order to apply it systematically to the target domain. As a result, principled patterns of lexical and grammatical polysemy emerge in language use and—through social and linguistic diffusion processes—become consolidated as conventional structures in the language. We will see in due course that benefaction and related notions form precisely such domains of experience that regularly correlate with the experience of purposeful activity and hence show up in the encoding of purpose across languages. In sum, the explanatory constructs applied in this paper do justice to the recognition by many typologists that there are “close correlations between universal preferences in structure with universal preferences in cognition and communication” (Bickel 2007:€240). Although such motivated structural patterns
 Karsten Schmidtke-Bode
are grounded in language use, it is crucially the mechanisms of diachronic change that lead incrementally to their conventionalised synchronic status. A truly ‘integrative functionalism’ thus requires us to pay due attention to the historical dimension inherent in current typological distributions. After all, “even though language function is often cited as an explanation for language structure, this must be interpreted as a shorthand for language function explaining diachrony which in turn explains structure” (Moravcsik 2007: 37–38). With these theoretical assumptions in mind, we are now in a position to begin to describe and motivate the role of benefactives in the typology of purpose clauses.
2.â•… Benefactives as gestalt features of purpose clauses The starting point for a construction-based typology of purpose clauses is to record the morphosyntactic properties that make up the unique ‘gestalt’ of each construction. In order for a clause to be interpreted purposively, it is usually (though by no means always) necessary that some symbolic unit in the clause is associated with the semantics of intentional action. Not surprisingly, most constructions in the sample exhibit a morpheme specifically dedicated to the expression of purpose, which we may thus call a purpose marker or ‘primary gestalt feature’. There is a fascinating diversity in the formal shape of such markers, ranging from well-known conjunctions and verbal affixes over purposive auxiliaries to participles and suprasegmental marking. In addition to such a primary purpose marker, however, many constructions come with another gestalt feature. If we classify these ‘secondary’ elements into recurrent semantico-syntactic categories, the following picture emerges (cf. Table 1). Most of these elements fall out fairly straightforwardly from the conceptual and syntactic peculiarities of purpose clauses. Nominalised and infinitival verb forms, for example, are indicative of a high degree of syntactic integration of the purpose clause into its associated matrix, which is well-motivated in functional terms (cf. Cristofaro 2003). Temporal and modal markers are perfectly compatible with the semantics of purpose: an intended (= desiderative, optative) outcome is necessarily posterior to the action that is meant to bring it about. At the same time, purposes are hypothetical in nature and there is no implication that the intention will actually be realised. This leaves us with a class of allative and dative markers. Admittedly, this class is comparatively small, but we should not dismiss its members as entirely arbitrary properties in a minority of purpose clauses. For one thing, allative and dative case markers do not show up in their canonical function of marking arguments in the clause; instead, they appear on the purposive verb itself, as in example (1) from Kannada (Dravidian: India).
Benefactives in purpose clauses 
Table 1.╇ Additional gestalt features in purpose clauses Semantico-syntactic type of feature
absolute frequency
relative frequency (in %)
nominalised verb form or masdar infinitive verb form future or sequential tense-aspect markers irrealis, hypothetical or potential mood (irr/hyp/pot) desiderative or optative mood (des/pot) intentional mood (int) subjunctive mood (subj) allative marker dative marker subordinator other no additional element
22 15 12 9 6 2 12 2 8 9 19 102
10.1 6.9 5.5 4.1 2.8 0.9 5.5 0.9 3.7 4.1 8.7 46.8
total
218
100.0
(1) Praka:∫ jarmanige [enjiniyaring o:duvudakka:gi] ho:gidda:ne. Prakash Germany.dat engineering study.msd.dat.for go.n:pst.prf.3sg.m ‘Prakash has gone to Germany to study engineering.’ (Sridhar 1990: 73)
As can be seen, the dative case is suffixed to a nominalised verb which in turn is headed by a postposition. An analogous example showing allative case marking on the predicate is found in Basque (isolate: Spain): (2) Zer ez dute egiten Espainiako euskualdunek╇ […] eskuara beren what not aux do.impf Spain.rel Basques.erg Basque their sor-mintzaia galtzerat ez uzteagatik? birth-language lose.noml.all not let.noml.det.purp ‘What will the Basques from Spain not do not to let the Basque language, their native tongue, die?’ (Hualde & Ortiz de Urbina 2003: 742)
It appears that whenever a (nominalised) predicate is case-marked in a purpose clause, it strongly tends to be the allative or dative case that is employed. As is well-known, nominalisations generally serve to construe a verbal action as a nominal entity, and we shall return below to the question of what motivates this particular construal in purpose clauses. For now, we can strengthen our argument against the arbitrariness of allative and dative marking in purpose constructions by recognising that purpose markers themselves are often identical or historically derived from allatives and/or
 Karsten Schmidtke-Bode
datives. A quantitative approach to this question reveals the following patterns of polysemy that purpose markers establish with the domains we are interested in (cf.€Table€2): Table 2.╇ Polysemy of purpose markers Purpose marker identical to…
absolute frequency
relative frequency (in %)
locative marker locative and allative marker allative marker allative and benefactive marker dative/benefactive marker dative/goal marker n.i. no marker of this domain
1 4 7 4 41 6 6 149
0.5 1.8 3.2 1.8 18.8 2.8 2.8 68.3
total
218
100.0
n.i. = no information on multifunctionality available.
In the following, I will exemplify and discuss some of the recurrent coding patterns from Table 2 (§2.1), before they will receive an explanation in terms of the integrativefunctional framework outlined above (§2.2).
2.1â•… Principled polysemy patterns One of the most well-known polysemies exists between allative and purpose. An illustrative example from our sample is found in Maori (Austronesian: New Zealand):
(3) E /
mai hither
haere move i
ana /
ahau 1 teetahi a.
ki to
te the
taaone town
[ki to/
te the
hoko barter
koha]. gift
‘I am going to town to buy a present.’
(Bauer 1993: 66)
The connecting line indicates the polysemy of the allative morpheme ki: whereas in the main clause it appears in its basic directional function (‘to town’), it also introduces the subordinate purpose clause. In this latter function, it usually occurs with the definite article, with which it might be said to have formed a complex purposive conjunction. English to in its most basic function is an allative preposition, but has also become a very frequent purpose marker. In a most recent publication, Rice & Kabata (2007) argue that the metaphorical extension to purpose is, in fact, the most widespread pattern of polysemy that allative markers develop: “The single most prevalent cohort
Benefactives in purpose clauses 
sense of an allative is to mark purpose” (Rice & Kabata 2007: 472). Although I fully agree with the authors’ conclusion, I would like to point out that their study is based on a rather idiosyncratic sample of 44 languages, which are not particularly well-balanced in terms of genealogical and geographical independence.2 In this sample, they found a striking overlap between purpose and allative in 25 out of 54 allative markers (46.3%). Not surprisingly, a more tightly controlled sample like the present one yields a still notable, yet more moderate relationship: recall from Table 1 that a direct association between allative and purpose is attested for only 15 out of 80 languages (18.8%). In my sample, therefore the association is even less likely to be due to sampling error. A closely related pattern is the conflation of locative, allative and purposive meanings into one formative. Tzutujil, a Mayan language of Guatemala, has two prepositions, pa(n) and ch(i), both of which govern infinitival purpose clauses. Example (4) shows pa in this function: (4) Ja nata7 b’enaq [pa tikoj chiij]. the my.father B3.has.gone purp plant cotton ‘My father has gone to plant cotton.’
(Dayley 1985: 380)
The more basic uses of pa include the spatial and directional prepositional meanings ‘in’, ‘on’ and ‘to’. Similarly, ch(i) has the locative meaning ‘at’ and the directional meaning ‘to’, along with the purposive ‘(in order) to’. An interesting pattern is found in the Chadic language Mina, where locative and allative parcel out the work to mark purpose clauses. While the locative preposition n is used as the primary purpose marker (i.e. like a conjunction), a so-called ‘goal extension’ suffix with chiefly allative meaning attaches to the verb as a secondary purpose marker: (5) Kwáykwáy à ndә` dáp nә` gr-á nook. hyena 3sg go only loc find-gl 1pl ‘Let the hyena go to find it for us.’ (Frajzyngier & Johnston 2005: 420)
Benefactive purpose markers, finally, form the quantitatively largest ‘cohort sense’ of purpose markers (cf. Table 2). At this point, we can see how the present investigation is truly complementary to Rice & Kabata’s study: while they found that purpose is the most important extension of allatives, in fact the “↜‘seed’ sense” or “tipping point […] licensing abstract usages in other domains” (Rice & Kabata 2007: 490), the opposite perspective reveals that purpose markers do not primarily overlap with allatives, but with benefactives. However, a closer look at the distribution also shows that some markers conflate all three senses. Therefore, it seems that benefactive purpose markers
2.â•… One may criticize, for instance, that their sample contains nine Indo-European languages, many of which are closely related, as well as three relatively similar Oceanic languages, and only very few languages from South America.
 Karsten Schmidtke-Bode
can cluster together with allatives or be independent of directed motion. In Abun (West Papuan: Indonesia), a benefactive preposition (6a) is used as a purposive conjunction (6b), but not for directed motion (6c): (6) a.
An fro nu yo wa men. 3sg prepare house det.indf for 1pl ‘She prepared a house for us.’
(Berry & Berry 1999: 205)
b. Marta tot su-gato gum do brer gwat ma Marta cut thing-rel name comp brer bring dir
[wa men git sugit mo]. for 1pl eat food on
‘Marta cut some brer (leaves) and brought them back for us to eat our food on.’ (Berry & Berry 1999: 205) c.
Án yo ma mo nu nde. 3pl neg come loc house neg ‘They did not come to the house.’
(Berry & Berry 1999: 66)
The Italian preposition per, similarly, covers benefactive and purposive, but not allative contexts. In (7), per introduces a purpose clause that is related to its matrix at the speech-act or interpersonal level (Hengeveld 1989): (7) Per dire la verità, non sono d’accordo. for say the truth not be.1sg agreeable ‘To be honest, I don’t agree.’
(Schwarze 1995: 300)
A more intricate example comes from Qiang (Tibeto-Burman: China). The formative -χua, commonly in combination with the adverbial suffix -Ži, represents one way of marking benefaction in Qiang (8a); at the same time, this complex marker has come to be used as an adverbial conjunction3 of both cause and purpose (8b): (8) a.
The: qa-χua-Ži sә gue-k. 3sg 1sg-ben-adv firewood chop-go ‘He went to chop wood for me.’
(LaPolla 2003: 240)
b. [The: stuaha t.hә-(s ŋuә)-χua-Ži, ] (qupu) 3sg food/rice eat-noml cop-for/because-adv 3sg
dzigu mep ˚ ˚˙ money look.for(earn)
‘In order to live, s/he seeks to earn more money.’
(LaPolla 2003: 240)
3.â•… Note that LaPolla (2003: 240) labels -χua-Ži a ‘postposition’ even though it appears to be a bound morpheme. He further argues that in order “to mark the purpose of an action, the postposition […] can be used […] after a clause, or a nominalised clause”, which effectively turns it into a conjunction. For this reason, I call it a conjunction here.
Benefactives in purpose clauses 
While thus in Abun, Italian and Qiang, allative marking is distinct from purpose and benefaction, speakers of other languages have conflated the three notions into one morpheme. Purpose clauses in Yagua (Peba-Yaguan: Peru) provide a case in point:
(9) a.
allative -jų = benefactive ray-jų 1sg-all ‘for me’
(Payne & Payne 1990: 379)
b. allative -ju = purposive
Jáásiy-janu-jų núúdya-jito˛˛o-jásiy. cut.grass-inf-all 1ex-arrive-pst.prox1 ‘We arrived there to cut grass.’
(Payne & Payne 1990: 274)
In sum, these recurrent systematic overlaps suggest that benefactives form part of a cluster of semantic distinctions which are related to one another in terms of a ‘family resemblance’ (Wittgenstein 1958: 66), a chain of similarity that forms a contiguous region in conceptual space (cf. Croft 2001: 92–94). This finding ties in with Blansitt’s (1988), who investigated locative, allative, dative and more general object case markers and proposed the following logical connections (cf. Fig. 1): Object—Dative—Allative—Locative Figure 1.╇ The conceptual space of goal-encoding devices (Blansitt 1988: 174)
Based on substantial cross-linguistic data, Blansitt argues that case markers can only cover a contiguous region in this space, i.e. disallowing, for example, a marker that covers dative and locative, but not allative functions. In view of the present sample, we are now justified to add purpose to this conceptual space. Haspelmath (2003:€213) has sketched out a possible configuration to draw a semantic map for English to (Fig. 2): predicative possessor
external possessor
direction
recipient
beneficiary
purpose
experiencer
to judicantis
Figure 2.╇ Conceptual space for typical dative functions, with a semantic map of English to
 Karsten Schmidtke-Bode
Although there are differences in the subtleties between individual proposals for the conceptual space of allatives, recipient-benefactives and purposives (cf. e.g. Heine 1991 and Rice & Kabata 2007 for yet more fine-grained maps), there appears to be a general consensus that the three functions are closely related. It should be noted again that the empirical evidence for this position is almost entirely based on data from historically unrelated languages, so genealogy can be excluded as a confounding factor. Likewise, the frequent morphosyntactic clustering of allative, dative and purposive functions is not a (macro-) areal effect. The following map (cf. Fig. 3) shows all languages from my database for which the available material provides clear evidence that at least one purpose clause construction shows traces of allative-dative marking (N = 39). As can be seen, languages of this kind are found in all macro-areas of the world. The absence of genetic and areal effects suggests that allative-dative marking in purpose clauses is a ‘convergent feature’ (Deacon 1997: 116) that has arisen independently in unrelated languages, precisely as an adaptation to functional pressures. We will now turn to the question of what these pressures may be and how various patterns of polysemy emerge and become consolidated over time.
Figure 3.╇ Languages with traces of allative and/or dative marking in at least one purpose clause construction4
4.â•… The more specific distributions in the map should be taken with great caution. It is the availability of information on patterns of polysemy and historical relationships that heavily biases the distribution to well-documented languages. For many Northern American languages, for example, such information cannot be reliably obtained. Their absence in the map cannot, therefore, be taken to exclude the possibility that allatives and datives are (or have been historically) involved in the construction of purpose clauses. All that the map is supposed to illustrate is the geographically widespread attestation of the pattern in question.
Benefactives in purpose clauses 
2.2â•… Motivating forces As was mentioned earlier, a prime candidate for external pressures in our ‘integrativefunctional’ framework is the interplay of language and cognition. Indeed, cognitive linguists have anchored recurrent coding patterns in language to the conceptualization of “prelinguistic and extralinguistic experiences” (Sweetser 1990: 7) that are shared across cultures and speech communities. For example, it has been argued that human beings share the ability for metaphorical conceptualization (cf. §1 above), and this, in turn, has been used to motivate polysemous extensions of lexical and grammatical morphemes. Most famous, perhaps, is the recruitment of spatial expressions for verbalising more abstract temporal concepts: although time constitutes as basic and as real an experience as (motion in) space, it is extremely difficult to construct mental models of time in its own terms (cf. Evans 2004). Instead, location and movement in space provide the necessary conceptual structure to reason and talk about time. Along the same lines, Lakoff and Johnson (1999: 190–91) propose that purposes are also conceptualised metaphorically. Specifically, they hypothesise that purposes are destinations. On this view, acting purposefully is understood as physical motion towards a destination; the desired result is achieved by metaphorically arriving at the destination. This metaphor, they go on to argue, is ultimately grounded in basic patterns of human experience. From our earliest childhood days onwards, we experience that desires and intentions can be fulfilled by a directed change of location: we need to reach out to grasp a bottle, go to the bathroom (or to a river) to wash our hands etc. In other words, the realisation of purposes is often tied to specific locations, and this experiential correlation leads to the emergence of a ‘primary metaphor’, a (near-) universal mapping having “minimal structure and aris[ing] naturally, automatically, and unconsciously through everyday experience” (Lakoff & Johnson 1999: 46). An immediate linguistic reflex of this salient conceptualization is the frequent occurrence of matrix verbs of motion governing purpose clauses (cf. Schmidtke-Bode 2009: 94); many languages do, in fact, have a distinct ‘motion-cum-purpose’ construction alongside a more generally applicable purpose clause. Apart from motion verbs, however, the choice of allative markers to encode purposes can also be argued to reflect the metaphorical construal of intentions. More specifically, the intended result state is construed as a goal, and hence an already existing allative marker of the language can be pressed into service for the expression of that goal, resulting in allative adpositions or case affixes flagging a purposive VP (more on this process below). This direct extension arguably applies to Maori ki and English to.5
5.â•… Notice that there is an alternative explanation that draws, not on metaphorical, but on metonymic reasoning. On this account, destination and purpose are not seen as two different experiential domains, but part of one and the same domain. In terms of Frame Semantics
 Karsten Schmidtke-Bode
With dative-benefactive markers, the story is slightly more complicated because different historical scenarios are logically conceivable. The conceptual motivation for extending allatives to recipient-benefactive contexts is easily framed in metaphorical terms: recipients are “human endpoints of a physical transaction” (Rice & Kabata 2007: 479–480), and beneficiaries can also be conceived of as goals towards which the action is directed. In other words, recipients and beneficiaries are human goals. It is this inherent goal-orientation that is also shared between recipient-beneficiaries and purposes. Therefore, it may well be that allative markers first encroach onto the domain of recipient-benefactive participant marking, before they become analogically extended to verbal purposive contexts. An alternative to this linear extension model is what Rice & Kabata (2007: 497) call ‘unilateral’ development. This scenario underlies the intricate semantic map proposed by the authors (Fig. 4): ALLATIVE
PURPOSE
REASON
INFIN
SUBORD MAN
CONCEPTUAL RECIPIENT
PERC POSS
EXP
LOCATIVE
ADR
BOUND BEN
TIME ABL DUR
Figure 4.╇ Polysemous extensions of allatives in 44 languages (Rice & Kabata 2007: 490)
On this view, allatives may give rise to purposive and recipient-benefactive extensions independently since each pathway is equally well-motivated. The synchronic outcome of both linear and unilateral developments can be a conflation of allative,
(cf. Fillmore 1982), the location in sentences like I’m going to the river constitutes a conceptual frame including the (desired) action typically associated with that location, e.g. bathe, fish, wash etc. This contiguity, in turn, can give rise to inference processes by which directed motion becomes associated directly with the accomplishment of the respective action itself. Put differently, the action comes to stand for the location, and as a result, the grammatical marker previously flagging the location comes to flag the purposive action, thereby being paradigmatically extended from nominal to verbal environments (to NP > to VP). In effect, however, both accounts boil down to a close experiential correlation of a location (as the endpoint of directed motion) and a specific goal tied to that location, so it is difficult to decide whether a metaphorical or a metonymic explanation should be preferred (cf. Schmidtke-Bode 2009 for a more detailed discussion).
Benefactives in purpose clauses 
benefactive and purposive meaning in one morpheme, as we have seen for Yagua, and it can be quite difficult to prioritise one of the two scenarios based on synchronic evidence only (cf. the case of Amele nu ‘go, all, ben, purp’ discussed in Schmidtke-Bode 2009: 188). In any case, however, it seems justified to state that these three semantic domains are unified by the more general notion of goal, and it is hence possible to find languages which have grammaticalized an all-embracing goal marker covering these functions simultaneously. Imonda (Border: Papua New Guinea) can serve as an illustration: (10) Tëla-l-na falgõ i-õb-n [iam maga-na uõl-m ]. husband-noml-poss bow clf-pl-pst later what-instr shoot-gl ‘She got her husband’s bow in order to shoot with it later.’ (Seiler 1985: 162)
It should also be acknowledged, however, that a three-way exponence of allative, benefactive and purposive in a single marker is not that common after all. We have rather seen that recipient-benefactive morphemes develop purposive senses independently from allative marking. (Notice that Haspelmath’s semantic map of dative functions does not take this option into account.) Empirical evidence for this tendency was amply provided by Genetti (1991) and in the literature on grammaticalization more generally (cf. Heine & Kuteva 2002: 55–57 for an overview). In order to motivate this historical pathway satisfactorily, it can be instructive to distinguish the hitherto conflated notions of reception and benefaction6 because, crucially, they may cluster with purposive morphemes in different ways. On the one hand, when recipient, benefactive and purposive marking overlap synchronically, the received view in the literature seems to be that the purposive sense draws directly on the recipient one (e.g. Shibatani 1996), motivated by the underlying goal orientation we encountered above. The benefactive sense, in this scenario, is basically a secondary extension of the recipient meaning, but not directly related to the purposive sense in diachrony (thus, another ‘unilateral development’). On the other hand, however, there are quite a few languages in which a purposive marker is identical with the benefactive morpheme, but completely distinct from the recipient marker typically used in ditransitive constructions (‘give’): Turkish için (ben-purp) is different from the dative marker -a (Kornfilt 1997: 220, 226), in the same way that Lango mε (‘for’) is different from bòt (‘to’) used in ditransitive clauses (Noonan 1992: 149). Similarly, the Ndyuka preposition fu (ben-purp) is not employed for indirect objects of transfer (Huttar & Huttar 1994: 115, 158), and Yaqui, too, separates a directional preposition for give-contexts from a benefactive-purposive one (Valenzuela 2004: 17). There is thus a more profound relationship between
6.â•… I would like to thank Masayoshi Shibatani for drawing my attention to this important point.
 Karsten Schmidtke-Bode
benefaction and purpose which does not primarily hinge on the notion of transfer. Rather, it seems that benefaction always implies the notion of purpose, as argued by Rice & Kabata (2007: 481): “when one acts for the benefit of another (or out of malevolence), he or she is usually acting purposefully”. Therefore, it seems that ‘dative’-like purposive markers arise either due to the goal-orientation of reception, or by virtue of the inherent purposive element of benefaction. These conceptual forces pave the way for ‘paradigmatic intraference’ (Croft 2006) in diachronic change. In this process, the recipient/benefactive adposition or case affix is functionally extended from participant to event marking, which entails a formal shift from nominal to verbal environments (just as with allative > purpose marking). In many cases, this extension also draws the former adposition into a different partof-speech paradigm, such that a benefactive preposition, for instance, joins the set of clause-initial conjunctions in the language. Perhaps the most conducive setting for paradigmatic intraference is that of nominalised verb forms denoting purposeful action, precisely because they are easily construed as paradigmatic alternatives to recipient or benefactive NPs. But the process is, of course, not restricted to nominalisations. In Lavukaleve (Solomons East Papuan), purpose clauses exhibit a bare verb stem (neither overtly non-finite nor nominalised) to which the benefactive postposition ham is suffixed. Thus we have minimal pairs such as the following: (11) a.
a-ham b. a-vala-ham 3sg.m.o-ben 3sg.m.o-pull-ben ‘for him’ (Terrill 2003: 161) ‘for pulling it/to pull it’(Terrill 2003: 439)
In English, benefactive for enters a more complex infinitival construction. It has its origins in Early Modern English, specifically in contexts where the preposition for is used in combination with a purposive infinitive: (12) A tent of purple velvet for the counsailers to mete in. (1548 HALL Chron., Hen. V, (an. 7) 65b; OED)
In this form, the construction may be analysed as consisting of a main clause including a benefactive NP and an associated purpose clause. The authors of the Oxford English Dictionary (OED) note, however, that the construction was reinterpreted and “early extended to include cases to which this analysis is inapplicable”, such as a nonbenefactive NP in a generally non-purposive environment (cf. also Cuyckens 1999; De Smet 2007): (13) It is high time for the Satyrist to pen something […] (1621 R. BRATHWAIT Nat. Embass. Ded., OED)
In even more modern cases, the for-NP cannot be analysed as belonging to the main clause; instead, for is usually treated as “a subordinator marking the start of the non-finite
Benefactives in purpose clauses 
clause” (Huddleston & Pullum 2002: 1203) and the former benefactive NP becomes the subject of the infinitival construction as a whole, with the semantics of benefaction typically being lost: (14) He called [for Ed to be sacked].
(Huddleston & Pullum 2002: 1203)
The progression from an infinitival benefactive-purposive construction to a more widely applicable complement clause represents a special case of the more general constructional intraference ‘from purposive to infinitive’ (Haspelmath 1989). Interestingly, the current verb-specific distribution of ‘for…to’-complements in English is to a large extent semantically constrained (albeit not entirely predictable) by the goalorientation inherent in the elements of the source construction (De Smet 2007: 73ff.). It should have become clear so far that the family of allatives, recipient-benefactivesdatives and purposives forms a network of conceptual and morphosyntactic interdependencies, the latter of which may arise in a variety of different ways. What all of them illustrate, though, is a tendency for economical coding in language. After all, the choice to recruit the same marker for the coding of conceptually related participants and events reduces the overall amount of distinct form-meaning pairings in the language (one of the ‘Minimize Form’ effects in Hawkins’ (2004)). In the following section, we will see that economical coding can go much further, with benefactive arguments occasionally replacing a full-fledged purpose clause altogether.
3.â•… Benefactives as metonymic devices in the expression of purpose A remarkable phenomenon at the interface of benefaction and purpose is the potential of benefactive/recipient/dative NPs to act as an economical shortcut for the expression of purposive relations. The two constructions to be discussed here have been mentioned—often merely in passing—at various points in the literature, yet neither have they been dealt with systematically in the context of purpose clauses, nor has any attempt been made to account for them in a unified way. As we will see shortly, our present framework provides a tool for doing so. In the first construction, a goal-directed activity is expressed by simply mentioning the goal participant, i.e. without spelling out the verbal part of the ‘underlying’ purpose clause. Thus in Imonda, the purpose of a motion event can be expressed by attaching the suffix -m, used for allative, benefactive and purposive marking, to the goal NP of the intended situation: (15) Tëh-ia-m uagl-fan. firewood-loc-gl go-prf ‘He has gone to collect firewood.’
(Seiler 1985: 161)
 Karsten Schmidtke-Bode
A similar example is provided by purposive NPs in Yidi\ (Pama-Nyungan: Queensland/ Australia): (16) Gawu: guga daŋgan buda:gu. tree.abs skin.abs take:off.imp blanket.purp ‘Take the Gawu: bark off [the Gawu: tree] for a blanket (i.e. to use it as a blanket).’ (Dixon 1977: 342)7
Such constructions are reminiscent of, and in fact analogous to, purposive for in English. Thus, (17) is the reduced form of (18): (17) He has gone for lunch. (18) He has gone in order to have/for having lunch.
As can be seen in the three cases, the goal NPs tëhiam, buda:gu and for lunch, respectively, come to stand for the whole verbal situation in which each is a participant, namely ‘collecting firewood’, ‘using something as a blanket’ and ‘having/eating lunch’. Given that an argument thus represents the whole semantic structure of which it is part, one may best characterise this morphosyntactic construction in terms of a metonymy. Similar to conceptual metaphors, metonymies have long been recognised by cognitive linguists as an important tool by which we organise our conceptual knowledge: just like metaphors, creating metonymies is an imaginative process that allows us to construe one conceptual entity in terms of another (cf. Lakoff/Johnson 1980). With metaphors, the two entities are from different domains (such as time and space), while with metonymies, both entities are from the same domain (cf. Leite 1994 for ample illustration). In language use, metonymic mappings are economical to the extent that the entity or category which is explicitly mentioned in discourse can activate another category in the same overall cognitive domain or model (an important corollary of the theory of Frame Semantics, cf. Fillmore 1982). Thus by simply adding a goal NP to a verb denoting directed action, a whole purposive situation—or more precisely, the cognitive model or frame of that situation— can be evoked. In (8) above, we already saw the complex postposition χua-Ži at work in both benefactive NPs and purpose clauses in Qiang. In precisely the way outlined here, this marker can also be attached to nouns that do not themselves denote beneficiaries but the (inanimate) goal or benefit of an action: (19) The: dzigu -χua-Ži ha-qә-Ži koŋtsuo-pә-k. ˚ 3sg money-ben(because-adv) dir-go-adv work-do-go ‘For the sake of money, s/he goes to work.’ (LaPolla 2003: 240)
7.â•… A strikingly similar phenomenon also occurs in Nunggubuyu (Gunwinyguan: Northern Territory, Australia), cf. Heath (1984: 208). Heath uses the term ‘case-spreading’ for such instances.
Benefactives in purpose clauses 
In fact, it is conceivable that such NPs, combining ‘indirect’ benefaction with purpose, provide the crucial link—i.e. the formal and semantic environment needed—for the development from a benefactive PP to a purposive conjunction.8 Of course, the absence of systematic diachronic evidence for this hypothesis across languages makes it impossible to go well beyond speculation. Nevertheless, this line of argumentation accords well with the emphasis laid on metonymic inferencing in the recent literature on grammaticalization (e.g. Brinton 1988; Traugott & König 1991; De Mulder 2002). Conceptual metonymies have come to be seen as ways of “indexing or pointing to meanings that may otherwise be only covert, but are a natural part of conversational practice” (Hopper & Traugott 2003: 92). As I have tried to show, purpose is an implied ‘natural part’ of benefaction, and hence the purposive component of meaning may well become salient over time and thenceforth emancipate, being analogically extended to the coding of purpose clauses.9 In the second construction to be discussed, a benefactive NP encroaches onto the temporal domain, and the conceptual link between them is again an implicit purposive situation. Haspelmath (1997) observed that a number of languages can express atelic temporal extension by a benefactive construction. More technically, an NP denoting duration in time is flagged by a benefactive adposition or case affix. English for provides a good starting point again (cf. also Fuhs submitted): (20) When fermentation has finished, the ‘green beer’ is run into conditioning tanks for a few days. (BNC A0A 31)
The OED labels such constructions ‘intended duration’ uses of for, and they contrast with ‘actual duration’ as in (21) The two great parties … had for a moment concurred. (1849 MACAULAY Hist. Eng. I. 166; OED)
It is challenging to trace the directionality in the historical emergence of this two-way polysemy. An account that derives intended from actual duration would be broadly compatible with Traugott’s concepts of increasing subjectification and internalisation
8.â•… Note, however, that if the benefactive-purposive marker is also used in causal contexts, as in the Qiang data, the diachronic pathways might be significantly different and, importantly, not unidirectional across languages, i.e. there are multiple chains of semantic extension that can bring about a syncretism of cause, benefaction and purpose (cf. Luraghi 2005). 9.â•… Whether or not the grammaticalization path from benefactive adposition or case marker to purposive conjunction is actually instantiated also depends on the available alternatives for marking purpose. English for, for instance, was not fully reanalysed as a purposive conjunction, arguably because the to-infinitive had been well-established in this function before (cf. Los 2005).
 Karsten Schmidtke-Bode
in semantic change (Traugott 1989: 34ff.). It is not straightforward, however, what may have motivated the use of benefactive for in contexts of actual temporal duration in the first place. What is more reasonable, I believe, is that benefactive for came to be used in contexts that implied a purposive situation. On this view, the temporal adverbial in a sentence like (22) I’m […] going back to Denmark for two weeks. (ICE-GB: S1A-089 #093:1:B)
would actually refer to the duration of the state resulting from deliberately going to Denmark (cf. also Haspelmath 1997: 128; Berthonneau 1991). The underlying situation, therefore, is that the person went back to Denmark for (the purpose of) staying/ vacating there over a period of three weeks. This scenario is analogous to the goal-NPs we saw above: here, too, a benefactive-marked argument is capable of evoking an entire purposive situation, the verbal part of which can easily be inferred from the context (i.e. from the mental model of the situation being spoken about) and is hence left implicit for economical reasons. Once established in this function of intended duration, the for+NP construction could be further broadened to actual duration. This process, in which the notion of purpose is lost, i.e. in which “a speaker analyzes out a semantic component from a syntactic element”, has been called ‘hyperanalysis’ by Croft (2006: 81). Although purposive temporal extent tends not to be systematically covered by reference grammars, there are quite a few instances of this construction both within and beyond my sample. In Turkish, the dative case suffix -(y)A basically expresses the typical locative, allative, recipient and benefactive meanings (cf. (23)), but also attaches to arguments expressing temporal extension (cf. (24)): (23) Bu kitab-ı Hasan-a al-dı-m. this book-acc Hasan-dat buy-pst-1sg ‘I bought this book for Hasan.’
(Kornfilt 1997: 226)
(24) Bir hafta-lığ-ın-a gel-dı-m. one week-abstr.n-cmpm-dat come-pst-1sg ‘I have come for one week.’
(Kornfilt 1997: 263)
Importantly, such constructions are morphologically different from other arguments expressing temporal extension. In (25), for instance, the NP iki yıl ‘(for) two years’ remains completely unmarked, i.e. in its bare form: (25) Londra-da iki yıl yaşa-dı-m. London-loc two year live-pst-1sg ‘I lived in London for two years.’
(Kornfilt 1997: 262)
Benefactives in purpose clauses 
The crucial semantic difference between the two constructions is that (24) implies the notion of purpose or intention. Unlike English for, then, Turkish has retained separate markers for intended and actual duration. Completely parallel examples can be found in Malayalam (Dravidian: India) and Italian: (26) żaan ivize oru massatteekkә vannirikkayaa^ә. 1sg here one month.dat/all come.perf1.impf2.prs ‘I have come here for one month.’ (Asher & Kumari 1997: 244) (27) Posso restare qui solo per un’ ora. can.1sg stay here only for one hour ‘I can only stay here for one hour.’
Admittedly, however, we also occasionally find constructions in which benefactives have developed various temporal uses, but without a purposive extension being attested for them. Thus the temporal uses of the dative case in Kolyma Yukaghir (cf. Maslova 2003: 99) and Gulf Arabic (cf. Holes 1990: 146) do not have any implications of purpose. In Wardaman, the dative case is recruited for repetitive temporal extension (‘number of times > 1’) only (cf. Merlan 1994: 75). At present, it seems to me that these patterns have a somewhat exceptional status and are, therefore, fairly hard to motivate from a purely functional view. In conclusion, the bulk of the data presented in this section suggests that Givón’s statement that purpose clauses “can disguise themselves in the syntactic form of benefactive objects in English” (Givón 1984: 132) is applicable to other languages as well. Moreover, I have tried to exploit the importance of conceptual metonymies in language use and grammaticalization in order to account for the relevant constructions.
4.â•… Negative purpose and the absence of benefactive marking Interestingly, there is a class of purpose clauses which is exempt from the previously mentioned tendencies. More precisely, we can observe the systematic absence of goal-encoding devices such as allatives and benefactives in negative purpose clauses (or ‘lest’-constructions). Of course, many languages express negative purpose by simply inserting a negative marker into an ‘ordinary’ purpose clause, so that the goal-marking inherent in the purposive morpheme or elsewhere in the clause remains untouched. However, when languages grammaticalize a construction specifically dedicated to the expression of negative purpose, it seems that neither allatives nor dative-benefactives have a chance to appear in such constructions. We will briefly survey this phenomenon and take first steps towards explaining it in our present theoretical framework.
 Karsten Schmidtke-Bode
Negative-purpose clauses are also called ‘avertive’ constructions since their matrix clause typically encodes that precautions are taken so as to avoid an ‘apprehensioncausing’ situation in the subordinate clause (Lichtenberk 1995: 298). The English construction with lest, shown in (28)–(29), is not only a typical representative of this clause type, but has also led to the widely-used Anglocentric name of the construction (‘lest’-clause): (28) He cut the remark out of the final programme [lest it should offend the listeners]. (BNC BLY 1232) (29) Take hede lest eny man deceave you. (1526 TINDALE Mark xiii, OED)
Our overall sample of 218 purposive constructions contains 20 negative-purpose clauses, distributed over 19 different languages. (30)–(32) give an impression of the morphosyntactic make-up of different ‘lest’-clauses: (30) negative auxiliary verb y + main verb in Yagua (Peba-Yaguan: Peru): … naada-y-numaa dáátya jiy-daasaada. â•… 3du-aux.neg-now know 2sg-mother.in.law ‘… so that your mother-in-law won’t know.’
(Payne & Payne 1990: 416)
(31) preposed ‘lest’-clause with double negative marking in Slave (Athapaskan: Canada) [Daniel yegúh ôále ch’á ] goghádehk’a. Daniel 3opt.find 4.neg lest 1sg.threw ‘I threw it so Daniel wouldn’t find it.’
(Rice 1989: 1262)
(32) avertive suffix in Martuthunira (Pama-Nyungan: Western Australia) Ngayu puni-layi-rru nyina-wirri manhamanha-ma-l.yarra nhuwala-a. 1sg.nom go-fut-now be-lest awkward-caus-sim 2du-acc ‘I’ll go now lest I’ll be making it awkward for you.’ (Dench 1995: 249)
These examples are indicative of the interesting fact that avertive clauses exhibit an entirely different grammatical shape and behaviour when compared to their positive counterparts. In brief, they show a slight preference for finite (rather than non-finite) verbs, often involve different and overtly coded (rather than implicit co-referential) subjects, and do not preferably select motion verbs in the corresponding matrix clauses. In the present context, it is most remarkable that allative and dative-related morphemes are also absent from avertive clauses. Neither do they surface as additional gestalt features in the clause, nor do they provide the historical source for the development of ‘lest’-markers. For positive purpose clauses, I argue in Schmidtke-Bode (2009) that their prototypical morphosyntactic characteristics have arisen as Â�functional
Benefactives in purpose clauses 
Â� adaptations to the conceptual properties of purposive relations and our experience with intentional actions. This approach can only be justified if it also accounts for avertive constructions. Therefore, it may be helpful to consider the scenarios typically expressed by such constructions. In the original study (Schmidtke-Bode 2009), this was achieved by compiling converging evidence from two data sources, i.e. translations of representative ‘lest’clauses from the typological sample, and a corpus-based query into English ‘lest’clauses in the British National Corpus (BNC). Both sources revealed unambiguously that the conceptual situation associated with negative purposes is prototypically such that an agent is acting (on some entity) in order to prevent some undesired situation from happening to her or to this entity. Crucially, this is not usually contingent on directed motion. Consequently, motion verbs are rather infrequent in main clauses of avertive constructions. In the corpus-based study, verbs of motion constitute a mere 4% of all matrix verbs in ‘lest’-constructions. And while it is not a typological rarity for languages to develop distinct ‘motion-cum-purpose’ constructions in positive contexts (e.g. Aissen 1984 on Tzotzil), none of the avertive constructions in my sample specifically attracts motion verbs. In terms of Stefanowitsch and Gries (2003), they rather seem to be ‘repelled’ by avertive clauses (indeed, a collostructional analysis on the corpus data from English yields significant degrees of repellence). Importantly, if avertive situations do not experientially correlate with directed motion, there is no basis for cross-domain mappings (or place-event metonymies, for that matter), and no motivation for allative (or goal-) marking arises in the first place. Avertive situations, on this account, are comparatively harder to construe as spatial goals. Benefactives, too, do not surface in distinctly grammaticalized avertive contexts, even though such situations typically contain a participant that directly benefits from the precautions expressed in the matrix clause. From the viewpoint of grammaticalization mentioned in §3, it is almost precluded that a benefactive adposition or case marker develops into an avertive morpheme, since this would require this morpheme to fuse with a negative marker (semantically or structurally) during the grammaticalization process (for [not-X] > for-not ‘lest’ [X]). I am not aware of any such development.10 Instead, there is some convincing evidence that ‘lest’-markers typically evolve from lexical or grammatical items that are associated with negative semantics right
10.â•… In Hungarian, the negative marker ne can indeed fuse with the purposive conjunction hogy ‘that’ to form what might be called a complex avertive conjunction ne-hogy ‘lest’. Crucially, however, hogy is a complementizer, not a benefactive morpheme (Kenesei et€al. 1998: 51).
 Karsten Schmidtke-Bode
from the start. Thus English lest is the modern form of OE lǽs þe ‘less-rel’, with less in its basic form typically being associated with an ‘undesirable’ NP (e.g. less trouble). This would provide an ideal environment for clausal extensions. Somewhat differently, the Tukang Besi (Austronesian: Sulawesi) conjunction bara ‘lest’ functions as a verb meaning ‘don’t’ in main clauses (Donohue 1999: 399) and may thus ultimately have a verbal source. For To’aba’ita, an Austronesian language spoken on the Solomon Islands, Lichtenberk (1995: 304, 320) reports that ada ‘lest’ is historically derived from a lexical verb meaning ‘see, look out, watch out’, i.e. a perception verbs that has a connotation of warning. The point is that the sources of ‘lest’-markers are inherently negative in one way or the other, and hence benefactives do not lend themselves well to being source constructions for their grammaticalization (interestingly, neither do malefactives). In sum, allative-based benefactives do not occur in grammaticalized ‘lest’-clauses because directed motion is not involved in the conceptualization of avertive situations. Benefactives proper do not overlap significantly with negative purpose from a conceptual point of view, and, moreover, they do not usually occur in structural environments that favour a reanalysis to avertive markers. For these reasons, the historical sources of ‘lest’-markers are fundamentally different from the goal-encoding devices that often give rise to positive purpose markers. In both conceptual and historical terms, then, the absence of allatives and dative-benefactives in ‘lest’-clauses is well-motivated.
5.â•… Conclusion This paper has shifted the attention from benefactives as a semantic participant role of NPs to the distribution and functions of benefactive morphemes in the context of larger grammatical constructions. More specifically, I have surveyed the rich functional potential of the family of ‘goal-encoding devices’, including allatives, datives, recipients and benefactives, in the typology of purpose clauses. Building on previous literature and a database of 80 languages, I outlined the roles of benefactives as both primary and secondary gestalt features in purpose clauses, their metonymic potential for the economical expression of purposive relations, and their systematic absence from avertive constructions. The cognitive-functional framework provided us with useful tools to develop a principled and coherent account for these phenomena, and for their diachronic evolution. With regard to the latter, it would now be most desirable to obtain more explicit data on patterns of polysemy and their historical connections in order to corroborate the hypotheses put forward in this and previous papers on the topic.
Benefactives in purpose clauses 
Abbreviations 1, 2, 3 abs abstr.n acc adv all b3 ben caus clf cmpm comp cop dat def dir du erg ex fut gl impf incl ind indf inf instr lest
1st, 2nd, 3rd person absolutive abstract noun accusative adverbial (case) allative (‘to’) person agreement marker benefactive causative classifier compound-marker complementizer copula dative definite directional, directive dual ergative exclusive future goal imperfective inclusive indicative indefinite infinitive instrumental avertive marker
loc locative m masculine msd masdar (verbal noun) n: n- non- (e.g. n:sg non-singular, n:pst nonpast) negative, negation neg nom nominative noml nominaliser obj object p/o transitive patient (patient-like argument of canonical transitive verb) perf perfect pl plural poss possessive prf perfect prfv perfective prox proximal, proximate pst past tense pst past purp purpose, purposive rel relative clause marker rem remote (past, future) sg singular sim simultaneous spec specific, specifier t/a tense-aspect marker
References Aissen, Judith L. 1984. Control and command in Tzotzil purpose clauses. BLS 10: 559–571. Asher, Ronald E. & Kumari, T.C. 1997. Malayalam. London: Routledge. Bauer, Winifred. 1993. Maori. London: Routledge. Berry, Keith & Berry, Christine 1999. A Description of Abun: A West Papuan Language of Irian Jaya. Canberra: Australian National University.
 Karsten Schmidtke-Bode Berthonneau, Anne-Marie. 1991. Pendant et pour, variations sur la durée et donation de la référence. Langue Française 91: 102–124. Bickel, Balthasar. 2007. Typology in the 21st century: Major current developments. Linguistic Typology 11: 239–251. Blansitt, Edward L. 1988. Datives and allatives. In Studies in Syntactic Typology [Typological Studies in Language 17], Michael Hammond, Edith Moravcsik & Jessica Wirth (eds.), 173–191. Amsterdam: John Benjamins. Brinton, Laurel. 1988. The Development of English Aspectual Systems: Aspectualizers and Postverbal Particles. Cambridge: CUP. Cristofaro, Sonia. 2003. Subordination. Oxford: OUP. Croft, William. 1995. Autonomy and functionalist linguistics. Language 71(3): 490–532. Croft, William. 2001. Radical Construction Grammar: Syntactic Theory in Typological Perspective. Oxford: OUP. Croft, William. 2006. Evolutionary models and functional-typological theories of language change. In The Handbook of the History of English, Ans van Kemenade & Bettelou Los (eds.), 68–91. Oxford: Blackwell. Croft, William & Cruse, D. Allan 2004. Cognitive Linguistics. Cambridge: CUP. Cuyckens, Hubert. 1999. Grammaticalization in the English prepositions to and for. In Cognitive Perspectives on Language, Barbara Lewandowska-Tomaszczyk (ed.), 151–161. Frankfurt: Peter Lang. Dayley, Jon P. 1985. Tzutujil Grammar. Berkeley CA: University of California Press. De Mulder, Walter. 2002. Grammaticalisation, métaphore et métonymie: Le cas d’aller. Special issue of Verbum 24(3): 229–246. De Smet, Hendrik. 2007. For…to-infinitives as verbal complements in Late Modern and Present-day English: Between motivation and change. English Studies 88(1): 67–94. Deacon, Terrence. 1997. The Symbolic Species: The Co-evolution of Language and the Human Brain. London: Penguin. Dench, Alan Charles. 1995. Martuthunira: A Language of the Pilbara Region of Western Australia. Canberra: Dept. of Linguistics, Research School of Pacific and Asian Studies, Australian National University. Diessel, Holger. 2001. The ordering distribution of main and adverbial clauses: A typological study. Language 77(2): 433–455. Dixon, Robert M.W. 1977. A Grammar of Yidi\. Cambridge: CUP. Donohue, Mark. 1999. A Grammar of Tukang Besi. Berlin: Mouton de Gruyter. Evans, Vyvyan. 2004. The Structure of Time [Human Cognitive Processing 12]. Amsterdam: John Benjamins. Fillmore, Charles J. 1982. Frame semantics. In Linguistics in the Morning Calm, The Linguistic Society of Korea (eds.), 111–137. Seoul: Hanshin. Frajzyngier, Zygmunt & Johnston, Eric. 2005. A Grammar of Mina. Berlin: Mouton de Gruyter. Fuhs, Stephan. Submitted. A corpus-based investigation of the aspectual character of the English durative adverbial construction. To appear in a volume on Quantitative Methods for Cognitive Semantics, Dylan Glynn & Kerstin Fischer (eds.). Berlin: Mouton de Gruyter. Genetti, Carol. 1991. From postposition to subordinator in Newari. In Approaches to Grammaticalization, Vol. II: Types of Grammatical Markers [Typological Studies in Language€19:2], Elisabeth Closs Traugott & Bernd Heine (eds.), 227–256. Amsterdam: John Benjamins. Givón, Talmy. 1984. Syntax: A Functional-Typological Introduction, Vol. I. Amsterdam: John Benjamins.
Benefactives in purpose clauses 
Goldberg, Adele. 2006. Constructions at Work: The Nature of Generalization in Language. Oxford: OUP. Harris, Alice C. & Campbell, Lyle. 1995. Historical Syntax in Cross-linguistic Perspective. Cambridge: CUP. Haspelmath, Martin. 1989. “From purposive to infinitive – a universal path of grammaticization.” Folia Linguistica Historica X/1–2: 287–310. Haspelmath, Martin. 1997. From Space to Time. Temporal Adverbials in the World’s Languages. Munich: Lincom Europa. Haspelmath, Martin. 2003. The geometry of grammatical meaning: Semantic maps and crosslinguistic comparison. In The New Psychology of Language, Vol. II, Michael Tomasello (ed.), 211–242. Mahwah NJ: Lawrence Erlbaum Associates. Hawkins, John A. 2004. Efficiency and Complexity in Grammars. Oxford: OUP. Heath, Jeffrey. 1984. Functional Grammar of Nunggubuyu. Canberra: Australian Institute of Aboriginal Studies. Heine, Bernd. 1991. The dative in Ik and Kanuri. In Studies in Typology and Diachrony: Papers Presented to Joseph H. Greenberg on his 75th Birthday [Typological Studies in Language€20], William Croft, Susanne Kemmer & Keith Denning (eds.), 129–149. Amsterdam: John Benjamins. Heine, Bernd & Kuteva, Tania. 2002. World Lexicon of Grammaticalization. Cambridge: CUP. Hengeveld, Kees. 1989. Layers and operators in Functional Grammar. Journal of Linguistics 25: 127–157. Holes, Clive 1990. Gulf Arabic. London: Routledge. Hopper, Paul J. 1987. Emergent Grammar. In Proceedings of the 13th Annual Meeting of the Berkeley Linguistics Society, Jon Aske, Natasha Beery, Laura Michaelis & Hana Filip (eds.), 139–157. Berkeley CA: University of California Press. Hopper, Paul J. & Traugott, Elisabeth Closs. 2003. Grammaticalization, 2nd edn. Cambridge: CUP. Hualde, José Ignacio & Ortiz de Urbina, Jon. (eds.). 2003. A Grammar of Basque. Berlin: Mouton de Gruyter. Huddleston, Rodney & Pullum, Geoffrey K. 2002. The Cambridge Grammar of the English Language. Cambridge: CUP. Huttar, George L. & Huttar, Mary L. 1994. Ndyuka. London: Routledge. Kenesei, István, Vago, Robert Michael & Fenyvesi, Anna. 1998. Hungarian. London: Routledge. Kornfilt, Jacklin. 1997. Turkish. London: Routledge. Lakoff, George & Johnson, Mark. 1980. Metaphors we Live by. Chicago IL: University of Chicago Press. Lakoff, George & Johnson, Mark. 1999. Philosophy in the Flesh: The Embodied Mind and its Challenge to Western Thought. New York NY: Basic Books. Langacker, Ronald W. 1999. Grammar and Conceptualization. Berlin: Mouton de Gruyter. LaPolla, Randy J. 2003. A Grammar of Qiang. Berlin: Mouton de Gruyter. Leite, N. 1994. Master metonymy list. Ms, University of California at Berkeley. Lichtenberk, František. 1995. Apprehensional epistemics. In Modality in Grammar and Discourse [Typological Studies in Language 32], Joan Bybee & Suzanne Fleischman (eds.), 293–327. Amsterdam: John Benjamins. Los, Bettelou. 2005. The Rise of the to-Infinitive. Oxford: OUP. Luraghi, Silvia. 2005. Paths of semantic extension – From cause to beneficiary and purpose. In Historical Linguistics 2003 [Current Issues in Linguistic Theory 257], Michael Fortescue, Eva Skafte Jensen, Jens Eric Mogensen & Lene Schøsler (eds.), 141–157. Amsterdam: John Benjamins.
 Karsten Schmidtke-Bode Maslova, Elena. 2003. A Grammar of Kolyma Yukaghir. Berlin: Mouton de Gruyter. Merlan, Francesca C. 1994. A Grammar of Wardaman. Berlin: Mouton de Gruyter. Moravcsik, Edith A. 2007. What is universal about typology? Linguistic Typology 11(1): 27–41. Nichols, Johanna. 2004. Samples for comparative grammar: A practical guide. Ms, University of California at Berkeley. Noonan, Michael. 1992. A Grammar of Lango. Berlin: Mouton de Gruyter. Payne, Doris L. & Payne, Thomas E. 1990. Yagua. In Handbook of Amazonian Languages, Vol. 2, Desond C. Derbyshire & Geoffrey K. Pullum (eds.), 249–474. Berlin: Mouton de Gruyter. Rice, Keren. 1989. A Grammar of Slave. Berlin: Mouton de Gruyter. Rice, Sally & Kabata, Kaori. 2007. Crosslinguistic grammaticalization patterns of the allative. Linguistic Typology 11: 451–514. Schmidtke-Bode, Karsten. 2009. A Typology of Purpose Clauses [Typological Studies in Language 88]. Amsterdam: John Benjamins. Schwarze, Christoph. 1995. Grammatik der italienischen Sprache, 2nd edn. Tübingen: Niemeyer. Seiler, Walter. 1985. Imonda. Canberra: Australian National University. Shibatani, Masayoshi. 1996. Benefactive constructions: A schema-based approach. Chicago Linguistic Society 32: 319–334. Sridhar, S.N. 1990. Kannada. London: Routledge. Stefanowitsch, Anatol & Gries, Stefan. 2003. Collostructions: Investigating the interaction of words and constructions. International Journal of Corpus Linguistics 8(2): 209–243. Sweetser, Eve. 1990. From Etymology to Pragmatics. Metaphorical and Cultural Aspects of Semantic Structure. Cambridge: CUP. Terrill, A. 2003. A Grammar of Lavukaleve. Berlin: Mouton de Gruyter Thompson, Sandra A. & Longacre, Robert E. 1985. Adverbial clauses. In Language Typology and Syntactic Description, Vol. II: Complex Constructions, Timothy Shopen (ed.), 171–218. Cambridge: CUP. Thompson, Sandra A. & Couper-Kuhlen, Elizabeth. 2005. The clause as a locus of grammar and interaction. Discourse Studies 7(4/5): 481–505. Traugott, Elizabeth Closs. 1989. On the rise of epistemic meanings in English: An example of subjectification in semantic change. Language 65: 31–55. Traugott, Elizabeth Closs & König, Ekkehard. 1991. The semantics-pragmatics of grammaticalization revisited. In Approaches to Grammaticalization [Typological Studies in Language 19:1], Elizabeth Closs Traugott & Bernd Heine (eds.), 189–218. Amsterdam: John Benjamins. Valenzuela, Lilian Graciela. 2004. The Syntax-Semantics Interface in Yaqui Complex Sentences: A Role and Reference Grammar Analysis. Ph.D. dissertation, The State University of New€York at Buffalo. Van Valin, Robert D., Jr. & LaPolla, Randy J. 1997. Syntax: Structure, Meaning and Function. Cambridge: CUP. Wittgenstein, Ludwig. 1958. Philosophical Investigations, 2nd edn. Translated by Gertrude Elizabeth Margaret Anscombe. Oxford: Blackwell.
Benefactive and malefactive uses of Salish applicatives Kaoru Kiyosawa & Donna B. Gerdts
University of Victoria/Simon Fraser University We survey benefactives and malefactives in Salish, a family of twenty-three languages in northwestern North America. For the most part, benefactives and malefactives are expressed via applicative constructions, which are classified into two types: redirective and relational. Redirective applicatives are formed on transitive bases, and their precise interpretation—as benefactive, delegative, or malefactive—depends upon the context of the situation and the semantics of the verb. Most transitive verbs form redirectives with benefactive meanings, but redirectives formed on transfer verbs often express malefactive meanings, especially when a source or possessor is the applied object. Relational applicatives are formed on intransitive bases. They frequently have malefactive or adversative meanings, especially with natural or psychological events, and only rarely express benefactive meanings.
1.â•… Introduction1 There are twenty-three Salish languages currently or historically spoken in British Columbia, Washington, Idaho, Montana, and Oregon. This paper surveys benefactives and malefactives based on primary data from the Central Salish language Halkomelem, as well as from secondary source data from nineteen other Salish languages.2 Much of
.â•… We would like to thank the speakers of the Island dialect of Halkomelem who shared data with us, especially Arnold Guerin, Ruby Peter, and Theresa Thorne. Funding was provided by Jacobs Fund, Phillips fund, SFU, and SSHRC. Thanks also to the Salishanists who shared data and knowledge with us: Dawn Bates, David Beck, Ewa Czaykowska-Higgins, Ivy Doak, Brent Galloway, Dwight Gardiner, Mercedes Hinkson, Tom Hukari, Paul Kroeber, Anthony Mattina, Nancy Mattina, Timothy Montler, Nile Thompson, Jan van Eijk, Honoré Watanabe, and especially the late Dale Kinkade. Thanks to Charles Ulrich for his comments and corrections. .â•… In making use of data from secondary sources, we updated transriptions to current standards and modified some morphological boundaries according to our own analysis of applicatives, and also we have added morphological boundaries when they are not provided in the original data. When the original data do not have morpheme glosses, we have provided them
 Kaoru Kiyosawa & Donna B. Gerdts
what we have to say about benefactives and malefactives in Salish languages is tied to the topic of applicatives. Applicative constructions are the main way to express these meanings, but they also express many other meanings as well. Salish applicatives are divided into two types, which we refer to as redirectives, following Kinkade (1980: 33) and relationals, following (Thompson & Thompson 1992: 73). In a typical redirective applicative, a suffix is added to a transitive base to produce a semantically ditransitive form, as in (1), while in a relational applicative, a suffix is added to an intransitive base to produce a semantically transitive form, as in (2):3
(1) Halkomelem (Gerdts 1988: 90) , ni‘ qw´әl-әłc-t-әs łә słéni‘ ‘ә kwθә sәplíl. aux bake-rdr-tr-3erg det woman obl det bread ‘He baked the bread for the woman.’
(2) Halkomelem (Gerdts 1988: 144)
ni‘ kwukw-me‘-θám‘š-әs. aux cook-rel-tr:1sg.obj-3erg ‘He cooked for me.’
The morphosyntax of applicative constructions is straightforward, and both types of applicatives result in the same surface syntax: the semantically oblique NP, in this case the benefactive, is the syntactic object, and thus is licensed as a direct argument in
by using dictionaries and grammars, and by referring to other data. We have also changed some of the glosses from the original sources in order to standardize them to the abbreviations used here. .â•… The following abbreviations and symbols are used in this paper. 1 2 3 art asp aux clt cs dem det emph erg fut imp impf intr
first person second person third person article aspect auxiliary clitic causative demonstrative determiner emphatic ergative future imperative imperfect intransitive
irr lnk loc lv mdl nm obj obl pass past perf pl poss prfx ptc pstn
irrealis linker locative linking vowel middle nominalizer object oblique passive past perfect plural possessive (unglossed) prefix particle positional
punct q rdr red rel refl rslt sg st sub tr vbl //…// =…
punctual question particle redirective reduplication relational reflexive resultative singular stative subject general transitive verbalizer morphophonemic form lexical suffix
Salish applicatives 
the clause or object agreement on the verb.4 The comparative/historical picture of the morphology is much more complicated because of the variety of forms and the various meanings they express in the different Salish languages. Kiyosawa (2006) gives a detailed discussion of this topic. We know a great deal about redirectives and their use to express benefactives, as detailed in Section 2. A complication, however, is that most Salish languages do not have a redirective morpheme that is dedicated to a benefactive meaning per se; redirective applicatives also expresses dative,5 source, and possessive applied objects. The interpretation of an applicative construction is often ambiguous, though the class of the base verb and the semantic context help to pinpoint the meaning. We know less about malefactives, and again no morpheme is specifically dedicated to expressing malefactive meanings. The pragmatics of the situation are crucial for distinguishing benefactive versus malefactive interpretations. Relational applicatives and their use to express benefactive and malefactive meanings are detailed in Section 3. The main use of relationals is to express psych and directional applicatives; their use to express benefactives and malefactives is much less common. One documented use of relationals is to express adversatives, which usually appear as passives. In Salish languages, applicatives are the major means of expressing benefactives and malefactives. As discussed in Section 4, other constructions, such as prepositional phrases or serial verbs are regarded as circumlocutions and would not normally be used. The one exception is that causatives are also used to express benefactives in a very limited set of cases. We conclude with a summary of our findings in Section 5. Salish languages, like many languages of the world, have no morphological forms dedicated solely to the
.â•… As discussed in in Kiyosawa (2006) and Gerdts and Kiyosawa (2007), applied objects in Salish applicative constructions have all the hallmarks of direct objects: they undergo passive, reflexive, reciprocal, and extraction. Although theme nominals in applicatives vary with respect to nominal marking, they lack all object properties: they never appear as object pronouns, passivize, etc. .â•… Throughout this paper, we are using “dative” as a semantic notion, not as a case term. In fact, there is little or no case marking in Salish languages. We use dative as a convenient cover term for recipient, goal of a speech act, and goal or purpose of an action, but not goal or endpoint of a motion. Such usage follows in the tradition of Fillmore (1968), Givón (1984) inter alia. We refer to applicatives with dative applied objects as dative applicatives, which invokes a parallelism to the Transformational Grammar term “dative movement”. Using dative in this way is not without controversy; as our editors pointed out to us, dative is more properly limited to the discussion of case marking.
 Kaoru Kiyosawa & Donna B. Gerdts
expression of benefactives or malefactives, and therefore the speech act context must supply much of the meaning.
2.â•… Redirective applicatives In the redirective applicative construction, the base is transitive, and the role of direct object is “redirected” from the theme to the applied object, which is a semantically oblique NP. Compare the following Shuswap examples:
(3) Shuswap (Dwight Gardiner p.c.) , , m-kúl-n-s > mimx. perf-make-tr-3sub det basket ‘She made the basket.’
(4) Shuswap (Gardiner 1993: 31) , , m-kúl-x-t-s > núxˇwәnxˇâ•›w tә mimx. perf-make-rdr-tr-3sub det woman obl basket ‘She made a basket for the woman.’
(3) is a simple transitive construction, and the agent is the subject and the theme is the object. (4) is a redirective applicative construction, and a semantically oblique NP, the benefactive, is the direct object while the theme is an oblique-marked NP.6 The verb in (3) is transitive and is suffixed with the general transitive suffix -n(t), the third-person transitive subject determines ergative agreement, and the theme ‘basket’ is a direct object. As discussed below, the redirective construction typically involves dative, benefactive, malefactive, or possessive applied objects. Example (4) is a benefactive applicative: the verb is suffixed with the redirective suffix -x(i), the benefactive ‘woman’ is the direct object, and the theme ‘basket’ appears with an oblique marker. (3) is syntactically transitive with two arguments: a subject and a (theme) direct object. And it is also semantically transitive with two participants. (4) is syntactically transitive as well, having two direct arguments: a subject and an applied object (the benefactive). However, (4) is semantically ditransitive, with three participants: a subject, a (benefactive) applied object, and a (theme) oblique. In this case, the redirective applicative suffix has allowed an increase in the semantic valence over the sentence with only a general transitive suffix in (3).
.â•… See Gerdts (to appear) for a detailed discuss of the syntax of ditransitive constructions in one Salish language, Halkomelem.
Salish applicatives 
Each Salish language has from one to three redirective suffixes. Table 1 shows data for the twenty Salish languages in our study organized by the five branches; the Interior Salish branch is further divided into two sub-branches—Northern and Southern:7 Table 1.╇ Salish redirective suffixes Branch
Sub-Branch
Bella Coola Central Salish
Tsamosan Tillamook Interior Salish
Northern Interior
Southern Interior
Language
Redirective
Bella Coola Comox Sechelt Squamish Halkomelem Nooksack Northern Straits Klallam Lushootseed Twana Upper Chehalis Cowlitz Tillamook Lillooet Thompson Shuswap Okanagan Kalispel Coeur d’Alene Columbian
-amk -‘әm -em -ši -as, -łc -ši -si -si -yi -ši -ši, -tuxâ•›wt, -tmi -ši, -tuxâ•›wt, -s -ši -xit -xi -xi -xi, -ł, -tuł -ši, -ł -ši, -ł, -tuł -xit, -ł, -tuł
The concept of redirective applicative—adding a third participant as a core argument—is a very old concept in Salish. The most common redirective suffix is reconstructed for Proto-Salish as *-xi by Kinkade (1998). It is found in all three branches, and, in fact, is the only redirective suffix in Northern Interior Salish and most of the Central Salish languages. Reflexes of this form (-si, -ši, -xi, -xit, -yi) appear in all but four of the languages (Bella Coola, Comox, Sechelt, and Halkomelem).8 Other suffixes
.â•… The following references were consulted for the information in the Tables 1 and 4 in this paper: Bates et al. (1994); Beaumont (1985); Carlson (1972, 1980); Carlson and Flett (1989); Davis and Saunders (1997); Doak (1997); Egesdal and Thompson (1998); Galloway (1997); Gerdts (1988); Hess (1967); Kinkade (1980, 1991, 2004); Kuipers (1967, 1974, 1992); A. Mattina (1994); N. Mattina (1993); Montler (1986); Thompson and Thompson (1992); Van Eijk (1997); Watanabe (2003). .â•… The suffix in the outlier language Bella Coola is very different in form and function from redirectives in other languages, as discussed in Kiyosawa (2006).
 Kaoru Kiyosawa & Donna B. Gerdts
have been added to the redirective system in some branches or individual languages and have usurped the functions of *-xi or added additional functions to the redirective applicative system. The following sections explore the semantics of the redirective construction and, in particular, their use to express benefactive and malefactive meaning.
2.1â•… Semantic roles of applied objects in redirective applicatives The semantic role of the applied object in the redirective applicative construction is usually dative, benefactive, source, or possessor, as elaborated in the following sections.
2.1.1â•… Dative Salish languages express recipients as direct objects:
(5) Sechelt (Beaumont 1985: 138)
, , yát-cí-čen-skâ•›wa ‘e še s‘íłten, we téqíš-axâ•›w. give-tr:2sg.obj-1sg.sub-fut obl det food if sit.down-2sg.sub ‘I’ll give you some food if you sit down.’
This is true whether the verb appears in a simple form, as in the above examples, or if it takes a redirective suffix, as in the following examples:
(6) Kalispel (Carlson 1980: 25) , xâ•›wíc-š-t-әn łu‘ Agnes łu‘ t yámxˇâ•›we‘. give-rdr-tr-1sg.sub art Agnes art obl basket ‘I gave a basket to Agnes.’
A typical dative applied object is a recipient of a transfer verb, such as ‘give’ (6). We use “dative” loosely, and include goals of speech acts (7) as well as goals or targets of actions in general (8)–(9):
(7) Halkomelem (Gerdts 1988: 92)
(8) Cowlitz (Kinkade 2004: 240)
, ni yә´θ-әs-t-әs łә Mary ‘ә kâ•›wθәn syays. aux tell-rdr-tr-3sub det Mary obl det:2sg.poss work ‘He told Mary about your job.’ ‘it cíxˇ-tuxâ•›w-c. perf show-rdr-1sg.obj ‘He showed it to me.’
(9) Thompson (L. Thompson & M. Thompson 1980: 28) , qâ•›w‘xítne. , , //qâ•›wәw-xi-t-â•›ø-enè// trap-rdr-tr-3sg.obj-1sg.sub ‘I set a trap for it [a particular animal].’
Salish applicatives 
2.1.2â•… Benefactive Redirective suffixes are used to form constructions expressing a classic benefactive meaning of doing something for someone’s benefit: (10) Cowlitz (Kinkade 2004: 234)
, ‘it sá‘-š-n ł tit †íqsn. perf make-rdr-tr obl det box ‘He made the box for him.’ (11) Comox (Watanabe 2003: 249)
, †әs-‘әm-θi tθәm ‘ә tә čuy. punch-rdr-tr:2sg.obj 1sg.sub:fut obl det child ‘I’ll punch the kid for you.’ (12) Halkomelem (Gerdts 1988: 95) , ni‘ θә´y-әłc-θámš-әs ‘ә kâ•›wθә nә-snә´xâ•›wәł. aux fix-rdr-tr:1sg.obj-3sg.sub obl det 1sg.poss-canoe ‘He fixed my canoe for me.’ (13) Upper Chehalis (Kinkade 1991: 10) ‘it ‘ә´xˇ-xˇ-tuxâ•›wt čn. ptc see/look.at-red-rdr 1sg.sub ‘I examined it for him.’ (14) Cowlitz (Kinkade 2004: 235) ,â•› ‘it q wałé‘-s-c. perf mark/write-rdr-1sg.obj ‘He signed [it] for me.’ (15) Kalispel (Carlson 1972: 89)
‘ul-ł-cí-n. //wil-ł-t-si-en// burn-rdr-tr-2sg.obj-1sg.sub ‘I burned it for you.’
(16) Okanagan (A. Mattina 1994: 208) , , kâ•›wu sәq-túł-t-s i‘ slíp. 1sg.obj split-rdr-tr-3sub art wood ‘He split wood for me.’ (17) Lillooet (Van Eijk 1997: 115) , , txâ•›wus-min-xí-c-kaxâ•›w ni n-cqáxˇ‘ a! look-rel-rdr-1sg.obj-2sg.sub det 1sg.poss-horse ptc ‘Look out for my horse for me!’
 Kaoru Kiyosawa & Donna B. Gerdts
(18) Columbian (Willett 2003: 139) , , xәkxíc st,ámka‘s t syáya‘. , //xәk-xit-s stamka‘-s// pick-rdr-3sub daughter-3sg.poss obl serviceberry ‘S/he picked some serviceberries for her/his daughter.’
All of the suffixes in Table 1 can be used to express benefactive meanings, except for -as in Halkomelem and -tmi in Upper Chehalis, which are used only for dative applicatives. As discussed further in Section 4 below, redirective suffixes are the main means for expressing benefactives in Salish languages.
2.1.2.1â•… Delegative.â•… Besides the classic benefactive meaning of doing something for someone’s benefit, Salish redirective suffixes can also be used to express delegative meanings. Thus, the Coeur d’↜Alene example in (19) is glossed with two interpretations— delegative or benefactive. (19) Coeur d’↜Alene (Doak 1997: 157) , níčšices xâ•›we pilí. , //níč-ši-t-s-es xâ•›we pili// cut-rdr-tr-1sg.obj-3sub det Felix ‘Felix cut (wood) instead of me/in my place.’/‘Felix cut (wood) for me.’
An overlap in these two meanings is understandable, since one is often doing a favor for someone when one does a task instead of them. Languages frequently conflate these two meanings, for example, as seen by the use of for for both in English. Also, our Halkomelem consultant has verified that multiple readings are available for all benefactive applicatives. Take the following, for example: (20) Halkomelem (Gerdts, f.n.) , , qâ•›wәl-әłc-θamә cәn ce‘ ‘ә kâ•›w sce:łtәn. bake-rdr-tr:2obj 1sub fut obl det salmon ‘I will bake some salmon for you.’
She says: “you can use this for your benefit in whatever way: for you to eat, because you are unable to do it for whatever reason, because you are too busy to do it and it needs to be done, because I am being substituted to do your job, and so on.” The precise meaning is determined by the context. However, the most normal or neutral reading in the absence of a context would be that the salmon is being cooked for the referent of the object to eat themselves rather than for the salmon to be cooked to give it to someone else to eat.
Salish applicatives 
The use of redirectives on intransitive—rather than transitive—bases is rare in Salish, as discussed in Kiyosawa (2006: Chapter 5). We have found a handful of examples of benefactive applicatives formed on intransitive activity predicates:9 (21) Comox (Watanabe 2003: 252) , , paya‘ štәm t ča-čah-am-‘әm-t-anapi. always 1pl.sub clt red(impf)-pray-mdl-rdr-tr-2pl.obj ‘We will always be praying for you (pl.).’
Two of these examples are translated with delegative meanings: (22) Okanagan (N. Mattina 1993: 272) kâ•›wu qâ•›wәlqâ•›wíl-x-t-s. 1sg.obj talk-rdr-tr-3sub ‘He talked for me (in my stead).’ (23) Okanagan (N. Mattina 1993: 272) , , kâ•›wu qәy=sqáxˇa‘-x-t-s. 1sg.obj burn=dosmestic.animal-rdr-tr-3sub ‘He branded for me (in my stead).’
The distinction between a classic benefactive and a delegative meaning is apparently derived from the meaning of the predicate and the context of the situation.
2.1.2.2â•… Malefactive.â•… Salish redirectives can also be used to expressive malefactive meanings. (24) Comox (Watanabe 2003: 251)
, †әpxâ•›w-a‘am-θ-as ‘ә tә tθ xˇapay. break-rdr-tr:1sg.obj-3sub obl det 1sg.poss stick ‘He broke my stick on me.’
(25) Thompson (L. Thompson & M. Thompson 1980: 28)
má‛xtimes tә s-zélt-ep. //má‛-xi-t-uym-es// break-rdr-tr-2pl.obj-3sub obl nm-dish-2pl.poss ‘He broke you people’s dish.’
(26) Shuswap (Kuipers 1992: 49) , x-lm[t]=ci-x-t-s prfx-close=mouth-rdr-tr-3sub ‘close door on somebody’
.â•… See also Example (2) above.
 Kaoru Kiyosawa & Donna B. Gerdts
While some examples are translated with only a benefactive reading or only a malefactive one, many examples are given more than one interpretation. Thus, whether the applied object bears the role of benefactive or malefactive often depends on the situation. Watanabe (2003: 251) states, ‘the choice [between benefactive and malefactive] seems to depend on the context’, and cites examples like the following: (27) Comox (Watanabe 2003: 251) , , , čәpxˇ-a‘am-θ-as ‘ә tә tθ qәsnay. dirty-rdr-tr:1sg.obj-3sub obl det 1sg.poss dress ‘She dirtied my dress [on me]./She dirtied my dress for me.’ (28) Comox (Watanabe 2003: 252) qâ•›wuqâ•›wu-‘әm-θ-as ‘ә tә tθ tiy. drink-rdr-tr:1sg.obj-3sub obl det 1sg.poss tea ‘He drank my tea for me [when I could not finish it].’/‘He drank up my tea [on me].’
The following Shuswap example also allows two interpretations: (29) Shuswap (Gardiner 1993: 21) m-sté(t)‘ә-x-t-sm-s tә xˇâ•›wәxˇâ•›w‘ú‘s. perf-drink-rdr-tr-1sg.obj-3sub obl beer ‘She drank the beer for/on me.’
In sum, the choice between benefactive and malefactive readings is contextual based on the pragmatics of the situation. The same event may have different readings depending on the opinion of the speaker regarding the situation.
2.1.3â•… Source Another use of redirective applicatives is to express source applied objects of transfer verbs such as ‘buy’ (30), ‘steal’ (31), and ‘take away’ (32): (30) Columbian (Kinkade 1980: 33.1) táw-ł-n. buy-rdr(-tr)-1sg.sub ‘I bought it from him.’ (31) Columbian (Kinkade 1982: 57) c-l әm-ł-cí-nn. ˙˙ impf-steal-rdr-tr:2sg.obj-1sg.sub ‘I am stealing it from you.’ (32) Squamish (Kuipers 1967: 253) , pí‘-ši-t-ka ta xˇәl‘tn-s! grab-rdr-tr-imp det pencil-3sg.poss ‘Take that (lit. his) pencil from him!’
Salish applicatives 
Many cases involving source applied objects carry the implication of malefaction, more specifically deprivation, since the event involves physically separating a theme from the source. Applicatives based on action verbs, such as ‘hide’ (33), ‘pull’ (34), and ‘keep’ (35), are especially prone to this interpretation: (33) Nooksack (Galloway 1997: 222) kâ•›wo-wát-as ‘íł kâ•›wa[l]-xyí-θ-as. someone aux hide-rdr-tr:1sg.obj-3sub ‘Someone hid something from me.’ (34) Columbian (Willett 2003: 256)
ncәkâ•›wakstúłn wa hacmíntn. ˙ w //n-cәkâ•› =akst-túł-t-n// pstn-pull=hand-rdr-tr-1sg.sub ptc rope ‘I pulled the rope out of his hand.’
(35) Shuswap (Kuipers 1974: 154; Kuipers 1992: 49) tәknem-x-t-s keep-rdr-tr-3sub ‘withhold from object/refuse to give something to somebody (object)’
The applied object in (36) can be dative or source:10 (36) Lushootseed (Hess & Bates 2004: 176) ‘u-kâ•›wәd-yí-t-әb čәd ti‘ił qâ•›włay‘. punct-take-rdr-tr-pass 1sg.sub det stick ‘She took that stick to me (i.e. whipped me).’/‘She took that stick from me.’
Both translations have malefactive connotations.
2.1.4â•… Possessive Many, but not all, Salish languages use redirective suffixes to form possessive applicative constructions (a.k.a. “possessor ascension” or “external possession” constructions). The applied object is interpreted as the possessor of the theme NP: (37) Comox (Watanabe 2003: 252) łәw-‘әm-θ-as ‘ә šә tθ aplәs-uł take.out-rdr-tr:1sg.obj-3sub obl det 1sg.poss apple-past , ‘ә tә kâ•›waxâ•›wa. obl det box ‘He took my apples from the box.’
.╅ The passive in Salish is often not reflected in the English translation.
 Kaoru Kiyosawa & Donna B. Gerdts
(38) Twana (Kinkade n.d.)
, kâ•›wәd-ší-d-čәd ә tә wәq´әb. get-rdr-tr-1sg.sub obl art box ‘I took his box.’
Possessors cannot be objects in Salish languages unless the verb takes applicative morphology.11 In Examples (37) and (38), the translation indicates that the theme NP is possessed by the applied object even though no possessive marking appears on that NP. In other examples of possessive applicatives, possessive marking appears on the theme NP: (39) Shuswap (Kuipers 1992: 49) mlmalqâ•›w-x-t-s tә citxâ•›w-s. paint-rdr-tr-3sub obl house-3poss ‘Hei paints the/hisj house for himj.’/‘Hei paints hisj house.’
Note that, since the same redirective suffixes are used to mark both benefactive applicatives and possessive applicatives, the applied object in examples like (39) can be interpreted as either the benefactive or the possessor. As discussed below, the redirective suffix -ł in Southern Interior Salish languages is best analyzed as primarily a possessive suffix. In fact, it is generally the case that possessive applicative constructions do not have simple possessor semantics, but rather have an additional semantic “kick” indicating that the possessor is affected by the action (cf. Fried 1999). So the applied object bears an additional role: dative (40), benefactive (41), malefactive (42), or source (43). (40) Okanagan (N. Mattina 1993: 277) , kâ•›wu c-xâ•›wic-ł-t i-kł-lkalát. 1sg.obj asp-give-rdr-tr 1sg.poss-irr-bread ‘Give me what will be my bread.’ (41) Kalispel (Vogt 1940: 34) yes-u:l-ł-t-ém. asp-burn-rdr-tr-intr ‘I am burning it for him.’/‘I am burning his…’ (42) Columbian (Kinkade 1980: 34) , wәlq wátkâ•›w-ł-c wa ‘in-lәtí. drink-rdr-tr:1sg.obj ptc 1sg.poss-tea ‘She drank my tea (after taking it away from me).’
.â•… An exception to this generalization is a construction with a lexical suffix construction—the Salish equivalent to noun incorporation—in which the semantic possessor of the lexical suffix is the object. See Gerdts and Hinkson (2004) for a discussion of the relation of lexical suffixes, applicatives, and external possession.
Salish applicatives 
(43) Okanagan (A. Mattina 1994: 212) , lut kâ•›wu a-ks-naqâ•›w-m-ł-t-әm in-kәwáp. not 1sg.obj 2sg.poss-fut-steal-rel-rdr-tr-intr 1sg.poss-horse ‘Don’t steal my horse from me.’
This double layer of semantics, together with the added complication that the possessive marking often appears on the theme in possessive applicatives, leads to a confusing range of translations for many examples, as discussed in more detail in Kiyosawa (2004).
2.2â•… The mapping of form and function It is not unusual for languages to have a single multi-purpose applicative morpheme that is used in a variety of applicative constructions such as dative, benefactive, and possessive. Languages with a single applicative include Swahili (Driever 1976), Mayan languages (Aissen 1987), and Mixean languages (Zavala 1999). However, it is also fairly common for a language to have two or more applicative morphemes, each specialized for use with applied objects in a limited range of the semantic roles. Languages that have several applicatives include Chickasaw (Munro 2000), Hakha Lai (Peterson 2007), Ilokano (Gerdts & Whaley 1993), Kinyarwanda (Kimenyi 1980), Nez Perce (Rude 1985), Tukang Besi (Donohue 1999), Upper Necaxa Totonac (Beck 2006), and Yimas (Foley 1991). In Salish, we can see both types of mappings: there are some general purpose applicative morphemes and some suffixes that tend to map to specific semantic roles. First we will discuss languages that have a single multi-purpose redirective suffix and then we will turn to languages that have more than one redirective suffix.
2.2.1╅ Languages with one redirective suffix Many Salish languages have only a single redirective suffix and the applied object takes on a variety of semantic roles. For example, in Squamish, applied objects in redirective applicatives have three semantic roles: dative (a), benefactive (b), and possessor/source€(c). (44) Squamish a.
, sát-ši-t-m čáxâ•›w ‘it. give-rdr-tr-pass 2sg.sub fut ‘He’ll give it to you.’
b. číł-ši-t-úmuł-ka! put.up.high-rdr-tr-1pl.obj-imp ‘Put it up high for us!’ , c. pí‘-ši-t-ka ta xˇ´әl‘tn-s! grab-rdr-tr-imp det pencil-3sg.poss ‘Take that (lit. his) pencil from him!’
(Kuipers 1967: 303)
(Kuipers 1967: 318)
(Kuipers 1967: 253)
 Kaoru Kiyosawa & Donna B. Gerdts
In some languages, applied objects can have four semantic roles. For example, Shuswap has only one redirective suffix, -xi, and the semantic role of the applied object can be dative (a), benefactive (b), malefactive (c), or possessor (d): (45) Shuswap , a. tw kә-mí-x-t-s. sell-rel-rdr-tr-3sub ‘He sells it to somebody.’ (Gardiner 1993: 23) , , ,w ,w b. c-kâ•› i-kâ•› ł-x-cm-e †k-ci‘! prfx-red-leave.food-rdr-tr:1sg.obj-imp irr-meat ‘Leave some meat for me!’ (Kuipers 1974: 222) , c. x-lm[t]=ci-x-t-s prfx-close=mouth-rdr-tr-3sub ‘close door on somebody’ (Kuipers 1992: 49) d. > Mary wik-x-t-sm-s tә n-qé‘čә. det Mary see-rdr-tr-1sg.obj-3sub obl 1sg.poss-father ‘Mary saw my father.’ (Gardiner 1993: 22)
In sum, Salish languages that have only a single redirective suffix use it to form applicatives in which the applied object has a variety of semantic roles.
2.2.2â•… Languages with more than one redirective suffix In contrast, when a language has more than one redirective suffix, the semantic roles associated with a certain suffix are more limited. Halkomelem illustrates this point. There are two redirective suffixes in Halkomelem, -as and -łc, and the applied object is always dative with -as, as in (46), and always benefactive with -łc, as in (47):12 (46) Halkomelem
, ni‘ ‘ám-әs-t-әs kâ•›wθә swíwlәs ‘ә kâ•›wθә púkâ•›w. aux give-rdr-tr-3sub det boy obl det book ‘He gave the boy the book.’ (Gerdts 1988: 115) , , w b. ni‘ ‘iw -әs-θamš-әs ‘ә kâ•› θә qeq-s. aux show-rdr-tr:1sg.obj-3sub obl det baby-3poss ‘She showed me her baby.’ (Gerdts & Hinkson 2004: 228) a.
(47) Halkomelem a.
ni‘ lә´xˇâ•›w-әłc-әt-әs. aux blanket-rdr-tr-3sub ‘He covered it with a blanket for him.’
(Gerdts 1988: 101)
.â•… Although the benefactive redirective suffix -łc partially resembles -ł, the possessive redirective applicative in Southern Interior languages, Halkomelem does not use -łc for possessive applied objects.
Salish applicatives 
, b. ni‘ qâ•›w´әl-әłc-t-әs łә słéni‘ ‘ә kâ•›wθә sәplíl. aux bake-rdr-tr-3sub det woman obl det bread ‘He baked the bread for the woman.’ (Gerdts 1988: 90)
While all of the Northern Interior languages have a single multi-purpose redirective suffix, Southern Interior languages have two or three different suffixes, each of which is used for a variety of redirective meanings. Nevertheless, as we discuss below, the redirective suffixes tend to align with applied objects bearing particular semantic roles. At first glance, it seems that the situation in Southern Interior Salish parallels the Northern Interior one: note that reflexes of all three redirective suffixes *-xi, -ł, and -tuł can appear in dative (48), benefactive (49), and possessive (50) applicatives:13 (48) a.
Kalispel (Carlson 1980: 25) , xâ•›wíc-š-t-әn łu‘ Agnes łu‘ t yámxˇâ•›we‘. give-rdr-tr-1sg.sub art Agnes art obl basket ‘I gave a basket to Agnes.’
b. Okanagan (N. Mattina 1993: 277) , kâ•›wu c-xâ•›wic-ł-t i-kł-lkalát. 1sg.obj asp-give-rdr-tr 1sg.poss-irr-bread ‘Give me what will be my bread.’ c.
Columbian (Willett 2003: 137)
yәrmәntúłn. //yәr-mi-tuł-t-n// push-rel-rdr-tr-1sg.sub ‘I pushed it to her/him.’
(49) a.
Okanagan (A. Mattina 1994: 211)
, ka‘kíc-x-t-m-әn t a-kł-qa‘xán. find-rdr-tr-2sg.obj-1sg.sub obl 2sg.poss-fut-shoes ‘I found you some shoes.’
b. Kalispel (Carlson 1972: 89)
‘ul-ł-cí-n. //wil-ł-t-si-en// burn-rdr-tr-2sg.obj-1sg.sub ‘I burned it for you.’
c.
Okanagan (A. Mattina 1994: 208) , , kâ•›wu sәq-túł-t-s i‘ slíp. 1sg.obj split-rdr-tr-3sub art wood ‘He split wood for me.’
.╅ We omit source applicatives from the discussion but include them in the summary below.
 Kaoru Kiyosawa & Donna B. Gerdts
(50) a.
Coeur d’Alene (Doak 1997: 167)
mé’â•›wšic. //me’â•›w-ši-t-Ø-s// break-rdr-tr-3obj-3sub ‘He broke something that belongs to another.’
b. Columbian (Kinkade 1980: 34)
má’â•›w-ł-c-xâ•›w ‘in-łkáp. break-rdr(-tr)-1sg.obj-2sg.sub 1sg.poss-pot ‘You broke my pot.’
c.
Okanagan (A. Mattina 1994: 211) , ti‘ c-әn-‘úłxâ•›w pit, uł uk-túł-t-s i‘ as.soon.as asp-loc-come.in Pete and see-rdr-tr-3sub art , ,, , tәtwít i‘ stәtqána‘k-s. boy art pistol-3sg.poss
‘As soon as Pete came in, he saw the gun that the boy had (not necessarily the boy’s gun).’
However, there are two types of evidence that suggest that the suffixes are in fact specialized semantically. First, when we look more closely at a sample of data, we find that the suffixes distribute according to function. As discussed in detail in Kiyosawa (2006), collecting examples of redirective applicatives from grammars and dictionaries of Southern Interior languages yields 210 tokens.14 The number of applied objects bearing each semantic role is given in Table 2: Table 2.╇ Applied Objects with Southern Interior Redirectives Redirective
Dative
Benefactive
Possessor
Source
Total
*-xi ╛↜-ł ╛↜-tuł
19 13 10
47 35 5
6 63 2
0 6 4
72 117 21
total
42
87
71
10
210
We can ascertain from these data which type of applied object tends to occur in applicative constructions with each redirective suffix. Reflexes of *-xi correlate with benefactives; the suffix -ł with possessors; and the suffix -tuł with datives. A ranked hierarchy of use can be given for each suffix:
.╅ All examples of the same verb root and suffix with the same translation in a language are counted as one token in this sample.
Salish applicatives 
(51) Hierarchies of applied object occurrence a.
Benefactive-oriented suffix: *-xi, Benefactive > Dative > Possessor
b. Possessive-oriented suffix: -ł Possessor > Benefactive > Dative > Source c.
Dative-oriented suffix: -tuł Dative > Benefactive > Source > Possessor
Benefactive meanings are the most common overall; they are the first or second most popular meaning for each suffix. Second, if we take examples of the redirective suffixes used with the same verb root in the same language, we find semantic contrasts. For example, in Okanagan, the verb root for ‘tie’ yields a benefactive applicative when suffixed with *-xi and a possessor applicative when suffixed with -ł. (52) Okanagan (N. Mattina 1993: 280)
, , Mary ’ac-xí-t-s i‘ t snkłca‘sqáxˇa‘ i‘ ttwit. Mary tie-rdr-tr-3sub art obl horse art boy ‘Mary tied the horse for the boy.’ , b. Mary ’ác-ł-t-s i‘ ttwit i‘ kәwáp-s. Mary tie-rdr-tr-3sub art boy art horse-3sg.poss ‘Mary tied the boy his horse.’15 a.
Also, we find that in Columbian the applied object is benefactive in the redirective constructions with reflexes of *-xi in (a), while it is dative in the -tuł redirective constructions in (b). (53) Columbian (Willett 2003: 136, 137) a.
kâ•›włnwílxtn. //kâ•›wułn=wil-xit-n// borrow=vehicle-rdr-1sg.sub ‘I borrowed a vehicle for her/him.’
b.
kâ•›włntúłn. //kâ•›wułn-tuł-t-n// lend-rdr-tr-1sg.sub ‘I loaned it to her/him.’
Similarly, in the following Okanagan examples, we see possessive applicatives with -ł in (a) contrasting with benefactive applicatives with -tuł in (b).
.╅ This translation from N. Mattina (1993) is intended to convey a possessor applicative meaning.
 Kaoru Kiyosawa & Donna B. Gerdts
(54) Okanagan (A. Mattina 1994: 208) , a. kâ•›wu síq-әł-t-s 1sg.obj split-rdr-tr-3sub ‘He split my wood.’ , b. kâ•›wu sәq-túł-t-s 1sg.obj split-rdr-tr-3sub ‘He split wood for me.’
, i-slíp. 1sg.poss-wood , i‘ slíp. art wood
In sum, although each redirective suffix in Southern Interior Salish languages can be used for a variety of meanings, there is nonetheless evidence that each tends to be associated with a particular semantics, and that there tends to be a single interpretation for any given combination of a verb root and a suffix. Thus the situation in languages with multiple redirective suffixes contrasts with that in languages with a single redirective suffix.
2.2.3╅ Summary: Benefactive as the core redirective concept There are ten different redirective suffixes in total in the Salish languages. Only *-xi can be reconstructed for Proto-Salish. Kiyosawa (2006) determines the core function of each applicative suffix by how frequently it appears in that function in the modern languages. She collected 447 tokens of verb plus suffix pairings from grammars and dictionaries and found the following distribution: Table 3.╇ Applied Objects with Redirectives Redirective
Benefactive
Dative
Possessor
Source
Total
*-xi 8 others
134 83
58 49
35 70
6 12
233 214
total
217
107
105
18
447
Notice that, overall, benefactive (or, more precisely, benefactive, delegative, and malefactive) is the most common meaning for redirectives, constituting half of the sample (217 of 447 tokens). Benefactive is also the most common meaning in the data with *-xi (134 out of 233 tokens or 58%). In sum, judging from the usage of its reflexes in the modern languages, ProtoSalish *-xi was probably associated with the semantic role of benefactive. But in many languages, its function has been extended to mark several kinds of applied objects, including datives, possessors, and sources. Thus, in these languages, its function is parallel to a general di-transitivizer, perhaps equivalent in semantic function to dative case in dependent-marking languages. Over time, other suffixes have been added to the redirective system in some branches or individual languages and these have usurped some of the functions of *-xi. The situation in Halkomelem is particularly interesting because it has no reflex of *-xi. The suffixes -as and -łc are most certainly
Salish applicatives 
innovative.16 The suffix -as expresses dative and -łc expresses benefactive applied objects, and the function and form and the suffixes have a one-to-one correspondence. Though the situation is not as clear-cut in Southern Interior as it is in Halkomelem, the redirective suffixes also tend to align with applied objects bearing particular semantic roles. The core function of the suffix -ł is to express a possessor as the applied object and that of the suffix -tuł is to express a dative applied object, though both suffixes are also used to express benefactives.
3.â•… Relational applicatives A second type of applicative construction, the relational applicative, is relevant to our discussion of benefactives and malefactives in Salish languages. A relational applicative construction adds a second argument to a clause whose non-applicative counterpart is intransitive. The resulting clause is a syntactically transitive construction in which a non-theme nominal is the direct object. Compare the intransitive construction in (55) with the applicative construction in (56): (55) Halkomelem (Gerdts 2004: 330) , , ni‘ nem kâ•›wθә swiw lәs. aux go det boy ‘The boy went.’ (56) Halkomelem (Gerdts 2004: 330) , ni‘ nәm-nәs-әs kâ•›wθә swiw lәs kâ•›wθә John. aux go-rel-3sub det boy det John ‘The boy went up to John.’
The clause in (55) is intransitive, while (56) is syntactically transitive, as evidenced by the third person ergative marker, and ‘John’, the goal of the motion, is the applied object. The semantic role of the applied object, goal in this case, is signaled by the relational suffix -nәs. Applied objects in relational applicatives are generally not semantic arguments of the predicate but rather have an indirect (or oblique) relationship to the event. For example, contrast the intransitive clause in (57) with the relational applicative in (58): (57) Halkomelem (Gerdts & Kiyosawa 2005b: 339) ni‘ cәn si‘si‘ ‘ә kâ•›wθә snәxâ•›wәł. aux 1sg.sub frighten obl det canoe ‘I was frightened at the car.’
.â•… The dative suffix â•‚as is grammaticalized from the lexical suffix ‘face’ (Gerdts & Hinkson 2004).
 Kaoru Kiyosawa & Donna B. Gerdts
(58) Halkomelem (Gerdts & Kiyosawa 2005b: 339)
, ni‘ cәn si‘si‘-me‘-t kâ•›wθә sqâ•›wәmey. aux 1sg.sub frighten-rel-tr det dog ‘I was frightened at the dog.’
In both sentences, the first-person subject is the experiencer of the psychological event. In (57), the stimulus of the event is expressed as an oblique, marked with the general oblique preposition ‘ә, but in (58) the stimulus is the applied object in a relational applicative construction, marked by the verbal suffix -me‘. The example in (58) is transitive, as seen by the presence of the transitive suffix on the verb. As discussed in Gerdts and Kiyosawa (2005a), the choice between expressing the NP as the object of a preposition or as an applied object depends on its discourse prominence and its animacy. Each Salish language has from one to three relational suffixes, and there are a total of seven different suffixes, as given in Table 4. Table 4.╇ Salish Relational Suffixes Branch
Sub-Branch
Bella Coola Central Salish
Tsamosan Tillamook Interior Salish
Northern Interior Salish
Southern Interior Salish
Language
Relational
Bella Coola Comox Sechelt Squamish Halkomelem Nooksack Northern Straits Klallam Lushootseed Twana Upper Chehalis Cowlitz Tillamook Lillooet Thompson Shuswap Okanagan Kalispel Coeur d’Alene Columbian
-m -mi, -ni -mí, -ni , -min, -ni -me‘/-mi‘, -nәs -ni -ŋiy, -nәs -ŋә, -nәs -bi, -di, -c/-s -ac -mis/-mn, -ni, -tas/-ts -mi(s), -ni, -t(a)s -әwi, -әs , -min/-min -mi -m(í) -min -mi -mi -mi
The most common relational suffix is reconstructed for Proto-Salish as *-mi by , Kinkade (1998) and reflexes of this form (-bi, -me‘, -mi, -mi‘, -min, -min, -mn, -mis, -ŋiy, -ŋә, -әwi) appear in all but three of the languages (Bella Coola, Nooksack, and
Salish applicatives 
Twana). Reflexes of the suffix *-mi are the only relational suffixes in Interior Salish languages, while most Central Salish languages and Tillamook have two relational suffixes, and Lushootseed and Tsamosan languages have three. The second most widespread relational suffix *-ni ( > -di, -ni) is attested in five of the Central Salish languages and in Tsamosan languages. The following sections briefly explore the semantics of the relational construction, especially their use to express benefactives and malefactives.
3.1â•… The semantics roles of relational applicatives In relational applicative constructions, the verb stem is usually intransitive, and the semantic role of the applied object is usually goal or direction of motion as in (59)–(60) or stimulus of a psychological or perceptual event as in (61)–(62).17 (59) Lillooet (Van Eijk 1987: 140) , , ka-łәxˇâ•›w-min-c-as a †ú‘ a. rslt-come.up-rel-1sg.obj-3sub rslt clt(well/so) rslt ‘He came right up to me.’ (60) Comox (Watanabe 1996: 335) , łukâ•›w-mi-θ-as. fly-rel-tr:1sg.obj-3sub ‘It’s flying towards me.’ (61) Okanagan (A. Mattina 1994: 221) ixí‘ ‘áyxˇâ•›w-t-mә-nt-s-әn. there tired-st-rel-tr-2sg.obj-1sg.sub ‘I am tired of you.’ (62) Halkomelem (Gerdts & Kiyosawa 2005b: 334) , , , cәq-me‘-t č ce‘ kâ•›wθә nәcәwmәxâ•›w ‘i ce‘ tecәl. surprise-rel-tr 2sg.sub fut det visitor aux fut arrive ‘You will be surprised at the visitors when they arrive.’
While there is a tendency for a suffix to be associated to a particular meaning within a language, across languages the same suffix is used to express various meanings.
.╅ See Gerdts (2004) for a discussion of Halkomelem directional applicatives. Gerdts and Kiyosawa (2005b) treat psych applicatives in Halkomelem, and Gerdts and Kiyosawa (2004) treat psych applicatives in Salish languages in general.
 Kaoru Kiyosawa & Donna B. Gerdts
The result is that the same meaning may be expressed by different relational suffixes in different languages, for example -mi in (63), -ni in (64), and -ac in (65): (63) Sechelt (Beaumont 1985: 104) , čásxˇém-mí-t-á-čexâ•›w te ‘úłqay? afraid-rel-tr-q-2sg.sub det snake ‘Are you afraid of the snake?’ (64) Nooksack (Galloway 1997: 222) ‘ił ‘a(s)-sí(y)‘say‘-ni-θ-as. aux st-afraid-rel-tr:1sg.obj-3sub ‘He’s afraid of me.’/‘He doesn’t trust me.’ (65) Twana (Kinkade n.d.) , , as-cípal-ac-bәš. st-fear-rel-1sg.obj ‘He’s afraid of me.’
In sum, the match of form to function in relational applicatives is complicated both within and across languages, like the situation found with the redirective applicatives.
3.1.1â•… Benefactive Motion verbs and psychological predicates account for the majority of examples of relational applicatives. However, relational suffixes also attach to a few activity verbs to form applicatives in which the applied object is the benefactive: (66) Halkomelem (Gerdts 1988: 144) ni‘ kâ•›wukâ•›w-me‘-θám‘š-әs. aux cook-rel-tr:1sg.obj-3erg ‘He cooked for me.’ (67) Tillamook (Egesdal & M. Thompson 1998: 253) , dә s-‘isleš-әw í-t-әw. art nm-sing-rel-tr-pass ‘Someone is singing for him.’ (68) Lillooet (Van Eijk 1997: 125) , , kâ•›wzús-min work-rel ‘to work for, to look after’ (69) Thomspon (L. Thompson & M. Thompson 1992: 75) , kâ•›wz=ús-m-me-ne. rough=face-mdl-rel(-tr)-1sg.sub ‘I work hard for him.’
Salish applicatives 
(70) Halkomelem (Gerdts, f.n.) ya:ys-me‘-t work-rel-tr ‘work for him/her’/‘work on him/her’ [e.g. a spiritualist working a cure]
Note that the preceding Halkomelem data can also have the meaning ‘work on’. The same use of ‘work’ with a relational suffix in found in Nooksack: (71) Nooksack (Adams et al. 2005: 9) ‘әháy-ni-t-әs tә i-ł xˇ´әł work-rel-tr-3sub art in-past sick ‘he will do work on the sick’
In sum, the benefactive use of the relational suffix is limited to at most a few verbs per language. For example, we have found only two in Halkomelem. Thus, these forms are best analyzed as lexicalized. Some Salish languages make use of a second strategy to express benefactives based on intransitive verbs; they attach a redirective suffix, which by definition should attach only to a transitive base, to activity predicates to form relational applicatives in which the applied object is a benefactive: (72) Lillooet (Van Eijk 1987: 312) , ‘í†-әm-xit sing-mdl-rdr ‘to sing for someone’ (73) Upper Chehalis (Kinkade 1991: 372) ‘it yús-š-c. perf work-rdr-tr:1sg.obj ‘He/she worked for me.’ (74) Comox (Watanabe 1996: 331) čił-im-‘әm-t-umuł-as Mary. dance-mdl-rdr-tr-1pl.obj-3sub Mary ‘Mary danced for us.’ (75) Comox (Watanabe 2003: 252) , , paya‘ štәm t ča-čah-am-‘әm-t-anapi. always 1pl.sub clt red(impf)-pray-mdl-rdr-tr-2pl.obj ‘We will always be praying for you (pl.).’
Again, this function is highly lexicalized, appearing on at most a couple of verbs per language.
 Kaoru Kiyosawa & Donna B. Gerdts
The inability of most intransitive verbs to form benefactive applicatives in Salish languages is not surprising given the cross-linguistic propensity noted by Shibatani (1996) for benefactive constructions to favor transitive bases.
3.1.2â•… Malefactive While relational applicatives are used in a limited way to express benefactives, their use for malefactive meanings is much more robust. Actually, many of the psychological events expressed as relational applicatives are negative ones. Take, for example, the list of psychological predicates occurring with the suffix -me‘ in the Island dialect of Halkomelem. Some are positive or neutral in meaning, depending on the context: , , , c әqme‘t ‘astonished, surprised at’, hilәkâ•›wme‘t ‘happy for’, ‘iyәsme‘t ‘happy for’, si‘әm me‘t ,w ‘respect’, hek â•› me‘t ‘remember’, siwәlme‘t ‘sense’, xâ•›wθtiwәnme‘t ‘think, decide about’, , štә‘e:wәnme‘t ‘think that way about’, xâ•›wqâ•›wәlәwәnme‘t ‘think about’. However, more , denote negative experiences: si‘si‘me‘t ‘afraid, frightened of ’, q,elme‘t ‘believe (lies)’, w xˇi‘xˇe‘me‘t ‘embarrassed, shy of , , ’,, kâ•› iłәme‘t ‘fed up with’,, ,mel,qme‘t ‘forget about’ , , , w wәwistәnәqme‘t ‘jealous of ’, sәl sәl qâ•› me‘t ‘lonely, sad for’, tetiyәqmәt ‘mad at’, qәl me‘t , ,, , , , , ‘miss’, qilәsme‘t ‘sad for’, sәwsәwme‘t ‘sad for’, t θәyk â•›wme‘t ‘startled at’, k â•›welәk â•›wme‘t , ‘suspicious of ’, qsәme‘t ‘tired of waiting for’, łciwsme‘t ‘tired of ’. The positive or negative effect of the event is determined by the denotation of the predicate; it is the subject and not the applied object that is affected. However, in other cases, the malefactive meaning is not supplied by the verb, which does not carry a negative connotation on its own, and thus the negative effect clearly arises as a result of the relational suffix. (76) Comox (Watanabe 1996: 337) qâ•›way-mi-θi tθәm. talk-rel-tr:2sg.obj 1sg.sub:fut ‘I’ll scold you.’ (77) Lushootseed (Hess & Bates 2004: 183) , , yәc-bí-d ti čáčas. tell-rel-tr det child ‘She told on the boy (and made a good story of it).’ (78) Klallam (Montler 1996: 262) , ‘әná-nәs-әŋ cn ‘a‘ cә sqáxˇә‘. come-rel-pass 1sg.sub obl det dog ‘The dog came at me.’ (79) Comox (Watanabe 1996: 335) łag-a-θut-mi-θ-as. leave-lv-tr:refl-rel-tr:1sg.obj-3sub ‘He walked/ran out on me.’/‘He ran away from me.’
Salish applicatives 
(80) Lushootseed (Hess & Bates 2004: 181) ‘úkâ•›wukâ•›w-bi-t-s. play-rel-tr-1sg.obj ‘They made fun of me.’ (81) Squamish (Kuipers 1967: 351) , na qxâ•›w=ús-mi-nt-as-wit. aux gathered=face-rel-tr-3sub-pl ‘They ganged up on him.’
The malefactive use of relational sufixes has been previously noted by only two researchers. Beaumont (1985: 105) says the Sechelt suffix -ni is used when the action performed by the subject “works to the disadvantage of someone else”, and Kuipers (1974: 46) notes that the Shuswap suffix -m(í) refers to an object that is affected indirectly, superficially, or malefactively by the action. However, it becomes obvious when data are assembled from across languages.
3.1.2.1â•… Source and malefaction.â•… Another class of verbs that form relational applicatives is the transfer predicates. In many Salish languages, transfer predicates are intransitive (often in middle or autonomous voice) rather than transitive, and thus form relational rather than redirective applicatives: (82) Sechelt (Beaumont 1985: 104) , qâ•›wímels-ni-t-št-kâ•›wa čems syíyaya ‘e če ‘úpan=ús. borrow-rel-1pl.sub-fut det:1pl.poss friend obl det ten=round.object ‘We’re going to borrow ten dollars from our friends.’ (83) Squamish (Kuipers 1967: 343) na kâ•›wúł(n)-ni-t-c-as. aux borrow-rel-tr-1sg.obj-3sub ‘He borrowed it from me.’ (84) Lushootseed (Bates et al. 1994: 172) qáda-di-d steal-rel-tr ‘steal from someone’ (85) Comox (Watanabe 2003: 256) , čәw u-ni-θ-as ‘ә tә tθ tala. steal-rel-tr:1sg.obj-3sub obl det 1sg.poss money ‘He stole money from me.’
As was pointed out in Section 2.1.2.2 above, redirectives with source applied objects often carry a malefactive meaning. This is because the applied object is being deprived of the theme.
 Kaoru Kiyosawa & Donna B. Gerdts
3.1.2.2â•… Adversative with natural phenomena.â•… Relational applicatives can be formed on predicates (nouns or verbs) denoting natural phenomena to express an adverse effect, such as discomfort from inclement weather, on the applied object: (86) Thompson (L. Thompson & M. Thompson 1992: 74) , c‘oz-mí-nt-i-s. dark-rel-tr-1pl.obj-3sub ‘It gets dark on us.’ (87) Comox (Watanabe 2003: 257) , čәł-ni-θay-әm. rain-rel-tr:1sg.obj-pass ‘I got rained on.’ (88) Halkomelem (Gerdts & Kiyosawa 2005b: 331) sqâ•›wәlqâ•›walxˇâ•›w-me‘-t-әm hail-rel-tr-pass ‘(he/she/it) get hailed on’ (89) Halkomelem (Gerdts & Kiyosawa 2005b: 331) , yәq-me‘-t-әm snow-rel-tr-pass ‘(he/she/it) get snowed on’
One example from Lillooet shows that the relational suffix can be suffixed to a noun to convey an attack by an animal: (90) Lillooet (Van Eijk 1997: 122) mixˇał-mín-әm bear-rel-pass ‘it was eaten by a bear, he was met by a bear, ran into a bear’
Relational applicatives of nature verbs tend to be used in the passive, rather than the active. Whatever the voice, the passive is agentless: the agent (or force) is part of the meaning expressed by the predicate. There is a small set of adversative constructions in some Salish languages that have similar properties: they always appear in an agentless passive, never as an active. For example, the Halkomelem roots for ‘get drowsy’ or ‘have a nightmare’ never appear as , active transitive forms (*cәtxˇâ•›wt, *xˇәyxˇayәk â•›wәst) but only in the passive: (91) Halkomelem (Gerdts, f.n.)
, , cәtxˇâ•›w-θa:m ce‘ ‘әw ni:xâ•›w mәq ‘ә θә xˇixâ•›wә. drowsy-tr:2pass fut lnk aux.2sub full obl det sea.urchin ‘You’re going to get drowsy when you get full of the sea urchin.’
Salish applicatives 
(92) Halkomelem (Gerdts, f.n.) , xˇәyxˇayәkâ•›wәs-θelәm. nightmare-tr:1pass ‘I had a nightmare.’ (93) Halkomelem (Gerdts, f.n.) , , , q, әlq,әlpnәctәm ‘get a cramp’ kâ•›wәlәtәm ‘get diarrhea’ cxâ•›watәm ‘be swollen, bloated’ (94) Squamish (Kuipers 1967: 83–84) , , tut,łmántm ‘be flea-ridden’ , c´әkâ•›wntm ‘be worm-infested’ , w ántm ‘ix ‘be infested with maggots’ , ˇ,â•› ic , čәčičántm ‘have convulsions’ xˇišačí‘ntm ‘have cramp in the arm’ ,, xˇәpk, â•›wántm ‘be in pain’ , xˇәpkâ•›wíwsntm ‘be rheumatic’
This is a special use of the passive. Normally, Salish passives carry no implication of adversative; that is, they are not ‘suffer’ passive constructions like those found in some Asian languages.18
3.2â•… The mapping of form and function Even though the distribution and usage of the suffixes paints a complex picture, several generalizations emerge from the above discussion. The concept of relational applicative, i.e. adding a non-theme participant as a core argument, thereby changing an intransitive verb into a transitive verb, must be very old in Salish, probably going back to Proto-Salish. Given the robustness of the suffix *-mi, in terms of the number of different branches that have reflexes of this suffix, the wide range of verb classes that they attach to, and the different semantic roles of the applied objects associated with them, it is likely that this morphology was associated with the relational applicative construction in Proto-Salish. The exact nature of the semantics of Proto-Salish *-mi can be debated. Was it a general transitivizer devoid of semantics functioning simply to license an object? Or was it associated with a particular verb class or verb classes, as reflected in the modern languages? The former would parallel its current use in the Northern Interior Salish languages. The latter would parallel its use in Central Salish languages like Halkomelem. In Halkomelem, *-mi is more productive on psych predicates and
.╅ See the papers by Radetzky and Smith and by Smith in this volume.
 Kaoru Kiyosawa & Donna B. Gerdts
other verbs of internal experience, and it is most commonly associated with applied objects with the semantic role of stimulus.19 In either case, the function of *-mi has changed over time, expanding and/or contracting in its range of meaning in the various languages. Other suffixes that have been added to the relational system in some languages have usurped some of the functions of *-mi or added other functions to the relational applicative system. The suffix *-ni may originally have been associated with transfer verbs, perhaps indicating a concept like source or malefactive. Other relational suffixes have minor usage within the family, usually localized to one sub-branch. In languages with multiple relational suffixes, there is often considerable overlap in the functions of the different relationals; there is not a one-to-one relationship between form and function. The layering of the relational applicative suffixes over time has created a complex system of relational applicatives in the modern Salish languages. The use of relational applicatives to express benefactive meanings is limited to only a few activity verbs. Several languages use redirective suffixes instead to express such meanings. In contrast, malefactive readings of relational applicatives are widely attested. As with redirective applicatives, the malefactive meaning may be tied in with the notion of deprivation of source applied objects of transfer verbs. Negative psychological events and adverse natural phenomena are also expressed with relational applicatives. The latter often appear as agentless passives, which are used in some Salish languages to express adverse events.
4.╅ Non-applicative expressions of benefactives In Salish languages, applicative constructions are the usual means of expressing benefactives, though it is also possible to use periphrastic constructions (Section€4.1). The only other morphological means used for expressing benefactives is the causative (Section 4.2).
. Gerdts and Hinkson (1996), approaching the problem from a Halkomelem internal viewpoint, posit the relational applicative to be ╂min, and speculate that it grammaticalized from the instrumental suffix ╂min, which was probably a lexical suffix historically. Gerdts and Kiyosawa (2005b) argue that instruments of activity verbs are semantically parallel to indirect causes or stimuli of psychological and perception predicates. Therefore, instrumental morphology came to be used for applicatives. This is also attested in Chickasaw (Munro 2000:€292).
Salish applicatives 
4.1â•… Periphrastic expressions of benefactive Dative and benefactive redirective constructions are obligatory in the sense that there is no non-applicative equivalent in which the theme occurs as an object and the applied object occurs as an oblique NP. So, for example, the benefactive in (95) cannot be expressed as a prepositional phrase, as in (96): (95) Halkomelem (Gerdts, f.n.) , , qâ•›wәl-әłc-θamә cәn ce‘ ‘ә kâ•›w sce:łtәn. bake-rdr-tr:2obj 1sub fut obl det salmon ‘I will bake some salmon for you.’ (96) Halkomelem (Gerdts, f.n.) , , , *qâ•›wәl-әt cәn ce‘ kâ•›w sce:łtәn ‘ә † nәwә. bake-tr 1sub fut det salmon obl det 2emph ‘I will bake some salmon for you.’
Nevertheless, it is possible to break down the two components of the event—the effect on the theme and the transfer of possession or benefit—and express each as a separate predicate. This can be accomplished by means of a serial verb construction as in (97) or conjoined clauses as in (98).20 (97) Halkomelem (Gerdts, f.n.) , , , qâ•›wәl-әt cәn ce‘ kâ•›w sce:łtәn xˇâ•›wte‘ ‘ә † nәwә. bake-tr 1sub fut det salmon go.toward obl det 2emph ‘I will bake some salmon for you.’ (98) Halkomelem (Gerdts, f.n.) , , qâ•›wәl-әt cәn ce‘ kâ•›w sce:łtәn ‘i‘ nił s-we‘-stamә (ce‘). bake-tr 1sub fut det salmon and 3emph nm-own-cs:2obj fut ‘I will bake some salmon and it will be for you.’
Circumlocutions can also be used to accommodate a dative and a benefactive in the same clause; Halkomelem does not allow more than one applicative suffix per verb. (99) Halkomelem (Gerdts, f.n.)
, ni‘ cәn ‘am-әs-t łәn ten ‘ә kâ•›wθә pukâ•›w aux 1sub give-rdr-tr det:2poss mother obl det book nił s-we‘-stamәt 3emph nm-own-cs:2obj ‘I gave your mother the book that is for you.’
.â•… Periphrastic constructions are also available for delegative and malefactive meanings.
 Kaoru Kiyosawa & Donna B. Gerdts
In the above example, information about the benefactive is given as a relative clause modifying the theme.
4.2â•… Causatives used as benefactives The causative suffix is used in a very limited set of cases to express benefactive meanings in at least two Salish languages—Bella Coola and Halkomelem. For example, the causative suffix in Bella Coola functions like a relational applicative, attaching to intransitive stems:21 (100) Bella Coola (Davis & Saunders 1997: 28) nuyamł-tu-s ti ‘imlk tx ti ‘immllki: tx. sing-cs-3sub art man dem art boy dem ‘The man made/let the boy sing.’/‘The man sang for the boy.’ (101) Bella Coola (Davis & Saunders 1997: 75) , tx-a-tu-s mat ‘aleks ti qlsxâ•›w tx. cut-intr-cs-3sub Matt Alex art rope dem ‘Matt made Alex cut the rope.’/‘Matt let Alex cut the rope.’/‘Matt cut the rope for Alex.’
In the above examples, both benefactive and causative meanings are possible. The benefactive reading of the object with the causative suffix is not always available. According to Davis and Saunders (1997: 28), it is only available for verbs that “impute the greater motility” to their objects. If the verb involves “any spontaneous, self-initiated performance”, the clause has a causative meaning only, not a benefactive (or delegative) one: (102) Bella Coola (Davis & Saunders 1997: 75) , qs-tu-s ti nus‘u:lxˇ tx ti ‘imlk tx. ill-cs-3sub art thief dem art man dem ‘The thief made the man ill.’ (*‘The thief was ill in the man’s place.’)
Gerdts and Hukari (2006) have also noted that the causative suffix in Halkomelem can be attached to a small class of transitive verbs to express a benefactive meaning. Compare the transitive (a) examples, in which the object is a source (103) or a goal (104), with the causative (b) examples, in which the object is a benefactive: (103) Halkomelem (Gerdts & Hukari 2006: 142–3) , , , a. nił łwet kâ•›wә ni‘ qe‘әn-t tθәn silә 3emph who det aux steal-tr det:2poss grandparent , ‘ә kâ•›wθә sew әn-s? obl det lunch-3poss
‘Who stole your grandfather’s lunch from him?’
.â•… Zúñiga (in prep.) also mentions the benefactive use of Bella Coola causatives, citing (100).
Salish applicatives 
, , , , , b. nem č ce‘ qәn-stәxâ•›w tθәn sәlsilә go 2sg.sub fut steal-cs det:2poss grandparent(pl) , ‘ә kâ•›wθә sciyә. obl det strawberry
‘You’re going to steal some strawberries for your grandparents.’
(104) Halkomelem (Gerdts & Hukari 2006: 143) , , a. cala‘ł-t č tθәn men ‘ә θәn snәxâ•›wәł. borrow/lend-tr 2sg.sub det:2poss father obl det:2poss canoe/car ‘Lend your father your car.’ , b. ni‘ ‘ә č cala‘ł-stәxâ•›w kâ•›wθә John ‘ә kâ•›w telә? aux q 2sg.sub borrow/lend-cs det John obl det money ‘Did you borrow some money for John?’
The causative suffix thus functions like a redirective applicative in these examples. A second class of examples with this use of causative involves denominal verb constructions. Halkomelem denominal verbs (Gerdts & Hukari 2004, 2008), formed by prefixing a verbalizer such as c- ‘make, have, get’ or txâ•›w- ‘buy’ to a noun, can take the causative suffix, yielding a benefactive, not a causative, meaning: (105) Halkomelem (Gerdts & Hukari 2004: 206) , ni‘ c-tθele‘-stәxâ•›w-әs tθә šәšiyәł-s. aux vbl-heart-cs-3sub det elder.sibling(pl)-3sg.poss ‘He made hearts for his older brothers.’ (106) Halkomelem (Gerdts & Hukari 2004: 206) , , , nem c-xˇәltәn-stamš! go vbl-pencil-cs:1obj ‘Go get me a pencil!’ (107) Halkomelem (Gerdts & Hukari 2004: 206)
, , ni‘ txâ•›w-sәplil-stәxâ•›w-әs łә słeni‘ kâ•›wθә memәnә-s. aux vbl-bread-cs-3sub det woman det children-3sg.poss ‘The woman bought bread for her children.’
Bella Coola denominal verbs formed by prefixing a verbalizer such as tam- ‘make, construct’ (Nater 1984: 93) also form benefactives by means of the causative suffix: (108) Bella Coola (Nater 1984: 40) , tam-cla-tu-m-xˇ! vbl-basket-cs-1sg.obj-imp ‘Make me a basket!’ (109) Bella Coola (Nater 1984: 40) , tam-‘akâ•›wna-tu-ti-m. vbl-sea.canoe-cs-3pl.obj-pass ‘Somebody made them a sea canoe.’
 Kaoru Kiyosawa & Donna B. Gerdts
Salish denominal verb constructions are syntactically intransitive, and thus the causative suffix functions like a relational applicative in these examples. In sum, we see that the causative suffix functions, at least in a limited way, as a benefactive suffix in Bella Coola and Halkomelem. For causative and applicative constructions to share morphology is not unexpected because both involve valenceincreasing operations (from intransitive to transitive, or from transitive to ditransitive). Further, both take non-theme objects: the causee and the benefactive respectively. Causees in many languages, like most applied objects in Salish languages, tend to be higher animates. In fact, in some languages, e.g. Hualapai (Ichihashi-Nakayama 1996), Mapudungun (Zúñiga this volume) Orizaba Nahuatl (Tuggy 1996), and Olutec (Zavala 2000), the same suffix can be used for both causative and applicative constructions.
5.â•… Conclusion Evaluating events as benefactive or malefactive is a central part of the human experience, and thus languages have a way of encoding this information. The usual means for expressing them in Salish languages is with applicative constructions, in which the affected person is cast as a clausal argument. Periphrastic expressions of benefaction and malefaction are rare and considered circumlocutions (cf. Section 4.1). Kiyosawa (2006) gives a thorough treatment of the form and function of the applicative suffixes found in twenty Salish languages. Applicative constructions can be grouped into two types. Redirective applicatives are formed on transitive bases and have dative, benefactive, source, and possessive applied objects (Section 2.1). Relational applicatives are formed on intransitive bases and have stimulus, goal, benefactive, and malefactive applied objects (Section 3.1). Each Salish language has at least one redirective suffix and one relational suffix. Only two applicative suffixes can be reconstructed for Proto-Salish: the redirective *-xi and the relational *-mi. The other suffixes have been innovated in sub-branches or individual languages. The innovated applicatives usurp or augment the functions of the two Proto-Salish applicatives, yielding a complex picture in the modern languages. Most redirective suffixes allow the addition of a benefactive applied object to a transitive event. In fact, as discussed in Section 2.2.3, benefactive was probably the central meaning of the Proto-Salish redirective suffix *-xi. Its use then spread to other types of applied objects. Only half of the Salish languages use applicatives for possessive applied objects (Section 2.1.4). Some branches have added new redirective suffixes for this purpose. Possessive applicatives in Salish usually carry an extra semantic “kick”: the possession (or deprivation of possession) of an object connotes a positive or negative effect on the possessor. This leads to a confusing range of translations for each example.
Salish applicatives 
Adding benefactive applied objects to transitive verbs is robustly attested in all Salish languages and seems to be generalized across all verb classes. The situation with intransitive verbs is more limited (see Section 3.1.1). Only a handful of activity verbs, with meanings like “sing for”, “pray for”, and “work for”, are used in this way. Some languages use relational suffixes, some use redirective suffixes, and at least one, Bella Coola, uses a causative suffix (see Section 4.2) for these meanings. Moreover, a single language may use different suffixes on different verbs. Therefore, the benefactive forms of intransitive verbs seem to be lexicalized. The precise interpretation of a redirective applicative—benefactive, delegative, or malefactive—is supplied by the context (see Section 2.1.2). Verb class semantics also contributes to the meaning. For example, transfer verbs with source applied objects (appearing with translations like “steal”, “take from”, etc.) often carry the connotation of deprivation and hence malefaction (see Section 2.1.3). The situation with transfer verbs is complicated by the fact that many Salish languages cast transfer verbs as formally intransitive constructions, for example, in middle or autonomous voice. In this case, they tend to use a relational rather than a redirective applicative to express the source (see Section 3.1.2.1). The relational suffix *-ni, which appears in two branches, seems in particular to be linked to source and malefactive applied objects. While benefactive meanings with relational applicatives are quite rare, malefactive meanings are more widely attested. Judging from its behaviour in Central Salish, the Proto-Salish relational suffix *-mi was probably most closely tied with the notion of the stimulus in a psychological predicate, often with a negative meaning. Especially in the case of speech act verbs, we see that the relational suffix adds a malefactive meaning (see Section 3.1.2). Also observed is the wide-spread use of relational suffixes to express the adverse effect of natural phenomena (see Section 3.1.2.2). These constructions are frequently cast in the passive voice and thus tie in with a small set of agentless passives with negative effect found in some Salish languages. In conclusion, expressing benefactive and malefactive meanings in Salish languages is a complex topic because no single morphological form is used to indicate them, and furthermore they are not morphologically differentiated from each other. However, compiling and examining benefactive and malefactive examples has proven to be an insightful exercise in understanding the use, history, development, and limits of the Salish applicative system.
References Adams, George, Galloway, Brent & Renteria, Catalina. 2005. A Nooksack story from 1956, how to become an Indian doctor. Papers for the 40th International Conference on Salish and Neighbouring Languages, University of British Columbia Working Papers in Linguistics 16:€1–31. Aissen, Judith L. 1987. Tzotzil Clause Structure. Dordrecht: Reidel.
 Kaoru Kiyosawa & Donna B. Gerdts Bates, Dawn, Hess, Thom & Hilbert, Vi. 1994. Lushootseed Dictionary. Seattle WA: University of Washington Press. Beaumont, Ronald C. 1985. She Shashishalhem, the Sechelt Language: Language, Stories and Sayings of the Sechelt Indian People of British Columbia. Penticton BC: Theytus Books. Beck, David. 2006. Control of agreement in multi-object constructions in Upper Necaxa Totonac. In Proceedings of the 11th Workshop on Structure and Constituency in the Languages of the Americas, Atsushi Fujimori & Maria Amelia Reis Silva (eds). Vancouver BC: UBC Working Papers in Linguistics. Carlson, Barry F. 1972. A Grammar of Spokane: A Salish Language of Eastern Washington. Ph.D. dissertation, University of Hawaii. Carlson, Barry F. 1980. Two-goal transitive stems in Spokane Salish. International Journal of American Linguistics 46: 21–26. Carlson, Barry F. & Flett, Pauline. 1989. Spokane Dictionary [Occasional Papers in Linguistics 6]. Missoula MA: University of Montana Press. Davis, Philip W. & Saunders, Ross. 1997. A Grammar of Bella Coola. [Occasional Papers in Linguistics 13]. Missoula MT: University of Montana Press. Doak, Ivy Grace. 1997. Coeur d’Alene Grammatical Relations. Ph.D. dissertation, University of Texas at Austin. Donohue, Mark. 1999. A Grammar of Tukang Besi. Berlin: Mouton de Gruyter. Driever, Dorothea. 1976. Aspects of a Case Grammar of Mombasa Swahili [Hamburger Philologische Studien 43]. Hamburg: Helmut Buske. Egesdal, Steven M. & Thompson, M. Terry. 1998. A fresh look at Tillamook (Hutéyu) inflectional morphology. In Salish Languages and Linguistics: Theoretical and Descriptive Perspectives, Ewa Czaykowska-Higgins & M. Dale Kinkade (eds.), 235–273. Berlin: Mouton de Gruyter. Fillmore, Charles J. 1968. Lexical entries for verbs. Foundations of Language 4: 373–393. Foley, William. 1991. The Yimas Language of New Guinea. Stanford CA: Stanford University Press. Fried, Mirjam. 1999. From interest to ownership: A constructional view of external possessors. In External Possession and Related Noun Incorporation Phenomena, Doris L. Payne & Immanuel Barshi (eds.), 137–163. Amsterdam: John Benjamins. Galloway, Brent D. 1997. Nooksack pronouns, transitivity, and control. Papers for the 32nd International Conference on Salish and Neighbouring Languages, 197–243. Port Angeles WA. Gardiner, Dwight. 1993. Structural Asymmetries and Pre-verbal Positions in Shuswap. Ph.D. dissertation, Simon Fraser University. Gerdts, Donna B. 1988. Object and Absolutive in Halkomelem Salish. New York NY: Garland. Gerdts, Donna B. 2004. Halkomelem directional applicatives. Papers for the 39th International Conference on Salish and Neighbouring Languages, University of British Columbia Working Papers in Linguistics 14: 189–199. Gerdts, Donna B. To appear. Ditransitive constructions in Halkomelem Salish: A direct object/ oblique object language. In Studies in Ditransitive Constructions, Andrej Malchukov, Martin Haspelmath, & Bernard Comrie (eds.). Berlin: Mouton. Gerdts, Donna B. & Hinkson, Mercedes Q. 1996. Salish lexical suffixes: A case of decategorialization. In Proceedings of the Conference on Conceptual Structure, Discourse, and Language, Adele Goldberg (ed.), 163–176. Stanford CA: CSLI. Gerdts, Donna B. & Hinkson, Mercedes Q. 2004. The grammaticalization of Halkomelem ‘face’ into a dative applicative suffix. International Journal of American Linguistics 70: 227–250.
Salish applicatives 
Gerdts, Donna B. & Hukari, Thomas E. 2004. Halkomelem denominal verbs. Papers for the 39th International Conference on Salish and Neighbouring Languages, University of British Columbia Working Papers in Linguistics 14: 201–214. Gerdts, Donna B. & Hukari, Thomas E. 2006. Classifying Halkomelem causatives. Papers for the 41st International Conference on Salish and Neighbouring Languages, University of British Columbia Working Papers in Linguistics 18: 129–145. Gerdts, Donna B. & Hukari, Thomas E. 2008. Halkomelem denominal verbs. International Journal of American Linguistics 74: 489–510. Gerdts, Donna B. & Kiyosawa, Kaoru. 2004. Salish psych applicatives. In Proceedings of the Sixth Workshop on American Indigenous Languages, Jeanie Castillo (ed.), 28–42. Santa Barbara CA: Santa Barbara Papers in Linguistics. Gerdts, Donna B. & Kiyosawa, Kaoru. 2005a. Discourse functions of Salish applicatives. Papers for the 40th International Conference on Salish and Neighbouring Languages, University of British Columbia Working Papers in Linguistics 16: 98–124. Gerdts, Donna B. & Kiyosawa, Kaoru. 2005b. Halkomelem psych applicatives. Studies in Language 29(2): 329–362. Gerdts, Donna B. & Kiyosawa, Kaoru. 2007. Combinatorial properties of Salish applicatives. Papers for the 42nd International Conference on Salish and Neighbouring Languages, University of British Columbia Working Papers in Linguistics 20: 176–219. Gerdts, Donna B. & Whaley, Lindsay. 1993. Kinyarwanda Multiple Applicatives and the 2-AEX. In Papers from the Twenty-Eighth Regional Meeting of the Chicago Linguistic Society, Costas P. Canakis, Grace P. Chan & Jeanette Marshall Denton (eds.), 186–205. Chicago IL: University of Chicago. Givón, Talmy. 1984. Syntax: A Functional-Typological Introduction, Vol. I. Amsterdam: John Benjamins. Hess, Thomas M. 1967. Snohomish Grammatical Structure. Ph.D. dissertation, University of Washington. Hess, Thom & Bates, Dawn. 2004. Lushootseed applicatives and their ilk. In Studies in Salish Linguistics in Honor of M. Dale Kinkade [Occasional Papers in Linguistics 17], Donna B. Gerdts & Lisa Matthewson (eds.), 172–196. Missoula MT: University of Montana Press. Ichihashi-Nakayama, Kumiko. 1996. The ‘applicative’ in Hualapai: Its functions and meanings. Cognitive Linguistics 7(2): 227–239. Kimenyi, Alexandre. 1980. A Relational Grammar of Kinyarwanda. Berkeley CA: University of California Press. Kinkade, M. Dale. 1980. Columbian Salish -xí, -ł, -túł. International Journal of American Linguistics 46: 33–36. Kinkade, M. Dale. 1982. Transitive inflection in Moses-Columbian Salish. Kansas Working Papers in Linguistics 7: 49–62. Kinkade, M. Dale. 1991. Upper Chehalis Dictionary [Occasional Papers in Linguistics 7]. Missoula MT: University of Montana Press. Kinkade, M. Dale. 1998. Comparative Salish seminar. Course notes, University of British Columbia. Kinkade, M. Dale. 2004. Cowlitz Dictionary and Grammatical Sketch [Occasional Papers in Linguistics 18]. Missoula MT: University of Montana Press. Kinkade, M. Dale. No date. Twana field notes, June 1972. Kiyosawa, Kaoru. 2004. The distribution of possessive applicatives in Interior Salish Languages. Papers for the 39th International Conference on Salish and Neighbouring Languages, University of British Columbia Working Papers in Linguistics 14: 241–252.
 Kaoru Kiyosawa & Donna B. Gerdts Kiyosawa, Kaoru. 2006. Applicatives in Salish Languages. Ph.D. dissertation, Simon Fraser University. Kuipers, Aert H. 1967. The Squamish Language: Grammar, Texts, Dictionary. The Hague: Mouton. Kuipers, Aert H. 1974. The Shuswap Language: Grammar, Texts, Dictionary. The Hague: Mouton. Kuipers, Aert H. 1992. The Shuswap complex transitivizers. Papers for the 27th International Conference on Salish and Neighbouring Languages, 49–53, Kamloops BC. Mattina, Anthony. 1994. -tułt, and more on Okanagan transitive forms: A working paper. Papers for the 29th International Conference on Salish and Neighbouring Languages, 204–231, Pablo MT. Mattina, Nancy. 1993. Some lexical properties of Colville-Okanagan ditransitives. Papers for the 28th International Conference on Salish and Neighbouring Languages, 265–284. Seattle WA. Montler, Timothy R. 1986. An Outline of the Morphology and Phonology of Saanich, North Straits Salish [Occasional Papers in Linguistics 4]. Missoula MT: University of Montana Press. Montler, Timothy R. 1996. Some Klallam paradigms. Papers for the 31st International Conference on Salish and Neighbouring Languages, 257–263. Vancouver BC. Munro, Pamela. 2000. The leaky grammar of the Chickasaw applicatives. Papers from the Regional Meetings, Chicago Linguistic Society 36: 285–310. Nater, Hank F. 1984. The Bella Coola Language [Mercury Series Paper 92]. Ottawa: National Museum of Man, Canadian Ethnology Service. Peterson, David A. 2007. Applicative Constructions. Oxford: Oxford University Press. Rude, Noel. 1985. Studies in Nez Perce Grammar and Discourse. Ph.D. dissertation, University of Oregon. Shibatani, Masayoshi. 1996. Applicatives and benefactives: A cognitive account. In Grammatical Constructions: Their Form and Meaning, Masayoshi Shibatani & Sandra A. Thompson (eds.), 157–194. Oxford: Clarendon Press. Thompson, Laurence C. & Thompson, M. Terry. 1980. Thompson Salish //â•‚xi//. International Journal of American Linguistics 46: 27–32. Thompson, Laurence C. & Thompson, M. Terry. 1992. The Thompson Language [Occasional Papers in Linguistics 8]. Missoula MT: University of Montana Press. Tuggy, David. 1996. Dative-like constructions in Orizaba Nahuatl. In The Dative, Vol. 1: Descriptive Studies, William van Belle & Willy van Langendonck (eds.), 407–452. Amsterdam: John Benjamins. Van Eijk, Jan P. 1987. Dictionary of the Lillooet language. Victoria BC: University of Victoria. Van Eijk, Jan P. 1997. The Lillooet Language: Phonology, Morphology, Syntax. Vancouver BC: UBC Press. Vogt, Hans. 1940. The Kalispel Language: An Outline of the Grammar with Texts, Translations, and Dictionary. Oslo: Det Norske Videnskaps-Akademi i Oslo. Watanabe, Honoré. 1996. Sliammon (Mainland Comox) transitive constructions with -‘әm, -ni, and -mi. Papers for the 31st International Conference on Salish and Neighbouring Languages, 327–338. Vancouver BC. Watanabe, Honoré. 2003. A Morphological Description of Sliammon, Mainland Comox Salish, with a Sketch of Syntax [Endangered Languages of the Pacific Rim Publication Series A2–040]. Osaka: Osaka Gakuin University, Faculty of Informatics. Willett, Marie Louise. 2003. A Grammatical Sketch of Nxa’amxcin (Moses-Columbia Salish). Ph.D. dissertation, University of Victoria.
Salish applicatives 
Zavala, Roberto. 1999. External possession in Oluta Popoluca (Mixean): Applicatives and incorporation of relational terms. In External Possession [Typological Studies in Language 39], Doris L. Payne & Immanuel Barshi (eds), 339–372. Amsterdam: John Benjamins. Zavala, Roberto. 2000. Inversion and Other Topics in the Grammar of Olutec (Mixean). Ph.D. dissertation, University of Oregon. Zúñiga, Fernando. In preparation. The Grammar of Benefaction.
Beneficiaries and recipients in Toba (Guaycurú) Marisa Censabella
Conicet*/Universidad Nacional del Nordeste This paper examines the coding properties of benefaction in Toba, such as case marking, agreement and word order, as well as the alignment patterns in derived ditransitive clauses. Toba does not have non-derived ditransitive verbs with three obligatory arguments (agent, theme and recipient) but uses applicatives in order to incorporate more than two arguments in a clause. In clauses where the arguments appear as lexical noun phrases, the recipient or beneficiary argument of the ditransitive clause behaves as the patient of the transitive one, appearing in the same functional slot and behaving alike as to verb agreement. Only with the derived verb ‘give’ is it possible to clearly distinguish between a recipient and a recipient-beneficiary. Thus, following Kittilä’s (2005: 277) classification, Toba is a beneficiary-prominent language.
1.â•… Introduction Toba is a Guaycuruan language spoken mainly in the Argentinean Chaco region (provinces of Chaco, Formosa, and Salta), as well as in some cities like Rosario and the surroundings of Buenos Aires. There are approximately 60,000 speakers with different levels of linguistic and communicative competence, which basically depends on the type of demographic concentration: in urban settlements, young Tobas are frequently less communicative competent; in rural ones, higher communicative competence is found, even among young people.1 Toba speakers—as well as linguists (Klein 1981)— identify four main dialects in the Chaco province: lañagashek, no’olgranaq, rapigeml’ek, and takshek, all mutually intelligible.2 It is an endangered language, despite the relatively *Consejo Nacional de Investigaciones Científicas y Técnicas 1.â•… Messineo (1991) mentions 500 Tobas in the surroundings of El Cerrito (Paraguay). Gordon (2005) estimates 700 Tobas in Paraguay, some 60 km northwest of Asunción, and 147 Tobas in southern Bolivia. The latter group has strong links to the Tobas in the province of Salta (Argentina). 2.â•… In this paper I present sentences and clauses from the lañagashek, no’olgranaq and rapigeml’ek dialects.
 Marisa Censabella
high number of speakers and the existence of programs of educación intercultural bilingüe ‘bilingual intercultural education’ in all the provinces where the Tobas live. An advanced stage of language shift is observed in all urban settlements, even in some rural ones, due to particular historical and social causes (Censabella 1999: 72–75). This paper will examine the coding properties of benefaction in Toba (case marking, agreement, and word order), the alignment patterns in derived ditransitive verbs, and a coordination strategy used to introduce beneficiary arguments. Finally, other applicatives and their distribution around the semantic notions of beneficiary and human goal are presented in order to understand the semantics of benefaction in the language. The examples below were taken mainly from narrative texts (collected and translated by myself) and from Buckwalter (1980). For expository reasons, I have simplified some sentences, testing their acceptability with two Toba consultants by means of elicitation techniques.
2.â•… Brief typological sketch Toba is a polysynthetic language with non-rigid word order (mainly VS, AVO, and OVA when O is pronominal), a verb/noun opposition, head marking, an alienable/inalienable possession distinction, and neither case marking nor an adjectival class of words. Toba has four types of nominal categorization: nominal classes or genders, nominal classifiers (to indicate the shape of same derived nouns), possessive classifiers, and demonstrative or deictic classifiers.3 The expression of number in nouns combines a category of morphological number with three differences (singular, paucal and plural), one of collectives, and another of distributives, while dual number is marked syntactically.
3.â•… The phoneme inventory of Toba is shown below (see Censabella 2002: 53 for more details). The graphemes in brackets are the ones used in this paper. plosives voiceless fric. voiced fric. laterals semi-con. tap nasals
labial p
w m
alveolar t s
pre-palatal t∫ [ch] ∫ [sh] Š
l
λ j
n [r] n
close open
palatal
n˜ front i e
back o a
velar k > [g]
uvular q t [g]
glottal ô h
Beneficiaries and recipients in Toba 
Table 1.╇ Active and middle voice personal markers4 Person
Active voice
Middle voice
1sg 2sg 3sg
sawi-; r-; Ø-; t-; w+ trn ……………………… − trn s…q qaw…i i-…ô; r-… ô; Ø-… ô; t-… ô; w-…ô + trn ……………………… − trn
ñann-
1pl 2pl 3pl
ñ…q qan…i n-…ô
The verb system is organized according to an active/middle opposition for all persons (see Table 1); the active voice presents an interesting split for the third person used to distinguish actions, events, different kinds of processes, and states. This continuum can be interpreted, mainly, as a scale of semantic transitivity. With non-derived verbs, the third personal marker i- appears with canonical transitive verbs and it is coreferential with the A; the complete clause needs plain or pronominal A and P arguments, with or without applicatives.5 The personal markers r-, ∆-, t-, and w- are coreferential with the S argument, and they may appear in monovalent intransitive clauses, but also in extended intransitive clauses (i.e., with an intransitive verb plus the applicative).6 There is no verbal tense, and the aspectual system consists in an obligatory perfective/imperfective opposition for all verbs. It has two groups of verbal adpositions, one
4.â•… Regarding their co-occurrence with voice, defined here as an inflectional category that signals the orientation of the verb within the participants of the clause, Toba verbal bases fit in three categories: (1) Active voice verbs: some prototypical transitive verbs (sa-lawat ‘I kill’/*ña-lawat; i-lo ‘he looks at’/*n-lo). (2) Active/middle voice verbs: some grooming verbs (s-ijo ‘I wash (sthg.)/ñ-ijo ‘I wash myself’); motion verbs with lexicalized applicatives (se-lariñi ‘I bury’/ñi-lar-shegem ‘I dig up’); stative verbs (w-etagi ‘he is inside somewhere’/n-etagi ‘he lives somewhere’); path verbs (i-wiô ‘he arrives somewhere’/n-iwiô ‘he arrives willingly somewhere’); transitive verbs (sa-sagawek ñi qomaô ‘I throw out the stone’/ña-sagawek ñi qomaô ‘I throw out the stone with anger or a strong emotion’). (3) Middle voice verbs: indirect middle verbs (n-ato ‘he gathers’/*r-ato; ñiôepe ‘I hunt’/*siôepe); involuntary bodily actions (ñi-jin ‘I cry’/*si-jin; n-qashen ‘he sneezes’/*sa-qashen). 5.â•… With semantically bivalent verbs, applicatives do not necessarily introduce a third argument. Instead, they may change the meaning of the event expressed by the verb. 6.â•… The following example illustrates the extended intransitive clause type (b): a.
RamaŠe ∅-kewo. b. 3sg 3-walk.inside ‘He goes home.’
RamaŠe ∅-kewo-lek na nojik. 3sg 3-walk.inside-appl:loc dem house ‘He walks around the house.’
 Marisa Censabella
of â•›“directional” and another of â•›“locative-movement oriented” paradigms, the elements of the latter functioning as applicatives. Valency changes are expressed by derivational (causatives) and syntactic (applicatives) mechanisms. The semantic notions of human goal, recipient and beneficiary are expressed via applicatives that increase or rearrange syntactic argument structure, showing some degree of semantic overlap in their morphological expression.7
3.â•… Reception and benefaction Kittilä (2005) distinguishes the semantic roles of recipient, beneficiary, and recipientbeneficiary on the basis of the notions of reception and benefaction. For this author, reception refers to the fact that as a result of an event a (concrete) entity enters a recipient’s sphere of control or domain of possession, as in ‘the teacher gave/sent me a book’… Benefaction, on the other hand, implies that the occurrence of an event is beneficial to an entity other than the agent itself. (Kittilä 2005: 273)
The beneficiary notion is further subcategorized as substitutive benefaction and concrete benefaction. The former refers to events where the benefaction “consists in not having to carry out the profiled event oneself… like ‘the teacher parked the car for me/on my behalf ’…”; the latter refers to events in which “the beneficiary does not ‘receive’ anything concrete but benefits from the event, as ‘the teacher lied for me’â•›”(Kittilä 2005: 273). Due to the fact that concrete benefaction is rather irrelevant— because in one way or another it is present in the three mentioned semantic roles—, Kittilä uses only two features in order to classify the recipient, beneficiary, and recipientbeneficiary: reception and substitutive benefaction: “… recipients (as a semantic role) are thus animate entities that receive something concrete transferred to their sphere of control or domain of possession. The prototypical recipient is illustrated by the recipient of the event ‘give’â•›” (Kittilä 2005: 274). Beneficiaries have, on the contrary, substitutive benefaction as their only positive feature value; the notion of reception is completely lacking in this case. Finally, the recipient-beneficiary represents a combination of recipient and beneficiary, and it thus comprises both reception and (substitutive) benefaction (Kittilä 2005: 275).
7.â•… See Censabella (2002) for a general morphosyntactic sketch of the language, Carpio (2004) for the expression of number, and Censabella (2006) and Carpio (2007) for a study of the active personal markers.
Beneficiaries and recipients in Toba 
Since recipients and beneficiaries are formally distinct in most languages of the world, Kittilä (2005: 276) suggests exploring whether the recipients-beneficiaries are coded like recipients or like beneficiaries instead. This is not a simple task when working with less-studied languages. Which are the semantic clues to decide that the reception of an entity is not beneficial for the recipient? Even though the definitions presented above are clear-cut, sometimes it is difficult to distinguish between a recipient and a recipient-beneficiary. Toba clearly encodes these two semantic notions only with the verb ‘give’ (see Example 2 below). With other transitive verbs and with intransitive verbs, this distinction is blurred.
4.â•… Coding properties of recipients and beneficiaries in Toba There are three main ways to express a functional domain like benefaction in a given language: case marking or flagging; agreement or indexing behavior, and word order (Malchukov et al. 2007: 6). I will address them in turn here.
4.1â•… Case marking or flagging In Toba, beneficiaries as well as other non-subject arguments of the recipient type are introduced via applicativization. The applicative suffix occupies the final positional slot in the verb complex, followed only by the affix specifying number of patient-like participants (see § 4.2). Toba has a series of applicatives that encode meanings related to the notions of reception and benefaction (1a–d): (1) a.
So jagajkjolek j-añ-i so l-qaja dem old.man 3trn-give-appl:rec dem 3poss-brother
so l-apoô. dem 3poss-poncho
‘That old man gives his brother the poncho.’
b. J-an-em so l-qaja so pan. 3trn-give-appl:ben dem 3poss-brother dem bread ‘S/he gives his brother bread.’ c.
Ra jale-pi ro-ôonatagan-ot seôeso saλaganek. dem man-col 3intr-work-appl:loc dem.pron rich.man ‘The men work for that rich man.’
d. Juan ro-ôon-agaa-lek ñi l-qaja. J. 3intr-sing-antipass-appl:loc dem 3poss-brother ‘Juan sings for his brother.’/‘Juan pays homage to his brother.’
 Marisa Censabella
In its non-derived form, the verb -an ‘give’ is transitive and requires the expression of two arguments in the clause, the agent and the theme (2a).8 In order to introduce a recipient or a beneficiary via applicativization, the speaker encodes this third participant as either the former or the latter. When the given object does not entail a special benefit for the recipient, the applicative used is -i (2b). On the contrary, when the given object is of special value for the recipient—either culturally and in itself, as in (2c), or contextually negotiated, as in (2f)—, the recipient-beneficiary is coded as a beneficiary by means of the applicative -em. The unacceptability of (2d) and the minimal pair shown in (2e–f) confirms our hypothesis. These examples also show that the encoding choice depends on pragmatic factors, i.e., the speaker chooses the applicative morpheme depending on his/her interpretation of the referred situation: if the recipientbeneficiary is starving or living in poverty and needs food, it will be interpreted as a beneficiary (2f); if the recipient-beneficiary does not particularly benefit from the reception of the bread, the recipient morpheme will be chosen. (2) a.
So jagajkjolek j-an so l-apoô. dem old.man 3trn-give dem 3poss-poncho ‘That old man gives the poncho (to someone).’
b. So jagajkjolek j-añ-i so l-qaja dem old.man 3trn-give-appl:rec dem 3poss-brother
so l-apoô. dem 3poss-poncho
‘That old man gives his brother the poncho.’
c.
So dem so dem
jagajkjolek j-an-em so l-awal old.man 3trn-give-appl:ben dem 3poss-grandchild no’onatagak. custom
‘That old man teaches his grandchild the tradition (lit. gives his grandchild the custom).’ d. *So jagajkjolek j-añ-i so l-awal dem old.man 3trn-give-appl:rec dem 3poss-grandchild
so no’onatagak. dem custom
‘That old man teaches his grandchild the tradition (lit. gives his grandchild the custom).’ e.
J-añ-i so l-qaja so pan. 3-give-appl:rec dem 3poss-brother dem bread ‘S/he gives his brother bread.’
8.â•… Non-derived ditransitive verbs do not exist in Toba.
Beneficiaries and recipients in Toba 
f. J-an-em so l-qaja so pan. 3-give-appl:ben dem 3poss-brother dem bread ‘S/he gives his brother bread.’ (with a benefactive reading: the brother is starving)
Before dealing with the encoding properties of beneficiaries with other transitive verbs and with intransitive verbs, however, it is necessary to consider the entire applicative paradigm in order to better understand the semantic differences between the recipient-beneficiaries and beneficiaries-like arguments and their expression in Toba. The applicative morphemes are the following: -ôa allative, -lek locative1 ‘over’, -gi locative2 ‘inside’, -ôot locative3 ‘under, in front of ’, -get directional1‘approaching, coming closer’, -ge directional2 ‘moving away’, and three morphemes for introducing more abstract notions, viz. -i recipient, -eô human goal, and -em benefactive. These affixes occupy the same slot in the verbal complex and are mutually exclusive. The examples in (3) show typical spatial and directional uses of the applicative markers; here the applicatives cannot be detached from the lexical base of the verb root: (3) a.
T-aja na nemaganagaki. 3-go.appl:all dem market ‘S/he goes to the market.’
b. T-alek a-ñi komaô. 3-go.appl:loc f-dem stone/mountain ‘S/he climbs the mountain.’ c.
We-to-ôot a-ra ipaq. 3-go-appl:loc f-dem tree ‘S/he is under the tree.’
The examples in (4) show the same applicatives, but arguably in less lexicalized uses, introducing arguments that can be interpreted as human locative goals: (4) a.
Ra-ta-ôa na l-qaja. 3-go-appl:all dem 3poss-brother ‘S/he goes to his brother.’/‘S/he goes with his brother.’
b. Ra-ta-lek so l-tagalaô. 3-go-appl:loc dem 3poss-chief ‘S/he pursues his chief.’ c.
Ra-tato-ôot so l-taôa. 3-go-appl:loc dem 3poss-father ‘S/he is in front of his father.’
The examples in (5) show human locative goals introduced by verbs that differ semantically from the path verbs seen in (4), with the same applicative markers. Example€(5c)
 Marisa Censabella
clearly refers to a benefactive event (a “deputative benefactive” in the sense of Van Valin€& LaPolla 1997: 383) where the rich man benefits from the men’s work: (5) a.
Ra-lemata-ja a-so l-jale. 3-be.angry-appl:all f-dem 3poss-daughter ‘S/he is angry with her/his daughter.’
b. Juan ro-ôon-agaa-lek ñi l-qaja. Juan 3-sing-antipass-appl:loc dem 3poss-brother ‘Juan sings for his brother.’/‘Juan pays homage to his brother.’ c.
Ra jale-pi roôonatagan-ot seôeso sa‡aganek. dem man-col 3-work-appl:loc dem.pron rich.man ‘The men work for that rich man.’
The recipient applicative -i (2b) seldom appears with verbal bases other than -an ‘give’. Thus, recipients and recipient-beneficiaries are also encoded by the locative and allative applicatives shown in Examples (3) to (5). In (6), with transitive verbal bases, Juan is a recipient-beneficiary, but in (6a) the locative applicative -ôot ‘under’—see (3c) and (4c)—indicates that a marked hierarchical relationship exists between the agent and the recipient-beneficiary: the latter is more “salient” or “important” than the former (6b). In (6c), the recipient-beneficiary customer is encoded as outranking the agent. In (6d), the recipient-beneficiary is encoded as a beneficiary. Example (6e) shows that a substitutive reading is also possible with a similar clause when the noun is an alienable possession. We focus on the type of relationship between the agent and the recipient or beneficiary chosen by the speaker here because Toba grammaticalizes this distinction by means of the locative applicatives. (6) a.
Se-pagaq-tee-tot ñi Juan na waloq. 1-deliver-prog-appl:loc dem J. dem cotton ‘I am giving Juan the cotton’. (Juan is the owner of the plantation)
b. Se-pagag-tee-tem ñi Juan na waloq. 1-deliver-prog-appl:ben dem Juan dem cotton ‘I am giving Juan the cotton’. (Juan is my friend) c.
ôam se-men-ot a-so in-kanasto. 2sg 1-sell-appl:loc f-dem 3.i.poss-basket ‘I sell you that basket.’ (You are a customer)9
9.â•… With the verb -men ‘sell’, Toba encodes customers as if they were bosses. Ideally, Example€ (6c) allows a substitutive reading, meaning ‘I sell the basket for you (= the sales clerk)’, but I was not able to confirm this interpretation via elicitation.
Beneficiaries and recipients in Toba 
d. ôam se-men-em ñi i-maô. 2sg 1-sell-appl:ben dem 3poss-home ‘I sell you my house’ (lit.: ‘I sell you my home’) (you are my friend, or someone that benefits with the giving of the house; substitutive reading very improbable) e. ôam se-men-em ñi nojik. 2sg 1-sell-appl:ben dem house ‘I sell you the house’ (Substitutive reading allowed: you are my friend (not my boss) and I sell the house on your behalf)
Other transitive verbs with which it is possible to find the benefactive applicative are ôagat ‘say’ (7) and -achagan ‘show’ (8): (7) a.
Juan i-ôagat seôeso. J. 3-say pron.dem ‘Juan says that.’
b. SomaŠe ajim i-ôagat-em ra n-chogonagak. 3sg 1sg 3trn-tell-appl:ben dem 3i.poss-dream ‘He tells me about the dream.’ c.
Juan ra-ôaqt-agan. J. 3-say-antipass ‘Juan talks.’
d. Juan ra-ôaqt-agan-em a-so ôalo. J. 3-say-antipass-appl:ben f-dem woman ‘Juan talks to the woman (i.e., says something from which the woman will benefit).’ (8) a.
RamaŠe j-achagan so l-ôonatak. 3sg 3-show dem 3poss-craftwork ‘He shows/displays his craftwork.’
b. RamaŠe ôam j-achagan-a so l-ôonatak. 3sg 2sg 3-show-appl:all dem 3poss-craftwork ‘He shows you his craftwork.’ c.
RamaŠe ôam j-achagan-ot so l-ôonatak. 3sg 2sg 3-show-appl:loc dem 3poss-craftwork ‘He shows you his craftwork.’ (you are the craftwork teacher)
d. RamaŠe ôam j-achagan-em so l-ôonatak. 3sg 2sg 3-show-appl:ben dem 3poss-craftwork ‘He shows you his craftwork for your benefit.’ (you will benefit from knowing about his craftwork) e.
RamaŠe ôam j-achagan-em so n-ôonatak. 3sg 2sg 3-show-appl:ben dem 3.i.poss-craftwork ‘He shows the craftwork on your behalf.’ (Substitutive reading)
 Marisa Censabella
The examples in (8) show that the applicative morpheme used in order to convey broad benefaction is the same employed to encode substitutive benefaction. Example (8e) allows a substitutive reading (like 6e above) when the theme is not possessed by the agent. With intransitivized verbs, it is possible to find more examples of substitutive benefaction, and it is always the applicative -em that encodes the beneficiary. (9) a.
RamaŠe i-ôon na no-ôon-agak. 3sg 3trn-sing dem 3.i.poss-sing-nom ‘He sings a song’
b. RamaŠe ro-ôon-agaa-tak 3sg 3intrn-sing-antipass-prog ‘He is singing’. c.
RamaŠe ro-ôon-agaa-teg-em so l-apiô. 3sg 3intrn-sing-antipass-prog-appl:ben dem 3.poss-grandfather ‘He is singing for his grandfather’. (Substitutive reading is also allowed)
d. RamaŠe ro-ôon-agaa-ta-lek so nogotolek. 3sg 3intrn-sing-antipass-cont-appl:loc dem child ‘He is singing for the child.’ (Substitutive reading is not allowed) e.
RamaŠe ro-ôon-agaa-toôot ñi Dios. 3sg 3intrn-sing-antipass-cont-appl:loc dem child ‘He is singing for God.’ (Substitutive reading is not allowed)
f.
RamaŠe ro-ôon-agaa-teg-eô ñi j-aqaja. 3sg 3intrn-sing-antipass-prog-appl:h.goal dem 1sg.poss-brother ‘He is singing with my brother.’
In (10a) we find a recipient-beneficiary interpretation, while (10b) corresponds to substitutive benefaction only. In the latter example, the substitutive reading is allowed if an intransitive personal marker r- is used instead of transitive i-. (10) a.
A-ra ôalo ajim i-wosh-agan-em. f-dem woman 1sg 3trn-cook-antipass-appl:ben ‘This woman cooks for me.’
b. A-ramaŠe ajim r-wosh-agan-em. f-3sg 1sg 3intr-cook-antipass-appl:ben ‘She cooks for me.’ (she works as a cook in my restaurant)
In summary, whenever a speaker wants to simultaneously foreground the notion of benefaction and background the notion of reception despite the saliency of the recipient, s/he chooses the benefactive applicative. Toba thus confirms Kittilä’s (2005: 295) prediction that “languages with few formally ditransitive verbs are more prone to beneficiary prominence”.
Beneficiaries and recipients in Toba 
4.2â•… Agreement or indexing In Toba, both patients and all other applied arguments trigger number agreement on the verb: (11) a.
Ajim sh-en-agat so l-jalek. 1sg 1-be.similar-caus dem 3poss-son ‘I imitate his son.’
b. Ajim sh-en-aqt-a so ˆalqa. 1sg 1-be.similar-caus-pl dem 3poss.son.pl ‘I imitate his two sons.’ (12) a.
J-an-em so l-qaja so pan. 3-give-appl:ben dem 3poss-brother dem bread ‘S/he gives his brother bread.’
b. J-an-em-a so l-qaja-qa so pan. 3-give-ben-pl dem 3poss-brother-pl dem bread ‘S/he gives his brothers bread.’ (13) …â•›qamiô sa-ôaqtagaa-teg-em-a de la tema 2pl 1-say-antipass-prog-appl:ben-pl about the subject
Še Nsogoj. dem Nsogoi
‘… I’m talking to you (pl) about the subject of the Nsogoi (cannibalistic woman).’
4.3â•… Word order Toba shows non-rigid word order, mainly VS in one-argument clauses, AVO in twoargument clauses (OVA when O is pronominal), and AVRT in three-argument clauses. Note that the applied arguments appear immediately after the verb complex, as in (2) above; however, when the recipient or the beneficiary is expressed by a personal pronoun, the applied argument appears immediately before the verb complex, as in (14b) and (15b): (14) a.
So jagajkjolek j-añ-i so l-qaja dem old.man 3trn-give-appl:rec dem 3poss-brother
so l-apoô. dem 3poss-poncho
‘That old man gives his brother the poncho.’
b. So jagajkjolek ajim j-añ-i so l-apoô. dem old.man 1sg 3trn-give-appl:rec dem 3poss-poncho ‘That old man gives me the poncho.’
 Marisa Censabella
(15) a.
So jagajkjolek j-an-em so l-awal dem old.man 3trn-give-appl:ben dem 3poss-grandchild
so no’onatagak. dem custom
‘That old man teaches (lit. gives) his grandchild the tradition.’
b. So jagajkjolek ajim j-an-em so no’onatagak. dem old.man 1sg 3trn-give-appl:ben dem custom ‘That old man teaches (lit. gives) me the trdition.’
Taking into account agreement as shown in (11)–(13), as well as word order of ditransitive clauses with plain and pronominal arguments in (14)–(15), Toba has secundative alignment, i.e., the recipient or the beneficiary is treated like the patient of a transitive clause.
4.4â•… Other characteristics In (16) and (17), the applicative -em is used to express the notion of malefaction. This interpretation is possible because the speaker selects a “negative” verb,10 either with an intransitive (16) or with a transitive (17) verbal base. With a “positive” transitive verb like -iltagan ‘free (someone)’ that selects an agent and a human goal (18a), the benefactive applicative encodes a maleficiary (18b). (16) Ajim r-ashimat-em so i-qaja. 1sg 3intrn-be.unwilling.to.share-appl:ben dem 1poss-brother ‘My brother is unwilling to share with me.’ (17) Ajim i-lla-peg-em ra i-aqtak. 1sg 3trn-prohibit-rep-appl:ben dem 1poss-word ‘He prohibits me to talk.’ (18) a.
ôAm j-iltagan. 2sg 3trn-free ‘S/he frees you’.
b. ôAm j-iltagan-em so Šilogojk. 2sg 3trn-free-appl:ben dem the.mean.man ‘S/he frees you from the mean man.’
The beneficiary applicative -em can also be used to express the cause of an event. (19) a.
A-ramaŠe i-lew. f-3sg 3-die ‘She dies.’
10.â•… A negative verb is a verb that has a negative meaning in itself without any negative morphology.
Beneficiaries and recipients in Toba 
b. A-ka ôalo i-lem ra l-koôok. f-dem woman 3-die.appl:ben dem 3poss-giving.birth ‘That woman dies because of giving birth.’ (20) a.
RamaŠe ∆-keôemaq. 3sg 3-be.wounded ‘He is wounded.’
b. aw-keôemag-am so l-kat. 2-be.wounded-appl:ben dem 3poss-piece.of.iron ‘You are wounded by the piece of iron.’ (21) SomaŠe sa-sa-jat-em. 3sg neg-1-know-appl:ben ‘I don’t know [sc. the truth] because of him.’
This meaning provides a clue about the relationship between beneficiaries and possessor markers, as Heine & Kuteva (2002: 54) suggest. Toba shows markers of possession (22) and of topicality (23) that resemble the benefactive applicative suffix: (22) So-m nashe l-waô. dem-poss chief 3poss-wife ‘The chief ’s wife.’ (23) So-m Ferocho j-apagagee-tak qoôoˆaga l-qajolek. dem-top Ferocho 3-teach-prog adv 3poss-little.brother ‘That Ferocho was teaching, long time ago, his little brother.’
5.â•… Coordination Instead of using a serialization strategy to express benefaction as many languages do, Toba shows a clause coordination strategy with recipient-like applicatives, which goes substantially beyond the use of the beneficiary applicative.11 This strategy, based on
11.â•… Creissels (this volume) defines serialization as follows: “I adopt the definition of serial verb constructions as complex predicates (i.e., monoclausal constructions involving two or more verbs) showing the following two characteristics: a. no linking element is present between the verbs involved in the construction; b. none of the verbs involved in the construction is in a form implying a non-autonomous status. […] Covert coordination may be the historical source of serial verb constructions, and in some languages the limit between covert coordination and serial verb constructions may be fuzzy, but a notion of serial verb construction that would not include this restriction would be too vague to be useful in the analysis of syntactic structures.”
 Marisa Censabella
paratactic juxtaposition of clauses with severe restrictions on the nominal arguments associated with each verb, could be explained by the inexistence of inherently ditransitive verbs and ‘oblique’ arguments (as Mithun 2004 suggests for some native American Indian languages); Toba speakers prefer Example (24a) to (24b): (24) a.
I-ro so l-jalek, i-ra-ôa ñi Resistencia. 3-take dem 3poss-son 3-take-appl:all dem R. ‘He takes/carries his son, he takes [him] to Resistencia.’
b. I-ra-ôa ñi Resistencia so l-jalek. 3-take-appl:all dem R. dem 3poss-son ‘He takes his son to Resistencia.’
The coordination strategy is preferred when the recipient or the beneficiary are expressed by lexical noun phrases, as in (25); in this sentence, the theme is prosodically attached to the first clause. On the contrary, a three-argument clause is preferred when the recipient or the beneficiary is expressed by a pronoun (26): (25) R-shihn-agan-a a-so l-awogo, j-añ-i 3poss-get-antipas-appl:all f-dem 3poss-flower 3-give-appl:rec a-so María. f-dem M. ‘S/he gets to that flower, s/he gives [it] to María.’ (26) Ajim ra-ôaka-ôa so i-qaja, ajim 1sg 3-offer-appl:all dem 1poss-brother 1sg j-an-em so anoq. 3-give-appl:ben dem food ‘He offers my brother to me, he gives me food.’
The coordination strategy is always preferred when arguments instantiated by lexical noun phrases are involved, as in Example (25), even with -an ‘give’. More “canonical” ditransitive clauses (i.e., three-argument clauses forming one prosodic unit) are used when the recipient or the beneficiary argument is expressed by a personal pronoun and the verb is -an ‘give’ (14a) or others similar to it, like agat- ‘tell’ (27) and achagan‘show’ (28): (27) Qalagayi chochi ajim ji-ôagat-em yi j-anoqoô. coord.caus emph 1sg 3trn-tell-appl:ben dem 1poss-birth ‘Thus, it is like that!, he tells me about my birth.’ (28) RamaŠe ôam j-achagan-em so l-ôonatak. 3sg 2sg 3trn-show-appl:ben dem 3poss-craftwork ‘He shows you his craftwork [for your benefit].’ (You will benefit from knowing about his craftwork.)
Beneficiaries and recipients in Toba 
In narrative texts, it is frequent to find two coordinated clauses instead of a ditransitive clause, but some changes are required: the transitive verb -ôagat ‘tell’ first reduces its valency by antipassivization and then allows a beneficiary argument to be introduced by the applicative -em: (29) …â•›qaq era ra la-ôaqtaganagak so i-taôa, coord ph.dem dem 3poss-story dem 1poss-father koôoλaq ajim ra-ôaqt-agan-em. adv 1sg 3intr-tell-antipass-appl:ben ‘… and this is my father’s story, long time ago he told [it] to me.’
6.â•… Conclusions In this preliminary sketch, I have examined the coding properties related to the expression of benefaction in Toba. The language does not have a class of non-derived ditransitive verbs with three obligatory arguments (agent, theme and recipient); even -an ‘give’ needs to undergo derivation in order to be used in three-argument clauses. In order to include more than two arguments in a clause in general, applicatives are necessary. In clauses where the arguments appear as lexical noun phrases, the alignment pattern of ditransitive clauses is secundative: the recipient or the beneficiary of the ditransitive behaves as the patient of the transitive (i.e., they are the primary object), occupying the same functional slot and triggering the same number agreement marking on the verb. Applicativization via locative and directional applicatives is the main strategy to introduce non-subject arguments in ditransitive clauses. There is a scale of saliency that governs the selection of one of the four morphemes related to human locative goals (from less salient to more salient arguments): -lek, -ôa, -eô, -ôot. Many verbs allow the alternation of at least two of these morphemes. Despite this saliency scale and its overlap with the recipient-beneficiary coding, whenever the speaker wants to highlight the benefit that a non-subject participant obtains from the event, it will be encoded as a beneficiary (i.e., with the same applicative morpheme -em used for a substitutive beneficiary reading in intransitive clauses). Only with the verb -an ‘give’ is it possible to make a clear distinction between a recipient and a recipient-beneficiary in Toba. With other verbs, the benefactive applicative -em is used to express both semantic notions. Thus, following Kittilä’s (2005: 277) classification, Toba is a beneficiaryprominent language.
 Marisa Censabella
Abbreviations 1 2 3 3.i all antipass appl ben caus col cont coord dem dem.pron emph f
first person second person third person indeterminate third person allative antipassive applicative benefactive causative collective continuous coordinator demonstrative demonstrative pronoun emphatic feminine
h.goal intr loc m nom pat pers.pron ph.dem pl poss prog rec rep trn unkw.ag
human goal intransitive locative middle voice nominalizer patient personal pronoun phrasal demonstrative plural possessive progressive aspect recipient repetitive transitive unknown agent
References Buckwalter, Alberto. 1980. Vocabulario Toba- Seguido de algunos Apuntes sobre la gramática del idioma toba. Roque Saénz Peña, Chaco (Argentina): Author’s edition. Carpio, María Belén. 2004. La categoría de número en la lengua toba (familia Guaycurú, Argentina). BA thesis, Universidad Nacional de Rosario, Argentina. Carpio, María Belén. 2007. Sistemas de alineación en toba (familia Guaycurú, Argentina). MA thesis, Universidad de Sonora, México. Censabella, Marisa. 1999. Las lenguas indígenas de la Argentina. Una mirada actual. Buenos Aires: Eudeba. Censabella, Marisa. 2002. Descripción funcional de un corpus en lengua toba (familia Guaycurú, Argentina). Sistema fonológico, clases sintácticas y derivación. Aspectos de sincronía dinámica. Ph.D. dissertation, Universidad Nacional de Córdoba, Argentina. Censabella, Marisa. 2006. Relaciones gramaticales en lengua toba (flia. Guaycurú, Argentina). In Actas VIII Encuentro Internacional de Lingüística en el Noroeste, Zarina Estrada et al. (eds.), 81–103. Hermosillo: Universidad de Sonora. Gordon, Raymond G., Jr. (ed.). 2005. Ethnologue: Languages of the World, 15th edn. Dallas TX: SIL International. . Heine, Bernd & Kuteva, Tania. 2002. World Lexicon of Grammaticalization. Cambridge: CUP. Kittilä, Seppo. 2005. Recipient-prominence vs. beneficiary-prominence. Linguistic Typology 9:€269–297. Klein, Harriet E. Manelis. 1981. Una gramática de la lengua Toba: Morfología verbal y nominal. Montevideo: Departamento de Lingüística, Universidad de la República.
Beneficiaries and recipients in Toba 
Malchukov, Andrej, Haspelmath, Martin & Comrie, Bernard. 2007. Ditransitive constructions: A typological overview. First draft version September 2007. . Messineo, Cristina. 1991. Variantes dialectales del complejo lingüístico toba. Hacia una nueva carta étnica del Chaco II: 13–22. Mithun, Marianne. 2004. The non-universality of obliques. Paper read at the Syntax of the World’s Languages (SWL 1) Conference. Leipzig, August 5–8. Handout version. Van Valin, Robert D., Jr. & LaPolla, Randy J. 1997. Syntax. Structure, Meaning and Function. Cambridge: CUP.
Benefactive and malefactive applicativization in Mapudungun* Fernando Zúñiga University of Zurich
This paper argues that two productive applicatives in Mapudungun, viz. -(l)el and -ñma, can be labeled benefactive and malefactive respectively, but only in a superficially impressionistic way. Based on a detailed survey of the literature on the language, as well as on synchronic evidence and some tentative internal reconstruction, I suggest that the yield of -ñma was originally probably neutral with respect to benefaction, and that it still is in some uses of this suffix. According to this account, the appearance of (l)el-applicatives (probably from elu- ‘give’ or el- ‘set’) gave rise to a privative opposition between the meaning of both morphemes with many bivalent predicates.
1.â•… Introduction1 The purpose of the present study is to evaluate the characterization of those applicative operations in Mapudungun that have received comparatively more attention in the literature, viz. “benefactive” -(l)el and “malefactive” -ñma, as well as to tentatively put the findings into cross-linguistic perspective. Section 2 sketches the fundamentals of Mapudungun morphosyntax that are necessary in the present context, and Section 3 briefly introduces the four applicative suffixes -tu, -ye, -(l)el and -ñma. Section 4 surveys and evaluates the literature on -l(el) and -ñma in greater detail. This study claims that -ñma probably was an applicative semantically unmarked for benefit or detriment, and that as -(l)el entered the picture with a semantics marked for benefit by default (possibly thanks to its etymon, either elu- ‘give’ or el- ‘create, set’, and presumably at a
*I am grateful to Seppo Kittilä and Matt Shibatani for their comments on earlier versions of this paper. 1.â•… Mapudungun is a language isolate currently spoken by approximately 150,000 people in southern Chile and Argentina. Several closely related dialectal varieties can be distinguished; this paper will focus on the main Chilean variety sometimes called Central Mapudungun.
Fernando Zúñiga
later point), there emerged a privative opposition for many, but not all, transitive predicates. Section 5 closes the paper.
2.â•… Some fundamentals of Mapudungun morphosyntax The language is agglutinative and head-marking; its rich verbal morphology indicates person and number of core arguments, inversion, lexical and grammatical aspect, spatial categories, tense, evidentiality, and deontic modality, among others. There are no adnominal markers of grammatical case for core arguments, and obligatory nominal number marking is restricted to nonsingular with adjectives and plural with (typically human) nouns. Verb forms are either finite or nonfinite, the latter showing less productivity as to tense and mood and marking one person less than the former. Processes that change syntactic valency are NP-incorporation, reflexivization, passivization, causativization, and applicativization. Mapudungun clause structure is best analyzed as distinguishing the following core syntactic functions, both of which can be instantiated by bare NPs: Focal Person (FP) and Satellite Person (SP, which can be either Primary or Secondary, cf. further down). The Focal Person controls the reference of WH-words in interrogative clauses and the explicit person marking on the verb (i.e. the one distinguishing three persons and three numbers). The (Primary) Satellite Person, by contrast, controls the relatively underspecified person marking on the verb (i.e. the one distinguishing person, but not number). Constituent order has not been thoroughly studied yet, but if both the FP and the SP are present as lexical NPs, the typical order appears to be [FP V SP], with [FP SP V] and [V SP FP] attested as well. In other words, the FP tends to precede the SP, and the latter has to be adjacent to the verb. The following examples illustrate the indexing pattern just mentioned:2 (1) a.
Weyel-küle-i-Ø ñi fotüm. swim-prog-ind-3.fp 1sg.psr son ‘My son (FP) is swimming.’
b. Weyel-küle-i-mi (eymi). swim-prog-ind-2sg.fp 2sg ‘You (sg) (FP) are swimming.’ c.
Leli-nie-fi-i-mi chi wentru.3 look.at-have-3sp-ind-2sg.fp art man ‘You (sg) (FP) watched/kept an eye on the man (SP).’
2.â•… Unless otherwise specified, examples come from my field notes. 3.â•… In this article, Mapudungun verbs are given in their underlying form for convenience; some syllabification, elision and assimilation rules apply. See Zúñiga (2006:105–106) for details.
Benefactive and malefactive applicativization in Mapudungun
d. Leli-nie-e-i-mi-mew chi wentru. look.at-have-inv-ind-2sg.fp-3.sp art man ‘The man (SP) watched/kept an eye on you (sg) (FP).’ e.
Elu-fi-i-mi kura chi kalku. give-3.sp-ind-2sg.fp stone art warlock ‘You (sg) (FP) gave a stone to the warlock (SP).’
f.
Elu-e-i-mi-mew kura chi kalku. give-inv-ind-2sg.fp-3.sp stone art warlock ‘The warlock (SP) gave you (sg) (FP) a stone.’
Examples (1a,b) show intransitive clauses where the only participant (ñi fotüm ‘my son’ and eymi ‘you (sg)’ respectively) is the FP; the 3rd person singular is zero-marked while the 2nd person singular is marked as -mi on the verb (the independent pronoun is normally used for focusing purposes only).4 In the direct transitive clause (1c) the 2nd person singular (actor) is the FP and chi wentru ‘the man’ (undergoer) is the SP. The definite and human undergoer triggers differential marking (-fi) on direct verb forms; 3rd person undergoer SPs are normally zero-marked on direct verb forms when nonhuman. In the inverse clause (1d) the 2nd person (undergoer) is the FP and chi wentru ‘the man’ (actor) is the SP; the former is marked as -mi, just as in (1c), and the latter is marked by -mew, unspecified for number. Inverse verb forms are additionally characterized by an inverse suffix (here: -e) that occurs in the same slot as the differential undergoer marker -fi. Examples (1c,d) are clauses featuring a ambitransitive/labile verb (leli- ‘look at’) and (1e,f) are clauses with a ditransitive verb (elu- ‘give’). Secondary SPs typically occur in the latter clause type (here: kura ‘(a) stone’) and also appear as bare NPs, but, unlike Primary SPs, they do not trigger any kind of verbal indexing. In sum, FP markers follow a neutral alignment pattern that does not distinguish the role of the participants (here: -Ø ‘3.fp’ and -mi ‘2.sg.fp’) but PSP markers distinguish A’s (-mew ‘3.A.psp’) from U’s (-Ø ~ -fi ‘3.U.psp’). Finally observe that Mapudungun has two causatives -m and -l, which will be relevant in the discussion of the applicative -(l)el in Section 4. The two causatives are, at least in principle, in complementary distribution: the former is used for low-control states of affairs and/or low-animacy causees (2a,b), and the latter appears in highcontrol and/or high-animacy contexts (2c,d). A number of verb roots can actually take both suffixes, so the system features both a split and a fluid dimension; see Golluscio’s (2007) study—from which the examples in (2) are taken—for more details.
4.â•… The element -mi is bimorphemic, viz. -m ‘2’ + -i ‘sg’, but since the morphological make-up of person-number marking is irrelevant for our present purposes I have chosen to give abbreviated glosses throughout the article. Most FP personal markers are bimorphemic; one of the notable exceptions is -n ‘1.sg.ind’, which also includes information as to mood.
Fernando Zúñiga
(2) a.
Are-i-Ø ta ko. get.hot-ind-3.fp part water ‘The water heated up.’
b. Are-m-i-Ø ta ko. get.hot-caus-ind-3.fp part water ‘S/he heated water up.’ c.
Aye-i-Ø ta malen. laugh-ind-3.fp part girl ‘The girl laughed.’
d. Aye-l-fi-i-Ø ta malen. laugh-caus-3.sp-ind-3.fp part girl ‘S/he made the girl laugh.’
3.â•… A first look at the applicatives of Mapudungun Applicatives add to the clause a non-agentive argument, which appears as either FP or SP depending on the factors that condition argument realization (roughly: 1st and 2nd persons are privileged FPs when compared to 3rd persons, and topics are privileged FPs when compared to non-topics).5 There are two relatively non-productive applicative suffixes that have received little attention in the literature: -tu and -ye. The former appears typically with a handful of verb roots that express motion and introduces an animate goal, whereas the latter introduces comitatives, topics of conversations and various other types of participants to the clause. There are cases, however, where neither -tu nor -ye have predictable yields. The two applicatives that are the center of attention here have been studied in some detail in the literature. Augusta (1903:61) characterizes them as follows: -l(el) simply increases transitivity but -ma ~ -(ü)ñma signals that “the action of the verb is performed to the detriment of somebody or something, or that it implies that one parts with a former possession” (my translation).6 By contrast, Salas (2006) labels these markers indirectizantes and describes the clauses thus derived as including an “indirect (ethical)” participant, as illustrated in€(3).
5.â•… Traditional definitions of applicativization agree that the prototypical use of such constructions consists in allowing a participant to appear as primary/direct object in a clause where it otherwise has less syntactic privileges. This is certainly true of Mapudungun applicatives in direct clauses, but inverse clauses seem to remap the grammatical functions in such a way that the applicatives introduce/promote a new subject instead of object. 6.â•… For details on the possible history of these two suffixes, the reader is referred to Zúñiga (i.p.).
Benefactive and malefactive applicativization in Mapudungun
The clauses shown there (taken from Salas 2006:123) are habitual passives used in order to give instructions as to how to perform some customary actions. The FP is ñi ütrar ‘its pips’ in (a) and fürkü ko ‘cold water’ in (b), and miyaye ‘jimson weed’ and fey ‘it’ respectively are the PSPs. The verb rüngü- ‘grind’7 in (a) normally takes an animate actor and an inanimate undergoer. Tuku- ‘put, set’ (b) is only apparently different; even though the English gloss might suggest an underlying meaning ‘put into’, in Mapudungun the verb takes an animate actor and an inanimate undergoer. (3) a.
Miyaye rüngün-üñma-nge-ke-i-Ø ñi ütrar. jimson.weed grind-appl-pass-hab-ind-3.fp 3.psr pip ‘One grinds the jimson weed pips.’
b. Fey tuku-lel-nge-ke-i-Ø fürkü ko. 3sg put-appl-pass-hab-ind-3.fp cold water ‘One adds cold water to it.’
With some verbs, the state of affairs can be seen either as causing the theme to approach this additional participant (-lel) or as separating the former from the latter (-ñma). In Salas’ view, this spatial displacement can be sometimes taken to mean detriment or benefit of the additional participant, but “the detriment/benefit contrast is secondary and accidental” (p. 123); in many cases these notions do not apply. The sentences in (4) are from Salas (2006:122–123): (4) a.
Ngilla-ñma-fi-i-n Antonio ñi kawellu. buy-appl-3.sp-ind-1sg.fp A. 3.psr horse ‘I bought Antonio’s horse from him.’
b. Ngilla-lel-fi-i-n Antonio ñi kawellu. buy-appl-3.sp-ind-1sg.fp A. 3.psr horse ‘I bought Antonio’s horse for him.’
The transitive verb ngilla- ‘buy’ takes a human actor (the buyer) and a non-human undergoer (the theme). When applicativized, the 1st person actor is the FP, Antonio is the PSP and triggers fi-marking on the verb, and ñi kawellu ‘his horse’ is the SSP. Antonio is the seller and parts with his horse in (a), while in (b) Antonio is the intended recipient of the horse and the 1st person bought it on his behalf.
7.â•… The verb ‘grind’ is normally found either as a simple root (rüngü-) or as a causative stem (rüngü-m-); the form rüngün- given by Salas is anomalous.
Fernando Zúñiga
4.â•… The applicatives -ñma and -l 4.1â•… The applicative -ñma The three old accounts of the two productive Mapudungun applicatives are rather sketchy. Valdivia (1606) mentions a suffix -nma only briefly and gives (5a) below as his only example. Febrés (1765) treats both -ñma and -l as if they were interchangeable and notes two allomorphs for the former, viz. -ñma or -nma, without stating the allomorphy conditions; one of his two examples is given here in (5b). Finally, few examples of the applicative -ma ~ -nma (the latter being the postvocalic allomorph) are found in Havestadt (1777); for our present purposes let me cite (5c) here: (5) a.
Dungu-ma-e-n! speak-appl-inv-1sg.ind.fp ‘Speak for me!’
(Valdivia 1606:45)
b. tañi weñe-nma-nge-n mew ta manshun ika 3.psr steal-appl-pass-1sg.ind.fp ppos part ox pl ‘because they (impers) stole the oxen from me’ (Febrés 1765:48–49) c.
Küpa-l-ma-a-e-y-mi. come-caus-appl-inv-ind-2sg.fp ‘I will bring [it] for you (sg).’
(Havestadt 1777:95)
According to Augusta’s grammar (1903), the allomorphs of -ñma are distributed as follows: -üñma after consonants (as in langüm-üñma- ‘kill on (somebody)’, from langüm- ‘kill’), -ñma after vowels (as in weñe-ñma- ‘steal from (somebody)’, from weñe- ‘steal’) and -ma after diphthongs (as in küdaw-ma- ‘work for (somebody)’, from küdaw- ‘work’). However, Augusta lists only few verbs that take -üñma in his dictionary (1916), e.g. kalpud- ‘pull out’, kim- ‘know’ and trolol- ‘hollow out’. The suffix -ñma appears postvocalically as predicted in his grammar, but it is not the case that the allomorph -ma only appears after diphthongs: it regularly appears after laterals (e.g. lol- ‘become furrowed’, wifül- ‘throw (liquid)’, püra-kawell- ‘mount a horse’; actually, trolol- is the only exception to this rule I found in the dictionary), nasals (e.g. ñiküm- ‘shelter (a house) from the wind’, patron- ‘find a boss’, lawen- ‘use (something) as medicine’, trangliñ- ‘frost, freeze’), g (e.g. treg- ‘spark’) and sometimes even fricatives (e.g. llipad- ‘(birds) fly towards (a prey)’, ñidüf- ‘sew’). Salas (2006:121f) mentions (i) the phonologically conditioned allomorphy one would expect based on the behavior of other verbal suffixes, viz. -ñma after vowels and -üñma elsewhere, and (ii) a morphologically conditioned allomorphy that explains the maform after causative -l(el) (6a). Also note that Salas postulates a homonymous “participative” suffix -ma that signals participation or personal involvement in the state of affairs on the part of the FP (6b):
Benefactive and malefactive applicativization in Mapudungun
(6) a.
Küpa-l-el-ma-nge-i-mi kuram tami ñuke. come-caus-appl-appl-pass-ind-2sg.fp8 egg 2sg.psr mother ‘They (impers) brought eggs for your (sg) mother.’ (Salas 2006:124)
b. Maw-ma-n. rain-???-1sg.ind.fp ‘I got rained on.’
(Salas 2006:127)
Similarly, Smeets (2008) labels one -ñma “indirect object” and describes it as changing ‘the patient into 1. a participant which is adversely affected by the event [… and] 2. a participant that is indirectly affected by the event, viz. through possession of or relationship with the patient (p. 276). According to this author, the allomorphs are -üñma after consonants, -ñma after vowels and -ma after diphthongs, l and sh. Smeets labels a second -ñma suffix “experience” and characterizes it as indicating that “the subject of an intransitive verb experiences the process or event denoted by the verb” (p. 301); -ñma is the postvocalic, -ma the postconsonantal allomorph and they are said to occur in a slot closer to the verbal root. One of her examples of each suffix is given in (7): (7) a.
I-ñma-nge-i-Ø kiñe sañchu ñi karukatu. eat-appl-pass-ind-3.fp one pig 1sg.psr neighbor ‘They (impers) ate a pig of my neighbor’s.’ (Smeets 2008:276)
b. Iñche aku-ñma-n
kiñe küme dungu.
1sg arrive-ma-1sg.ind.fp one good message ‘I received a nice message.’
(Smeets 2008:302)
Finally, Harmelink (1996:231) recognizes only one -ñma and labels it “indirect patient”. He identifies its allomorphs as postconsonantal -üñma and postvocalic -ñma, and two of his examples are given in (8). Note that Harmelink argues that -ñma is neutral between malefactive and benefactive and that its interpretation depends on the semantics of the verb stem and/or the context. (8) a.
Wülel-üñma-e-n-mew tañi fotüm. hit-appl-inv-1sg.ind.fp-3.sp 1sg.psr son ‘S/he hit my son.’
b. Kellu-ñma-e-i-mi-mew tami ñawe. help-appl-inv-ind-2sg.fp-3.sp 2sg.psr daughter ‘S/he helped your (sg) daughter.’
My own work with middle-aged urban speakers of Chilean Mapudungun suggests that they are not always sure as to (i) whether a verb can take -ñma or not, (ii) the
8.â•… Salas’ analysis is incorrect here: küpa-l-el- is trimorphemic instead of simply bimorphemic (küpa-lel-) as stated in his gloss. See Section 4.2 for details.
 Fernando Zúñiga
exact meaning such a form would have, and (iii) whether the correct form of the suffix is -üñma or -ma after consonants. There seems to be some dialectal variation as well: the speakers whose forms most closely resemble the ones given by Augusta are originally from the central zone (Region IX, in terms of the Chilean administrative divisions), whereas those speakers from northern and southern regions (VIII, X and XIV) tended to inconsistently produce forms that deviated from Augusta’s. However, the conditions of occurrence of Smeets’ “indirect object” suffix are incomplete as well, since not only diphthongs, l and sh are followed by -ma. Further observe that there are two different patterns of morphosyntactic behavior for constructions where -ma ~ -(ü)ñma occurs. Consider the following examples (adapted from Smeets 2008:277,302): (9) a.
Langüm-üñma-fi-i-Ø ñi fotüm. kill-appl-3.sp-ind-3.fp 3.psr son ‘S/he killed his son.’
b. Kon-ma-i-Ø trüfür ñi nge mew. enter-appl-ind-3.fp dust 3.psr eye ppos ‘S/he got dust in her/his eye.’
In both sentences, a new participant is introduced when compared to the clauses without -ma ~ -(ü)ñma. The non-applicative counterpart of (a) is langümfi ñi fotüm ‘s/he killed his son’, where no explitic reference is made to the father’s affectedness. Similarly, the counterpart of (b) is koni trüfür ñi nge mew ‘dust entered her/his eye’, an unidiomatic sentence structurally analogous to its English translation in that there are only two syntactic arguments: trüfür ‘dust’ and ñi nge ‘her/his eye’. Nevertheless, note that -ñma as applied to a transitive predicate (langüm- ‘kill’ in 9a) does not change the syntactic status of the original FP; it merely demotes the original PSP to SSP and introduces a new PSP. By contrast, when -ma is applied to an intransitive predicate (kon- ‘enter’ in 9b), the original FP is demoted to SSP and the new argument is the new FP. (This behavior is also found with the applicative -l(el), as shown in the next subsection.) Such a pattern is remarkably consistent; the examples in Salas (2006) and those I have elicited from my informants confirm the accuracy of virtually all relevant entries in Augusta (1916). The list of intransitive verbs that appear to frequently take -ma includes treg- ‘sparkle’, wün- ‘dawn’, pun- ‘become night’, dumiñ- ‘get dark’ and aku- ‘arrive here’. The examples found in Smeets (2008:302–303) also confirm this, but with two qualifications. First, she notes that such applicativized intransitives may take the agentless passive suffix -nge, a feature that was also noted by Havestadt for the intransitive verb la- ‘die’ and for the ditransitive verb elu- ‘give’ (1777:96; with his original wordings in Latin): (10) a.
La-ñma(-nge)-n kiñe kawellu. die-appl-pass-1sg.ind.fp one horse ‘One of my horses died/a horse died on me.’(mihi periit, mortuus est equus)
Benefactive and malefactive applicativization in Mapudungun 
b. Kiñe elu-ñma(-nge)-n. one give-appl-pass-1sg.ind.fp ‘One was given to me/I was given one.’ ( mibi [sic] datum est unum)
I interpret this variation as a sign of uncertainty on the part of Havestadt’s consultants as to the syntactic status of the affected participant at a time when, presumably, the productivity of the applicative -(ü)ñma with intransitives was higher than today. Second, Smeets (2008:303) gives one example that, in her analysis, shows that there might be two different slots after all: (11) Anü-ñma-nie-ñma-en ñi makuñ. sit.down-appl-have-appl-2sg → 1sg 1sg.psr coat ‘You (sg) sat on my coat.’
Interestingly enough, Smeets observes on the very same page that “the [suffix] -ma […] may form part of the stem”. As a matter of fact, the latter is precisely the analysis I would like to propose here as an alternative to hers, since I have not been able to find conclusive evidence of two different slots for -ñma with non-complex bases, and verb stem serialization is, albeit still ill-understood, a recognized word formation processes in the language (Salas 2006; Zúñiga 2006:178–181). The first complex stem anü-ma- ‘sit on (something)’ would then combine with the other stem nie-ñma- ‘have (something) for/on (somebody)’ to yield the complex base anü-ma-nie-ñma-. Summing up, I think it is safe to say that there are two -ñma’s, defined not only by their approximate allomorphy rules but also according to transitivity: (12) a.
Applying to transitives, leaving FP intact and introducing new PSP:
-ñma after vowels -ma after g, liquids, nasals and some fricatives -üñma elsewhere
b. Applying to intransitives, demoting original FP and introducing new one: -ñma after vowels -ma elsewhere
4.2â•… The applicative/causative -l(el) Valdivia (1606:45) mentions a causative with the variants -l for the southern and -l-ka for the northern dialects, with two examples: ngolli- and lakutu- ‘get drunk’ vs. ngollil-ka- and lakutu-l-ka- ‘make (someone) get drunk’.9 He further mentions a suffix -l that
9.â•… Only ngolli- means ‘get drunk’; lakutu- refers to a gift exchange ritual between namesakes (particularly, paternal grandfathers and their grandchildren, both called laku).
 Fernando Zúñiga
turns intransitive verbs (“neutros”) into transitives (“activos”) and gives examples built upon anü- ‘sit down’ and küpa- ‘come’ (1606:41). He does not seem to acknowledge -l as applicative. Febrés’ (1765:48–49) few examples of the use of this applicative include the following (note the variation between -lel and -l with the same verb in 13a/13b; cf. also further down): (13) a.
Kintu-lel-fi-nge kawellu tami chaw. look.for-appl-3.sp-2sg.imper horse 2sg.psr father ‘Look for your (sg) father’s horse!’
b. Kintu-l-u-fi-nge tami kawellu. look.for-appl-refl-3.sp-2sg.imper 2sg.psr horse ‘Look for your (sg) horse!’ c.
Ketra-lel-nge-la-n mapu. plow-appl-pass-neg-1sg.ind.fp land ‘They (impers) did not plow my land.’
Havestadt (1777:93–94) says -lel is an honorific marker and gives one Example (14): (14)
Fey pi-lel-e-n tami chaw. 3sg say-appl-inv-1sg.ind.fp 2sg.psr father ‘Please tell your (sg) father in my name.’ (in the original: dignetur, non gravetur dicere suo patri meis verbis, meo nomine)
Furthermore, he implicitly suggests that -l can appear without recognizable semantic yield in some cases and that -l and -lel are merely variants of the same formative (pp.€162–164). Augusta (1903:61–62) says that -l is the postvocalic allomorph of the applicative and that -el occurs elsewhere: katrü- ‘cut’ gives katrü-l-, küpal- ‘bring’ gives küpal-el(remember that küpal- is bimorphemic: küpa- ‘come’ + -l ‘causative’), nentu- ‘extract, get out’ gives nentu-l- and recursively nentu-l-el-. Pi- ‘say, tell’ apparently takes -lel: fey pi-lel-en tami chaw ‘tell (it) to your father’ (the same example as Havestadt’s in (11) above), and with kintu- ‘look for’ both -l and -lel are grammatical without noticeable difference in meaning: kintu-l(el)-en tañi chumpiru ‘look for my hat’. Further consider Augusta’s example are-l- ‘lend (something to someone)’, which takes a FP, an animate PSP and an inanimate SSP, as opposed to are-nge-l-m- ‘lend (something)’, which only takes a FP and an inanimate PSP. The latter verb includes a passive, the applicative and a low-animacy/control causative: ‘cause (something) to be lent (to someone)’; *aredoes not exist with the meaning ‘lend (something)’. Similar, but not as neatly built, is pe-nge-l- ‘show (something to someone)’, derived from pe- ‘see (something)’.
Benefactive and malefactive applicativization in Mapudungun 
Salas’ (2006) brief analysis of -l(el) has already been addressed in Section 3 above; this author does not provide any additional details as to allomorphy conditions. Harmelink (1996:232–233) distinguishes the uses of -l with intransitives (i.e. high animacy/control causative, e.g. tremo- ‘be/get healthy’ vs. tremo-l- ‘heal (tr.)’) from those with transitives (i.e. applicative, e.g. küpal-el- ‘bring (something to someone)’) and proposes a unified account for it as “transitivity increase” marker. Harmelink gives the following allomorphy rules: -l after vowels, -ül after w, and -el elsewhere. Here the sequence -l-el is, contrary to what Salas says, not morphologically but actually phonologically conditioned. Smeets (2008) distinguishes between the causative (pp. 299–301) and the “beneficiary” (pp. 278–280) suffixes with respect both to their meaning and to their templatic position within the verb complex. The former has the allomorphs -l (after vowels) and -ül (after consonants), which in turn “alternates with -el” with some verbs (e.g. küdaw- ‘work’, trupef- ‘get frightened’ and leg- ‘hit (itr.)’). However, some verbs only take -el (her list consists of af- ‘end’, aywiñ- ‘become a shadow’, kim‘know’, ariñ- ‘get burned’, mutrur- ‘bump against (something)’, wim- ‘get used’, adÂ� ‘become beautiful’ and lingar- ~ fülang- ‘become white’).10 By the same token, i- ‘eat’ is said to take -lel. With some ambitransitive verbs, -l does not introduce a causer to the clause but signals unintentional action instead: llocho- ‘loosen/get loose’, ngüfo- ‘make/get wet’, and wacho- ‘break (itr./tr.)’.11 By contrast, Smeets’ beneficiary “changes the patient of an event into the beneficiary”: its marker is -l ~ -lel after vowels and -el elsewhere. Unlike Salas, Smeets finds that the applicative sequence -lel-ma is infrequent and the sequence -ñma-ñma is ungrammatical. She gives an example of the infrequent sequence, viz. anü-künu-lel-ma-fi-n (sit.down-leaveappl-appl-3.sp-1.sg.ind.fp) ‘I left (it) for him’, as well as one of an “unreliable” instance of the sequence -lel-el: anü-künu-lel-el-fi-n (sit.down-leave-appl-appl-3. sp-1.sg.ind.fp) ‘I left (it) for him’. Finally, Smeets mentions that wülel- ‘hit (tr.)’ (cf. Example 9 above) might be derived from wül- ‘hand, give’, and that both the latter verb and elu- ‘give’ do not appear with -l.
10.â•… Her example traf- ‘fit, be joined, gather (itr.)’ vs. trap-el- ‘tie’ is different because there is a low-animacy/control causative formation with -m as well: trap-üm ‘gather (tr.)’. Since this is the only instance I am aware of where root-final f becomes p before -l, I prefer to explain this alternation as an “anomalous” analogy to or back-formation from trapümün. 11.â•… Smeets (1989:397) gives katrü- ‘cut’ as belonging to this category, i.e. katrü-l- would mean ‘cut unintentionally’. Nevertheless, according to my informants this non-applicative reading is ruled out; this may be an idiolectal peculiarity of the speech of Smeets’ informants or a dialectal trait.
 Fernando Zúñiga
My informants’ responses to the forms given by Augusta’s dictionary and other authors mentioned above suggest that there is both a general pattern and interesting exceptions. The general pattern is the following: -l is used after vowels and -el elsewhere. The exceptions are i- ‘eat’, ina- ‘follow’, kullkü- ‘reserve (domestic animals)’, pe- ‘see’ and pi- ‘say’, all of which take -lel instead of -l (Augusta does not mention kintu- ‘look for’ with applicative morphology), and the verbs given in (15), all of which take -ül: (15) a.
miaw- ‘walk’ ñawñaw- ‘meow’ puw- ‘arrive there’ ünew- ‘feel uncomfortable’ piaw- ??12 kamañkiaw- ‘go around serving as a shepherd’ küñatuyaw- ‘go around hand in hand’
b. mefür- ‘shake’ püllüf- ‘splinter’ trüker- ‘roll one’s eyes’ trünüy- ‘vibrate’ trupef- ‘get scared’ patrüg- ‘spread (tr.)’ nüwf- ‘hand by extending one’s arm’
The verbs in (15a) end in aw, ew, iw or uw—kamañkiaw- and küñatuyaw- due to the perambulative suffix -(ki)yaw-, but observe that there are some verbs that take -el under the same phonological conditions: küdaw- ‘work’, fi-ñma-w- ‘get distressed, become anxious’, rew- ‘be stirred’ and wew- ‘win (itr./tr.)’; also llow- ‘receive, accept’, which ends in Vw, takes -el instead of -ül. Intransitives and transitives do not behave differently, although the meaning of miaw-ül- is not the expected one (‘make someone walk’): ‘take something along’. The verbs in (15b), by contrast, end on a consonant or a glide but do not take -el. Verbs that do take -el, like traf- ‘gather (itr.)’ and trag- ‘pass from one hand to another’, are in opposition to verbs ending in the same non-vowels (labial and velar) but taking -ül instead, like nüwf- ‘hand by extending one’s arm’ and patrüg- ‘spread (tr.)’. Here it seems that the segment ü in the root syllable triggers assimilation in the suffix, which would account for püllüf-, trünüy-, mefür-, pütef-, and trüker-. However, there are exceptions to this rule: trupef- takes -ül in spite of not having ü in the root, but luyüf- ‘become shiny’ takes -el in spite of being like nüwf-. Furthermore, with pütef‘sprinkle chicha’, both pütef-el- and pütef-ül- are attested with the same meaning, and of course all verbs listed with -ül in (15b) whose root-final segment is f can be found with -el as well, depending on the specific informant one asks. Finally, there is one verb where both derivations appear with a difference in meaning: the ambitransitive ütrüfmeans either ‘lay down’ or ‘throw’, hence ütrüf-el- ‘throw (something to someone)’ and ütrüf-ül- ‘throw away’. Some of my urban informants have confirmed the opposition
12.â•… The simple verb piaw- does not seem to be in use anymore; piaw-ül- means ‘bother (somebody)’.
Benefactive and malefactive applicativization in Mapudungun 
between ütrüf-el- and ütrüf-ül- while others identify ütrüf-ül- with the meaning ‘throw (something to someone)’ as the only possible derivation. I have not been able to find conclusive evidence supporting Smeets’ claim that there are two different l-suffixes with different allomorphy rules and in two different slots of the verbal template. Therefore, I must come to Harmelink’s conclusion with a slightly revised allomorphy rule: there is only one suffix that (i) causativizes intransitives and ambitransitives and (ii) applicativizes transitives and is realized as -l after vowels, -ül after w and -el elsewhere. The exceptions to this general rule are as follows: with some frequently used monosyllabic roots like i-, pe-, and pi-, but also with some others like ina-, kullkü- and kintu-, the allomorph -lel is used (which is homophonous with the causative-applicative sequence -l-el); with some roots including ü and ending in f or r, -el is realized as -ül, and, to the extent that ütrüf-el- and ütrüf-ül- both exist with different meanings, this latter example would be the only one suggesting a possible grammaticalization of wül- ‘hand (something)’. For all other forms either elu- ‘give’ or, perhaps, el- ‘put, create, array’ is the probable etymon.
5.â•… The ñma-l(el) opposition Based on the discussion in the preceding section, I think the intuition of the older authors was on the right track: both -ñma and -l(el) were originally transitivizers. When applied to intransitive predicates, -ñma introduces a participant affected by the event and -l(el) introduces a causer: (16) a.
Lleg-ma-n fotüm. be.born-appl-1sg.fp.ind son ‘A son was born to me.’
b. Aye-l-fi-i-Ø ta malen. laugh-caus-3.sp-ind-3.fp part girl ‘S/he made the girl laugh.’
(Golluscio 2007)
The semantics of -ñma and -l(el) becomes similar when they are applied to transitive predicates: ruka-ñma- means ‘build a house over (something) to protect it’ (cf. ruka‘build a house’), and ruka-l- can mean not only the same as ruka-ñma- but also ‘build a house for (somebody)’. Even though the documented history of the language does not allow me to be more conclusive about this, the evidence suggests that the semantics of -ñma was originally neutral (i.e. covering both benefaction and malefaction) while the semantics of -l(el) was not. Whether the etymon of the latter suffix is el- ‘create, put’ or elu- ‘give’, the meaning [cause PSP to have SSP] is arguably still active in pairs like ngilla-ñma- ‘buy from (somebody)’ vs. ngilla-lel- ‘buy for (somebody)’.
 Fernando Zúñiga
Such a privative opposition between an unmarked -ñma and a marked -l(el) accounts for all cases I am aware of where these suffixes can appear with the same predicate, as well as those cases where only one or the other is found. In examples like those in (17) (Smeets 2008:276), then, the malefactive interpretation is derivative; -ñma does not express that the event takes place to the affectee’s detriment and is thus semantically parallel to the Spanish dative construction given in the additional translations: (17) a.
Nü-ñma-fi-n. take-appl-3.sp-1sg.fp.ind ‘I took it away from her/him.’ (Sp. se lo tomé/quité)
b. Femngechi weñe-nie-ñma-rke-fi-i-ngün ñi mapu thus steal-have-appl-rep-3.sp-ind-3pl.fp 3.psr land
pu mapuche. pl M.
‘In that way, I heard, they (pl) kept robbing the Mapuche of their land.’ (Sp. les robaban su tierra a los mapuches)
By contrast, the benefactive interpretation in cases like those in (18) is actually the original meaning of -l(el) if we consider that this suffix might well be the grammaticalized version of el- ‘create’ or elu- ‘give’: (18) a.
Nie-l-eyu. have-appl-1sg → 2sg ‘I hold it for you (sg).’
b. Kintu-lel-en mamüll. look.for-appl-2sg → 1sg wood ‘You (sg) looked for (fire)wood for me.’
(Smeets 2008:278)
(Salas 2006:122)
There are cases occasionally found in more recent texts where -l(el) does not have a plausible interpretation in terms of benefaction or transfer. I interpret cases like these as evidence of a more progressive, i.e. dative-like, status of -l(el). Note that contact with Spanish is a likely source of this extension—especially in (19), which is a translation from Spanish: (19)
welu katrü-l-nge-lle-i-Ø trawü-no-a-el but cut-appl-pass-cert-ind-3.fp gather-neg-fut-nfin ‘but they (pl) are prevented from gathering’ (Ercilla y Zúñiga 2006 I.6) (original: mas esles impedido el allegarse)
It is not uncommon for one and the same morpheme to have both a causative and an applicative function, as mentioned frequently in the literature and discussed in some
Benefactive and malefactive applicativization in Mapudungun 
detail in Peterson (2007:64–66). In Bella Coola, for example, some verbs can take endings from the so-called causative paradigm and be ambiguous as to its exact interpretation (20a) while others exclude even the deputative benefactive (i.e. applicative) reading (20b) (Davis & Saunders 1997:28): (20) a. b.
Nuyamł-tus ti-’imlk-tx ti-’immllkī-tx. sing-3sg → 3sg deic-man-deic deic-boy-deic ‘The man sang for the boy.’/‘The man made the boy sing.’ Q’s-tus ti-nus’ūlχ-tx ti-’imlk-tx. be.ill-3sg → 3sg deic-thief-deic deic-man-deic Only: ‘The thief made the man ill.’
More relevant in the present context, however, are the following Hualapai examples (Ichihashi-Nakayama 1996:228,231), where unergatives like ‘sing’ (21a) are applicativized when transitivized by the applicative suffix -o ~ -wo whereas unaccusatives like ‘cry’ (21b) are normally causativized instead: (21) a.
Jean-ch ba ma-swa:d-o-y-k-i. J.-subj pl.obj 3 → 2-sing-appl-fut-3-aux ‘Jean will sing for you all.’
b. Ma:-ch nya mi-mi:-wò-ng-wi-ny. 2sg-subj 1sg 2 → 1-cry-caus-2-aux-pst ‘You (sg) made me cry.’
Summing up, the formatives -ñma and -l(el) introduce an additional participant to the clause, which is best described as broadly non-patientive. With -ñma, this participant is portrayed as affected by the event—either in a unspecified manner or, if the option of a l(el)-construction is available, in an adverse way. With -l(el), the additional participant is a causee with high control/animacy intransitives (i.e. an agentive participant) and a beneficiary with transitives. In other words, -l(el) is a causative and a somewhat imperfect benefactive applicative, while -ñma is a transitivizer whose malefactive character with transitives arises as a result of the opposition with -l(el). My impression is that the current trend is to use intransitives with -ñma less frequently, and possibly also to use -l(el) forms more freely, i.e. without implying that there is some benefit for the affectee. If these transitivizers in fact developed along the lines proposed here, Mapudungun provides an interesting example of how a particular section of verbal paradigms may come about. It might prove fruitful to study valency-changing operations in other languages from the perspective of semantic/pragmatic privative vs. equipollent oppositions in order to better understand how they originated, how they really work, and perhaps even where they are headed.
 Fernando Zúñiga
Abbreviations appl art autoben aux caus ind inv nfin obj part pass pl ppos psp
applicative article autobenefactive auxiliary causative indicative inverse nonfinite object particle passive plural postposition primary satellite person
deic du fp fut impers psr pst refl rep sg sp ssp subj
decitic dual focal person future impersonal possessor past reflexive reportative singular satellite person secondary satellite person subject
References Augusta, Félix José de. 1903. Gramática araucana. Valdivia: Imprenta Central J. Lampert. Augusta, Félix José de. 1916. Diccionario araucano. Mapuche-español, español-mapuche. Santiago: Cerro Manquehue. Davis, Philip & Saunders, Ross. 1997. A Grammar of Bella Coola [Occasional Papers in Linguistics 13]. Missoula MT: University of Montana. Ercilla y Zúñiga, Alonso de. 2006. La Araucana. Ta Awkan Mapu mew. Versión bilingüe castellanomapuzugun. Santiago: MINEDUC, Programa de Educación Intercultural Bilingüe. Febrés, Andrés. 1765. Arte de la Lengua General del Reyno de Chile. Lima: Calle de la Encarnación. Golluscio, Lucía. 2007. Morphological causatives and split intransitivity in Mapudungun. International Journal of American Linguistics 73(2): 209–238. Harmelink, Bryan. 1996. Manual de aprendizaje del idioma mapuche: aspectos morfológicos y sintácticos. Temuco: Ediciones Universidad de la Frontera. Havestadt, Bernhard. 1777. Chilidúg´u sive tractatus linguae chilensis. Leipzig: B.G. Teubner. Ichihashi-Nakayama, Kumiko. 1996. The ‘applicative’ in Hualapai: Its functions and meanings. Cognitive Linguistics 7(2): 227–239. Peterson, David. 2007. Applicative Constructions. Oxford: OUP. Salas, Adalberto. 2006. El mapuche o araucano. Fonología, gramática y antología de cuentos. 2nd, revised edn. Santiago: Centro de Estudios Públicos. Smeets, Ineke. 1989. A Mapuche Grammar. Ph.D. dissertation, University of Leiden. Smeets, Ineke. 2008. A Grammar of Mapuche. Berlin: Mouton de Gruyter. Valdivia, Luis de. 1606. Arte y confesionario de la lengua de Chile. Leipzig: B.G. Teubner. Zúñiga, Fernando. 2006. Mapudungun—el habla mapuche. Introducción a la lengua mapuche, con notas comparativas y un CD. Santiago: Centro de Estudios Públicos. Zúñiga, Fernando. In preparation. An exploration of the diachrony of Mapudungun valencychanging operators.
The benefactive semantic potential of ‘caused reception’ constructions A case study of English, German, French, and Dutch Timothy Colleman Ghent University
English, German, French and Dutch all exhibit a ditransitive construction in which the verb is combined with a subject and two NP objects (either unmarked or with overt dative and/or accusative case), which typically encode the agent, theme and recipient of a ‘caused reception’ event. In construction grammar terms, Goldberg (1995) posits ‘Agent successfully causes Recipient to receive Patient’ as the basic sense of the (English) ditransitive. Each of these ditransitive constructions, however, can also be used to encode a variety of other scenes in addition to basic transfer of possession events. This paper investigates the extent to which they can be used to encode benefactive events, i.e. events where an agent carries out an action (involving a patient) for the benefit of another person. The constructions in question will be shown to be subject to distinct constraints in this respect. Kittilä’s (2005, this volume) distinction between recipientbeneficiaries and other beneficiaries, for instance, turns out to be an important factor in English and (certain varieties of) Dutch, but not (or to a much lesser extent) in German and French. In addition, it will also be shown that in several of the languages under discussion, the semantic range of the benefactive ditransitive is subject to language-internal variation.
1.â•… Introduction As a starting point for this paper, we begin by observing the close conceptual similarity between the semantic roles of recipient and beneficiary, which motivates why many languages encode events of caused reception and (certain) events of benefaction by means of the same construction, or, put differently, why in many languages recipient markers double up as beneficiary markers; see Shibatani (1996), Newman (1996: 95–97, 217–223) and Kittilä (2005), inter alia, for examples from various languages and general discussion. For instance, while the Thai example in (1a) below illustrates the use of the construction with the secondary verb hây ‘give’ to encode a fairly prototypical act of caused reception, the examples in (1b) and (1c) illustrate the use of the same
Timothy Colleman
construction to encode an event in which an action is performed for somebody’s benefit (also see Jenny this volume). (1) a.
kháw sòŋ còdmăaj hâj phŷan he send letter give friend ‘He sends his friend a letter.’
b. dεεŋ jiŋ nóg hâj sùdaa pn shoot bird give pn ‘Deng shoots a bird for Sudaa.’ c. dεεŋ paj talàad hâj sùdaa pn go market give pn ‘Deng is going to the market for Sudaa.’ (Kittilä 2005: 285, cited from Bisang 1992: 366)
The present paper looks into a number of prototypical ditransitive ‘caused reception’ constructions in English, Dutch, German and French and investigates to which extent they can be used to encode events of benefaction as well.1 The constructions to be investigated are (i) the English double object construction with two unmarked NP objects exemplified in (2a) below; (ii) the German construction with dative and accusative objects in (2b); (iii) the French construction with a direct object and a dative clitic in (2c); and (iv) the Dutch double object construction exemplified in (2d).
(2) a.
The man gave the woman an expensive ring.
b. Der Mann hat dem Jungen einen Laptop geschenkt. the.nom man has the.dat boy a.acc laptop presented ‘The man has given the boy a laptop (as a present).’ c.
Son oncle lui a envoyé une somme d’ argent. his/her uncle him/her.dat has sent a sum of money ‘His/her uncle has sent him/her a sum of money.’
d. De winkelier verkocht mijn vader een fles wijn. the shopkeeper sold my father a bottle wine ‘The shopkeeper sold my father a bottle of wine.’
The examples in (2) illustrate the basic function of the respective ditransitive constructions, viz. to encode a three-participant transfer of possession event in which an agent causes a (typically inanimate) theme to pass into the possession of a (typically animate) recipient. Goldberg’s (1995) construction grammar analysis of the English
1.â•… Following Malchukov et al. (2007), among others, we use the term ditransitive in a broad sense throughout this paper, as referring to any three-argument construction with an agent-like, a themelike and a recipient-like argument. In construction grammatical studies, the term ditransitive construction is sometimes used in a narrower sense, as a synonym for double object construction (i.e. a construction with two unmarked NP objects), see e.g. Goldberg (1995) and Colleman (2009).
Benefactive semantic potential of ‘caused reception’ 
double object construction, for instance, posits ‘Agent successfully causes Recipient to receive Patient’ as the basic sense of this construction. However, while the above examples all involve verbs of giving in the strict sense of the word, it is well-known that the constructions in question can also be combined with verbs from a number of other semantic classes, which do not denote the causation of an actual transfer of possession. To give but two examples, each of the constructions in (2) also accommodates a variety of verbs of communication (e.g. tell, ask, phone, etc. in English; sagen ‘say’, berichten ‘report’, erzählen ‘tell’, etc. in German; dire ‘say’, déclarer ‘state’, répondre ‘answer’, etc. in French; vertellen ‘tell’, zeggen ‘say’, meedelen ‘communicate’, etc. in Dutch) as well as a number of verbs which denote the prevention rather than the causation of possession (e.g. refuse and deny in English; verweigern ‘refuse’, vorenthalten ‘withhold’, etc. in German; refuser ‘refuse’, dénier ‘deny’, etc. in French; weigeren ‘refuse’, ontzeggen ‘deny’, etc. in Dutch); also see the semantic classification of three-participant events presented in Margetts & Austin (2007). In construction grammar terms, the combinations of these ditransitive constructions with verbs from other classes than verbs of giving instantiate various extended subsenses rather than the basic ‘caused reception’ sense: English double object clauses with refuse and deny, for instance, are associated with the subsense ‘Agent causes Recipient not to receive Patient’ in Goldberg (1995). In terms of the semantic map approach advocated in Haspelmath (2003), inter alia, the indirect or dative objects in the constructions exemplified in (2) cover a region in semantic space which next to the recipient function also includes a number of related functions such as addressee and deprivee (cf. also Malchukov et al. 2007). The basic question to be answered here is whether this region also includes beneficiary functions. Sections 2 to 5 deal with each of the above ‘caused reception’ constructions in turn, in the order in which they are presented in (2). Most attention in this overview will be devoted to the double object constructions of English and Dutch, since these display the most interesting semantic constraints and language-internal variation. The discussion will repeatedly refer to the various subtypes of the general beneficiary role distinguished in Van Valin & LaPolla (1997), Kittilä (2005) and Kittilä & Zúñiga (this volume); it will be shown that in several of the languages under discussion, recipientbeneficiaries (see below) are treated differently from other subtypes. Section€ 6 summarizes the main findings.
2.â•… Benefactive uses of the English double object construction 2.1â•… An “intended reception” constraint The English double object construction can be used to encode events of benefaction, but not without important restrictions, as shown by the grammaticality contrast between (3) and (4).
Timothy Colleman
(3)
a. b. c. d.
(4)
a. b. c. d.
The man bought the boy an apple. Shall I fry you some eggs? My wife knitted me a nice sweater. Albert made Erica a picture. *John opened Mary the door. (Shibatani 1996: 170) *Uncle Jim cleaned Margaret the windows. (Allerton 1978: 30) *Uncle Jim watched Margaret a television programme. (Allerton 1978: 23) *Sue fixed Dick the radiator. (Jackendoff 1990: 196)
While the indirect object referent is clearly involved as the beneficiary of the subject’s actions in both (3) and (4) – as shown by the possibility of a paraphrase with a benefactive for-PP in all cases – the double object construction is fine in (3) but not in (4). This contrast has been signalled by many authors and is often described in terms of an “intended reception” constraint: for the double object construction to be possible in English, the beneficiary has to be involved as the intended recipient of the direct object referent (e.g. Allerton 1978; Wierzbicka 1988; Jackendoff 1990: 195–196; Langacker 1991: 360; Wechsler 1995; Goldberg 2002; Nisbet 2005). In (3a), for instance, the subject’s buying action is aimed at a subsequent transfer of an apple to the boy in question. Similar intended transfers are at stake in the other clauses in (3), but not in the clauses in (4). This constraint can be rephrased in terms of Kittilä’s (2005) distinction between recipient-beneficiaries and (pure) beneficiaries: the English double object construction only accommodates events of recipient-benefaction, i.e. events in which the nature of benefaction is such that the beneficiary ultimately receives something by instigation of the agent. The ungrammatical examples in (4) involve a different kind of benefaction, which does not involve an intended transfer but in which the agent carries out an action instead of the beneficiary. This is labelled deputative benefaction in Van Valin & LaPolla (1997: 383–84) and substitutive benefaction in Kittilä (2005; cf. also Newman’s 1996: 220 “true benefactives”). Goldberg (1995, 2002) incorporates this view in her semantic network model of the English double object construction: she distinguishes a sub-sense ‘Agent intends to cause recipient to receive patient’, which covers double object clauses with verbs of creation (including preparation) such as knit, make, bake, fry, cook, etc. and with verbs of obtaining such as buy, get, find, etc. These two classes (or three, depending on whether creation and preparation are treated as a single class or not) are invariably included in the overviews of verb classes conventionally associated with the English double object construction offered in the literature (also see Green 1974; Pinker 1989 and Levin 1993, inter alia). Within these classes, the construction can be used more or less productively. It should be noted, however, that contrary to what is sometimes suggested (e.g. in Pinker 1989), a verb need not belong to one of these classes to be eligible for use in the benefactive double object construction. This is nicely illustrated by the examples from Langacker (1991) in (5).
Benefactive semantic potential of ‘caused reception’
(5) a. *I cleared him the floor. b. I cleared him a place to sleep on the floor.
(Langacker 1991: 360)
While clear, in isolation, is not a verb of creation, preparation or obtainment, the event in (5b) can nevertheless be construed as an act of recipient-benefaction and can hence be encoded by means of the double object construction, in contrast to the event in (5a). In Langacker’s terms: “[C]learing a floor does not per se have any possessive implications. However, clearing someone a place to sleep on the floor makes him a possessor in the sense of having that place at his disposal for a particular purpose” (1991: 360). Shibatani (1996: 162–164) discusses several other cases where ditransitive expressions involving the same verb receive different grammaticality judgements depending on the construability of the situation according to the basic ‘give’ schema. He argues that, crosslinguistically, situations which approximate the ‘give’ schema ‘NP1 causes NP2 to have NP3’ are typically expressible as benefactives – which is to be understood as constructions in which the beneficiary role is encoded as an argument rather than as an adjunct – while situations which are harder to construe in ‘give’ terms become harder to express as benefactive expressions, and he presents English as an example of a language which is quite restrictive in this regard (i.e. in which situations need to match the ‘give’ schema closely in order to be expressible as benefactives, see below for further discussion). (5b) also shows that the intended transfer need not involve a concrete entity which enters into a relation of prototypical possession with the beneficiary. Some more examples involving non-prototypical transfers are given in (6). (6) a. Can you call me a taxi please? b. Frederick sang Gertrude a love song.
In (6a), while calling somebody a taxi is an act of obtainment very much like buying somebody a book or getting somebody a drink, the outcome desired by the speaker is in this case of course not that s/he comes to possess a taxi in the strict sense of the word, but that a taxi is made available to her/him for a particular stretch of time and for a particular purpose. Similarly, the performance event in (6b) does not lead to a situation in which Gertrude has the song at her permanent disposal: rather, she is involved in the denoted event as the metaphorical “receiver” of Frederick’s song, in the sense of the person to whom it is addressed, its intended percipient (cf. Green 1974: 97). Such metaphorical recipients are by no means exclusive of benefactive uses of the double object construction. Clauses with verbs of giving, for instance, can denote metaphorical rather than concrete transfers as well (e.g. He gave her a kiss, She gave him her opinion on the matter, etc.; cf. Goldberg 1995: 141–150), as do double object clauses with verbs such as tell, show, quote, etc. If the actual ‘caused reception’ event denoted by double object clauses with give etc. can be of a metaphorical nature, it is hardly surprising that the event of recipient-benefaction denoted by benefactive double object
Timothy Colleman
clauses can involve the intended causation of metaphorical rather than material �reception as well.
2.2â•… Language-internal variation The above intended reception constraint comes with a certain amount of inherent fuzziness, i.e. whether a given event can be construed as involving intended causation of reception—or, in Shibatani’s (1996) terms, as approximating the basic ‘give’ schema—is a matter of degree rather than kind. This explains why, in terms of Allerton (1978: 23), “sentences containing such [benefactive double object] constructions exhibit no clear boundary between acceptable and unacceptable but a whole range of degrees from fully unacceptable to fully acceptable”. Allerton’s conclusion is based on the findings from a preliminary questionnaire study in which he asked fifty British university students to judge a series of forty test sentences of the pattern Could you V me NP?. While examples such as Could you pour me a cup of coffee? and Could you paint me a picture? were accepted (virtually) unanimously and examples such as Could you taste me this wine? and Could you teach me a class? were quite consistently rejected, there were also a lot of cases which received mixed judgements. Some examples are given in (7). (7) a. % Could you iron me these shirts? b. % Could you sharpen me the knife? c. % Could you wash me the dishes?
(Allerton 1978: 25)
The examples from the WWW in (8), quoted in Fellbaum (2005: 223), illustrate that such clauses indeed occur in actual language use. (8) a. Well, the rest is his story? Honey, can you iron me a shirt?? b. You’re a good boy, Joe. Now get busy and wash me some dishes.
According to Fellbaum, such web data reveal the semantic constraints on the benefactive double object construction posited in the literature “to be soft and subject to frequent violation and extension, raising the possibility of a constraint with fuzzy edges” (2005: 209). Indeed, the notion of intended reception is stretched considerably in the examples above. In (7a), for instance, the shirts very probably already belonged to the beneficiary before the event, so there is no intended transfer of possession in the strict sense of the word. On the other hand, however, something previously unavailable to the beneficiary would be made available to them as a result of the desired act, viz. some ironed shirts, and it would seem that this suffices to satisfy the intended reception constraint for some speakers but not for others, hence the mixed judgements. The same applies, mutatis mutandis, to the other clauses in (7) and (8), in which the
Benefactive semantic potential of ‘caused reception’
product that becomes available to the beneficiary as a result of the agent’s action is a sharpened (hence ready for use) knife or a set of clean dishes. The inherent fuzziness in the intended reception constraint is not the only source of language-internal variation in English, however. First, consider the following quote from Kay (1997) on the example in (9) below, framed in the context of a critical assessment of Goldberg’s (1995) analysis of the semantic range of the English double object construction.
(9) I’ll peel you an orange. A sentence like [9] can answer for your desire for orange pulp, juice or peel, but also to your wish to see if the new orange peeler works, or to your desire to know whether my injured hand has regained its dexterity. I need not have in mind that you will receive orange products to utter [9] felicitously, only that you will benefit in some way from the act of my peeling an orange. (Kay 1997: 11–12)
This is a telling quote because it suggests that, in Kay’s idiolect, there need not be an event of intended reception involved for the double object construction to be possible at all. While the above examples with iron etc. illustrate the fuzzy edges of the intended reception constraint, the peel example on its “demonstrative” interpretation (i.e. the orange is peeled to demonstrate the working of the orange peeler or the dexterity of the peeler’s hand) presents a downright violation of this constraint: in such events, nothing is made available to the beneficiary, which hence does not qualify as a recipient-beneficiary, not even under a broad interpretation of that notion. Nor is this a clear case of substitutive benefaction, as there is no suggestion that the beneficiary would have been expected to peel the orange herself otherwise. Fawcett (1987: 149) introduces the term pleasee for this kind of beneficiary which is neither a recipient-beneficiary nor a substitutive beneficiary (cf. also the notion of plain benefaction in Van Valin & LaPolla 1997: 383–84, defined as “the action of the actor provides [the beneficiary] with amusement, enjoyment or other kind of benefit”). The examples in (10) below involve this pleasee subtype of beneficiary as well. These uses were first signalled by Green (1974), who observes that they do not fit in any of the conventional semantic classes of double object verbs and categorizes them as “expressions which can be used to denote acts intended to be symbolic of the subject’s devotion to the indirect object or something of the sort” (1974: 95). (10)
a. They’re going to kill Reagan a hippie. (Green 1974: 95) b. Sam promised to move/crush his lover a mountain. (Green 1974: 96) c. All you have to do to gain my confidence is rob me a couple of banks. (Oehrle 1976: 111) d. You say you’re a fast weeder. OK, can you weed me that garden over there in (Takami 2003: 206) two hours? If you can, you’re hired.
Timothy Colleman
Takami (2003: 211–212) labels this as the “derring-do” subtype of benefaction, in which the agent performs a brave or otherwise noteworthy action in order to make an impression on another person, who is meant to witness the act (or to see its results) and be pleased by it. Though several authors have noted that these uses are marginal in the present-day language and that their acceptability is subject to wide dialectal variation (e.g. Oehrle 1976: 111–114; Goldsmith 1980: 428; Goldberg 1995: 95–96), they still show that at least for some speakers of contemporary English, the double object construction can encode other subtypes of benefactive events than just recipientbenefaction. Goldberg (1995: 150–51) argues that the beneficiary can be construed as an intended recipient even in these “derring-do” cases, viz. through a metaphor which understands actions which are performed for the benefit of a person as objects which are transferred to that person (the beneficiary is the “receiver” of the performed action, so to speak). However, as pointed out by Takami (2003: 208–209), this metaphorical explanation is somewhat problematic, as it predicts that the double object construction should always be acceptable whenever an action is performed for somebody’s benefit, which is obviously not the case. Even for those speakers who accept the uses in (10), clauses such as * Sue fixed Bill the radiator and the other instances of substitutive benefaction listed in 4 above, for instance, are still ungrammatical.2 In sum, there are several indications that the intended reception constraint on the use of the benefactive double object construction is more strictly adhered to by some speakers of English than by others. It is perfectly possible that this language-internal variation (partly) follows regional lines – the above “derring-do” use, for instance, seems typical of colloquial varieties of American English – but, to my knowledge, this has not been systematically investigated. As it happens, the double object construction has been shown to display other “special” uses as well in certain regional varieties of English. See for instance Webelhuth and Dannenberg’s (2006) discussion of the construction in (11) below, which is typical of southern vernacular varieties of American English (cf. also Christian 1991; Wolfram & Christian 1976). (11) a. He shot him a pheasant. b. I only need to sell me a dozen more toothbrushes. c. I’m gonna drink me a couple of beers and watch me a movie.
Though there is of course a certain semantic likeness, this construction differs from the benefactive instances discussed above in that the indirect object – which is obligatorily coreferential with the subject here – does not refer to a person for whose benefit the
2.â•… In Green’s idiolect , for instance, the clause J. baked Mr. Lubin cakes “is appropriate if the cakes are intended as a gift to Mr Lubin, if the baking is intended as a performance for Mr Lubin, or if the baking is intended as a ritual act in honor of Mr Lubin; but not if J. was a substitute for Mr Lubin, or if J. was an employee of Mr Lubin” (Green 1974: 97–98).
Benefactive semantic potential of ‘caused reception’
act is carried out but rather serves to underscore or intensify the agentivity of the subject referent (cf. Webelhuth & Dannenberg 2006: 36). Still, it is another illustration that in certain varieties of English, the double object pattern covers a wider region in semantic space.
3.â•… Benefactive uses of the German dative-accusative construction The examples in (12) and (13) below illustrate the use of the German ditransitive construction with dative and accusative objects to encode events of benefaction. German grammars traditionally distinguish between Dativobjekte ‘dative objects’, which correspond to a semantic argument of the verb, and freie Dative ‘free datives’, which do not have argument status and can be freely added or deleted (see e.g. Helbig 1981 for an overview). According to this view, the dative nominals in (12) and (13) belong in the latter category, since kaufen ‘buy’, bauen ‘build’, tragen ‘carry’, öffnen ‘open’, etc. are not readily thought of as lexically selecting a recipient, beneficiary or addressee participant, in contrast to verbs such as geben ‘give’, sagen ‘say’, verkaufen ‘sell’, schenken ‘present’, etc. More precisely, these datives instantiate the so-called dativus commodi, one of the traditional subtypes of the free dative. This would imply that (12) and (13) do not instantiate the same grammatical construction as (2b) above, since they do not contain a dative object. However, following Wegener (1985: 113–19), Eisenberg (1999: 286–93) and Draye (1996: 174–76), inter alia, we depart from this traditional view here and treat the dative nominals in clauses with kaufen, bauen, tragen, etc. as genuine dative objects which introduce an additional core argument in the clause (also see Hole 2006, 2007). This approach is consonant with the construction grammar theory of argument structure constructions, which recognizes that the number of argument roles associated with a particular construction need not coincide with the number of argument roles of all verbs that can enter in the construction (cf. Goldberg 1995, 2002). (12) a.
Der Mann kauft seiner Freundin einen Ring. the.nom man buys his.datˉ girlfriendˉ an.acc ring ‘The man buys his girlfriend a ring.’
b. Bestellst du mir einen Kaffee? order you.nom me.dat a.acc coffee ‘Can you order me a coffee?’ c.
Sie haben dem Direktor ein Haus gebaut. they.nom have the.dat director a.acc house built ‘They have built the director a house.’
(13) a.
Der Mann öffnet dem Jungen die Tür. the.nom man opens the.dat boy the.acc door ‘The man opens the door for the boy.’
Timothy Colleman
b. Peter trägt der Tante den Koffer. Peter carries the.dat aunt the.acc suitcase ‘Peter carries the suitcase for his aunt.’ c.
Er mäht seinem Vater den Rasen. he.nom mows his.dat father the.acc lawn ‘He mows the lawn for his father.’
As has been observed before – by Shibatani (1996), among others – the semantic possibilities of this German benefactive construction are wider than those of the equivalent construction in English. Crucially, there is no intended recipient constraint in German: the examples in (12) and (13), which denote prototypical events of recipientbenefaction and of substitutive benefaction, respectively, are equally unproblematic. These two types correspond with the two subtypes of the traditional dativus commodi category distinguished in Zimmermann (1985): the dativus dispositivus, where the beneficiary gets something at his or her disposal as a result of the agent’s action, and the dativus benefactivus, where the agent substitutes for the beneficiary as the performer of the denoted action. According to Wegener (1985: 66–67), the only semantic condition on the use of the benefactive ditransitive in German is that the referent of the accusative object be construable as an effected or affected entity. This is illustrated by the contrast in€(14) below. (14) a.
Otto beobachtet die Kollegen für seinen Chef. Otto observes the.acc colleagues for his.acc boss ‘Otto observes the colleagues for his boss.’
b. *Otto beobachtet seinem Chef die Kollegen. Otto observes his.dat boss the.acc colleagues (Wegener 1991: 85)
As shown by the acceptability of the construction with a benefactive adjunct in the form of a prepositional phrase headed by für ‘for’ in (14a), it is perfectly imaginable that somebody observes something or somebody for the benefit of a third party. Still, as illustrated in (14b), the beneficiary of such an event cannot be encoded as a dative nominal in a ditransitive construction, because the denoted event has no tangible effect on the accusative object’s referent. Hens (1995: 158–159) discusses this semantic restriction in terms of Hopper and Thompson’s (1980) well-known transitivity scale: the benefactive ditransitive construction only accommodates highly transitive events. A further contrast between English and German is that unlike the indirect object of the English double object construction, the dative object of the German construction can encode the maleficiary of an action as well. Compare the examples in (15) and (16) below. While the accusative object referents in (15) are affected for the benefit of the dative object referent, the examples in (16) denote events in which
Benefactive semantic potential of ‘caused reception’
the action is performed to the detriment of the dative referent, at least on the most likely interpretation. (15) a. Er wäscht seinemˉ Vaterˉ das Auto. he washes his.dat father the.acc car ‘He washes the car for his father.’ b. Ich habe ihm sein Fahrrad repariert. I have him.dat his.acc bicycle fixed ‘I fixed his bicycle for him.’ (16) a.
Sie haben ihm das Auto zerstört. they.nom have him.dat the.acc car destroyed ‘They destroyed his car (on him).’
b. Sie haben ihm das Bier umgekippt. they.nom have him.dat the.acc beer over.tipped ‘They tipped over his beer (on him).’
As pointed out by several authors, the construction itself does not specify whether the dative object referent is involved as a beneficiary or as a maleficiary, nor does the verb, so that the benefactive/malefactive distinction is a matter of world knowledge and interpretation (see e.g. Shibatani 1996: 183–184; Wegener 1985: 266; Hens 1995: 152–153). In an isolated reading, the dative objects in (16) are most plausibly interpreted as maleficiaries, but both clauses allow an alternative interpretation in which the action is carried out for the benefit of the dative object referent (think of someone who willingly has his car wrecked in the context of an insurance scam, for instance). In this way, many ditransitive clauses can be interpreted either way, see (17) for an ambiguous example from Shibatani (1996: 184). (17) Man hat dem Emil den ganzen Garten umgegraben. they.nom have the.dat Emil the.acc whole.acc garden up.dug ‘They dug up the whole garden for/on Emil.’
This is different in English: as pointed out by Green (1974: 92) and Goldberg (1995: 146), among others, an example like (18) below is only acceptable on the (unlikely) interpretation that John likes burnt steak and that Mary was burning one for him in order to prepare it for his consumption. It is not acceptable in a context where Mary has malicious intentions. In other words, the English construction does specify that the subject’s act is beneficial and cannot be used to encode malefactive events. (18) Mary burned John a steak.
4.â•… Benefactive uses of the French dative construction The semantic possibilities of the French construction with a dative clitic are closer to those of the German dative-accusative construction discussed above than they are
Timothy Colleman
to those of the English double object construction. Just like German, French has a number of constructions which are traditionally labelled as free datives (datifs libres or étendus), including beneficiary datives. (19) lists a number of examples in which the dative clitic encodes the intended recipient of the entity created or selected by the subject referent, but as illustrated in (20) and (21), there need not be a recipientbeneficiary involved for the construction to be possible. (19) a.
Je leur ai cuit des gâteaux. I them.dat have baked indef.pl cookies ‘I baked them some cookies.’
b. Je lui ai trouvé un emploi. I her/him.dat have found a job ‘I found her/him a job.’ (20) a.
Je lui ai repassé sa chemise. I her/him.dat have ironed her/his shirt ‘I’ve ironed her/his shirt for her/him.’
b. Je lui ai rajeuni ses plantes. I her/him.dat have revived her/his plants ‘I revived her/his plants for her/him.’ (21) a.
On leur a démoli leurs maisons. they them.dat have demolished their houses ‘They demolished their houses (on them).’
b. On lui a tué son père. they her/him.dat have killed her/his father ‘They killed her/his father (on her/him).’
In both (20) and (21), the dative clitic does not refer to an intended recipient but to the person who has an interest in the entity which is affected by the agent’s action (usually, but not necessarily, the dative referent is the possessor of the direct object referent).3 In the most likely interpretation, the agent’s act is beneficial to the dative referent in (20) and detrimental to the dative referent in (21), but, just like in German, the benefactive/ malefactive distinction is a matter of world knowledge, so the reverse interpretations are not a priori impossible (cf. Leclère 1976). 3.â•… Barnes (1985: 161) supplies an example in which the dative object referent is not the possessor of the direct object referent. Example (i) below allows a reading in which the sweater belongs to the agent rather than to the dative object referent, if the dative clitic refers to, for instance, the kid’s mother who had knitted the sweater for him. (i)
Le gossei luij a démoli soni pull. the kid him/her.dat has destroyed his sweater ‘The kid destroyed his sweater (on her).’
Benefactive semantic potential of ‘caused reception’ 
Barnes (1985) reports that in French too, the semantic range of the benefactive construction is subject to language-internal variation: In straightforward cases (i.e. those generally accepted by all speakers), the dative object is typically the recipient of a thing which is created, selected, or altered, or the possessor of something which is destroyed or damaged … For some speakers, the benefactive dative is also possible where the action is simply performed for the person, in the sense of ‘in his or her place’. More marginal, but accepted by some speakers, are sentences … where all one can say is that the dative object is capable of being pleased or displeased by the event, because she or he desired that it take place or is somehow inconvenienced by its occurrence. (Barnes 1985: 172–173)
The examples in (22) and (23) below illustrate these less generally accepted uses. The events denoted in (22) are clear examples of substitutive benefaction: the agent carries out an action instead of the dative referent. This use is also noted by Herslund (1988: 271), who labels it the bénéficiaire remplacé, “celui pour qui l’action a lieu, dans le sens que celui-ci aurait dû, ou pu, l’exécuter”. (22) a. %Il lui a goûté le vin. he her/him.dat has tasted the wine ‘He tasted the wine for her/him.’ b. %Je veux bien lui regarder le document I want well her/him.dat look the document (pour voir s’il y a quelque chose d’important là-dedans). ‘I’m willing to look at the document for her/him (to see if there’s anything important in it).’ (Barnes 1985: 172) (23) a. %Paul lui a enfin écouté cette symphonie. Paul her/him.dat has finally listened this symphony ‘Paul finally listened to this symphony for her/him.’ b. %Il leur a râté son bac. he them.dat has flunked his baccalaureat ‘He flunked his baccalaureat exam on them.’
(Barnes 1985: 172)
The examples in (23) illustrate the limited case in which the dative object simply encodes the person who is pleased or displeased by the denoted event.
5.â•… Benefactive uses of the Dutch double object construction 5.1â•… Standard Netherlandic Dutch: A heavily constrained construction Of the constructions discussed so far, the English double object construction is clearly the most limited in semantic range when it comes to the encoding of events of benefaction. Even in those varieties of English in which the intended reception
Timothy Colleman
constraint is less strictly adhered to, the construction cannot be used to encode acts of substitutive benefaction, for instance, nor can it be used to encode malefactive events. In Shibatani’s (1996: 169–173) cursory survey of the cross-linguistic variation seen in benefactive constructions, English is presented as the most restrictive of the seven or eight languages under discussion. Indeed, many of the constructions discussed elsewhere in this volume cover a far wider array of benefactive events than the English double object construction. It can be observed, however, that in Dutch, the semantic possibilities are even more limited than in English. The accepted view is that in present-day standard (Netherlandic) Dutch, the benefactive double object construction is a marked construction which is possible with a small set of rather infrequent verbs related to food provision or preparation only, such as (een drankje) inschenken ‘to pour (a drink)’, (een bord) opscheppen ‘to dish up (a plate)’ and in formal language also (een maaltijd) bereiden ‘to prepare (a meal)’, see (24) for construed examples from the literature.4 (24) a.
De ober heeft Piet een kop koffie ingeschonken. the waiter has Pete a cup coffee in.poured ‘The waiter has poured Pete a cup of coffee.’ (Schermer-Vermeer 1991: 219)
b. Zal ik jou eens een lekker bord boerenkool opscheppen? shall I you part a nice plate borecole up.dish ‘Shall I dish you up a tasty plate of borecole?’(Schermer-Vermeer 1991: 219) c.
Ze had ons een heerlijke maaltijd bereid. she had us a delicious meal prepared ‘She had prepared us a delicious meal.’ (Haeseryn et al. 1997: 1165, labelled “formal”)
In view of this, Verhagen (2002) observes that standard Dutch does not have a productive benefactive double object construction and marks the example in (25a) below as ungrammatical. (25b–d) also illustrate instances of the double object construction which are awkward in the present-day (Netherlandic) standard language, although the indirect object referent is clearly involved as a recipient-beneficiary, which shows that the Dutch construction is even much more heavily constrained than its English counterpart (but see Section 5.2 below for discussion of the language-internal variation attested in Dutch). (25) a. *Jan maakte haar een boterham. (Verhagen 2002: 415) John made her a sandwich ‘John made her a sandwich.’
4.â•… In addition, there are a number of fixed expressions such as iemand een kool stoven (lit. to stew someone a cabbage) and iemand een poets bakken (lit. to bake someone a prank), both of which mean ‘to play a trick/joke on someone’, which derive from benefactive/malefactive uses of the double object construction historically.
Benefactive semantic potential of ‘caused reception’
b. *Mijn moeder heeft mij een mooie trui gebreid. my mother has me a nice sweater knitted ‘My mother has knitted me a nice sweater.’ c. *Zijn reclasseringsambtenaar heeft hem een nieuwe baan gevonden. his probation.officer has him a new job found ‘His probation officer has found him a new job.’ d. *?Ik heb mijn zoontje een speelgoedauto gekocht.5 I have my son a toy.car bought ‘I have bought my son a toy car.’
Colleman (submitted) introduces a “contiguity” constraint to account for this restricted distribution. Prototypical events of recipient-benefaction can be decomposed into two sub-events: (i) the creation or selection of an entity by the agent and (ii) the subsequent transfer of this entity to the recipient-beneficiary. This is adequately captured by the following semantic paraphrase of the benefactive double object construction, adapted from Geeraerts (1998: 196): ‘An Agent carries out a preparatory action (involving a Patient), with a view to a subsequent transfer of this Patient to a Recipient’. It seems that in present-day standard (Netherlandic) Dutch, the two sub-events almost need to coincide for the double object construction to be possible, i.e. there must be a maximum degree of contiguity between the preparatory action and the actual transfer. This largely restricts the possibilities to cases such as iemand een drankje inschenken ‘to pour someone a drink’ and iemand een bord groenten opscheppen ‘to dish someone up a plate of vegetables’, in which the preparatory action and the transfer are virtually indistinguishable, so that the overall events could just as well be construed as single acts of giving. Interestingly, verbs of food provision such as inschenken ‘pour’ and opscheppen ‘dish up, ladle out’ can also be used in a number of other constructions which are normally reserved for verbs of giving. For instance, the attested instances in (26) below illustrate their use in the so-called receptive or indirect-passive construction with krijgen ‘get’ (cf. the structurally and semantically similar kriegen/bekommenpassive in German). (26) a.
Wie melk uit een gekleurde fles kreeg ingeschonken, zou who milk from a coloured bottle got in.poured would
raar opkijken. (Reformatorisch Dagblad, 15/01/2000) strange up.look
‘Someone who was poured milk from a coloured bottle (lit. who got poured milk), would be much surprised.’
5.â•… Colleman’s (in press) preliminary corpus investigation suggests that with kopen ‘buy’, the benefactive double object construction is not altogether impossible for some speakers of standard Netherlandic Dutch, but it is definitely a very marginal use.
Timothy Colleman
b. Niet iedereen zal erop staan dat hij zijn zabaglione uit not everybody will on.it stand that he his zabaglione from
een peperdure massief koperen zabaglionepan krijgt opgeschept. a high.price solid copper zabaglione.pan gets up.dished
‘Not everybody will insist to be served his zabaglione (lit. to get dished up his zabaglione) from a high-price solid copper zabaglione pan.’ (De Volkskrant, 18/12/2001)
This is a construction which is typically found with verbs of giving and certain verbs of communication only, and which certainly does not accommodate more central members of the classes of creation/preparation and obtainment verbs such as maken ‘make’, bakken ‘bake’, kopen ‘buy’, vinden ‘find’, etc.6 As such, the occurrence of clauses such as (26) corroborates that the events of food provision denoted by three-argument clauses with inschenken ‘pour’ and opscheppen ‘dish up, ladle out’ can be construed as relevantly similar to single acts of giving rather than to more prototypical events of recipient-benefaction such as buying a book for someone or baking a cake for someone. This semantic similarity to more central classes of double object verbs explains why in the most restrictive varieties of Dutch, including the present-day standard variety, these verbs of food provision are the only verbs which can be used in the benefactive double object construction. In terms of Shibatani’s (1996) ‘give’-based account, present-day standard Dutch is an even more limited case than English, i.e. in order to be encodable as a double object expression, benefactive events need to present an even closer approximation of the basic ‘give’ event than is the case in English.
5.2â•… Diachronic and regional variation The discussion in the previous paragraph has already hinted at the existence of extensive language-internal variation in the semantic range of the benefactive double object construction in Dutch, too (cf. our discussion of language-internal variation in English in Section 2.2). Indeed, while the benefactive construction is heavily
6.â•… That is, even in those varieties of Dutch in which verbs such as kopen ‘buy’, maken ‘make’, etc. can be used in the benefactive double object construction (see Section 5.2), these verbs are still excluded from the indirect-passive construction. The construed examples in (i) and (ii) are unacceptable in all varieties of the language. (i) *De jongen kreeg een boek gekocht van zijn vader. The boy got a book bought of his father ‘The boy was bought a book by his father.’ (ii) *De jongen kreeg een taart gebakken van zijn vader. the boy got a cake baked of his father ‘The boy was baked a cake by his father.’
Benefactive semantic potential of ‘caused reception’
constrained in the present-day standard language, its semantic possibilities are much wider in a number of other varieties of the language. First, it can be observed that in earlier phases of the language, the benefactive ditransitive was not at all limited to a handful of verbs of food provision or preparation of the kind illustrated in (24) above, but could be combined productively with all kinds of verbs to encode various subtypes of benefactive events. (27) below lists a small number of attested examples from various periods; many more such examples can be found via the citation sections of the Middle Dutch Dictionary (Verwijs & Verdam 1885–1952) and the Dictionary of the Dutch language (De Vries, Te Winkel et al. 1882–1998). Examples (27a) and (27d) denote events of recipient-benefaction, (27b) involves the pleasee subtype of beneficiary and (27c) involves a substitutive beneficiary.7 (27) a.
Daer cochte Joseph sinen here den vijften scoof there bought Joseph his.dat lord the.acc fifth part
van der vrucht. of the harvest
‘There Joseph bought his lord the fifth part of the harvest.’
(c. 1300, Maerlant, Rhimed Bible)
b. Wat ic huer doe k=in can=se niet ghepaeyen. what I her do I=neg can=her neg content ‘Whatever I do for her, I cannot content her.’ (16th C, anonymous ballade) c. …dat hy hem de deur opende, en in een kamer voerde. ╇ â•›that he him the door opened and in a room led ‘… that he opened him the door, and led him into a room.’ (N. Heinsius, De vermakelyken avanturier, 1695) d. [Ik] zal de deur aanwijzen, waar-voor zij mij de I shall the door to.point where-for she me the
sleutel maken moet. key make must
‘I shall point to the door for which she has to make me a key.’ (J.F. Oltmans, De Schaapherder, 1838)
Some of these possibilities have been preserved in regional varieties of the present-day language. According to Haeseryn et al. (1997: 1165), construed examples such as (28a) and (28b) below are typical of Southern Dutch, i.e. of language varieties spoken in Belgium and in the southern provinces of The Netherlands (examples labelled as “regionally marked” in Haeseryn et al. 1997: 1165).
7.â•… Note that, strictly speaking, Middle Dutch examples such as (27a) do not involve the double object construction (in its narrow sense of a construction with two unmarked NP objects) but rather its predecessor in Dutch grammar, viz. the dative + accusative construction.
Timothy Colleman
(28) a.
De hoogleraar kocht zijn vrouw een gouden armband. the professor bought his wife a gold bracelet ‘The professor bought his wife a gold bracelet.’
b. Mijn vrouw heeft me een trui gebreid. my wife has me a sweater knitted ‘My wife has knitted me a sweater.’
There have been a series of questionnaire and/or survey-based studies which illustrate that such uses indeed occur in local dialect and regional substandard varieties of southern, and as it happens also eastern, Dutch (cf. Van Bree 1981; Cornips 1994 and Colleman & De Vogelaer 2003). In addition, Colleman (in press) presents the results of a preliminary corpus investigation into the distribution of the benefactive double object construction in the two national varieties of the language, i.e. standard Netherlandic Dutch and supra-regional Belgian Dutch.8 These results corroborate that the construction is still used fairly productively in Belgian Dutch: in addition to the small set of verbs of the inschenken ‘pour’ type discussed above, there is an open class of creation/preparation and obtainment verbs which are eligible for use in the double object construction in Belgian Dutch as well. (29) lists a number of attested examples with obtainment verbs such as kopen ‘buy’ and halen ‘get, fetch’ and creation/preparation verbs such as bakken ‘bake’, build ‘bouwen’ and smeden ‘forge’. (29) a.
Volgens O. kocht de la Brassine hem ooit voor according.to O. bought de la Brassine him once for
100.000 frank nieuwe kleren in een “boetiek” aan 100.000 francs new clothes in a boutique on
de Anspachlaan. the Anspach.Avenue
‘According to O.i, de la Brassinej once bought himi 100,000 francs worth of new clothes in a “boutique” in the Anspach Avenue.’ (De Standaard, from the CONDIV-corpus of written Dutch) b. Schat wil=‘k ik u nog.een jenever-ke halen? darling shall=I.clit I you another geneva-dim get ‘Darling, shall I get you another geneva?’ (from the Belgian component of the Corpus of Spoken Dutch)
8.â•… Since Belgian Dutch can hardly be said to possess a standard variety of equal status with standard Netherlandic Dutch, the more neutral term supra-regional Belgian Dutch is used here instead, which should be understood as referring to, for instance, the language of Belgian newspapers.
Benefactive semantic potential of ‘caused reception’
c. De mensen uit het dorp bakken ons een pizza in de houtoven. the people of the village bake us a pizza in the wood.oven ‘The villagers are baking us a pizza in the wood oven.’ (from the 2003 brochure of the Belgian tour operator Anders Reizen) d. Bouw me een pretpark, papa! build me an amusement.park daddy ‘Build me an amusement.park, daddy!’ (headline of an article about spoiled Hollywood kids in Het Nieuwsblad/De Gentenaar, 31/10/2005) e.
Zijn er nog vaklui te vinden die je een zwaard smeden are there still craftsmen to find that you a sword forge
van echte topkwaliteit? (De Morgen, 20/08/2003) of real top.quality
‘Are there still craftsmen to be found who can forge you a top quality sword?’
This illustrates that the contiguity constraint discussed in the previous paragraph is less of an issue in the Belgian variety of the language. Although further research into the exact distribution is required, it would seem that the semantic possibilities of the benefactive double object construction in supraregional Belgian Dutch are roughly equivalent to the generally accepted possibilities in English: it is enough for the beneficiary to be construable as the intended recipient of the entity affected or effected by the agent for the double object construction to be possible. Indeed, the clauses in (29) all denote fairly typical events of recipient-benefaction.9 In some local eastern varieties, the possibilities are even wider than in supra-regional Belgian Dutch. Cornips (1994) discusses the much freer use of various indirect object constructions, including the benefactive double object construction, as one of the major grammatical characteristics which distinguish the variety of Dutch spoken in Heerlen – a town in the very south east of the Dutch province of Limburg, close to the Dutch-German border – from the present-day (Netherlandic) standard language. She quotes several
9.â•… It should be added, however, that although the semantic range of the construction in Belgian Dutch may be similar to that of the present-day English construction, the Belgian construction is not nearly as frequent as its English counterpart in actual language use. In Stefanowitsch & Gries’s (2003) corpus investigation, at least two out of the thirty English verbs that are revealed to be most strongly attracted to the double object construction according to their “collexeme analysis” method, instantiate the benefactive double object construction, viz. buy and cook (and perhaps also earn). This shows that, in English, the benefactive use of the double object construction is well-established frequencywise. In Belgian Dutch, by contrast, the construction is quite infrequent. In a corpus totalling some 20 million words of Belgian text, for instance, Colleman (in press) found only one double object example of bakken ‘bake’, and only three double object examples of bouwen ‘build’ (out of total frequencies of 554 instances for bakken and 2339 instances for bouwen).
Timothy Colleman
observed instances which are similar to the Belgian examples in (29) above in denoting events of recipient-benefaction, but in addition, her material contains several observed instances where the construction is used to encode events of benefaction which do not involve a recipient-beneficiary. Some relevant examples are listed in (30). (30) a.
Nou die [man] heeft me dat instrument … onder handen well that man has me that instrument under hands
genomen, daar kon je niets meer taken there could you nothing anymore
aan zien. on see
(Cornips 1994: 183)
‘Well, that man put his hands to that instrument for me and you couldn’t see anything wrong with it anymore (i.e. he took on the task of repairing the instrument for me).’ b. Doe me gauw de verwarming aan want ik verrek do me quickly the heating on because I perish
van de kou. of the cold
‘Quickly, turn the heating on for me, because I’m perishing with cold.’
c.
Dat heeft die man me allemaal mooi in het Duits vertaald. that has that man me all nicely in the German translated ‘That man has translated it all nicely into German for me.’ (Cornips 1994: 174)
(Cornips 1994: 184)
d. Als ik iemand een rapport heb opgemaakt van feiten when I someone a report have made.up of facts
die [ik] geconstateerd heb. that I ascertained have
(Cornips 1994: 174)
‘When I have made a report for/on someone, about facts that I have ascertained.’
The first example in (30) is similar to some of the German and French examples quoted above in that the beneficiary is not a recipient but the possessor of an entity which is acted upon by the agent for his benefit; in this particular case, the agent repairs a musical instrument for the speaker (compare with the German examples with wäschen ‘wash’ and reparieren ‘fix’ in 15 and the French examples with repasser ‘iron’ and rajeunir ‘revive’ in 20 above). In the second and third example, the agent turns the heating on for the sake of the beneficiary and acts as his translator, respectively, so, again, there is no intended causation of reception involved. Both clauses allow a substitutive benefaction interpretation, though other readings are possible as well. Finally, (30d) shows that just like the relevant German and French constructions, the Heerlen construction can also encode events involving a maleficiary: this particular example is ambiguous between a beneficial (i.e. the indirect object refers to the person for whom the report is made,
Benefactive semantic potential of ‘caused reception’
the agent’s boss, for instance) and a maleficial reading (i.e. the indirect object refers to the person on whom the agent files a report). We refer to Cornips (1994: 173–204) for more examples and further discussion; also see Van Bree (1981: 308–310) for a number of similar instances observed in local dialects from the Twente region, which is also to be situated in the very east of The Netherlands. In sum, the different regional varieties of Dutch form a cline with regard to the semantic possibilities of the benefactive double object construction. In present-day standard Netherlandic Dutch, the construction is heavily constrained, to the effect that it can only be used with a handful of verbs of food provision of the inschenken ‘pour’type. In Belgian Dutch, however, which is generally considered the more conservative variety, the construction can still be productively combined with all kinds of verbs of creation/preparation and obtainment to encode events of recipient-benefaction, though it should be added that such uses are quite infrequent in actual language use. Finally, there are a number of local language varieties from the eastern periphery of The Netherlands in which the double object construction can be used to encode a wider array of benefactive/malefactive events than just recipient-benefaction, just like in neighbouring German.10
6.â•… Conclusion Goldberg (2006: 199) observes that many languages have a ditransitive construction with a core ‘transfer’ meaning similar to the English double object construction, but that, compared to English, “the formal pattern typically has a somewhat wider or narrower range of meaning associated with it” (also see Malchukov et al. 2007: 51ff). It is evident from the above discussion that one of the areas of cross-linguistic differentiation is the extent to which these ‘transfer’ constructions can also be used to encode events which involve a beneficiary rather than a prototypical recipient as the third participant (next to the agent and patient/theme participants). Even within our small sample of constructions from four closely related Indo-European languages, we have seen a lot of cross-linguistic variation in the extent to which the investigated ‘caused reception’ constructions extend into the benefactive/malefactive domain. The German construction with dative and accusative objects and the French construction with a direct object and a dative clitic both cover a wide semantic range of benefactive/malefactive events.
10.â•… In addition to this diachronic and regional variation, there is also register variation. Even in Netherlandic Dutch, the benefactive double object construction is still used with some degree of productivity in certain text genres with a tendency for formal, archaic language (cf. Colleman in press for further details; also see the occurrence of ditransitive iemand een maaltijd bereiden ‘to prepare someone a meal’ in formal style, as illustrated in 24c above).
Timothy Colleman
In comparison, the English and Dutch double object constructions are more restrictive: both are subject to an intended reception constraint which limits their semantic range to events which, in terms of Kittilä (2005) involve a recipient-beneficiary. In addition, the Dutch construction is also subject to a contiguity constraint, which narrows down the possibilities to certain benefactive events of food provision, at least in present-day standard (Netherlandic) Dutch. It is tempting to relate the observed contrasts between the four languages under discussion to the presence or absence of overt dative case marking on the recipient/beneficiary, though it goes without saying that we need far more data from a large set of unrelated languages to state this with any degree of certainty.11 In addition, it has been shown that several of the languages under discussion display interesting language-internal variation in this domain. Especially in Dutch, there is wide regional variation in the semantic range of the benefactive double object construction, with the present-day Netherlandic standard variety being situated at the most restrictive end of the scale, a number of local dialects from the eastern periphery of The Netherlands being situated at the most permissive end, and supra-regional Belgian Dutch occupying a middle position. In English and French too, there are indications that there is a cline from fully acceptable to fully unacceptable benefactive uses, with different speakers drawing the line at different cut-off points (and though we have not discussed this in so many words, there is no reason to assume that this should be different in German). This latter kind of language-internal variation can be associated with the nature of the semantic constraints on the respective benefactive constructions: whether or not a benefactive situation is construable as involving ‘intended reception’, for instance, is a matter of degree rather than kind. As such, our results are consistent with Shibatani’s (1996) ‘give’-based account of benefactives: the acceptability of benefactive constructions varies with their construability in terms of the ‘give’ schema, both crosslinguistically and intra-linguistically.
References Allerton, David J. 1978. Generating indirect objects in English. Journal of Linguistics 14: 21–33. Barnes, Betsy. 1985. A functional explanation of French nonlexical datives. Studies in Language 9: 159–195.
11.â•… In this perspective, it is interesting to note that both in English and in Dutch, the semantic possibilities of the benefactive construction were much wider in earlier varieties of the language which did mark the beneficiary with dative case. See the examples from earlier phases of Dutch in (27) above and, for English, see e.g. Visser’s (1963: 626–632) discussion of the so-called indirect object of advantage in Old English and subsequent periods.
Benefactive semantic potential of ‘caused reception’ 
Bisang, Walter. 1992. Das Verb im Chinesischen, Hmong, Vietnamesischen, Thai und Khmer. Tübingen: Narr. Christian, Donna. 1991. The personal dative in Appalachian speech. In Dialects of English: Studies in Grammatical Variation, Peter Trudgill & Jack K. Chambers (eds.), 11–19. London: Longman. Colleman, Timothy. 2009. Verb disposition in argument structure alternations. A corpus study of the Dutch dative alternation. Language Sciences. 31: 593–611. Colleman, Timothy. In press. Lectal variation and constructional semantics: Benefactive ditransitives in Dutch. In Recent Advances in Cognitive Sociolinguistics, Dirk Geeraerts, Gitte Kristiansen & Yves Peirsman (eds.). Berlin: Mouton de Gruyter. Colleman, Timothy & De Vogelaer, G. 2003. De benefactiefconstructie in de zuidelijk-Nederlandse dialecten. Taal en Tongval 15–16: 184–208, theme issue. Cornips, Leonie. 1994. Syntactische variatie in het Algemeen Nederlands van Heerlen. Amsterdam: IFOTT. De Vries, Matthias, Te Winkel, Lammert Allard. et al. (eds.) 1882–1998. Woordenboek der Nederlandsche Taal.’s Gravenhage: Martinus Nijhoff. Draye, Luk. 1996. The German dative. In The Dative, Vol. 1: Descriptive Studies [Case and Grammatical Relations across Languages 1], William Van Belle & Willy Van Langendonck (eds.), 155–215. Amsterdam: John Benjamins. Eisenberg, Peter. 1999. Grundriß der deutschen Grammatik, Bd. 2: Der Satz. Stuttgart: Metzler. Fawcett, Robin P. 1987. The semantics of clause and verb for relational processes in English. In New Developments in Systemic Linguistics, Vol. 1: Theory and description, Michael A.K. Halliday & Robin P. Fawcett (eds.), 130–183. London: Francis Pinter. Fellbaum, Christiane. 2005. Examining the constraints on the benefactive alternation by using the World Wide Web as a corpus. In Evidence in Linguistics: Empirical, Theoretical, and Computational Perspectives, Stephan Kepser & Marga Reis (eds.), 209–240. Berlin: Mouton de Gruyter. Geeraerts, Dirk. 1998. The semantic structure of the indirect object in Dutch. In The Dative, Vol. II: Theoretical and Contrastive Studies [Case and Grammatical Relations across Languages 2], Willy Van Langendonck & William Van Belle (eds.), 185–210. Amsterdam: John Benjamins. Goldberg, Adele E. 1995. Constructions: A Construction Grammar Approach to Argument Structure. Chicago IL: University of Chicago Press. Goldberg, Adele E. 2002. Surface generalizations: An alternative to alternations. Cognitive Linguistics 13: 327–356. Goldberg, Adele E. 2006. Constructions at Work. Oxford: OUP. Goldsmith, John. 1980. Meaning and mechanism in grammar. In Harvard Studies in Syntax and Semantics 3, Susumu Kuno (ed.), 423–448. Cambridge MA: Harvard University Department of Linguistics. Green, Georgia M. 1974. Semantics and Syntactic Regularity. Bloomington IN: Indiana University Press. Haeseryn, W., Romijn, K., Geerts, G., De Rooij, J. and Van den Toorn, M.C. (eds.). 1997. Algemene Nederlandse Spraakkunst. 2d edn. Groningen: Martinus Nijhoff & Deurne: Wolters Plantyn. Haspelmath, Martin. 2003. The geometry of grammatical meaning: Semantic maps and crosslinguistic comparison. In The New Psychology of Language, Vol. 2, Michael Tomasello, (ed.). 211–242. Mahwah NJ: Lawrence Erlbaum. Helbig, Gerhard. 1981. Die freien Dative im Deutschen. Deutsch als Fremdsprache 18: 372–74. Hens, Gregor. 1995. Ditransitive constructions in German. Ph.D. dissertation, University of California at Berkeley.
Timothy Colleman Herslund, Michael. 1988. Le datif en français. Louvain: Peeters. Hole, Daniel. 2006. Extra argumentality – affectees, landmarks, and voice. Linguistics 44: 383–424. Hole, Daniel. 2007. (Unifying ‘Possessor’ and) ‘Beneficiary’ datives as binders. Paper read at the Workshop on the Typology of Benefactives and Malefactives, University of Zurich, October 25–26, 2007. Hopper, Paul & Thompson, Sandra A. 1980. Transitivity in grammar and discourse. Language 56(2): 251–299. Jackendoff, Ray. 1990. Semantic Structures. Cambridge MA: The MIT Press. Kittilä, Seppo. 2005. Recipient-prominence vs. beneficiary-prominence. Linguistic Typology 9:€269–97. Langacker, Ronald W. 1991. The Foundations of Cognitive Grammar, Vol. 2: Descriptive Application. Stanford CA: Stanford University Press. Leclère, Christian. 1976. Datifs syntaxiques et datif éthique. In Méthodes en grammaire française, Jean-Claude Chevalier & Maurice Gross (eds.), 73–95. Paris: Klincksieck. Levin, Beth. 1993. English Verb Classes and Alternations: A Preliminary Investigation. Chicago IL: University of Chicago Press. Malchukov, Andrej, Haspelmath, Martin & Comrie, Bernard. 2007. Ditransitive constructions: A typological overview. Ms. Margetts, Anna & Austin, Peter. 2007. Three-participant events in the languages of the world: Towards a crosslinguistic typology. Linguistics 45: 393–451. Newman, John. 1996. Give. A Cognitive Linguistic Study. Berlin: Mouton de Gruyter. Nisbet, Tim. 2005. Benefactives in English: Evidence against argumenthood. Reading Working Papers in Linguistics 8: 51–67. Pinker, Steven. 1989. Learnability and Cognition: The Acquisition of Argument Structure. Cambridge MA: The MIT Press. Oehrle, R. 1976. The Grammatical Status of the Dative Alternation. Ph.D. dissertation, MIT. Schermer-Vermeer, E. Cornelia. 1991. Substantiële versus Formele Taalbeschrijving: Het Indirect Object in het Nederlands. Amsterdam: Universiteit van Amsterdam. Shibatani, Masayoshi. 1996. Applicatives and benefactives: A cognitive account. In Grammatical Constructions: Their Form and Meaning, Masayoshi Shibatani & Sandra A. Thompson (eds.), 157–194. Oxford: OUP. Stefanowitsch, Anatol & Gries, Stefan. 2003. Collostructions: Investigating the interaction of words and constructions. International Journal of Corpus Linguistics 8: 209–243. Takami, Ken-Ichi. 2003. A semantic constraint on the benefactive double object construction. English Linguistics 20: 197–224. Van Bree, Cornelis. 1981. Hebben-constructies en Datiefconstructies binnen het Nederlandse Taalgebied. Leiden: University of Leiden. Van Valin, Robert D., Jr. & LaPolla, Randy J. 1997. Syntax: Structure, Meaning, and Function. Cambridge: CUP. Verhagen, Arie. 2002. From parts to wholes and back again. Cognitive Linguistics 13: 403–439. Verwijs, Eelco & Verdam, Jakob. 1885–1952. Middelnederlandsch Woordenboek. ‘s Gravenhage: Martinus Nijhoff. Visser, Fredericus Theodorus. 1963. An Historical Syntax of the English Language, Part I: Syntactical Units with one Verb. Leiden: E.J. Brill. Webelhuth, Gert & Dannenberg, Clare. 2006. Southern American English personal datives. The theoretical significance of dialectal variation. American Speech 81: 31–55. Wechsler, Stephen. 1995. The Semantic Basis of Argument Structure. Stanford CA: CSLI.
Benefactive semantic potential of ‘caused reception’
Wegener, Heide. 1985. Der Dativ im heutigen Deutsch. Tübingen: Narr. Wegener, Heide. 1991. Der Dativ – ein struktureller Kasus? In Strukturen und Merkmale syntaktischer Kategorien, Gisbert Fanselow & Sascha W. Felix (eds.), 70–103. Tübingen: Narr. Wierzbicka, Anna. 1988. The Semantics of Grammar [Studies in Language Companion Series€18]. Amsterdam: John Benjamins. Wolfram, Walt & Christian, Donna. 1976. Appalachian Speech. Arlington VA: Center for Applied Linguistics. Zimmermann, Annette. 1985. Zur Syntax und Semantik des dativus commodi bei transitiven Verben. Deutsch als Fremdsprache 22 (1): 29–40.
Beneficiary coding in Finnish Seppo Kittilä
University of Helsinki This paper examines the beneficiary coding in (Standard) Finnish. Finnish has an elaborate system of coding benefaction. Finnish uses two cases (allative and ablative) for this, in addition to which benefaction can be coded by an array of postpositions. As will be shown in the paper, the markers are not in free variation, but formal variation has a semantic basis. For example, the three types of beneficiaries (plain beneficiary, deputative beneficiary and recipient-benefactive) distinguished by Van Valin & LaPolla (1997) are relevant to beneficiary coding in Finnish. In addition, I will discuss beneficiary coding in light of other features, such as animacy, volition and transitivity of the denoted event.
1.â•… Introduction This paper examines the beneficiary1 coding in Standard Finnish; seven grammaticalized/ productive ways of coding benefaction will be illustrated in Section 2. Section 3 will then provide a semantic analysis of these benefactive constructions. The Finnish benefactive will first be studied in light of some earlier typologies of beneficiaries, the primary source for which is represented by the typology proposed by Van Valin & LaPolla (1997). In Section 4, I will examine some other features relevant to the notion of benefaction in Finnish. Section 5 summarizes the central findings of the paper. The paper aims at giving a comprehensive overview of benefactives in Finnish by examining the notion from various perspectives. The notion of beneficiary/benefactive used in this paper is in accordance with the definition outlined in the introduction to this volume.
2.â•… The formal mechanisms of beneficiary coding in Finnish In this section, I will illustrate the beneficiary coding in Finnish from a primarily formal perspective, even though the presentation will also touch upon the semantics of
1.â•… In this paper, the label beneficiary is used for referring to a semantic role, whereas the label benefactive is used whenever the reference is to the formal coding of beneficiaries.
 Seppo Kittilä
benefactives, since this is essential to showing that the seven benefactive constructions illustrated below are semantically distinct. A closer semantic discussion follows in the subsequent sections. As noted above, Finnish may code benefaction in seven ways. First, two case forms, allative and ablative are used, as shown in (1a) and (1b): (1) a. henkilö raken-si talo-n yksilö-lle person.nom build-3sg.pst house-acc individual-all ‘A person built a house for an individual.’ b. vanhempi pes-i lapse-lta käde-t parent.nom wash-3sg.pst child-abl hand-pl ‘A parent washed the child’s hands.’ (lit. from the child the hands)
The allative codes benefaction productively, most typically reception-benefaction (see Kittilä 2005, 2006 and also below). This polysemy, along with the frequent occurrence of the allative in benefactive constructions, is understandable because the allative also codes recipients. On the other hand, the ablative occurs more rarely as a marker of benefaction, but in (1b) it is the only way of coding the role of beneficiary. The use of the ablative as a benefactive marker is possible only in a limited number of cases. The ablative codes beneficiaries basically in two cases: with grooming events targeted at body parts and in cases in which the denoted event involves substitution of an entity with another one (see (5d) for an example). In addition, Finnish employs five (productive) postpositions for beneficiary coding (with semantic differences, as will be shown below). Consider: (1) c.
henkilö raken-si talo-n yksilö-n puolesta person.nom build-3sg.pst house-acc individual-gen instead.of ‘A person built a house instead of an individual.’
d. henkilö raken-si talo-n yksilö-n vuoksi person.nom build-3sg.pst house-acc individual-gen for ‘A person built a house for an individual.’ e.
henkilö raken-si talo-n yksilö-n takia person.nom build-3sg.pst house-acc individual-gen because.of ‘A person built a house because of an individual.’
f.
henkilö raken-si talo-n yksilö-ä varten person.nom build-3sg.pst house-acc individual-part for ‘A person built a house for an individual.’
g. henkilö raken-si talo-n yksilö-n ilo-ksi person.nom build-3sg.pst house-acc individual-gen joy-transl ‘A person built a house for an individual.’
Beneficiary coding in Finnish 
The postposition puolesta (as in (1c)) simply states that an action was performed by someone instead of the beneficiary (see below for a more detailed discussion). Vuoksi (1d) and takia (1e) are close to puolesta in expressing deputative benefaction (see van Valin & LaPolla 1997: 383), but, as will become clear below, there are other differences between the three postpositions; in principle, puolesta expresses deputative beneficiaries only, whereas takia and vuoksi are associated with other types of benefaction as well, in addition to which their use is more restricted in other ways. The use of the postposition varten implies that something was done for the beneficiary for some specific purpose, and varten (1f) usually also involves some kind of reception (even though the latter may be conceptual). Iloksi expresses plain benefaction (Van Valin & LaPolla 1997), i.e. the fact that something was done or happened for the purpose of entertaining the beneficiary.2 A formal feature that may interest some readers interested in Finnish grammar in general is that all postpositions in (1c–g) except varten govern genitive case. This variation is, however, not discussed in any way in this paper, since it has no semantic basis from a synchronic perspective, and this discussion would therefore not add anything to our understanding of beneficiary coding. As shown, both case forms and adpositions may be used for beneficiary coding, which is only expected in light of the fact that cases and adpositions express similar functions also in many other cases (e.g. the coding of spatial relations). However, we should note that the two mechanisms also have clear differences. In the case of casemarked benefactives, the beneficiary semantics of the construction is a combination of verbal semantics and case marking of the beneficiary. In other words, allative and ablative code benefaction only in favorable conditions, and we cannot say that coding benefaction would be their primary, let alone only, function. It is therefore best to speak of constructional benefactives in these cases: coding benefaction is a joint effort of all clausal constituents. On the other hand, the postpositions illustrated in (1c–g) are more directly related to coding benefaction, and they alone suffice for rendering a given sentence a benefactive construction. The verbal semantics and the semantics of other clausal constituents are less relevant to this. Moreover, the semantic connection to benefaction is more evident with postpositions, for example, with ilokseni the association with plain benefaction is obvious due to the lexical meaning of the stem ilo ‘joy’. These differences are relevant and merit a study of their own, but this discussion lies outside the scope of this paper.
2.â•… Finnish has also other ways of expressing this kind of plain benefaction, such as riemuksi ‘to the great joy’, but these are semantically very close to iloksi and are therefore not discussed separately in this paper.
 Seppo Kittilä
3.â•…Plain beneficiary, deputative beneficiary and recipient-beneficiary in Finnish 3.1â•… Preliminaries Van Valin & LaPolla (1997: 383) distinguish three types of beneficiaries3 that they label as plain beneficiary, deputative beneficiary and benefactive-recipient.4 The plain beneficiary comprises cases in which “the action of the actor provides them (plain beneficiaries, SK) with amusement, enjoyment or other kind of benefit” (Van Valin & LaPolla: ibidem). Examples thus include cases such as the parent is reading to the child or the phonetician is dancing for us. Deputative beneficiary refers to cases in which someone is replacing the beneficiary as the actor of an event, as in the father opened the window instead of his son. Recipient beneficiaries are beneficiaries (Song (this volume) uses the label ‘engager-beneficiary’ for this type of beneficiaries) that can also be seen as recipients. Examples include cases such as the parent baked the child a cake, in which the cake is baked for the child, who is also the recipient of the event. In what follows, I will briefly discuss the coding of these beneficiary roles in Finnish in the order they have been introduced above.
3.2â•… Plain beneficiary Three of the benefactive markers illustrated in Section 2 can be used for coding plain beneficiaries.5 Examples are given in (2): (2) a.
presidentti laula-a kansalaiste-n ilo-ksi president.nom sing-3sg.pres citizens-gen joy-transl ‘The president is singing for (lit. to the joy of) the citizens.’
b. presidentti laula-a kansalais-i-lle president.nom sing-3sg.pres citizen-pl-all ‘The president is singing to the citizens.’
3.â•… It may be disputed whether the three types of beneficiaries suggested by Van Valin & LaPolla should be taken as distinct semantic roles, or rather as different nuances/senses of one and the same role/the same benefactive meaning. It is not the goal of this paper to offer a solution to this problem. 4.â•… Van Valin & LaPolla use the label benefactive-recipient for the last type of benefactives, but in this paper the label recipient-beneficiary is favored. 5.â•… A question that possibly arises in cases where a single type of benefaction can be coded in multiple ways, is which of the distinct ways of coding can be regarded as primary. This is an interesting question, but this would (to my mind) require a detailed corpus study, whence an attempt to answer this question is not made anywhere in this paper. Instead, the focus lies on the semantic differences between the mechanisms.
c.
Beneficiary coding in Finnish 
presidentti laula-a kansalaiste-n vuoksi president.nom sing-3sg.pres citizens-gen for ‘The president is singing for the citizens.’
The first two constructions are productive, whereas vuoksi ‘for’ codes plain beneficiaries in a lower number of cases. Examples (2a) and (2b) express rather similar meanings. In both, the president is singing, which entertains the citizens. There are, however, also differences between (2a) and (2b). Clause (2a) can be used if the president is singing among other people, for example at a concert. The singing is spontaneous, but it brings joy to the citizens. The fact that the president sings among common people and thus sees him/herself as equal to other people can also be the reason for the joy in (2a). In (2b), the singing is more directed and planned, which means that the citizens have gathered somewhere with the specific purpose of enjoying the president’s singing. The president is also more likely to sing on a stage or at a location specifically dedicated for this purpose. (2a) can also have this meaning, while (2b) cannot be used to express nondirected singing. (2a) and (2b) also differ from each other in that with iloksi the agent may be unaware of the fact that his/her actions bring joy to someone else, whereas the agent needs to be aware of the beneficial consequences of his/her actions if the allative is used. Examples (2a) and (2b) clearly express plain beneficiaries only, and other types of benefaction are not relevant. In (2c), in turn, another feature is in the foreground. Vuoksi usually adds the nuance that the agent is doing something special for the beneficiary, which is also the typical reading of (2c). (2c) can, however, have a plain beneficiary reading in favorable contexts, but this meaning is not the primary one. (2c) would then mean that the citizens wanted to see the president sing, and s/he finally agrees. (2c) may also mean that the president is collecting funds for his/her people by singing in public, in which case the notion of recipient-benefaction is also present. As noted above, the allative and iloksi are both productive ways of coding plain beneficiaries. They are, however, not in free variation, and also vary in their productivity, cf. (3) a.
vanhempi piir-si hevose-n lapse-n ilo-ksi parent.nom draw-3sg.pst horse-acc child-gen joy-transl ‘A parent drew a horse to the child’s joy.’
b. vanhempi piir-si hevose-n lapse-lle parent.nom draw-3sg.pst horse-acc child-all ‘A parent drew a horse for the child.’ c.
lapsi nukkuu vanhempi-en-sa iloksi child.nom sleep.3sg.pres parent.pl-gen-3psr joy-transl ‘The child is sleeping, which makes the parents happy.’
d. ??lapsi nukkuu vanhemmi-lle-en child.nom sleep.3sg.pres parent.pl-all-3psr ??‘The child is sleeping for/to his/her parents.’
 Seppo Kittilä
(3a) illustrates the paradigm case of plain beneficiary; a parent draws a horse for the purpose of amusing/entertaining the child. (3b) can in principle denote the same event, but the use of the allative adds the nuance that the picture produced is transferred to the child’s domain of possession, and/or that the child will be able to use the entity transferred to him/her for some specific purpose (for example, drawing a picture of a horse was a part of his/her homework). The reading that this was done only for the purpose of entertainment is not possible in (3b). (3b) may involve plain benefaction as well, but this is never the only meaning expressed. This difference is primarily due to transitivity. As will be discussed in 3.4 (and see also Kittilä 2006), the allative typically codes recipient-beneficiaries in Finnish. Consequently, the argument in the allative is seen as a recipient-beneficiary in (3b). This occurs whenever the denoted event involves production of an entity, as in ‘a parent baked the child a cake’. The allative thus codes plain beneficiaries only in cases that lack any kind of concrete outcome. Examples (3c) and (3d) illustrate another kind of restriction on the use of allative in the function of coding plain beneficiaries. They are in principle similar to (3a) and (3b), but despite this, (3d) is not a felicitous construction. Transitivity is not helpful here, since both examples involve an intransitive verb. The most important reason for this lies in the nature of the act performed. Singing (along with reading, dancing and acting) is an act that can be ‘directed’ at other people. This makes the allative grammatical in (3b) but ungrammatical in (3d): sleeping is an inactive state that cannot be directed at other people. However, it is in order to note that the mere active nature does not suffice for making the use of the allative possible, since verbs such as juosta ‘run’ and kävellä ‘walk’ do not allow an allatively coded beneficiary either; these events also lack the quality of being directed at other participants (the allative is possible with these verbs, but it can only code a goal of motion). It is also important to note that the use of the allative in cases such as (3b) further implies ‘mutual awareness’. This means that the agent is aware of the fact that an act performed by him/her is perceived (and enjoyed) by others, and the beneficiary knows that the agent is performing the action for him/her. These restrictions do not apply to iloksi (‘to the joy of ’), which can be used for referring to any event bringing joy to another participant. Consequently, iloksi can also appear in (3c), which is natural, for example, if parents have finally managed to put their child to bed, and s/he is finally sleeping. The fact that iloksi is also possible with very inactive and completely non-directed actions/states, also has the consequence that awareness is not prerequisite for its use (see below).
3.3â•… Deputative beneficiary The second major type of beneficiaries discussed by Van Valin and LaPolla is represented by deputative beneficiaries. The notion comprises cases in which the referent of the subject is substituting the beneficiary as the agent of the denoted event. Typical
Beneficiary coding in Finnish 
examples thus comprise cases such as the janitor mowed the lawn instead of the dentist, in which the dentist was supposed to mow the lawn but the janitor replaced him as the agent of this event. Five of the mechanisms illustrated above can be employed for coding deputative beneficiaries. Examples are given in (4): (4) a. talonmies vaihto-i lampu-n puolesta-ni janitor.nom change-3sg.pst lamp-acc instead-1sg.psr ‘The janitor changed the lamp instead of me.’ b. talonmies vaihto-i lampu-n takia-ni janitor.nom change-3sg.pst lamp-acc because.of-1sg.psr ‘The janitor changed the lamp because of me.’ c.
talonmies vaihto-i lampu-n vuokse-ni janitor.nom change-3sg.pst lamp-acc for-1sg.psr ‘The janitor changed the lamp for me.’
d. vanhempi pes-i lapse-lta käde-t parent.nom wash-3sg.pst child-abl hand-pl.acc ‘The parent washed the child’s hands (instead of him/her).’ e.
talonmies ava-si ove-n minu-lle janitor.nom open-3sg.pst door-acc 1sg-all ‘The janitor opened the door for me.’
In all the cases in (4), an action or an event is carried out in someone else’s stead, but the exact nature of benefaction varies. Puolesta (4a) is both a neutral and a productive way of stating that an action is carried instead of someone; other features are not relevant in (4a). Also takia expresses deputative benefaction primarily. The main difference to puolesta is that takia adds a negative nuance to deputative benefaction; in both (4a) and (4b) the janitor changed the lamp instead of me, but in the latter s/he did it reluctantly. This reluctance is manifested also in the fact that such adverbs as mielellään ‘with pleasure’ and innoissaan ‘enthusiastically’ are rather infelicitous with (4b). Furthermore, the notion of reason/indirect cause is strong with takia. Vuoksi ‘for’ can also code deputative benefaction (even though it is usually associated with other types of benefaction), but in contrast to puolesta it is not neutral. The use of vuoksi implies that the agent is doing a really big favor to the beneficiary in changing the lamp for him/her. The notion of reason or indirect cause is thus strong with vuoksi, too. This also means that the agent acted volitionally and his/her main purpose was to do something for the beneficiary. Consequently, adverbs such as vahingossa ‘by accident’ cannot appear in clauses with vuoksi. The ablative expresses typical instances of deputative benefaction in cases such as (4d). The meaning of the ablative in this function is close to puolesta, and the ablative could be replaced with puolesta without any major change in meaning in (4d), for example. Last, the allative can also express deputative benefaction. This use is, however,
 Seppo Kittilä
not productive and the allative is the least deputative-like of the mechanisms that can express this meaning. First, this is possible only with intransitive verbs (see (3b)). Second, even though (4e) does not involve any concrete reception, and we cannot thus speak of a true recipient-beneficiary, the beneficiary in (4e) also displays recipientlike (on engager-beneficiary-like) features. For example, (4e) is felicitous if someone opens a door for me with the purpose of me entering or exiting a room. This does involve deputative benefaction, since the act of opening a door was carried out in my stead, but (4e) cannot be used to refer to deputative benefaction only. This is further manifested in the fact that (4e) cannot be substituted with a puolesta-clause without a semantic change. As noted above, puolesta ‘instead of ’ and the allative differ from each other in that the use of allative usually implies that the denoted event has some kind of specific purpose. This is usually evident with cases such as ‘open a door for someone’. However, there are also instances in which the notion of specific purpose is not necessary for the allative to be used. Examples are found in (5), cf. (5) a. sairaanhoitaja ava-si potilaa-lle/koira-lle ikkuna-n nurse.nom open-3sg.pst patient-all/dog-all window-acc ‘The nurse opened the window for/instead of the patient/dog.’ b. vanhempi laitto-i lapse-lle kirja-n hylly-yn parent.nom put-3sg.pst child-all book-gen shelf-ill ‘A parent put the book onto the shelf for the child.’ (lit.to the child’) c.
hovimestari pysäkö-i auto-n presidenti-lle butler.nom park-3sg.pst car-acc president-all ‘The butler parked the car instead of the president.’
d. talonmies vaihto-i mei-lle lampu-n janitor.nom change-3sg.pst 1pl-all lamp-acc ‘The janitor changed the lamp for us.’
In (5), the allative codes primarily deputative benefaction. The denoted events are all such that they do not usually involve any kind of recipient-benefaction or any kind of specific purpose that the beneficiary could engage with. The lack of reception is further manifested in the fact that in (5) the allative could be replaced with puolesta without any change in the nature of benefaction. This is not possible in case of genuine receipient-benefaction. However, allative and puolesta are not in free variation in (5), either, but replacing the allative with puolesta has other kinds of semantic consequences. The use of the allative usually implies that the beneficiary was not an expected agent in the denoted event at all. In (5a) and (5b), beneficiaries are improbable agents due to physical properties; patients in a hospital often have limited capabilities of instigating events, while dogs usually do not have the knowledge needed to open a window and children also lack physical capabilities (size, strength, know-how) necessary for performing certain events. Presidents of states, in turn, are not expected
Beneficiary coding in Finnish 
agents in many events due to their high social status. In (5d), the unexpectedness of the beneficiary as agent is motivated somewhat differently; it is not the incapability of the beneficiary that determines the use of the allative, but rather the fact that it is a part of the janitor’s job to change lamps. The use of puolesta in (5) would imply that both the agent and the beneficiary could have instigated the given event, and even that the beneficiary was the more probable agent, but the agent chose to replace him/her as the agent of the event. Therefore, puolesta would be odd especially with koira ‘dog’, since dogs are unexpected agents of opening, and since the use of puolesta implies that the beneficiary could have performed the denoted action, its use in (5a) would be marginal at best. We may add that the nature of deputative benefaction is somewhat different in (5). In a strict sense, there is no deputative benefaction, since the beneficiary is not an expected agent in the denoted event. The differences between puolesta and allative are manifest also in the fact that both of them may appear in a single clause, as in (5e): (5) e.
hoitaja ava-si ikkuna-n potilaa-lle nurse.nom open-3sg.pst window-acc patient-all
lääkäri-n puolesta doctor-gen instead.of
‘The nurse opened the window for the patient instead of the doctor.’
In (5e), puolesta expresses true deputative benefaction, while the patient is not an expected agent in the event. However, since an event also occurs in favor of the beneficiary in (5a–d), and since the benefaction is primarily ‘substitutive’ in nature, I have opted for viewing cases such as (5) as instances of deputative benefaction.
3.4â•… Recipient-beneficiary The last beneficiary type of Van Valin & LaPolla is illustrated by recipient-beneficiaries, which are defined very much in the same way as in Kittilä (2005) (see also Song, this volume). Recipient-beneficiaries are participants that are not only beneficiaries, but they can also be viewed as recipients, because they receive something transferred to their sphere of control or domain of possession. The role thus comprises both recipientlike and beneficiary-like features. Typical examples are represented by cases such as the dentist baked the phonetician a cake, in which the phonetician both benefits from the event in not having to bake the cake him/herself (deputative benefaction), and in receiving a cake transferred to him/her as a result of the baking event. The allative is clearly the most frequent way of coding recipient-benefaction in Finnish. Examples are found in (6): (6) a.
henkilö leipo-i piiraka-n yksilö-lle person.nom bake-3sg.pst pie-acc individual-all ‘A person baked an individual a cake.’
 Seppo Kittilä
b. vanhempi raken-si talo-n lapse-lle-en parent.nom build-3sg.pst house-acc child-all-3psr ‘A parent built his/her child a house.’
The examples in (6) describe a transitive (effective) event, whose concrete outcome is transferred to another participant’s (beneficiary) domain of possession. The notion of transfer is strong in (6a) and (6b), which is rather unsurprising, since the allative also codes recipients of such verbs as antaa ‘give’, lähettää ‘send’ and ostaa ‘buy’. The notion of reception is common for both recipients and recipient-beneficiaries, but the roles are distinguished by (deputative) benefaction, which is relevant to recipient-beneficiaries only. Nevertheless, the similarities are evident enough to make similar coding possible (see Kittilä 2005 for the cross-linguistic coding of recipientbeneficiaries). In addition to the allative, also varten ‘for’ codes recipient-beneficiaries productively: (6) c.
henkilö leipo-i piiraka-n yksilö-ä varten person.nom bake-3sg.pst pie-acc individual-part for ‘A person baked a pie for an individual.’
d. vanhempi raken-si talo-n las-ta-an varten parent.nom build-3sg.pst house-acc child-part-3psr for ‘A parent built his/her child a house.’
Both (6a–b) and (6c–d) involve reception and (deputative) benefaction. This is manifested, for example, in the fact that (6a) and (6c) can be used interchangeably in many cases. However, also (6a) and (6c) involve subtle differences. The main difference is that the notion of transfer is more strongly present with the allative. Allatively coded beneficiaries are almost always seen as true recipients, and (6a) and (6b) thus both involve actual transfer of possession. Clauses with varten can also have this meaning, and often do, but with varten the focus is rather on the specific purpose the event has. It is the intention of the agent to instigate an event of transfer, but this may also fail to occur for some (usually unspecified) reason. This difference is manifest between (6b) and (6d). The most likely reading of (6b) is that the house was transferred to the child’s possession and the child also moved into the house. In (6d), on the other hand, the natural reading is that the parent has built his/her child a house, but the child never moved into the house and thus no actual transfer of possession occurred. (6d) can also have the reading that the house was built for the child, but it was never the parent’s intention that the house be transferred to the child’s domain of possession, in which case the specific purpose of the building event is focused on. The motivation may, for example, be that the parent can accommodate his/her child occasionally. The differences between allative and varten are manifest also in (6e) and (6f): (6) e.
hän leipo-i piirako-i-ta juhli-a varten s/he bake-3sg.pst pie-pl-part party-part for ‘S/he baked pies for the party.’
Beneficiary coding in Finnish 
f. *hän leipo-i piirako-i-ta juhli-lle s/he bake-3sg.pst pie-pl-part party-all (S/he baked pies for the party)
Baking pies for a party has a specific purpose (the pies will be eaten by the guests at the party) and an unstated recipient (the organizers of the party), but the party itself cannot be the recipient, which excludes the use of the allative, and makes (6f) ungrammatical. The allative and varten ‘for’ usually express direct recipient-benefaction that involves a concrete transfer of an entity to the sphere of control of the recipientbeneficiary (even though this may only be temporary, or it may only be the intention of the agent without actual transfer, as shown above). Vuoksi, takia and the ablative can also code recipient-benefaction, but the nature of reception is clearly different and less direct in nature. Consider: (7) a.
lapse-t sö-i-vät oliive-ja isä-n-sä vuoksi child-pl eat-pst-3pl olive.pl-part father-gen-3psr for ‘The children ate olives for their father.’
b. lapse-t sö-i-vät oliive-ja isä-n-sä takia child-pl eat-pst-3pl olive.pl-part father-gen-3psr for ‘The children ate olives for their father.’ c.
sairaanhoitaja vaihto-i potilaa-lta lakana-t nurse.nom change-3sg.pst patient-abl sheet-pl ‘The nurse changed the patient’s sheets.’
In contrast to (6a–f), the patient does not enter the recipient-beneficiary’s sphere of control or domain of possession in (7a–c). Example (7a) is felicitous, for example, in case that the father is having a very important dinner, during which the children agree to eat olives, even though they do not usually eat them. Eating olives may be important, for example, because the boss of the father likes olives very much, and they are thus served. The benefaction may, for instance, mean that the father gets a promotion in the office. The big favor nuance of vuoksi is very strong in (7a), since the children agree to do something they would not usually do. Example (7b) is similar to (7a) with the difference that in (7b) eating olives is conceptualized as less volitional. In (7c), genuine transfer is also lacking, but the notion of engager-benefaction (see Song in this volume) is strong. Example (7c) also involves deputative benefaction, which is completely lacking in (6a–f) and (7a–b). The semantic differences between allative and varten (i.e. the focus on the reception and the focus on the specific purpose) are further manifested in volitionality and awareness. The allative may code recipient-beneficiaries whenever an event involves some kind of transfer, and features such as volitionality, awareness and animacy are not relevant. On the other hand, varten always implies that the agent is
 Seppo Kittilä
aware of the fact that s/he is doing something beneficial for someone (vuoksi and takia are similar to varten in this respect). Consider: (8) a.
Usain Bolt juoks-i minu-lle satase-n PN run-3sg.pst 1sg-all hundred-acc ‘Usain Bolt’s running earned me 100 Euro.’(Lit. Usain Bolt ran me 100 Euro)
b. Usain Bolt juoks-i satase-n minu-a varten PN run-3sg.pst hundred-acc 1sg-part for ‘Usain Bolt ran a hundred Euro for me.’ c. d.
auto hajo-si minu-lle vaimo-n car.nom break.down-3sg.pst 1sg-all wife-acc ‘The car’s breaking down got me a wife.’ (Lit. The car broke down a wife to me) *auto hajo-si vaimo-n minu-a varten car.nom break.down-3sg.pst wife-acc 1sg-part for (The car’s breaking down got a wife for me) (Lit. The car broke down a wife for me)
Example (8a) is felicitous, for example, if I have made a bet for Usain Bolt and he wins a race, which brings me 100 Euro. Usain Bolt is completely unaware of the fact that his winning the race earned me 100 Euro. In (8b), this reading is not possible. (8b) is felicitous, for example, if Usain Bolt has agreed to run 100 meter race for me and he knows that by running this race he will earn me 100 Euro. The differences in awareness can be stressed by using adverbs expressing (un)awareness. Adverbs of unawareness, such as tietämättään ‘without knowing/realizing’ are possible in (8a), but not in (8b). Adverbs of awareness, such as tietäen ‘knowing/realizing’ are possible both in (8a) and (8b), but the agent can be unaware of the consequences of his actions only in (8a). Example (8c) includes an inanimate agent, but since the event in question involves some kind of transfer, the use of the allative is possible (even though it should be noted that examples like (8c) are extremely rare in spontaneous speech, but they cannot be ruled out as utterly ungrammatical). Inanimate entities are not capable of volitionality and they cannot be aware of the consequences of their actions, which renders (8d) utterly ungrammatical.
4.â•… Other features of the Finnish beneficiary 4.1â•… Preliminaries In this section, I will discuss some interesting features of beneficiary coding in Finnish. Some of the facets discussed below have been already touched upon in previous sections, but in this section the chosen perspective is different. I will discuss the effects of the scrutinized features on the mechanisms of beneficiary coding irrespective of the type
Beneficiary coding in Finnish 
of benefaction they express. The investigated features include the effects of verbal semantics, animacy (of both agent and beneficiary), willingness of the agent and awareness of the agent/beneficiary.
4.2â•… The effects of verbal semantics As has been shown above, Finnish benefactive constructions are semantically distinct. This has the natural consequence that not all mechanisms are possible in all cases; for example, the use of the ablative as a benefactive marker is limited. This variation in acceptability is only natural, since the nature of benefaction depends (among other things) on the nature of the event and its other participants. In what follows, I will examine the effects of verbal semantics on the nature of benefaction and the acceptability of benefactive constructions in Finnish. The scrutinized features include transitivity (referring here to the number of arguments required by a verb) and more general (verb specific) features of verbal semantics. As shown in (2)–(3), the allative may code both recipient-beneficiaries and plain beneficiaries. As was also noted above, allative codes plain benefaction only with intransitive verbs, with transitive verbs (especially in the case of verbs of creation, such as ‘bake’ and ‘paint’) the referent of the allative is interpreted as a recipient-beneficiary. This is expected, since the resulting construction is formally a typical ditransitive clause with an allatively coded argument interpreted as a recipient, which is the typical reading of the allative in ditransitives. In other words, the nature of benefaction is determined by verbal semantics, more precisely by transitivity. Another example is provided by ditransitive clauses. The occurrence of an extra allative in a ditransitive clause would produce an ambiguous construction with two animate allatives, which is blocked in Finnish (see Kittilä 2007 for a cross-linguistic examination of similar cases). The other case form employed for benefaction coding, namely ablative, appears only in transitive clauses, i.e. transitivity makes a very important contribution here, too. On the other hand, I have not been able to find any transitivity-determined restrictions on the use of any of the benefactive postpositions. This is not unduly surprising; case-marked arguments are often more intimately integrated in the clause structure, and they can often be seen as parts of the verb valency. In Finnish, arguments required by verb semantics may appear, for example, in nominative, accusative, allative, illative, elative, ablative and adessive cases. Despite their form, case-marked arguments are integral parts of verbal semantics, which is manifested also in the coding of benefaction. In Finnish, postpositions, on the other hand, always code additional information and peripheral roles that are not integral parts of verb semantics. This has the consequence that they may appear in clauses freely regardless of number and marking of other clausal constituents. One clause may also include more than one benefactive postposition.
 Seppo Kittilä
The closer semantic relation of verbs and case-marked arguments is relevant also in other ways. Even though allative and ablative are so-called semantic cases, and thus semantically richer and more independent from the verb than purely grammatical cases, such as nominative and accusative, their meaning is nevertheless largely dependent of verbal semantics. This is especially evident for the allative, which can code a variety of semantic roles depending on verbal semantics (and also the nature of other participants), consider: (9) a.
isä raken-si lapse-lle legotalo-n father.nom build-3sg.pst child-all lego.house-acc ‘The father built the child a Lego-house.’
b. henkilö ava-si yksilö-lle ikkuna-n person.nom open-3sg.pst individual-all window-acc ‘A person opened a window for the individual.’
In (9a), the allative codes a recipient-beneficiary, while in (9b) the most natural reading would be deputative beneficiary. With intransitive verbs the most natural reading is plain beneficiary, and as shown in (3c–d), with intransitive verbs the event has to be such that it can be directed at other people. The use of the ablative as a beneficiary marker is also largely determined by verbal semantics. The ablative can code benefaction only with a very limited set of verbs, basically with verbs of change and some grooming verbs. The first verb type allows focus on the removal part of the event (as in ‘the nurse changed the patient’s sheets’), which enables the use of the ablative as a marker of benefaction. In the latter case, the use of the allative would yield the wrong reading, which can be avoided by using the ablative. In all other cases, the semantics of the ablative is incompatible with benefaction, and the ablative is understood as referring to a source/maleficiary, cf. (10) a.
talonmies vaihto-i mei-ltä lampu-n janitor.nom change-3sg.pst 1pl-abl lamp-acc ‘The janitor changed the lamp for us/our lamp’
b. vanhempi pes-i lapse-lta tuka-n parent.nom wash-3sg.pst child-abl hair-acc ‘The parent washed the child’s hair (for him/her)’ c.
varas ve-i minu-lta auto-n thief.nom take-3sg.pst 1sg-abl car-acc ‘The thief took my car from me’
In (10a) and (10b), the ablative codes primarily the role of deputative beneficiary, but the role also involves some traits of recipient-beneficiary. In (10c), the referent of the ablative bears the role of maleficiary/source (which is much more typical of the ablative).
Beneficiary coding in Finnish 
Benefactive postpositions are semantically more specific than case forms (which is evident for postpositions of Finnish in general, see Ojutkangas 2008). This means that these postpositions are more directly associated with a specific type of benefaction, and their meaning does not vary according to the semantics of the clause they appear in, which is also a clear difference between case forms and postpositions. For example, in (11) the semantics of puolesta remains constant despite the differences in verbal (and clausal) semantics, cf. (11) a. hän juok-si/kaatu-i puolesta-ni s/he.nom run-3sg.pst/fall.down-3sg.pst instead.of-1sg.psr ‘S/he ran/fell down instead of me.’ b. hän leipo-i kaku-n puolesta-ni s/he.nom bake-3sg.pst cake-acc instead.of-1sg.psr ‘S/he baked a cake instead of me.’ c.
hän maala-si talo-n puolesta-ni s/he.nom paint-3sg.pst house-acc instead.of-1sg.psr ‘S/he painted the house instead of me.’
The changes in transitivity of verbs and verb semantics are completely irrelevant to the nature of benefaction in (11); puolesta codes deputative beneficiaries irrespective of other features of the clause it appears in. The semantically constant and verb-independent nature of postpositions does not mean, however, that the use of postpositions is completely independent of all other properties of clauses. The restrictions attested for postpositions are only differently motivated. The acceptability of constructions with benefactive postpositions is determined rather by the semantics of the given postposition and its compatibility with the denoted event; the events vary according to the nature of benefaction they allow. Hence, the meaning of clauses may vary, but the nature of benefaction remains constant as long as the denoted event allows a certain type of benefaction. The benefactive postpositions of Finnish display variation in the nature of benefaction they code, which has the natural consequence that the restrictions on their use also vary. Iloksi is the least restricted benefactive postposition in Finnish, which is expected, since all events (including weather and natural phenomena, such as wind, fire and earthquake) can happen to someone’s amusement. The notion is further highly subjective, which also contributes to the productive nature of iloksi; even events seen as a nuisance by some may be viewed as beneficial by others. Also puolesta, takia and varten can appear in most clauses (see below for restrictions motivated differently). Puolesta and takia both express deputative benefaction and, since the great majority of actions can be performed in someone else’s stead, it is only natural that these postpositions appear readily in benefactive constructions. Excluded are actions not performable for someone else (most bodily functions), reflexive events and actions/happenings not controllable
 Seppo Kittilä
or executable by humans (such as weather phenomena). The productive nature of varten is also rather unsurprising, since most events can have a specific purpose, which mainly determines the use of varten. As long as an event can be viewed as having a specific purpose, the use of varten is possible. In principle, the use of vuoksi is also rather unlimited, but the ‘doing a big favor’-nuance makes clauses with vuoksi rather odd in many cases. For example, daily routines do not lend themselves to this kind of reading easily, which limits the use of vuoksi.
4.3â•… Animacy 4.3.1â•… Preliminaries Animacy (or rather humanness in the present context) is a feature that is often associated with beneficiaries (see, for example, Blake 1994: 70; Lehmann et al 2000: 93). This is understandable, since only animate participants are capable of using the result of an event for a specific purpose or enjoying an act of plain benefaction. Moreover, the agent of a beneficiary situation is often also animate. This is also to be expected, since only humans may, for example, volitionally and deliberately replace another participant as the agent of an event and thus bear the role of deputative beneficiary. In this section, I will examine the effects of animacy on beneficiary coding in Finnish. I will first examine the effects of (in)animacy on the agent,6 which is followed by a discussion of the animacy effects on the beneficiary. Before proceeding, it is in order to note that in all the discussed cases animate agents and beneficiaries are more natural than the examined instances of inanimate agents and beneficiaries, and all benefactive constructions allow an animate agent/beneficiary, while inanimate agents/beneficiaries are infelicitous with some of the mechanisms. The discussion thus focuses on cases in which inanimate agents/beneficiaries are in principle possible. Moreover, the discussion below only includes cases with an inanimate agent/beneficiary. 4.3.2â•… Animacy of the agent Inanimate agents may occur in four of the benefactive constructions of Finnish, which means that inanimacy of the agent blocks the use of roughly half of the constructions. Only the postpositions varten, takia and vuoksi (if the intended meaning is benefaction, see below) are infelicitous with inanimate agents. Examples of benefactive constructions with an in animate agent are found in (12):
6.â•… The label ‘agent’ is used in a neutralized fashion in this section, it comprises both animate and inanimate causes of events irrespective of whether they can be seen as typical agents or not.
Beneficiary coding in Finnish 
(12) a. b.
tuuli ava-si ikkuna-n vahtimestari-n puolesta wind.nom open-3sg.pst window-acc janitor-gen instead.of ‘The wind opened the window instread of the janitor’ tuli poltta-a pui-ta ilo-kse-ni fire.nom burn-3sg.pst wood-part joy-transl-1sg.psr ‘The fire is burning wood to my joy’
c.
auto hajo-si minu-lle vaimo-n car.nom break.down-3sg.pst 1sg-all wife-acc ‘The car’s breaking down got me a wife’ (Lit. The car broke me down a wife)
d. sade pes-i minu-lta tuka-n rain.nom wash-3sg.pst 1sg-abl hair-acc ‘The rain washed my hair (instead of me)’
Example (12a) is fully grammatical, since puolesta expresses deputative benefaction in a plain fashion; all other types of benefaction are absent, which makes it possible for inanimate agents to surface as the subject of clauses with puolesta. Inanimate entities cannot be volitional agents in beneficiary situations, but they can cause certain events to happen that were supposed to be performed by another agent. There are even fewer restrictions on the use of iloksi, since all events in the world (including weather phenomena) can be enjoyed by someone and thus be denoted by a clause with iloksi. The decisive semantic feature of the allative (in the function of coding recipientbeneficiary) is the transfer of an entity to the domain of possession or sphere of control of an allatively coded participant. This renders the exact nature of the transfer (including animacy of the agent) less relevant, and also allows inanimate agents to surface as the subject of constructions such as (12c). The use of the ablative as a marker of benefaction is generally rather limited (see 4.2.), but as shown in (12d) animacy it is not a prerequisite for its use in this function. Examples in (12) are grammatical, since the mechanisms in question are semantically rather neutral, or they focus on other features making the exact nature of the agent less relevant. They do not involve any agentivity-related features, which would render them incompatible with inanimate agents. This distinguishes the three remaining postpositions from the mechanisms illustrated in (12). First, only humans (and perhaps some higher animals) are capable of instigating events with a specific purpose, which excludes inanimate agents from clauses with varten: with the allative the focus is on the transfer alone, while varten also requires volitionality from its agent (see also (8)). Vuoksi and takia, on the other hand, both require emotional involvement from the agent, which excludes them from benefactive constructions with inanimate agents. For example, in (12a) the use of vuoksi would imply that the wind opened the window instead of the janitor, even though it would have had better things to do, while takia would mean that the wind acted very reluctantly, both of which are infelicitous readings with inanimate entities.
 Seppo Kittilä
4.3.3â•… Animacy of the beneficiary Beneficiaries are usually animate entities, since genuine benefaction implies that the beneficiary is able to use the result of an event for his/her purposes. This generalization is also confirmed by Finnish data, since there are clear restrictions on the occurrence of inanimate beneficiaries. Inanimate entities referring to an institution consisting of human beings (such as school, office, company etc.) appear rather normally as beneficiaries, while genuinely inanimate entities (such as car, house, table etc.) are far less felicitous (and usually ungrammatical) beneficiaries. Consider: (13) a.
joukkue pela-si koulu-n/*auto-n puolesta team.nom play-3sg.pst school-gen/*car-nom instead.of ‘The team played instead of the school/*instead of the car’
b. rehtori tanss-i koulu-n/*talo-n ilo-ksi principal.nom dance-3sg.pst school-gen/house-gen joy-transl ‘The principal danced to the joy of the school/*the house’
The beneficiaries in (13) are in principle inanimate entities, but there are obvious differences in their nature. Inanimate entities with a potential animate referent are more probable beneficiaries, since, for example koulu in (13) can refer to the pupils of the school. The pupils of the school are potential agents, whence deputative benefaction is possible. Moreover, they are capable of enjoying an act of dancing, which renders (13b) grammatical. These arguments are thus very similar to genuine animate arguments, which explains their use as beneficiaries. On the other hand, inanimate entities without a potential animate referent are not possible beneficiaries, since they lack the agentive features needed for genuine (in this case deputative and plain, respectively) benefaction. Below, I will limit the discussion to genuinely inanimate entities, since examination of inanimate entities with an animate referent would not add anything to our understanding of beneficiaries. As shown in (12), three of the beneficiary coding mechanisms are incompatible with inanimate agents. It is interesting that the exact opposite seems to hold for inanimate beneficiaries; varten, vuoksi and takia occur with inanimate beneficiaries, while the rest of the mechanisms are ungrammatical with genuinely inanimate entities (metaphorical uses aside). Consider: (14) a.
henkilö laihdutt-i 20ˉ kiloa tuoli-n vuoksi/takia person.nom lose.weight-3sg.pst 20ˉ kilos chair-gen for ‘A person lost 20 kilos of his/her weight for his/her chair.’
b. henkilö raken-si lämmittime-n auto-a varten person.nom build-3sg.pst heater-acc car-part for ‘A person built the heater for the car.’
Example (14a) is possible in a situation in which a person has noticed that his/her chair is close to falling apart due to the weight of the person sitting on it. As a result
Beneficiary coding in Finnish 
of losing weight, the chair remains intact, which is naturally highly beneficial. Both vuoksi and takia are possible here, with the subtle difference that takia is seen as somewhat negatively biased. In (14b), the agent performs an act of building a heater, which is then transferred to the car’s sphere of control (at least conceptually). Example (14b) is thus best considered an instance of recipient-benefaction. The event of transfer is beneficial for the car, for example, due to the fact that a well-heated car burns less fuel and can be started more easily. The main difference to canonical recipient-benefaction lies in the inanimacy of the beneficiary. The notion of empathy is probably relevant to cases such as (14); inanimate entities can be regarded as beneficiaries if the agent has some kind of affective relation with them. For example, in (14a), the chair is very important to the agent, because of which it is conceptualized as more animate, which certainly also contributes to the acceptability of (14a).
4.4â•… Volitionality and awareness The last features to be discussed here are volitionality and awareness. I will restrict the discussion to animate entities in this section, since the distinction between volitionality/ awareness and non-volitionality/unawareness is relevant only to animate entities, inanimate entities inherently lack these features. I will first discuss the notions from the viewpoint of the agent, which is followed by a discussion of the beneficiary. In principle, volitionality of the agent is rather irrelevant to the beneficiary coding in Finnish. Only takia ‘for/because of ’ usually implies the lack of volitionality, as was shown in (4a–b) in light of the differences between puolesta ‘instead of ’ and takia; the former is neutral as regards volitionality, while the latter always implies a lower degree of volitionality of the agent. All other mechanisms of beneficiary coding usually imply that the agent is acting volitionally. However, it is possible to modify the degree of volitionality with adverbs. All the mechanisms allow unagentive adverbs, such as vastentahtoisesti or tahtomattaan ‘unwillingly/against one’s will’, which means that volitionality is not an entailed feature of the agent in benefactive constructions. The variation in agency is not unduly surprising, since events in general can be performed willingly or unwillingly, and events with a beneficiary are no exception. By contrast, the awareness of the agent makes a more important contribution to the coding of benefaction in that most benefactive constructions are incompatible with agents that are not aware of the beneficial nature of their actions. This is especially evident with vuoksi ‘for’, takia and varten ‘for’, the semantics of which imply agency and awareness. For example, vuoksi usually means that the agent is making a really big favor to the beneficiary, which reading is possible only if the former is aware of the identity of the latter. Takia implies reluctance, which is possible also only if the agent is aware of his/her actions (see also (4a) and (4b)). Varten is used when an event has a specific purpose, which is also closely related to awareness. The rest of the mechanisms can also be used in cases in which the agent is not aware of the
 Seppo Kittilä
beneficial nature of his/her actions, but the mechanisms vary as regards how freely they allow unaware agents. Puolesta and ilokseni ‘to my joy’ are both rather liberal in this regard. Puolesta only states that someone substituted the beneficiary as the agent of an event, which does not entail awareness. I can, for example, mow the lawn instead of someone else without knowing that this was his/her task, or any event I am involved in can be enjoyed by others without me knowing about this. The ablative is similar to puolesta in primarily denoting deputative benefaction, which should make its use possible with unaware agents as well. However, the ablative usually appears with agents that are aware of the consequences of their actions. The reason for this probably lies in the fact that events that allow ablatively coded beneficiaries usually involve high degrees of awareness, as in ‘the parent washed the child’s hands for/instead of him/her’. But in cases such as ‘the janitor changed the lamp for us’, the agent can also be unaware of deputative benefaction, exactly as with puolesta. The last of the mechanisms, namely the allative case, is split as regards the awareness of agent depending on the nature of benefaction. As shown in (8d), the use of the allative does not entail awareness if it denotes recipient-benefaction. On the other hand, the allative and iloksi ‘to the joy of ’ both of which denote plain benefaction, differ from each other with respect to the awareness of the agent; allative implies awareness, while iloksi does not (see also (2)). As regards the volitionality and awareness of the beneficiary itself, it seems that the nature of benefaction largely determines the (un)grammaticality of benefactive constructions. These features are completely irrelevant to deputative beneficiaries and recipient-beneficiaries. This is understandable, since someone can substitute the beneficiary as an agent of an event without the beneficiary being aware of this. In a similar vein, recipient-benefaction is possible even if the beneficiary is not aware of it. I can, for example, bake a cake for someone without him/her being aware of this. It is, however, in order to note here that the insignificance of volitionality and awareness refers here only to performing the beneficial event; the beneficiary does not need to be aware of the fact that it has been substituted as an agent of an event or that an event of recipientbenefaction has occurred for him/her. Beneficiary coding is nevertheless possible. However, actual benefaction implies that the beneficiary is aware of the occurrence of a beneficial event; otherwise, genuine benefaction is not possible. For example, an instance of deputative benefaction is genuinely beneficial only when the beneficiary becomes aware of the beneficial event in question. The irrelevance of volitionality and awareness is only expected, since beneficiaries are (especially when the denoted beneficial event is in progress) rather passive participants, and they may even be absent when the event that eventually results in benefaction occurs. Plain benefaction differs from the two other types of benefaction as regards the contribution of volitionality. The beneficiary usually needs to be animate and also aware of the occurrence of the beneficial event in case of plain benefaction. This follows, since the beneficiary needs to perceive the beneficial event in question for being able to enjoy it, which entails awareness, and also animacy. However, the two ways of coding
Beneficiary coding in Finnish 
plain benefaction vary according to how prominent the contribution of awareness is. Consider: (15) a.
minä laula-n häne-n ilo-kse-en I.nom sing.pres-1sg s/he-gen joy-transl-3psr ‘I sing to his/her joy.’
b. minä laula-n häne-lle I.nom sing.pres-1sg s/he-all ‘I sing for (lit. to) him/her.’
The use of iloksi is possible only if the beneficiary him/herself is aware of the beneficial event. On the other hand, the awareness of the agent is completely irrelevant to the use of iloksi, as was noted above. Allative in the function of coding plain benefaction is the exact opposite of iloksi. Allative can be used also when the beneficiary is not aware of the fact that an event is occurring for his/her benefaction, whereas the awareness of the agent is a prerequisite for the use of the allative for coding plain benefaction. Whether the beneficiary actually enjoys the event meant as beneficial for him/her is less relevant. The two mechanisms of plain benefaction coding are thus mirror images of each other; iloksi is more beneficiary-focused, while the allative is more agent-oriented and focused on the fact an agent is actively and volitionally doing something that is meant to entertain a participant external to the agent him/herself.
5.â•… Summary and discussion 5.1â•… Summary In Section 3, the Finnish benefactive constructions were discussed in light of Van Valin and LaPolla’s (1997) typology. This discussion is summarized in Table 1.7 Table 1.╇ Schematic representation of the Finnish beneficiary coding in light of Van Valin & LaPolla (1997) Allative
takia
vuoksi
Ablative
iloksi
puolesta
varten
Plain
+
(+)
(+)
−
+
−
−
Deputative Recipient
+ +
+ (+)
(+) (+)
(+) +
− −
+ −
− +
7.â•… In Tables 1 and 2, + means that a notion is relevant to the examined mechanism, while – means the lack of it; (+) means that a feature is usually relevant, but may be absent as well, while (−) refers to the opposite of this.
 Seppo Kittilä
As can be seen in Table 1, the allative can be productively used for coding all the three types of benefaction suggested by Van Valin & LaPolla. This does not mean, however, that the allative could always be used to code any of the types, but its productivity varies drastically and is conditioned by other features, such as verbal semantics. Takia and vuoksi can also code all three types, but as Table 1 implies, the use of these postpositions for coding benefaction is far from being productive (see also the discussion below). The ablative is not possible with plain benefaction, but may code the other types of beneficiaries (even though its use is generally limited in this function). Puolesta, iloksi and varten are the semantically most specific markers of benefaction, since they can only be used for coding one type of benefaction. Section 4 examined other features relevant to the coding of benefaction in Finnish, but which are independent of the nature of benefaction per se. This discussion is summarized in Table 2: Table 2.╇ Schematic summary of the features discussed in Section 4 Allative
Ablative
puolesta
varten
Animacy agt
−
−
−
+
Animacy ben
+
+
+
(+)
Volitionality agt
−
−
−
+
takia
vuoksi
iloksi
+
+
−
(+)
(+)
+
−
+
−
Volitionality ben
−
−
−
−
−
−
(+)
Awareness agt
−
(−)
−
+
+
+
−
Awareness ben Event semantics
− +
− +
(−) −
− +
(−) (−)
− (−)
+ −
A few generalizations can be drawn from Table 2. The most evident of these is that beneficiaries usually need to be animate in Finnish. The animacy of the agent is less relevant, even though it is necessary to state that agents are usually animate, as is typical of transfer events in general. The contribution of volitionality and awareness is not significant either, as Table 2 clearly shows. As regards the relevance of verbal semantics to beneficiary coding, case forms are more sensitive to this than postpositions. This is especially evident with the allative that can code all three types of benefaction discussed in Section 3, but the actual reading of the allative varies according to the semantics of the denoted event. Puolesta and iloksi are in principle completely independent of verbal semantics and the nature of the denoted event, which is expected in light of their semantics. Vuoksi and takia are rather interesting in this respect, since they primarily code ‘big favor’ and reluctant agents respectively, but the actual nature of benefaction depends on the semantics of the denoted event. As such, the semantics of the postpositions does not vary in any significant way according to the semantics of the denoted event, but the nature of benefaction is determined by verbal semantics.
Beneficiary coding in Finnish 
5.2â•… Rationale I end the paper by briefly discussing the rationale of beneficiary coding in Finnish. I will first discuss the rationale behind the use of different mechanisms for coding the type of benefaction they code and then address the features examined in Section 4. The allative is a highly polysemous case in and across languages (see e.g. Blansitt 1988 and Rice & Kabata 2007), which is manifested also in Finnish beneficiary coding; the allative can code all three types of benefaction discussed in Section 3. The coding of recipient-benefaction by the allative is expected, since the allative also codes recipients. Recipients and recipient-beneficiaries have features in common, which explains the identical coding of these roles (see Kittilä 2005 for a more detailed discussion of this). The coding of plain benefaction is also understandable in light of the basic function of the allative. The allative can code plain benefaction only when the beneficial event is directed at the beneficiary, which is in accordance with the typical semantics of recipients as endpoints of transfer. The coding of deputative benefaction seems somewhat less natural, but can perhaps also be explained by recipient coding. Recipients of transfer events are inactive and non-initiating participants (as opposed to agents), exactly as are deputative beneficiaries coded by the allative that are not expected agents in the denoted event to begin with. They are not targeted participants in the same way as recipients, but they nevertheless have some features in common with recipients, which explains the coding of (specific instances of) deputative benefaction by the allative. The use of the ablative for coding benefaction seems rather odd at first, and it is also rather rare cross-linguistically (similar cases are attested only in a handful of languages, including Mekeo (Alan Jones, p.c.), Gawwada (Mauro Tosco, p.c.), Korean (Jae Jung Song, p.c.) and Kuuk Thaayorre (Alice Gaby, p.c.)). However, this polysemy becomes understandable if we consider the two instances in which the ablative can be employed for coding benefaction. First, ablative can code events of exchange (see (10a)). These events always involve two parts, the removal of the old entity (such as a broken lamp) and insertion of the new entity (such as a working lamp). It is possible to focus on either of these two parts, which renders ablative coding of the beneficiary possible. Second, ablative can code instances of deputative benefaction in which the beneficial action is targeted at a body part of the beneficiary, as in (4d) and (10b). The use of the ablative, instead of the ‘more normal allative’, can be explained by the fact that the use of the allative would yield a wrong kind of reading. Consider the examples below: (16) a.
vanhempi pes-i lapse-lta tuka-n parent.nom wash-3sg.pst child-abl hair-acc ‘The parent washed the child’s hair (for him/her).’
b. !vanhempi pes-i lapse-lle tuka-n parent.nom wash-3sg.pst child-all hair-acc !‘The parent washed the child’s hair.’
 Seppo Kittilä
(16a) is a typical example of a benefactive construction with an ablative. In (16b), the ablative is replaced with the allative, which has clear semantic consequences: (16b) is not ungrammatical, but its reading is clearly different from (16a). The natural reading of (16b) is that the child got some hair as a result of the washing event, the child was bald before the event occurred. This is a relatively odd reading, since washing is normally not an act that causes hair to grow. However, (16b) would be possible, for example, in case the child’s hair is completely covered with dirt, and after washing it looks more like a real hair again. The strong association between allative case and reception is very strong, whence the ablative is used in cases such as (16a). As was noted above, cases are usually more dependent on verbal semantics, which causes case polysemy also in beneficiary coding. Postpositions are semantically more specific, which means in this context that a detailed discussion of the rationale behind the use of different postpositions for coding different instances of beneficiary would not add very much to our understanding of beneficiary in Finnish (on the other hand, a cross-linguistic investigation may reveal interesting aspects, i.e. which instances of benefaction are coded by distinct/specific adpositions in the world’s languages). In other words, stating that puolesta codes deputative benefaction because this is compatible with its semantics does not reveal any new aspects of its semantics or benefaction in general. The only thing of potential interest here is illustrated by the fact that Finnish has very specific and very productive postpositions for coding deputative benefaction (puolesta) and plain benefaction (iloksi). Similar cases are attested at least in English and German, so this is not typologically a unique phenomenon. As shown in Table 2, the seven mechanisms of beneficiary coding of Finnish differ according to the features examined in Section 4. Generally, animacy is more relevant to beneficiaries than to agents of beneficial events, which can be explained by the fact that only animate entities can genuinely benefit from events and use their outcome for a specific purpose. On the other hand, animate entities can benefit from events caused by inanimate entities, which makes the animacy of agents less relevant. The allative is strongly associated with reception, and the focus is thus on the beneficiary, which renders the exact nature of the agent less important and allows inanimate agents. The animacy of the beneficiary is also understandable in light of the focus on reception and the fact that the allative codes animate recipients. It would be odd if inanimate beneficiaries were possible, whereas inanimate recipients were excluded. The ablative is similar to the allative as regards the effects of animacy on its use, which probably follows at least to some extent from the fact that the ablative codes animate sources and can therefore be seen as the mirror image of the allative. The ablative is, however, somewhat more intimately associated with animacy of the agent due to the nature of the events that allow ablatively coded beneficiaries. The other discussed features are not especially relevant to allative and ablative. As noted above, puolesta ‘instead of ’ is a neutral way of
Beneficiary coding in Finnish 
coding deputative benefaction, which can also be seen in Table 2. The animacy of the beneficiary is only natural, since beneficiaries coded with puolesta are potential agents of events, which are substituted by other entities as agents. Typical agents are animate (human), which accounts for the animacy of the beneficiary in this case. Varten ‘for’—in contrast to the allative, which also codes recipient-benefaction—highlights the specific purpose of the denoted event. This strongly implies agency and volitionality from the agent (see (8a) and (8b)). On the other hand, varten may occur with inanimate beneficiaries in some cases (see (8a–d)). This follows, since the focus is more on agentrelated features, which gives the beneficiary more ‘freedom’. Vuoksi ‘for’ and takia ‘for/ because of ’ are also very agent-oriented postpositions in beneficiary coding, which is verified by Table 2, since the features of Table 2 are more relevant to the agent than to the beneficiary. Volitionality distinguishes between the two; takia highlights the lack of volitionality, while the semantics of vuoksi strongly implies it. As such, takia and vuoksi occur readily with inanimate arguments, but their referents are in this case typically seen as reasons or indirect causes. Vuoksi also allows inanimate beneficiaries in cases, in which an inanimate entity is somehow personified. Iloksi ‘to the joy’ is very clearly split between agent- and beneficiary-related features. Its felicity is not in any way determined by agent-related features, but iloksi is usually possible only with volitional animate beneficiaries that are aware of the beneficial event occurring. This division is understandable, since any event, regardless of its instigator/cause, can be enjoyed by animate entities.
Acknowledgements I would like to thank the following colleagues (in alphabetical order) for their comments on some of my Finnish examples: Marja-Liisa Helasvuo, Tuomas Huumo, Emmi Hynönen, Jaakko Leino, Jari Sivonen and Katja Västi. I also thank the students of my Semantiikan ja pragmatiikan jatkokurssi (Advanced semantics and pragmatics) held at the University of Helsinki, spring 2008 for fruitful discussions of and valuable comments on some of my ideas (benefaction was one of the topics discussed thoroughly on this course). Discussions with Jae Jung Song during my visit at The University of Otago and with the co-editor of the book and the co-author of the first chapter, Â�Fernando Zúñiga, face-to-face at conferences (in Zürich, Joensuu, Leipzig, Helsinki and Tartu) and in dozens and dozens of mails have also helped me to clarify many issues related to benefaction, not only in Finnish, but also across languages or independently of Â�individual languages. Finally, I express my gratitude to Academy of Finland (grant number 1127724) for providing funding for this study.
 Seppo Kittilä
References Blake, Barry. 1994. Case. Cambridge: CUP. Blansitt, Edward. 1988. Datives and allatives. In Studies in Syntactic Typology [Typological Studies in Language 17], Michael Hammond, Edith Moravcsik & Jessica Wirth (eds.), 173–191. Amsterdam: John Benjamins. Kittilä, Seppo. 2005. Recipient prominence vs. beneficiary prominence. Linguistic Typology 9(2): 269–297. Kittilä, Seppo. 2006. On distinguishing between recipient and beneficiary in Finnish. In Grammar from the Human Perspective: Case, Space and Person in Finnish [Current Issues in Linguistic Theory 277], Marja-Liisa Helasvuo & Lyle Campbell (eds.), 129–152. Amsterdam: John Benjamins. Kittilä, Seppo. 2007. A typology of tritransitives: Alignment types and motivations. Linguistics 45(3): 455–508. Lehmann, Christian, Shin, Yong-Min & Verhoeven, Elisabeth. 2000. Person Prominence and Relation Prominence. On the Typology of Syntactic Relations with Special Reference to Yucatec Maya. Munich: Lincom Europa. Ojutkangas, Krista. 2008. Mihin suomessa tarvitaan sisä-grammeja (What for does Finnish need sisä- ‘inside’ grams). Virittäjä. 112(3): 382–400. Rice, Sally & Kabata, Kaori. 2007. Cross-linguistic grammaticalization patterns of the allative. Linguistic Typology 11(3): 451–514. Van Valin, Robert D., Jr. & LaPolla, Randy J. 1997. Syntax. Cambridge: CUP.
Benefactives in Laz René Lacroix
The paper deals with the expression of benefactives in Laz, an endangered South Caucasian language spoken in North-East Turkey. Laz has two ways of encoding benefactives: an applicative derivation and a postposition (šeni). In the preliminary sections, I present the system of cross-referencing affixes, the major alignment types and the system of valency operators. Then I describe the morphosyntax and semantics of the applicative derivation and the postposition used to encode beneficiaries, before some tentative remarks about the factors that may condition the choice between the applicative and the postposition. Both the applicative and the postposition have a range of other functions: the applicative may express the recipient, the locative and the possessor; the postposition may express the cause, the purpose and also has the meaning ‘about, concerning’.
1.â•… Introduction This paper deals with the expression of benefactives in Laz, a South Caucasian Â�(Kartvelian) language spoken in North-East Turkey.1 Estimates about the number of speakers range from 45,000 (Andrews 1989) to 500,000 (Holisky 1991). Laz is an unwritten endangered language. Although young people still understand it, they speak only Turkish. According to some scholars (Marr 1910; Čikobava 1936), Laz is divided into three main dialects. Kutscher (2001) suggests that Ardeşen variety constitutes a fourth dialect. The data presented here is from the dialect of Arhavi. My study is based on a corpus of published texts (Dumézil 1937, 1967 and 1972; Ž>ent’i 1938; Q’ipšiŠe 1939; K’art’ozia 1972 and 1993) as well as data from my work with native speakers. If not otherwise stated, the examples are from my informants.
1.â•… I would like to thank Denis Creissels and Winfried Boeder for their valuable suggestions in the preparation of this article. Of course, I am responsible for all shortcomings and inaccuracies. In the transcription of Laz, the apostrophe marks glottalized consonants; Š¦ corresponds to [dŠ], Š to [dz] and c to [ts].
René Lacroix
Laz has two ways of encoding benefactives: an applicative derivation and a postposition. Section 2 outlines the system of cross-referencing affixes in Laz; in Section 3, I �present the major alignment types of this language and in Section 4, the system of valency operators; then I describe the morphosyntax and semantics of the �applicative derivation and the postposition used to encode beneficiaries (Sections 5 and 6); in Section 7, I make some remarks about the factors that may condition the choice between the applicative and the postposition in the expression of the beneficiary, before the �conclusion in Section 8.
2.â•… Cross-referencing affixes The morphological structure of finite verb forms in Laz may be summarized as follows: −3 −2 −1 ╇╛0 ╇╛1 ╇╛2
preverbs cross-referencing prefixes valency operators root causative and other suffixes TAM and cross-referencing suffixes
The two grammatical categories which will be of importance to the rest of the discussion are cross-referencing affixes and valency operators. There are two sets of cross-referencing affixes, glossed by roman figures (“Set I” and “Set II”). The following table gives the paradigm of Set I affixes, without phonologically conditioned allomorphs. The verb -Šir- ‘see’ in the present tense with a 3rd person object is taken as an example (3rd person objects are not marked in the verb; see Table 2). This verb takes the thematic suffix -om. Thematic suffixes appear in certain tenses only, like the present and the imperfect; they do not appear in the aorist. The system of cross-referencing in the future differs from that of the other tenses and will not be presented here. Table 1.╇ Set I cross-referencing affixes CR prefixes 1sg 2sg 3sg 1pl 2pl 3pl
root
CR suffixes
examples
-s/n/u -t -t -an/nan/es/n
b-Šir-om ‘I see him’ Šir-om Šir-om-s b-Šir-om-t Šir-om-t Šir-om-an
b-
b-
Benefactives in Laz
For the 3rd person singular and plural, the choice between the different allomorphs is conditioned by verb class and tense. For instance, the verb in Example (3), which belongs to another class than ‘see’, takes the Set I 3rd person singular suffix -n. Set II affixes, as they appear in direct verb forms, are presented in Table 2, Â�without phonologically conditioned allomorphs.2 There is no difference between 3rd person singular and 3rd person plural. The verb ‘see’ in the present tense is again taken as an example, with a 3rd person singular subject. The prefix m- cross-references 1st person and the prefix g- second person. The suffixes -t, -an, -nan, -es and -n indicate plurality. Table 2.╇ Set II cross-referencing affixes CR prefixes 1sg 2sg 3 1pl 2pl
root
CR suffixes
examples
mgmg-
-t or -an/nan/es/n -t or -an/nan/es/n
m-Šir-om-s g-Šir-om-s Šir-om-s m-Šir-om-an g-Šir-om-an
‘he sees me’ ‘he sees you’ ‘he sees him/them’ ‘he sees us’ ‘he sees youpl’
The choice between -an, -nan, -es and -n is again conditioned by verb class and tense. The choice between -t on the one hand and -an/nan/es/n on the other depends on person combinations: if Set I is 1st or 2nd person, then Set II suffix is -t; if Set I is 3rd person, then Set II suffix is -an, -nan, -es or -n. This is shown in the following table (R = root). See Lacroix (2008) for a diachronic explanation of this system of plural marking.
Set II 1pl Set I
1sg 2sg 3sg
m-R-t m-R-an/nan/es/n
examples 2pl g-R-t g-R-an/nan/es/n
g-Šir-om-t ‘I see youpl’ m-Šir-om-t ‘you see us’ g-Šir-om-an ‘he sees youpl’
2.â•… In Laz, direct verb forms may be defined as forms occurring in constructions where 3rd person arguments cross-referenced by Set I affixes trigger number agreement in the verb. By contrast, in “inversion”, 3rd person arguments cross-referenced by Set II affixes trigger number agreement (see below).
René Lacroix
3.â•… Alignment types 3.1â•… Full NPs Sentence 1 illustrates the basic transitive construction. The A argument (berek) is in the ergative and is cross-referenced on the verb by a Set I affix (-u); the O argument (ocxo¦Š) is in the absolutive and is cross-referenced by a Set II affix: (1) Bere-k ocxoŠˇ me-Ø-tk’oč-u. boy-erg comb.abs pv-ii3-throw-aor.i3s ‘The boy threw the comb.’
(Dum37.10.3)
Non-3rd person objects are overtly cross-referenced on the verb: (2) Hentepe-k si var g-Šir-om-an. demp.pl-erg 2s neg ii2-see-ths-i3p ‘They will not see you.’
(Dum37.51.3)
In this paper, the term “object” refers only to the syntactic role fulfilled by ocxo¦Š ‘comb’ in (1). A transitive construction is one which involves an object. The S argument is cross-referenced by Set I affixes. Certain intransitive verbs take an absolutive subject (3) while others take an ergative subject (4). (3) Mk’yapu xrock-u-n. jackal.abs die-ths-i3s ‘The jackal is dying.’
(Ž>.96.26)
(4) K’oči-k čind-um-s. man-erg sneeze-ths-i3s ‘The man sneezes.’
The comparison of (3) with (1) shows that the alignment of absolutive-S verbs is of the mixed type: S behaves like O with respect to case marking (both are in the absolutive), but it behaves like A with respect to cross-referencing (both are cross-referenced by Set I affixes). On the other hand, the comparison of (4) with (1) shows that the alignment of ergative-S verbs is accusative: S behaves like A according to case marking (both are in the ergative) and cross-referencing (both are cross-referenced by Set I affixes). Prototypically, then, Set I affixes are used to cross-reference the A and S arguments, while Set II affixes cross-reference O arguments. These two sets have been called by some authors “subject” and “object” markers (Holisky 1991; Harris 1985; Tuite 1998; Boeder 2005). I prefer to avoid this terminology since there are constructions known as “inversion” where the argument cross-referenced by so-called “object” markers (i.e. Set II affixes) exhibits subject properties. This is the case, e.g. in the perfect:
Benefactives in Laz
(5) Nak’o c’ana yen baba-čkimi var m-i-Šir-u-n! how.many year be.i3s father-poss1s neg ii1-sfx-see-ths-i3s ‘How many years I haven’t seen my father!’ (Dum67.XII.93)
In this sentence, although the 1st person is cross-referenced by a Set II affix (m-), it must be recognized as the (non-canonical) subject, for several reasons which will not be dealt with here (these include word order and number agreement in the verb). Consequently, I prefer to use the more neutral terms “Set I” and “Set II”, which do not prejudge the actual functions of these affixes.
3.2â•… 1st and 2nd person pronoun split 1st and 2nd person pronouns have the same form in the ergative, absolutive and dative cases:
ergative, absolutive and dative
1sg 2sg 1pl 2pl
ma si čku tkva
The examples below show that the 2nd singular pronoun si may function as A, O and S. In the paradigm of Set I cross-referencing affixes, the 2nd person singular is the only cell that has no overt making (see Table 1). (6) Si mu ču-me? 2s what wait-ths ‘What are you waiting for?’ (7) Ma si e-k-č’op-are. 1s 2s pv-ii2-marry-fut.i1/2s ‘I will marry you.’ (8) Si guruni ye-i? 2s donkey be-int ‘Are you a donkey?’
(Dum67.XX.42)
(Dum37.121.6)
(K’art’72.193.26)
Thus, the alignment of 1st and 2nd person pronouns is neutral (A = O = S). 1st and 2nd person cross-referencing, however, remains accusative.
4.â•… Valency operators The immediate pre-root slot (–1) may contain one of five valency operators, each of them consisting of one vowel: o- indicates that the verb is transitive; it is also used to derive applicative forms; i- marks middle voice; i/u- mark mainly applicative
René Lacroix
derivation; a- marks, among other things, applicative derivation and middle voice simultaneously. Two valency operators cannot occur together. Such markers are often called “preroot vowels” or “preradical vowels” in Kartvelian linguistics. These terms refer to the phonological rather than morphosyntactic level. Since the general function of these markers is to indicate the valency of the verb they occur in, the term “valency operator” is preferred here. The valency operators marking applicative derivation (i/u-, o- and a-) are examined in the following section. In general, the different functions of the marker i-, which is used in the other languages of the family as well, are not given a unified account in the literature on these languages. In Lacroix (forthcoming), I argue that these functions correspond to a domain covered by “middle” markers in other languages. Compare for instance (9a) and (9b). In (9b), the addition of i- to the verb yields an intransitive verb with anticausative reading – a subtype of the middle. (9) a.
Bozo-k nek’na ge-nk’ol-um-s. girl-erg door pv-close-ths-i3s ‘The girl closes the door.’
b. Nek’na ge-i-nk’ol-e-n. door pv-mid-close-ths-i3s ‘The door closes.’
The term “middle” is taken here in a broad sense, corresponding to that of Kemmer’s (1993). It covers intransitive constructions such as the anticausative, but also includes transitive constructions, as in (10): 3 (10) Ma oxoi b-i-k’od-um. 1s house i1-mid-build-ths ‘I build a house for myself.’
Transitive verbs may be classified in two groups according to whether they involve the valency operator o- or not. This operator (glossed tr for “transitive”) is not correlated to any semantic distinction; its presence is lexically conditioned. Some examples are given below. Some transitive verbs which do not involve the valency operator op-šinax-am i1-hide-ths ‘I hide sth’ p-kos-um i1-sweep-ths ‘I sweep sth’ me-m-¦Šγon-am pv-i1-send-ths ‘I send sth’ p-xazi-um i1-prepare-ths ‘I prepare sth’ p-č’ar-um i1-write-ths ‘I write sth’ 3.â•… Thus, the term “middle” is not taken here in the sense it has in traditional Georgian grammar, for instance in Tschenkéli (1958, Lesson 28: “Mittelverben”).
Benefactives in Laz
Some transitive verbs which involve the valency operator ob-o-rd-am i1-tr-raise-ths ‘I raise sb’ b-o-rg-am i1-tr-plant-ths ‘I plant sth’ b-o-gzal-em i1-tr-send-ths ‘I send sb’ b-o-čil-am i1-tr-marry-ths ‘I marry him (to a girl)’ go-b-o-kt-am pv-i1-tr-turn-ths ‘I turn sth’ Causative verb forms include the vowel o- by default: (11) o-mt’-in-am-s / *mt’-in-am-s tr-flee-caus-ths-i3s ‘he makes it flee, he chases it away’
(Dum67.XXIII.8)
5.â•… The applicative derivation 5.1â•… Morphosyntax Sentence (12b) is an example of the applicative derivation. It can be compared with (12a), the transitive construction from which it is derived. In (12b), case marking of the A and O arguments is the same as in the corresponding non-derived construction, but the second set of cross-referencing affixes represents the applicative argument (beremušis) marked by the dative case; furthermore, the verb contains a mark of applicative derivation (u-). (12) a.
Baba-k oxoi do-k’od-u. father-erg house pv-build-aor.i3s ‘The father built a house.’
b. Baba-k bere-muši-s oxoi d-u-k’od-u. father-erg child-poss3s-dat house pv-ii3.appl-build-aor.i3s ‘The father built a house for his son.’
The applicative derivation can apply to transitive as well as intransitive verbs; see (13) and (14b). (13) Sum c’ana-s ma m-i-čališ-i! three year-dat 1s ii1-appl-work-imp ‘Work for me for three years!’4
(Ž>.34.16)
4.â•… As we see in this example, the dative may be used to form adjuncts (sum c’ana-s ‘three years’, oxoi-s ‘at home’).
René Lacroix
The operator symbolized by i/u- is realized as i- when the applied argument is 1st or 2nd person and u- when it is 3rd person: m-i-čališ-u g-i-čališ-u u-čališ-u
ii1-appl-work-aor.i3s ii2-appl-work-aor.i3s ii3.appl-work-aor.i3s
‘he worked for me’ ‘he worked for you’ ‘he worked for him’
These constructions are very similar to applicative derivations in that they encode the presence of an additional core argument (other than an agent as in the causative). The difference between Laz applicatives and prototypical applicatives is that in Laz, the applicative argument is not case-marked in the same way as O (see Peterson 2007: 1).5 As a consequence, the applicative derivation does not transitivize intransitive verbs (see the intransitive construction in 14b). Some authors, however, extend the notion of applicative to include such non-canonical applicative mechanisms, and I adopt this solution (Dixon & Aikhenvald 2000: 15).6 In the literature on Kartvelian languages, the operator u- illustrated in (12b and 13) is known as “objective version”. The applicative argument has a special syntactic status. Like core terms of non-derived constructions (A, O and S), it is cross-referenced on the verb, and hence cannot be considered an oblique. On the other hand, it differs from A, O and S arguments by its dative marking. This suggests recognizing a fourth core syntactic role, which can be symbolized by E (standing for “extension to core”), following Dixon & Aikhenvald (2000: 3). Applicative arguments, like O arguments, do not trigger number agreement in the verb. In Example (26), for instance, the plurality of the applicative argument hentepes is not indicated in the verb. The applicative in i/u- is used, among other things, to express the beneficiary. Its semantics is examined in the next section. The valency operator o- also is used to derive applicatives. Although it is homonymous with o- indicating transitive verb forms (see above), it must be distinguished from it on a formal ground. The applicative with o- is not very widespread. It is a locative applicative, indicating a movement toward something or a contact. Its semantic scope is thus more restricted than the applicative in i/u- (see next section). Compare sentence (14a), which involves a non-derived intransitive verb (¦Šant’u) and (14b), where the same root appears in an applicative verb form, marked by the operator o-. The applicative argument is nek’na-s ‘door-dat’.
5.â•… The dialect of Ardeşen has lost the dative and ergative cases (Dumézil 1972; Kutscher 2001). As a consequence, neither the applicative argument nor the O are case-marked. In this variety, then, the applicative construction is closer to a prototypical applicative. 6.â•… Important references about applicatives include, among others, Alsina & Mchombo (1990 and 1993), Donohue (1999), Baker (1988) and Austin (2005), in addition to Peterson (2007).
Benefactives in Laz
(14) a.
Xasani dido ora-s Šabuni Šˇan-t’u Hasan much time-dat sick lie-impft.i3s ‘Hasan was lying sick for a long time.’
b. Didi kva n-o-Šˇan-t’u nek’na-s. big stone pv-appl-lie-impft.i3s door-dat ‘There was a big stone against the door.’
(Ž>.5.8)
(K’art’72.166.6)
This example shows that in Laz, the applicative construction is not restricted to human participants, as it is in other languages (Polinsky 2005). Two differences can be pointed out between o- indicating transitive verb forms and o- indicating applicative. First, verb forms with transitive o- cannot be shown to derive from more simple forms (except for the causative). By contrast, verb forms like noŠ¦ ant’u in (14b) result from the addition of o- to a more basic verb. Secondly, transitive o- indicates the presence of an absolutive argument (the object), while applicative o- indicates the presence of a dative argument (the applicative argument). The valency operator a- is used, among other things, to derive applicatives from middle verbs. Compare (9b) above with (15), which involves the addition of a dative argument, bič’is ‘boy’, bearing the semantic role of maleficiary: (15) Bič’i-s ixi-te nek’na gy-a-nk’ol-u. boy-dat wind-instr door pv-appl.mid-close-aor.i3s ‘Because of the wind, the door closed on the boy.’
Laz does not possess an instrumental or a comitative applicative, contrary to many languages with an applicative derivation (Peterson 2007: 67, 202; Polinsky 2005). Instruments and comitative participants are oblique arguments expressed respectively by the instrumental case and the postposition k’ala: (16) K’ilič’i-te divi-s ti n-o-k’vat-u. sword-instr giant-dat head pv-tr-cut-aor.i3s ‘He cut off the dragon’s head with a sword.’ (17) Si čkuni k’ala gyari var čk’om-i. 2s 1p with meat neg eat-aor ‘You didn’t eat with us.’
(K’art’93.139.9)
(Ž>.49.3)
Tests used for determining which object properties are retained, in the applicative construction, by the base object and the applicative argument include relativization and passivization. In the i/u-applicative derivation, both the base object (nena ‘word’ in€[18]) and the applicative object (bee ‘child’ in [19]) may be relativized: (18) A>a-k hamu-s-na u-c’-u-dort’un agha-erg demp-dat-sub ii3.appl-tell-aor.i3s-pprf nena ko-g-a-šin-u. word pv-pv-appl.mid-remember-aor.i3s
‘He remembered the words that the agha had told him.’
(Ž>.34.28)
René Lacroix
(19) Yei-na b-u-xen-i bee-k i-bga-s. place-sub i1-ii3.appl-make-aor child-erg mid-cry-i3s ‘The child I made some place for cries.’
The passive, as the middle, is marked by the operator i-, which could have been labeled “mediopassive”, but passive constructions occur rarely in Laz, and I simply use the term “middle”. I have not found in my corpus any example of passivization of the base object or the applicative argument, neither have I been able to elicit such constructions with informants. Only one argument can be cross-referenced by Set II affixes. In all examples of transitive applicative construction I have observed, the applicative argument is crossreferenced, not the object. In this respect, there is an asymmetry, in the applicative construction, between the object and the applicative argument, which is more like the object of the monotransitive construction. In all examples, however, the object is 3rd person. Further investigation should determine whether a transitive applicative construction can involve a 1st or 2nd person object and if, in such a case, the object would be cross-referenced. In fact, at least with the non applicative verb -č- ‘give’, crossreferencing in Laz is sensitive to the person hierarchy. The verb ‘give’ belongs to a small group of ditransitive verbs which are lexically specified as ditransitive and do not involve the applicative derivation (Lacroix 2007): -č- (without preverb) -č- (with preverb me- or mo-) -č- (with preverb gama-) -č- (with preverb ge-) -c’ir- -gur- -k’itx- -kun-
‘feed’ ‘give’ ‘sell’ ‘hit with’ ‘show’ ‘teach’ ‘ask’ ‘dress’ (‘put a garment on sb’)
As a rule, these verbs take only 3rd person inanimate objects. The verb ‘give’, however, can take an animate object, in which case it means ‘marry (a girl) to someone’. With this verb, cross-referencing is sensitive to the person hierarchy 1st > 2nd > 3rd. Of the theme and the recipient, the one which stands higher in the hierarchy is cross-referenced; the other is not. (20) shows that when the theme and the recipient are 1st and 3rd person, the verb cross-references only the 1st person, whatever its semantic role: recipient in (20a), theme in (20b). (21) shows that when the theme and the recipient are 2nd and 3rd person, the verb cross-references only the 2nd person, whatever its semantic role. Eventually, (22) shows that when the theme and the recipient are 1st and 2nd person, the verb cross-references only the 1st person, whatever its semantic role.
Benefactives in Laz 
With the verb ‘give’, the preverb is mo- when the recipient is 1st person and me- when it is 2nd or 3rd person. (20) a.
1st recipient > 3rd theme Ck’ar mo-m-č-i-ya! water pv-ii1-give-imp-quot ‘Give me some water!’
b.
1st theme > 3rd recipient Ma ha bere-s ko-me-m-č-i-a! 1s demd boy-dat pv-pv-ii1-give-imp-quot ‘Give me to this boy!’
(Ž>.22.1)
(21) a.
2nd recipient > 3rd theme Puci ko-me-k-č-are. cow pv-pv-ii2-give-fut.i1/2s ‘I will give you a cow.’
(Ž>.7.17)
b.
2nd theme > 3rd recipient Hemu-s me-k-č-are-a. demp-dat pv-ii2-give-fut.i1/2s-quot ‘I will give you to him.’
(22) a.
1st recipient > 2nd theme Baba-skani-k si ma va mo-m-č-ase. father-poss2s-erg 2s 1s neg pv-ii1-give-fut.i3s ‘Your father won’t give you to me.’
b.
1st theme > 2nd recipient Baba-k var me-m-č-am-s. father-erg neg pv-ii1-give-ths-i3s ‘My father won’t give me to you.’
(Dum37.102.4)
(Ž>.122.31)
(Dum37.61.1)
Note that the verb ‘give’ may take a beneficiary expressed by a postposition phrase headed by šeni ‘for’: (23) Xeyati-ši šey-epe šeni dido para me-č-am-s. charity-gen thing-pl for lots.of money pv-give-ths-i3s ‘He gives lots of money for the charities.’
(Dum67.LI.25)
5.2╅ Semantics In this section, I examine the different meanings of the applicative marked by the �vowels i/u-.
5.2.1â•… Benefactive/malefactive Van Valin & La Polla (1997: 383–384) distinguish between “plain beneficiary” (do sth to amuse/please sb), “deputative beneficiary” (do sth in sb’s place) and “recipient
René Lacroix
beneficiary” (perform some action with the result that sth is given to sb). The i/uapplicative may express all these types of beneficiaries: (24)
Plain beneficiary Hasteri biyapa-ti u-bir-am-s… such song-too ii3.appl-sing-ths-i3s ‘And she sings for him such a song…’
(25)
Deputative beneficiary Mo-m-č-i do ma do-g-i-naxv-a-ya. pv-ii1-give-imp and 1s pv-ii2-appl-wash-opt-quot ‘Give me (the linens), I will wash it for you.’
(Ž>.163.5)
(Dum67.VII.8)
(26) Recipient beneficiary Bozo-k xe-muši-te hentepe-s girl-erg hand-poss3s-instr demp.pl-dat
k’ahve d-u-gub-um-s. coffee pv-ii3.appl-boil-ths-i3s
‘The girl makes coffee for them with her own hands.’
(Dum37.114.9)
The i/u-applicative argument may also be a maleficiary: (27)
K’ui g-i-ntxo-es nek’na-s tudele. hole ii2-appl-dig-aor.i3p door-dat under ‘They have dug a hole under the door (for you to fall in it).’ (lit. ‘they have dug you a hole under the door’)
(K’art’72.156.1)
(28) K’oči-k dolokunu n-u-xir-u mola-ši oxor¦Šˇa-s. man-erg clothes pv-ii3.appl-steal-aor.i3s mullah-gen woman-dat ‘The man stole the clothes of the mullah’s wife.’ (Ž>.66.2)
5.2.2â•… Other meanings Laz has two types of ditransitives verbs. As we have seen above, there is a class of lexically specified, morphologically non-derived ditransitives (‘give’, ‘show’, ‘ask’, etc.). A second class of ditransitives is represented by i/u-applicatives. Recipients may be expressed by dative arguments of both types of verbs. Applicative verbs involving a recipient include -yon- ‘take someone to someone’ and -n¦Š>on- ‘send something to someone’ (29). These verbs cannot occur in a ditransitive construction without the operator i/u-. (29) Bič’i-k mack’indi-muši bozo-s ko-n-u-n¦Šˇ>on-u. boy-erg ring-poss3s girl-dat pv-pv-ii3.appl-send-aor.i3s ‘The boy sent the ring to the girl.’
(Ž>.51.35)
The beneficiary/recipient polysemy of applicative construction is widespread among languages, as noted by Peterson: “if a language has a construction which could be
Benefactives in Laz
characterized as an applicative, it is most common that the semantic role of the applicative object will be that of recipient and/or beneficiary/maleficiary” (Peterson 2007: 40). The applied argument of the i/u-applicative construction can bear a locative role, either allative (30) or stative (31); the locative applicative argument may be animate (32): (30) Hac’i hen umčane bič’i mcxuli-s k-el-u-xed-u. then superl eldest boy pear.tree-dat pv-pv-ii3.appl-sit-aor.i3s ‘Then the eldest son sat down near the pear tree.’ (Ž>.107.20) (31) Ham kyöi-s ar didi γal gel-ul-u-n demd village-dat one big river pv-pass-ths-i3s hac’ hemu-s goc’-u-x-e-n. now demp-dat pv-ii3.appl-sit-ths-i3s ‘In this village a big river flows; now he is sitting in front of it.’
(Dum37.90.13)
(32) Uča k’oči-ti hantepe-s k-el-u-xed-u. black man-too demp-dat pv-pv-ii3.appl-sit-aor.i3s ‘The man in black sat down near them [the father and his son].’ (Dum67.XIII.16)
The i/u-applicative is used in the “external possessor” construction (or “possessor raising” construction), where the possessor is expressed as the applied argument instead of appearing as a genitive NP: padišahi ‘sultan’ in (33), xo¦Ša ‘hodja’ in (34). In (34), the applicative argument xo¦Ša bears at the same time the role of maleficiary. (33) Ar padišahi-s ont’ule-s ar didi one sultan-dat field-dat a big
oškur u-dg-i-t’u-doren. apple.tree ii3.appl-stand-sth-impft.i3s-evd
‘A sultan had a big apple tree in his field.’
(34) Xo¦Šˇa-s mčxuri o-b-u-mčk’om-a-t-ya! hodja-dat sheep pv-i1-ii3.appl-eat-opt-pl-quot ‘Let’s eat the hodja’s sheep!’
(Dum37.78.3)
(Ž>.114.13)
This type of polysemy is attested elsewhere, as Polinsky remarks: “additional semantic functions that may be associated with the applied object include possessor” (Polinsky 2005). In some idiomatic expressions, the dative argument of a verb with operator i/ubears the semantic role of experiencer: (35) Oxor¦Šˇƒa-s guri ko-m-u-xt-u. woman-dat heart pv-pv-ii3.appl-come-aor.i3s ‘The woman got angry.’ (lit. ‘heart came to the woman’) (36) Bozo-s men¦Šˇel n-u-l-u-dort’un. girl-dat strength pv-ii3.appl-fall-aor.i3s-pprf ‘The girl had no more strength.’ (lit. ‘the girl’s strength fell’)
(Ž>.43.14)
(Dum37.28.1)
René Lacroix
Formally, such constructions must be kept apart from the i/u-applicative constructions we have seen so far since here, the dative argument triggers number agreement. Compare in this respect (35) and (37); the verb in (37) takes the plural suffix -es indicating the plurality of bozopes. This is an example of “inversion” (see note 2). (37) Bozo-pe-s gui m-u-xt-es. girl-pl-dat heart pv-ii3.appl-come-aor.pl ‘The girls got angry.’
In some cases, it is difficult to draw a clear boundary between the possessor raising construction and applicative constructions including an experiencer; consider (38): (38) Nana do baba do-m-i-γur-u-ya. mother and father pv-ii1-appl-die-aor.i3s-quot ‘My mother and my father died.’
(Ž>.52.15)
Compare (38) with (39), where the possessor appears as a genitive complement: (39) Baba-čkuni do-γur-u. father-poss1p pv-die-aor.i3s ‘Our father died.’
(Dum67.I.418)
The i/u-applicative is used in a number of other verbs, among which are the following: c’v g Š¦ ox u¦Š tk’ob
‘tell sb sth’ ‘resemble sb’ ‘call sb’ ‘listen to sb’ ‘hide from sb’
č’and xezmet’ p’aramit γarγal
‘invite sb’ ‘serve sb’ ‘talk to sb’ ‘scold sb’
The operator i/u- may be lexicalized, as in -škv- ‘swallow’, a verb which takes an object but no applicative argument: (40) Mgey-epe-k puci-š xorci k’ala wolf-pl-erg cow-gen meat with
bere-ti gy-u-škv-es. boy-too pv-ii3.appl-swallow-aor.i3p
‘The wolves swallowed the cow’s meat and the boy.’
(Dum67.II.96)
The beneficiary, although in most cases expressed as an applicative argument, may be alternatively instantiated as an oblique, as we will see below (Section 6). The i/u- applicative construction is thus optional with certain verbs (Peterson 2007: 45). There are many cases, however, where the applied argument cannot be alternatively instantiated as an oblique. This holds for verbs such as -c’v- ‘tell sb sth’, -¦Šox- ‘call sb’, -yon- ‘take sb
Benefactives in Laz
to sb’ or -xt- ‘go’. Sentence (41), for instance, cannot be replaced by (42), where oxor¦Ša ‘woman’ is expressed as a dative oblique not cross-referenced in the verb: (41) Oxor¦Ša-s guri ko-m-u-xt-u. woman-dat heart pv-pv-ii3.appl-come-aor.i3s ‘The woman got angry.’ (lit. ‘heart came to the woman’)
(Ž>.43.14)
(42) *OxorŠˇa-s guri ko-mo-xt-u. woman-dat heart pv-pv-come-aor.i3s ‘The woman got angry.’
The operator i/u- is of ancient origin. It has cognates in the other languages of the family and is reconstructed for the proto-language (Fähnrich 2007: 209, 416).
6.â•… The postposition šeni In addition to appearing as an applied argument, the beneficiary may be expressed as an adjunct with the postposition šeni. This postposition governs the absolutive with nouns (44), the genitive with 1st and 2nd person pronouns (45) and the oblique stem with demonstrative pronouns (which serve as 3rd person pronouns; Example 46). It covers the same benefactive meanings as the i/u-applicative: (43) Recipient beneficiary Ore ko-b-Šir-i, a žur o-p-čk’om-i, cabbage pv-i1-see-aor one two pv-i1-eat-aor ar čkva arkadaši šeni k-e-b-zd-i. one more friend for pv-pv-i1-take-aor ‘I saw some cabbages, ate one or two and took another for my friend.’ (Dum67.LIV.16) (44) Plain beneficiary Ma-ya eči c’ana-s-ya k’oči 1s-quot twenty year-dat-quot man
šeni b-i-čališ-am-t’i-a. for i1-mid-work-ths-impft-quot
‘I work for this man for twenty years.’
(45)
Deputative beneficiary Si čkimi šeni meda-xt-i! 2s 1s.gen for pv-go-imp (I don’t want to go downtown.) ‘Go there for me!’
(Ž>.126.14)
René Lacroix
The šeni phrase is not cross-referenced on the verb. Since final vowels may be deleted, šeni may appear as šen. The postposition šeni has a range of other meanings, viz. cause (46–47), purpose (48), ‘about, concerning’ (49) and ‘in exchange for’ (50): (46) Hemu šeni pimpili do-m-i-bγ-es demp for beard pv-ii1-appl-shave-aor.i3p k’ibir-epe ko-do-m-i-ck’-es. tooth-pl pv-pv-ii1-appl-tear.off-aor.i3p ‘For this reason (lit. ‘for that’), they shaved my beard and tore my teeth off.’ (Q’ip.35.32)
Šeni may take a clausal complement, which takes the enclitic subordinator na generally attached to the word immediately preceding the verb (sentences (18), (19), (54a) and (56) illustrate this subordinator). As with nominal complements (46), the postposition phrase indicates a cause. It seems that šeni can be also used to express the maleficiary€(61). (47) Arslan-epe k’ala-na i-rd-u šen lion-pl with-sub mid-grow-aor.i3s for arslani-š skiri ko-gy-o-d-u-doren. lion-gen son pv-pv-tr-name-aor.i3s-evd ‘Since he had grown up with lions, he named him “Lion’s son”.’
(Dum67.I.101)
(48) Mtuti mkveri-š o-čk’om-u šen bear flour-gen pv-eat-vn for karmat’e-s ama-xt-u-dort’un. mill-dat pv-go.in-aor.i3s-pprf ‘The bear had come into the mill to eat the flour.’ (49) Hac’i bič’i šeni p-č’ar-a-t! now boy for i1-write-opt-pl ‘Now let’s write about the boy!’
(Dum37.21.9)
(K’art’72.167.37)
(50) Ažliya-s me-p-č-am-t ar vedre ck’ari šeni. dragon-dat pv-i1-give-ths-pl one bucket water for ‘We give him to the dragon in exchange for one bucket of water.’ (K’art’72.177.18)
The postposition šeni is transparently related to the ablative case -šen, which, among other things, expresses spatial and temporal origin: (51) Du>uni-skani šeni P’oli-šen kart’ali mo-m-i-č’ar-i. wedding-poss2s for Istanbul-from letter pv-ii1-appl-write-imp ‘Write a letter to me from Istanbul when you get married.’ (Q’ip.27.8)
Benefactives in Laz
(52) Ham ora-šen ar c’ana mek’-il-u-dort’un. demd time-from one year pv-pass-aor.i3s-pprf ‘Since that time, one year had passed.’
(Dum37.41.9)
In almost all Laz texts published so far, the postposition šeni is not written separately from its complement. Since final /i/ may drop, there is often nothing in the orthography that distinguishes the postposition šeni from the ablative case -šen in these texts: what is written oxoišen can represent the ablative case (oxoi-šen ‘house-abl’) or the postposition (oxoi šen ‘house for’). Asking informants to repeat such examples provides a way to disambiguate, since the suprasegmental properties of an NP with the ablative case and an NP with šeni differ. In nominals, the stress falls on the penultimate syllable, as exemplified below with oxoi ‘house’, and the postposition does not affect stress placement in the noun. There is, thus, a difference of stress assignment in oxoíšen ‘house-abl’ and oxói šén ‘house for’. absolutive ergative dative genitive allative ablative instrumental
singular oxói oxói-k oxói-s oxoí-ši oxoí-ša oxoí-šen oxoí-ten
plural oxo-épe oxo-épe-k oxo-épe-s oxo-epé-ši oxo-epé-ša oxo-epé-šen oxo-epé-ten
7.â•… The choice between the applicative derivation and the postposition 7.1â•… Free variation We have seen so far that beneficiaries and maleficiaries can be expressed either as applied arguments in an applicative construction or as postpositional adjuncts. With certain verbs and in certain syntactic positions, either the postposition or the applicative can be used: (53) a.
Baba-k čkimi šeni dol-o-x-u-ya. father-erg 1s for pv-tr-bury-aor.i3s-quot ‘Our father buried it for me.’
b. Baba-k ma para dolo-m-i-x-u. father-erg 1s money pv-ii1-appl-bury-aor.i3s ‘My father buried money for me.’
(Dum67.XX.47)
René Lacroix
(54) a.
Bozo-muši šeni ntxiyi-ši xeč’ep’a-s-na girl-poss3 for walnut-gen shell-dat-sub
meš-i-nt’r-asen porča gor-um-s. pv-mid-be.contained.in-fut.i3s dress look.for-ths-i3s
‘He looks for a dress which could be contained in the shell of a walnut for his daughter.’ (Dum67.V.61)
b. Bozo-muši-s porča u-go-um-s. girl-poss3-dat dress ii3.appl-look.for-ths-i3s ‘He looks for a dress for his daughter.’
For both the postposition strategy (53a) and the applicative strategy (53b), my informant accepts a recipient beneficiary reading and a deputative beneficiary reading. More work has to be done in order to determine exactly what conditions the choice between the two constructions here, and in what extent topicality plays a role (Peterson 2007: chap. 4). There are contexts, however, where only the postposition can be used. These are examined below.
7.2â•… Restrictions on the use of the applicative construction First, there are syntactic contexts where the applicative construction cannot be used to express the beneficiary. The verbal noun and the participles (i.e. non-finite verb forms) do not contain any slot for valency operators and cross-referencing affixes. Hence, they do not allow the applicative construction. This explains the occurrence of the postposition in sentences such as the following: (55) Skani šen e-xtim-u-s b-i-čališ-ar. 2s for pv-go.up-vn-dat i1-mid-work-fut.i1/2s ‘I will try to go up for you.’
(Dum37.101.13)
Here, the 2nd person singular beneficiary, which depends on the verbal noun extimu, cannot be expressed as an applicative argument.7 In (56), the occurrence of the postposition may be explained by the fact that one of the beneficiaries is 3rd person while the other is the 1st person subject. In the case of a subject
7.â•… In this sentence, postpositional marking could be explained by saying that skani is a demoted argument in the sense of relational grammar (cf. Harris 1985). However, the beneficiary in this construction can be marked by the dative as well:
Nana-čkimi-s pukui me-č-am-u minon. mother-poss1s-dat flower pv-give-sfx-vn I.want ‘I want to give flowers to my mother.’
Benefactives in Laz
doing something for its own benefit (‘X buys sth for oneself’), the verb does not include the applicative marker, but the operator i- marking middle voice, as illustrated in (57).8 (56) Na-u-γ-u-t’u altuni-ši muši šeni do sub-ii3.appl-have-ths-impft.i3s gold-gen emph/refl for and oxorŠˇa-muši šeni k’ay šey-epe k-e-č’op-u. woman-poss3s for good thing-pl pv-pv-take-aor.i3s ‘With the gold he had, he bought good things for himself and his wife.’ (Dum67.XIV.12) (57) Mčk’udi e-b-i-č’op-ae. bread pv-i1-mid-take-fut.i1/2s ‘I will buy some bread for myself.’
In (56), then, a conflict arises, since the verb cannot contain more than one valency operator, i.e. it cannot contain simultaneously the middle voice operator i- indicating that one of the beneficiaries is the subject, and the applicative operator u- indicating that the other beneficiary is 3rd person. The conflict is resolved by using the postposition, so that no valency operator has to be expressed on the verb. Secondly, there are lexical restrictions on the use of the applicative. The verb zd ‘take’ in the applicative derivation has a maleficiary reading: ‘take sth from sb (without him being aware of it)’ (58a). To express the benefactive counterpart (‘take sth for sb’), the postposition šeni has to be used (58b). (58) a.
Šˇuma-čkimi-s ore k-e-b-u-zd-i. brother-poss1s-dat cabbage pv-pv-i1-ii3.appl-take-aor ‘I took my brother’s cabbage (and he is not aware of it).’ *‘I took a cabbage for my brother.’
b. Šˇuma-čkimi šeni ore k-e-b-zd-i. brother-poss1s for cabbage pv-pv-i1-take-aor ‘I took a cabbage for my brother.’
By contrast, with the verb č’op ‘take, buy’ (already illustrated in [57]) the applicative derivation has both the readings ‘take/buy from sb (maleficiary)’ and ‘take/buy for sb (beneficiary)’: (59) a.
Hem šey-epe-čkuni xanŠˇi-s ko-y-u-č’op-i. demd thing-pl-poss1p khan’s.boss-dat pv-pv-ii3.appl-take-imp ‘Take these things of ours from the khan’s boss.’ (Dum67.XII.129)
b. Didi bozo-pe-s šey-epe ko-y-u-č’op-u. older girl-pl-dat thing-pl pv-pv-ii3.appl-take-aor.i3s ‘He bought the things for the older girls.’ (Dum67.IV.13)
8.â•… Recall that the operator i- marking middle voice is different from the applicative operator i-, which indicates that the applicative argument is 1st or 2nd person (see Section 5.1).
René Lacroix
7.3â•… Co-occurrence of the postposition and the applicative Finally, there are some occurrences of sentences in which both the applicative and the postposition may be used simultaneously:9 (60) Padišahi-k bere-muši šen sultan-erg son-poss3s for hac’i šendon ar t’axti d-u-xen-ap-u-dort’un. then from one throne pv-ii3.appl-make-caus-aor.i3s-pprf ‘Since that time, the sultan had had a throne done for his son.’
(Dum37.44.10)
Sentence (61) is another example of a construction involving simultaneously the postposition and the applicative derivation. Here, the postpositional NP has the semantic role of maleficiary. (61) Mtugi šeni-ti kapani ko-gy-u-dg-u. mouse for-too trap pv-pv-ii3.appl-lay-aor.i3s ‘And she set a trap to catch the mice.’
(Dum67.VIII.39)
8.â•… Conclusion We have seen that, in Laz, the beneficiary can be expressed as the applicative argument in an applicative construction marked on the verb by the operator i/u- (Section 5), or as a šeni postposition phrase (Section 6). The former has the syntactic status of a core argument; the latter is an oblique. Certain applicative arguments may alternatively be instantiated as obliques; others cannot (Section 7). Both the applicative construction and the postposition have a range of other meanings. If not lexicalized, the i/u-applicative argument indicates beneficiary/ maleficiary, recipient, locative/allative and possessor. In some idiomatic expressions, “applicative” arguments have the role of experiencer; they must be distinguished from genuine applicative arguments in that they trigger number agreement. The postposition šeni indicates, among other things, beneficiary/maleficiary, cause and purpose. Although both belong to very common types of polysemy, the polysemy of the applicative construction and that of šeni are not identical. This suggests that
9.â•… Here again, postpositional marking could be explained by saying that bere-muši is a demoted argument in the sense of Relational Grammar. However, it is cross-referenced on the verb by the operator u- ‘ii3.appl’ – a property of core arguments. Furthermore, the benefactive can also be marked by the dative in this construction:
Padishai-k bere-s d-u-xen-ap-u altuni-shi eyeri. sultan-erg child-dat pv-ii3.appl-do-caus-aor.i3s gold-gen saddle ‘The sultan had a saddle done for his son.’ (ŽΓ.121.36)
Benefactives in Laz 
they developed from different source meanings, and that the cases of overlapping result from convergence. However, the lack of conclusive comparative data makes it difficult to put forward a hypothesis about the starting point of the evolutions that resulted in the competition between the applicative construction and šeni for expressing beneficiary/maleficiary.
Abbreviations abs aor appl caus cr dat demd demp emph/refl erg evd fut gen imp impft instr int mid neg
absolutive aorist applicative causative cross-referencing dative demonstrative determiner demonstrative pronoun emphatic/reflexive pronoun ergative evidential future genitive imperative imperfect instrumental interrogative middle negation
opt pl, p poss pprf pv quot s sfx sub superl ths tr vn i ii 1 2 3
optative plural possessive pluperfect preverb quotative singular suffix subordinator superlative thematic suffix transitive verbal noun Set I cross-referencing affix Set II cross-referencing affix 1st person 2nd person 3rd person
For Dumézil 1937, Ž>ent’i 1938 and K’art’ozia 1993, the references of the examples give the page and the line of the Laz text, so that Dum37.10.3 means Dumézil 1937, page€10, line 3. For Dumézil 1967, the references give the text number and the line number provided by Dumézil, so that Dum67.XXIII.8 means Dumézil 1967, text XXIII, line€8.
References Alsina, Alex & Mchombo, Sam A. 1990. The syntax of applicatives in Chichewa: Problems for a Theta Theoretic asymmetry. Natural Language and Linguistic Theory 8: 493–506. Alsina, Alex & Mchombo, Sam. 1993. Object asymmetries and the Chichewa applicative construction. In Theoretical Aspects of Bantu Grammar, Sam Mchombo (ed.), 17–45. Stanford CA: CSLI.
René Lacroix Andrews, Peter A. (ed.). 1989. Ethnic Groups in the Republic of Turkey. Compiled and edited with the assistance of Rüdiger Benninghaus. Wiesbaden: Reichert. Austin, Peter K. 2005. Causatives and applicative constructions in Australian aboriginal languages. <www.hrelp.org/aboutus/staff/peter_austin/AustinCausatives.pdf>. Authier, Gilles & Haude, Katharina (eds.). Forthcoming. Ergativity and Voice. Berlin: Mouton de Gruyter. Baker, Mark. 1988. Theta Theory and the syntax of applicatives in Chichewa. Natural Language and Linguistic Theory 6: 353–389. Boeder, Winfried. 2005. The South Caucasian languages. Lingua 115(1–2): 5–89. Čikobava, Arnold. 1936. Č’anuris gramat’ik’uli analizi t’ekst’ebiturt (A grammatical analysis of Laz, with texts). Tbilisi: Mecniereba. Dixon, Robert M.W. & Aikhenvald, Alexandra Y. (eds.). 2000. Changing Valency: Case Studies in Transitivity. Cambridge: CUP. Donohue, Mark. 1999. A Grammar of Tukang Besi. Berlin: Mouton de Gruyter. Dumézil, Georges. 1937. Contes lazes. Paris: Travaux et mémoires de l’Institut d’Ethnologie, XXVII. Dumézil, Georges. 1967. Documents anatoliens sur les langues et les traditions du Caucase, I:. Récits lazes (dialecte d’Arhavi). Paris: Presses Universitaires de France. Dumézil, Georges. 1972. Textes en laze d’Ardeşen. Bedi Kartlisa, Vol. XXIX-XXX. Fähnrich, Heinz. 2007. Kartwelisches Etymologisches Wörterbuch. Leiden: Brill. Harris, Alice C. 1985. Diachronic Syntax: The Kartvelian Case [Syntax and Semantics 18]. New York NY: Academic Press. Harris, Alice C. (ed.). 1991. The Indigenous Languages of the Caucasus, Vol. 1: Kartvelian Languages. Denver CO: Academic Resources Corporation. Haspelmath Martin, Dryer, Matthew S., Gil, David & Comrie, Bernard (eds.). 2005. The World Atlas of Language Structures. Oxford: OUP. Holisky, Dee Ann. 1991. Laz. In Harris (ed.). Kemmer, Suzanne. 1993. The Middle Voice [Typological Studies in Language 23]. John Benjamins. K’art’ozia, Guram. 1972. Lazuri t’ekst’ebi (Laz texts). Tbilisi: SSSR MAG. K’art’ozia, Guram. 1993. Lazuri t’ekst’ebi (Laz texts). Tbilisi: Mecniereba. Kutscher, Silvia. 2001. Nomen und nominales Syntagma im Lasischen. Eine deskriptive Analyse des Dialekts von Ardeşen [Lincom Studies in Caucasian Linguistics 17]. München: Lincom Europa. Lacroix, René. 2007. Ditransitive Constructions in Laz. Communication at Conference on Ditransitive Constructions, Max Planck Institute for Evolutionary Anthropology, Leipzig, Germany, 23–25 November. Lacroix, René. 2008. The origin of Set II plural cross-referencing suffixes in the South Caucasian languages. Communication at Décembrettes 6, International Morphology Conference, Université Bordeaux 3, Bordeaux, France, 4–5 December. Lacroix, René. Forthcoming. Laz Middle Voice. In Ergativity and Voice, Gilles Authier & Katharina Haude (eds.) Marr, Nicolas. 1910. Grammatika čanskago (lazskago) jazyka (Grammar of Čan (Laz)). Materialy po jafetičeskomu jazykoznaniju 2. St. Petersburg: Akademija. Čikobava, Arnold. 1936. Č’anuris gramat’ik’uli analizi (t’ekst’ebiturt) (A grammatical analysis of Laz (with texts)). Tbilisi: Sakartvelos SSR Mecnierebata Ak’ademiis Gamomcemloba. Peterson, David A. 2007. Applicative Constructions. Oxford: OUP. Polinsky, Maria. 2005. Applicative constructions. In The World Atlas of Language Structures, Martin Haspelmath, Mathhew Dryer, David Gil & Bernard Comrie (eds.). Oxford: OUP.
Benefactives in Laz
Q’ipšiŠe, Ioseb. 1939. Č’anuri t’ekst’ebi (Laz texts). Tbilisi: SSRK’ Mecnierebata Ak’ademiis Pilialis Gamomcemloba. Tschenkéli, Kita. 1958. Einführung in die georgische Sprache, Band I: theoretischer Teil. Zürich: Amirani Verlag. Tuite, Kevin. 1998. Kartvelian Morphosyntax: Number agreement and morphosyntactic orientation in South Caucasian Languages. Studies in Caucasian Linguistics, vol. 12. München: Lincom Europa. Van Valin, Robert D., Jr. & LaPolla, Randy J. 1997. Syntax. Structure, Meaning and Function. Cambridge: CUP. Ž>ent’i, Sergi. 1938. Č’anuri t’ekst’ebi (arkabuli k’ilok’avi) (Laz texts (dialect of Arhavi)). Tbilisi: SSRK’ Mecnierebata Ak’ademiis Pilialis Gamomcemloba.
Benefactive and malefactive verb extensions in the Koalib verb system* Nicolas Quint In Koalib (Niger-Congo, Kordofanian branch, Heiban family), a basic verb can give rise to several verb extensions (VE), among which a benefactive and a malefactive. Section one deals with the main derivational characteristics (affixation, tonology, productivity, valency) of both VE. Section two proposes a more in-depth scrutiny of the morphology of those extensions in Koalib, examining various particular cases (lack of basic verb, double derivation, irregularities). Section three focuses on the semantics of Koalib bene-/malefactive VE, putting special emphasis on the semantic roles encoded by each extension, and on the diverse nuances of meaning they can convey. The conclusion lays the stress on the interest of Koalib (and more generally Kordofanian languages) data for typological studies of the notions of benefaction and malefaction in human language.
0.â•… Introduction Koalib is a Niger-Congo language (Kordofanian branch, Heiban family) spoken by approximately 100,000 people who live in or come from the North-Eastern part of the Nuba Mountains (in the South-Kordofan Province of the Republic of Sudan). The Koalib area comprises the towns of Delami, Abri, Umm Berembeita and Umm Heitan. Koalib itself is divided in several dialects, some of which differ considerably from each other (Quint 2006: 23–25). All the examples presented in this paper are taken from ]èrε´εqε¡ (anglicised under the form rere, see Schadeberg 1981: 11, 13), the Koalib variety traditionally spoken in and around Abri (at the very center of the Koalib area) and the main basis of written modern Koalib. In Koalib, a basic (or simple) verb (henceforth: BV) can give rise to several derived forms, most of which are produced (as in many Niger-Congo languages, see Hyman
*I am most indebted to Seppo Kittilä and Fernando Zúñiga for their useful comments on earlier drafts of this paper. I also want to thank here Marie Lerat and Bart Jacobs, for their efficient bibliographic collaboration and last but not least, Siddig Ali Karmal Kokko, my main informant for Koalib, whose unvaluable help has enabled me to understand many of the structures of his mother tongue.
Nicolas Quint
2007; Williamson & Blench 2000: 12–13) using suffixed verb extensions, e.g. a benefactive (BEN) and a malefactive (MAL), as can be seen in the following examples: (1)
BV nyìimí1 ‘steal’ kwé-nyìimí ]èeqó clf-steal.ipfv.3sg goat.o ‘He will steal a goat.’
(2)
Benefactive nyíim!¡ccí ‘steal for (= to the advantage of)’ kwé-nyHˇim!¡ccí Kwókkò]wó ]èeqó clf-steal.ben.ipfv.3sg Kwókkò.o goat.o ‘He will steal a goat for Kwókkò (= to Kwókkò’s advantage, e.g. to help Kwókkò pay his dowry).’
(3)
Malefactive nyíim!¡t!¡ ‘steal from (= at the expense of)’ kwé-nyHˇim!¡t!¡ Kwókkò]wó ]èeqó clf-steal.mal.ipfv.3sg Kwókkò.o goat.o ‘He will steal a goat from Kwókkò (= at Kwókkò’s expense).’
This article aims at giving a detailed account of the main characteristics of the benefactive and malefactive verb extensions in contemporary Koalib. In Section 1, I will focus mainly on the general derivational morphosyntactic characteristics of the benefactive and malefactive verb extensions in Koalib. In Section 2, I will examine the morphology of these extensions in some detail. In Section 3, I will address the diversity of semantic values encoded by each of the two verb extensions at stake. Finally, I will stress the issue of why Koalib (and Kordofanian languages) is of special interest for the study of bene-/malefaction.
1.â•… A morphosyntactic account of the Koalib benefactive and malefactive 1.1â•… Morphological characteristics: Standard derivational processes The segmental and tonal devices that allow for the production of a bene-/malefactive verb extension2 are summarized in Table 1:
. Most of the Koalib examples presented herein are transcribed using the phonological spelling I have devised for ŋèrέεqὲ (for more details, see Quint 2006: 208–210). Note in particular that the Koalib letter ¸t always stands for a voiceless retroflex stop [z]. 2.â•… For the sake of clarity, within the scope of this article (except for a few cases discussed in 3.3.4.), I have decided to focus on the analysis of one of the three main aspect-movement forms of the Koalib verb, namely the Centrifugal Imperfective (CFI), which expresses the fact that an action is not completely performed yet when the utterance takes place (Imperfective) and the realisation of this action is conceived as moving away from the spatial origin
Benefactives and malefactives in Koalib
Table 1.╇ Morphological characteristics of bene-/malefactive verb extensions in Koalib Verb extension
Suffix
Tone melody
Benefactive Malefactive
-/(V)ccÉ/ -/(V)tÀ/
H-(L)n-H H-(L)n-L
A = /a, !/ E = /ε, e, i/3
(4)
H = high tone L = low tone
n≥1
Benefactives3 a. BV nyε¡]lε´ ‘taste’ > BEN nyé]làccé ‘taste (something) for s.o.’ b. BV ¡ffpε´ ‘chop (wood)’ > BEN ´ffpèccέ ‘chop (wood) for s.o.’ c. BV ù!´ ‘milk (a cow…)’ > BEN ú!¡ccí ‘milk (a cow) for s.o.’
(5) Malefactives: a. BV èmné ‘eat something (mainly sorghum porridge)’ > MAL émnàtà ‘eat unduly something (mainly sorghum porridge) belonging to s.o.’ b. BV pùutí ‘be hot/ warm’ > MAL púut!¡t!¡ ‘be too hot for s.o.’
In casual speech the -/(V)tà/ CFI malefactive suffix is quite frequently reduced to -/à/:
(6) BV ùuqí, ‘cut something’ > MAL úuq!¡t!¡, or úuq!¡ ‘cut something (body part) to s.o.’
1.2â•… Comparison of the two verb extensions 1.2.1â•… Tone melody Benefactive and malefactive are the only Koalib verb extensions associated with a H(L)n-tone melody (see Table 1), while all other extensions are characterised by a (L)n-H melody.
considered in the speech situation (Centrifugal). The reasons for this restriction are the following: (a) morphologically, the CFI seems to be the basic form of the Koalib verb from which the two other aspect-movement forms (namely Centripetal Imperfective and Perfective) can be shown to be derived (mainly through suffixation); (b) the typical tone melodies associated to benefactive and malefactive in the CFI are levelled in the remaining forms, probably due to morphological interactions with the aspect-movement makers, and (c) my description of Koalib verb morphology (see Quint forthcoming) is not completed yet and I€prefer to limit my analyses to forms that I am best able to explain in morphological terms, which is the case of the CFI. 3.â•… For both A and E, the choice of the vowel in the suffix depends on vowel harmony (for more details about harmony rules in Koalib, see Quint 2006: 34–42).
Nicolas Quint
(7) Passive: ùunùnní (LLH) ‘be harvested’ < BV òoné ‘harvest’ (8) Reciprocal: àmqàtècé (LLLH) ‘love each other’ < BV àmqé ‘love’
This commonality between the benefactive and the malefactive is fundamental, for it shows that the semantic complementarity (for s.o.’s benefit/ at s.o.’s expense) of the two verb extensions at stake is concretely reflected in the morphology of the language. Such fact justifies a parallel treatment of the benefactive and malefactive throughout this article.
1.2.2╅ Productivity Both benefactive and malefactive verb extensions are widely attested in contemporary Koalib and seem to be synchronically productive. However, as shown in Table 2, the benefactive has a higher frequency than malefactive in the Koalib vocabulary:4 the benefactive/malefactive ratio is higher than 2 (449/200 = 2,3), which seems to �prove that the benefactive verb extension is significantly more productive than the malefactive one. Table 2.╇ Relative frequency of bene-/malefactive verb extensions in my Koalib lexical database Number of items Basic verbs Benefactives Malefactives Total verbs Total items in database
360 449 200 1970 5113
In the case of the benefactive, I have found a handful of cases of benefactive verb extensions associated with verb roots borrowed from Sudanese Arabic (SA), the dominant language in Northern Sudan and in particular in South Kordofan, the place where most Koalibs live or come from: (9) SA /‘jagli/ ‘he roasts/ fries’ > BV ¡!klí ‘roast (coffee)’ > BEN !´klìccí ‘roast (coffee) for s.o.’
I have no similar example for malefactive but, as the integration of Arabic verb roots into the Koalib verb system is quite exceptional (probably due to the huge �morphological differences between both languages), this fact does not allow us to draw
4.â•… The content of my own Koalib lexical database is based on 16 months of intensive fieldwork among Koalib native speakers and the exhaustive scrutiny of more than 900 pages of texts (folktales, schoolbooks, Bible translations, etc.) written in this language. Given the amount of linguistic material used and the time of exposure to spoken Koalib, the results presented in Table 2 can be held to be statistically relevant with respect to the reality of the language.
Benefactives and malefactives in Koalib
any positive conclusion about the relative productivity of the malefactive and the benefactive in Koalib.
1.2.3â•… Valency As a general rule, both benefactive and malefactive trigger an increase of one unit of the number of arguments governed by the verb, as can be seen in the Examples (1) to (3) of Section 0, repeated and renumbered below for convenience: (10)
BV nyìimí ‘steal’ kwé-nyìimí ]èeqó clf-steal.ipfv.3sg goat.o ‘He will steal a goat.’
(11)
> BEN nyíim!¡ccí ‘steal for (= to the advantage of)’ kwé-nyHˇim!¡ccí Kwókkò]wó ]èeqó clf-steal.ben.ipfv.3sg Kwókkò.o goat.o ‘He will steal a goat for Kwókkò (= to Kwókkò’s advantage).’
(12)
> MAL nyíim!¡t!¡ ‘steal from (= at the expense of)’ kwé-nyHˇim!¡t!¡ Kwókkò]wó ]èeqó clf-steal.mal.ipfv.3sg Kwókkò.o goat.o ‘He will steal a goat from Kwókkò (= at Kwókkò’s expense).’
In this canonic case, the basic verb nyìimí ‘steal’ is transitive and takes one object (accusative patient, here ]èeqó ‘goat.o’). Both benefactive and malefactive verb extensions of nyìimí are ditransitive and take one more object (bene-/maleficiary; here: Kwókkò]wó ‘Kwókkò.O’) inflected in the same case as the accusative patient and regularly placed between the verb and the patient. However, as can be seen in Table 3, ditransitive constructions are almost the norm (73%) for the benefactives while their occurence is almost three times lower for the malefactives (28%). Therefore, in terms of valency, both extensions behave in a significantly different way, perhaps due to the fact that the very semantic notions of benefaction and malefaction are not equally compatible with two objects – namely patient + bene-/maleficiary.5
5.â•… The connection between the degree of ditransitivity and the semantic content of the notions of benefaction and malefaction has been suggested to me both by Denis Â�Creissels and Daniel Hole (2007-10-26: p. c.). The existence of such a link seems to be implicitly confirmed by the few sources I have been able to find that deal (often marginally) with this question: Amberber (2000: 321–325), while discussing Amharic, a language which, like Koalib, possesses both a benefactive and a malefactive verb extension, insists on the fact that “unaccusative” (i.e. intransitive) [basic] verbs can readily produce malefactive extensions [and appaÂ� rently no parallel benefactives], such as ‘become night’ > ‘become night at s.o.’s disadvantage’
 Nicolas Quint
Table 3.╇ Relative frequency of ditransitive items in Koalib basic verbs, as well as benefactive and malefactive verb extensions Total number Basic verbs Benefactives Malefactives
360 449 200
Nb. ditrans. 1 329 55
% dit. 0% 73% 28%
Nb. ditrans. = number of ditransitive items (= any verb which regularly takes two objects in at least one acception) % dit. = percentage of ditransitive items in relation to the total number of items for each verb type.
As shown in Table 4, both benefactive and malefactive extensions can be produced from either intransitive or transitive basic verbs. Nevertheless, one can observe a certain contrast between the distributional patterns of each extension according to the valency of the basic verb (BV). Benefactives are comparatively more frequently produced from transitive BV (79%) than from intransitive BV (66%), which corroborates the general trend noticed by Shibatani (1996: 160). Malefactives, on the other hand, are comparatively more frequently produced from intransitive BV (45%) than from transitive BV (37%). Therefore, if we assume (see (10), (11), (12) above) that both benefactive and malefactive VE generally trigger an increase of one unit in the scale of transitivity, Table 4 tends to show that, quite predictably, (a) as benefactive VE are preferably derived from transitive basic verbs, ditransitive (= transitive plus one argument) benefactives are statistically favoured, and (b) as malefactive VE are preferably derived from intransitive basic verbs, transitive (= intransitive plus one argument) malefactives are statistically favoured. Thus, the valency of the basic verb is one of the factors accounting for the fact that benefactives are much more likely to be ditransitive than malefactives (see Table 3). At any rate, it should be noted that valency of the basic verb is clearly not the only cause of the strong difference of relative frequency between ditransitive benefactive and male� factive verb extensions, given that (i) the percentage differentials reflecting the effect of
(p. 323). Therefore, it seems that in some instances Amharic favours the production of transitive (i.e. non ditransitive) malefactive verb extensions, which may account for lower percentages of ditransitive malefactives in Amharic in relation to what can be observed for benefactives. For Kinyarwanda, Lemaréchal (1998: 212) signals that, “with verbs meaning ‘steal’ or ‘take (from)’, which presuppose that the stolen object is one of the actants’ possess, it is useless to resort to an applicative derivation” (my translation), from which we can infer that, in Kinyarwanda, applicative forms are disfavoured for some verbs typically associated with the notion of malefecation, such as ‘steal’.
Benefactives and malefactives in Koalib 
the valency of the basic verb on the productivity of bene-/malefactive are lower than 20% (16% and 18% respectively, see TIPD in Table 4), and (ii) the percentage differential reflecting the relative frequency of ditransitivity according to the bene-/malefactive nature of the verbal extension is 62% (i.e. [73–28] / 73), see % dit. in Table 3). Table 4.╇ Respective frequency of malefactive and benefactive verb extensions in Koalib according to the valency of the basic verb6 Valency of the Basic verb Intransitive Basic verbs Number of items Total Basic verbs from which: ≥ 1 Benefactive VE ≥ 1 Malefactive VE
%I
Transitive Basic verbs Number of items
%T
TIPD
149
100%
146
100%
0%
98 67
66% 45%
115 54
79% 37%
16% 18%
≥ 1 Benefactive VE: basic verbs for which at least one benefactive verb extension is attested in my database. / ≥ 1 Malefactive VE: basic verbs for which at least one malefactive verb extension is attested in my database. / %I = percentage of intransitive basic verbs for which at least one item of the verb extension at stake is attested. / %T = percentage of transitive basic verbs for which at least one item of the verb extension at stake is attested. / TIPD: Transitive - Intransitive Percentage Differential = |%T-%I| / the biggest of the two values (%T or %I), i.e. TIPD = [79–66] / 79 for benefactives and [37–45] / 45 for malefactives.
2.â•…Towards a better understanding of the morphology of bene-/malefactives in Koalib 2.1â•… Phenomena attested both in the benefactives and the malefactives 2.1.1â•… Lack of the basic verb In some cases (approximately 10 benefactives and 25 malefactives in my database), the basic verb is missing in present-day Koalib and we cannot but speculate
6.╅ In Table 4, the labels intransitive and transitive respectively apply to exclusively intransitive items (an intransitive construction is the only option for all recorded meanings) and exclusively transitive items (a transitive construction is the only option for all recorded meanings). For the sake of clarity, the minority of basic verbs that can alternatively be intransitive or transitive according to their meaning have not been taken into consideration. The numbers and percen� tages presented in Table 4 are based on a somewhat more recent version (2009-03-19, containing 5620 lexical items) of my Koalib lexical database than the one used for the rest of this article.
 Nicolas Quint
about the original root from which the attested verb extensions are presumably derived: (13) BEN k!¥qnycí ‘frighten/cheat s.o.’ < BV *kAqn-, *kAqny (14) MAL kwí1ny!¡t!¡ ‘shell (the sesame)’ < BV *kwE1ny-
Such cases show that the derivational mechanisms which allow for the production of the benefactives and the malefactives are quite old, as it must have taken a reasonable time span for the system first to produce bene-/malefactive extensions from BV and then to lose the latter (probably through gradual replacement by other lexemes with similar meanings).
2.1.2â•… Double derivation A bene-/malefactive verb can be produced from another verb extension (95 benefactives and 15 malefactives in my database). (15) BEN ´!rrìnnìccí ‘happen to s.o.’ < passive !`rrìnní ‘happen (passive form = be done)’ < BV !`rrí ‘do’ (16) BEN rúucìccí ‘turn something into something else at s.o.’s profit’7 < transitive rùucí ‘change something into something else’ < BV ´ffrf¡ ‘become’ (17) MAL ´!!qìnn!¡t!¡ ‘take something back from s.o.’ < reflexive ¡!!qìnní ‘take back (one’s belongings)’ < BV ¡!!qí ‘bring back’
Conversely, a bene-/malefactive can be used (although this is quite rarer: ca. 30 occurrences in my database) as a new basic verb to produce another verb extension: (18) Passive yìccìnní ‘be watered (animal)’ < BEN yìccí ‘water (animal)’ < BV yìí ‘drink (something)’ (19) Reflexive bàlccànné ‘pour a liquid onto oneself ’ < BEN bâlccé ‘pour a liquid to/onto s.o.’ (20) Immediate èt]àccànné ‘send immediately s.o to s.o. else’ < BEN ét]àccé ‘send s.o to s.o. else’ < MAL ét]à(tà) ‘send s.o.’
Note that it is not uncommon (25 occurrences) to find a benefactive verb derived from a malefactive (cf. ét]àccé < ét]àtà in (20) above) but the reverse (a malefactive derived from a benefactive) is not attested at all. This is another element suggesting that the benefactive derivation is more productive than the malefactive one in modern Koalib (cf. 1.2.2).
7.â•… E.g. the stones into children for Abraham in the Bible, see Luke III:8.
Benefactives and malefactives in Koalib 
2.2â•… Phenomena attested only for benefactives 2.2.1â•… Irregular forms Irregular forms are quite rare (ca. 10 occurrences) and apparently always phonologically motivated by the presence of a nasal consonant in the last syllable of the basic verb. (21) BEN mânycé ‘cook for (s.o.)’ < BV màané ‘cook’ (22) BEN qε¥nycε´ ‘kill (an animal) for (s.o.)’ < BV ¡εqnyε´ ~ qε¡εnyε´ ‘kill’
The expected forms should have been *máanàccé (cf. BEN kánnàccé, ‘distribute something to several people’ < BV kànné, ‘distribute something’) in (21) and either *ε´qnyèccε´ or *qε´εnyèccε´ (cf. BEN kε´¸t¸t èccε´, ‘put/lay something (down) for s.o.’ < BV k¡ε¸t¸t´ε, ‘put/lay something (down)’) in (22). The consonantic basic verbs (i.e. items whose Centrifugal Imperfective (CFI) form ends with a consonant) with final -/m, \/ (transcribed here as m and ny respectively) regularly produce benefactives with a -/LE/ suffix: (23) BEN ókrf`mjε´ [ógrf`mL´ε] ‘crush (an egg) belonging to s.o. between one’s hands’ < BV òkrf´m ‘crush (an egg between one’s hand)’ (24) BEN íkqìnycí [ígqì\Lí] ‘lance s.o.’s (abscess)’ < BV ìkqíny ‘lance (an abscess)’
2.2.2â•… Loss of some derivational characteristics For a handful of items, the typical benefactive tone melody has been lost and we can only know that those verbs are (or were) benefactive through other clues. I will call henceforth such items old benefactives (OB) or debenefacted verbs, as opposed to full benefactives, i.e. those verbs which display every characteristic of the standard benefactive verb extension (see Table 1). (25) OB yìccí ‘give s.o. (something) to drink’ < BV yìí ‘drink (something)’
Here the benefactive suffix-[ccí] (see Table 1) and the fact that the valency of yìccí (ditransitive) is higher than that of its probable morphological source yìí (transitive) tend to show (see 1.2.3) that yìccí, although being now deprived of the typical benefactive tone melody, must have been a full benefactive in the past. (26) OB rε`ccε´ or `εrccε´ [ε`rcε´] ‘sprinkle s.o./something (with something else)’ < BV rε`´ε, ‘id.’
There does not seem to be any significant difference in terms of valency between the two verbs, i.e. both are transitive. The only actual clues showing that rε`ccε´ ~ `εrccε´ must have been previously a full benefactive are the suffix -[(c)cε´] and the obvious similarity of meaning between the BV and the old benefactive.
 Nicolas Quint
A less advanced case in the loss of the benefactive derivational characteristics is !¥mLí [!¥mjí] ‘leave/abandon (s.o./something)’, whose original basic verb seems to be missing.8 Two arguments strongly favour the hypothesis of a benefactive origin for this item. First, !¥mjí has the typical H-(L)n-H (with n = 1 and the falling tone of !¥ being analysed as H+L) benefactive tone melody (cf. Table 1). Second, the benefactive suffix regularly appears under the form -/LE/ when the basic verb ends in -/m/, as showed above in (23) with BEN ókrf`mjε´ < BV òkrf´m. Nevertheless, in contemporary Koalib, !¥mjí cannot be analysed anymore as a full benefactive, since this verb may be used as a basic verb to produce a new benefactive verb extension, namely !¢mjìccí ‘leave something for (s.o.)’ and there is no other example in my database of a benefactive being produced from another benefactive.
2.3â•… Conclusions of the morphological overview The cases discussed above show that the bene-/malefactive verb extensions are fully integrated in the Koalib morphological system and seem to be quite old phenomena, as the loss of basic verbs and the existence of irregular and (partly) debenefacted verbs seem to attest. Furthermore, the higher productivity of the benefactive derivation as well as the fact that the morphology of the benefactive is more diversified than that of the malefactive suggest that the former verb extension is more central to the Koalib derivational system than the latter.
3.â•… Benefactive and malefactive in Koalib: Semantic considerations 3.1â•… Semantic roles 3.1.1â•… Semantic roles encoded by the benefactives In Koalib, the benefactive extension seems to encode all three semantic roles distinguished for benefaction by Kittilä (2005: 272–276): Recipient9 (27) BEN !´!pìccí ‘bring something to s.o.’ < àapé ‘bring something’ (28) BEN úucìccí ‘send s.o. to s.o. else’
< òocé ‘send s.o.’
8.â•… The expected BV should have the form */A(A)M(E)/ [a(a)m(e)~!(!)m(i)], but the only verb corresponding to this reconstructed root is áamè, ‘be sweet/nice’, the meaning of which does not seem to have anything in common with !ˆmjí. 9.â•… It is worth mentioning here that, against all odds, the Koalib (irregular) verb ínt!`(t!`) ‘give’ displays the main tonal and affixal characteristics of malefactives. Therefore, like in several other languages of the world (Kittilä 2005: 277–278), the Koalib item most commonly Â�associated with the meaning of ‘give’ cannot be used to illustrate the semantic role labelled recipient by Kittilä (2005) for benefaction.
Benefactives and malefactives in Koalib 
Recipient-beneficiary (29) BEN !΄ccìccí ‘build s.o. something, build something for s.o.’ < àccé ‘build something (a house)’ (30) BEN íklìccí ‘make something become green (i.e. paint a house green) for the sake of s.o./make something belonging to s.o. become green’ < íklì ‘become green’ (31) BEN úkwqìccí ‘sew something for s.o.’ < òkwqé ‘sew something’ (Pure) beneficiary (32) BEN áŋqìccí ‘to draw (water) for (= instead of) s.o. (who is absent)’10 < àŋqé ‘to draw (water)’ (33) BEN útqìccí ‘to gamble for (= instead of) s.o. (in order to bring him/her good luck)’ < ùtqí, ‘to gamble’
Thus, as regards benefaction, Koalib is a neutral language (Kittilä 2005: 284–286), as one and the same morphological device (the benefactive extension) can be used to encode the three semantic roles of recipient, recipient-beneficiary and (pure) beneficiary.
3.1.2â•… Semantic roles encoded by the malefactives For malefactive verbs, it seems that only the semantic role of source11 (i.e. the malefactive equivalent of Recipient for a benefactive) is effectively attested: (34) MAL áapètà ‘take something from s.o.’ < àapé ‘bring something’
In some instances, the semantics of the malefactive verb may induce us to perceive a semantic component akin to substitutive malefaction: (35) MAL óonàtà ‘harvest unduly the crop of s.o. (instead of him/her)’ < òoné ‘harvest’
If you harvest unduly s.o.’s crop, this means that you harvest it (unduly) instead of him/her, but not on his/her behalf. Therefore, from a semantic point of view, what prevails here is not substitution but extraction, i.e. if we rephrase Kittilä’s words (2005: 273) in a malefactive context: “as a result of an event, a (concrete) entity [here: the
10.â•… In this case, the (pure) beneficiary interpretation is clearly favoured, as there also exists a second benefactive verb extension derived from the same basic verb (àŋqé), namely áŋqàccé ‘draw water for s.o. (who is present)’, this meaning being preferably associated with the semantic roles of recipient and recipient-beneficiary. For more details about double benefactives and their possible association with a specific semantic role, see also Section 3.2, in particular (40). 11.â•… Due to the fact that Koalib formally distinguishes benefactives and malefactives, I have preferred to use specific labels for the semantics of malefaction. I give here the malefactive labels I have devised, each one followed (in brackets) by its benefactive counterpart in Kittilä (2005): source (recipient), maleficiary (beneficiary), source-maleficiary (recipient-beneficiary), extraction (reception), substitutive malefaction (substitutive benefaction).
 Nicolas Quint
crop] leaves a source’s [here: the legitimate owner of the crop/field] sphere of control or domain of possession.” In fact, it seems that, at least in Koalib, the notion of substitution is just not compatible with the malefactive extension, which therefore cannot encode the semantic roles of maleficiary and source-maleficiary. The fact that the range of semantic roles is narrower for malefactives than for benefactives could account (at least partly) for the higher productivity of benefactive derivation in Koalib (see Table 2 and the discussion in 1.2.2).
3.2â•… Double benefactives In some cases, two benefactives can be produced from one and the same basic verb (there is not any clear, convincing example for malefactives). This obtains for at least several tens (between 30 and 50) of basic verbs. Such double benefactives (DB) are particularly interesting from a semantic point of view. Sometimes, the two benefactives seem to be absolute synonyms: (36) BV kε`¸t¸t΄ε ‘put/lay something (down)’ > BEN kε΄¸t¸tèccε΄ ~ kít¸¸tìccí ‘put/lay something (down) for s.o.’
In other instances, however, there does exist a difference in meaning between both forms, and this difference is not always of the same nature. First, the BV can be polysemous; in such cases, each member of the pair seems to have been ascribed to one particular meaning or semantic potentiality of the basic verb: (37) BV àaké 1. ‘marry (one woman)’; 2. ‘do something (work…)’ > áakàccé ‘marry a woman among a certain family/group’↜, corresponding to 1. > !΄!kìccí ‘do something (work…) for s.o.’↜, corresponding to 2. (38) BV `εqnyε΄ ~ qε`εnyε΄ ‘kill’ > qε¥nycε΄ ‘kill (an animal) for (s.o.)’ (the concrete meaning of BV is maintained) > éqnyàccé ‘forgive s.o. (for something)’, which has a more abstract meaning (‘kill/erase [in one’s mind] the bad actions that s.o. did to you’), not attested (to my knowledge) in the BV itself
Second, there is a distinction between main and secondary benefactives. In this kind of pairs, one of the double benefactives can be held as the main benefactive for it can combine with many different arguments while the use of the other—the secondary benefactive—is restricted to one or a few idioms (39). Actually, the main vs. secondary distinction is probably a variant of the pairs based on the polysemy of the BV (cf. 1.), in which one of the double benefactives has become semantically so specialised that it is now hard to determine its exact meaning (i.e. independently from the idioms in which it appears).
(39)
Benefactives and malefactives in Koalib 
BV òmmé ‘catch (animals)’ > main benefactive úmmìccí ‘catch (animals) for s.o.’ > secondary benefactive ómmàccé attested in only two idioms such as: ómmàccé kwícì yε`e-lá catch.ben.ipfv human.o eye.pl.o-up ‘make sheep’s eyes at s.o.’
Finally, some pairs of double benefactives seem to correspond to two of the basic semantic types of benefactives distinguished by Van Valin & LaPolla (1997: 382–384), namely the deputative-benefactive (“when the action is performed instead of s.o.”) and the benefactive-recipient (“when the beneficiary is also the recipient of the action”): (40) BV còocé ‘filter (the beer)’12 > deputative-benefactive cóocàccé ‘filter (the beer) for s.o. (= do the work instead of him/her)’ > benefactive-recipient cúucìccí ‘filter (the beer) for s.o. (= in order to give him/ her it to drink, i.e. make him/her the recipient of the action of filtering)’.
Synchronically, from a morphological point of view, it is rather hard to explain how these various pairs of double benefactives have emerged in the Koalib system. Similarly, from a semantic point of view, it is also hard to understand how each member of such pairs has come to be associated to one particular meaning. In some instances however, that member of a pair of DB which contains high vowels (i.e. /i, !, u/) might be derived from a factitive extension of the BV, as the use of high vowels is one of the most commonly used derivational tools in Koalib (see Quint 2006: 38) to produce factitives from BV containing low vowels (i.e. /e, ε, a, f, o/): (41)
BV 1àaré ‘go/come using a roundabout way’ > Factitive 1!`!rí ‘bring/take s.o. using (= make s.o. use) a roundabout way’ > BEN (from 1àaré) 1áaràccé ‘go/come round something’ > BEN (from 1!`!rí) 1!΄!rìccí ‘make s.o.’s animals use a roundabout way’
In (41), the meaning of each verb makes it clear that the two benefactives at stake are derived from the BV and from the factitive respectively; one cannot speak of double benefactives, for neither benefactive is directly derived from the same BV. Nonetheless, it is quite probable that some of the DB pairs attested in synchrony may be the result of the coexistence at an earlier stage of the language of a BV-derived benefactive (with low vowels) and a factitive-derived benefactive (with high vowels).
12.â•… Following Kittilä’s (2005) nomenclature, cóocàccé would encode the semantic role of (pure) beneficiary and cúucìccí the one of recipient-benefactive. In the discussion of (40) however, I have maintained the labels of Van Valin & LaPolla, because I think that such labels and their definition suit particularly well the case at stake.
 Nicolas Quint
3.3â•… Non-bene-/malefactive meanings for bene-/malefactive verb extensions 3.3.1â•… Benefactives with malefactive values and the reverse In some cases, the meaning associated with a benefactive verb extension is clearly malefactive: (42) BEN ákràccé ‘break something (bed/body part…) belonging to s.o.’ < BV àkré ‘break’13
The reverse (i.e. a malefactive with benefactive meaning) also happens: (43) MAL áamètà ‘please s.o., be liked by s.o.’ < BV áamè ‘be sweety/nice’
Such instances are interesting in that they question the very accuracy of the “benefactive” and “malefactive” labels applied to the two Koalib verb extensions considered in this article. However, statistically, most of benefactives and malefactives attested in Koalib actually convey respectively a notion of benefaction and malefaction in relation to the meaning of the basic verb. Therefore, the fact that some few (probably no more than 40 in my database) specific items have developed a semantic value much distant from the dominant one for the extension at stake is not enough per se to dismiss the generic labels “benefactive” and “malefactive” for the verb extensions examined here. Besides, for a human being at least, the action of breaking something (see (42)) is probably more easily viewed as having a malefactive connotation than the reverse and conversely, the fact of being nice (see (43)) is more spontaneously conceived of as having a benefactive value. Therefore the respectively non-benefactive and nonmalefactive “abnormal” meanings of ákràccé and áamètà seem to be the result of a conflict between the meanings inherent to the bene-/malefactive verb extensions and the peculiar semantics of the basic verbs such extensions combine with.
3.3.2â•… Simple increase of valency Quite frequently, in particular—but not only—in the case of benefactive extensions, the use of a bene-/malefactive extension seems to convey no particular bene-/malefactive meaning but to be used as a simple tool to increase the valency of a given verb:
13.â•… One could arguably find contexts in which ákràccé may be used with a benefactive meaning. For instance, if I broke s.o.’s arm so that he could collect the insurance money, in light of the semantic environment, the fact of breaking s.o.’s arm could be viewed as having a benefactive consequence and therefore the person whose arm would have been broken would actually be the beneficiary of the verbal action. However, although being theoretically possible, such cases of benefactive interpretations of ákràccé seem to be much less favoured than malefactive interpretations in the daily use of the language: in my database, the semantic contexts in which ákràccé is attested are consistently malefactive.
Benefactives and malefactives in Koalib 
(44) BV kànné ‘distribute something’ > BEN kánnàccé ‘distribute something to s.o.’ (45) MAL úrb!`t!` ‘lie in ambush’ > BEN úrb!`ccí ‘set an ambush for s.o.’ (46) Reflexive !`qnyìnní ‘hide (behind something)’ > BEN !΄qnyìnnìccí ‘hide from s.o. (behind something)’ (47) Transitive ànt¸àcé ‘say something (transitive)’ > BEN ánt¸àccé ‘say something to s.o. (ditransitive)’
In the case of benefactives, this “plus-one-argument use”, although sporadically attested in the case of simple verbs (see (44)), is particularly frequent for double benefactives ((45), (46), (47)), i.e. when a benefactive is derived from another verb extension. From such cases, it appears that, in a derived verb BV-VE1-VE2, in which two verb extensions (VE1 and VE2) have been added to the same basic verb, when VE2 is a benefactive, the benefactive semantic value of VE2 tends to be lost, while the resulting BV-VE1-VE2 verb retains the morphological (see 1.1) and syntactic (plus one argument, see 1.2.3) characteristics typically associated with the benefactive. As regards malefactives, the “plus-one-argument use” is attested but exceptional. In fact, the following instance is one of the few convincing ones I have found: (48) MAL !΄urìt!` ‘be easily carried (because it is light) by s.o.’ < BV !΄urì ‘be light’
From a semantic point of view, it is hard to see any malefaction in the meaning of !΄urìt!`. However, the structure of this verb closely parallels the malefactive item ní!`t!` ‘be too heavy for s.o.’ > BV nîi ‘be heavy’. The fact of ‘being too heavy for s.o.’ is clearly understandable as a semantically malefactive derivation from the basic meaning ‘be heavy’. What happens here is that it is much more common to think about something as being ‘too heavy’ for s.o. than about something being ‘too light’ for s.o. Therefore, the non-malefactive meaning of !΄urìt!` is probably due mainly to the semantics of !΄urì, and the same seems to obtain with the few remaining cases of malefactives deprived of malefactive value.
3.3.3â•… Spatial uses As in other languages of the world,14 one can find in Koalib several examples in which the semantic value of the linguistic devices otherwise used to express bene-/malefaction is essentially spatial, i.e. associated with the notions of movement/position: (49) BEN íilìccí ‘draw nearer from s.o.’ < BV `εεlε΄ ‘go/come’
14.â•… See, for instance, Censabella (Examples (1), Rapold (Example (32)), and Schmidtke-Bode (Table 2) in this volume.
 Nicolas Quint
(50) BEN t!΄!ŋìccí, ‘draw nearer (while moving upwards) from some place’ < BV tàaŋé ‘go upwards’ (51) MAL éelàtà ‘go/come and meet s.o.’ < BV `εεlε΄ ‘go/come’ (52) MAL érlàtà ‘stand near from s.o.’ < BV èrlé ‘stand’ (53) MAL !΄llìnn!`t!` ‘go/come out from a group and turn towards s.o. (= extract oneself towards s.o.)’ < reflexive !`llìnní ‘go/come out from a group (= extract oneself)’ < BV àllá ‘take out/extract’
The above verb extensions all come from basic verbs whose meaning is closely linked with motion or position, which seems to allow us to think that, as for the cases seen in 3.3.1, the semantics of the BV accounts for a great part of the spatial uses of BEN and MAL. However, two other elements must be taken into consideration here. First, there are other instances in which the meaning of the BV is not obviously spatial and the use of the VE seems to be triggered by an extra spatial marker, especially for malefactives: (54) MAL ŋáalètà ‘yawn’ < BV ŋàalé ‘yawn’
To my knowledge, both BV ŋàalé and its malefactive counterpart ŋáalètà mean exactly the same, i.e. ‘yawn’. However, the form ŋáalètà is obligatorily used in association with the postposition -lá ‘up(wards)’, whose basic meaning is obviously spatial: kwé-ŋããalètà-lá ‘he will yawn’. Second, given the fact that the notion of oriented movement is absolutely central to the whole structure of Koalib verb morphology (see Footnote 2), one cannot exclude the possibility that the attested spatial uses of MAL and BEN verb extensions in modern Koalib may as well represent the original semantic value of such extensions. In the absence of diachronic data, however, I cannot test the validity of this hypothesis.
3.3.4â•… Causal interrogative use Benefactive and malefactive extensions can happen to be used in causal interrogations of the following type: (55) áatà kw-òrò kw-ûrl!`lìcc!`ŋ!`? what clf-become.pfv.s3sg clf-get.drunk.ben.pfv.rel.s2sg.o3sg ‘Why did you get drunk?’ (lit. ‘what + is [it] + [this thing] that you got drunk for it’, with kw-ûrl!`lìcc!`ŋ!` < BEN úrl!`lìccí < BV ùrl!`lí ‘get drunk’) (56) áatà kw-òrò kw-ôoqàtéŋà what clf-become.pfv.s3sg clf-go/come.down.mal.pfv.rel.s2pl.o3sg -kwò Kálkè(ŋwó)? -rel Delami(.o) ‘Why did you come down (from your mountains) to Delami?’ (lit. ‘what + is [it] + [this thing] that you (all) got down for it + Delami’, with kw-ôoqàtéŋà < MAL óoqàtà < BV òoqé ‘go/come down’)
Benefactives and malefactives in Koalib 
Apparently, the causal interrogative use seems to be indifferent to the semantic notions of bene-/malefaction, as a benefactive and a malefactive derived from the same basic verb may both be used for causal interrogation in exactly the same context:15 (57) kw-ào kw-àlò, áatà kw- !¥! rìcc!`ŋ!`? clf-woman clf-poss2pl what clf-cry.ben.pfv.rel.s2sg.o3sg ‘Hey woman, why are you crying?’ (see John XX:13) (lit. “your woman what + [is this thing] that you cried/are crying16 for it’, with kw-!¥! rìcc!`ŋ!` < BEN !¢!rìccí < BV àaré ‘cry’) (58) kw-ào kw-àlò, áatà kw- âaràtèŋà? clf-woman clf-poss2pl what clf-cry.mal.pfv.rel.s2sg.o3sg ‘Hey woman, why are you crying?’ (see John, XX:15) (lit. ‘your woman what +[is this thing] that you cried/are crying for it’, with kw- âaràtèŋà < MAL áarètà < BV àaré ‘cry’)
In my whole corpus, I have only eleven occurrences (ten benefactives and one malefactive)17 of such causal interrogative use in Koalib. This use is strictly limited to causal interrogations beginning with áatà (kw-òrò) (in one case also with ŋ-!`!nì áatà, lit. ‘this affair] clf-be.called.pfv.s3sg + what’, i.e. ‘what is it?’). In particular, such use of bene-/malefactive verb extension is not attested in non-causal questions, and in the answer to questions like (55) and (56), one can use neither a malefactive nor a benefactive: (59) áatà kw-òrò kwé-tìinyìcc!΄ŋ!`? what clf-become.pfv.s3sg clf-be.afraid.ben.pfv.rel.s2pl.o3sg Question: ‘What are you afraid of?’ (lit. ‘what + is [it] + [this thing] that you (all) got afraid for it’, with kwé-tìinyìcc!΄ŋ!` < BEN tíinyìccí < BV tε΄εnyε` ‘be afraid’) nyíŋ!` lè-tèenyà kèerâoá. we clf-be.afraid.pfv.s1pl lion.o Answer: ‘We are afraid of the lion’ (lit. ‘we + got afraid + as concerns the lion’, with lè-tèenyà < BV tε΄εnyε` ‘be afraid’)
15.â•… Both examples come from Wa@d wiyaŋ (1993: 235) a Koalib translation of the New Testament, which was mainly produced by father Samwiil J. Angollo, a native speaker of Koalib (rere dialect). The grammaticality and tone of those sentences have been carefully checked in April-May 2008 with Siddig Ali Karmal Kokko, my main informant for Koalib. 16.â•… The Koalib Perfective, probably due to the fact that it is independent from the movement (contrary to Imperfective, which is either Centrifugal or Centripetal: see Footnote 2), can be used with a progressive value, like in (57) and (58). It would probably be possible to find a more suitable label than ‘Perfective’ (i.e. one encompassing both the perfective and progressive value) for this Koalib aspect-movement form, but such question goes much beyond the scope of this article: it will be dealt with in Quint (forthcoming). 17.â•… Those numbers are based on my computerised corpus, which does not include (57) and (58), that I have collected during my last mission in the Koalib-speaking area (April-June 2008).
 Nicolas Quint
As all the examples of this section show, the causal interrogative use seems to be quite systematically associated with the perfective. Although I have tried to check it, I am not entirely sure that such use would be allowed in imperfective contexts. For certain verbs, such as úrl!`lìccí, the causal interrogative use is the only attested use of the bene-/ malefactive verb extension, which means that the basic form (centrifugal imperfective) presented here is a reconstructed one, i.e. *úrl!`lìccí. According to my current knowledge of Koalib, I have no convincing explanation to link the causal interrogative use to the rest of the semantic values associated with bene-/malefactive verb extensions. However, it should be said that in several Bantu languages, such as Zulu (Canonici 1996 [1994]: 120; Doke 1992 [1927]: 140–141) and Swahili (Lodhi 2002: 11), a very similar causal interrogative use is also observed (and explicitly mentioned by grammarians) for the applicative verb extension,18 the semantic value (Creissels 2006: 75–77; Schadeberg 2003: 74) of which corresponds quite closely to both Koalib benefactive and malefactive verb extensions: (60) ba-khal-el-a-ni clf.t/a-cry-appl-thv-q ‘What are they crying for, why do they cry?’ (Canonici 1996: 120)
Such a specific similarity between Bantu and Koalib (a Kordofanian language) may be due to the fact that they are genetically related, since they belong to the Niger-Congo phylum.19
3.3.5╇ Cultural uses In some cases, the actual meaning of the benefactives and malefactives seems to be due to cultural idiosyncrasies inherent to the Koalib community. The two extensions built on BV òoré ‘have sex with s.o.’ are a case in point. On the one hand, there is MAL óoràtà ‘have sex with s.o. else’s wife’; at least according to the dominant moral values of the western world, such a meaning is easily conceivable as having a malefactive connotation. On the other hand, there is BEN úurìccí ‘have sex with s.o. else’s daughter/ son’. Why should this latter meaning be encoded by a benefactive verb extension in Koalib? No contemporary speaker seems to be able to answer this question. We can only hypothesise that this particular benefactive verb extension probably bears witness
18.â•… I am much indebted to Laura J. Downing (2008–08-26: p.c.) who, during a presentation of a draft of this paper, made me aware of this similarity between Koalib bene-/malefactives and Bantu (namely Zulu) applicative and suggested me further readings on that topic. 19.â•… For a more substantial account of the question of the relationship between Kordofanian and the rest of Niger-Congo, see Quint (2006: 3–12).
Benefactives and malefactives in Koalib 
of old moral codes which have been lost by today’s Koalib but whose reality is still alive in the verb morphology of the language.
3.4â•… Summary of the semantic overview This semantic overview of the two extensions at stake shows clearly that, although it is objectively true that the Koalib benefactives and malefactives are usually associated respectively with the core notions of benefaction and malefaction, both verb extensions can convey a variety of meanings which may be quite distinct from such core-notions.
4.â•… Conclusion As has been shown in the previous sections, Koalib verb morphology represents a valuable example for those interested in the understanding of benefaction and malefaction, because Koalib verb morphology distinguishes explicitly (through verbal extensions) benefactive and malefactive synthetic forms, which does not seem to be that common among the world’s languages.20 Furthermore, in Koalib, both benefactive and malefactive derivations display a rich range of morphological and semantic variants which will undoubtedly enrich the debate about how and why such extensions can develop in individual languages. In conclusion, I would like to emphasise that this brief account of bene-/malefactive verb extensions in Koalib clearly illustrates that “descriptive and comparative work on Kordofanian languages is still in its infancy” (Schadeberg 1989: 79). In particular, many of the semantic and morphological characteristics I have exemplified and discussed above cannot be easily compared with other Kordofanian languages because of the lack of accurate, available data on these languages.21 It is highly desirable that Kordofanian studies will develop in the future for, as we do not dispose of any diachronic data for Koalib, a comparative approach is probably the only possibility to understand better the origins and present uses of the benefactive and the malefactive in Koalib as well as in those Kordofanian languages in which such verb extensions are actually attested. 20.â•… I have already (see Footnote 5) mentioned the case of Amharic (Amberber: 2000). Some other cases have been mentioned during the Zürich workshop on the Typology of benefactives and malefactives, such as Abaza (Radetzky & Smith, this volume), Gumer (Völlmin, this volume) and Mapudungun (Zúñiga, this volume). 21.â•… Suffice it to say that the most complete grammatical studies we dispose of for Heiban languages (the family Koalib belongs to) were published several tens of years ago (Black 1971; Stevenson 1956–1957; 1943; 1942; Meinhof 1966 [1943–1944]) and that none of them bears mention of tone rules.
 Nicolas Quint
Abbreviations appl ben bv cfi clf db ipfv mal o ob pfv
applicative benefactive basic verb Centrifugal Imperfective classifier (noun class marker) Double Benefactive Imperfective malefactive Object Old benefactive perfective
pl poss q rel s sa sg t/a thv ve
plural possessive question particle/marker relative subject Sudanese Arabic singular tense/aspect marker thematic vowel verb extension
Bibliography Amberber, Mengistu. 2000. Valency-changing and valency-encoding devices in Amharic. In Changing Valency. Case Studies in Transitivity, Robert M.W. Dixon & Alexandra Y. Aikhenvald (eds.), 312–332. Cambridge: CUP. Black, Mr. & Mrs. K. 1971. The Moro Language, Grammar and Dictionary. Khartoum: University of Khartoum, Faculty of Arts. Canonici, Noverino. 1996 [1994]. Zulu Grammatical Structure. Durban: University of Natal. Creissels, Denis. 2006. Syntaxe générale, une introduction typologique, 2: La phrase. Paris: Lavoisier. Doke, Clement Martin. 1992 [1927]. Textbook of Zulu Grammar. Cape Town: Maskew Miller Longman/The University of the Witwatersrand. Hyman, Larry. 2007. Niger-Congo verb extensions: overview and discussion. In Selected Proceedings of the 37th Annual Conference on African Linguistics, Doris L. Payne & Jaime Peña (eds.), 149–163. Somerville MA: Cascadilla Proceedings Project. Kittilä, Seppo. 2005. Recipient-prominence vs. beneficiary-prominence. Linguistic Typology 9(2): 269–297. Lemaréchal, Alain. 1998. Théories de la transitivité ou théories de la valence: Le problème des applicatifs. In La transitivité, André Rousseau (ed.), 203–218. Villeneuve d’Ascq: Presses Universitaires du Septentrion. Lodhi, Abdulaziz. 2002. Verbal extensions in Bantu (the case of Swahili and Nyamwezi). Africa & Asia 2: 4–26. 25 September 2008. Meinhof, Carl. 1966[1943–1944]. Das Heiban in Kordofan. Zeitschrift für Eingeborenen-Sprachen XXXIV: 94–130. (Nendeln (Liechtenstein): Kraus Reprint). Quint, Nicolas. 2006. Phonologie de la langue koalibe. Paris: L’Harmattan. Quint, Nicolas. Forthcoming. Morphologie de la langue koalibe. Paris: L’Harmattan. Schadeberg, Thilo. 1981. A Survey of Kordofan, Vol. 1: The Heiban Group. Hamburg: Helmut Buske.
Benefactives and malefactives in Koalib 
Schadeberg, Thilo. 1989. Kordofanian. In The Niger-Congo Languages, John T. Bendor-Samuel (ed.), 66–80. Lanham MA: University Press of America. Schadeberg, Thilo. 2003. Derivation. In The Bantu Languages, Derek Nurse & Gérard Philippson (eds.), 71–89. London: Routledge. Shibatani, Masayoshi. 1996. Applicatives and benefactives: A cognitive account. In Grammatical Constructions: Their Form and Meaning, Masayoshi Shibatani & Sandra A. Thompson (eds.), 157–194. Oxford: OUP. Stevenson, Roland C. 1942. The Tira Language. Ms. Stevenson, Roland C. 1943. The Otoro Language. Ms. Stevenson, Roland C. 1956–1957. A survey of the phonetics and grammatical structure of the Nuba Mountain languages. Afrika und Übersee, Band XL, Heft 2: 73–115, Band XLI, Heft 1/2: 27–65, Band XLI, Heft 3: 117–196. Van Valin, Robert D., Jr. & LaPolla, Randy J. 1997. Syntax. Structure, Meaning and Function. Cambridge: CUP. Wa@d wiyaŋ (Koalib translation of the New Testament, mainly produced by Angollo, S.J.). 1993. Khartoum: The Bible Society in Sudan. Williamson, Kay & Blench, Roger. 2000. Niger-Congo. In African Languages. An Introduction, Bernd Heine & Derek Nurse (eds.), 11–42. Cambridge: CUP.
Benefactives and malefactives in Gumer (Gurage) Sascha Völlmin Universität Zürich
Gumer (West-Gurage, South Ethiosemitic) has three suffixes to mark objects on the verb: ‘primary object’, ‘benefactive’, ‘malefactive/locative/instrumental’. BEN and MAL can occur with every verb to express in a rather broad sense that the event is ‘to one’s benefit’ or ‘to one’s detriment’. BEN covers the benefactive senses ‘recipient’, ‘deputative’, and ‘plain’. Generally, BEN and MAL express non-core participants. The three suffixes are mutually exclusive: if there are two concurring objects, the most salient candidate is chosen (which usually is BEN or MAL). Overt nominals BEN are marked with yә-, the same prefix that marks recipients. MAL are marked with yә- or bә-. A construction with a subordinate form of barәm ‘say’ (“saying for NP”) also expresses benefactives.
0.â•… Introduction This article is a descriptive overview of the benefactive and malefactive constructions in Gumer (gwәmarә), a variety of the Western Gurage cluster of dialects/languages (South Ethiosemitic), which are spoken in the Gurage administrative zone south-west of Ethiopia’s capital Addis Ababa. In the 16th edition of the Ethnologue (Lewis 2009) Gumer is listed as one dialect of Sebat Bet Gurage along with Chaha, Ezha, Gura, Gyeto, and Muher. Chaha is the variety closest to Gumer with only minor dialectal differences (cf. Völlmin forth.). This article is exclusively based on own fieldnotes of Gumer. However, due to the fact that the West Gurage dialects/languages are rather closely related, the principles of the benefactive and malefactive constructions are (presumably) the same in all of these varieties, the main differences lying in the shape of the morphemes. The first part presents the subject and object affixes of the Gumer verb. The following sections discuss the semantics and use of the benefactive and malefactive suffixes, the question of their argumenthood, and the choice of the right object suffix in case there is more than one possibility. Section 5 deals with the marking of overt NPs. Finally, preceding some concluding remarks, the construction with the verb barә ‘say’ to express benefactives is discussed.
Sascha Völlmin
1.â•… Verb morphology Gumer features a rather complex verbal morphology. There are three basic verb stems or TAM forms: perfective, imperfective, and jussive/imperative. By adding pre- and/or suffixes further forms are built such as two different future tenses or subordinate forms. Besides the obligatory marking of subject in all of these forms there are three different (optional) possibilities to mark objects on the verb: primary object, benefactive, and malefactive-locativeinstrumental (see also Section 2.2). This quite rich verbal system is further complicated by morphophonological processes, especially palatalization and labialization.
1.1â•… Subject marking Table 1 below shows the subject markers of the three basic verb stems, the position of which is indicated by Σ. The perfective is suffix conjugated, i.e. the subject marker follows the stem. The imperfective and the jussive/imperative are prefix conjugated, in the plural, however, they are extended by suffixed number/gender markers. There is one special form, the so-called impersonal (ips), a hallmark of Gurage languages that corresponds to English one or German man. Unlike the other persons the impersonal does not have a corresponding independent prounoun and cannot occur with any other overt subject. In the perfective there is no affixed subject marking as such on the verb, the only formal marking being both palatalization and labialization in the stem. The imperfective and the jussive have a (default) 3rd person marking y(ә)-, whereas the impersonal imperative lacks this subject marking (like the ‘personal’ imperatives). Finally, some subject markers have allomorphs when followed by object suffixes, as does the 1.sg.sbj of the imperfective after prefixes (see notes to Table 1). Table 1.╇ Subject markers pfv 1.sg 2.sg.m 2.sg.f 3.sg.m 3.sg.f 1.pl 2.pl.m 2.pl.f 3.pl.m 3.pl.f ips
1
Σ-xw Σ-xә Σ-xy Σ-ә Σ-әc Σ-nә2 Σ-xu Σ-xma Σ-o3 Σ-әma ΣPAL + LAB
ipfv
jus/imp
ә-Σ4 t-i-Σ t--i-ΣPAL y-i-Σ t-i-Σ n-i-Σ-nә2 t-i-Σ-o3 t-i-Σ-әma y-i-Σ-o3 y-i-Σ-әma y-i-ΣPAL + LAB
n-i-Σ Σ ΣPAL yә-Σ t-i-Σ n-i-Σ-nә2 Σ-o3 Σ-әma yә-Σ-o3 yә-Σ-әma (yә-)ΣPAL + LAB
1. -x- before object markers 2. -ne- before object markers 3. -әw- occasionally before 3.sg.m.obj 4. -n- after prefixesâ•…â•… pal, lab: palatalization, labialization in the stem
Benefactives and malefactives in Gumer
A note concerning the vowel i is in order here. This vowel is epenthetic throughout the whole language with only few exceptions, i.e. it is used to split consonant clusters of three or more consonants, but also CC word initially and some CC word internally and word finally. Despite the fact that its occurence and position is usually predictable, it is nontheless represented in the tables in two cases: in Table 1 with the subject prefixes, because they mostly occur with i, as the majority of verbs begin with a consonant; and in Table 2 below with the suffixes that consist of a CCC series and therefore necessarily occur with the epenthetic vowel.
1.2â•… Object marking There are three possibilities to mark objects on the verb: primary object (obj), benefactive (ben), and malefactive-locative-instrumental (mal). For simplicity and because this paper deals with benefactives and malefactives, the latter is glossed mal only. As can be seen in Table 2, there are two sets for each of these three object suffixes, one called ‘light’ and the other ‘heavy’: The ‘light’ set occurs after all singular (subject) forms except 2.sg.f.sbj, the ‘heavy’ set after the rest, i.e. plural subjects, 2.sg.f.sbj, and after ips. There are no ips object markers. For the perfective the 2nd persons obj of the ‘light’ version have an additional -na-, which is missing in the imperfective and jussive. The difference between the ‘light’ and the ‘heavy’ sets is often the strengthening of a consonant, (e.g. ben:2.sg.m -nxә vs. -nkә), but sometimes there is no difference at all (e.g. ben:3.sg.f -la). In the case of the 3rd persons obj there is a change from -n to -i/y, whereas the morpheme pair of 1.sg.obj is -e vs. -n. Finally, note that the 1.pl obj and ben suffixes do not differ. Table 2.╇ Object markers obj
1.sg 2.sg.m 2.sg.f 3.sg.m 3.sg.f 1.pl 2.pl.m 2.pl.f 3.pl.m 3.pl.f
ben
mal
‘light’
‘heavy’
‘light’
‘heavy’
‘light’
‘heavy’
-e -(na)xә1 1 -(na)xy -n -na -ndә -(na)xu1 -(na)xma1 -no -nәma
-n -kә -ky -i -ya -ndә -ku -kma -yo -yәma
-ni -nxә -nxy -lә -la -ndә -nxu -nx-ima -lo -lәma
-ni -nkә -nky -lә -la -ndә -nku -nk-ima -lo -lәma
-βi -βxә -βxy -wә -βa -β-indә -βxu -βx-ima -βo -βәma
-pi -βkә -βky -pwә -pa -p-indә -βku -βk-ima -po -pәma
1. with (na) → pfv, without (na) → ipfv/jus/‘light’: after singular S, except 2.sg.f.sbj/‘heavy’: after plural€S, 2.sg.f.sbj, impersonal
 Sascha Völlmin
The three object sets obj, ben and mal consist in principle of a specialized semantic marker followed by a personal suffix. The ‘light’ and ‘heavy’ markers for mal are -β/p respectively (3.sg.m -w/pw are the labialized forms, see below), whereas both the ‘light’ and ‘heavy’ markers for ben are -l in the 3rd and -n in the 1st and 2nd persons (in 1.pl the -n fused with the personal suffix -ndә). The case marking of obj is slightly less straightforward: the 3rd persons have a ‘light’ -n and a ‘heavy’ -i/y, in the other persons, however, there is no overt semantic marker, except the already mentioned -na- in the ‘light’ suffixes of the perfective. The ‘light’ 3.sg.m suffixes of obj (-n) and ben (-ә) as well as the 3.sg.m suffixes of mal (-ә) in both the ‘light’ and ‘heavy’ variants (but not the ‘heavy’ ones of obj and ben) trigger labialization of labials and velars preceding them. A floating feature [round] starts from these suffixes and docks to the first labializable consonant to the left. In the mal, this rightmost consonant is the semantic marker -β/p, the labialized forms being -w/pw. In the other cases, the labialized consonant is either the rightmost labializable consonant of the verb root, irrespective of its position, or the velar of the perfective subject markers -xә 2.sg.m.sbj or -x 1.sg.sbj. The latter results in -xw, which is identical to the subject marker that is not followed by an object suffix. Remember, however, that 1.sg.sbj followed by any other suffix is -x only, thus the labialization is indeed triggered by the obj/ben suffix. Finally, if there is no labializable consonant, no labialization takes place (e.g. 1b/c below). The object suffixes appear at the end of the verbal complex, except for an obligatory suffixed -m in the perfective in matrix clauses. This -m can be analysed as an affirmative marker (aff) as it is not present in negated forms. The examples below show the verb odә ‘tell’ in the perfective without (1a) and with the object suffixes (1b–d). The three different object suffixes are mutually exclusive: it is not possible to combine two or all of them (1e). (1) a.
Od-ә-m. tell.pfv-3.sg.m.sbj-aff ‘He told [it].’
b. Od-ә-n--im. tell.pfv-3.sg.m.sbj-3.sg.m.obj-aff ‘He told [it to] him.’ c.
Od-ә-lә-m. tell.pfv-3.sg.m.sbj-ben:3.sg.m-aff ‘He told [it] for him.’
d. Od-ә-wә-m. tell.pfv-3.sg.m.sbj-mal:3.sg.m-aff ‘He told [it to] him (to his detriment).’ e. *Od-ә-n-lә-m. tell.pfv-3.sg.m.sbj-3.sg.m.obj-ben:3.sg.m-aff (intended: ‘He told [it] to him for him.’)
Benefactives and malefactives in Gumer 
2.â•… Semantics of BEN and MAL The opposition between benefactive and malefactive can be characterized as ‘for the benefit of ’ vs. ‘to the detriment of s.o./s.th.’↜. These meanings are rather broad, and the corresponding ben and mal suffixes can be added to basically every verb, as e.g. zәnәβә ‘rain’: (2) a.
Zәnәw-ә-lә-m. rain.pfv-3.sg.m.sbj-ben:3.sg.m-aff ‘It rained for him/it/them (e.g. flowers).’
b. Zәnәβ-ә-wә-m. rain.pfv-3.sg.m.sbj-mal:3.sg.m-aff ‘It rained on him (e.g. when surprised by sudden rain).’
Events like raining are not positive or negative as such, but if one wishes to express that there is someone or something positively or negatively affected by it, it will be done so by adding the corresponding ben or mal suffixes. The semantics of a verb may restrict the use of some of these suffixes to very specific contexts or meanings, but it does not necessarily preclude them from appearing. The verb tamә ‘be tasty, be sweet’, e.g., is normally perceived as being positive, so one might expect it not to occur with a malefactive. Consider, however, the following contrasting examples: (3) a.
∫әrәt y-i-tәm. food 3.sg.m.sbj-be.tasty.ipfv ‘The food is tasty.’
b. ∫әrәt y-i-tәm-e. food 3.sg.m.sbj-be.tasty.ipfv-1.sg.obj ‘I find the food tasty.’ c.
∫әrәt y-i-tәm-ni. food 3.sg.m.sbj-be.tasty.ipfv-ben:1.sg ‘The food is tasty (i.e. I prepared it well).’
d.
∫әrәt y-i-tәm-bi. food 3.sg.m.sbj-be.tasty.ipfv-mal:1.sg ‘The food is tasty to my (=the cook’s) detriment.’ or: ‘The food is too sweet (i.e. I have added too much sugar, honey, etc.).’
Without any suffix (3a), one simply states that some food tastes good. The obj suffix (3b) expresses the experiencer of the tasty food. With the ben suffix (3c), the verb gains the additional meaning that one (i.e. the beneficiary) has succeeded in cooking food that is tasty. Finally, Example (3d) with the mal suffix may be less usual, but the important point is that it is not completely ruled out. It could express, e.g., that one intended to prepare bad food but it turned out to be good tasting, which is to one’s detriment or against one’s intention. Another thinkable interpretation involves the verb’s literal meaning ‘be
 Sascha Völlmin
sweet’: it can be understood that the food is too sweet because the cook (=maleficiary) added (accidentally or through ignorance) too much sugar or the like.
2.1â•… Different types of benefactive (and malefactive) The benefactive in Gumer covers all three senses of benefactives proposed by Van Valin & LaPolla (1997: 383f.): recipient benefactives (‘so that they have it’), deputative benefactives (‘so that they wouldn’t have to’), and “plain” benefactives (‘to amuse them, or other kinds of benefit’). Sentence (4), e.g. can mean both that somebody slaughtered a sheep for somebody else to have it, and that somebody did it in somebody’s stead. Furthermore, it can also be a “plain” benefactive meaning that a sheep was slaughtered in honour for someone, or even (but in this case less likely) that the slaughtering took place for one’s amusement only. (4) T’ay ant’--ixw--ilә-m. sheep cut.pfv-1.sg.sbj-ben:3.sg.m-aff ‘I slaughtered a sheep for him.’
There is at least one other use of the benefactive that does not fit entirely the above categorization (unless it is considered a subtype of “plain” benefactive) and deserves special attention. It can express that one (i.e. the beneficiary) carried out an a action successfully. The above (3c) is an example of this use, another one being (5): (5) K-irr--ixw-it bә-mәrf gәppa-ni-m. thread-def in-needle enter.pfv[.3.sg.m.sbj]-ben:1.sg-aff ‘The thread entered (the eye of) the needle.’
The following example could be understood in the same way, i.e. the beneficiary was successful in healing his child: (6) T-ikә-na fәzәz-ә-ni-m. child-1.sg.poss become.better.pfv-3.sg.m.sbj-ben:1.sg-aff ‘My child got better (recovered from illness).’
Generally, however, this sentence simply expresses that the child recovered from some illness and that this is good for the participant marked as beneficiary. This rather general “plain” meaning of benefactives, i.e. that “something is good or positive for someone”, is also represented in the following two examples: (7) Zәnga-na sar-ә-ni-m. argument-1.sg.poss please.pfv-3.sg.m.sbj-ben:1.sg-aff ‘(The people) liked my argument (and this is good for me).’ (8) B-ema yә-cona әrc tәrәssa-ni-m. in-way attr-sit.pfv[.3.sg.m.sbj] boy get.up.pfv[.3.sg.m.sbj]-ben:1.sg-aff ‘The boy sitting in the way stood up for me (e.g. that I can pass).’
Benefactives and malefactives in Gumer 
Since it does not seem reasonable to postulate malefactive equivalents of recipient and deputative benefactives, malefactives can largely be regarded as the reverse of the “plain” use of benefactives, i.e. as very general meaning that “something is bad or negative for someone”. Consider the following illustrative examples: (9) a.
T-ikә-na gad-ә-βi-m. child-1.sg.poss become.hungry.pfv-3.sg.m.sbj-mal:1.sg-aff ‘My child is hungry (and this is bad; I am worried).’
b. Gwәncә cәnә-βi-m. hyena come.pfv[.3.sg.m.sbj]-mal:1.sg-aff ‘A hyena came to me (which is dangerous).’ c.
Gәnә-na cәnә-xә-βi-m. land-1.sg.poss come.pfv-2.sg.m.sbj-mal:1.sg-aff ‘You came to my country (i.e. against my will; you attacked me).’
d. Bazәra cәnә-βi-m. guest come.pfv[.3.sg.m.sbj]-mal:1.sg-aff ‘A guest came to me (and this causes me some work).’ e.
I-n-jәppәr-xu-pi! neg-finish-2.pl.m.sbj-mal:1.sg ‘Don’t finish on me! (e.g. the food, before I come).’
2.2â•… Malefactive – locative – instrumental It has been mentioned that the malefactive has also the function of a locative and an instrumental. From the point of view of grammaticalization, however, it would be more appropriate to state it the other way round. The mal suffix -β/p is etymologically connected to the preposition bә-, which expresses both locations ‘in, at’ and instruments ‘with’. Examples (10) and (11) show the use of a locative and a instrumental respectively. (10) I-xa x-ira-wә! water put.imp[.2.sg.m.sbj]-mal:3.sg.m ‘Put some water in it!’ (11) Y-i-kәfwc-i-pwә. 3.sbj-open.ipfv.ips-mal:3.sg.m ‘One opens with it.’ (e.g. as an explanation for a bottle opener)
Without sufficient context the meaning of a mal suffix is ambiguous: it can be locative, instrumental, or malefactive: (12) K’-it’-ir-wә! kill.imp[.2.sg.m.sbj]-mal:3.sg.m ‘Kill [it] in it!’ or ‘Kill [it] with it!’ or ‘Kill [it] to his detriment!’
 Sascha Völlmin
3.â•… Core argument vs. non-core argument The examples so far showed that the ben and mal suffixes on the verb do not correspond to basic arguments in such a sense that they only express additional participants that are not sanctioned by the basic verb semantics. This criterion, which only concerns the verb suffixes and is not meant to say anything about the syntactic status of the suffixes or the correspondent overt NPs, is admittedly not very clear-cut.1 The point I want to make here is that the objects belonging to the semantic core of (a) transitive verbs (including experiencers) and (b) recipients and addressees of ditransitive verbs like aβә ‘give’, barә ‘say’ and odә ‘tell’ are all treated alike as obj (rather than ben or mal): (13) Wәxe k’ar-u bwar-ә-n--im. good k’ar-cop.3.sg.m.sbj say.pfv-3.sg.m.sbj-3.sg.m.obj-aff ‘He said to him “It is good”.’
Only in a few cases are there ‘core’ participants (or participants that could be conceived of as semantically obligatory) marked with ben on the verb. In the case of naxә ‘send’, e.g. the recipient is treated as ben (14a) due to the fact that the “object” being sent can also be a person, thus other than with the ditransitive verbs above a participant marked as obj will be understood as the theme rather than as the recipient (14b). (14) a.
F-irank nax-ni! money send.imp[.2.sg.m.sbj]-ben:1.sg ‘Send me money!’
b. F-irank nax-e! money send.imp[.2.sg.m.sbj]-3.sg.m.obj ‘Send me [with] money (e.g. to someone) !’
Another example is the verb t’afә ‘write’: here the primary object can only be the object being written, e.g. a letter, whereas the recipient of the letter has to be expressed as ben. (In this case ben can also be a real beneficiary, i.e. the one in whose stead the letter is written). Such instances can be described as follows: ben is also used for participants that cannot be marked as obj anymore, in case there is another participant that already occupies this position. Similarly, the verb dәwwәlә ‘call (by telephone)’ seems to occur with a ben “object”: (15) Dәwwәl-ә-ni-m. call.pfv-3.sg.m.sbj-ben:1.sg-aff ‘He called me.’
1.â•… Also note that I do not regard ben and mal as applicatives, the main reason being the fact that they are not obligatory. See Section 4.
Benefactives and malefactives in Gumer 
As a matter of fact, however, the original meaning of this verb – a loan from Amharic, where it shows the same argument structure – is ‘ring (a bell)’. Thus, the primary object is a bell and the person being called is the one for whom the bell is rung, which has to be marked as ben.
4.â•… Choice of object suffixes The three object suffixes obj, ben, and mal are mutually exclusive (cf. Example 1e). When more than one suffix is possible, one has to be chosen over the other. Usually the competing pairs are obj vs. ben or obj vs. mal. The primary object suffix is absent with indefinite (16), non-specific, or generic (17) objects (here, the object slot is empty and could be filled with ben or mal): (16) Bora ant’--inә-m. ox cut.pfv-1.pl.sbj-aff ‘We slaughtered an ox.’ (17) Kw-itara -ink’w-ira y-i-c’әn. chicken egg 3.sg.m.sbj-lay.ipfv ‘Chickens lay eggs.’
Definite objects are usually marked on the verb (albeit not always: it seems that the less animate/human and/or specific an object is, the more likely it is for the verb to occur without an object suffix). (18) Bet-xw-ita nәkәβ-o-yi-m. house-def find.pfv-3.pl.m.sbj-3.sg.m.obj-aff ‘They found the house.’
In sentences with a definite object and a beneficiary or maleficiary, the latter two seem to rule out the primary object: (19) a.
Astәmari-xw-it nәkәβ-o-yi-m. teacher-def find.pfv-3.pl.m.sbj-3.sg.m.obj-aff ‘They found the teacher.’
b. Astәmari-xw-it nәkәβ-o-lә-m. teacher-def find.pfv-3.pl.m.sbj-ben:3.sg.m-aff ‘They found the teacher for him.’ c.
Astәmari-xw-it nәkәβ-o-pwә-m. teacher-def find.pfv-3.pl.m.sbj-mal:3.sg.m-aff ‘They found the teacher to his detriment.’
If ben and mal are not present in the sentence as overt nominals (like in Example€19), the corresponding suffixes necessarily have to win over the direct object, because
 Sascha Völlmin
otherwise there would not be any expression of benefaction or malefaction at all. However, also with overt nominals the ben and mal suffixes usually override the obj: (20) T’ay-xw-ita y-әrc-әta s-iyә-lә-m. sheep-def y-son-3.sg.m.poss buy.pfv[.3.sg.m.sbj]-ben:3.sg.m-aff ‘He bought the sheep for his son.’
Yet it is also possible to mark the obj instead of the ben on the verb: (21) T’ay-xw-ita y-әrc-әta s-iyә-n--im. sheep-def y-son-3.sg.m.poss buy.pfv[.3.sg.m.sbj]-3.sg.m.obj-aff ‘He bought the sheep for his son.’
This variation has to do with information structure. In (21) the conversation seems to be about the sheep rather than the boy, it is more topical or more in the center of attention in some way, whatever this means exactly. Moreover, it is sometimes even possible to have no object marking at all, the explanation for which has to be left open here. (22) T’ay-xw-ita y-әrc-әta s-iyә-m. sheep-def y-son-3.sg.m.poss buy.pfv[.3.sg.m.sbj]-aff ‘He bought the sheep for his son.’
One might tend to see the suffixes as applicatives that increase the valency of a verb to introduce an additional participant. However, Examples (21) and (22) show that overt nouns denoting beneficiaries can also occur without the corresponding ben morpheme on the verb, and therefore it is not a required operation to change the valency. Usually beneficiaries seem to be human and also definite, thus the reason for the often coocurrence of overt NP and verb suffix has to be due to the specificity or definiteness of the referents than anything else – very much the same as it is the case with primary objects (see e.g. 18). (With maleficiaries the situation is a bit different, cf. Example 27).
5.â•… Marking of overt NPs Overt non-subject NPs can be prefixed with yә-. Generally, objects are not marked, whereas recipients and addressees always take the prefix yә-. (23) Rut zena-xw-it yә-gwәppay-әxyta od-әc-n--im. R. news-def y-brother-3.sg.f.poss tell.pfv-3.sg.f.sbj-3.sg.m.obj-aff ‘Rut told the news to her brother.’
Beneficiaries, like recipients and addressees, are also always marked with yә-. (24) Yә-gwәppay-әxyta basta ∫әkәt-әc-lo-m. y-brother-3.sg.f.poss pasta prepare.pfv-3.sg.f.sbj-ben:3.pl.m-aff ‘She prepared pasta for her brothers.’
Benefactives and malefactives in Gumer 
Additionally, animate definite objects can also be marked with yә-. Largely, this is for disambiguation in cases where an NP could be mistaken for the subject. In (25a), the marking of the hyena as an object is necessary; without yә- it would be understood as the subject (25b). (25) a.
Yә-gwәncә-xw-it k’wәt’t’әr-ә-n--im. y-hyena-def kill.pfv-3.sg.m.sbj-3.sg.m.obj-aff ‘He killed the hyena.’
b. Gwәncә-xw-it k’wәt’t’әr-ә-n--im. hyena-def kill.pfv-3.sg.m.sbj-3.sg.m.obj-aff ‘The hyena killed him.’
The marking becomes more obligatory the higher referents are on a hierarchy of humanness and definiteness. Proper names are virtually always marked with yә-, even if no disambiguation is needed, as in (26) where the subject is a first person. (26) Yә-Kәbbәdә k’әt’t’әr-xw--in--im. y-K. kill.pfv-1.sg.sbj-3.sg.m.obj-aff ‘I killed Kebbede.’
Maleficiaries can take two different prefixes: bә- and yә-. Among other readings the sentences in (27) mean that a bottle (=the subject) fell due to the father’s inattention or the like. (27) a.
B-abba-na t’әrmwiz wәt’әk’-ә-wә-m. b-father-1.sg.poss bottle fall.pfv-3.sg.m.sbj-mal:3.sg.m-aff
b.
Y-abba-na t’әrmwiz wәt’әk’-ә-wә-m. y-father-1.sg.poss bottle fall.pfv-3.sg.m.sbj-mal:3.sg.m-aff both: ‘A bottle fell due to my father’s inattention/My father (inadvertedly) dropped a bottle.’
In (27a) the maleficiary is marked with bә- and therefore should be considered as a peripheral argument, whereas in (27b) it is somewhat promoted and closer to being a core argument as it is marked with yә-, which also markes beneficiaries, recipients and some animate/definite primary objects.
6.â•… Benefactives with barә ‘say’ To express benefactives there is a construction that can be rendered literally ‘X do Y saying for Z’. The beneficiary is marked with yә- and occurs as the complement of the verb barә ‘say’ in the form of a so-called converb or a temporal subordinate verb with the prefix t- (‘when’). (28) Yә-xwta bar-o-m t’ay ant’-o-m. y-3.sg.m say-3.pl.m.sbj-cvb sheep slaughter.pfv-3.pl.m.sbj-aff ‘They slaughtered a sheep for him.’
 Sascha Völlmin
There does not seem to be a significant difference to the benefactive without the verb barә, although Crass & Meyer (2008: 243) call this construction, which also exists in other Ethiopian languages, “benefactive focus”. However, other than (28), their examples represent cleft sentences, which may explain the focal interpretation. Apart from possible differences in pragmatics, there are also syntactic reasons to express a benefactive with barә. This is the case when there is more than one participant marked with yә- (obj and ben), or when there is more than one beneficiary in one clause, which then becomes ambiguous if not ungrammatical. In (29) there is a beneficiary and a definite (human) object, both of which should be marked with yә-. (29) ???Yә-Kәbbәdә y-Abbәβә k’wәt’t’әr-ә-n-i/lә-m. y-Kebbede y-Abbebe kill.pfv-3.sg.m.sbj-3.sg.m.obj/ben:3.sg.m-aff (intended: ‘He killed Kebbede for Abbebe.’/‘He killed Abbebe for Kebbede.’)
One might force an interpretation onto this sentence (‘He killed Kebbede for Abbebe’ or ‘He killed Abbebe for Kebbede’), but the grammaticality of sentences with two yә- is highly in doubt. In natural speech such cases are expressed with the barә-construction: the beneficiary is detached from the main sentence and becomes the complement of (a subordinate form of) barә – in (30) as a temporal clause (lit. ‘when he says, saying’). (30) Arәgga yә-Kәbbәdә [y-Abbәβә A. y-Kebbede [y-Abbebe t-i-βir] k’↜wәt’t’әr-ә-n--im. when-3.sg.m.sbj-say.ipfv] kill.pfv-3.sg.m.sbj-3.sg.m.obj-aff ‘Aregga killed Kebbede for Abbebe.’
With this construction there is no ambiguity anymore, even though both arguments are still marked with yә-: the complement of barә is clearly the beneficiary, the other one being the primary object. Note that the object suffix on the main verb refers to the primary object rather than to the beneficiary, which as the complement of a subordinate verb is no longer in the scope of the main verb. Finally, if there is more than one beneficiary in a sentence, the construction with barә seems to be obligatory. In (31) the male proper name Kәbbәdә and the female ben suffix on the verb are intended to represent two distinct beneficiaries. However, the difference in gender causes a conflict that normally could only be solved if Kәbbәdә were interpreted to be the name of a woman instead of a man, as the verb suffixes reflect coreferentiality with overt noun phrases. Thus (31) is usually not acceptable as being a well-formed sentence. (31) ???K’awa-xw-it yә-Kәbbәdә s-iyә-x-la-m. coffee-def y-K.(m) buy.pfv-1.sg.sbj-ben:3.sg.f-aff (intended: ‘I bought the coffee for her in behalf of Kebbede.’)
Benefactives and malefactives in Gumer 
To solve this problem one has to remove the (overt) beneficiary from the main clause and put it as the complement of barә, which is in the form of a so-called converb in (32). (32)
K’awa-xw-it [yә-Kәbbәdә ba-xw--im] s-iyә-x-la-m. coffee-def y-K. say-1.sg.sbj-cvb buy.pfv-1.sg.sbj-ben:3.sg.f-aff ‘I bought the coffee for her in behalf of Kebbede.’ (more lit. ‘I bought the coffee for her saying “for Kebbede”.’)
Being in two different clauses, the verb suffix and the overt beneficiary are no longer coreferential, but there are clearly two independent beneficiaries.
6.â•… Conclusions Gumer—like other Gurage and Ethiosemitic languages—has three different sets of object markers on the verb: primary object (obj), benefactive (ben), and malefactivelocative-instrumental (mal). Generally, definite objects belonging to the core of a verb (including experiencers, recipients and addressees) are marked as obj, whereas ben and mal encode additional participants. The three object suffixes are mutually exclusive. If there is more than one potential candidate in a clause, usually ben and mal win over obj. There are also exceptions to this general rule, including the complete lack of object suffixes. Most probably, these variations are determined by information structure, but there might also be other reasons, which cannot be accounted for accurately so far. Overt NPs can be marked with the prefix yә-. Here, other than with the verb suffixes, beneficiaries and recipients/addressees are treated the same and always occur with yә-. Primary objects in a high position on a hierarchy of humanness and definiteness can also be marked with yә-. This seems to serve only the disambiguation of subjects and objects, but the obligatoriness of yә- with primary objects still deserves further investigation. Overt NPs denoting maleficiaries, on their part, can be marked either with yә- or bә-. Often the two prefixes seem to be in free variation, but some data not presented in this article suggest that the use of bә- is only possible when additionally to the malefactive meaning there is also some locative meaning involved or at least imaginable, whereas ‘pure’ malefactives occur only with yә-. As for the semantics, the suffixes ben and mal encompass the respective notion of benefaction and malefaction in a very broad sense and can in principle be used with every verb. ben expresses all shades of meanings like recipient, deputative, and “plain” benefactives, also including the expression of “successful accomplishment of an action”. mal, in addition to its locative and instrumental meanings, conveys all possible nuances of malefaction. Nevertheless, despite the fact that their use is very free, there might be some restrictions or special uses and meanings of ben and mal that need more clarification.
 Sascha Völlmin
Finally, there is a special means to express benefactives: the beneficiary can occur as a complement NP of a subordinate form of the verb barә ‘say’. On the one hand, this biclausal construction seems to be obligatory in ambiguous cases in order to clearly distinguish the beneficiary from a primary object or when there are two different beneficiaries in one sentence. On the other hand, it is also very frequently used when no disambiguation is needed at all. Although there does not seem to be a significant difference to the ‘normal’ benefactive, further research may still reveal some semantic or pragmatic differences.
References Crass, Joachim & Meyer, Ronny. 2008. Ethiopia, In A Linguistic Geography of Africa, Bernd Heine & Derek Nurse (eds.), 228–250. Cambridge: CUP. Lewis, Paul M. (ed.). 2009. Ethnologue: Languages of the World, 16th edn. Dallas TX: SIL International. Van Valin, Robert D., Jr. & LaPolla, Randy. 1997. Syntax. Structure, Meaning and Function. Cambridge: CUP. Völlmin, Sascha. Forthcoming. Some dialectal differences between Gumer and Chaha (Gurage). In Language Contact in Ethiopia, Joachim Crass & Ronny Meyer (eds.). München: Lincom
A “reflexive benefactive” in Chamba-Daka (Adamawa branch, Niger-Congo family) Raymond Boyd
Centre National de la Recherche Scientifique The “Nnakenyaare” lingua franca dialect of Chamba-Daka, a geographically and linguistically peripheral language of the Adamawa branch of Niger-Congo spoken in northeastern Nigeria, has not only beneficiary/recipient verb arguments but also a “reflexive benefactive” or autobenefactive marked by a copy pronoun from the possessive paradigm. To enhance typological understanding, Chamba-Daka argument structure is briefly described with focus on the importance of the animacy feature and the ambiguities arising from the existence of three different directly construed objects. The range of nonpossessive uses of the possessive paradigm, including the reflexive benefactive, is then set out. Conclusions include a discussion of the relations among autobenefaction, middle voice, and applicative derivation.
1.â•… Introduction Chamba-Daka (CD) is a peripheral member of the Adamawa branch of Niger-Congo. Various dialects of this language are spoken by between 100,000 and 200,000 people in Adamawa and Taraba States in northeastern Nigeria. This paper concerns the lingua franca dialect called Nnakennyaare.1 The purpose of this paper is to describe and define a pronominal usage current in CD, which I call “reflexive benefactive” (RB). The notion of benefactivity as characterizing a semantic argument can, of course, be conceived in more than one way. It can be taken very narrowly as contrasting with receptivity, the latter always associated with an argument of a particular class of verbs containing ‘give’. Again, it can be taken as an umbrella term covering both receptivity and various other forms of affectedness, and will then also be recognizable by association with the verb ‘give’. There is at least one more way of understanding benefactivity, namely, as an additional property assigned
1.â•… My work on this language has been done in collaboration with Isa Sa’ad, a native of the village of Gurum, Adamawa State. Thanks are due to the Nigerian National Commission for Museum and Monuments for their sponsorship.
Raymond Boyd
to an argument with another semantic role in the utterance, expressing the attribution of benefit or detriment to that argument in that role in the predicative context. This is the sense used for the RB. In CD, benefactivity as an additional property can only be assigned to a grammatical subject whose semantic role may vary but will be neither patient nor beneficiary. The formal mark of this property is a pronominal copy of the subject. The semantics of benefactivity in this use is much more difficult to define than in the case of benefactive arguments in the other senses. This is unavoidable, as the nature of the semantic roles of arguments is to a large extent induced by the governing predicative term, usually a verb. Benefactivity as an additional feature of another role is not defined by the predicate, but depends greatly on pragmatic factors. This leads to senses which may be extremely hard to capture in translation, and may be only hinted at or not attempted at all for some examples given below particularly since any translation will be out of context.2 Indeed, context alone can clarify the utterance features chosen to render the desired nuances in translation. A few characteristic kinds of contrast are cited in 3.5.2 below, but elsewhere the reader must accept or assume the benefactive sense. A discussion of the full range of senses would require much more space than is available here, and doubtless also require much more knowledge than we currently possess. It remains to be decided whether this type of benefactivity should be grouped together with middle verbs in a category different from that of primary benefactive arguments.3
2.â•… The semantic structure of verbal utterances The canonical CD affirmative verbal utterance contains a verb in the “absolute” (i.e., lexical) form with no TAM specification whatsoever.
2.â•… Actually, the import of the benefactivity in some of these utterances might not be transparent to other native speakers and would require interrogation of the original speaker to obtain a valid explanation, often a task with unpredictable results in a primarily oral culture. 3.â•… It is impossible to grasp fully what is involved in the CD phenomena described here without an overall view of argument structure in this language. I presented to the Zürich workshop a version of this paper whose first part dealt at length with pertinent aspects of this topic, while its second part described the range of uses of the possessive pronominal paradigm other than as a noun modifier, among them the RB. The restriction on overall length imposed by the publisher and the equitable allotment of space among the contributors prevents the full version from being presented here. It is nevertheless available on my website, rblanguesdafrique.info. Here I condense the discussion of the general setting (part 2) and focus on the RB (part 3). A conclusion (part 4) provides some comparative perspectives.
A “reflexive benefactive” in Chamba-Daka
(1) í k7¡¡7 g66n tím buu, (í saam sf¢) 3pl seek medicine tree pl 3pl find neg ‘they are looking/looked for a remedy (and they can’t/didn’t find one)’
The canonical order is SVO as in (1), and there must be a subject unless the referent is a known third person/thing, in which case S is ø. Animate subjects of all other persons must be marked by a subject index directly preposed to the verb. Only third-person inanimate nominal subjects are not followed by a subject index.
2.1â•… Transitive verbs The category of transitive verbs is defined by the canonical possibility (i.e., the logical structural possibility) of taking three direct (unmarked) objects, hence SVOOO. These objects are syntactically direct, i.e., they are unmarked nonsubject arguments; but each object has a distinct semantic role which is associated with relative position in the utterance. In the first object position stands the benefactive object (BO); in the second, the patient object (PO); and in the third, the object I have called “relational” (RO). This last object designates that “with respect to which” the process takes place. The semantic role of the BO may be either that of beneficiary or that of recipient stricto sensu. Thus, (2a) is constructed in the same way as (2b): (2) a.
waa-kù túm cook-3sgBO foodPO ‘[he/she] cooked staple food for him/her’
b. nyaa-kù túm give-3sgBO foodPO ‘[he/she] gave staple food to him/her’
In the most general semantic terms, the BO is the participant who is affected or concerned by the process without undergoing it. There would seem to be a strict complementary distribution of the animacy feature between the animate BO and the inanimate RO. It could thus conceivably be argued that this is the sole semantic distinction required and that in CD the two are otherwise semantically identical, whereby the RO would be “the inanimate argument which is affected or concerned by the process without undergoing it”. If it were not for this possible complementary distribution (requiring confirmation from a much larger corpus) and the role of the RO with intransitive verbs, the latter might be treated as a peripheral argument or semantic adjunct. Owing to ambiguities arising from noun phrase structure (where again semantic features are assigned to slots in word order and segmental marking is almost totally absent), utterances with a sequence of three direct nominal objects are extremely rare in practice. In the strictest sense, not one canonical utterance of this type is
Raymond Boyd
attested in the present corpus. There are, however, a few noncanonical examples, such as (3):
(3) í mà Ù. nwúù n6¢t pak7n tíi 3pl fut U. wifeBO bagPO put in:inf headRO ‘they put a bag on U.’s wife’s head’
Ambiguity (e.g., an interpretation which would have nәʹt tíi as ‘top of bag’) is avoided here by transposition of the BO and PO to a position before the verbonominal form used in this future construction. If other objects are present, the RO cannot undergo this transposition. Pronominalization of an O provides another means of avoiding ambiguity. Of the three possible objects, one and only one may be pronominal rather than nominal. This pronominal is enclitic to the verb, and the word order constraints remain. In other words, if the BO is pronominalized, it may be followed by two nominal objects (PO and RO). If, however, the PO is pronominalized, there can be no BO and any following object is necessarily the RO. If the RO were to be pronominalized (there are no attested cases of this and, if the RO must indeed be inanimate, it would presumably be exceptional or impossible, as there is no proper inanimate enclitic), there would be no following object.
2.2â•… Intransitive verbs The “intransitive” verb category is defined by the ability to take at most two objects, which are necessarily BO and RO excluding a PO. A third category of verbs is made up of intransitives derived by suffixation. These derivatives which I call “resultatives”, can take no more than one inanimate direct object, the RO. This use of the term “intransitive” may initially seem strange; nevertheless, it is easy to see that the transitive/intransitive contrast in CD rests solely on the preÂ� sence (transitive) or absence (intransitive) of a PO. This corresponds to the ordinary sense of this contrast. The peculiarities in CD are that both the BO and the PO are direct (i.e., unmarked), and that there can be a third nonoblique object. There is no principled reason why any verb should not take its full number of direct objects, although pragmatics determine that most do not or do so only rarely. Hence, there is no good reason to speak of ditransitive or even tritransitive verbs as if these were grammatical categories.
2.3â•… Expansions Transitivity is defined for CD as inherent patient-object valency; intransitivity, as absence of this valency. For both kinds of verbs, beneficiary and relational objects are expansions. Let us look now at an instructive example of an intransitive verb which
A “reflexive benefactive” in Chamba-Daka
can take BO and/or RO expansions, làà ‘stay, remain’. It has ordinary objectless uses with and without locative complements: (4) í làà r7 Yoolà d7´¡7 n háá 3pl stay loc town dem until ‘they stayed long there in Yola’ (5) `n làà `n k7¡¡7 nyáa-r7¡7 ? 1sg stay 1sg seek what-q ‘what am I staying around for?’, lit. ‘I stay I seek what?’
Yet an optional BO can be found with this verb:
(6) àán pέn àán laà-m àán dem thing dem stay-1sgBO dem ‘that is the thing that remains for me (to do)’
There are also various expressions in which the same verb takes an object which is sematically relational rather than a beneficiary: làà súún yàà, stay | husband | compound, ‘be a married woman’, lit. ‘stay husband’s compound’; or làà gàng, stay | chief, ‘be chief, occupy the chiefship’, etc. Hence, with both BO and RO:
(7) laa-bú gàng stay-3plBO chiefRO ‘he ruled as their chief ’, lit. ‘he stayed as chief for them’
There is, however, another important and problematic usage of làà, viz., làà láam ‘(to) sleep’, where it has only its “internal object”4 láam ‘sleep’ as an argument. A priori, it would seem better làà still be considered intransitive with láam as its relational object. Some support for this choice might be gained from examples such as: (8) làà tunsì púrùm láam stay send.out fatigue sleep ‘he slept to get rid of his fatigue, he refreshed himself by sleeping’ (lit. ‘he lay and removed fatigue through sleep’)
Here láam can easily be construed as RO to both làà and tunsì púrùm. At the same time, làà púrùm láam ‘fall asleep from fatigue’, lit. ‘lie sleep of fatigue’, is also a valid expression, another classic case of the ambiguity, adumbrated in 2.1 above, arising from the juxtaposition of two nominals outside S position. While (7) ‘be chief for them’ is quite natural and provides a good prima facie example of an intransitive verb taking both BO and RO, the native speaker strongly 4.â•… This is the nominal associated with a verb root to express a specific “unitary”, often “generic” sense. In languages in this region, the internal object is often etymologically related to the verb as in this case.
Raymond Boyd
resists expanding làà láam by adding a BO to obtain an utterance such as *laakù láam ‘he slept for him’. This resistance might, however, be attributed rather to semantic strangeness and inapplicability than to strict agrammaticality. But there is still another analytical possibility available, viz., that some nonderived verbs are such that they take a single obligatory object which saturates the object structure. These verbs would form utterances with S V O structure where O is an object with no contrastive semantic value (unlike resultative verbs, which are assumed to be intransitive and take a RO but no BO). The PO/RO contrast would become inapplicable in this case and hence, the categorization of the verb as transitive or intransitive as well.
2.4â•… A consideration of the saturator object interpretation If this interpretation were to be adopted, other verbs which resemble internal object constructions in taking a very limited range of objects might be categorized in the same way. Thus, jùt ‘bathe, take a bath’ which requires an object denoting a liquid, usually w¢ffk ‘water’ but sometimes, for example, things such as g66n tím, medicine | tree, ‘herbal decoction’. This verb behaves exactly like làà in not taking a beneficiary (indeed, what would ‘taking a bath to someone else’s benefit or detriment’ exactly imply?), but differs from làà in never appearing (outside the reduplicative intensive construction) without an object. For this reason, it is impossible to demonstrate any original intransitivity. This categorization nevertheless faces a major obstacle with examples such as: (9) a à w¢ffk jurεn wúu wèè jé pát 2sg fut water wash:inf body 2sgPoss just all ‘you must wash your whole body’
Here wúu wèè ‘your body’ seems clearly to stand as RO, so that w¢ffk ‘water’ as inanimate must be the PO (barring the possibility of two ROs), and jùt ‘bathe’ is therefore transitive. Hence, it must be assumed that there is both a category of obligatory single-object verbs which are used exclusively with a specific inanimate object, and a category of transitive verbs which are used only with a semantically limited range of objects. The former are originally intransitive, like làà ‘stay’, and have a specific sense when used with their “default object”. They share with the latter and with resultatives the property of inability to take a BO. The question now is, Are there verbs with ordinary transitive usages which exclude a BO when they take an internal or semantically restricted object? There are quite a few transitive verbs taking any one or more of the three canonical objects, e.g., d¢fp ‘set (fire to something)’ (10a), which can also be used with an internal object (10b): (10) a.
Y7´¡7 p n7¡ bu í d¢fp wùù d7´¡7 n Yebbi person pl 3pl set.fire room dem ‘the Yebbi people set fire to that house’
A “reflexive benefactive” in Chamba-Daka
b. í d¢fp yísi d7´¡7 n gáà dàà sè 3pl set.fire fire dem loc down first ‘first they get the fire burning inside [the hole]’
In (10a), the object is unrestricted: one can burn anything inflammable. In (10b), the object is restricted: one can only get fire burning. While neither usage is found with a BO in the available corpus, such utterances can be elicited for both without difficulty. These cases of “generic object” are thus distinct in their behavior from specialized transitive verbs such as jùt ‘bathe’ and indeed, from idiomatic V O expressions of a “metaphorical” nature, cf. tiksì ‘take, bring, put, set down (a load)’, the causative derived from tìk ‘go, come down’, but tiksì púú ‘have a miscarriage’ (púú ‘belly’). A more systematic study of the lexicon would be required to learn which expressions are compatible with a BO. Nevertheless, if discursive practice and not strict agrammaticality is involved, such a study will yield indeterminate results, as some expressions may allow a BO under unusual semantic conditions which do not immediately come to the mind when the native speaker is consulted.
2.5â•… Semanticosyntactic ambiguity Let us examine in greater detail the causatives derived from intransitive verbs, for example laksì (irregularly from làà ‘stay’) ‘lodge (someone in a place); lay (something down or on its side); install (someone in a position), give (someone a title)’. This verb gives rise to a new problem of grammatical interpretation in the event (7) is assigned the structure (S) V BO RO. In (11), does gàng remain a RO as it stood to the underived verb, whereby the pronominal may be interpreted as either a BO or a PO? If a BO, must the causative be considered to remain intransitive like the base verb? Or does gàng now become a PO? (11) í laksì-kù gàngì 3pl install-3sg chief:real ‘they made him their chief ’
The V BO RO solution leaving laksì as an intransitive verb seems undesirable. In its other uses, laksì is clearly transitive and little is to be gained by categorizing it Â�differently in (11). The problem is then whether V BO PO or V PO RO is the “right” choice. In fact, there is no good test of which syntactic interpretation should apply, not only with causatives derived from intransitives, but even with some transitive base verbs. Indeed, there is nothing to distinguish, for example, (12) a.
pirì-kù mum put-3sg mouth ‘he notified him’, lit. ‘he put him an announcement’ (BO PO ?)
Raymond Boyd
from (12) b. gur-kù mànáà seize-3sg friend ‘he befriended him’, lit. ‘he took him as a friend’ (PO RO ?)
other than a perceived semantic distinction, which may be psychologically very real, but remains a tenuous basis, to say the least, for assuming a syntactic contrast.5 To make a choice on formal grounds, we would need to have some context in which POs and ROs were treated differently. If one such exists, it is subtle, and I have not yet come upon it. The simplest procedure would, of course, be to elicit from a speaker both an utterance adjoining a RO to (12a) and a refusal to adjoin a RO to (12b) since mànáà already is the RO. Excluding the possibility of an RO nevertheless presents the same problem as excluding a BO in object saturation: How can it be proved that the speaker’s negative response is grammatically rather than pragmatically grounded? An even more serious problem is created by utterances where, unlike (12), the psychological reality of the distinction is blurred. There are indeed expressions where a fundamental syntactic indeterminacy seems to arise because their semantic content is not sufficiently precise for one interpretation or the other to prevail indisputably. Take the case of káng ‘watch over, guard, protect’: (13) í mà m7´´7m bu kangεn kàà ràán 3pl fut children pl guard:inf as dem ‘they take care of the children thus’
There is another expression using the same verb with two objects: (14) kang-kù wáa b7´¡7 tíi guard-3sg hand mod:with head ‘he protected him from attack’
which may be construed either as ‘spread protectively his handPO for himBO over his head’ or as ‘protected himPO by use of his handRO over his head’, the first being the simplest (perhaps the “default”) double-object construal and the second being the direct expansion of the construction with only the animate PO. Metaphor certainly contributes to the opacity of syntactic interpretation; likewise with many idiomatic expressions, as can be seen from the question of the proper syntactic interpretation of ordinary utterances like (15).
5.â•… It may be noted that both verbs have corresponding reciprocals gurk¡7n mànáà ‘befriend each other’ pirk¡7n mum ‘invite each other’.
A “reflexive benefactive” in Chamba-Daka
(15) í dfm-kù láam 3pl greet-3sg sleep BO PO ‘they greet him’
Indeed, this utterance could easily be assigned the structure S V PO RO: literally ‘they greet him with respect to sleep’. Nevertheless, given the use of the expressions df¢f¢m láam ‘greet (people)’ and láam dffmεn ‘(fact of) greeting (people)’, láam appears much more like the PO, with the person greeted as BO. The structure with RO expansion thus seems to revert easily to the more “compact” BO PO structure, whatever the “logical” interpretation of the expression may be.
2.6â•… Summary Here is a summary of the major features of CD semanticosyntax which have been set out above and are pertinent to the understanding of the reflexive benefactive: –â•fi –â•fi –â•fi
absence of a beneficiary/recipient contrast in the canonical utterance structures; close association of the beneficiary/recipient with the semantic feature of animacy; uncertainty regarding the syntactic role, hence the semantic content, of the first nominal/pronominal object in any sequence of two objects.
In this system where the three object positions are so rarely filled, the BO PO sequence seems to have preferential status with transitive verbs. It is more “compact” than any other sequence, both in the sense that two nominals in BO PO position are semantically closer to the Modifying Noun – Modified Noun sequence used in CD, and in the sense that the RO is a “different kind” of object from the other two. Hence, provided the PO be animate, the PO RO sequence, even when etymologically justified, is easy to reinterpret as BO PO. Furthermore, though the fact is hard to prove, the BO PO sequence may well exercise a kind of attraction on the BO RO sequences found with intransitive verbs, inviting crossover from the intransitive to the transitive category. One further remark is required to make explicit the system of CD verb derivational morphology, which has been evoked on several occasions. This morphology includes suffixes marking the pluractional and the causative (with some overlap, given that the causative suffix is sometimes used with pluractional sense). There is also a “resultative” expressing being or becoming and a reciprocal (formally the resultative of the pluractional), both of which yield intransitive verbs. All these derivatives are semantically definable and productive. All other derivational suffixes are, however, semantically imprecise (though some sets of semantically related verbs can be found with each one), unproductive, limited to certain phonological environments, and often used for verbs for which no base verb is attested. We can thus note that there is no proper applicative derivation in CD.
Raymond Boyd
3.â•… Reflexive benefactives The CD reflexive benefactive (RB) is a pronominal paradigm formally identical with the possessive paradigm, which is as follows:
1 mὲ wòò
2 wèè bèè ~ vèè
3 kèè bèè
mèè bèè
3.1â•… Possessives as copulae In addition to its proper possessive use, as for example in (9), this paradigm has two, even three, “secondary” uses. Thus, in the absence of a subject index, the third person singular possessive can be seen as a copula or as a presentative of a following noun phrase with the sense ‘this is a matter of, this concerns, this means’, e.g., (16) p7´n àán pfk súnnà sf¢, pfk fàríllà sf¢ ràn, thing dem be(come) tradition neg be(come) obligation neg dem ‘something that is neither tradition nor moral obligation kèè pén àán kàà t7´´7m wffn wèè cop thing dem like heart want:inf 2sgPoss can be defined as something that depends on your free will’ (17) mínε7´-n ny7´´7n nwúù bu b7´7n ´7¡7 tèè gà, LogsgIdp-Logsg see wife pl mod mod.loc there top ‘(said,) “I’ve seen some women out there, and kèè n7¡-dffk7´n bu p7´n sf¢, gàng bu p7´n cop person-commoner pl thing neg chief pl thing this is not a matter for commoners but for royalty”’
First- and second-person copulae require the presence of a pronominal subject. (18) `m m7¡ n7¡-nòòní 1sg cop person-one ‘I am a solitary, isolated person’
The second person singular copula is found without extension for the presentation of a noun phrase (19), but takes an -n extension for the presentation of a circumstantial phrase (20): (19) s¢fgà n wèè l7´rùm sf¢ otherwise 2sg cop male neg ‘otherwise, you are not a man’ (= ‘unless you are a man’) (20) wii míí dá/án bóó, n wèen wàà vis7´7n kóó kóbò tèè bé 2sgIdp child dem top 2sg cop:ext with ask:inf or kobo there neg ‘you, that child, cannot ask for so much as one kobo’
A “reflexive benefactive” in Chamba-Daka 
Comparable examples are available for the full pronominal paradigm including the logophoric. According to person and number, certain formal variations involving the presence and/or the form of the extension appear in the copula usage, for reasons which are as yet unclear. It is, however, manifest that this usage is in complementary distribution with the RB to which we now turn.
3.2â•… Possessives as reflexive benefactives As the discussion in 3.1 shows, when the possessive pronominals are used as copulae with subject indices or even nominal subjects, they have attributive sense. The attribute is generally nonverbal. Use with verbal propositions as attribute occurs exceptionally, but these should certainly be interpreted as having undergone nominalization. When possessive pronominals are postposed to the verb group in a verbal utterance, one obtains the RB. The construction is reflexive because the possessive form is coreferential with the subject of the verb. It is benefactive (or malefactive) in the sense that the referent of the subject is affected in a particular way by the verb process. If “referent of the subject” is replaced by “referent of a nonpatient object” in this broad semantic definition, we obtain a viable definition of both the object of the applicative verb form in Niger-Congo (or at least in many Benue-Congo and some Adamawa-Ubangi languages) and the BO in languages like CD. In other words, the RB effectively assigns the same semantic feature to the subject as the BO role does to an object argument. If the semantic charge of the RB had been expressed by some kind of verbal morphology, this usage would doubtless have been more appropriately called a “middle voice”.6 It must, however, be observed that the range of semantic features designated crosslinguistically by the latter expression is very large and not limited to autobenefaction. If there is a subsuming semantic function of the middle, it may be (in utterances with at least one animate denotatum) to convert the agent (sometimes diathetically) into an experiencer; or, if the agent retains control over the process, to attribute to the agent the additional role of experiencer; or to the experiencer, the responsibility of an agent. This function can be taken as a semantic transposition of “affectedness of the subject as the essential characteristic of the middle”, an affirmation attributed by Mous and Fufa (in preparation) to Klaiman (1991). The autobenefactive is nevertheless apparently peculiar within the generality of the middle sense. Thus, Mous and Fufa (in preparation), for whom “body orientation is central in the meaning of the derivational middle”, note that the autobenefactive is like an epidemic: once a language community has been exposed to the option of including this semantic sense in the meaning of the
6.â•… The literature also refers to an “autobenefactive”, an acceptable alternative.
Raymond Boyd
middle marker (through contact with a language for which this sense is productive for the corresponding middle marker), it expands on it and the middle marker [becomes] productive for this meaning.
In Eastern Cushitic where there is a morphological autobenefactive, Mous and Qorro (2000:166) quote Hayward (1975: 209) in referring to indirect or autobenefactives middles, the category that Hayward in his study on Eastern Cushitic middle voice calls “the middle-voice function” par excellence”.
These remarks suggest that the autobenefactive is not just immediately integrated into the semantics of the middle, but can even take it over entirely. Furthermore, in languages like CD where the RB is not a part of verb morphology, the idea of the affectedness of the body as the core meaning of this structure is far from evident. Some doubt about the middle as semantic monolith therefore seems justified. We shall return to the semantics of the RB in 3.2.5 and more generally in part 4 after a succinct formal presentation.
3.2.1â•… Syntactic properties Unlike the BO, the RB is compatible with derived -7¡n resultatives: (21) g¢ffn tun7¡n kèè nàà póó s7¡ sáng gà 3sgIdp go.out 3sgBen in grass first again top ‘whenever he goes out into the bush once again’
A third-person RB may be associated with an inanimate subject: (22) bùm kú dák7´n kèè s7¡ place 3sg clean:inj 3sgBen first ‘it has to dawn beforehand’
Although propositions with a RB and no more than one direct object are more frequent, the RB is compatible with the presence of a both a BO and a PO. (23) nak-á wèè nòòní míí sin do-1plBO 2sgBen onePO small just ‘do7 just one short one [i.e., sing a short song] for us’
The RB is not considered a suffix because, unlike object pronominals, it can be moved to preinfinitival position, where it stands before any preposed PO or BO, in the future form. Indeed, this is by far the preferred position: (24) `m mà m7¡ nyìín wáa kurk7´7n sin 1sg fut 1sgBen now hand clap:inf just ‘I’ll now just applaud him’ (= ‘I’ve decided just to go along with him, not contradict him’)
7.â•… The CD imperative, like the English one, has no segmental representation of the 2sg subject.
A “reflexive benefactive” in Chamba-Daka
Postposition is nevertheless also observed: (25) mín77´-n mà vallì-k¡ffn mèè LogsgIdp-Logsg fut help-2/3sg:inf LogsgBen ‘(said,) “I’ll give you a hand, I’m willing to help you”’
The semantic import of the RB may prevail over blocks of discourse resulting in a serial usage: (26) Kóó à dàa JàngPúlì máá, kóó da KúrúmJíi máá, `m màà or loc down:loc JP top or on KJ top 1sg go ‘Down this way to JangPuli or even up to KurumJii, I couldn’t make m7¡ sf¢. `N g7¡t m7¡ gà, `n dùk m7¡ tii B. baan. 1sgBen neg 1sg go 1sgBen top 1sg finish 1sgBen at B. farm it. If I try to go, I can get no farther than B.’s farm. G7rúm m7¡ dùk kèè ¡7 tii B. baan. movement 1sgPoss finish 3sgBen loc at B. farm My journey stops there at B.’s farm.’
3.2.2â•… Morphological features: -n extensions The RB often bears a -n extension, but this seems in general to be a copy of another morphological marker in the same utterance. Indeed, CD has several markers of the form -(7)n bearing one of the three contrastive level tones, among them the infinitive and the subject focalizer suffixes, both of which bear mid tone, though the formation of the infinitive also requires neutralization of the lexical tone of the verb root. Both of these suffixes can be found copied on RBs: (27) exemplifies copying from the focalizer and (28), from the infinitive marker. (27) mínε7´-n vεt-k¡ffn mèen LogsgIdp-Logsg leave-2/3sg:foc LogsgBen:ext ‘(said,) “I’m the one who is leaving you”’ (28) í mà baan bèen 2pl fut come:inf 2plBen:ext ‘you people will come’
3.2.3â•… Morphological features: Absence of agreement An additional interesting phenomenon involving the RB is the use of a 3sg in association with a 3pl subject. It is easily assumed (though there is no strict proof of the hypothesis) that this usage handles situations in which the 3pl subject index has no specific referent but means ‘one, everyone’: (29) s¡f¡fn gà, í mà kèè nyìín pir7´7n tii páá dance top 3pl fut 3sgBen now return:inf at grounds ‘for dancing, everyone now goes back to the meeting grounds’
Raymond Boyd
3.2.4â•… The reflexive benefactive with object agreement? In 3.1, I referred to a possible third secondary use of the possessive pronominals. Indeed, the 1sg and Logsg possessives often directly follow the object pronoun form of the same persons. This usage is extremely frequent, and is conceivably euphonic in origin. (30) Súu kú vεr-ùm m7¡ rì S. p7¡ God 3sg leave-1sg 1sgBen with S also ‘may God grant life to me and [my child] S.’ (31) ´7n, pàná, kú tíí-m mèè jé that please 2/3sg plait-Logsg:inj LogsgBen just ‘(she) said, “Please just plait [my hair] for me”’
These pronominals also occur, certainly analogically with the RB, with -n extension as in (32). (32) -ὶ juk-úm mèén yàà tèè 2pl:imp look.at-Logsg LogsgBen:ext compound there ‘(said,) “Watch my compound for me”’
Other object pronouns, particularly 2sg, may nevertheless appear in the same construction as in (33): (33) kóó nak-à wèè p7´n àán haaní sf¢fn /gà pát, n7¡-wàrí or do-2sg 2sgBen thing dem be.proper neg:dem top all person-big ‘even if he behaves improperly towards you, he’s still your elder’
Other instantiations are rare or unattested, cf. with 3pl: (34) n¡fk gà, Súu kú nak-bú bèè báràkà 1sgIdp top God 3sg do-3pl 3plBen happiness ‘for my part, [I say] may God bless them’
For the time being, it is hard to know exactly what semantic import should be accorded to this usage, but it is conceivable that at least in some cases, it senses as a variant of the RB, hence stands for kèè in (33–34). Another possibility is, however, that the possessive pronominal is developing as a mark of the BO to counter some of the uncertainties of syntactic interpretation described in part 2 of this paper.
3.2.5â•… Semantic properties The examples presented thus far give a sampling of the semantic range of the RB. Let us look at a couple more examples with some common verbs—sát ‘say’ (35–36), nyíí ‘know’ (37), nyέέn ‘see’ (38), báá ‘come’ (39), and nyangsì ‘spoil’(40)—and contrast them with the same utterance without RB to see exactly how nuances are added.
(35) (36)
A “reflexive benefactive” in Chamba-Daka
`n sar-à m7¡rì 1sg say-2sg 1sgBen:real ‘I assure you’ (rather than ‘I say to you’) n sát wèè g¡f, n k77 n ny77¡-m tèè sf¢ 2sg say 2sgBen pfv 2sg seek 2sg give-1sg there neg ‘you promised [me], but you didn’t go and get me [any]’ (rather than ‘you told me…’)
The RB expresses the subject’s commitment to the content of his discourse. (37) n¡fkiì-n nyii m7¡ pέn d7´¡7 n gáà nàà àán sf¢, 1sgIdp-1sg know 1sgBen thing dem loc in dem neg ‘I don’t know what happened there, ´n wupsì wáa r7´ da p7´n 3pl hide hand mod on thing people are keeping quiet about things’ (‘I haven’t been able to find out’ rather than simply ‘I don’t know’)
The RB expresses frustration of the effort to acquire information (and whatever benefit may derive from that information). (38) `n ny7´´7n m7¡ n7¡´7 r7 yìsá gà, t7n7´n F. sf¢, sf¢ dúkà 1sg see 1sgBen person loc door top pass F. neg neg all ‘except for F., I’ve never seen any of you at my door’ (‘at my door as visitors’, not simply ‘in my doorway’)
The RB expresses a seeing which brings pleasure or advantage. (39) kù `m máà sar7n `n nóó, I. d7´¡7 n à kèè baan fct 1sg fut:fct say:inf 1sg say I. dem fut 3sgBen come:inf ’that makes me say to myself, I. will come’ (‘and when he does, I’ll find out the truth’ rather than simply ‘he will come’ with no implication)
The RB expresses the appropriateness of the subject’s arrival, not necessarily to his own benefit but rather to the advantage of those who await him. (40) ´n nàk p7´n dáan gà, í nyangsì kèè p7´n 3pl do thing dem top 3pl spoil 3sgBen thing ‘it’s a waste of money’ (‘if people do that, they do it to their own detriment’, not simply ‘they spoil something or other’)
The RB expresses the malefactive effect of the action on the agents.
4.â•… Concluding remarks: Benefactives, the applicative and the middle voice It is generally accepted that Proto-Niger-Congo (PNC) had derivational verb extensions (Hyman 2007:151). These would have included an “applicative verbal derivation”
Raymond Boyd
(often taken to be synonymous with “benefactive”), included among the “valencymodifying” derivatives, i.e., those which add (or remove) a direct object argument in a particular semantic role. This view of the applicative is grounded on Â�phenomena observed in the Bantu languages. My own experience drawn mainly from the Â�Adamawa-Ubangi branch of NC suggests a rather different conception of the applicative, viz., that its primary function was not the addition of a syntactic argument but rather exclusively semantic: it allowed nuancing of the base verb with the sense ‘specialization of the process for a particular circumstance or purpose’. This type of applicative derivation is found (though not necessarily as a NC inheritance) in Zande, a Ubangi language, where there is a recipient/beneficiary contrast marked by different prepositions. In this language, for example, pe ‘speak’ has a derived pluractional pek‘speak repeatedly’, which itself has a derived applicative peked- ‘tell, narrate’, i.e., speak at length for a specific purpose. Likewise, pas- ‘cook’ (formally causative but without a base verb) has a derived applicative pasad- for which a variety of senses are reported. I was told it was used for ‘make a decoction’. The Gores (1952) have ‘heat up, cook twice’, while Lagae and Van den Plas (1921/25) give ‘cuire d’une manière incomplète pour empêcher la corruption’. All of these definitions are specific kinds of “cooking” done for particular reasons in specific circumstances. The specialization of the process can sometimes be accounted autobenefactive as in the example from Boyd (1995) cited by Hyman (2007): gbe ‘pull, attract, stretch out, brandish, suck’, whose derived applicative gbed- has an intensive-type meaning, ‘grasp, hold on strongly (including in sexual relations), suck, smoke’. Another case is the intransitive ug- ‘be dry, dry out’ whose applicative ugud- remains intransitive with the sense ‘be thin, slim down’. In none of these cases is the Zande applicative (which I have previously called benefactive, Boyd 1995) a valency-modifying morpheme; its effect is purely semantic. The applicative derivation, despite its apparent usefulness, is nevertheless less productive in Zande than the pluractional and the causative. Furthermore, it is semantically imprecise: the intensive sense often predominates to the extent that some formal applicatives are used as the sole pluractional derivative of their base verb. The very fact that the applicative was somehow semantically distinct from other extensions may thus have contributed to its rarification or demise. In many NC languages, for example some in the Bantu A zone (the ones closest to CD), it seems to have fallen into total disuse. In other cases, for example Emai (an Edoid Benue-Congo language of Nigeria, Ron Schaefer, p.c.) and the central, eastern, and southern Bantu languages (Hyman 2007:157), the applicative seems to have been “conserved” by being semantically denatured and transformed into a mere valencyaugmenting marker of the beneficiary/recipient argument. Still more complex phenomena exist, as Mous (2006) shows for Bantu A44 Tunen, which seems to have retained the applicative for adding valency while developing a
A “reflexive benefactive” in Chamba-Daka
“middle derivation” which is semantically indistinguishable from a new applicative, showing even examples of the drift towards the intensive sense observed in Zande. Nevertheless, the combined use of the applicative and middle markers allows for a form of diathesis whereby an animate beneficiary object or, in the absence of animate objects, an inanimate patient object can become syntactic subjects. These data suggest that the NC applicative as understood here is semantically parallel to the Cushitic middles as described by Mous and Fufa (in preparation) just insofar as its general sense ‘do in a specific way, to a specific end’ takes on the specifically autobenefactive nuance ‘do for one’s own ends, to one’s own advantage’. Syntactically speaking, however, the valency-augmenting type of applicative is closer to the diathetic functions of the Cushitic middles whose use is not limited to activities whereby an animate argument affects himself but extends to uses where notional inanimates (as body parts) affect animate benefactives (their possessors). A fuller understanding of these parallels may come from further study of the structural interplay which may exist within derivational systems. In Niger-Congo, the Tunen case is already an eloquent example, and a closer examination of theÂ� Atlantic languages from this standpoint may prove helpful. Mous and Fufa (in preparation) look at the interrelations of passive, middle, and causative verb morphology in Cushitic. The interplay of derivational verb morphology with case marking and the Â�conflation or contrast of beneficiary and recipient arguments is likewise of cruÂ�cial importance. Against this background, the specificity of the CD phenomena discussed here lies in showing the autonomy of autobenefaction, which is only one possible nuance of NC applicatives and Cushitic middles. They suggest that autobenefaction can be a perceived semantic need which is satisfied independently of any morphological system. The main features to be noted in CD are: 1. the association of a reflexive benefactive with conflated beneficiary/recipient arguments; 2. the co-ocurrence of a reflexive benefactive with only the most vestigial applicative verb derivation of the Zande type (a half dozen verbs suffixing -lì which have a possible semantic affinity with base verbs); 3. the use of the possessive paradigm to express the reflexive benefactive in verbal utterances, contrasting with a copula-like use in nominal utterances. The third point is of some importance as the Adamawa (“Kebi-Benue”) language Tupuri may have a similar reflexive benefactive structure using not the possessive pronouns but a pronominal paradigm reserved for this function, referred to by Ruelland (1992:194–5) only in distributive terms as “BC3” and glossed as ‘(lui)-même’. Further
Raymond Boyd
examples from Fiorio (2007) suggest the third person singular member of the paradigm is commonly used for nominal predication, in addition to its apparent autobenefactive use in verbal propositions. A comparable use of a specific pronominal paradigm as a reflexive benefactive may also be present in the closely related language Mambay as described by Anonby (2008). All of these cases must of course be seen in the perspective of the well-known Intransitive Copy Pronoun in Chadic and similar features in non-Chadic languages which have not yet been identified as such (Zande in the Ubangi subbranch constitutes an example). The fact that such constructions are restricted to intransitive verbs suggests limitation to a middle-verb usage. In terms of regional influence on CD and Chadic, the existence of a middle voice in Adamawa Fulfulde should also not be ignored. Attested extension to transitive constructions is thus far limited to Adamawa. The insufficiently studied languages of the Adamawa Plateau and the contiguous regions may yet reveal other pertinent phenomena for the semanticosyntax of benefaction.
Abbreviations used in word-for-word glossing 1,2,3 2/3 Ben cop dem ext fct foc fut imp pl Poss
first, second, third person pfv perfective nonspeaker in logophoric Idp independent pronoun propositions inf infinitive reflexive benefactive inj injunctive copula loc locative demonstrative Log logophoric pronoun extension mod modal marker factitive neg negative focalizer q polar (or redundant) future interrogative imperative that opener of reported speech plural top topicalizer possessive sg singular
References Anonby, Erik John, 2008. Mambay (Niger-Congo, Adamawa): Phonology and Morphology. Ph.D. dissertation, University of Leiden. Boyd, Raymond, 1995. De l’expression à l’expressivité en morphologie: Analyse comparée de la dérivation verbale en zande et en nzakala. Afrikanistische Arbeitspapiere 43: 5–36.
A “reflexive benefactive” in Chamba-Daka
Fiorio, Elisa, 2007. L’étranger chez les tupuri du Tchad: Étude ethnolinguistique de la notion de l’étranger d’une population de langue Adamawa. Thèse de doctorat, Institut National des Langues et Civilisations Orientales (INALCO). Gore, B.C. & Gore, Mrs. B.C. 1952. Zande and English dictionary. London: Sheldon (rev. edn by H.P. Bullen). Hayward, Richard J. 1975. Middle voice verb forms in Eastern Cushitic. Transactions of the Philological Society 1973: 203–224. Hyman, Larry M., 2007. Niger-Congo verb extensions: Overview and discussion. In Selected Proceedings of the 37th Annual Conference on African Linguistics, Doris L. Payne & Jaime Peña (eds.), 149–163. Somerville MA: Cascadilla Proceedings Project. Klaiman, Miriam H. 1991. Grammatical voice. Cambridge: Cambridge University Press. Lagae, C.R. & van den Plas, V.H. 1921/25. La langue des Azandé, 3 Vols. Ghent: Veritas. Mous, Maarten, 2006. Voice in Tunen: The so-called passive prefix bé-. In Evidence and Counterevidence, Festschrift Frederik Kortlandt, Vol. I [SSGL 32–33], Alexander Lubotsky, Jos Schaeken & Jeroen Wiedenhof (eds.), 1–10. Amsterdam: Rodopi. Mous, Maarten & Fufa, Tolemariam. In preparation. The derivational middle in African languages. Mous, Maarten & Qorro, Martha. 2000. The middle voice in Iraqw. In Lugha za Tanzania, Languages of Tanzania. Studies Dedicated to the Memory of Prof. Clement Maganga, Kulikoyela K. Kahigi, Yared Kihore & Maarten Mous (eds.), 157–176. Leiden: CNWS. Ruelland, Suzanne. 1992. Description du parlet tupuri de Mindaore, Mayo-Kebbi (Tchad): Phonologie, morphologie, syntaxe. Thèse pour le doctorat d’état ès-lettres. Lille: Atelier national de reproduction des thèses.
Beneficiary and other roles of the dative in Tashelhiyt* Christian J. Rapold
Max Planck Institute for Psycholinguistics This paper explores the semantics of the dative in Tashelhiyt, a Berber language from Morocco. After a brief morphosyntactic overview of the dative in this language, I identify a wide range of its semantic roles, including possessor, experiencer, distributive and unintending causer. I arrange these roles in a semantic map and propose semantic links between the roles such as metaphorisation and generalisation. In the light of the Tashelhiyt data, the paper also proposes additions to previous semantic maps of the dative (Haspelmath 1999, 2003) and to Kittilä’s 2005 typology of beneficiary coding.
1.â•… Introduction This paper explores the semantics of the dative in Tashelhiyt with a focus on beneficiaries, the dative being the main strategy to encode beneficiaries in the language. A range of semantic roles is treated such as possessor, experiencer, distributive and judicantis, and a semantic map of the dative in Tashelhiyt is proposed. Further, the semantic links between the prototypical function of the dative, recipient, and the other functions are described. Based on the typology of beneficiary, recipient and recipient-beneficiary encoding in Kittilä (2005) and the semantic maps of the dative in Haspelmath (1999, 2003), the paper proposes additions and amendments to both lines of research in the light of the data discussed here. Finally, it is argued that Tashelhiyt is a neutral Â�language with respect to Kittilä’s (2005) typology. Tashelhiyt is spoken in South Morocco by six to eight million people, which makes it the biggest Berber (Afroasiatic) language. Basic typological facts relevant for the present topic include a basic VSO constituent order, anaphoric agreement on the verb and a nominative-accusative alignment system. The language has a two-term case *I would like to thank Maarten Kossmann and Harry Stroomer for valuable comments on earlier versions of this article.
Christian J. Rapold
system with a marked nominative and an unmarked accusative case (also known as construct state/état d’annexion and free state/état libre respectively (Sasse 1984)). The nominative case mainly marks postverbal subjects and the complement of most preÂ� positions, while the accusative is mainly used for direct objects, preverbal subjects and the citation form. Adverbial relations are often encoded by prepositions, one of which is traditionally called “dative” in Berberology.1 This article has the following outline: Section 2 presents a description of the morphosyntax of the dative, followed by an account of the semantics of the dative in Section 3, which forms the main part of the article. Specifically, Section 3 addresses the main semantic types of Tashelhiyt beneficiaries in the light of Kittilä’s 2005 typology. Section 4 summarises the main results of the paper. An appendix, finally, provides some statistical data on the use of the dative in Tashelhiyt.
2.â•… Morphosyntax of the dative As in all major Berber languages, dative marking is the primary strategy to mark beneficiaries/maleficiaries in Tashelhiyt. A comparable phenomenon is also found in Moroccan Arabic and other Arabic varieties (Caubet 1993; Brustad 2000). Consider the following Tashelhiyt illustration:
(1)
Agr- i-fsi=d 3sm-untie./=
izikr .rope
i-mqqur-n, …2 =his.brother=. .-be.big.-. ‘He untied the rope for his elder brother,…’(#74) (context: The elder brother had just let him down into a well on a rope.) 2 i=gwmas=lli
1.â•… Even though it designates a clitic preposition and not a bound morphological case strictly speaking, the traditional term “dative” will be retained in this paper. As will be shown below, this label fits well semantically. 2.â•… Orthographic conventions: j = IPA [Š], š [∫], > [t], x [χ], ε [↜ô], h [ħ], other consonants: ˙ C [Cô], r [n], y [j], CC [Câ•›˜]. Following Stroomer (1994), simple clitics (such as the dative i=) ˙ and clitic PPs are represented as separate orthographic words. Glossing conventions: Verb stems are glossed morphologically; the Aorist stem, which is unmarked for aspect, is taken as the unmarked form and is not glossed. – The translations of example sentences are taken from the indicated sources and are only occasionally adapted to be more literal.
Semantic roles of the dative in Tashelhiyt
The dative is encoded by different allomorphs depending on the type of complement it governs. These types are lexical noun phrases (as in (1)), pronouns, and gaps, which leave the preposition stranded. Preceding lexical complements, the dative is realised as i, before pronominal complements as a, and as a stranded preposition it is realised as mu or mi. The following examples illustrate the use of a dative preceding a pronominal complement (2) and a gap (3): (2) aylli> a-s t id iwi-n… (pronominal complement) when dat-3s do.3sm vent bring.pfv-3pm ‘When they had brought it to him,…’ (#88)
Prepositional phrases with pronominal complements follow complex placement rules that are described in detail in Dell and Elmedlaoui (1989). Personal pronouns governed by a preposition are expressed by a special paradigm of suffixes. There are also free-standing, pragmatically marked pronouns, but in Tashelhiyt (unlike e.g. in Kabyle (Chaker 1983:250–253)) these cannot be governed by a preposition (Galand 1990:281). (3) ukan ur ssn-> mai [mu Øi (gap complement) then neg know.aor/pfv-1s who/what â•›dat rad fl-> tasarut]. fut leave.aor/pfv-1s sf.key ‘I don’t know to whom I must leave to key.’ (#249) (lit.: …the one [to whom I…].)
Stranded prepositions occur only in non-subjective relative clauses. Traditionally, combinations of a dative preposition and a pronoun have been seen as “indirect pronouns” in their own right, parallel to the direct pronouns (e.g. Aspinion 1953; Boumalk 2003; Mountassir 2003). A further analysis of the dative markers has been proposed by Galand (1969, 1988 i.a.), but this will not be dealt with here. In this paper, along with Stumme (1899), Bentolila (1981) and Dell and Elmedlaoui (2002), I view the marker a in dative phrases as a preposition, but this choice has no direct Â�consequences for the further analysis presented below.
2.1â•… Clitic doubling of lexical NPs Lexical dative noun phrases may be doubled by a co-referential clitic dative phrase in the same clause (4). Clitic doubling is found with about 7% of all lexical dative NPs in
Christian J. Rapold
the corpus investigated (55 out of 788 tokens. More details on the corpus are given in Section 3). Its precise semantic and discourse pragmatic factors remain to be investigated. The clitic dative PP and the lexical dative NP do not have to be adjacent, but may be separated by other constituents. (4) clitic dat PP lexical dat PP ls-nt a-si i tsliti ilbad nn-s,…. ˙ put.on.clothes-3pf dat-3s dat sf.nom.bride clothes gen-3s ‘They clothe the bride with her clothes,…’ (#139)
2.2â•… Sentence-level dative phrases A further syntactic feature of dative phrases is that they are not restricted to the clause level, but can also occur at the sentence level, in which case they pertain semantically to the illocution rather than the predication or the proposition. Consider the following example, where the sentence-level dative phrase is separated from the rest of the sentence by a comma and even precedes the vocative summoner: (5) i rbbi, a hnna, ma t-gi-t? ˙ dat God voc my.dear who/what 2-be.pfv-2s ‘Please (tell me), oh dear, who are you?’ (original translation) (#258)
This usage of the dative has been found with only one word in the corpus, rbbi ~ rbbi ˙ ‘God’. Note that even if such phrases were synchronically analysed as interjections, they would probably still derive from sentence-level dative phrases diachronically.
3.â•… Semantics of the dative Although beneficiary is a frequent semantic role of dative phrases in Tashelhiyt, it is by no means their only one. For a better understanding of the beneficiary in this language, therefore, it is necessary to know with what other roles beneficiary has to compete in the interpretation of a given dative phrase. This section attempts to sketch the whole semantic range of the dative in Tashelhiyt as reflected in the corpus, with special focus on the beneficiary and its closely related roles. Sections 3.1–3.6 present the main semantic types of beneficiaries and closely related categories. Sections€3.7–3.16 are devoted to the other roles of the dative. The data in this section is drawn from a corpus of approximately 83,000 words, containing more than 4,000 datives. The corpus is taken from published texts (Amard 1997;
Semantic roles of the dative in Tashelhiyt
Stroomer 2001a, 2001b, 2002) and comprises several literary genres (for a review of Stroomer 2001b see Rapold 2005).3 Various semantic definitions and characterisations of the dative are found in the literature. In a pioneering article on the dative in Polish, Wierzbicka (1986:420) tentatively proposes the following cross-linguistic semantic formula for the dative:
(6) “something happened not because person Z wanted it one could assume that person Z would think/feel/know something because of that”.4
Wierzbicka adds that to “ensure that [this] description of the dative has full predictive power it is necessary to specify the precise set of semantic formulae which can generate no more and no less than the entire range of dative sentences acceptable in a given language” (1986:420). For Polish, she lists no less than thirty-one such formulae, with headings such as “causing to see” and “accidental changes in appearance”. To establish such an exhaustive list for Tashelhiyt goes beyond the scope of this paper, since in combination with lexical dative phrases alone (788 tokens in the corpus), 162 different verbs are found. While there will always remain a need for such lists, the various items can be grouped into larger units of semantic roles which range somewhere between the very general formula in (6) and the very specific ones just mentioned. A list of such semantic roles of the dative in Tashelhiyt is given in Figure 1, organised in the form of a semantic map which builds on a map of the dative proposed by Haspelmath (2003)5 (for the theoretical rationale and problems of semantic maps, see Anderson 1982; Croft 2003; Haspelmath 2003; de Haan 2004; Cysouw forth. i.a. A more elaborate version of the Tashelhiyt map is found in Figure 3 below). The Tashelhiyt findings do not contradict Haspelmath’s semantic map, but suggest a number of additional functions that should be incorporated. These roles are preceded or underlined by dotted lines in Figure 1. Their position in the network is tentative or remains to be established on the basis of cross-linguistic evidence.
3.â•… The corpus includes 62 folktales, fairy tales, legends, 27 (short) animal fables, 11 other stories (jokes, etc.), 33 ethnographic texts, 14 poems, and 55 riddles (consisting of single sentences). 4.â•… This definition is limited to non-lexically governed uses of the dative (Wierzbicka 1986:425f.) 5.â•… An earlier version of the same map (Haspelmath 1999) does not contain purpose, but includes addressee, which is conflated with the role of recipient (“recipient/addressee”). This suggests that the position of addressee in the network is yet to be established by further cross-linguistic data. The earlier version also includes the role of ethical dative, linked to experiencer.
Christian J. Rapold
possessor
direction
recipient
beneficiary/ maleficiary/ general affectee
judicantis
purpose
experiencer
––– ––– ––– ––– ––– –––
- source-maleficiary substituted-assisted concretely benefited unintending causer distributive [2 subroles]
locative Figure 1.╇ A semantic map of the dative in Tashelhiyt (I) (built on Haspelmath 1999, 2003)
The various roles in Figure 1 form a network and can be derived from one another through mechanisms of semantic change, the semantic core of the dative being extended along various dimensions. The main mechanism of extension involved is generalisation, a process of semantic change by which some features of the original meaning are faded out while the original role is still properly included in the semantic specification of the more general, derived role. Which of the roles in Figure 1 will be selected for a given dative phrase depends on the cotext, the context and the knowÂ� ledge of the world of the addressee. As evidenced in the data below, almost all roles of the Tashelhiyt dative can be substituted by the semantically very general role of general affectee (Section 3.8). However, for the sake of cross-linguistic comparison and in view of the “precise set of semantic formulae” called for by Wierzbicka in the quote above, it is worth exploring which specific semantic interpretations the dative can have in Tashelhiyt. In the following, each role is briefly commented and illustrated and its link to other roles briefly accounted for. The latter attempt is mainly based on Geeraerts (1998). recipient is taken as the starting point of the description, since at least around 66% of all dative phrases in the corpus are recipients/addressees (or recipient-beneficiaries) (cf. the Appendix, Section 3). Moreover, this is fairly generally assumed to be the prototypical role of the dative (Van Langendonck 1998:212, 218). Wierzbicka (1986:386), for instance, claims that the semantic core of the dative can be captured by the following semantic formula:
Semantic roles of the dative in Tashelhiyt
(7) “X did something with thing Y wanting person Z to come to have it something happened to Y because of that one could assume that Z would come to have Y because of that”
This characterisation of the semantic core largely coincides with Geeraerts’ more tangible prototype of a “recipient of a benefactive material transfer” (1998:189), although for the present purposes recipient and beneficiary are of course to be disentangled. Note also that the synchronic prototype is not necessarily diachronically older than the more peripheral roles.6 The description further first treats the roles that (typically) pertain to human or animate referents, since these constitute the overwhelming majority of all dative phrases in the corpus: about 75% of all lexical datives are animate; in pronominal and gapped dative complements the percentage is even higher (cf. Appendix). This is also largely in line with Wierzbicka’s definition of the dative cited above (6), which explicitly limits the dative to humans.
3.1â•… Recipient Entities carrying the semantic role of recipient can be defined as “animates that receive something concrete transferred to their sphere of control” (Kittilä 2005:274). Whether the reception itself is beneficial or detrimental to the recipient is irrelevant. The following examples feature a basic (8) and a derived verb (9) with a recipient: (8) lli> a-s tn t-fka,…. when dat-3s do.3pm 3sf-give.pfv ‘When she had given them [=clothes] to him,…’ (#172) (9) i-tahl tt wuday lli mu 3sm-marry.pfv do.3sf sm.nom.Jew res.ref dat t-zz-nza-m tiglay. 2-caus-be.sold.pfv-2pm pf.egg ‘The Jew [to whom you used to sell eggs] married her.’
The semantic links emanating from this semantic core of the dative are discussed with the relevant roles below.
3.2â•… Beneficiary Based on Kittilä (2005:273), I define the beneficiary as an entity for which the occurrence of an event is encoded as being beneficial. This is illustrated by the following
6.â•… For this reason, the arrows indicating the directionality of diachronic semantic change in Haspelmath (2003:234) have been omitted in Figure 1.
Christian J. Rapold
example, where the beneficiary, the bride, is thought to enjoy (i.e. benefit from) the fact that the people sing. (10)
(context: The people go out and bring out the bride to the place of rejoicing.) a a-s sharraw-n mddn. ˙˙ ipv dat-3s sing.ipv-3pm pm.people ‘The people sing for her.’ (#138)
While for Kittilä (2005:273) the beneficiary may not be coreferential with the agent, others distinguish between allocentric and autocentric (or: autobenefactive) benefaction. Whatever terminology one prefers, both categories are encoded with a dative in Tashelhiyt (10, 11), the former being much more frequent. (11) bxxr-nt a-snt s lbxur … (autocentric) burn.incense.aor/pfv-3pf dat-3pf inst incense ‘They burn odoriferous plants for themselves,…’ (context: preparation of a ritual during Ramadan) (#143)
Note also that the event involving a beneficiary may happen spontaneously, with an inanimate subject or force as the grammatical subject: (12) sul a ya-k i-ttujad unzar, a yafllah, … ˙ ˙ still comp dat-2sm 3sm-be.ready.ipv sm.nom.rain voc sm.peasant ‘The rain is not yet ready for your benefit, peasant,…’ (#198)
From a cross-linguistic perspective, two notable characteristics of benefactive constructions in Tashelhiyt are that the beneficiary can be inanimate (13, 14), and the benefactive event itself may be stative (14): (13) i-bnu tigmmi lli, i-skr a-s 3sm-build sf.house res.ref 3sm-do/make.aor/pfv dat-3s yat ttaqa. ˙˙ f.one small.window ‘He built a house and made a small window in it.’ (#16) (lit.: …made a small window for it.) (14) (‘A basket full of “Sair” dates,’)
i-li ya-sn yan bu_fqqus > wammas, 3sm-be.located dat-3pm m.one B. in sm.nom.middle ‘there is one Boufeqqous date in their middle’ (a riddle) (#183) (lit.: ….for them in the middle.)
3.3â•… Recipient-beneficiary Recipient-beneficiaries combine the roles of recipient and beneficiary. Like the beneficiaries, they occur in allocentric and autocentric uses:
(15)
Semantic roles of the dative in Tashelhiyt
arki> a-sn t-ssnu-t mad ššta-n; … (allocentric) until dat-3pm 2-cook-2s who/what eat.ipv-3pm ‘until you cook something to eat for them;…’ (#49) (context: ‘you’ = the mother, ‘them’ = her children)
(16) (‘There is no fraternity here, the one that finds something’) (autocentric) y-af t i ugayyu nn-s. 3sm-find do.3sm dat sm.nom.head gen-3s ‘finds it for himself.’ (#221)
Kittilä (2005) presents a typology of the encoding of beneficiaries and the closely related categories of recipient and recipient-beneficiary. Depending on how these three categories pattern formally in a language, languages can be divided into beneficiaryprominent, recipient-prominent, tripartite, fluid and neutral languages. In the first two cases, recipient-beneficiaries formally pattern with beneficiaries and recipients respectively, while in the last case, all three categories are encoded in the same way. As evidenced by the above Tashelhiyt data, all three semantic categories are encoded in the same way, as far as their main encoding strategy—dative marking—is concerned. This classifies the language as a neutral one in Kittilä’s 2005 typology. Given that the position of recipient-beneficiary in the semantic map of the dative remains to be established on the basis of cross-linguistic evidence, its semantic links to other roles are not specified here. The same holds for the other roles that are preceded or underlined by dotted lines in Figure 1.
3.4â•… Subroles of beneficiary The beneficiary role (Section 3.2) subdivides into two roles, concretely benefited beneficiary and substituted beneficiary.
3.4.1â•… Concretely benefited beneficiary Concrete benefaction refers to “pure” benefaction, neither implying that the beneficiary does not have to carry out an event (as in substitutive beneficiaries), nor that there is a concrete result that is received by the beneficiary (as in recipient-beneficiaries). The event in question is simply portrayed as benefiting the beneficiary, e.g. by the fact that he can “use the result for his/her purposes” (Kittilä 2005:273). This role is illustrated by (10) and (12) above. 3.4.2â•… Substituted beneficiary With substitutive benefaction, the beneficiary does not have to carry out the profiled event himself, in this sense benefiting from the occurrence of the event. Whether the result of the event itself is beneficial or not, is less important (Kittilä 2005:273).
Christian J. Rapold
Consider the following example, in which the subject, the master of a servant, does not have to heat his own water because his servant does this job for him: (17) (‘He had a (black) servant who’) ar a-s i-ss-r>a aman n lud u. ˙ ipv dat-3s 3sm-caus-be.warm pm.water gen ritual.ablution ‘heated water for his ritual ablutions.’ (#87)
Note that Tashelhiyt does not distinguish whether substitutive benefaction completely frees the beneficiary from carrying out the task from whether he is merely significantly assisted. In this sense, it would be more appropriate to speak of substitutive-assistive benefaction. This is illustrated by the following example, where the protagonist, a jackal, will not have to do the threshing all alone, as the others do part of it on his behalf: (18) (‘↜“I have some threshing to do, give me a helping hand.” […]’) i-dalb a-sn tiwizi ad a-s srut-n ˙ 3sm-ask.aor/pfv dat-3pm sf.assistance comp dat-3s thresh.aor/pfv-3pm ‘he asked their assistance with threshing.’ (lit.: …that they thresh for him.) (#117)
Substitutive benefaction and recipient-beneficiaries At this point, it may be useful to add a note on the characterisation of the recipientbeneficiary in Kittilä (2005:275), according to whom “substitutive benefaction is a crucial part of Rbs [= recipient-beneficiaries (CJR). Italics by CJR].” However, based on the following example (and many others in the corpus), this claim seems to have to be revised: (19) ar a-s i-ttxxi umušš lli ddhb. ipv dat-3s 3sm-excrete.ipv sm.nom.cat res.ref gold ‘the cat excreted gold for him.’ (#14)
It is clear that the protagonist in (19) was not expected to excrete gold, an action that is then done by the cat on his behalf. Rather, the benefaction involved in this event is one of concrete benefaction. This suggests that while there are clear cases of recipient-beneficiaries involving substitutive benefaction, as Kittilä rightly observes, there is another subtype of recipient-beneficiaries that involves concrete benefaction. Â�Recipient-beneficiaries thus mirror the basic distinction between substitutive and concrete Â�benefaction found in beneficiaries.
3.5â•… Maleficiary Maleficiaries are the conceptual counterpart of beneficiaries, representing an entity to whose detriment an event is encoded as occurring. Whether a dative phrase is interpreted as a beneficiary or a maleficiary depends purely on the cotext, the context and the general knowledge of the world. In the following example, these factors make it
Semantic roles of the dative in Tashelhiyt 
clear that the death of someone in the family is seen as a negative event for other members of the family: (20) (‘What is the matter with you, that you weep like this, oh king?’) […] ur a-k i-mmut yan z> dar-k? neg dat-2sm 3sm-die.pfv m.one from at-2sm ‘Has any member of your family died?’ (#215) (lit. ‘Hasn’t one from those with you died to your detriment?’)
As with beneficiaries, the subject may be an inanimate noun acting as force in a spontaneous event: (21) af-n d tilist, t-qqn a-sn (-anim sub) find-3pm vent sf.snow 3sf-close.aor/pfv dat-3pm imi n tgmmaw. sm.mouth gen pf.nom.house ‘They find the snow blocking the entrances of their houses.’ (#154)
3.6â•… “Source-maleficiary” A semantic counterpart of recipient-beneficiaries is constituted by what could be termed “source-maleficiaries”, which combines the role of source/ablative with that of maleficiary in the same way as recipient-beneficiaries combine the roles of recipient and beneficiary. Consider the following example where something is stolen from the affectee: (22) ar d i-ttasi tiglay > ddu ugdid ann. […] ˙˙ ipv vent 3sm-take.ipv pf.egg in under sm.nom.bird demdist ar tnt i-ttasi > ljib nn-si i sarq-nnhari. ˙ ipv 3pf 3sm-take.ipv in pocket gen-3s dat Day-Thief. ‘[Day-Thiefi] took the eggs away from under that bird. [Night-Thief followed him and] took them from hisi pocket.’ (without him noticing it) (#35)
This example also nicely illustrates how specific semantic features trigger the presence or absence of dative marking. The data sentence features two possessors, a bird and a thief, from whom the same eggs are stolen. Even though both possessors are animate, only the human one is signalled by dative marking as being indirectly affected by the event.
3.7â•… Unintending causer The dative also marks entities that unintentionally cause an event. This is the case in (23), where bread is made to fall by unintended causation of the animate protagonist. The causer, which is adversely affected by the event, is encoded in a dative phrase.
Christian J. Rapold
On a cross-linguistic basis, it is useful to separately identify this type of states of affairs, but from a language internal perspective it is possibly best seen as a special case of the maleficiary role (3.5). (23) (context: He took [one loaf of bread] in order to eat it. Just when he wanted to take a bite, the horse reared and gave him a fright.) t-dr a-s tngult ti>wrdin. (-anim sub) ˙ ˙ 3sf-fall.aor/pfv dat-3s sf.nom.loaf.of.bread pf.behind ‘The load of bread fell down behind to his detriment.’ (#30)
3.8â•… General affectee According to Haspelmath (1999:113) “it seems that we cannot easily conceive of ‘neutral’ affectedness that is neither negative nor positive.” In the corpus, however, it is rather frequently difficult to decide whether an event is benefactive or malefactive for an affected entity. This is the case in (24), where the target of the action, a jacket, is clearly affected by the whistling, but otherwise seems to be indifferent as to whether it is a benefactive or malefactive action. In other cases, a dative phrase with plural reference may refer to a group of people for some of which the event is benefactive, while for others of the same group it is malefactive. This suggests that it is more appropriate to use the category of a general affectee that encompasses the notions of both Â�beneficiary and maleficiary. (24) (The king took the jacket and was amazed to see) ad ukan a-s i-sins:, ar i-ttrqas. ˙ ˙ comp then dat-3s 3sm-whistle.aor/pfv ipv 3sm-dance.ipv ‘that if somebody whistled at it, it began to dance.’ (#84)
As noted above, the role of general affectee is semantically so general that most other roles of the dative in Tashelhiyt could be subsumed under it (though at the cost of meaningful cross-linguistic comparison of language specific Datives). The role that conceptually most clearly resists this is the distributive (3.15). Semantically, general affectee, beneficiary and maleficiary are derived from recipient by metaphorisation involving the shift from material to abstract transferred entities (Geeraerts 1998:190, 193).
3.9â•… Judicantis The dativus iudicantis or judger’s dative encodes the entity according to which some state of affairs is evaluated.
Semantic roles of the dative in Tashelhiyt
(25) i> t-dr tasswikt ad […] ˙ if 3sf-fall.aor/pfv sf.(nom).nut.tree demprox.sp t-a::u: a>. 3sf-be.far.aor/pfv dat/do.1p ‘If this nut tree falls, it will be too far away for us [to be able to chop it up].’ (#326) (26) mad a-k y-uf-n: xtann n>d xti? who/what dat-2sm 3sm-surpass.pfv-sub.rc sf.dem or pf.dem ‘Which do you prefer: the [girl] here or the ones here [= paradise girls under the speaker’s armpit]?’ (lit.: Which is better according to you:…) (#89)
At this point it may be noted that the reference point of a comparative construction is not marked with a dative, but with a range of other means, including the preposition f ‘on’, the direct object of the verb af ‘surpass’, and the possessor of the noun ugg(w)ar ‘more than’. Semantically, the judicantis role depends on that of general affectee and derives from it through metaphorisation from an affected to an evaluating entity.
3.10â•… Addressee Dative phrases with the role of addressee are very common in the corpus, mainly in combination with the verb ni ‘say’, but also with εawd ‘tell, do again’ and some less frequent verba dicendi. (27) t-nna a-s: “a uddi, i-šša t uwtil!” 3sf-say.pfv dat-3s voc my.dear 3sm-eat.pfv do.3sm sm.nom.hare ‘She said to him: “Oh dear, the hare ate it!’ (#234)
Although the exact position of addressee in the semantic map is yet to be established by cross-linguistic data, the semantic process that could derive addressee from recipient is one of metaphorisation: the recipient of a material transfer is metaphorised into a recipient of a communicative transfer (Geeraerts 1998:193).
3.11â•… Experiencer Experiencers are typically coded as subjects (28), but dative phrases occur with some verbs (29): (28) t-frh tm>art lli, … ˙˙ 3sf-be.happy.aor/pfv sf.nom.woman res.ref ‘The woman was happy, …’ (#84)
Christian J. Rapold
(29) “iwa, mad a-k i-dhr-n?” […] “ur ˙ well who/what dat-2sm 3sm-appear.aor/pfv-sub.rc neg yyi i-dhr yat: ….” ˙ dat/do.1s 3sm-appear.aor/pfv f.one ‘Well, what do you think? […] I don’t think anything: …’ (#230)
The semantic link from recipient to experiencer (to locative) is constituted by generalisation with respect to the amount of control that the dative-marked referent can exert over the transferred entity. As a recipient, he actively receives it and takes it into his sphere of control, while as an experiencer, he passively undergoes the effect of the transfer and has no control over it.
3.12â•… Locative The dative is not often interpreted as having a locative role in the corpus and the precise semantic constellations that evoke it remain to be investigated. Alternatively, the locative in such cases could be analysed as an inanimate general affectee that happens to be a location (or Ground, in Talmy’s terminology (Talmy 1991)). This option is of course ruled out if the affectee is conceived of as animate by definition. Consider the following examples: (30) y-awn i udrar ann. 3sm-climb dat sm.nom.mountain that ‘He went up that mountain.’ (on a horse) (#21) (31) asi-n kra n idammn, g-n tn i lεtbat. take-3pm a.little gen pm.blood put-3pm do.3pm dat sf.doorstep ‘They take some blood and put it on the doorstep.’ (context: ritual during the construction of a house) (#157)
The position of locative in the semantic map is debatable and has to be decided on the basis of cross-linguistic evidence. It could be conceived of being linked via generalisation to experiencer, representing entities that are still lower on the agency and control hierarchy. As inanimates, they are merely affected by the transfer event, acting as some spatial ground (Talmy 1991) for the event.
3.13â•… Direction Direction is generally marked by the allative/instrumental preposition s as in (32). The dative is used only with a couple of verbs, such as huwd ‘go down’ (33). (32) i-ddu s ssuq. 3sm-go allat market ‘He went to the market.’ (#23)
Semantic roles of the dative in Tashelhiyt
(33) i-huwd ufrux lli i yan lxla… 3sm-go.down sm.nom.boy res.ref dat m.one desert ‘The boy went down to a desert…’ (#54)
The extension from recipient to direction involves a generalisation in the spatial dimension of the prototype: the notion of transfer of an entity is faded out, but the orientation of the movement is maintained (Geeraerts 1998:194–195).
3.14â•… Possessor Haspelmath (2003) distinguishes between the functions of predicative and external possessor in his semantic maps, the relevant parts of which look as follows: predicative possessor
external possessor
recipient
beneficiary
Figure 2.╇ Predicative and external possessor in Haspelmath (2003)
I conflate the two, since “predicative” and “external” are (semantico-)syntactic, not semantic concepts and as such have no place in a semantic map. However, for the sake of comparison with Haspelmath’s pioneering work, I treat both cases separately in the following. In both cases, the semantic process leading from recipient to possessor is one of generalisation by which the transfer part of the prototype definition (including the orientation of the movement) is bleached. Consequently, the focus simply lies on the fact that some entity is in the in the sphere of control of some other entity. Note, finally, that it remains to be proven whether the bona fide “external possessor” is really primarily about possession. This question is addressed in Section€3.14.2 below.
3.14.1â•… Predicative possessor The predicative possessor is usually marked by the preposition dar ‘at, next to (French chez)’ (35). Use of the dative to mark this function is rare and seems to be confined to verbless predications (34). Alternatively, dar ‘at next to’ can be used to predicate Â�possession in verbless sentences as well (35). (34) ikkattinn yan urgaz, ism a-s ihya. ˙ there.was.once m.one sm.nom.man sm.name dat-3s I. ‘There once was a man, his name was Ihya.’ (#222)
Christian J. Rapold
(35) i-lla yan ugllid, dar-s kratt lbnat, … ˙˙ 3sm-be.located.pfv m.one sm.nom.king at-3s three pf.daughter ‘There once was a king, he had three daughters.’ (#215)
3.14.2â•… External possessor In external possessor constructions, the possessor does not occur in the NP that is headed by the possessee, but externally as a constituent of the clause in its own right. (What some would analyse as) such constructions are common in Tashelhiyt, the possessor being encoded as a dative phrase. This is noteworthy, for according to König and Haspelmath (1997; cf. also Haspelmath 1998, 1999:109–110, 116), dative external possessors are a characteristic feature of the European linguistic area, and Haspelmath (1999:119) claims that “the non-Indo-European languages just outside Europe Â�generally have only NP-internal possessors[.]” (italics by CJR).7 Consider the following examples, where the possessee is a direct object and the possessor is encoded in a dative phrase: (36)
i-bbi i ul:wmad sbεa walsiwn, … ˙ 3sm-cut.aor/pfv dat sm.nom.snake seven pm.nom.tongue ‘He cut out the seven tongues of the snake,…’ (#271)╇ (dat = n, do = body part) (lit.: He cut the seven tongues to the detriment of the snake.)
(37)
εla rbbi, ad a-k n-kks alln! ˙ by God comp dat-2sm 1p-take.away.aor/pfv pm.eye ‘We surely have to put out your eyes.’ (#26)â•… (dat = pron., do = body part) (lit.: …put out the eyes to your detriment.)
Such constructions alternate with cases where the possessor is encoded NP-internally by a possessive pronoun suffix or by means of a genitive phrase (36’ below). However, it is questionable whether in the so-called external possessor construction possession is really the main semantic role of the dative phrase. Impressionistically, its role is rather that of a general affectee (beneficiary/maleficiary), while ‘possessor’ is at best an implicature that can be cancelled. Affectedness, by contrast, cannot be cancelled in cases like (36). Whether this hypothesis is indeed correct for Tashelhiyt is a subject for further research. One argument against viewing “pure” possession as the main
7.â•… Haspelmath has the dative case in mind, but the formal and semantic properties of the dative preposition in Tashelhiyt would seem to justify its inclusion here, all the more as Haspelmath himself includes the Maltese dative marker lil- as an example of dative external possessor marking. This marker has been analysed as a preposition by Maltese specialists (Vanhove (1993:424), Borg and Azzopardi-Alexander (1997:61, 103, 143)).
Semantic roles of the dative in Tashelhiyt
semantic role of the dative in such cases is the fact that the dative may co-occur with a possessive affix in the same construction (Šarić 2002:16; for Tashelhiyt data see below (41, 42). For the sake of cross-linguistic comparison, however, such cases are listed as “external possessor constructions” here, but this should be taken as no more than a (perhaps misleading) label for constructions in which the dative phrase can be Â�interpreted as a possessor. (36′) i-bbi sbεa walsiwn n ul:wmad, … ˙ 3sm-cut.aor/pfv seven pm.nom.tongue gen sm.nom.snake ‘He cut out the seven tongues of the snake,…’ (#271)╇ (dat = n, do = body part)
In (36) and (37) the possessee is a body part of the possessor. Other types of possessed entities found include kin members and clothing items. Clitic doubling of external possessors External possessors can be doubled by a clitic (cf. 2.1): (38) bbi-n a-si ixf i um>ari n uzru. ˙ cut.aor/pfv-3pm dat-3s sm.head dat sm.nom.chief gen sm.nom.A. ‘They cut the head of the chief of Azrou.’ (#271)
The possessee can be encoded as a direct object (38), a PP (39), or a subject of an intransitive clause (40). At least in the first two cases, there can be an additional Â�possessive marker referring to the possessor on the possessee (41, 42). (39) ar a-s tt>wm-n lmlayka lhnna > ifassn. ˙ ipv dat-3s paint.ipv-3pm pm.angel henna in pm.hand ‘[When he was sleeping] angels used to paint his hands with henna.’ (#35) (40) (I warn you:)
bnadm nna [mi Ø zgzaw-n-t twallin] person demprox.addr dat be.blue.pfv-3pm-f pf.nom.eye ‘Somebody with blue eyes [-don’t go with him]!’ (lit.: …to whom the eyes are blue.)
(41)
t-ssudmj agayyu nn-si i twayyai nn-sj, … 3sf-kiss.aor/pfv sm.head gen-3s dat sf.nom.maidservant gen-3s ‘She kissed the head of her maidservant,…’ (#68) (lit.: Shej kissed heri head for herj maidservanti.)
(42) ffi-n a-s irdn f idar n nn-s ˙ ˙ pour.aor/pfv-3pm dat-3s pm.wheat on pm.foot gen-3s ‘they pour wheat on his feet.’ (lit.: …on the feet “for” him.) (#137)
Christian J. Rapold
Covert possessees The possessee of an external possessor need not be overtly expressed. Consider the following two examples, which feature the same verb. In (43) the possessee is overt, while in (44) it is understood from the context. (45) is a further case in point. (43) t-ssird a-s adar yat tm>art […] (ben/ep, do) ˙ 3sf-wash.aor/pfv dat-3s sm.foot f.one sf.nom.woman ‘One womani […] washes herj [right] foot’ (#282) (explaining wedding rituals) (44) (‘Let’s get back to the bride. She too gets up on the seventh day and’] ssird-nt a-s tm>arin. (ben/ep, no overt do) wash.aor/pfv-3pf dat-3s pf.nom.woman ‘the women wash her.’ (#139) (explaining wedding rituals) (lit.: …wash Ø for her)
The referent of the covert direct object in (44) is the body of the possessee, as can be inferred from cultural knowledge. The referent of the understood direct object in (45) is probably the courting manoeuvres of the possessee. (45) (One day, the muezzin fell in love with the girl.) ur a-s t-ri, (ben/ep, no overt do) neg dat-3s 3sf-want ‘(But the girl) did not want him.’ (original translation) (#16)
Covert possessees and syntactic reanalysis The fact that the possessee can be covert, as seen in the previous section, explains why some transitive verbs more often than not occur with a dative phrase as the sole undergoer.8 Verbs of this kind include the ones listed in (46). In the English or French translation the referent of the dative phrase is usually encoded as a direct object (as in€(44, 45)). (46)
hzzm ‘tie up’ ˙ qqs ‘sting, prick’ izwir ‘precede’
In some cases, the frequent use with a dative phrase but without a direct object may eventually be leading to polysemy. This seems to hold for the verbs in (47), where the rightmost column gives the typical meaning in the context indicated. Some verbs, finally, may represent the lexicalisation of the frequent use without an overt direct object (48). They can only be used intransitively; the only way to encode an undergoer is by means of a dative phrase.
8.â•… For the present purpose, “transitive” is understood as being able to take a direct object and being passivisable.
Semantic roles of the dative in Tashelhiyt
(47) tr. verb with overt do with dat, without overt do rzm ˙ >r
‘open’ ‘read’
‘release, set free’ ‘call’
(48) itr. verb >rs ‘slaughter’ sll ‘hear’ ggr ‘touch’
That the verb >rs ‘slaughter’ in (48) may formerly have been transitive, meaning something like ‘cut (the throat)’, is supported by the gesture typically accompanying it: a swift movement of the hand or index across one’s throat.9
3.15â•… Distributive A cross-linguistically less common function of the dative is that of marking a distributive phrase. This function has a temporal and a pluractional use.
3.15.1â•… Temporal
(49) lmawta, nttat ur akkw i-dhir mššk ˙ dead 3sm neg totally 3sm-appear.neg how.much/many a i-ttmmtat-n i wass. comp 3sm-die.ipv-sub.rc dat sm.nom.day (#362) ‘As for people killed, it is not completely clear how many died each day.’
The clearly temporal distributive use of the dative may have developed from cases as in (50), where the dative phrase can also be interpreted as an inanimate beneficiary: (50)
ar as i-ttsd ar tallwizt i wass. ˙˙ ipv dat-3s 3sm-lay.(egg).ipv sf.gold.coin dat sm.nom.day ‘[The bird] laid a gold coin for him every day.’ (original translation) (lit.: …for the day.) (#224)
In the corpus, this function occurs only with the noun ass ‘day’, but this is also by far the most frequent temporal adverbial that occurs with a distributive reading. Other means of marking distributive quantification in time adverbials include the modifier kraygat ~ kaygat (< kra i-ga tt ‘some 3sm-be.located.pfv 3sf:do’) and the periphrastic construction X ar X ‘X after X’.
3.15.2â•… Pluractional The dative phrase i yan/yat (dat m.one/f.one) expresses pluractional events. Depending on the context, this can involve several non-subject (51) or several subject referents (52).
9.â•… Thanks to Maarten Kossmann for this piece of information.
Christian J. Rapold
Note that the distributive subject reading is not confined to intransitive verbs, contrary to a widespread cross-linguistic tendency in pluractionals. (51) i-fka a-sn tummazt n lgrgaε i yan. ˙ 3sm-give.pfv dat-3pm sf.handful gen nuts dat m.one ‘He gave them a handful of nuts each.’ (#11) (distributive pp) (52) fki-n a-s da> amrriq i yan. ˙˙ give.pfv-3pm dat-3s again sm.slap.in.face dat m.one ‘They each slapped him in the face again.’ (#11) (distributive sub)
3.16â•… purpose and multiple datives By way of closing this section on the various functions of the dative in Tashelhiyt, it may be noted that while the functional range of the dative is considerably wider in this language than in the semantic map in Haspelmath (2003), there is one function in that map, purpose, which is not covered by the dative. If purpose typically pertains to events, then there are no clear cases in our corpus where a dative marks nominalised verbs or clauses that encode a purpose. This function is typically rather encoded in a subordinate clause introduced by the complementiser ad: (53) (‘He went off with them’) at tn i-jlu. comp 3pm 3sm-leave.behind ‘in order to get rid of them.’ (#11)
Note, finally, that several dative phrases with various functions may co-occur in a single clause. Example (54) features a substitutive beneficiary and a general affectee, (55) a recipient/addressee and a recipient-beneficiary, (56) two concrete beneficiaries, and (57) a beneficiary and a locative: (54) (context: The new wife of the king seems to be mute. He calls his advisor and says:) t-skr-t i (., .) >ira yy i only .1s 2-do/make./-2s xtad arki> [a t-sawal]. this.one until 3sf-speak. ‘Just make this one speak speak for me.’ (lit. ‘[I wish] only that you do to that one for me so that she speaks.)
(55) id_u_flan, dalb-n a-wn illit-un (rec/add, rec–ben) ˙ so.and.so ask.aor/pfv-3pm dat-2pm daughter-poss:2pm i yiwit-sn. dat sm.nom.son-poss:3pm (context: marriage arrangement) ‘The So-and-so ask your daughter from you for their son.’
Semantic roles of the dative in Tashelhiyt 
(56) anna t-ri-t ad a-m amz-> axddam (2x concrete ben) ˙ int 2-want.aor/pfv-2s comp dat-2sf take-1s sm.worker htta i >wad nn-m ma tt i-ttεdal-n? ˙ even dat this.one gen-2sf who/what do.3sf 3sm-fix.ipv-sub.rc ‘Do you want that I take on a worker for you even for that [i.e. pubic hair] of yours, one who takes care it?’ (#240) (57)
g-n a-s tt i uxsas nn-s zund tikrzit. (ben, loc) put-3pm dat-3s do.3sf dat sm.nom.head gen-3s like sf.turban ‘They put it on his head like a turban.’ (description of a ritual) (#138) (original translation: …qu’on lui pose sur la tête comme un turban.)
4.â•… Conclusion and questions for further research After an overview of the morphosyntax of the dative in Tashelhiyt, this paper has discussed the semantic roles of this category as attested in a text corpus containing over 4,000 datives. These roles have been presented in a semantic map and semantic links leading from the prototypical dative function of recipient to the other functions have been proposed. The main findings are summarised in Figure 3 below, which incorporates the roles treated and the specific semantic links between them. Roles that are preceded or underlined by dotted lines are not found in Haspelmath (2003); their position in the map is tentative or remains to be established on the basis of cross-linguistic evidence. This study has further identified Tashelhiyt as a “neutral language” in Kittilä’s (2005) typology of beneficiary, recipient and recipientbeneficiary encoding, given that all three functions are encoded in the same way as far as their main encoding strategy— dative marking—is concerned. Moreover, the Tashelhiyt data show that, contrary to Kittilä’s (2005:275) claim, the benefaction involved in recipient-beneficiaries need not be substitutive, but can also be concrete benefaction. The findings and hypotheses of this article only scratch the surface of the semantics of the dative in Tashelhiyt. The following points in particular await further research: First, it is likely that the range of semantic roles identified here is not complete. The research should be extended to other text genres and a larger database. Second, the precise factors determining or favouring a specific semantic interpretation of the dative remain to be investigated. This is especially important for those cases where more than one semantic role would yield a possible interpretation of a given utterance. Third, the relation between the dative and competing encoding strategies waits to be described. In the case of the general affectee, beneficiary and maleficiary, this mainly concerns the preposition f ‘on’.
Christian J. Rapold
possessor
G2 direction
G1
recipient M2 G3
M1
beneficiary/ maleficiary/ general affectee
M3
judicantis
purpose
experiencer G3 locative
––– ––– ––– ––– ––– –––
– source-maleficiary substituted-assisted concretely benefited unintending causer distributive [2 subroles]
Semantic links between roles generalisations:
metaphorisations:
G1=transfer, but not its orientation, bleached
M1=shift from material to abstract transferred entity
G2=transfer including its orientation bleached
M2=shift from material to communicative transferred entity
G3=decreasing control over transferred entity
M3=shift from affected to evaluating entity
Subroles of : -temporal -pluractional
Figure 3.╇ A semantic map of the dative in Tashelhiyt (II) (built on Haspelmath 1999, 2003)
Appendix: Statistics This appendix presents some statistical data about the corpus of datives on which this article is based. More specifically, it treats the frequency of datives by type of complement (Section 1), animacy in dative phrases (Section 2) and the co-occurrence of the dative with selected verbs (Section 3). Datives occurring with first person pronouns are disregarded in the following, due to the syncretism of these structures with the direct object pronouns (yyi ‘dat/do.1s’, a> ‘dat/do.1’). 1. Frequency of datives and types of dative complements Roughly every twentieth word is or contains a dative marker in the corpus. The datives subcategorise according to morphosyntactic contexts as shown in Table 1:
Semantic roles of the dative in Tashelhiyt
Table 1.╇ Frequency of datives by type of complement pronouns
nouns
gaps
total
3,307 (80%)
788 (19%)
38 (1%)
4,133 (100%)
tokens
tokens
2sm 196
2sf 41
3s 2,428
2pm 43
2pf 1
3pm 533
3pf 65
2. Animacy in dative phrases About 75% of all lexical dative PPs are animate. In pronominal and gapped dative complements the amount of animates is even higher. This also holds for third person pronouns. Table 2.╇ Animacy in lexical dative complements
no. of occurrences
+anim
–anim
total
608
180
788
3. Co-occurrence of the dative with selected verbs While the semantic role of beneficiary is most frequently encoded with a dative, this is not the most frequent interpretation of that category in running texts. Table 3 presents the three verbs with the highest co-occurrence rate with the dative: Table 3.╇ Verbs with the highest co-occurrence rate with the dative
ini ‘say’ fk ‘give’ awi ‘bring’ Grand total
pronouns
nouns
gaps
total
2,100 242 72
196 69 29
4 4 1
2,300 315 102 2,717
The three verbs in Table 3 co-occur with a dative in (virtually) all instances. Almost all tokens of dative PPs listed in Table 3 are recipients/addressees (or recipient-beneficiaries). In addition to the three verbs in Table 3, there are also other verbs in the corpus that co-occur with dative that is interpreted as a recipient(-beneficiary), e.g. ml ‘show’, azn ‘send’, εawd ‘tell’. This means that at least around 66% of all dative PPs are recipients/addressees (or recipient-beneficiaries). In other words, all the other semantic functions in Figure 3 share only around 34% of all dative occurrences (1,416 out of 4,133 tokens).
Christian J. Rapold
The 34% of all dative phrases that are not represented in Table 3 co-occur with a wide range of verbs. For lexical dative phrases alone (788 tokens), 162 different verbs are found. Many verbs occur just once; some of the more frequent ones include s> ‘buy’ (6€times) ssnu ‘cook’ (6), šš ‘eat’ (5).
Abbreviations addr: aff: all: aor: aor/pfv: cl.: dir: direct: ep: int: ipv:
addressee obl: general affectee pfv: (neither beneficiary nor maleficiary) pf: allative pm: aorist sf: Aorist stem (unspecified for aspect) sm: = pfv stem res.ref: clitic directional preposition (s) sfx.: directional clitic (allative/ventive) sub_rc: external posssessor interrogative vent: Imperfective
oblique Perfective plural feminine plural masculine singular feminine singular masculine resumed referent (“aforementioned”) suffix subjective relative (i.e. in a relative clause) ventive
References Anderson, Lloyd B. 1982. The ‘perfect’ as a universal and as a language-particular category. In Tense–Aspect. Between Semantics and Pragmatics [Typological Studies in Language 1], Paul Hopper (ed.), 227–264. Amsterdam: John Benjamins. Aspinion, Robert. 1953. Apprenons le berbère. Initiation aux dialects chleuhs. Rabat: Félix Moncho. Bentolila, Fernand. 1981. Grammaire fonctionelle d’un parler berbère. Aït Seghrouchen d’Oum Jeniba (Maroc) [Langues et civilisations à tradition orale 46]. Paris: SELAF. Borg, Albert J. & Azzopardi-Alexander, Marie. 1997. Maltese [Descriptive Grammars]. London: Routledge. Boumalk, Abdallah. 2003. Manuel de conjugaison du tachelhit (langue berbère du Maroc) [Tira – Langues, littératures et civilisations berbères]. Paris: L’Harmattan. Brustad, Kristen. 2000. The Syntax of Spoken Arabic: A Comparative Study of Moroccan, Egyptian, Syrian, and Kuwaiti Dialects. Washington DC: Georgetown University Press. Caubet, Dominique. 1993. L’arabe marocain [Études chamito-sémitiques]. Paris: Peeters. Chaker, Salem. 1983. Un parler berbère d’Algérie (Kabylle): Syntaxe. Aix en Provence: Université de Provence. Croft, William. 22003. Typology and Universals [Cambridge Textbooks in Linguistics]. Cambridge: CUP.
Semantic roles of the dative in Tashelhiyt
Cysouw, Michael. Forthcoming. Building semantic maps: The case of person marking. In New Challenges in Typology, Matti Miestamo & Bernhard Wälchli (eds.). Berlin: Mouton. Dell, François & Elmedlaoui, Mohammed. 1989. Clitic ordering, morphology and phonology in the verbal complex of Imdlawn Tashlhiyt Berber, Part I. Langues Orientales Anciennes, Philologie et Linguistique 2: 165–194. Dell, François & Mohammed Elmedlaoui. 2002. Syllables in Tashlhiyt Berber and in Moroccan Arabic [Kluwer International Handbooks of Linguistics 2]. Dordrecht: Kluwer. Galand, Lionel. 1969. Types d’expansion nominale en berbère. Cahiers Ferdinand de Saussure 25:€83–100. (Reprinted in Galand 2002:193–210). Galand, Lionel. 1988. Le berbère. In Les langues dans le monde ancien et moderne, Vol. 3: Les€langues chamito-sémitiques, Jean Perrot (ed.), 207–242. Paris: Editions du CNRS. Galand, Lionel. 1990. Du nom au verb d’état. Le témoinage du berbère. In Proceedings of the Fifth International Hamito-Semitic Congress, 1987, Vol. I: Hamito-Semitic, Berber, Chadic [Veröffentlichungen der Institute für Afrikanistik und Ägyptologie der Universität Wien 40], Hans G. Mukarovsky (ed.), Vienna: Universität Wien, Institute für Afrikanistik und Ägyptologie. (Reprinted in Galand 2002: 273–283). Galand, Lionel. 2002. Etudes de linguistique berbère [Collection linguistique 83]. Louvain: Peeters. Geeraerts, Dirk. 1998. The semantic structure of the indirect object in Dutch. In The Dative, Vol. 2: Theoretical and Contrastive Studies [Case and Grammatical Relations across Languages 3], Willy van Langendonck & William van Belle (eds.), 185–210. Amsterdam: John Benjamins. Haan, Ferdinand De. 2004. On representing semantic maps. , April 4, 2008. Haspelmath, Martin. 1998. How young is standard average European? Language Sciences 20(3):€271–287. Haspelmath, Martin. 1999. External possession in a European areal perspective. In External Possession [Typological Studies in Language 39], Doris L. Payne & Immanuel Barshi (eds.), 109–135. Amsterdam: John Benjamins. Haspelmath, Martin. 2003. The geometry of grammatical meaning: Semantic maps and crosslinguistic comparison. In The New Psychology of Language. Cognitive and Functional Approaches to Language Structure, Vol. 2, Michael Tomasello (ed.), 211–242. Mahwah NJ: Lawrence Erlbaum Associates. Kittilä, Seppo. 2005. Recipient-prominence vs. beneficiary-prominence. Linguistic Typology 9:€269–297. König, Ekkehard & Hasplemath, Martin. 1997. Les constructions à possesseur externe dand les langues de l’Europe. In Actance et valence dans les langues d’Europe [Empirical Approaches to Language Typology/EUROTYP 20–2], Jack Feuillet (ed.), 525–606. Berlin: Mouton de Gruyter. Langendonck, Willy Van. 1998. The dative in Latin and the indirect object Dutch. In The Dative, Vol. 2: Theoretical and Contrastive Studies [Case and Grammatical Relations across Languages 2], Willy Van Langendonck & William Van Belle (eds.), 211–259. Amsterdam: John Benjamins. Mountassir, Abdallah El. 2003. Dictionnaire des verbes Tachelhit-Français. (parler berbère du sud du Maroc) [Tira – Langues, littératures et civilisations berbères]. Paris: L’Harmattan. Rapold, Christian. 2005. Review of: Harry Stroomer, An Anthology of Tashelhiyt Berber Folktales (South Morocco) [Berber Studies 2]. Cologne: Rüdiger Köppe, 2001. Journal of African Languages and Linguistics 26(1): 108–111.
Christian J. Rapold Sasse, Hans-Jürgen. 1984. Case in Cushitic, Semitic and Berber. In Current progress in AfroAsiatic linguistics: Papers of the Third International Congress [Current Issues in Linguistic Theory 28], James Bynon (ed.), 111–125. Amsterdam: John Benjamins. Stroomer, Harry. 1994. Morphological segmentation in Tachelhiyt Berber (Morocco). Études et documents berbères 11: 91–96. Šarić, Ljiljana. 2002. On the semantics of the ‘dative of possession’ in the Slavic languages: An analysis on the basis of Russian, Polish, Croatian/Serbian and Slovenian examples. Glossos 3: 1–22. Stumme, Hans. 1899. Handbuch des Schilhischen von Tazerwalt. Grammatik – Lesestücke€ – ˙ Gespräche – Glossar. Leipzig: J.C. Hinrichs. Talmy, Leonard. 1991. Path to realization: A typology of event conflation. Proceedings of the Berkeley Linguistics Society 17: 480–519. Vanhove, Martine. 1993. La langue maltaise: Études syntaxiques d’un dialecte arabe “périphérique” [Semitica viva 11]. Wiesbaden: Harrassowitz. Wierzbicka, Anna. 1986. The meaning of a case: A study of the Polish dative. In Case in Slavic, Richard D. Brecht & James S. Levine (eds.). Columbus OH: Slavica.
Corpus Amard, Pierre. 1997. Textes berbères des des Aït Ouaouzguite (Ouarzazate, Maroc). Edited and annotated by Harry Stroomer. ([Collection] “Bilingues”). Aix-en-Provence: Edisud. Stroomer, Harry. 2001a. Textes berbères des Guedmioua et Goundafa (Haut Atlas, Maroc), basés sur les documents de F. Corjon, J.M. Franchi et J. Eugène. ([Collection] “Bilingues”). Aixen-Provence: Edisud. Stroomer, Harry. 2001b. An Anthology of Tashelhiyt Berber Folktales (South Morocco) [Berber Studies 2]. Cologne: Rüdiger Köppe. Stroomer, Harry. 2002. Tashelhiyt Berber Folktales from Tazerwalt (South Morocco). A Linguistic Reanalysis of Hans Stumme’s Tazerwalt Texts with an English Translation [Berber Studies 4]. Cologne: Rüdiger Köppe.
Benefactive strategies in Thai Mathias Jenny
Universität Zürich In Thai grammatical and semantic relations are expressed by syntactic or lexical means, such as constituent order and grammaticalized nouns and verbs. The topic of the present study are benefactive strategies in Thai, i.e. linguistic means to describe situations which involve an entity peripheral to, but in some way affected by, the event described by the predicate. Different benefactive strategies are compared in terms of form and function with special emphasis on the semantic differences. The Thai data show an important distinction between ‘direct benefactive’ (the beneficiary is directly involved in the event as recipient or experiencer), and ‘indirect benefactive’ (the activity is performed for his sake without him necessarily being aware of it) expressions.
0.â•… Introduction Thai is a highly isolating language with only minimal morphology besides compounding. Grammatical and semantic relations are expressed by syntactic or lexical means, including constituent order. One important strategy to express grammatical and semantic relations employed in Thai is the use of grammaticalised nouns and verbs to indicate relations, including different types of benefactive. These grammaticalised lexemes are syntactically, but not always functionally, distinguishable from real prepositions (see Bisang 1992 on co-verbs). The topic of the present study is benefactive strategies in Thai, i.e. linguistic means to describe situations which involve an entity peripheral to, but in some way affected by, the event described by the predicate. The approach to be followed is function to form, using general linguistic terminology rather than theory-specific notions. The questions to be addressed are the following: (1) What strategies are used in Thai to express beneÂ� factive situations, (2) which of these strategies can be seen as grammaticalised, and (3) what semantic differences are there between the different grammaticalised benefactive expressions. The results especially of (3) will be compared with the typologies of benefactives suggested by Van Valin & LaPolla (1997), Kittilä (2005) and Song (2007). The sources for the Thai data include own fieldwork as well as published primary sources (printed texts, online corpora).
Mathias Jenny
1.â•… Types of benefactive Different authors have suggested different typologies of benefactive expressions, based on semantic and morphosyntactic considerations. Van Valin & LaPolla (1997) classify benefactives as belonging to one of the three following types: 1. Recipient Benefactive 2. Deputative Benefactive 3. Plain Benefactive According to Van Valin & LaPolla (1997: 383f), recipient benefactives describe actions which are carried out by the actor with the intention that the beneficiary then has an object. Deputative benefactive actions are carried out “with the intention that the beneficiary not do the action”, while plain beneficiaries are explained as “something like ‘the action of the actor provides them with amusement, enjoyment or other kind of benefit’â•›”. These subcategories of benefactive events are supposed to have cross-linguistic validity, but other authors have suggested (slightly) different subtypes. Song’s (2007) classification of Korean benefactive situations is similar to Van Valin & LaPolla’s, the main difference being that the “Recipient-Benefactive” is labelled “Engager Benefactive”, based on the fact that in Korean the recipient does not necessarily have to receive some object, but has to engage with it in some way. Kittilä (2005) proposes a feature-based model of benefactive types, with the Parameters [±reception] and [±substitutive benefaction]. Again, the types of benefactives arrived at by Kittilä’s typology are similar to the ones postulated by Van Valin & LaPolla and Song, with minor differences in morphosyntactic behaviour and semantic fields covered. Kittilä uses his typology to classify languages as recipient-prominent, beneficiary-prominent and neutral. While these three typologies provide good tools for the general classification of benefactives, it will be seen that they are not sufficient for Thai and, presumably, similar languages. In the following chapters I will investigate the different types of benefactive constructions in Thai in terms of semantics and syntactic behaviour. I take the notion of benefactive to cover an entity, usually human or highly animate, which is not a direct participant of the event expressed by the predicate, i.e. is not part of the semantic structure of the predicate, but is (positively or negatively) affected by the event. While recipient or pure beneficiary may be taken as prototypical or basic (and in many cases historically earlier) types of this category, the notion is often extended to cover affectees of other kinds. It will be seen that one of the relevant markers in Thai, viz. hây ‘give’, has gone the whole way of grammaticalisation from recipient to include the ‘goal of a directed action’ which does not need to be animate. The other main marker of benefactive
Benefactive strategies in Thai
events, the originally causal preposition phш ˆ ә, has undergone a shift in meaning from cause to purpose. Both markers cover different but sometimes overlapping aspects of the benefactive category.
2.â•… Benefactives in Thai 2.1â•… Prepositions Thai has a few function words that can be labelled real prepositions, i.e. grammatical morphemes that do not have lexical semantics, at least in synchronic perspective (some may be shown to be derived from older relator nouns or verbs). The following three are used to mark benefactive and related notions: 1. Kὲε/kàp ‘to’. The former is used in formal and literary styles, while the latter, lit. ‘with’, replaces it in more colloquial language use. Both forms merge in the informal pronunciation kàô/kә in spoken Thai. This form can be taken as a general marker of indirect (or dative) objects, often used redundantly after benefactive expressions involving hây ‘give’ as V2 (see 2.3). Example (1) illustrates the use of kә indirect object marker. (1) Khaˇw '¡ffk kә phfˆf wâa yàak pay riәn nfˆfk. 3 tell dat father say want go learn outside ‘He told his father that he wants to study abroad.’
2. Saˇmràp ‘for’, from Khmer sfmrap ‘to use for, apply to’. This preposition is frequently used in formal contexts and can be seen as translation equivalent of English for in most uses, including highly abstract notions like ‘for today’, ‘as for me’, etc., as seen in Example (2). As a subordinator, sˇamràp is used to introduce purposive clauses without overt subject. These purposive clauses often function as attributes to noun phrases. (2) Mây mii khfˇfŋ.khwaˇn saˇmràp câw.phâap l??y. neg exist present for host emph ‘[We] haven’t got a present for the host.’
3. Phш ˆ ә ‘for, for the benefit of, for the sake of ’. In older texts, as well as in closely related languages such as Lao and Shan, this preposition means ‘because of ’, a use that is still listed in standard Thai dictionaries but hardly found in the contemporary language. Its use in modern Thai, both spoken and written, is as a marker of a beneficiary which does not usually coincide with recipient and which is not directly affected by the act expressed by the main verb, but benefits in an indirect
 Mathias Jenny
way from its result. If the object of phш ˆ ә is inanimate, the normal interpretation is ‘for the sake of ’, which may be the underlying common function of this preposition in modern Thai. I introduce the term indirect or remote beneficiary for this participant and will explore the uses of this marker in more detail in Chapter€3. As a subordinator, phш ˆ ә introduces subjectless purposive clauses (similar to saˇmràp), but clauses introduced by phшˆ ә are not normally used in attributive function. Especially in formal texts and in older documents, such clauses almost always contain the prospective or irrealis marker càô, often together with the clause nominaliser thîi. The vast majority of instances in the category ‘old language’ of the Chulalongkorn online Thai corpus (which covers documents from the 13th to the early 20th century) has phш ˆ ә either as causal preposition or as causal/purposive conjunction in combination with càk/càô. Of the three prepositions mentioned above, saˇmràp and phш ˆ ә can by themselves express benefactive (or related) meanings, while kὲε/kàp does not have this force. The latter also lacks the capability of introducing subordinate clauses. Only phш ˆ ә appears to have benefactive as its core meaning in the present-day language.
2.2â•… Serial verb constructions Another strategy to express benefactive relations in Thai is by means of (complex) verbal semantics, usually composed of two simple verbs in a serial verb construction (SVC). Typically, the first verb describes an activity while the second introduces a beneficiary, usually in a rather concrete and semantically explicit way. Benefactive V2s belong to the general class of V2s introducing a secondary participant which is not licensed by the primary verb, such as goal, instrument, path, etc. SVCs with a V2 introducing a beneficiary or other peripheral participant contain two cores with shared subject but independent objects. Example (3) shows deputative benefaction, while (4) and (5) have a recipient beneficiary. (3) Khaˇw pay tәlàat thεεn phш ˆ әn. 3hum go market replace friend ‘He goes to the market in his friend’s stead.’ (4) Mεˆε sш ˆ ш khәnoˇm liәŋ lûuk. mother buy sweets feed child ‘The mother buys sweets for her children (to eat).’ (5) Phfˆf sш ˆ ш naˇŋsш ˆ ш fàak lûuk. father buy book entrust child ‘The father bought a book for his son (as a present).’
Benefactive strategies in Thai 
The expression in (6) is comparable to what Peterson (2007:18) describes as “additional benefactive” in Hakha Lai. (6) 1εεŋ sш ` ә 1am. ˆ ш kεεŋ phш Daeng buy curry set.aside Dam ‘Daeng bought some curry for Dam (besides buying some for herself).’
This construction is very frequent in colloquial Thai and can be seen as partly grammaticalised, covering also events that do not involve the act of actually setting aside something and later giving it to someone. It can also be used in (additional) deputative benefactive expressions, indicating that the subject is doing some activity for himself and does it on behalf of the beneficiary as well. Although this use is attested in colloquial Thai, not all speakers accept it as idiomatic or grammatical, as indicated by the question mark on (3a). (3) a. ?Khaˇw pay tәlàat phY ¡ ә phшˆәn. 3hum go market set.aside friend ‘He goes to the market for his friend, too.’
Depending of the semantics of the predicate and the context, the use of phш ˆ ә in deputative expressions is understood as real additional benefactive (‘do something for oneself and do some more for someone else’) or it leads to a metaphorical reading along the lines ‘do something and think of someone else who would like to do the same thing’. The verb used to introduce the beneficiary can be either transitive, as in the case of thεεn ‘replace’, or, less frequently, ditransitive as líәŋ ‘feed so. with sth.’, as shown in (7) and (8): (7) Khaˇw thεεn phш ˆ әn. 3hum replace friend ‘He replaces his friend.’ (8) Mεˆε líәŋ khәnoˇm lûuk. mother feed sweets child ‘Mother gives sweets to the children.’
The verb líәŋ has inherent benefactive semantics, i.e. the beneficiary role is part of the semantic structure of the verb. The word order V DO IO is fixed, but ditransitive verbs rarely occur with both objects expressed overtly. An SVC as in (4) is considered more elegant and is preferred over the double object construction. The most common verb used in a wide range of benefactive constructions is the ditransitive hây ‘give’. Especially if the beneficiary is left unexpressed, hây can replace the more concrete V2s, as seen in Examples (3′)–(6′).
 Mathias Jenny
(3´) Khaˇw pay tәlàat hây. 3hum go market give ‘He goes to the market in someone’s stead.’ (4´) Mεˆε sY ´ Y khәnoˇm hây. mother buy sweets give ‘The mother buys sweets for someone (to eat).’ (5´) Phfˆf sY ´ Y naˇŋsY ¦ Y hây. father buy book give ‘The father bought a book for someone (as a present).’ (6´) 1εεŋ sY ´ Y kεεŋ hây. Daeng buy curry give ‘Daeng bought some curry for someone.’
Example (9) shows the full verb use of hây with both theme and recipient overtly expressed, the latter optionally marked as indirect object. This construction sounds stilted to many native speakers. Again, the use of an SVC is the most common means to make the expression sound more natural. The verb ôaw ‘take’ in (10) does not add any semantic material but rather helps to avoid the double object construction.1 The verb ‘give’ retains its semantics and each verb in (10) has only one overt object while sharing the subject. (9) Nák.riәn hây 1ffk.máay (kә) khruu. student give flower (dat) teacher ‘The student gives a flower to her teacher.’ (10) Nák.riәn ôaw 1ffk.máay hây khruu. student take flower give teacher ‘The student gives a flower to her teacher.’
2.3â•… Hây ‘give’ As mentioned in 2.2, hây started out as a ditransitive verb, a function which is still present in the modern language. Sentences like (10) may be seen as the starting point of more grammaticalised (i.e. less semantically concrete) uses of hây as V2, expressing first a recipient and consequently also a beneficiary and general experiencer. Other grammatical functions of hây include purposive subordination, usually including a change of subject as a kind of causative (jussive, permissive) with an optional overt subject. As constituents (including verbs) can be dropped in Thai, it is not always obvious which function is intended, and often more than one function can be present
1.â•… A further effect of the introduction of ôaw to the construction is the fronting and therefore focusing of the object NP.
Benefactive strategies in Thai
at the same time. The most important distinguishing feature is the optional oblique marker kàp/kә, which cannot be used with hây as a subordinator. Sentence (11a) can be interpreted as either expressing benefactive or (causative/purposive) subordination. The addition of the optional preposition (11b) or the subordinate verb (11c) disambiguates the construction. (11) a.
Mεˆε rf¢fŋ phleeŋ hây lûuk (faŋ). mother sing song give child listen ‘The mother is singing a song for her child (to listen).’
b. Mεˆε rf¢fŋ phleeŋ hây kә lûuk (*faŋ). mother sing song give dat child listen ‘The mother sings a song for her child.’ c.
Mεˆε rf¢fŋ phleeŋ hây (*kә) lûuk faŋ. mother sing song give dat child listen ‘The mother sings a song for her child to listen.’
The common semantic element of the constructions involving hây can be labelled purposive. The semantic structure of [VP hây] can be formalised as do΄[ (x, y) purp LS]
where LS is a logical structure of any kind (see Van Valin & LaPolla 1997:passim). Spelled out in this formalism, (11c) would be (11) c´. [do΄ (mother, [sing΄[ (mother, song)])] purp [do΄[ (child, [listen΄ (child, song)])]
In (11c) the subjects of the matrix and the dependent clause are overtly expressed. As known or retrievable participants may be dropped in Thai, the presence of the puropsive marker hây itself can be enough to imply a change of subject, as seen in (12) and (13). (12) Khaˇw yàak pay mYәŋ thay. 3hum des go land Thai ‘He wants to go to Thailand.’ (13) Khaˇw yàak hây pay mYәŋ thay. 3hum des give go land Thai ‘He wants (me, you, him, her, us, them) to go to Thailand.’
The allocentric purposive reading (‘action directed toward another participant’) of hây has been generalised in Thai to the extent that [VP hây] is normally interpreted as ‘action performed for someone else’. The V2 hây has developed into a benefactive verb marker, rather than a benefactive preposition (or co-verb, cf. Bisang 1992:366). As seen in Section 2.2, this interpretation includes the different subtypes of benefactive (recipient, deputative, experiencer benefactive). More often than not, the
Mathias Jenny
beneficiary is not overtly expressed. When it is expressed, it may optionally be marked as indirect object only if the beneficiary is recipient or experiencer, not in deputative contexts. The actual semantics of the V2 hây is known only from the context, both linguistic and extra-linguistic, as seen in (14) with recipient benefactive and (15) with deputative benefactive. (14) Thâa wâaŋ phoˇm càô sfˇfn phaasaˇa thay hây (kә) khun. if free 1m fut teach language Thai give (dat) 2 ’If you have time, I will teach you Thai.’ (15)
Thâa mây wâaŋ phoˇ m càô sfˇfn phaasaˇa thay hây (*kә) khun. if neg free 1m fut teach language Thai give (dat) 2 ‘If you don’t have time, I will teach your Thai lesson (for you).’
While the determining factor in (14) and (15) is extra-linguistic knowledge, in other cases linguistic factors alone, such as construction type, determine the role of the participant introduced by hây. This is illustrated in (16) and (17). The default reading of the former is deputative while the latter is most naturally interpreted as recipient. There is nothing in the semantics of the verb or the theme that would favour one or the other reading. (16) Sòŋ còt.maˇay hây phfˆf. send letter give father ‘Send a letter for father.’ (rarely: ‘Send a letter to father’) (17) Sòŋ ŋ?n hây phfˆf. send money give father ‘Send money to father.’ or ‘Send money for father.’
The reason for the asymmetry in interpretation lies in the fact that there are special constructions in Thai for the expression of communication (16ʹ). If a speaker chooses not to use this readily available construction, the hearer reinterprets the expression as the nearest plausible reading, which in this case is the deputative benefactive. As seen from the ungrammaticality of (17ʹ), this construction is not available with themes other than means of communication, so (17) retains its default interpretation with hây marking a recipient. (16´) Sòŋ còt.maˇay pay haˇa phfˆf send letter go seek father ‘Send a letter to father.’ (17´) *Sòŋ ŋ?n pay haˇa phfˆf send money go seek father
Not being restricted to recipient benefactive contexts, hây can be used with intransitive verbs. The normal interpretation in these expressions is as deputative or plain
Benefactive strategies in Thai
benefactive, again depending on the context and the semantics of the primary verb. The sentence in (18) is an (adapted) example from a modern Thai novel. (18)
Hàak cam.pen khâa càô taay hây ôeŋ 1ây. if necessary 1fam fut die give 2fam get ‘If necessary I can/am ready to die for you.’ (‘in your place’: deputative; ‘for your sake’: pure ben)
Iwasaki & Ingkaphirom (2005:340) list a few instances of benefactive intransitive expressions in their reference grammar of Thai, e.g. ‘that woman smiled at Nikom’ and ‘my elder brother will be ordained for our mother’, which are acceptable because “a smile can be taken as directed to others” or because “in the Thai cultural context becoming a monk is one of the most venerable acts directed towards a mother”. Sentences like ‘this dog died for his master’ or ‘my friend cried for me’ are labelled unacceptable (ibid.). The unacceptability of these expressions without a context lies rather in the agentivity of the subject than in the cultural context, as Example (18) above shows, where the normally non-agentive predicate taay ‘die’ is reinterpreted as ‘be ready to die’ to fit the construction.2 A further generalisation involves the use of hây in non-benefactive contexts. These can have a malefactive interpretation in given contexts, as in (19), again taken from a modern novel. Other constructions allow either malefactive or neutral reading, depending on semantic and pragmatic factors, as seen in (20) and (21). The general tendency seems to be toward hây as a neutral marker of a human or human-like affectee of an action executed by a human or human-like agent. (19) Khâa kuu thammay? Kuu tham ôәray hây mYŋ rY ¦ Y? kill 1fam why 1fam do what give 2fam q ‘Why do you (want to) kill me? What have I done to you?’ (20) Khaˇw haˇn laˇŋ hây fεεn. 3hum turn.around back give lover ‘He turned his back on his girlfriend.’ (21) Khaˇw haˇn laˇŋ hây mfˇf. 3hum turn.around back give doctor. ‘He turned his back to the doctor.’
Sentences (22) and (23), both taken from the SEAlang online corpus show that hây (kàp) has gone a long way away from benefactive toward a marker of affectee and then
2.â•… A counter example to the “unacceptable” sentence with ‘cry for someone’ is the following, from the lyrics of a popular Thai song: ‘These teardrops I shed for the sky and the (falling) leaves, but I will never cry for you’ (…r¢f fŋ.hây hay kàp th??).
Mathias Jenny
pure directional indicating the goal of a directed action. The restriction to only animate participants is lost. Unlike other functions of hây, this construction requires an overt object, i.e. hây here has become more preposition-like, though it still co-occurs with the dative marker kàp/kә. (22) Khâa khәyàw kampân hây kàp sênkhf¡fpfáa 1fam shake fist give dat horizon ‘I shook my fist at the horizon.’ (23) Thúk khon tfˆŋ thûmthee chiiwít cítcay hây kàp kaanthffphâa. every man must devote life heart give dat weaving ‘Everyone must devote their lives and souls to weaving.’
The semantic extensions of hây can be summarised as follows. It should be remebered that the different stages described here coexist in modern Thai and, with the exception of stage iva, have co-existed throughout the recorded history of the language. Table 1.╇ Development of hây constructions Stage
Description
Formula
Example
Stage I
Plain transfer of object (theme)
hây θ (kә) REC
(9)
Stage II
Transfer of object (theme)
V θ hây (kә) REC
(10)
Stage III
a. Transfer of result of action b. Transfer of action
V hây (kә) EXP V hây (*kә) DEP
(14) (15)
Stage IV
Action directed toward affectee
V hây (kә) AFF
(19–21)
Stage IVa
Action directed toward goal
V hây (kә) GOAL
(22, 23)
Stage V
Action directed toward situation (purposive)
V hây S
(11c)
Unlike SVCs with semantically full V2s, [VP hây] describes a single allocentric action. It involves a single core, with a peripheral participant which is an object of the whole verb complex rather than of V2 alone. This can be seen by the possibility to extract the beneficiary from [VP hây] expressions, but not from other SVCs. Extraction of an argument is mostly done for pragmatic reasons such as topicalisation or focusing. The difference in the extractability of the beneficiary in general SVCs and [VP hây] is illustrated in (24) and (25). (24)
*Lûuk khon lék lὲô, mεˆε sY ´ Y nom maa líәŋ. child cl small foc mother buy milk come feed Intended: ‘It’s her youngest child that the mother bought milk for.’ Also: *Líәŋ lûuk khon lék lὲô, mεˆε sY ´ Y nom maa.
(25)
Benefactive strategies in Thai
¢ffŋ phleeŋ hây. Lûuk khon too lὲô, mεˆε rf child cl big foc mother sing song give ‘It’s for her oldest child that the mother is singing a song.’ But: *Hây lûuk khon too lὲô, mεˆε rf¢fŋ phleeŋ.
The syntactic behaviour and the wide applicability of the [VP hây] constructions show the high degree of grammaticalisation of hây, while at the same time it is still used as a main verb with full semantics.3
3.â•… Features of hây and phY ¥ ә compared This section takes a closer look at the two constructions that can be seen as grammaticalised means to mark benefactive events, viz. the [VP hây] construction and the preposition phш ˆ ә. Although both can be labelled benefactive, they exhibit important differences, both syntactically and semantically.
3.1â•… Syntactic features Unlike arguments of verbs, which can be moved or omitted for pragmatic reasons, NPs functioning as objects of prepositions must always be overtly expressed and occur in situ. Thus, there are no such things as floating or stranded prepositions in Thai. This restriction is true only for prepositions which are not of verbal origin or have lost their verbal characteristics. As seen in (23) above, the beneficiary can be moved to the sentence initial contrastive topic/focus position, leaving the V2 benefactive marker hây in sentence final position. This is not possible with the preposition phш ˆ ә, as the ungrammaticality of (26) shows. (26) *Lûuk khon too lὲô, mεˆε rf¢ffŋ phleeŋ phY ¥ ә. child cl big foc mother sing song for
Syntactically, hây is part of the verb complex, while phш ˆ ә forms a constituent with the following noun phrase. That hây does not have the function of a preposition can be seen from the fact that V+hây expressions often take an oblique argument introduced by the real preposition kàp/kὲε, as mentioned above. The beneficiary of V+hây expressions is never introduced by phш ˆ ә. The occurrence of both hây and phш ˆ ә in the same clause is only allowed with two beneficiaries, one direct (recipient, deputative, experiencer) and one indirect. The order in this case is always V hây (kә) (NP) phш ˆ ә NP, as seen in (27).
3.â•… The closely related Shan language has gone one step further, replacing the cognate verb h!Y in its full verb function by the semantically more specific p!¦n, originally ‘share’, retaining h!Y only as V2 with the same functions as Thai hây.
Mathias Jenny
(27) Lûuk rf¢fŋ phleeŋ hây (kә) mεˆε phYˆ ә phfˆf. child sing song give (dat) mother for father ‘The child sings a song for his mother for his father’s sake.’
Here, the mother is experiencer-beneficiary and the father is indirect beneficiary. If the first beneficiary argument is omitted, the interpretation is still that the child sings the song for someone else for his father’s sake, not for his father to listen. In negated contexts, both V+hây and phш ˆ ә+NP constructions lead to ambiguity concerning the scope of the negation. Most commonly, the scope of the negation is seen as covering the whole verbal complex in V+hây constructions. A reading with narrow scope negation of only the main verb is more difficult to obtain, obviously due to the fact that the non-performance of an act cannot lead to its result being transferred to someone. The ambiguity that remains in negated V+hây constructions is between performing an act but not for the beneficiary indicated and not performing the benefactive act at all. Relevant examples are given in (28) and (29). (28) Phoˇm mây sY ´ Y naˇŋsY ¦ Y hây phш ˆ әn. 1m neg buy book give friend ‘I am not buying the book for my friend.’ ??‘I am not buying the book, (and I do it) for my friend.’
The most natural interpretation of (28) would be that I am buying the book for myself (or someone else), but not for my friend. Another common reading is that I am not buying the book at all. (29) 1èk khon níi mây kh??y yím hây khray sák khon. child cl this neg exper smile give who just cl ‘This child never smiles at anyone.’
Here again the natural interpretation is not that the child never smiles, but rather that she does not smile at other people. With phш ˆ ә the situation is different. Here the scope of the negation can be over the verb alone (narrow scope), over the benefactive phrase alone or over the whole verbal complex, including the beneficiary. Depending on the context and the semantics of the predicate, one of the possible readings is usually favoured. (30) Lûuk mây rf¢ffŋ phleeŋ phш ˆ ә mεˆε. child neg sing song for mother ‘The child does not sing a song for his mother.’
In (30), the child may be singing, but he is doing this for his own entertainment, not for his mother to be happy. Alternatively, he might be supposed to sing, for example in honour of his mother, but he refuses to do so. The third interpretation, ‘the child does not sing (refrains from singing) for his mother’s sake’, is possible if it is clear from the context that the mother does not want him to sing. This reading is more easily
Benefactive strategies in Thai
obtained in (31), where the mother can be seen as benefiting in some way from the child not smoking. (31) Lûuk mây sùup 'ùôrìi phш ˆ ә mεˆε. child neg smoke cigarette for mother ‘The child refrains from smoking for his mother’s sake.’
The phrasal negation mây 1ây can optionally be used to partly disambiguate the different interpretations and favour the readings with non-narrow scope verb negation, i.e. the negation covering the whole verbal expression or the benefactive phrase alone, as seen in (31ʹ). (31ʹ)
Lûuk mây 1ây sùup 'ùôrìi phш ˆ ә mεˆε. child neg get smoke cigarette for mother ‘The child does not smoke for his mother (but for his own joy).’ Not: ‘The child refrains from smoking for his mother’s sake.’
3.2â•… Semantic features There are important differences in use between hây and phш ˆ ә, which can be explained by the historical development of the two markers. The prototypical argument introduced by hây is a recipient, either of a physical entity (theme) or, in a more abstract sense, the experiencer of the result of an action. This argument is prototypically highly animate or human, and receives the theme or result of the action consciously. In more recent developments this restriction is weakened or completely absent, as seen above, but the use with inanimate arguments is still rather marginal in the language. On the other hand, phш ˆ ә originally marked the cause or reason of an activity. This cause does not need to be conscious of its being the cause of an action, therefore does not need to be human or animate. Later the function of phш ˆ ә was shifted from causing entity to purpose, a semantic shift that is not uncommon in the world’s languages.4 The beneficiary argument introduced by phш ˆ ә may or may not be conscious of and is not directly affected by the action; therefore the label indirect benefactive seems appropriate for this preposition. The difference in consciousness of the beneficiary leads to a distinction in Thai which is not seen in the typologies proposed for benefactives in other languages, viz. direct vs. indirect benefactive (see Section 1 for a summary of the relevant typologies). The difference is illustrated in Examples (32) and (33): (32) Mεˆε rf¢ffŋ phleeŋ hây luˆuk. mother sing song give child ‘The mother is singing a song to her child.’
4.â•… Cf. English for, which can be used as a causal subordinator or French pourquoi ‘why’, i.e. ‘for what’ or ‘because of what’.
 Mathias Jenny
(33) Mεˆε rf¢ffŋ phleeŋ phш ˆ ә lu ˆ uk. mother sing song for child ‘The mother is singing a song for her child.’
While the child must be listening (or at least be present and able to listen) to the song his mother is singing in (32), in (33) the child does not necessarily have to be present. The mother simply sings in order that the child benefits in some way. She may be a street singer, collecting money for her child, or she may be singing at a school event because her child wants her to go on stage, although he may not himself be at the event. While it sounds odd to add ‘… but the child doesn’t know it’ to (32), it is perfectly natural to do so in (33). Another (related) difference is the notion of recipient, which is present in V+hây constructions if the main verb allows such a reading, whereas phш ˆ ә does not imply a recipient participant, even if the main verb seems to favour this interpretation. This difference follows naturally from the first one concerning consciousness of the affectee. The recipient of a physical theme is normally conscious of the fact that he receives something. Compare the different interpretations of (34) and (35): (34)
Mεˆε sY ´ Y phoˇnlәmáay hây phfˆf. mother buy fruit give father ‘The mother bought fruit for the father.’ or ‘The mother bought the father fruit.’
(35) Mεˆε sY ´ Y phoˇnlәmáay phш ˆ ә phfˆf mother buy fruit for father ‘The mother bought fruit for the father.’
In (34) the father receives the fruit, while in (35) this is not necessarily the case. Maybe the mother wants the father to eat healthy food so she bought the fruit to please him. In this context the original causal meaning of phш ˆ ә is still clearly present. Example (36) further illustrates the direct versus indirect affectedness of the object. While in (36a) the verb tὲŋ.tuә is used as an intransitive (reflexive) predicate, in (36b) and (36c) an object is introduced by the V2 hây and the preposition phш ˆә respectively. It is obvious that in (36b) the child is directly and, in this case, physically affected by his mother dressing him, so the situation is best described using the direct benefactive construction. In (36c), by contrast, the father might be pleased to see his wife dressed up (at least that is her intention), but he is not directly involved in the activity or the result of it. The indirect benefactive construction is therefore the natural choice in this context. (36) a.
Mεˆε tὲŋ.tuә. mother dress ‘The mother gets dressed.’
Benefactive strategies in Thai 
b. Mεˆε tὲŋ.tuә hây lûuk. mother dress give child ‘The mother dresses the child.’ c.
Mεˆε tὲŋ.tuә phш ˆ ә phfˆf. mother dress for father ‘The mother dresses up for the father.’
In Example (37a), the teacher may have his hands full and the student thus opens the door for him to enter the room, while in (37b) the students merely perceives it as the teacher’s wish that the door be open. (37) a.
Nák.riәn p?`?t prәtuu hây khruu. student open door give teacher ‘The student opened the door for the teacher.’
b. Nák.riәn p?`?t prәtuu phш ˆ ә khruu. student open door for teacher ‘The student opened the door for the teacher(’s sake).’
The use of hây seems to be restricted with intransitive predicates, whereas phш ˆ ә can occur also with intransitive verbs, as seen in (38a). The only natural interpretation of (38b) is as deputative benefactive, in which case the beneficiary would normally be omitted. (38) a.
Chaˇn yùu phш ˆ ә th??. 1fam stay for 2fam ‘I live (my life) for you.’
b. Chaˇn yùu hây (?th??). 1fam stay give (?2fam) ‘I will stay (here) for (you, so you can go out).’
Probably, the main factor causing the restriction is the possible result of an action which is absent in the case of yùu, rather than transitivity. In order to get a plain or recipient benefactive reading, the predicate must express an event that has a result which can conceivably be transferred to the beneficiary.
4.â•… Conclusion Thai employs different strategies to express benefactive situations. The most important ones, and the ones exhibiting the highest degree of grammaticalisation, are the secondary verb hây ‘give’ and the preposition phш ˆ ә ‘for’. While the use of the former has been extended to express ‘altruistic’ or ‘allocentric’ actions in general, as well as to introduce an affectee participant (recipient, experiencer) or ‘goal of action’ to a situation, the latter
 Mathias Jenny
is largely restricted to benefactive situations with the beneficiary possibly unaware of the benefactive event and only indirectly affected by it. In terms of the typologies presented in Section 1, hây covers all types of benefactives (recipient, deputative, pure benefactive), while phш ˆ ә usually excludes recipient and deputative beneficiaries due to the fact these are directly affected by and aware of the benefactive action. The main difference between the two constructions is that hây introduces a conscious, direct beneficiary and phш ˆ ә an indirect, possibly unconscious beneficiary. This dichotomy is not accounted for in the typologies of benefactives proposed by other authors, as illustrated in Section 1, making language specific categories necessary. My proposal for the Thai system presented in this paper is given in Table€2. The additional benefactive as described in Section 2.2 is added here tentatively as its grammatical status is not entirely clear and should be subject to further investigation. Table 2.╇ Benefactive system of Thai Concrete
Grammaticalised Direct
Cause
phr¢f ô
Plain Ben.
saˇmràp
hây (kә)
Recipient Ben. Deputative Ben.
hây, líәŋ, fàak thεεn
hây (kә) hây
Indirect
Additional
(phш ˆ ә) phш ˆә phш `ә phш `ә
References Bisang, Walter. 1992. Das Verb im Chinesischen, Hmong, Vietnamesischen, Thai und Khmer [Vergleichende Grammatik im Rahmen der Verbserialisierung, der Grammatikalisierung und der Attraktorpositionen]. Tübingen: Gunter Narr. Iwasaki, Shoichi & Ingkaphirom, Preeya. 2005. A Reference Grammar of Thai. Cambridge: CUP. Kittilä, Seppo. 2005. Recipient-prominence vs. beneficiary-prominence. Linguistic Typology 9(2): 269–297. Peterson, David. 2007. Applicative Constructions. Oxford: OUP. Song, Jae Jung. 2007. Korean benefactive particles and their meanings. Ms, University of Otago. Van Valin, Robert D., Jr. & LaPolla, Randy J. 1997. Syntax. Structure, Meaning and Function. Cambridge: CUP. Online corpora of Thai texts: Chulalongkorn corpus: http://www.arts.chula.ac.th/~ling/cgi-bin/ThaiConc/ SEAlang corpus: http://sealang.net/thai/bitext.htm
Korean benefactive particles and their meanings Jae Jung Song
University of Otago This article investigates the grammar of Korean benefactive constructions with particular reference to the benefactive particles, eykey, (l)ulwihay(se) and taysin. The meaning contribution of these particles is shown to be so specific that their use can be explained by direct reference to their individual meanings. In particular, the particle eykey, unlike the other two particles, is subject to certain restrictions. The existence of these restrictions can be better understood once the meaning of the particle is properly recognized (i.e. the goal engaging with the theme to its benefit, rather than the goal exercising possessive control over the theme). This recognition is argued to render avoidable or unnecessary a certain special constraint previously proposed for the constructions. Also discussed briefly in the guise of a conclusion are the implications of the findings of the article for Construction Grammar, in which constructional meaning reigns over the meanings of individual items.
1.â•… Introduction Korean has as many as three benefactive particles, namely eykey, (l)ulwihay(se) and taysin, as illustrated in (1), (2) and (3), respectively.1 1.â•… These are (postpositional) particles, but they will—in line with the convention in Korean linguistics—be linked up with preceding nominals by means of a hyphen in order to clearly identify the nominals they go with. Abbreviations used in glosses: acc ben dat d.ben gen imp ind loc neg nom
accusative benefactive dative deputative-benefactive genitive imperative indicative locative negative nominative
p.ben pf pl pst purp q rel top voc
plain-benefactive phonological filler plural past purposive question relative topic vocative.
Jae Jung Song
(1) kiho-ka ai-eykey swukcey-lul hay-cwu-ess-ta Keeho-nom child-ben homework-acc do-give-pst-ind ‘Keeho did the homework for the child.’ (2) kiho-ka ai-lulwihay(se) swukcey-lul hay-ess-ta Keeho-nom child-ben homework-acc do-pst-ind ‘Keeho did the homework for the child.’ (3) kiho-ka ai-taysin swukcey-lul hay-cwu-ess-ta Keeho-nom child-ben homework-acc do-give-pst-ind ‘Keeho did the homework for the child.’
Surprisingly, there seems to be little or no detailed discussion of these benefactive particles, which tend to be lumped together under the heading of benefactive function.2 In the general literature (e.g. Sohn 1994:243–244 and Chang 1996:55), they are all characterized as encoding a broad meaning of benefaction. Specific studies on Korean benefactive constructions (e.g. Shibatani 1994; You 1997), on the other hand, place their focus exclusively on the use of the auxiliary verb cwu-, literally meaning ‘give’, as in (1), with little or no discussion of the particles themselves. The present chapter will demonstrate that the particles, illustrated in (1), (2) and (3), do not encode one and the same meaning of benefaction—as may be suggested by the fact that the three particles are all translated into the English preposition for—but different meanings of benefaction, with different grammatical characteristics. In point of fact, the particles each define the benefactive role so distinctly and precisely that their “indexing” function (e.g. Hopper & Thompson 1980:290–292; Wierzbicka 1980, 1981, 1988) must be properly recognized. This indexing view, which attributes meanings directly to individual items such as the benefactive particles in question, contrasts sharply with what has been proposed in Construction Grammar (or CxG) or other constructional approaches to grammar (e.g. Shibatani 1994, 1996 in particular). One of CxG’s major theoretical claims is that constructions themselves have meaning independently of the individual items that make up those constructions (Fillmore 1985; Fillmore & Kay 1993; Goldberg 1995, 2006; Fried & Östman 2004a and b, and Östman & Fried 2005 inter alia). Thus the present work is a contribution not only to the study of Korean grammar but also towards understanding
2.â•… The particle eykey dates from Middle Korean (the 15th century), and originally had a locative meaning, lost completely in Present Day Korean; the particle taysin, attested as early as in Middle Korean, is built on two Chinese elements (tay ‘instead’ and sin ‘body’); and the particle (l)ulwihay(se), attested for the first time in the late 17th century, consists of the native accusative particle ((l)ul), one Chinese element wi ‘fake’, and two native elements hay ‘do’ and se ‘as’ (based on Jae-il Kwon, p.c.).
Korean benefactive particles
a debate between the indexing and the constructional approach to meaning in general (e.g. van der Leek 2000; Boas 2003). The rest of the chapter is organized as follows. In Section 2, a general semantic typology of benefaction, as a point of departure, will be provided largely on the basis of Van Valin and LaPolla (1997) and also on Kittilä (2005 and 2006). In Section€3, a good number of examples involving the benefactive particles will be discussed with a view to providing an overall description of Korean benefactive constructions, i.e. the grammar of Korean benefactive constructions. Special attention will be paid to the restrictions on the use of the benefactive particle eykey (e.g. (1)), as opposed to the freely occurring benefactive particles (l)ulwihay(se) (e.g. (2)) and taysin (e.g. (3)). In Section 4, the meaning of the benefactive particle eykey will be redefined so as to provide an explanation for the restrictions on the occurrence of that particle. It will be argued that the particle eykey is used to encode the goal engaging with the theme, not the goal exercising possessive control over the theme (pace Shibatani 1994, 1996). The chapter closes with a summary and conclusion in Section€5.
2.â•… Benefactive and benefaction: A general semantic typology Benefactive is a semantic role that is assumed by an entity which benefits from another entity’s action. Benefaction is a concept that describes a situation or an event in which someone receives benefit from someone else’s action (or the object arising from someone else’s action). As Kittilä (2005:270) points out, however, “[t]he label [be it benefactive or benefaction] is usually taken for granted and not defined in any detail […]” (cf. Jackendoff 1990; Newman 1998 and 2005). There may be at least two (somewhat conflicting) reasons for this state of affairs. First, benefaction is a concept intuitively easy to understand: Someone benefits from someone else’s action. Second, the benefactive role is so closely intertwined with the recipient role that it is sometimes difficult to separate these two roles “although they may be distinguished conceptually” (Newman 1996:219). For instance, the two roles seem to imply each other in that when something comes into someone’s sphere of control or domain of possession, the latter is likely to make some use of (i.e. benefit from) the former. In fact, both the benefactive and recipient roles tend to be “present as part of the meaning of phrases described in grammars as benefactive or recipient” (Newman 1996:211; also see Kittilä 2005). Nonetheless, the benefactive role, separately from the recipient role, needs to be examined further with a view to arriving at a better understanding of benefaction. Broadly speaking, there seem to be at least three different ways Y can benefit from some action or object. First, Y may get enjoyment or pleasure out of X’s action, and this is taken to be some kind of benefit for Y. Second, Y may benefit from X’s action
Jae Jung Song
because X did what Y should or could have done. Third, Y may benefit from receiving some object arising from X’s action and making use of it. These different ways, when interpreted in the context of benefactive situations (that is, the agent intending for the beneficiary to benefit), give rise to three Â�different types of benefaction, namely (a) plain-benefactive; (b) deputative-benefactive; and (c) recipient-benefactive, as captured in Van Valin and LaPolla’s (1997:382–384) tripartite typology of benefaction. Van Valin and LaPolla (1997:384) define plain-benefactive to be “something like ‘the action of the [agent] provides [the beneficiary] with amusement, enjoyment or other kind of benefit’â•›” (cf. Newman’s (1996:218) “true” benefactive). This is exemplified in (4).
(4) Maria sings for the children.
In (4), Maria’s singing is done for the children’s enjoyment, or Maria sings, for example, because the children might not be able to fall asleep unless she does. In other words, the children receive benefit from Maria’s singing in the form of enjoyment or inducement. Deputative-benefactive refers to situations in which the agent does something “with the intention that the beneficiary not do the action” (Van Valin & LaPolla 1997:384). In (5), for example, Lizzie (or the agent) signed the contract in her boss’s (or the Â�beneficiary’s) stead.
(5) Lizzie signed the contract for her boss.
The sentence in (5) can potentially have a plain-benefactive reading as well (e.g. Lizzie signed the contract for her boss’s pleasure or amusement), although it is much less likely than the deputative-benefactive reading. The last type of benefaction, recipientbenefactive, is demonstrated by (6), in which the (intended) recipient is also a beneficiary in the sense that the recipient is likely to benefit from the pie that the agent baked, e.g. by eating it. Said differently, Kylie in (6) assumes a ‘hybrid’ role, that is a recipient-cum-beneficiary.
(6) Sean baked Kylie a pie.
The sentence in (6) can be slightly changed into (7), where Kylie is marked by the preposition for.
(7) Sean baked a pie for Kylie.
While in (7) Kylie can still be taken to be a(n intended) recipient (i.e. Sean baked a pie for Kylie to eat it), it can be understood also as either a plain or a deputative beneficiary. In other words, (7) can also be ambiguous between (8) and (9). (8) Sean baked a pie for Kylie’s pleasure. [plain-benefactive: Kylie wanted Sean to cook instead of getting a take-away and he baked a pie to please her]
Korean benefactive particles
(9) Sean baked a pie in Kylie’s stead. [deputative-benefactive: Sean baked a pie as a favor for Kylie because she was too busy to bake a pie herself]
Kittilä (2005), on the other hand, recognizes two types of beneficiary: (a) pure beneficiary; and (b) recipient-beneficiary. He draws upon two basic notions, i.e. reception and benefaction, to account for these two benefactive types. Reception refers to “[the] result of an event [in which] a (concrete) entity enters a recipient’s sphere of control or domain of possession” whereas benefaction describes “the occurrence of an event [that] is beneficial to an entity other than the agent itself ” (Kittilä 2005:273). Benefaction, however, is to be understood further in terms of substitutive benefaction and concrete benefaction. Kittilä’s substitutive benefaction is akin to Van Valin and LaPolla’s deputative-benefactive, whereas his concrete benefaction, contrary to what its label may suggest, involves a situation “in which the beneficiary benefits from an event in that s/he can make some use of its result, but in contrast to reception, without receiving anything concrete” (Kittilä 2005:273). Concrete benefaction does not seem to have a distinguishing function because recipients, beneficiaries and recipient-beneficiaries can all potentially have this particular property. Recipients are distinguished from beneficiaries because the former involve reception, not (substitutive) benefaction, whereas the latter involve (substitutive) benefaction, not reception. Recipients are also taken to be different from recipient-beneficiaries because, although both involve reception, it is only the latter that involve (substitutive) benefaction. Lastly, beneficiaries and recipient-beneficiaries are distinguished from each other because the former do not involve reception whereas the latter do. In Kittilä’s typology of benefaction, beneficiaries always seem to be deputativebeneficiaries as well, because both beneficiaries and recipient-beneficiaries have the property of substitutive benefaction, as opposed to recipients, which lack that property. This, however, is odd, because some beneficiaries are what Van Valin and LaPolla (1997) refer to as plain beneficiaries, as exemplified in (4); the sentence in (4) means (10), although it also has the less likely meaning of (11). (10) Maria sings for the children’s enjoyment. [plain-benefactive] (11) Maria sings in the children’s stead. [deputative-benefactive]
Thus it seems that substitutive benefaction is not a basic conceptual notion to be used for the understanding of benefaction, but it is a proper type of benefaction in its own right. In this chapter, therefore, Van Valin and LaPolla’s semantic typology of benefaction will be adopted, whereby the three different types of benefaction are identified: (a) plainbenefactive; (b) deputative-benefactive; and (c) recipient-benefactive. Needless to say, this tripartite typology of benefaction is not meant to be exhaustive or complete. There may be other (sub)types of benefaction. Further modification, based on in-depth investigations of individual languages, cannot be ruled out (for a similar point, see Kittilä 2005:295; and also see 4.2 in particular).
Jae Jung Song
3.â•…The three benefactive particles in Korean: eykey, (l)ulwihay(se) and taysin The present section will be concerned mainly with the syntax and semantics of the three benefactive particles, eykey, (l)ulwihay(se) and taysin, although attention will be paid also to the auxiliary verb cwu- ‘give’ (or its humble form tuli-),3 to be Â�utilized in conjunction with two of the particles, namely eykey and taysin. Each of these Â�particles exclusively encodes one of the three different types of benefaction recognized in Van Valin and LaPolla’s typology.
3.1â•… Benefactives in Korean: Forms and meanings The three particles, to be differently glossed below, are all glossed as ben in (1), (2) and (3) above. This suggests that the meaning of the preposition for in English is somewhat vague or underspecified, because these sentences express different types of benefaction. First, the benefactive particle eykey in (1) is actually what is recognized in the literature as the dative particle, as exemplified in (12). (12) kiho-ka ai-eykey chayk-ul cwu-ess-ta Keeho-nom child-dat book-acc give-pst-ind ‘Keeho gave the book to the child.’
In (12), the dative particle is used to mark the nominal ai as someone who received the book from the agent Keeho. Recall that the same particle is also used to encode the recipient-benefactive role in (1). (The particle eykey will continue to be glossed as dat in the rest of this chapter.) In (1), the homework was done by Keeho for the child’s benefit. That is, Keeho, not the child, did the homework, but importantly the homework is also understood to have been “transferred” from Keeho’s domain of possessive control into the child’s (cf. Shibatani 1994, 1996). (This concept of possessive control, however, will be revisited in 4.1, for it is inadequate at least for Korean.) The second benefactive particle, (l)ulwihay(se), encodes the plain-benefactive (p.ben) role, as in (2). Keeho’s action (i.e. doing the homework) provided the child with “amusement, enjoyment or other kind of benefit” (Van Valin & LaPolla 1997:384). The third benefactive particle taysin, illustrated in (3), is used to express the deputativebenefactive (d.ben) role. Keeho did the homework with the intention that the child not
3.â•… The humble form tuli- must be selected in lieu of the plain form cwu- when the recipient has higher social and/or economic standing than the referent of the subject nominal (e.g. Song 2005:93–95).
Korean benefactive particles
(have to) do the homework. The child benefited from not having to do what she was supposed to do because Keeho did it in her stead. When the beneficiary nominal is marked by the particle eykey or taysin, the main verb cannot stand on its own but has to be augmented by the auxiliary verb cwu- ‘give’.4 This is illustrated in (13) and (14). (13) *kiho-ka ai-eykey swukcey-lul ha-ess-ta Keeho-nom child-dat homework-acc do-pst-ind ‘Keeho did the homework for the child (i.e. to have it).’ (14) *kiho-ka ai-taysin swukcey-lul ha-ess-ta Keeho-nom child-d.ben homework-acc do-pst-ind ‘Keeho did the homework in the child’s stead.’
Note that the sentence in (14) is grammatical under the reading: ‘Keeho, instead of the child, did the homework.’ In this ‘grammatical’ reading, however, the child is not the beneficiary of Keeho’s action, but merely a potential agent who would have performed the action. Suppose that the teacher initially wanted the child to do the homework, but she eventually asked Keeho to do so. In this situation, the child is not a deputative beneficiary but merely a potential agent. If, on the other hand, the sentence in (14) is to be understood with the child as a deputative beneficiary (i.e. Keeho did the homework with the intention that the child not do so), it will be ungrammatical, because the deputative-benefactive meaning requires the use of the auxiliary verb cwu-. The benefactive particle (l)ulwihay(se), on the other hand, appears without the Â�support of the auxiliary verb cwu-, as in (2).5 This does not mean that they do not co-occur at all. They certainly do. However, the use of the auxiliary verb cwu- in conjunction with the plain-benefactive particle (l)ulwihay(se) indicates that there is a recipient- or deputative-beneficiary ‘lurking in the background’, as it were. For instance, consider: (15) kiho-ka ai-lulwihay(se) swukcey-lul hay-cwu-ess-ta Keeho-nom child-p.ben homework-acc do-give-pst-ind ‘Keeho did the homework for the child(’s pleasure).’
4.â•… What this suggests is that ai[-eykey] (13) and ai[-taysin] in (14) are arguments, not Â�adjuncts. The reason why these two sentences are ungrammatical is that the argument status of the said NPs cannot be supported by the main verb ha- ‘do’ (i.e. a two-place predicate) alone (cf. (16) and (17)). 5.â•… This indicates that in (2) the NP ai[-lulwihayse] is an adjunct, not an argument (cf. note 3). Thus, (2) will be fully grammatical even if the said NP is left out completely or is not at all recoverable from the context. Note, in this context, that the verb ha- is a two-place predicate.
Jae Jung Song
The sentence in (15) means that the child is not the one who benefited directly from the homework done, but merely a plain beneficiary who benefited from the event in which Keeho did the homework for a third person (to have it) or in a third person’s stead. The person who benefited directly from Keeho’s action may thus be either a recipient beneficiary or a deputative beneficiary—albeit unspecified in (15). This explains why a recipient or deputative beneficiary can, in fact, be added to (15) for the sake of completeness, as in (16) or (17), respectively. (‘Keeho’ is a male name and ‘Yonghee’ a female name.) (16) kiho-ka ai-lulwihay(se) yenghi-eykey Keeho-nom child-p.ben Yonghee-dat swukcey-lul hay-cwu-ess-ta homework-acc do-give-pst-ind ‘For the child(’s pleasure), Keeho did the homework for Yonghee (to have it).’ (17) kiho-ka ai-lulwihay(se) yenghi-taysin Keeho-nom child-p.ben Yonghee-d.ben swukcey-lul homework-acc
hay-cwu-ess-ta do-give-pst-ind
‘For the child(’s pleasure), Keeho did the homework in Yonghee’s stead.’
In (16), Yonghee benefited from receiving the completed homework from Keeho, whereas in (17), Yonghee benefited from Keeho doing the homework in her stead. In both (16) and (17), however, it was the child who received benefit (e.g. pleasure) from what had taken place between Keeho and Yonghee.
3.2â•… The (underspecified) meaning of the auxiliary verb cwuAs has been shown in 3.1, the auxiliary verb cwu- ‘give’ must be used in conjunction with eykey or taysin-marked nominals. This indicates that the auxiliary verb contributes something to the benefactive construction. What can this be? One could say that it is the meaning of non-plain benefaction (that is, deputative- or recipient-benefactive), because the plain-benefactive particle (l)ulwihay(se) does not call for the use of the auxiliary verb cwu-. In CxG, this meaning will be taken to be constructional: The benefactive construction, built on the verb cwu-, supplies the beneficiary argument to be integrated with the arguments licensed by the other (i.e. main) lexical verb (e.g. ha(y)- ‘do’ as in (1)) (e.g. Shibatani 1994, 1996). The problem with this constructional analysis is, however, that what the benefactive construction is claimed to contribute to meaning is not specific or precise enough to determine the exact type of benefaction (is it deputative- or recipient-benefactive?). It is only in the context of the benefactive particles that it becomes possible to tell whether one is dealing with either a deputative or a recipient beneficiary. The most the auxiliary verb cwu- contributes to meaning is to signal the presence of an underdetermined non-plain beneficiary (argument).
Korean benefactive particles 
This can be demonstrated in at least three ways. First consider (18), in which the beneficiary is left unexpressed, for example, because the speaker mistakenly assumes that the hearer already knows the identity of the beneficiary. (18) kiho-ka swukcey-lul hay-cwu-ess-ta Keeho-nom homework-acc do-give-pst-ind ‘Keeho did the homework (for someone).’
If (18) is heard in a situation like this, there will be genuine ambiguity around the nature of benefaction. The unspecified beneficiary can be either a deputative or a recipient beneficiary. This can be shown to be the case by the following two questions, either of which can be used to ask for the type of benefaction to be clarified. (19) a.
yenghi-eykey hay-cwu-ess-e? Yonghee-dat do-give-pst-q ‘Did he do it for Yonghee (to have it)?’
b. yenghi-taysin hay-cwu-ess-e? Yonghee-d.ben do-give-pst-q ‘Did he do it in Yonghee’s stead?’
The fact that these questions can be interchangeably used in response to (18) is a strong piece of evidence that the meaning of benefaction (or the constructional meaning of benefaction in CxG) is underspecified by the auxiliary verb cwu-. Similarly, so-called alternative questions can be asked in order to identify the nature of benefaction, as in: (20) kiho-ka yenghi-eykey swukcey-lul hay-cwu-ess-ni Keeho-nom Yonghee-dat homework-acc do-give-pst-q ani-myen yenghi-taysin swukcey-lul hay-cwu-ess-ni not-if Yonghee-d.ben homework-acc do-give-pst-q ‘Did Keeho do the homework for Yonghee to have it or in Yonghee’s stead?’
The question in (20) clearly demonstrates that the meaning contributed by the auxiliary verb cwu- is underspecified. If the exact nature of benefaction were directly encoded by the auxiliary verb, the need to ask questions such as (20) would not arise in the first place. Finally, the underspecification of benefaction in (18) is also supported by the fact that the benefactive particles eykey and taysin can co-occur in one and the same sentence, as in: (21) kiho-ka yenghi-taysin ai-eykey swukcey-lul hay-cwu-ess-ta Keeho-nom Yonghee-d.ben child-dat homework-acc do-give-pst-ind ‘Keeho did the homework for the child (to have it) in Yonghee’s stead.’
It is not possible to argue that the auxiliary verb cwu- contributes an exact meaning of benefaction to the sentence in (21), because the two different types of benefaction, i.e.
 Jae Jung Song
deputative-benefactive and recipient-benefactive, are expressed in one and the same sentence. It is very difficult to claim that the benefactive construction, based on the verb cwu-, contributes the two different meanings of benefaction simultaneously.6 The function of the auxiliary verb in benefactive constructions is, therefore, merely to indicate the existence of the non-plain benefactive role, the exact nature of which is to be determined by the benefactive particle itself.
3.3â•… The intransitivity restriction on the recipient-benefactive particle eykey So far, the benefactive particles have been exemplified in the context of transitive sentences. This does not mean that they cannot appear in the context of intransitive sentences as well. Two of the particles, (l)ulwihay(se) and taysin, certainly can, but there is a restriction on the use of the particle eykey. This is illustrated by the following three examples. (22) *kiho-ka yenghi-eykey sicang-ey ka-cwu-ess-ta Keeho-nom Yonghee-dat market-to go-give-pst-ind ‘Keeho went to the market for Yonghee.’ (23) kiho-ka yenghi-lulwihay(se) sicang-ey ka-ess-ta Keeho-nom Yonghee-p.ben market-to go-pst-ind ‘Keeho went to the market for Yonghee(’s pleasure).’ (24) kiho-ka yenghi-taysin sicang-ey ka-cwu-ess-ta Keeho-nom Yonghee-d.ben market-to go-give-pst-ind ‘Keeho went to the market in Yonghee’s stead.’
6.â•… The sentence in (21) can also mean something like ‘Keeho, instead of Yonghee, did the homework for the child (to have it)’. In this alternative reading, the particle taysin merely encodes the potential agent role. Moreover, these two benefactive particles can also co-occur together with the plain-benefactive particle, (l)ulwihay(se), as in: (i)
kiho-ka chelsu-lulwihay(se) yenghi-taysin ai-eykey Keeho-nom Chelsoo-p.ben Yonghee-d.ben child-dat
swukcey-lul hay-cwu-ess-ta homework-acc do-give-pst-ind
The sentence in (i) is not easy to translate idiomatically into English, but it means something like: ‘Keeho did the homework for the child to have it in Yonghee’s stead for Chelsoo’s pleasure’. Thus three people benefited from Keeho’s work, but the types of benefaction that they received are all different.
Korean benefactive particles
What is interesting in this context is that the auxiliary verb cwu- can be used in conjunction with intransitive verbs even when no beneficiary is ‘overtly expressed’ (Shibatani 1994:57), as in: (25) kiho-ka sicang-ey ka-cwu-ess-ta Keeho-nom market-to go-give-pst-ind ‘Keeho went to the market (for someone).’
Examples like (22) and (25) lead Shibatani (1994:52) to suggest that the “omitted” or suppressed benefactive nominal in (25) is the one marked by the recipient-benefactive particle eykey. This, however, is strange, because if and when the benefactive nominal appears on the surface, as in (22), the sentence becomes totally ungrammatical. While it is to be discussed in 4.4, it needs to be pointed out here that the omitted nominal in (25) is not recipient-benefactive (or marked by eykey), but deputative-benefactive (or marked by taysin). (The omitted nominal cannot be plain-benefactive (or marked by (l)ulwihay(se)) because of the presence in (25) of the auxiliary verb cwu- (cf. You 1997:467)). Thus, while agreeing that the auxiliary verb can co-occur, without the support of an expressed benefactive nominal, with intransitive verbs as in (25), this chapter disagrees with Shibatani’s (1994, 1996) proposal that the benefactive role suppressed in (25) is recipient-benefactive.
3.4â•… Further restrictions on the recipient-benefactive particle eykey The discussions contained in 3.1, 3.2 and 3.3 might suggest that the benefactive particles can be used without any restrictions in the context of transitive sentences. Nothing is further from the truth, as already noted by Shibatani (1994, 1996) (and also by You (1997)). One of the benefactive particles, namely recipient-benefactive eykey, actually appears in some but not other cwu-based transitive sentences, while the other two benefactive particles can occur freely in transitive sentences. Consider the following examples, with the particle eykey (adapted from Shibatani (1994:43)). (26) kiho-ka yenghi-eykey mwun-ul yel-e-cwu-ess-ta Keeho-nom Yonghee-dat door-acc open-pf-give-pst-ind ‘Keeho opened the door for Yonghee.’ (27) *kiho-ka yenghi-eykey mwun-ul tat-a-cwu-ess-ta Keeho-nom Yonghee-dat door-acc close-pf-give-pst-ind ‘Keeho closed the door for Yonghee.’ (28) *kiho-ka yenghi-eykey changmwun-ul yel-e-cwu-ess-ta Keeho-nom Yonghee-dat window-acc open-pf-give-pst-ind ‘Keeho opened the window for Yonghee.’
Jae Jung Song
(29) kiho-ka yenghi-eykey ipwul-ul kkal-a-cwu-ess-ta Keeho-nom Yonghee-dat futon-acc lay.out-pf-give-pst-ind ‘Keeho laid out the futon for Yonghee.’ (30) *kiho-ka yenghi-eykey ipwul-ul kay-e-cwu-ess-ta Keeho-nom Yonghee-dat futon-acc fold.up.put.away-pf-give-pst-ind ‘Keeho folded up the futon and put it away for Yonghee.’
Opening a door for someone (26) is acceptable for the eykey + cwu-based benefactive construction, but closing a door for someone (27) is not. Similarly, laying out a futon for someone (29) can be accommodated by the same benefactive construction, but folding up a futon and putting it away for someone (30) cannot be. (In traditional Korean rooms, known as ontol pang, futons are spread for sleeping and stored away in wardrobes at other times.) What is more obfuscating is the contrast in grammaticality between (26) and (28). It is not possible to use the eykey + cwu-based benefactive construction in the case of opening a window for someone (28), while it is perfectly grammatical to draw upon the same construction in the case of opening a door for someone (26). Thus “even the same verb shows a measurable difference in the acceptability depending on the situation being described” (Shibatani 1994:44). Therefore, the real issue is not “concerned with the transitivity of the verb base [or the main verb]” (Shibatani 1994:44). Note, however, that the other two particles, i.e. (l)ulwihay(se) and taysin, do not behave like the particle eykey at all, as can be seen in the following examples, in which they are freely used to describe the same ‘unacceptable’ situations illustrated above. (31) kiho-ka yenghi-lulwihay(se) mwun-ul tat-ass-ta Keeho-nom Yonghee-p.ben door-acc close-pst-ind ‘Keeho closed the door for Yonghee(’s comfort).’ (32) kiho-ka yenghi-lulwihay(se) changmwun-ul yel-ess-ta Keeho-nom Yonghee-p.ben window-acc open-pst-ind ‘Keeho opened the window for Yonghee(’s comfort).’ (33) kiho-ka yenghi-lulwihay(se) ipwul-ul kay-ess-ta Keeho-nom Yonghee-p.ben futon-acc fold.up.put.away-pst-ind ‘Keeho folded up the futon and put it away for Yonghee(’s comfort).’ (34) kiho-ka yenghi-taysin mwun-ul tat-a-cwu-ess-ta Keeho-nom Yonghee-d.ben door-acc close-pf-give-pst-ind ‘Keeho closed the door in Yonghee’s stead.’ (35) kiho-ka yenghi-taysin changmwun-ul yel-e-cwu-ess-ta Keeho-nom Yonghee-d.ben window-acc open-pf-give-pst-ind ‘Keeho opened the window in Yonghee’s stead.’
Korean benefactive particles
(36) kiho-ka yenghi-taysin ipwul-ul kay-e-cwu-ess-ta Keeho-nom Yonghee-d.ben futon-acc fold.up.put.away-pf-give-pst-ind ‘Keeho folded up the futon and put it away in Yonghee’s stead.’
4.╅ The recipient-benefactive particle eykey: Restrictions and meaning Section€3 has identified three problems associated with Korean benefactive constructions that need to be accounted for. First, the benefactive particle eykey cannot be used in the context of cwu-based intransitive sentences. There is thus a co-occurrence restriction on the recipient-benefactive particle eykey and the auxiliary verb cwu- in intransitive sentences. Related to this co-occurrence restriction is the second problem. Recall that in Shibatani (1994) intransitive sentences which are ungrammatical because of the violation of the co-occurrence restriction are said to be rendered grammatical if the eykeymarked nominal is removed. This, however, is not correct. Why it is not correct will need to be demonstrated. To that end, the second problem will be examined in conjunction with the first. Third, it needs to be explained why some transitive sentences allow the co-occurrence of the auxiliary verb cwu- and the particle eykey while other transitive ones do not. As already illustrated, even transitive sentences based on same verbs may fluctuate between grammaticality and ungrammaticality (or between acceptability and unacceptability). These problems are all directly related to the restrictions on the particle eykey; the other two particles (l)ulwihay(se) and taysin can be used, without any restrictions, in not only intransitive but also transitive sentences. This strongly suggests that a solution (or solutions) to the problems should be sought, as will indeed be in 4.2, from (the meaning or function of) the recipient-benefactive particle eykey itself. But first how the problems have previously been dealt with (i.e. Shibatani 1994) needs to be discussed briefly as a backdrop for the solution to be developed in this chapter.
4.1â•…The benefactive construction defined: The goal’s possessive control over the theme Shibatani (1994) provides a solution to the three problems mentioned above. His solution is based on his definition of the benefactive construction (or what he calls the giveschema): The agent (coded as the subject) creates on behalf of the goal (coded as the indirect object) a possessive situation in which the goal exercises control over the theme (coded as the direct object). (Note that the beneficiary is referred to as the goal in Shibatani’s definition; for the sake of consistency the latter term will hereafter be adopted, where possible.) If a scene or situation described by a given sentence agrees
Jae Jung Song
with this definition, that sentence will be grammatical. If it does not in any respect, the sentence will turn out to be ungrammatical. What ultimately ‘dictates’ grammaticality (or acceptability) is the goal’s possessive control over the theme (Shibatani 1994:46). This is not surprising because the result of the agent’s action, as described in Shibatani’s definition, is the goal’s ability to exercise possessive control over the theme. Note that Shibatani (1994) does not define the meaning of the recipient-benefactive particle eykey per se but, in the spirit of CxG, ascribes the meaning of the goal directly to the cwubased benefactive construction. In other words, the goal is contributed to the overall meaning by the benefactive construction. First, the recipient-benefactive particle eykey cannot be used in cwu-based intransitive sentences, as in (22) above, because the intransitively expressed situation does not involve an entity (i.e. theme) which the goal can exercise possessive control over (Shibatani 1994:46, 68–69). In (22), the lexical verb ka- ‘to go’, being an intransitive verb, does not provide a theme but only an agent, although the constructional meaning of benefaction, provided by the cwu-based benefactive construction, contributes a goal to the overall meaning. There is, therefore, no theme that the goal can link itself up with. This lack of correspondence is taken to explain the ungrammaticality of (22). Second, the reason why the deletion of the eykey-marked nominal renders (22) grammatical, as in (25) above, is that the goal (contributed by the cwu-based benefactive construction) does not have to be associated with anything concrete (i.e. the theme) because the former is not formally realized (i.e. omitted) (Shibatani 1994:60–61). Thus the absence of the theme causes no problems provided that the goal remains unexpressed. Third, when and where the theme is not something that literally comes into the goal’s domain of possessive control, whether or not the recipient-benefactive particle eykey is permitted in cwu-based transitive sentences depends on whether or not the theme can be “metonymically” construed as coming under the goal’s possessive control (Shibatani 1994:67–68). Consider the relevant examples in (26), (27), (29) and (30) above. When X opens a door for Y, it is expected that Y will enter or exit the room through the opened door. Said differently, “the passage created by the opening of the door” is metonymically construed as something that comes under Y’s possessive control (Shibatani 1994:68). On the other hand, when X closes a door for Y, it is not expected that Y will control or “possess” the use of the door. For example, people cannot go through a closed door (i.e. no passage created by X’s action). To wit, there is the metonymic construal of possessive control in (26), whereas there is no such construal in (27). This difference is taken to explain the contrast in grammaticality between the two sentences. A similar comment can be made of the difference in grammaticality between (29) and (30). Spreading a futon for Y leads one to infer that Y will sleep on it, whereas folding up a futon and storing it away (in a wardrobe) for Y does not give rise to a similar inference. In the former case the spread futon is metonymically construed as coming into Y’s domain of possessive control whereas in the latter case such a construal is not plausible.
Korean benefactive particles
There is, however, some difficulty with this explanation based on the concept of possessive control. First, there are situations to which the concept of possessive control does not seem to apply. In particular, there are benefactive sentences in which the theme may already be in the goal’s domain of possessive control at the time of the agent’s action. In such sentences, the question of the theme coming under the goal’s possessive control does not arise. Consider the following two examples ((37) taken from Shibatani (1994:44) and also You (1997:459)). (37) na-nun hanako-eykey kwutwu-lul takk-a-cwu-ess-ta I-top Hanako-dat shoe-acc polish-pf-give-pst-ind ‘I polished the shoes for Hanako.’ (38) yenghi-ka kiho-eykey meli-lul kkakk-a-cwu-ess-ta Yonghee-nom Keeho-dat hair-acc cut-pf-give-pst-ind ‘Yonghee cut the hair for Keeho.’
The normal interpretation of (37) is that the speaker polished Hanako’s shoes for Hanako herself. In this interpretation, the shoes belonged to Hanako at the time of the polishing action. It is even likely that Hanako kept the shoes on her feet while they were being polished. In this situation, one cannot speak of X creating a possessive situation over the shoes for Y, who already owns them. This difficulty with the concept of possessive control is even more evident in (38). The sentence cannot mean that Yonghee cut someone else’s hair for Keeho. The hair that Yonghee cut was part of Keeho. One cannot claim that Yonghee created a possessive situation over Keeho’s hair on his behalf when the hair belonged to him before, during and after the event. Even if Yonghee may be said to have had some kind of non-possessive control over Keeho’s hair while she was cutting it (i.e. Yonghee holding the strands of hair to be cut), one still cannot say that Keeho lost possessive control of his own hair in any way or at any time. To wit, the theme cannot be said to come under the goal’s possessive control when the former is already in the latter’s domain of possessive control.7
7.â•… An anonymous referee argues that the examples in (39) and (40) do not vitiate Shibatani’s concept of possessive control, because “â•›‘the new condition/appearance of shoes that are polished and hair that is cut’ [can also be metonymically construed as coming] under Y’s possessive control”. In order to accept this argument, however, one must accept that Y does not have possessive control over some of the (physical) properties of Y’s own inalienable (e.g. hair) or quasi-inalienable (e.g. shoes worn on feet) entities—at least not until X has created a possessive situation in which Y exercises control over them—while at the same time enjoying possessive control over the entities themselves. If so, one must also be ready to accept the logical possibility (or conclusion) that the entities in turn—metonymically speaking—lack possessive control over their own properties (because the properties are still under X’s possessive domain). This strikes one as incongruous. More to the point, the concept of engaging, as developed in this chapter, renders otiose Shibatani’s ‘selective’ metonymy—selective in the sense that metonymy is invoked whenever deemed necessary. Furthermore, the need to
Jae Jung Song
Second, although Shibatani (1994, 1996) does not discuss intransitive sentences with deputative-benefactive taysin or plain-benefactive (l)ulwihay(se), it needs to be emphasized that intransitive sentences are compatible with these two benefactive particles, as already discussed in 3.3. Consider the intransitive sentence in (24) above, with the deputative-benefactive particle taysin. (Recall that this particle—like the recipientbenefactive particle eykey but unlike the plain-benefactive particle (l)ulwihay(se)— requires the use of the auxiliary verb cwu-.) While there is no theme for the goal to exercise possessive control over in (24), just as in (22), the deputative-benefactive particle taysin occurs freely in cwu-based intransitive sentences. This is problematic and, more importantly, calls into question the constructional meaning of benefaction itself. Presumably, the goal is contributed to the overall meaning by the cwu-based benefactive construction in (24), just as in (22), but despite the lack of a theme the sentence is grammatical. Why isn’t the goal required to link up with a theme in (24), as it is in (22)? Whatever the answer to this question may be, one thing is glaringly obvious. The difference in grammaticality between the two sentences is directly related to the only linguistic difference between them (i.e. the different benefactive particles used). This strongly suggests that attention be paid not so much to the constructional meaning of benefaction as to the meanings of the individual particles themselves.
4.2â•…The recipient-benefactive particle eykey defined: The goal’s engaging with the theme In view of the difficulty discussed in 4.1, it will be proposed here that the function of the recipient-benefactive particle eykey be recognized as encoding the goal engaging with the theme. But what does it mean that the goal engages with the theme? The definition of engaging to be adopted here is an everyday one (e.g. Oxford English Dictionary). Engaging involves two entities. When Y engages with Z (e.g. The student engages with the book or The driver engages with the clutch), what Y does is to establish a connection or contact with Z. This connection or contact must be, to borrow Goldberg’s (1995:39–43) words, “humanly relevant”. What Y does in relation to Z when Y engages with Z must be something that humans are expected to do in relation to Z. Said differently, Y’s engaging with Z must be understood in terms of activities or scenes that make up human experience. When, for instance, reading a book, humans engage with it in the sense that they are making a connection with it (or more precisely its content). This connection is also humanly relevant, because reading is what humans
delimit the conditions under which metonymy must be invoked in Shibatani’s theory should not be lost sight of. Thus, contrary to the referee’s position, there indeed is much “point to these examples”.
Korean benefactive particles
do in relation to books. There may be other things that humans do in relation to books, e.g. photocopying, shelving, writing, etc. These activities are all part and parcel of human experience (of books). If, on the other hand, Y is unable to make a humanly relevant Â�connection or contact with Z, then Y cannot be said to engage with Z. What the agent (i.e. X), on the other hand, does is to create a situation in which the goal (i.e. Y) engages with the theme (i.e. Z) to the goal’s benefit, whether Y consumes Z (e.g. eating a pie), watches Z (e.g. watching a DVD) or makes use of Z (e.g. walking through an opened door). The restrictions on the recipient-benefactive particle eykey, summarized at the beginning of Section 4, must thus be understood in terms of this concept of engaging. The recipient-benefactive particle eykey is not permitted when and where the goal cannot be seen to engage with the theme to its benefit either because it is difficult, if not impossible, to imagine, in terms of human experience, that what the goal does in relation to the theme is humanly relevant (see 4.3) or because there is no theme to start with (see 4.4). Before the concept of the goal’s engaging with the theme is shown to account for the restrictions on the occurrence of the recipient-benefactive particle eykey, it needs to be pointed out that the compound term ‘recipient-benefactive’ is a misnomer. Reception in Kittilä’s sense (see Section 2), involving the concept of possessive control, is not necessarily part of the meaning of this particle—at least in the context of the cwu-based benefactive construction—but engaging is (also cf. de Stadler (1998:283–284) for the inadequacy of possessive control in some Afrikaans benefactive sentences). For this reason, the term ‘engager-benefactive’ should be used instead. The particle eykey will hereafter be referred to as engager-benefactive (although it will continue to be glossed as dat). The engager beneficiary is to be distinguished from the plain or deputative beneficiary, which is to be encoded by (l)ulwihay(se) or taysin, respectively.
4.3â•…Restrictions on eykey-marked nominals in cwu-based transitive sentences Whether or not the goal engages with the theme (in a humanly relevant manner) in benefactive situations can be illustrated by one of the problems summarized above: the incompatibility of some cwu-based transitive sentences with the use of the engager-benefactive particle eykey. The relevant examples to be referred back to are (26), (27), (29) and (30) above. If X opens a door for Y as in (26), Y is reasonably expected to engage with the door by going through it. This is a mundane human activity. But what also makes this activity humanly relevant engaging is that Y is making use of the door as opened by X. Closing a door for Y, on the other hand, does not involve such engaging. When X closes a door for Y in (27), Y may do little or nothing with it, let alone go through it. Said differently, while closing a door in itself is a normal human activity (i.e. part of human experience), it is very difficult to imagine Y making use of
 Jae Jung Song
(or engaging with) a closed door. Literally, the door is shut behind Y. What is lacking in (27) is the goal’s engaging with the theme, which explains the ungrammaticality.8 The difference in grammaticality between (29) and (30) can be similarly explained. If X spreads a futon for Y, the reasonable expectation, in terms of human experience, is that Y will engage with the futon to Y’s benefit by sleeping on it. If, on the other hand, X folds up a futon and then puts it away for Y (most probably after Y having slept on it), Y cannot be expected to engage with the futon (any more) in any humanly relevant manner, because it has been folded up and stored away in a wardrobe. There is no humanly relevant connection or contact to speak of. Once again, it is the lack of the goal’s engaging with the theme that explains the ungrammaticality of (30), as opposed to the grammaticality of (29). Now consider the sentence in (28) above. This ungrammatical sentence is even more interesting, especially as opposed to (26) (i.e. opening the door). Humans do not normally make use of opened windows like opened doors. For example, they don’t enter or exit a room through a window. People normally open windows to let fresh air in. The reason why it is unacceptable to say (28) is that it is difficult to imagine, in terms of human experience, that humans engage with opened windows to their benefit. However, as Shibatani (1994: 68) points out, the ungrammatical sentence in (28) may be accepted in a certain situation, for instance, in which X opens a window for Y so that Y can enter or exit a room through it because the door has jammed and won’t open. In this unusual but possible situation, Y can, again, be seen to be engaging with the window to Y’s benefit. In other words, getting in or out though an opened window can potentially be a humanly relevant experience, albeit attested under very special circumstances. What about the problematic situations in (37) and (38), which initially highlighted the difficulty with the concept of possessive control (i.e. Shibatani 1994, 1996)? If X polishes Y’s shoes for Y, Y is expected to do something in relation to the polished shoes in a humanly relevant manner. For instance, it is reasonable to assume, in terms of human experience, that X polishes shoes for Y to wear (or to engage with) them. Similarly, if X cuts Y’s hair, there is the expectation that Y will engage with his or her own
8.â•… Note that it is very difficult, not impossible, to imagine Y engaging with a closed door. Thus, if, as suggested by an anonymous referee, Yonghee were to paint the door while closed, the acceptability of the sentence in (27) would improve greatly, as in: (i)
phayinthu-lul cal chil-ha-lswuiss-key kiho-ka yenghi-eykey paint-acc well paint-do-able-purp kiho-nom Yonghee-dat
mwun-ul tat-a-cwu-ess-ta door-acc close-pf-give-pst-ind
‘Keeho closed the door for Yonghee so that she could paint it well.’
Korean benefactive particles 
hair (as cut by X) by the simple fact that the hair is an inalienable part of Y. Whether Y likes it or not, Y has to wear (or to engage with) Y’s own hair. The reader will note that in (38) the goal (or Keeho) cannot help engaging with the theme (or Keeho’s own hair), because the latter is an inalienable part of the former (irrespective of what the agent has done). This contrasts with the situation in (37), where the theme is not an inalienable part of the goal. What this difference suggests is that the theme in (37) (i.e. shoes) can be engaged with not only by its possessor (i.e. Hanako) but also by someone else, whereas the theme in (38) (i.e. hair) can be engaged with, at least in terms of normal human experience, by its possessor only. This in turn suggests that when the theme is modified by a genitive phrase expressing someone other than the goal as the possessor, the sentences in (37) and (38) are expected to behave differently. The prediction is that under the said circumstances (37) will be acceptable but (38) will not. Indeed it is borne out by (39) and (40). (39) na-nun hanako-eykey yoshimi-uy kwutwu-lul takk-a-cwu-ess-ta I-top Hanako-dat Yoshimi-gen shoe-acc polish-pf-give-pst-ind ‘I polished Yoshimi’s shoes for Hanako (to engage with them).’ (40) *yenghi-ka kiho-eyke chelswu-uy meli-lul kkakk-a-cwu-ess-ta9 Yonghee-nom Keeho-dat Chelsoo-gen hair-acc cut-pf-give-pst-ind ‘Yonghee cut Chelsoo’s hair for Keeho (to engage with it).’
The use of the engager-benefactive particle eykey on the nominal hanako ‘Hanako’ in (39) indicates that Hanako (would) somehow engage(d) with Yoshimi’s shoes in the given benefactive situation. The understanding is that Hanako would do something humanly relevant in relation to Yoshimi’s shoes (e.g. wearing them, as humans indeed borrow other humans’ shoes and wear them). The sentence in (40), on the other hand, is ungrammatical precisely because it is extremely difficult to imagine Keeho engaging with the hair naturally attached to someone else’s (i.e. Chelsoo’s) head in the described
9.â•… An anonymous referee points out that (40) is no more difficult to accept than (39) if it refers to a situation in which Keeho needs some hair to make a wig. This, however, is his or her confusion between the use of cwu- as an auxiliary verb, and of cwu- as a full verb. In the latter case (i.e. cwu- as a full verb), the sentence in (40) would mean ‘Yonghee cut Chelsoo’s hair and gave it to Keeho’ (i.e. Keeho as the recipient of Chelsoo’s hair, not as the beneficiary of Yonghee’s action); this is why, for such an interpretation, Keeho-eykey can optionally appear just before the full verb cwu-, as in: (i)
yenghi-ka chelswu-uy meli-lul kkakk-a, kiho-eykey cwu-ess-ta Yonghee-nom Chelsoo-gen hair-acc cut-and Keeho-dat give-PST-ind ‘Yonghee cut Chelsoo’s hair, and gave [it] to Keeho.’
This optional ‘movement’ of Keeho-eykey is not possible under the intended reading of (40), because it would intervene between the main and auxiliary verb.
 Jae Jung Song
benefactive situation when the engager-benefactive particle eykey is used to encode the nominal kiho ‘Keeho’. The only way to render (40) grammatical is to use a different benefactive particle, namely the deputative-benefactive particle taysin, as in: (41) yenghi-ka kiho-taysin chelswu-uy meli-lul kkakk-a-cwu-ess-ta Yonghee-nom Keeho-d.ben Chelsoo-gen hair-acc cut-pf-give-pst-ind ‘Yonghee cut Chelsoo’s hair in Keeho’s stead.’
To wit, the goal, marked by the engager-benefactive particle eykey, is reasonably expected to engage with the theme to its benefit in a humanly relevant manner, and it is the agent that makes such engaging possible.
4.4â•… The non-co-occurrence in intransitive sentences of cwu- and eykey First, reconsider (22) and (24) above. As already argued in 3.2, the function of the auxiliary verb cwu- is to signal the presence of an underdetermined non-plain beneficiary and nothing more or less. In other words, the benefactive role in the construction is largely underspecified with respect to its exact nature, which is instead expressed by the benefactive particle used in the construction. The reason why (24) is grammatical while (22) is not should thus be explained on the basis of which benefactive particle is chosen in a given sentence. The concept of the goal’s engaging with the theme can also explain why intransitive sentences, based on the auxiliary verb cwu-, cannot cooccur with the engager-benefactive particle eykey. In (22), Keeho went to the market. Going to the market in or by itself does not involve or create anything that the goal (or Yonghee) can engage with in any humanly relevant manner. Whether Keeho went to the market or not, the market is located where it has been, and Yonghee is left with nothing to engage with, let alone the market, in the first place. There is no theme— expressed or unexpressed—arising from the agent’s action. The engager-benefactive particle eykey, however, encodes the goal engaging with the theme to its benefit. It thus makes sense why the engager-benefactive particle is not acceptable in the context of (22). On the other hand, the deputative-benefactive particle taysin is acceptable in (24), because Keeho went to the market in Yonghee’s stead. In other words, Keeho went to the market with the intention that Yonghee not (have to) go there herself, and as a direct consequence she benefited from his action. The semantic role of Yonghee here is precisely what the deputative-benefactive particle taysin is supposed to encode. The second problem, mentioned at the beginning of Section 4, concerns the suggestion that the ungrammatical sentence in (22) can be rendered grammatical if the benefactive nominal is omitted, as in (25) (Shibatani 1994:52–55). The present chapter, on the other hand, argues that the omitted benefactive nominal in (25) is not the one marked by the engager-benefactive particle eykey, but the one marked by the deputativebenefactive particle taysin. (Note that the omitted benefactive nominal in (25) cannot
Korean benefactive particles 
be the one marked by the plain-benefactive particle (l)ulwihay(se), which does not requires the use of the auxiliary verb cwu-; cf. 3.1.) In other words, the grammatical sentence in (25) should be related to the grammatical sentence in (24), not to the ungrammatical sentence in (22). To put it differently, (25) is grammatical, because it is based on the grammatical sentence in (24). Thus (25), with the benefactive nominal removed, really means something like: ‘Keeho went to the market (in someone’s stead)’. There is evidence in support of the alternative suggestion. It is possible to retain the deputative-benefactive particle taysin in the context of (24) if and when the referent of the omitted noun phrase is readily recoverable from the context, linguistic or otherwise. For instance, it is possible to say: (42) kiho-ka taysin sicang-ey ka-cwu-ess-ta Keeho-nom d.ben market-to go-give-pst-ind ‘Keeho went to the market in (someone’s) stead.’
The sentence in (42) is ‘derived’ from (24), as evidenced by the retention of the particle taysin. However, the engager-benefactive particle eykey cannot be ‘retained’ likewise, as shown in: (43) *kiho-ka eykey sicang-ey ka-cwu-ess-ta Keeho-nom dat market-to go-give-pst-ind ‘Keeho went to the market for someone.’
The ungrammaticality of (43) has to do with the fact that the engager-benefactive particle eykey cannot stand alone without an associated noun phrase, but the important point being made here is that, if anything can be fully or partially removed, with the grammaticality of the sentence kept intact, it is the deputative-benefactive nominal, not the engager-benefactive one, that can. Similarly, it is possible to ask (44) in response to (25) if and when it proves to be impossible to recover the identity of the understood beneficiary from the context. (44) nuku-taysin who-d.ben ‘In whose stead?’
However, it is not possible to mark the interrogative pronoun nuku with the engagerbenefactive particle in response to (25), as in: (45) #nuku-eykey who-dat ‘For whom (to engage with it)?’
If the omitted noun phrase in (25) is an engager-benefactive one, the question in (45) should not be unacceptable or infelicitous (as indicated by #).
 Jae Jung Song
4.5â•… Engager-benefactive and the constraint on theme-sharing Shibatani (1994:62–66) draws attention to ungrammatical benefactive sentences such as (46) (with two goals, yenghi-eykey and ai-eykey, and one theme ku kes-ul; cf. *John gave Mary a book to Jane in English). (46) *kiho-ka yenghi-eykey ku kes-ul ai-eykey iyaki.hay-cwu-ess-ta Keeho-nom Yonghee-dat the thing-acc child-dat tell-give-pst-ind ‘Keeho gave Yonghee the benefit of telling the child the thing.’
This leads him to put forth “the constraint on theme-sharing”. This constraint is needed because “one and the same object cannot be transferred to two or more parties at the same time” (Shibatani 1994:63); a theme cannot be associated with more than one goal. It is also designed to explain why (47) has the meaning that it has (Shibatani 1994:48–51).10 (47) kiho-ka yenghi-eykey ku kes-ul iyaki.hay-cwu-ess-ta Keeho-nom Yonghee-dat the thing-acc tell-give-pst-ind ‘Keeho gave someone (unspecified) the benefit of telling Yonghee the thing.’ 11
The sentence in (47) is grammatical only in the sense of the English translation provided (Shibatani 1994:49). The reason why that is so is claimed to be that the theme of the lexical verb iyaki.hay- does not have to be associated with the goal contributed by the benefactive construction, with the latter not explicitly expressed (Shibatani 1994:48–51). This ensures that the constraint on theme-sharing be complied with. Note that the existence of the unspecified person (or the unexpressed goal) is taken to be implied by the use in (47) of the auxiliary verb cwu-. This can be made clearer if (47) is compared with (48), in which the lexical verb iyaki.hay- alone is used as a ditransitive verb, i.e. with an agent, a goal and a theme. (48) kiho-ka yenghi-eykey ku kes-ul iyaki.hay-ss-ta Keeho-nom Yonghee-dat the thing-acc tell-pst-ind ‘Keeho told Yonghee the thing.’
Moreover, the reason why (47) does not mean that Keeho gave Yonghee the benefit of telling someone (unspecified) the thing is that the theme of the lexical verb iyaki. hay- ‘to tell’ has to be linked with the goal contributed by the benefactive construction,
10.â•… An anonymous referee asks if the unspecified person in (47) could be understood as corresponding to a (missing) taysin-phrase, as in (3). This is true, and the taysin-phrase can also be explicitly expressed (because, in conformity with Shibatani’s constraint on themesharing, it does not compete with the eykey-phrase). However, this is not the point of Shibatani’s discussion (i.e. the use of more than one eykey-marked noun phrase in the -eykey + cwubenefactive construction). 11.â•… The benefit referred to here could be some kind of pleasure that the unspecified person (e.g. a parent) draws from Keeho’s (e.g. a brother) telling Yonghee (e.g. a sister) the thing (e.g. a fable).
Korean benefactive particles 
and also with the goal of the lexical verb itself. This is said to violate the constraint on theme-sharing (Shibatani 1994:51). It needs to be pointed out, however, that in the ‘ungrammatical’ reading of (47) (i.e. Keeho gave Yonghee the benefit of telling someone (unspecified) the thing) the goal contributed by the lexical verb is as much unexpressed as is the goal contributed by the benefactive construction in the ‘grammatical’ reading of (47) (i.e. Keeho gave someone (unspecified) the benefit of telling Yonghee the thing). It is not clear, however, why the constraint on theme-sharing should not apply equally to the ‘ungrammatical’ reading. On the alternative account based on the concept of engaging, the sentence in (47) does not encode the situation in which Keeho gave Yonghee the benefit of telling someone (unspecified) the thing, because in that situation what was said was not intended for Yonghee to hear or engage with in the first place. The third party or the unspecified person is the one for whom the message was intended. There is no engaging with the theme (i.e. what was told) on the part of Yonghee. Therefore, the use of the engagerbenefactive particle eykey would be inappropriate for Yonghee. In fact, the only way to rescue (47) while retaining the meaning ‘Keeho gave Yonghee the benefit of telling someone (unspecified) the thing’ is to use the deputative-benefactive particle taysin or the plain-benefactive particle (l)ulwihay(se) in lieu of the engager-benefactive particle eykey, as in (49) (of course, the latter option without the auxiliary verb cwu-). (49) a.
kiho-ka yenghi-taysin ku kes-ul iyaki.hay-cwu-ess-ta Keeho-nom Yonghee-d.ben the thing-acc tell-give-pst-ind ‘Keeho told someone (unspecified) the thing in Yonghee’s stead.’
b. kiho-ka yenghi-lulwihay(se) ku kes-ul iyaki.hay-ess-ta Keeho-nom Yonghee-p.ben the thing-acc tell-pst-ind ‘Keeho told someone (unspecified) the thing for Yonghee(‘s pleasure).’
In (49), Yonghee is understood not as someone who heard or engaged with what Keeho had to say, but as someone who benefited from Keeho’s telling the story to an unidentified person in her stead (49.a) or as someone who received pleasure from Keeho’s telling the story to an unidentified person (49.b). On the other hand, the sentence in (47) has the ‘grammatical’ meaning that Keeho gave someone (unspecified) the benefit of telling Yonghee the thing. On this reading, the goal of the lexical verb is the only entity that can be reasonably expected to engage with the theme of the lexical verb. If Keeho told Yonghee the thing, the least she would have done is to process what he said to her. Thus the use of the engagerbenefactive particle for Yonghee is compatible with the ‘grammatical’ reading of the sentence in (47). By the same token, (46), with two eykey-marked goals, is ungrammatical, because Yonghee is not the one that Keeho’s story was told to. It was instead intended for the child to hear or to engage with. Thus it makes sense why in (46) the goal nominal yenghi cannot attract the engager-benefactive particle eykey.
 Jae Jung Song
To conclude, what the constraint on theme-sharing is designed to capture can be accounted for if the function of the particle eykey is properly recognized as encoding the goal engaging with the theme. This suggests that the constraint on theme-sharing may not be needed, because it falls out directly from the meanings of the benefactive particles themselves.
5.â•… Conclusion Korean benefactive constructions, with particular reference to the three benefactive particles, eykey, taysin and (l)ulwihay(se), have been investigated with the conclusion that the meaning contribution of these benefactive particles is so specific and significant that their distribution can be explained by reference to their individual meanings. Thus, bringing in the meaning of individual items—to borrow the words of van der Leek (2000: 309)—“through the back door so to speak, in the form of constructional senses, is not […] the right solution” (see Langacker (2005:147–155) on the difficulty of constructional meaning itself; also Van Valin (2007:236)). The right solution, as proposed here, is to give credit where it is due, i.e. the individual benefactive particles. The same position has been defended strongly by Boas (2003). While investigating English and German resultative constructions, he (2003:98) arrives at the conclusion that what is described as constructional meaning (e.g. in Goldberg (1995)) “fall(s) out naturally from the rich lexical-semantic information associated with individual verbs rather than being attributed to distinct but related constructions (emphasis added)”. Therefore, “there is no need for a construction to add information to a verb’s semantics because this information is already contained in the [lexical-semantic description of that individual verb] (emphasis added)” (Boas 2003:110) (also Butler (2009), and Mairal Usón and Ruiz de Mendoza Ibáñez (2009) for their Lexical Constructional Model, which highlights the importance of lexical semantics by integrating it with constructional semantics; but cf. Fried (2004), and Fried and Östman (2004b), who arrive at a more or less the same conclusion within CxG, but choose to deal with them differently). One theoretical conclusion to be drawn from this chapter is that the indexing function of the benefactive particles cannot be understated or ignored (Hopper & Thompson 1980; Wierzbicka 1980,1981,1988). The added advantage of this indexing view is that the constraint on theme-sharing can be shown to be avoidable or unnecessary.
Acknowledgements I would like to express my thanks to Seppo Kittilä, John Newman, John Taylor, Â�Fernando Zúñiga, and two anonymous referees for their constructive comments on
Korean benefactive particles 
earlier, much longer drafts of this chapter, and to Jae-il Kwon for information on the diachrony of the benefactive particles. None of these scholars should be held responsible for the shortcomings and inconsistencies that may remain in the chapter. Lastly, I would like to record my indebtedness to Shibatani (1994, 1996) for inspiring my own work on Korean benefactive constructions, although I have ended up disagreeing more than agreeing with him.
References Boas, Hans C. 2003. A Constructional Approach to Resultatives. Stanford CA: CSLI. Butler, Christopher S. 2009. The Lexical Constructional Model: Genesis, strengths and challenges. In Deconstructing Constructions [Studies in Language Companion 107], Christopher Butler S. & Javier M. Arista, 117–151. Amsterdam: John Benjamins. Butler, Christopher S. & Javier Martín Arista (eds.) 2009. Deconstructing Constructions [Studies in Language Companion Series 107]. Amsterdam: John Benjamins. Chang, suk-jin. 1996. Korean [London Oriental and African Language Library 4]. Amsterdam: John Benjamins. de stadler, Leon G. 1998. The indirect object in Afrikaans. In The Dative, Vol. 1: Descriptive Studies [Case and Grammatical Relations across Language 1], William Van Belle & Willy Van Langendonck, 251–288. Amsterdam: John Benjamins. Fillmore, Charles. 1985. Frames and the semantics of understanding. Quaderni di Semantica 6: 222–253. Fillmore, Charles & Kay, Paul. 1993. Construction grammar. Ms, University of California, Berkeley. Fried, Mirjam. 2004. Predicate semantics and event construal in Czech case marking. In Construction Grammar in a Cross-language Perspective [Constructional Approaches to Language 2], Mirjam Fried & Jan-Ola Östman, 87–119. Amsterdam: John Benjamins. Fried, Mirjam, Östman, Jan-Ola (eds.). 2004a. Construction Grammar in a Cross-Language Â�Perspective [Constructional Approaches to Language 2]. Amsterdam: John Benjamins. Fried, Mirjam & Östman, Jan.-Ola. 2004b. Construction grammar: A thumbnail sketch. In Construction Grammar in a Cross-Language Perspective [Constructional Approaches to Language 2], Mirjam Fried & Jan-Ola Östman, 11–86. Amsterdam: John Benjamins. Goldberg, Adele E. 1995. Constructions: A Construction Grammar Approach to Argument Structure. Chicago IL: The University of Chicago Press. Goldberg, Adele E. 2006. Constructions at Work: The Nature of Generalization in Language. Oxford: OUP. Hopper, Paul & Thompson, Sandra A. 1980. Transitivity in grammar and discourse. Language 56: 251–299. Jackendoff, Ray. 1990. Semantic Structures. Cambridge MA: The MIT Press. Kittilä, Seppo. 2005. Recipient-prominence vs. beneficiary-prominence. Linguistic Typology 9: 269–297. Kittilä, Seppo. 2006. On distinguishing between recipient and beneficiary in Finnish. In Grammar from the Human Perspective: Case, Space, and Person in Finnish [Current Issues in Linguistic Theory 277], Marja-Liisa Helasvuo & Lyle Campbell, 129–152. Amsterdam: John Benjamins.
 Jae Jung Song Langacker, Ronald W. 2005. Constructions grammars: Cognitive, radical, and less so. In Cognitive Linguistics: Internal Dynamics and Interdisciplinary Interaction, Francisco José Ruiz de Mendoza Ibáñez & Mária Sandra Peña Cervel, 101–159. Berlin: Mouton de Gruyter. Mairal Usón, Ricardo & Francisco José Ruiz de Mendoza Ibáñez. 2009. Levels of description and explanation in meaning construction. In Deconstructing Constructions [Studies in Language Companion Series 107], Christopher S. Butler & Javier M. Arista, 153–198. Amsterdam: John Benjamins. Newman, John. 1996. Give: A Cognitive Linguistic Study. Berlin: Mouton de Gruyter. Newman, John. 1998. Recipients and ‘give’ constructions. In The Dative, Vol. 2: Theoretical and Contrastive Studies [Case and Grammatical Relations across Languages 2], Willy Van Langendonck & William Van Belle (eds.), 1–28. Amsterdam: John Benjamins. Newman, John. 2005. Three-place predicates: A cognitive linguistic perspective. Language Sciences 27: 145–163. Östman, Jan-Ola & Fried, Mirjam (eds.). 2005. Construction Grammars: Cognitive Â�Grounding and Theoretical Extensions [Constructional Approaches to Language 3], Amsterdam: John Benjamins. Shibatani, Masayoshi. 1994. Benefactive constructions: A Japanese-Korean comparative perspective. Japanese/Korean Linguistics, Vol. 4, Noriko Akatsuka (ed.), 39–74. Stanford CA: CSLI. Shibatani, Masayoshi. 1996. Applicatives and benefactives: A cognitive account. In Grammatical Constructions: Their Form and Meaning, Masayoshi Shibatani & Sandra A. Thompson, 157–194. Oxford: Clarendon Press. Sohn, Ho-Min. 1994. Korean. London: Routledge. Song, Jae Jung. 2005. The Korean Language: Structure, Use and Context. London: Routledge. van der Leek, Frederike. 2000. Caused-motion and the ‘Bottom-Up’ role of grammar. In Constructions in Cognitive Linguistics [Current Issues in Linguistic Theory 178] Ad Foolen & Frederike van der Leek, 301–331. Amsterdam: John Benjamins. Van Valin, Robert D., Jr. 2007. Review of: Adele E. Goldberg, Constructions at work: The nature of generalization in language. Journal of Linguistics 43: 234–240. Van Valin, Robert D., Jr. & LaPolla, Randy J. 1997. Syntax: Structure, Meaning and Function. Cambridge: CUP. Wierzbicka, Anna. 1980. The Case for Surface Case. Ann Arbor MI: Karoma. Wierzbicka, Anna. 1981. Case marking and human nature. Australian Journal of Linguistics 1: 43–80. Wierzbicka, Anna. 1988. The Semantics of Grammar [Studies in Language Companion Series 18]. Amsterdam: John Benjamins. You, Seok-Hoon. 1997. Argument structure changes in the Korean benefactive construction. Japanese/Korean Linguistics, Vol. 6, Ho-Min Sohn & John Haig, 455–472. Stanford CA: CSLI.
Malefactivity in Japanese Eijiro Tsuboi
University of Tokyo This paper attempts to shed new light on malefactivity in Japanese by examining how the adversative interpretation obtains in the Japanese passive constructions, both as a fully grammaticalized constructional meaning and a context-dependent interpretation. Our analysis reveals the general mechanisms underlying the adversative semantics and concomitant syntactic behavior such as valency increase, as well as elucidating the way they interact with language-particular features characteristic of Japanese. The overall picture gained by our analysis points to the importance of viewing adversative constructions in a broader perspective, as something which emerges from the interaction of various factors in the dynamic processes of contextualization and grammaticalization.
1.â•… Introduction Japanese has a rich inventory of benefactive and malefactive constructions, but probably the best-known construction is its adversative passive,1 which has attracted wide attention of typologists because of its intriguing syntactic and semantic properties.2 As its another name, “indirect passive,” suggests, the passive subject is only indirectly affected by the event denoted by the verb, and even intransitive verbs can form adversative passive sentences: (1) a.
Taro-wa musuko-ni shooshins-are-ta. Taro-top son-dat get.a.promotion-pass3-pst (Lit.) ‘Taro was got promotion by son.’ ‘Taroi’s son got a promotion, much to hisi chagrin.’
1.â•… Since the malefactive construction discussed in this paper is widely known as adversative passive, we will mainly use “adversative” to refer to its malefactive meaning. 2.â•… See Wierzbicka (1979), Siewierska (1984), Shibatani (1985), Keenan (1985), Klaiman (1991), Shibatani (1994), Kim (1994), Shibatani (2000), among others. Shibatani (2000) provides a particularly detailed account of the Japanese passive constructions where various issues such as valency increase in the Japanese adversative passive and the malefactivity of some transitive-based passive sentences are discussed. 3.â•… The passive morpheme has two phonologically-conditioned variants, -rare- and -are-, the former attached to verbs ending with vowels, and the latter to verbs ending with consonants.
 Eijiro Tsuboi
b. Musuko-ga (*Taro-o) shooshins-ita. son-nom (Taro-acc) get.a.promotion-pst ‘(Taro’s) son got (*Taro) a promotion.’
(Tsuboi 2000)
The intransitive verb shooshinsuru ‘get a promotion’ in (1) takes only one argument noun phrase, as can be seen in (1b). It is important to note here that (1a) is not an impersonal passive, which in some languages can be formed based on intransitive verbs, since here the valency increases, rather than decreases, by one, which defies the standard definition of passive as a valency-decreasing construction. (2a) is an example of the Japanese adversative passive based on a transitive verb. As (2b) shows, the passive subject is not an argument of the base verb hiku ‘play a musical instrument’: (2) a.
Taro-wa Hanako-ni piano-o hik-are-ta. Taro-top Hanako-dat piano-acc play-pass-pst ‘(Lit.) Taro was played piano by Hanako.’ ‘Taro was adversely affected by Hanako’s playing the piano.’
b. Hanako-wa (*Taro-o) piano-o hii-ta. Hanako-top (*Taro-acc) piano-acc play-pst ‘Hanako played (*Taro) the piano.’
The semantics of the Japanese adversative passive has also received much attention, since the adversative interpretation is obligatory even when pragmatic considerations suggest otherwise. (1a) cannot but be taken adversatively even when Taro is known to be a doting father who loves his son blindly, and (2a) is also unequivocally adversative even when Hanako is a piano master renowned for her soul-stirring performance. This obligatory adversative interpretation in the Japanese adversative passive is remarkable considering the fact that the passive morpheme is the same as the one for the regular, neutral passive, as can be seen in (3). (3) Taro-wa Hanako-ni home-rare-ta. Taro-top Hanako-dat praise-pass-pst ‘Taro was praised by Hanako.’
The origin of the passive morpheme -(r)are- is unknown, and although it has been hypothesized that it is related to the Old Japanese existential verb ar- ‘be,’ it is unlikely that it originates in some lexical verb with a malefactive meaning like ‘suffer,’ as is the case with adversative constructions in some other languages, so where the adversative sense comes from constitutes an interesting topic for discussion. The Japanese adversative passive is a fully grammaticalized construction which presents the passive subject as adversely affected (in the sense explained below) who is otherwise not actually affected by the action done by the dative-marked agent. However, as we will see below, adversative connotations can sometimes arise even when the passive subject is directly involved in the event denoted by the verb, in apparently unstable, context-sensitive ways. What this paper aims to explore is how adversative
Malefactivity in Japanese 
meanings show up in the Japanese passive, including those cases where adversative interpretations are not necessarily part of the conventionally fixed meanings of the constructions as it is in the adversative passive.
2.â•… Nature of adversity in the Japanese adversative passive What is first to be noted about the Japanese adversative passive when it is compared with adversative or malefactive constructions in other languages is that it does not just mean something malefactive happening to the passive subject, and that malefactivity or adversity does not warrant the use of the construction. As characterized by Davies (1995), Javanese and Indonesian adversatives, exemplified by (4) and (5) respectively, are “unaccusative-based,” and the adversity expressed by the adversative constructions of these languages can be that of spontaneously occurring events or natural phenomena, like a rock falling on a car, but the Japanese adversative passive cannot generally be formed based on unaccusatives, and this can be seen in the unacceptability of examples like (8) and (9), even though they describe adversative situations. (4) Montor-e Amir ke-tiba-n watu. car-def A ad-fall rock ‘Amir’s car got fallen on by a rock.’ (5) *Kertas-e Amir ke-playo.nan bocahbocah. paper-def A ad-run children (Amir’s paper got run on by the children.) (6) Mobil itu kejatuhan batu. car that affect.fall rock ‘The car got fallen on by a rock.’ (7) *Sri kebicaraan pemuda. S affect.speak young.man (Sri endured a young man’s speaking.) (8)
(Davies 1995) ochi-rare-ta.4
*Tokyo-no hitobito-wa kinou kaminari-ni Tokyo-gen people-top yesterday lightning-dat fall-pass-pst (Lit.) ‘People in Tokyo were fallen by lightning yesterday.’ *‘Lightning occurred in Tokyo yesterday, adversely affecting the people.’
4.â•… One of the reviewers pointed out the acceptability of (i) and (ii): (i)
Koumo tabitabi kaminari-ni ochi-rare-tewa kanawanai. this.much often lightning-dat fall-pass-comp more.than.I.can.bear (Lit.) ‘I can’t stand being fallen by lighting this often.’/‘I can’t stand any more lightning.’
 Eijiro Tsuboi
(9) *Taro-wa pan-ni urikire-rare-ta. Taro-top bread-dat be.sold.out-pass-pst (Lit.) ‘Taro was sold out by bread.’ *‘Bread was sold out, adversely affecting Taro.’
(Tsuboi 2000)
The crucial requirement for using the Japanese adversative passive is not just adversity, but rather the perceived existence of a responsible actor who could have avoided doing what he or she did, but failed to, and annoyed the subject. Since the adversative effect and the sense of annoyance felt by the subject would not have arisen if the actor had behaved properly, the actor is held responsible for what happened, and is construed as the “culprit” who did the act of malefaction to the subject. This explains the dative marker on the actor noun phrase, which in the Japanese passive constructions in general marks the agent who does something to the passive subject.5,6 This transitive construal of an
(ii) Konnani hayaku pan-ni urikire-rare-ru-towa omot-tei-nakat-ta. like.this quicklybread-dat be.sold.out-pass-npst- comp think-ipfv-neg-pst (Lit.) ‘I didn’t expect to be sold out by bread so quickly.’/‘I didn’t expect bread to sell out so quickly.’ It is interesting to note that the acceptability of these examples relies on the existence of the second (main) clauses, and that these main clauses explicitly express or strongly imply a sense of annoyance and thus fulfill the requirement for adversative passivization which would be left unsatisfied by the passive clauses alone. 5.â•… There is in fact another kind of passive in Japanese, which is relatively new in the history of Japanese and has developed under a heavy influence of translations from European languages (cf. Kinsui 1997). It differs from the other, native kind of passive in marking the agent noun phrase with a complex expression, -niyotte, instead of the dative marker -ni, and “describes an activity of a human being or human beings from a completely neutral standpoint,” (Kinsui 1997: 776) unlike the ni-marked native type of passive which is generally accompanied with emotional overtones. As can be seen in the dative marking on the agent noun phrase and the emotional, subjective nature of the affectedness expressed, the adversative passive and other passive constructions discussed in this paper all belong to the native type. 6.â•… There is an exceptional weather expression ame/yuki ga furu ‘rain/snow NOM fall’ which takes an inanimate subject but allows adversative passivization: (i)
Taro-wa ame /yuki-ni fur-are-ta Taro-top rain/snow-dat fall-pass-pst
node gaishutsu-deki-nakat-ta because go.out-can-neg-pst
(Lit.) ‘Because Taro was fallen by rain/snow, (he) couldn’t go out.’ ‘It rained/snowed, forcing Taro to stay home.’
(Tsuboi 2000)
Tsuboi (2000) suggests that this has something to do with the durative and dynamic nature of raining and snowing, which is often a constituent element of self-sustained activities. See Tsuboi (2000) for more discussion of weather verbs.
Malefactivity in Japanese 
intransitive event is made possible by a folk theory about event causation like the one in (10), which is a common, everyday logic we often rely on when we say “YOU did it to me!” to accuse someone who has inadvertently caused trouble to us. (10) Folk model of event causation: The person who could have prevented the occurrence of an event which has caused trouble and therefore can be criticized and blamed for his or her failure to do so, is the one who caused it to happen and are regarded as having done the adverse act to the affectee.
Put differently, the Japanese adversative passive involves a transitivity-increasing construal of the event, and what could have been taken to be an intransitive event of someone acting in some way is construed from the viewpoint of the passive subject as a transitive event of himself or herself getting annoyed or inconvenienced by the person acting that way. There is a force-dynamic element here, since the person is perceived to have acted against the wishes of the subject or have upset the social expectations which would have prevented the occurrence of the event had it not been for what was done, and evokes in the passive subject some negative emotional reactions such as accusing feelings or a grudge against the wrongdoer. When this requirement is not met, as with unaccusative predicates, it is inappropriate to use the Japanese adversative passive. Next important point about the Japanese adversative passive is the subjective nature of the adversity expressed in it. As explained above, the Japanese adversative passive means that the subject is annoyed by the dative-marked actor. So the nature of adversity is subjective in the sense that it concerns a subjective feeling of annoyance, but it is even more so since the construal of the actor as someone doing the annoying act to the subject crucially depends on the perception of the subject. For example, (11) is perfectly acceptable if Taro’s father did not choose to die on his own will, nor did he intend to cause any adversative effects on his beloved son. However, he can still be held responsible and blamed for the consequences of his death, since according to the idealized Japanese cultural norms concerning paternal obligations, if a father dies too young before finishing his paternal duties, it counts as an act of negligence to his children. For further details about the nature of adversity in the Japanese adversative passive, see Tsuboi (2000). (11)
Taro-wa chichioya-ni shin-are-ta. Taro-top father-dat die-pass-pst (Lit.) ‘Taro was died by father.’ ‘Taro’s father died on him.’
3.â•… Adversativity and transitivity increase As explained above, the Japanese adversative passive is characterized by a lack of direct affectedness on the passive subject and an adversative interpretation, and Kuno (1983) explains this correlation as a requirement to provide the otherwise uninterpretable
 Eijiro Tsuboi
passive subject with a status of affectee, and the same idea is further developed by Shibatani (1994) and applied to various phenomena involving extra-thematic noun phrases in many languages. As for the reason why the interpretation required to integrate the extra-thematic noun phrase into clausal syntax is not benefactive but adversative, it has been sometimes suggested (e.g. Shibatani 1994) that a benefactive interpretation is blocked by independently existing benefactive constructions, such as -temorau construction. However, this line of argument does not seem to be valid for the Japanese adversative passive, since it fails to explain why adversative effects are not sufficient to make sentences like (8) and (9) acceptable, nor can it account for the reason why sentences like (12) is still adversative (and therefore sounds odd to the extent that a husband is expected to be happy when his wife recovers from illness) despite the fact that the verb naoru ‘recover from illness’ cannot enter into -temorau benefactive construction which is possible only with verbs expressing volitional acts. The blocking account would predict that a benefactive interpretation should not be blocked in (12), which is in fact wrong.7 (12) Taro-wa tsuma-ni naor-are-ta. Taro-top wife-dat recover.from.illness-pass-pst ‘Taro was adversely affected by his wife recovering from her illness.’
When we take a broader perspective and consider the integration of the passive subject into clausal semantics as a case of transitivity increase, we notice that the association of adversativity with transitivity increase is not limited to the Japanese adversative passive. Consider the following English sentence: (13) John dropped the key on his way to the station.
This may seem an ordinary, run-of-the-mill type of transitive sentence, but in fact it is a rather atypical one, in that the subject probably did not do anything to the key and it just dropped. Therefore, in the sense that the subject did nothing, he is external to the
7.â•… It was suggested by a reviewer that the following sentence might be acceptable: (i)
Tsuma-ni hayaku naot-temorat-te, hontooni tasukat-ta. wife-dat quickly recover.from.illness-ben-cvb really be.helpful-pst ‘I was greatly relieved by/benefited from my wife recovering from illness quickly.’
However, it sounds marginal to me, and will be more naturally expressed using a different, -kureru construction, where tsuma in (i) appears as a nominative NP rather than a dative-marked NP. For a critical comment on another familiar account of the adversative interpretation which attributes it to our pessimistic perception that what comes uninvited without our effort is generally adversative, see Tsuboi (2001).
Malefactivity in Japanese 
event of the key dropping and thus could have been left out in linguistic coding of the event. Nonetheless, the subject of a non-intentional transitive sentence like (13) still can be construed as agent and expressed as such, just as the dative-marked agent noun phrase in the Japanese adversative passive is, based on the same logic described in (10): John is regarded as being in a position to exercise control over and take proper care of his belongings and therefore when something that should not happen does happen to his belongings, he is considered to have caused it. Languages may differ as to how far the category of transitive sentences can be extended, and Japanese exhibits a remarkable readiness with which this transitive construal is applied to various kinds of situations. In Japanese, not only events like the one described in (13), where there is no one other than the subject who can be held responsible for the occurrence of the event, but events where it is quite easy to name the actual causer can also be expressed using transitive sentences. See (14) and (15), where the actual causal relation is overridden by the responsibility attributed to the subject in determining who is to be chosen as the ultimate causer of the event. The contrast with the odd literal English translations shows the greater availability of transitive construal based on the notion of responsibility in Japanese. (14)
Taro-wa (yakuza-ni nagur-are-te) ha-o ot-ta. Taro-top (gangster-dat hit-pass-cvb) tooth-acc breakvt-pst (Lit.) ‘Taro (was hit by gangster and) brokevt tooth.’ ‘Taro (got hit by a gangster and) got his tooth/teeth broken.’
(15)
Watashi-wa kuushuu-de ie-o yai-ta. I-top air.raid-loc house-acc burnvt-pst (Lit.) ‘I burntvt house in air raid.’ ‘(Due to my fault,) I had my house burnt down in an air raid.’
(Tsuboi 2000)
This wider applicability of transitive construal is arguably a contributing factor in grammaticalization of the Japanese adversative passive. However, there is a further interesting point about this attribution of responsibility in transitive construal that has not received much attention, which is its curious bias for adversative events. Note that the situation described in (13), (14), and (15) and those described in other similar examples discussed in the literature dealing with this type of nonprototypical transitive sentences are all adversative in nature. According to a common definition of causation, the causer of an event should be the one without whom the event would not have occurred, but since the role played by the unintentional “causer” of an adversative event is no more causal than that played by the unintentional “causer” of a non-adversative event, there should be no reason for the privileged status of adversative events in construability as transitive events. However, as Tsuboi (2001) discusses in detail, this special role of adversativity in increased-transitivity construal
 Eijiro Tsuboi
can be observed in various phenomena in other languages. Although it is just assumed and left implicit when we talk about responsibility, it makes more sense to talk about responsibility for adverse outcomes than for beneficial results. Adversative events are those events that should be prevented from occurring and whose occurrence brings to the fore the existence of someone who is to blame for what happened. This can be seen from the effortlessness with which a transitive construction can be employed to recognize one’s responsibility by saying I killed my son when I had forgotten to meet my son at the station to take him back home, thereby allowing the son to be killed in a bombing by terrorists, whereas unless we are ascertaining the logical causal relation of events in hindsight it would sound unnatural or even funny to say I saved my son’s life when I was told that the station to which I had forgotten to take my son was blown up to pieces by terrorists. Although the notion of responsibility imposed on the individual who is perceived to be in control of and thus causally connected to the occurrence of an event is crucially involved in the transitive construal of an intrinsically intransitive event, adversativity of the event fills the gap in the link from causal connection to transitive construal.
4.â•… Context-dependent adversative sense in the retained-object passive As explained in the previous section, the Japanese adversative passive has as its conventionalized meaning construal of someone who is external to the event as being involved as a victim of annoying act, thereby integrating the person into the clause as the passive subject. Japanese has another interesting passive construction, exemplified by (16), which is structurally no different from (2), since in both sentences the base verb is a transitive verb with its accusative object retained in its original position, and the subject is not an argument of the base verb. The difference lies in that, unlike the accusative noun phrase in (2), the accusative noun phrase in (16) refers to a body part of the passive subject, so the passive subject is directly involved in the event of someone acting on his body part, and in this respect it is different from adversative passive sentences like (1) and (2). Shibatani (1994: 463) states that “[w]hen a body-part is involved, no obvious adversative meaning is detected, while the adversity reading is associated with a non-body part passive form,” and the adversative sense felt in (16) is attributed to the adversative nature of the lexical meaning of the verb. (16) Taro-wa Jiro-ni atama-o nagur-are-ta. Taro-top Jiro-dat head-acc hit-pass-pst ‘Taro was hit on the head by Jiro.’
(Shibatani 1994: 463)
Since the adversative sense is evoked to give an interpretation to the otherwise unintegrated passive subject, it is naturally expected that passive sentences involving body
Malefactivity in Japanese 
parts receive neutral interpretations. However, when we try to find straightforwardly non-adversative, neutral examples, it turns out that such examples are actually not easy to come by. For example, consider (17). Having one’s own hair washed or shampooed by someone else is not necessarily adversative, but (17) is not neutral but adversative in connotation, and would not be used unless we think that Taro was annoyed by Jiro washing his hair, or some other similar situations. This is not an isolated exception that holds only with atama ‘head,’ and (18) and (19) are similar examples involving ashi ‘leg,’ and kata ‘shoulder’ respectively: (17) Taro-wa Jiro-ni atama-o araw-are-ta. Taro-top Jiro-dat head-acc wash-pass-pst ‘Taro was adversely affected by Jiro washing his hair.’ (18) Taro-wa Jiro-ni ashi-o sasur-are- ta. Taro-top Jiro-dat leg-acc stroke-pass-pst ‘Taro was adversely affected by Jiro stroking his leg.’ (19) Taro-wa mago-ni kata-o mom-are-ta. Taro-top grandchild-dat shoulder-acc massage-pass-pst ‘Taro was adversely affected by his grandson massaging his shoulders.’
These sentences seem to suggest that even when the passive subject is directly involved in the event denoted by the verb, adversative interpretations may obtain in some cases. However, the situation is further complicated by the fact that in some circumstances, the adversative sense seems to disappear. In other words, unlike (1) and (2), which are either intransitive-based or transitive-based with a non-body-part noun and do not show context dependence and are always adversative, transitive-based passive sentences with body part nouns may or may not have adversative connotations depending on the context. This can be seen in the contrast between (17), (18), and (19) on the one hand and (20), (21), and (22) on the other. The latter sentences are different from the former in that the verbs are in the imperfective. When compared with the former sentences, it is easier to get neutral readings for the latter sentences, which can be taken to describe what a reporter might have seen without implying adversative effects on the passive subjects. (20) Watashi-ga mi-ta tokiniwa, Taro-wa Jiro-ni atama-o I-nom see-pst when Taro-top Jiro-dat head-acc araw-are-tei-mashi-ta. wash-pass-ipfv-pol-pst ‘When I saw Taro, Jiro was washing his hair.’ (21) Watashi-ga mi-ta tokiniwa, Taro-wa Jiro-ni ashi-o I-nom see-pst when Taro-top Jiro-dat leg-acc sasur-are-tei-mashi-ta. stroke-pass-ipfv-pol-pst ‘When I saw Taro, Jiro was stroking his leg.’
 Eijiro Tsuboi
(22) Watashi-ga mi-ta tokiniwa, Taro-wa mago-ni I-nom see-pst when Taro-top grandchild-dat kata-o mom-are-tei-mashi-ta. shoulder-acc massage-pass-ipfv-pol-pst ‘When I saw Taro, his grandchild was massaging his shoulders.’
These examples may make us wonder whether this type of passive should be regarded as adversative passive, but in fact it is a wrong question which presupposes that it is either adversative or not. The construction in question is better understood to be left unspecified in that respect, and the eventual interpretation as an adversative or a neutral sentence is what the hearer or the reader reads into the sentence. (23) and (24) help to clarify this point. (23) Taro-wa Jiro-ni yokuyattato kata-o tatak-are-ta. Taro-top Jiro-dat as.if.to.say. ‘Good job!’ shoulder-acc pat-pass-pst ‘Jiro patted Taro on his back to praise him for his good job.’ (24) Taro-wa namae-o yob-are-ta. Taro-top name-acc call-pass-pst ‘Taro was called.’
(23) has kata ‘shoulder,’ and the verb is not in the imperfective, but this sentence is not adversative in meaning. The key to understand the reason for non-adversative meaning here is the fact that the real target of the action here is not the body part, the shoulder itself, but its possessor, the passive subject. What the dative agent did was to praise him for what he had done, and the patting on his shoulder is done because it is a customary way of praising a person for his or her good job. In this sense, the target of the action here is not Taro’s shoulder per se, but Taro himself. (24) is a similar example, whose accusative noun phrase, namae ‘name,’ belongs to the broader class of inalienably possessed things, and when Taro’s name is called, Taro himself is called. What is more interesting in this connection is the examples in (25) and (26), both of which are taken from Kuno (1983), with slight changes. In these examples, since the accusative noun phrases refer to another person or a thing distinct from the passive subjects, the acts done to the objects do not involve the passive subjects in the usual sense of the term, but these do not evoke adversative readings. Kuno (1983) makes a perceptive observation about these examples, and points out that the typical situation for (25) is one where praising a child can be taken to be an act of praising the parent, in this case the father of the child, for his paternal guidance and the education given to the child which make his son a person worthy of praise. Example (26) is most appropriate when the person who has received the souvenir expresses his or her appreciation directly to the passive subject.
Malefactivity in Japanese 
(25) Taro-wa kyooshi-ni musuko-o home-rare-ta. Taro-top teacher-dat son-acc praise-pass-pst ‘The teacher praised Taro’s son (and the education and parental guidance Taro had given to his son, making Taro feel proud and happy).’ (26) Watashi-wa miyage-o taisoo yorokob-are-ta. I-top souvenir-acc very.much like-pass-pst ‘They liked my souvenir very much (and thanked me).’
(Kuno 1983)
These examples show that if the passive subject is the actual target of the action, the retained-object passive can be semantically neutral even when the retained object is not a body part noun. When we turn back to (17), (18), and (19), we realize that they differ from the examples we have just seen, most clearly (25) and (26), in that the target of the action expressed in each sentence is the body part, not its possessor. If it is the case that when the subject is the actual target of the action, the retained-object passive is semantically neutral, then, this leads us to a kind of contrapositive inference that when a retained-object passive sentence is adversative in meaning, the subject is actually not directly affected, even when the accusative noun phrase is a body part expression. This means that direct involvement in the event does not necessarily lead to a neutral interpretation, and even when the possessor of a body part is factually involved in the action done to it, the possessor may not be presented as directly affected in the same way that the subject of an ordinary, neutral passive sentence is, and in this sense, we may say that even body parts can be treated like alienably possessed items in the Japanese retained-object passive. But this in turn leaves the neutral readings in (20), (21), and (22) unaccounted for, and in order to understand the reason for their neutrality, we need to look at the ambiguous status of body parts. In one sense body parts are integral parts of us, but in another sense, they are our possessions, things belonging to our personal sphere, which we manipulate to do various things, like using our hands to pick up a book from a desk. Therefore, when seen through the eyes of a detached, outside observer, an action done to a body part of a person can be seen as an action done to the person, based on the physical unity of the body part and the person, but from the perspective of the person in question, the ego is separate from the body as the locus of consciousness, and monitors, as it were, what happens to his body. From this perspective, body parts are grouped together with other things in his personal sphere, conceptually distinct from the ego. So, if body parts as expressed in retained-object passive sentences can be considered to be more or less on a par with regular alienable possessions, then it is no wonder that sentences like (17), (18), and (19) are adversative, for the same reason that adversative passive sentences like (2) are adversative in meaning. Thus, an adversative reading and a neutral reading are both potentially possible, and the question to be asked now is, what determines the preferred reading in each sentence. In order to answer this question, we need to think about the function of the
 Eijiro Tsuboi
Japanese passive. Although much more detail is obviously necessary to fully characterize it, we can safely say that it is a marked construction employed to put the affectee in focus and describe what happens to the affectee as a result of the action denoted by the verb phrase. This inherent affectee-orientedness is further reinforced when the event is perfective, as in cases like (17), (18), and (19) which have a past tense marker -ta. As DeLancey (1981) points out, whereas the imperfective aspect is employed to present the event as one which is to be completed by an actor, and therefore A-oriented, the perfective aspect serves to focus on the result of the process which is usually realized in the patient, and thus P-oriented. Thus, it is quite natural for the result or change to be foregrounded in a perfective passive sentence, thereby leading the hearer or the reader to incorporate the viewpoint of the passive subject, from whose perspective an action to the body part is not an action to himself but to his possession, resulting in an adversative interpretation. However, when the verb is in the imperfective, what is highlighted is not the result, which is yet to be gained since the event is not completed, but the ongoing process of something being done to a body part of the passive subject, which can be observed and described solely from an external viewpoint, free from the need to incorporate the passive subject’s feelings. As we have seen, when the event is described from an external viewpoint, the passive subject and the body part are not particularly conceptually distinguished, and this leads to the interpretation that the passive subject is directly involved and affected in the event of something being done to his body part. Since the passive subject is directly affected, there is no need to evoke an adversative reading, and sentences like (20), (21), and (22) can receive neutral interpretations.
5.â•…Context-dependent adversative sense and contextually understood affectee In the previous section, we saw how adversative sense appears or disappears in retained-object passive sentences depending on whether the subject is construed as directly affected or not, but even in those cases in which the passive subject is unambiguously the direct target of the action, adversative interpretations can sometimes arise in still another type of Japanese passive, exemplified by (27) and (28).8
8.â•… This type of passive with an unexpressed affectee was first pointed out by Masuoka (1991). If the dative agent marker -ni is replaced with -niyotte, these sentences are acceptable as neutral passive sentences; see Footnote 3. The subject of this type of passive often takes the nominative marker -ga, instead of the topic marker -wa, probably because an unexpressed affectee can most easily be taken to be the overall (though hidden) subject of predication when there is no other candidate competing for a topic status.
Malefactivity in Japanese 
(27) Tsumetakunatteita ryouri-ga/-wa tsuma-ni atatamenaos-are-ta. gone.cold dish-nom/-top wife-dat reheat-pass-pst (Lit.) ‘The dish which had gone cold was reheated by wife.’ ‘I (or possibly someone else) was annoyed by my (or someone else’s) wife reheating the dish which had gone cold.’ (28) Ie-no tobira-ga/-wa booto-ni nandomo ker-are-ta. house-gen door-nom/-top rioters-dat repeatedly kick-pass-pst (Lit.) ‘The door of the house was kicked by rioters repeatedly.’ ‘Someone (associated somehow with the door, possibly the owner of the house) was annoyed by the rioters kicking the door.’
There is one important difference between these sentences and those we have seen so far, and in this type of passive, it is not the subject but some contextually understood person who is adversely affected. Since the passive subject is the direct target of the action, these sentences should have neutral interpretations just as their literal English translations do, but actually the adversative interpretations prevail, with varying degrees of acceptability depending on how natural such an adversative construal is.9 It is understandable that such implicit reference to a concealed affectee is evoked to make the sentence conform to the general pattern of the Japanese passive which commonly describes how a human patient is affected, and the adversative interpretation is also expected since the unexpressed person is external to the event (which is why it is not expressed) and can only be construed as being involved in the event as an adversely affected person. But what is worth noting here is the fact that when (27) and (28) are turned into converbal subordinate clauses as in (29) and (30), they no longer need to receive adversative interpretations. Since the understood affectee is not directly involved in the event, cancellation of adversative connotation here needs a different account from the one given to the neutral readings in (20), (21), and (22). (29) Tsumetakunatteita ryouri-wa tsuma-ni atatamenaos-are-te yugeotate-tei-ta. gone.cold dish-top wife-dat reheat-pass-cvb steaming-ipfv-pst ‘The dish which had gone cold, having been reheated by my wife, was steaming again.’ (30) Sono ie-no tobira-wa bouto-ni nandomo ker-are-te the house-gen door-top rioters-dat repeatedly kick-pass-cvb hekon-dei-ta. get.dented-ipfv-pst ‘The door of the house, having been kicked by the rioters repeatedly, was dented.’
9.â•… This does not change if the subject is a continuing topic in the current discourse, which suggests the inadequacy of the approach which seeks to provide a universal characterization of voice constructions in terms of a semantic map defined by relative salience or topicality of A and P, wherein passive is characterized as a construction with more topical P and less topical A (cf. Croft 2001). For a critical discussion of the semantic map approach in general and its application to voice phenomena in particular, see Tsuboi (2008).
 Eijiro Tsuboi
Japanese has a fairly strong animacy restriction in describing two-participant events, both in the active and the passive, unlike some languages like English where inanimate things or abstract entities can relatively easily assume the subject status in transitive sentences if they are the causes of the events, as in the derailment killed more than one hundred commuters. Japanese strongly resists such constructions and prefers to choose a human being as subject, and describes the event in a way comparable to English more than one hundred commuters died because of the derailment, with the focus on the human participant. We may say that the readily conferrable causer status of a responsible actor in Japanese transitive sentences we saw in Section 3 as contrasted with the peripheral role given to inanimate “causers” is another facet of the central role played by sentient human beings in linguistic coding of transitive events in Japanese.10 As for the passive, Japanese also exhibits a strong preference to a human subject, and until the time when Japanese developed the ni-yotte-marked passive mentioned in Footnote 3, inanimate passive sentences were generally avoided, and were only possible when they had no dative-marked agent noun phrases and took verbs with resultantstate suffixes (cf. Kinsui 1991), which we take to be necessitated to background the otherwise salient human agents. Now, returning to the contrast between (27) and (28) on the one hand and (29) and (30) on the other, we notice that the passive clauses of the former sentences have high values with respect to several of Hopper & Thompson’s (1980) transitivity parameters: they are perfective and realis, and are the main, foregrounded clauses. Since the adversative sense in the Japanese adversative passive is based on a force-dynamic, highly transitive construal of the event, we may think that it is relatively easily evoked in high transitivity contexts, like the ones in (27) and (28). But when the passive clause is changed into a converbal clause, the clause-final tenseaspect-modality morphemes are stripped away, making it like an English gerund, and the assertive force of the clause is gone as well, significantly lowering the transitivity of the clause. When the passive clause is in a low transitivity context, a force-dynamic adversative interpretation need not be evoked, which is why (29) and (30) are neutral in connotation. This way of thinking about the fluctuations in interpretation seems to be supported by the fact that other low transitivity factors such as indefiniteness of time reference and a lesser degree of individuation of the participants can also serve the function of suppressing adversative interpretations where they might otherwise be expected. Thus, while (31) describes a perfective event involving specific participants and can be accepted only as a sentence implying that someone was somehow adversely affected or maybe chagrined by Poe’s writing of the first detective story, (32) only states the nonexistence of a certain type of novel, with an indefinite agent, and is not adversative
10.â•… For this point, see Ikegami (1994).
Malefactivity in Japanese 
in meaning. Similarly, (33) is not adversative either, and the agent is generic in reference, and it is not about a specific event and thus indefinite in temporal reference. (31)
Saishono suirishousetu-wa Edgar.Allan.Poe-ni kak-are-ta. first detective.story-top Edgar.Allan.Poe-dat write-pass-pst (Lit.) ‘The first detective story was written by Edgar Allan Poe.’ ‘I (or someone else) was adversely affected by Poe writing the first detective story.’
(32) Sono shuno shousetu-wa koremade darenimo kak-are-tei-nai. the kind novel-top so.far by.whoever write-pass-ipfv-neg ‘The kind of novel has never been written by anybody.’ (33) Sono zasshi-wa wakamono-ni yoku yom-are-tei-ru. the magazine-top young.people-dat often read-pass-ipfv-prs ‘The magazine is widely read among young people.’ (Tsuboi 2002)
6.â•… Conclusion The Japanese adversative passive has been grouped together with comparable constructions in other East Asian languages which are used to describe adversative events, but as we saw in Section 2, even the very notion of adversity varies among languages and the mere reference to their shared feature of adversity does not reveal much about them. What we need to do to tap their potential for providing a promising subject for areal study of East Asian languages is to make finer-grained analyses of individual constructions to clarify the role played by language-particular features and reveal the general mechanisms underlying their adversative semantics and concomitant syntactic behavior such as valency increase. This paper attempts to shed new light on malefactivity in Japanese by examining how the adversative interpretation obtains in the Japanese passive constructions, both as a fully grammaticalized constructional meaning and a context-dependent interpretation, as well as what is involved in the valency increase characteristic of the Japanese adversative passive. Whereas in the adversative passive the grammaticalization of adversative meaning may have been motivated in part by the requirement of adversity for the integration of an event-external affectee, in the other types of passive constructions with adversative connotations, the passive subject is the target of the action and an adversative interpretation is not particularly expected, but is imposed nonetheless. This context-dependent meaning augmentation by the speech act participants, which is often a precursor to a grammaticalization process, is related to the “coolness” (in the sense of Huang 1984 cited in Bisang 1998) of Japanese which tends to expect or require active audience participation, and is encouraged by the requirement of a human affectee for a Japanese passive sentence. Also relevant in this connection is the
 Eijiro Tsuboi
nonfinite subordination pattern (cf. Haspelmath & König 1998) of Japanese in cases like (29) and (30), which makes it possible for otherwise adversatively interpreted clauses to be used non-adversatively for the purpose of clause linking. Taken together, what we have discussed in this paper points to the importance of viewing adversative constructions in a broader perspective, as something which emerge from the interaction of various factors in the dynamic processes of contextualization and grammaticalization, but further research needs to be conducted on adversative constructions in other languages in order to lay down a solid basis for building a more detailed and clearer picture of the general category of adversative constructions.
References Bisang, Walter. 1998. Adverbiality: The view from the Far East. In Adverbial Constructions in the Languages of Europe, Johan van der Auwera (ed.), 641–812. Berlin: Mouton de Gruyter. Croft, William. 2001. Radical Construction Grammar. Oxford: OUP. Davies, William D. 1995. Javanese adversatives, passives and Mapping Theory. Journal of Linguistics 31: 15–51. DeLancey, Scott. 1981. An interpretation of split ergativity and related phenomena. Language 57: 626–57. Haspelmath, Martin & König, Ekkehard. 1998. Concessive conditionals in the languages of Europe. In Adverbial Constructions in the Languages of Europe, Johan van der Auwera (ed.), 563–640. Berlin: Mouton de Gruyter. Hopper, Paul & Thompson, Sandra A. 1980. Transitivity in grammar and discourse. Language 56: 251–299. Huang, C.T. James. 1984. On the distribution and reference of empty pronouns. Linguistic Inquiry 15: 531–574. Ikegami, Yoshihiko. 1994. The agent and the sentient: A dissymmetry in linguistic and cultural encoding. In Origins of Semiosis, Winifried Nöth (ed.), 325–337. Berlin: Mouton de Gruyter. Keenan, Edward L. 1985. Passive in the world’s languages. In Language Typology and Syntactic Description, Vol. 1: Clause Structure, Timothy Shopen (ed.), 243–281. Cambridge: CUP. Kim, Kyunghwan. 1994. Adversity and retained object passive constructions. Japanese/Korean Linguistics 4: 331–346. Kinsui, Satoshi. 1991. Judoo bun no rekishi ni tuite no ichi koosatsu. Kokugogaku 164: 1–14. Kinsui, Satoshi. 1997. The influence of translation on the historical development of the Japanese passive construction. Journal of Pragmatics 28: 759–779. Klaiman, M.H. 1991. Grammatical Voice. Cambridge: CUP. Kuno, Susumu. 1983. Shin Nihon Bunpoo Kenkyuu. Tokyo: Taishukan. Masuoka, Takashi 1991. Judoo hyoogen to shukansei. In Nihongo no Boisu to Tadoosei, Yoshio Nitta (ed.), 105–121. Tokyo: Kurosio. Shibatani, Masayoshi. 1985. Passives and related constructions: A prototype analysis. Language 61: 821–848. Shibatani, Masayoshi. 1994. An integrational approach to possessor raising, ethical datives, and adversative passives. BLS 20: 461–486.
Malefactivity in Japanese 
Shibatani, Masayoshi. 2000. Boisu. In Nihongo no Bunpoo, I: Bun no Kokkak, Yoshio Nitta, Shinjiro Muraki, Masayoshi Shibatani & Masato Yazawa, 117–186. Tokyo: Iwanami Shoten. Siewierska, Anna. 1984. The Passive: A Comparative Linguistic Analysis. London: Routledge Kegan & Paul. Tsuboi, Eijiro. 2000. Cognitive models in transitive construal in the Japanese adversative passive. In Constructions in Cognitive Linguistics, Ad Foolen & Frederike van der Leek (eds.), 283–300. Amsterdam: John Benjamins. Tsuboi, Eijiro. 2001. Asymmetry in event construal: Adversativity in epistemic assessment and force dynamics. The Proceedings of the First Seoul International Conference on Discourse and Cognitive Linguistics: Perspectives for the 21st Century, 877–889. Seoul: Discourse and Cognitive Linguistic Society of Korea. Tsuboi, Eijiro. 2002. Jueisei to ukemi. In Ninchigengogaku, I: Jishoo Koozoo, Y. Nishimura (ed.), 63–86. Tokyo: University of Tokyo Press. Tsuboi, Eijiro. 2008. Semantic maps and grammatical imagery: Universal and language-specific aspects of grammatical meanings. In From Gram to Mind : Grammar as Cognition, Jean-Remi Lapaire, Guillaume Desagulier & Jean-Baptiste Guignard (eds.), 133–145. Bordeaux: Presses Universitaires de Bordeaux. Wierzbicka, Anna. 1979. Are grammatical categories vague or polysemous? The Japanese ‘adversative’ passive in a typological context. Papers in Linguistics 12: 111–162.
Index
A Actor╇ 59, 205, 207, 225, 248, 378, 422–423, 430, 432 Addressee╇ 60, 221, 227, 324, 326, 355–356, 363, 370, 374 Adjunct╇ 30, 36, 228, 285, 287, 333, 399 Adposition╇ 4–9, 34–35, 38, 50–51, 63, 131, 134, 137, 141, 187, 247 Adversative╇ 5, 90, 99, 101, 113–114, 149, 172–173, 419–434 Affectee╇ 5, 21, 71, 88, 98–99, 102, 115, 216–217, 356, 361–364, 366, 370–371, 385, 390–391, 423–424, 430–431, 433 Affectedness╇ 4, 6, 15–16, 20–22, 25, 32, 99–101, 210, 331, 341–342, 362, 366, 390 Affectedness construction╇ 73, 98–99, 101, 105–106, 114–115 Agent/causer animate╇ 2, 7, 207, 260, 263, 266, 269, 333, 339, 347, 361, 389 inanimate╇ 7, 16, 85, 256, 260–262, 268–269, 333, 342, 347, 358, 361, 380, 422, 432 involuntary╇ 22, 187, 262 Agentive benefactive╇ 75–78, 81, 83–84, 86, 88–89, 91–92 Agreement╇ 26, 149–150, 186, 189, 195–196, 199, 273, 275, 278, 284, 290, 343–344, 351 object agreement╇ 26, 149, 344 Alignment patterns╇ 185–186
Ambiguity╇ 328, 334–335, 337, 388, 401 Ambitransitive verbs╇ 205, 213–215 Animacy of the beneficiary╇ 4, 6, 15–16, 255, 257, 262–264, 266, 268–269, 333, 339 Animacy restriction╇ 432 Applicative╇ 5, 10, 13, 21, 26, 29–37, 40–42, 46–48, 53–57, 60, 62–63, 98, 110, 116, 147–152, 154–179, 186–194, 196–197, 199, 203, 205–206, 208, 210–217, 271–272, 275–285, 287–291, 312, 326, 339, 341, 345–347 redirective╇ 147–154, 156–165, 168–169, 171, 174–175, 177–179 relational╇ 147–149, 165–174, 176, 178–179, 288, 290, 333–335 Areal distribution╇ 97, 116 Auxiliary╇ 40, 42–43, 45, 50, 54, 58, 61, 74–75, 77, 79–81, 88, 100, 115, 140, 148, 218, 394, 398–403, 405, 408, 411–415 Autobenefactive╇ 4, 26, 29, 33, 57–60, 63, 218, 331, 341–342, 346–348, 358
285, 289–290, 295–313, 317–319, 321–323, 327–333, 339–341, 344–348, 357–358, 362, 377–381, 383–385, 387–409, 411–417, 419, 424 direct╇ 2, 90, 255, 267, 376–378, 385, 387, 389–390, 392, 400, 412 double benefactive╇ 314 indirect╇ 2, 21, 137, 171, 206, 209, 361, 377, 379–380, 388–390, 392 Benefactive particle╇ 25, 53, 60, 92, 393–395, 398–406, 408–409, 411–413, 415–417 Benefactive periphrasis╇ 29, 31–33, 42, 44, 47, 49, 57–59, 62–63, 115 Benefactive-recipient╇ 14, 130, 132–133, 135, 222–223, 233–235, 237, 239, 245, 248–249, 255, 307, 378, 384, 396–398, 400, 402–403, 405–406, 408–409 Biclausal constructions╇ 26, 34, 47, 330 Body parts╇ 246, 267, 297, 347, 366–367, 426–430
B Benefactive╇ 2–5, 7, 14, 18, 26–27, 29–35, 37–48, 50–61, 63, 71–93, 97–116, 121–124, 126–129, 132–139, 141–142, 147–150, 152–156, 158–165, 167–170, 174–179, 191–194, 196–197, 199, 203, 209, 216–217, 219, 221–224, 226–237, 239–240, 245–248, 257, 259–261, 263–265, 268, 271–272, 281,
C Case╇ 1, 4–5, 7–8, 17–18, 20, 24–25, 62, 98, 104, 115, 124–125, 129, 137–139, 141, 185–186, 189, 240, 246–247, 257–259, 264, 266–268, 274–275, 277–279, 286–287, 352 ablative╇ 5, 7, 20, 115–116, 245–247, 251, 255, 257–258, 261, 264–268, 286–287, 361
Index absolutive╇ 274–275, 279, 285, 287 accusative╇ 5, 25, 97, 219–220, 227–228, 274–275, 299, 426, 428–429 allative╇ 5, 7, 13, 17, 24, 115, 121, 124–135, 191, 245–247, 249–258, 261, 264–269, 287 dative╇ 7, 12, 18, 20–27, 62, 102–105, 121, 124–126, 129–130, 138–139, 219–221, 227–229, 235, 277–279, 282–285, 351–374, 398, 420, 422–425 dative of interest╇ 21 ergative╇ 83, 165, 274–275, 287 genitive╇ 104, 247, 283–285, 287, 411 instrumental╇ 41, 174, 279, 287, 317–319, 323, 329, 364 locative╇ 129, 132, 138, 191, 317–319, 335, 364, 370 nominative╇ 257–258, 352, 430 Case marking╇ 4–5, 62, 185–186, 189, 247, 274, 277, 320, 347 Causation╇ 22–23, 25, 221, 224, 238, 361, 423, 425 direct╇ 22–23, 25, 224, 238 indirect╇ 22–23, 25, 361 Causee╇ 18, 22, 24–25, 178, 205, 217 Clitic doubling╇ 353, 367 Complex event╇ 42, 46–47 Complex predicate╇ 34–35, 37, 40, 48 Compound╇ 40, 42, 53–56, 63, 335, 377 Consciousness of the beneficiary╇ 15–17, 250, 255, 257, 263–266, 389–390, 392 Converb╇ 40, 42, 44–45, 48, 53–54, 62, 327, 329, 431–432 Coordination╇ 38, 46, 186, 197–198 Core argument╇ 151, 173, 204, 227, 278, 290, 324, 327 Cross-reference╇ 274
D Deputative-benefactive╇ 14, 31, 72, 192, 217, 247Â�–248, 250–253, 258–259, 264–265, 282, 285, 307, 322–323, 380–381, 383–385, 391–392, 396–403, 408–409, 412–413, 415 Disambiguation╇ 21, 62, 287, 327, 329–330, 383, 389 Ditransitive╇ 1, 25, 74–75, 78, 92, 133, 148, 150, 178, 185–186, 190, 194, 196, 198–199, 205, 210, 220–221, 223, 227–229, 235, 257, 280, 282, 299–300, 324, 334, 381–382, 414 clauses╇ 1, 25, 74–75, 92, 133, 195–196, 198–199, 220–221, 227–229, 235, 257, 353, 381 derived╇ 42–43, 53, 56, 148, 150, 185–186, 199, 277, 295, 300, 305, 307, 346, 357 underived╇ 4, 205, 210, 280, 282, 337, 381 Double-object construction╇ 26, 47, 220–228, 230–237, 239–240, 338, 381–382 E Engager-benefactive╇ 11, 15, 248, 252, 378, 409, 411–415 Event benefactive╇ 75–76, 85–92 Experiencer╇ 21–22, 35–36, 90, 166, 283–284, 321, 341, 363–364, 382–384, 387–389, 391 G Give╇ 6, 9, 11, 17–18, 31–60, 63, 72, 77–81, 85–89, 99–101, 106–116, 133, 152, 157–161, 188–192, 195–196, 198–199, 205, 215–216, 219–220, 223–224, 234, 280–282, 286, 378–379, 381–391, 394, 398–405, 411–415
Goal╇ 23–24, 35, 126–127, 131–136, 138–139, 141–142, 165, 176, 191, 250, 386, 405–412, 414–416 H Head marking╇ 186 I Intransitive clause╇ 13, 155, 165, 169–171, 178–179, 187, 205, 210–215, 257–258, 276–278, 300–301, 334–337, 367, 402–403, 405–406, 408, 412 extended╇ 187 Inverse clause╇ 205–206 L Language change╇ 46 M Malefactive (also maleficiary)╇ 5, 11–13, 20, 32–33, 98–116, 147–151, 154–158, 170–171, 178–179, 216–217, 258, 281–283, 289–290, 295–302, 304–306, 308–313, 318, 322–323, 325, 360Â�–362, 385, 419–421 Metaphor╇ 116, 126, 131–132, 136, 223, 226, 337–338, 362–363 Metonymy╇ 135–137, 141–142, 406–408 Motion verbs╇ 128, 131, 140–141, 167–168, 187, 206, 310 O Object╇ 25, 44, 79, 86, 129, 133, 147–150, 152, 156–176, 178–179, 199, 209–210, 220–240, 274, 279–280, 283–284, 299–300, 317–320, 324–330, 333–339, 341–342, 344, 346–347, 363, 366–368, 378–382, 384, 386, 390, 405, 426, 429–430 animate╇ 23, 257, 280, 283, 325, 327, 333, 338–339, 347, 373
applied╇ 149–150, 152, 156–176, 178–179, 195, 283–285, 287 definite╇ 205, 325–329 direct╇ 86, 150, 152, 165, 220, 222, 230, 325, 333–334, 342, 346, 366–368, 387, 405 inanimate╇ 207, 280, 333–334, 336, 380 indirect╇ 44, 133, 209–210, 221–222, 226, 228, 237–239, 379, 382, 384, 387, 405 primary╇ 79, 199, 206, 317–319, 324–330 Oblique╇ 4, 148, 150, 166, 175, 198, 278–279, 284–285, 290, 383, 387 P Passive╇ 31–32, 74–75, 89–90, 100–101, 112–114, 172–174, 179, 210, 212, 233, 280, 298, 302, 419–433 Patient╇ 2, 4–5, 22, 25, 30, 32, 59, 62, 75, 195–196, 209, 213, 221–222, 233, 299, 332–334, 430–431 Periphrastic constructions╇ 29–37, 39–40, 42, 44, 47–48, 53, 56, 58–60, 62–63, 77, 99, 174–175, 369 Plain beneficiary (also plain benefactive)╇ 14, 31, 34, 247–250, 257–258, 264–268, 281–282, 285, 322–323, 391–392, 396–400, 403, 408, 413, 415 Polysemy╇ 5–7, 18–25, 126, 130–132, 137, 246, 267–268, 282–283, 290, 306, 368 Possession╇ 14, 19–20, 49, 157–165, 175, 186, 192, 197, 209, 219–221, 223–224, 253–255, 283–284, 363, 365–368, 395, 397, 405–411, 428–430 alienable╇ 186, 192, 429 external╇ 19–20, 129, 157–158, 283, 365–368 inalienable╇ 186, 407, 411
Index predicative╇ 365 Possessive control╇ 49, 393, 395, 398, 405–410 Purpose╇ 36–37, 121–132, 134–142, 149, 252, 254–255, 260–261, 263, 286, 290, 370, 379, 389 Purposive╇ 121–124, 126–142, 379–380, 382–383, 386 R Reason╇ 6–7, 16, 22–23, 36, 251, 254, 269, 389, 428–429 Reception╇ 2, 6, 19, 133–134, 188–190, 194, 219–226, 238–240, 252, 254–255, 268, 357, 397 Recipient╇ 4, 6, 13–16, 18–19, 35–36, 48, 50, 52, 74–75, 115, 132–135, 152, 188–192, 194–199, 219–223, 226–228, 230–233, 248, 254–255, 265, 267, 280–283, 304–305, 324, 326–327, 346–347, 355–365, 370–371, 378–379, 382–384, 389–390, 395–397 Recipient-beneficiary (see also benefactive-recipient)╇ 13–16, 132–133, 188–192, 194, 199, 222, 225, 232–235, 238–240, 248–250, 252–255, 257–258, 267, 285, 288, 305, 307, 358–361, 370–371, 396–400, 402–403, 405–406, 408–409 Reconstruction╇ 42, 151, 165–166, 203, 285, 312 Reflexive╇ 57, 78, 85, 259, 302, 309–310, 331, 339–341, 344, 347–348, 390 Resultative╇ 336, 339, 342, 416 S Self-benefactive╇ 4, 57, 76, 78–89, 91, 106–108 Serial verb construction╇ 7–10, 37–38, 40–42, 50, 63, 109–110, 116, 175, 197, 380–387
Shared-benefit construction╇ 71, 83–84, 91 Source╇ 35, 47, 50, 152, 156–159, 162–164, 171, 174, 176, 178–179, 258, 268, 303, 305–306, 361 Specific benefactive marker╇ 7, 10–11, 31, 72, 114, 159, 259, 266, 268, 321, 355–356, Subject╇ 31–32, 57, 75, 80, 83–86, 91, 135, 150, 170–171, 209, 222, 240, 261, 273–275, 288–289, 317–320, 327, 332–333, 340–343, 352, 358, 360–361, 379–383, 385, 405, 419–433 Subordination╇ 126, 140, 286, 370, 379–380, 382–383, 431, 434 T Take╇ 26, 29, 33, 35, 41, 57–59, 63, 79, 81–82, 106–108, 156–157, 179, 198, 216, 258, 282, 284–285, 289, 302, 305, 307, 337, 361, 364, 366, 371, 382, 396 Theme╇ 3, 11, 14–15, 149–150, 157–159, 171, 175–176, 190, 194, 198–199, 207, 219–220, 280–281, 382, 384, 386, 389–390, 395, 405–412, 414–416 Three-argument constructions╇ 195, 198–199, 234 Transfer of possession╇ 49, 175, 219–221, 224, 233, 254–255, 261, 365, 386 Transitive verb╇ 25, 32, 49, 148, 150, 172–173, 176, 187, 190–193, 199, 207, 210–215, 275–279, 299–303, 333, 336, 339, 368–369, 420, 426 Transitivity╇ 13, 25, 187, 206, 211, 228, 245, 250, 257, 259, 300, 391, 404, 423–424, 432 Transitivity increase╇ 150, 206, 213, 300, 423–424
Index U Undergoer╇ 3, 52, 98, 205, 207, 368 Underspecified benefactive marker╇ 12–13, 33, 265–266 Unintentional action╇ 22, 116, 213, 263, 361–362, 425 V Valency╇ 29–36, 38–39, 48–49, 51–53, 55, 59–61, 150, 199, 204, 257, 275–279, 288–289, 299–301, 303, 334, 346–347 Valency change╇ 31
Valency increase╇ 178, 188, 299, 308, 326, 419–420, 433 Verb extension╇ 296–299, 301–305, 308–309, 311–312, 314 Verb morphology╇ 29, 42, 98, 158, 204, 272, 297, 301, 304, 309–310, 313, 318–319, 341–343, 347 Verbal origin╇ 41–42, 50, 52, 56, 174, 216, 304, 387, 420 Verbal semantics╇ 9, 21, 34, 59–60, 106, 147, 179, 191, 209, 221, 225, 247,
257–259, 266, 268, 324, 332, 380–388, Voice╇ 82, 85, 171–172, 179, 187, 275–276, 289, 331, 341–342, 345, 348, 431 Volitionality╇ 15–16, 23, 60, 251, 255–256, 260–261, 263–266, 269, 424 W Word order╇ 33, 45, 48, 51–52, 74, 185–186, 189, 195–196, 204, 275, 333–334, 351, 377, 381, 387
Typological Studies in Language A complete list of titles in this series can be found on the publishers’ website, www.benjamins.com 92 Zúñiga, Fernando and Seppo Kittilä (eds.): Benefactives and Malefactives. Typological perspectives and case studies. 2010. x, 440 pp. 91 Fiedler, Ines and Anne Schwarz (eds.): The Expression of Information Structure. A documentation of its diversity across Africa. 2010. xii, 383 pp. 90 Traugott, Elizabeth Closs and Graeme Trousdale (eds.): Gradience, Gradualness and Grammaticalization. 2010. ix, 306 pp. 89 Gildea, Spike and Francesc Queixalós (eds.): Ergativity in Amazonia. v, 317 pp. + index. Expected April 2010 88 Schmidtke-Bode, Karsten: A Typology of Purpose Clauses. 2009. xii, 229 pp. 87 Cyffer, Norbert, Erwin Ebermann and Georg Ziegelmeyer (eds.): Negation Patterns in West African Languages and Beyond. 2009. vi, 368 pp. 86 Mahieu, Marc-Antoine and Nicole Tersis (eds.): Variations on Polysynthesis. The Eskaleut languages. 2009. ix, 312 pp. 85 Givón, T. and Masayoshi Shibatani (eds.): Syntactic Complexity. Diachrony, acquisition, neurocognition, evolution. 2009. vi, 553 pp. 84 Newman, John (ed.): The Linguistics of Eating and Drinking. 2009. xii, 280 pp. 83 Corrigan, Roberta, Edith A. Moravcsik, Hamid Ouali and Kathleen M. Wheatley (eds.): Formulaic Language. Volume 2. Acquisition, loss, psychological reality, and functional explanations. 2009. xxiv, 361 pp. 82 Corrigan, Roberta, Edith A. Moravcsik, Hamid Ouali and Kathleen M. Wheatley (eds.): Formulaic Language. Volume 1. Distribution and historical change. 2009. xxiv, 315 pp. 81 Corbett, Greville G. and Michael Noonan (eds.): Case and Grammatical Relations. Studies in honor of Bernard Comrie. 2008. ix, 290 pp. 80 Laury, Ritva (ed.): Crosslinguistic Studies of Clause Combining. The multifunctionality of conjunctions. 2008. xiv, 253 pp. 79 Abraham, Werner and Elisabeth Leiss (eds.): Modality–Aspect Interfaces. Implications and typological solutions. 2008. xxiv, 422 pp. 78 Harrison, K. David, David S. Rood and Arienne Dwyer (eds.): Lessons from Documented Endangered Languages. 2008. vi, 375 pp. 77 Seoane, Elena and María José López-Couso (eds.): Theoretical and Empirical Issues in Grammaticalization. In collaboration with Teresa Fanego. 2008. x, 367 pp. 76 López-Couso, María José and Elena Seoane (eds.): Rethinking Grammaticalization. New perspectives. In collaboration with Teresa Fanego. 2008. x, 355 pp. 75 Frajzyngier, Zygmunt and Erin Shay (eds.): Interaction of Morphology and Syntax. Case studies in Afroasiatic. 2008. v, 234 pp. 74 Kurzon, Dennis and Silvia Adler (eds.): Adpositions. Pragmatic, semantic and syntactic perspectives. 2008. viii, 307 pp. 73 Ansaldo, Umberto, Stephen Matthews and Lisa Lim (eds.): Deconstructing Creole. 2007. xii, 292 pp. 72 Næss, Åshild: Prototypical Transitivity. 2007. x, 240 pp. 71 Nedjalkov, Vladimir P. (ed.): Reciprocal Constructions. With the assistance of Emma Š. Geniušienė and Zlatka Guentchéva. 2007. xxiii, 2219 pp. (5 vols.). 70 Zúñiga, Fernando: Deixis and Alignment. Inverse systems in indigenous languages of the Americas. 2006. xii, 309 pp. 69 Aranovich, Raúl (ed.): Split Auxiliary Systems. A cross-linguistic perspective. 2007. vii, 277 pp. 68 Abraham, Werner and Larisa Leisiö (eds.): Passivization and Typology. Form and function. 2006. x, 553 pp. 67 Veselinova, Ljuba N.: Suppletion in Verb Paradigms. Bits and pieces of the puzzle. 2006. xviii, 236 pp. 66 Hickmann, Maya and Stéphane Robert (eds.): Space in Languages. Linguistic Systems and Cognitive Categories. 2006. x, 362 pp.
65 Tsunoda, Tasaku and Taro Kageyama (eds.): Voice and Grammatical Relations. In Honor of Masayoshi Shibatani. 2006. xviii, 342 pp. 64 Voeltz, F. K. Erhard (ed.): Studies in African Linguistic Typology. 2006. xiv, 426 pp. 63 Filimonova, Elena (ed.): Clusivity. Typology and case studies of the inclusive–exclusive distinction. 2005. xii, 436 pp. 62 Couper-Kuhlen, Elizabeth and Cecilia E. Ford (eds.): Sound Patterns in Interaction. Crosslinguistic studies from conversation. 2004. viii, 406 pp. 61 Bhaskararao, Peri and Karumuri Venkata Subbarao (eds.): Non-nominative Subjects. Volume 2. 2004. xii, 319 pp. 60 Bhaskararao, Peri and Karumuri Venkata Subbarao (eds.): Non-nominative Subjects. Volume 1. 2004. xii, 325 pp. 59 Fischer, Olga, Muriel Norde and Harry Perridon (eds.): Up and down the Cline – The Nature of Grammaticalization. 2004. viii, 406 pp. 58 Haspelmath, Martin (ed.): Coordinating Constructions. 2004. xcv, 578 pp. 57 Mattissen, Johanna: Dependent-Head Synthesis in Nivkh. A contribution to a typology of polysynthesis. 2003. x, 350 pp. 56 Shay, Erin and Uwe Seibert (eds.): Motion, Direction and Location in Languages. In honor of Zygmunt Frajzyngier. 2003. xvi, 305 pp. 55 Frajzyngier, Zygmunt and Erin Shay: Explaining Language Structure through Systems Interaction. 2003. xviii, 309 pp. 54 Aikhenvald, Alexandra Y. and R.M.W. Dixon (eds.): Studies in Evidentiality. 2003. xiv, 349 pp. 53 Givón, T. and Bertram F. Malle (eds.): The Evolution of Language out of Pre-language. 2002. x, 394 pp. 52 Güldemann, Tom and Manfred von Roncador (eds.): Reported Discourse. A meeting ground for different linguistic domains. 2002. xii, 425 pp. 51 Newman, John (ed.): The Linguistics of Sitting, Standing and Lying. 2002. xii, 409 pp. 50 Feigenbaum, Susanne and Dennis Kurzon (eds.): Prepositions in their Syntactic, Semantic and Pragmatic Context. 2002. vi, 304 pp. 49 Wischer, Ilse and Gabriele Diewald (eds.): New Reflections on Grammaticalization. 2002. xiv, 437 pp. 48 Shibatani, Masayoshi (ed.): The Grammar of Causation and Interpersonal Manipulation. 2002. xviii, 551 pp. 47 Baron, Irène, Michael Herslund and Finn Sørensen (eds.): Dimensions of Possession. 2001. vi, 337 pp. 46 Aikhenvald, Alexandra Y., R.M.W. Dixon and Masayuki Onishi (eds.): Non-canonical Marking of Subjects and Objects. 2001. xii, 364 pp. 45 Bybee, Joan and Paul J. Hopper (eds.): Frequency and the Emergence of Linguistic Structure. 2001. vii, 492 pp. 44 Voeltz, F. K. Erhard and Christa Kilian-Hatz (eds.): Ideophones. 2001. x, 436 pp. 43 Gildea, Spike (ed.): Reconstructing Grammar. Comparative Linguistics and Grammaticalization. 2000. xiv, 269 pp. 42 Diessel, Holger: Demonstratives. Form, function and grammaticalization. 1999. xii, 205 pp. 41 Frajzyngier, Zygmunt and Traci S. Walker-Curl (eds.): Reciprocals. Forms and functions. Volume 2. 2000. xii, 201 pp. 40 Frajzyngier, Zygmunt and Traci S. Walker-Curl (eds.): Reflexives. Forms and functions. Volume 1. 2000. xiv, 286 pp. 39 Payne, Doris L. and Immanuel Barshi (eds.): External Possession. 1999. ix, 573 pp. 38 Siewierska, Anna and Jae Jung Song (eds.): Case, Typology and Grammar. In honor of Barry J. Blake. 1998. 395 pp. 37 Giacalone-Ramat, Anna and Paul J. Hopper (eds.): The Limits of Grammaticalization. 1998. vi, 307 pp. 36 Newman, John (ed.): The Linguistics of Giving. 1998. xv, 373 pp. 35 Givón, T. (ed.): Grammatical Relations. A functionalist perspective. 1997. viii, 350 pp. 34 Givón, T. (ed.): Conversation. Cognitive, communicative and social perspectives. 1997. viii, 302 pp. 33 Fox, Barbara A. (ed.): Studies in Anaphora. 1996. xii, 518 pp.
32 31 30 29 28 27 26 25 24 23 22 21
Bybee, Joan and Suzanne Fleischman (eds.): Modality in Grammar and Discourse. 1995. viii, 575 pp. Gernsbacher, Morton Ann and T. Givón (eds.): Coherence in Spontaneous Text. 1995. x, 267 pp. Downing, Pamela A. and Michael Noonan (eds.): Word Order in Discourse. 1995. x, 595 pp. Kahrel (PJK), Peter and René van den Berg (eds.): Typological Studies in Negation. 1994. x, 385 pp. Givón, T. (ed.): Voice and Inversion. 1994. viii, 402 pp. Fox, Barbara A. and Paul J. Hopper (eds.): Voice: Form and Function. 1994. xiii, 377 pp. Lord, Carol: Historical Change in Serial Verb Constructions. 1993. x, 273 pp. Svorou, Soteria: The Grammar of Space. 1994. xiv, 290 pp. Perkins, Revere D.: Deixis, Grammar, and Culture. 1992. x, 245 pp. Kemmer, Suzanne: The Middle Voice. 1993. xii, 300 pp. Payne, Doris L. (ed.): Pragmatics of Word Order Flexibility. 1992. viii, 320 pp. Downing, Pamela A., Susan D. Lima and Michael Noonan (eds.): The Linguistics of Literacy. 1992. xx, 334 pp. 20 Croft, William, Suzanne Kemmer and Keith Denning (eds.): Studies in Typology and Diachrony. Papers presented to Joseph H. Greenberg on his 75th birthday. 1990. xxxiv, 243 pp. 19:2 Traugott, Elizabeth Closs and Bernd Heine (eds.): Approaches to Grammaticalization. Volume II. Types of grammatical markers. 1991. xii, 558 pp. 19:1 Traugott, Elizabeth Closs and Bernd Heine (eds.): Approaches to Grammaticalization. Volume I. Theoretical and methodological issues. 1991. xii, 360 pp. 18 Haiman, John and Sandra A. Thompson (eds.): Clause Combining in Grammar and Discourse. 1988. xiii, 428 pp. 17 Hammond, Michael, Edith A. Moravcsik and Jessica Wirth (eds.): Studies in Syntactic Typology. 1988. xiv, 380 pp. 16 Shibatani, Masayoshi (ed.): Passive and Voice. 1988. xi, 706 pp. 15 Austin, Peter (ed.): Complex Sentence Constructions in Australian Languages. 1988. vii, 289 pp. 14 Hinds, John, Shoichi Iwasaki and Senko K. Maynard (eds.): Perspectives on Topicalization. The case of Japanese WA. 1987. xi, 307 pp. 13 Never published. 12 Nedjalkov, Vladimir P. (ed.): Typology of Resultative Constructions. Translated from the original Russian edition (1983). Translation edited by Bernard Comrie. 1988. xx, 573 pp. 11 Tomlin, Russell S.: Coherence and Grounding in Discourse. Outcome of a Symposium, Eugene, Oregon, June 1984. 1987. viii, 512 pp. 10 Ransom, Evelyn N.: Complementation: its Meaning and Forms. 1986. xii, 226 pp. 9 Bybee, Joan: Morphology. A Study of the Relation between Meaning and Form. 1985. xii, 235 pp. 8 Slobin, Dan I. and Karl Zimmer (eds.): Studies in Turkish Linguistics. 1986. vi, 294 pp. 7 Craig, Colette G. (ed.): Noun Classes and Categorization. Proceedings of a symposium on categorization and noun classification, Eugene, Oregon, October 1983. 1986. vii, 481 pp. 6 Haiman, John (ed.): Iconicity in Syntax. Proceedings of a symposium on iconicity in syntax, Stanford, June 24–26, 1983. 1985. vi, 402 pp. 5 Rutherford, William E. (ed.): Language Universals and Second Language Acquisition. 1984. ix, 264 pp. 4 Chisholm, William, Louis T. Milic and John A.C. Greppin (eds.): Interrogativity. A colloquium on the grammar, typology and pragmatics of questions in seven diverse languages, Cleveland, Ohio, October 5th 1981-May 3rd 1982. 1984. v, 302 pp. 3 Givón, T.: Topic Continuity in Discourse. A quantitative cross-language study. 1983. vi, 492 pp. 2 Haiman, John and Pamela Munro (eds.): Switch Reference and Universal Grammar. Proceedings of a symposium on switch reference and universal grammar, Winnipeg, May 1981. 1983. xv, 337 pp. 1 Hopper, Paul J. (ed.): Tense-Aspect. Between semantics & pragmatics. 1982. x, 350 pp.
