Constructional Reorganization
Constructional Approaches to Language The series brings together research conducted within different constructional models and makes them available to scholars and students working in this and other, related fields. The topics range from descriptions of grammatical phenomena in different languages to theoretical issues concerning language acquisition, language change, and language use. The foundation of constructional research is provided by the model known as Construction Grammar (including Frame Semantics). The book series publishes studies in which this model is developed in new directions and extended through alternative approaches. Such approaches include cognitive linguistics, conceptual semantics, interaction and discourse, as well as typologically motivated alternatives, with implications both for constructional theories and for their applications in related fields such as communication studies, computational linguistics, AI, neurology, psychology, sociology, and anthropology. This peer reviewed series is committed to innovative research and will include monographs, thematic collections of articles, and introductory textbooks.
Editors Mirjam Fried
Department of Slavic Languages & Literatures, Princeton University, USA
Jan-Ola Östman
Department of Scandinavian Languages & Literature, University of Helsinki, Finland
Advisory Board Peter Auer
Seizi Iwata
Hans C. Boas
Paul Kay
William Croft
Knud Lambrecht
Charles J. Fillmore
Michael Tomasello
Adele E. Goldberg
Arnold M. Zwicky
University of Freiburg, Germany University of Texas at Austin, USA University of New Mexico, USA Int. Computer Science Institute, Berkeley, USA Princeton University, USA
Volume 5 Constructional Reorganization Edited by Jaakko Leino
Osaka City University, Japan University of California, Berkeley, USA University of Texas at Austin, USA Max Planck Institute for Evolutionary Anthropology, Germany Stanford University, USA
Constructional Reorganization Edited by
Jaakko Leino University of Helsinki
John Benjamins Publishing Company Amsterdam / 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 Constructional reorganization / edited by Jaakko Leino. p. cm. (Constructional Approaches to Language, issn 1573-594X ; v. 5) Includes bibliographical references and index. 1. Grammar, Comparative and general. 2. Linguistic change. I. Leino, Jaakko. P151.C635 2008 417.7--dc22 isbn 978 90 272 1827 8 (Hb; alk. paper)
2007045478
© 2008 – 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
chapter 1 Grammar as a Construction Site Jaakko Leino
1
1. Introduction 1 2. Chapter by chapter overview 3 3. Main themes of the volume 5 References 7 chapter 2 Resolving form-meaning discrepancies in Construction Grammar Hans C. Boas
11
1. Introduction 11 2. Syntactic properties of verbs 14 3. Form-meaning relations among abstract constructions 16 4. AHTY as an idiomatic construction 18 5. The architecture of (Construction) Grammar 28 6. Conclusions and outlook 32 References 34 chapter 3 Language change, variability, and functional load: Finnish genericity from a constructional point of view Pentti Leino and Jan-Ola Östman 1. 2. 3. 4. 5. 6. 7.
Overview 37 Constructions and language change 38 The empirical issue 39 Conditions of usage 40 The inception, emergence, generalization, and acceptance of a construction 41 The construction in relation to other constructions 43 The sä passive as a construction 46
37
vi
Table of contents
8. Theoretical implications for language change 48 9. Language change in terms of constructions 50 52 References chapter 4 Precategoriality and argument structure in Late Archaic Chinese Walter Bisang
55
1. Introduction – prerequisites for describing precategoriality in a Construction Grammar framework 55 2. Argument structure constructions 65 3. The pragmatic use of precategoriality 77 4. Conclusion and outlook 83 References 86 chapter 5 Variations in Japanese honorification – deviations or a change in the making? Yoshiko Matsumoto
89
1. Introduction 89 2. O-V-suru construction: Use 1 – Nonsubject honorifics 92 3. O-V-suru construction: Direction toward performative honorifics 97 4. Conclusion 100 References 102 chapter 6 Constructing reasoning: The connectives för att (causal), så att (consecutive) and men att (adversative) in Swedish conversations Jan Lindström and Anne-Marie Londen
105
1. Introduction 105 2. Syntactic composition and distribution 109 3. Functions of the connective att 111 4. Functions of för att in conversational sequences 116 5. Functions of så att in conversational sequences 128 6. Functions of men att in conversational sequences 138 7. Summary and conclusions 145 References 149 Subject index Index of constructions
153 155
chapter 1
Grammar as a Construction Site Jaakko Leino University of Helsinki, Finland
1.
Introduction
The present volume consists of several different applications of the Construction Grammar framework (henceforth CxG) to areas such as language change, variation, and the internal organization of grammar: in short, the object of study is the constantly changing language system. Thereby, the book addresses two main questions which are of paramount interest to linguists working with the notion of grammatical construction: Where do constructions come from? And, how are the grammatical constructions in a given language organized to form a coherent whole commonly referred to as “grammar”? When we address these language-related issues, the more general question of cognitive organization arises. How does the human cognitive apparatus store the apparently immense amount of information that a speaker of any given human language has about his/her language? How is this information related to any “non-linguistic” knowledge that the speaker has? And, specifically, how does language change interact with this organization? The crucial theoretical question is: how can a constructional approach deal with such questions? The book brings together the latest developments in grammatical theory – specifically within the CxG framework – on the one hand, and empirical findings and data, language-specific research traditions, and cross-language issues, on the other. It may be slightly misleading to refer to “the CxG framework”, however. This term should not be taken to imply that CxG is a unified formalism. Rather, it is best thought of as a rather loose school of thought. The work of Charles Fillmore, Paul Kay and their collaborators (e.g. Fillmore 1988; Fillmore, Kay & O’Connor 1988; Fillmore & Kay 1996; Kay & Fillmore 1999; Fried & Östman 2004b) may rightfully be described as central to CxG, but in addition, there is a lot of work which contributes to the CxG enterprise by assuming a particular ob-
Jaakko Leino
ject of study, and a corresponding point of view, while holding on to the common construction-based conception of language. A fairly good idea of the variety of approaches covered by the term Construction Grammar is given by e.g. Östman & Fried (2005), Croft (2001, Chapter 1), and Goldberg (2006, Chapter 10). Probably the most widely known “branches of the CxG tree” are the Goldbergian argument structure construction approach (e.g. Goldberg 1992, 1995, 2002, 2006) and the Radical Construction Grammar line of thought of William Croft (e.g. Croft 2001, 2005). Goldberg’s work is centered on argument structure constructions, and more recently, on language acquisition and on psycholinguistic evidence for the CxG approach. Croft has a typological orientation, and he also goes to the very roots of grammatical thought and the background assumptions shared by most approaches to grammar and language in general. Other notable lines of work within CxG cover various aspects of language, such as information structure (e.g. Lambrecht 1994, 1996, 2001; Lambrecht & Lemoine 2004; Michaelis & Lambrecht 1996), cross-language issues (Fried & Östman 2004a), language acquisition (e.g. Tomasello 2003, Johnson 1999, Kauppinen 1999), discourse and pragmatics (e.g. Fischer 2001, Östman 2005, Fried & Östman 2005, Fujii 2001, Günthner & Imo 2006), corpus studies (e.g. Stefanowitsch & Gries 2003; Gries 2003, 2005), language variation and change (e.g. Israel 1996; P. Leino & al. 2001; J. Leino 2003; Fried 2005; J. Leino & Östman 2005), etc., to cite some examples of the existing research – the list by no means pretends to be exhaustive or cover all the different lines of work within CxG. Some of the current work focuses primarily on syntax, some approaches have a strong interest in semantics as well, and others are oriented more towards discourse and pragmatics. Overall, CxG has effectively grown from its original status, an approach to syntax and semantics, into a comprehensive view of language. Yet, CxG is probably best thought of as a collaborative enterprise rather than a coherent theory: a considerable part of the work done within CxG is the result of combining the notion of grammatical construction, and the insights of CxG in general, with other approaches to language and grammar. Accordingly, a lot of those previous works which fall within the scope of the present volume combine a constructional approach with other elements, such as grammaticalization research (e.g. Traugott 2003; Diewald 2006; Rostila 2006; Hilpert 2007), and many of the contributions in this volume are also related in several ways to a number of diachronically oriented works done within cognitive linguistics more generally (e.g. Sweetser 1990; Geeraerts 1997; Kemmer 2001 to give a few examples).
2.
Grammar as a Construction Site
Chapter by chapter overview
The present volume is composed of independent but interrelated contributions, each of which serves as one chapter of this volume. The contributions have been selected, edited, and organized in such a way as to form a compact and coherent whole. However, no heavy streamlining or homogenizing has been made either with regard to the theoretical approach or to the linguistic substance of the papers. The reason for this is simply the fact that one of the specific aims of this volume is to present various construction-based approaches to language, various interpretations of CxG; this theoretical versatility is thus a goal to be aimed at rather than a negative feature to be avoided. Chapter 2 of this volume, Resolving Form-meaning Discrepancies in Construction Grammar by Hans C. Boas, investigates different strategies for determining the multiple meanings of sentences that all exhibit the same syntactic frame, in this case, the ‘a hole through Y’ construction. A central theme of the chapter is fitting morphosyntactic and semantic generalizations together with lexical items and their specific properties. Boas shows that it is not always possible to automatically assign an abstract constructional meaning to a given sentence based on a high-level Goldberg-type construction (caused-motion, ditransitive, etc.). Instead, Boas demonstrates that it is important to pay close attention to the lexical semantic properties of verbs capable of occurring in a specific syntactic frame. The chapter builds upon Boas’ earlier work (Boas 2003) on lexical mini-constructions and the interconnections of grammar and lexicon. Pentti Leino’s and Jan-Ola Östman’s contribution in Chapter 3, Language change, variability, and functional load: Finnish genericity from a constructional point of view, discusses the feasibility of explaining language change and variability with the help of grammatical constructions and CxG. The analysis focuses on an emergent construction in Finnish where the second person singular form is being used as a generic form (cf. English You just drive). Leino and Östman show that a full understanding of the mechanisms involved in the emergence of this construction, as well as the mechanisms of language change in general, require an integration of syntactic and semantic analyses with discourse and areal information, and an acceptance of a view where the language system (i.e. the grammar) is in a constant flux. Specifically, the authors show how language change and variability are orchestrated through an intricate interplay between the two forces of the form-meaning constellations as impersonal constructions, on the one hand, and functionally motivated discourse patterns (Östman 1999, 2005), on the other. Chapter 3 touches upon two important theoretical and methodological issues. The first one is delimiting the object of study, pointing out the units of de-
Jaakko Leino
scription and discussion: what is it exactly that we speak about when we discuss a particular linguistic phenomenon? Is it, for example, a given verb form, one of its specific uses, or perhaps a combination of the particular form, the type of context where it is used, and the semantic and pragmatic interpretation that it receives in that type of context? The second, closely related issue is that of integrating different aspects of language in a single explanatory account: not only referring to syntactic and semantic features, but also taking into account discourse factors, areal phenomena, the division of labor (or sharing of functional load) between different constructions, and the like. In addition to taking a CxG approach, Chapter 3 also draws heavily on insights from Cognitive Grammar. Overall, the study stresses the necessity of pragmatic explanations in grammar, and presents CxG as a functional theory of language. In Chapter 4, Precategoriality and argument structure in Late Archaic Chinese, Walter Bisang shows that words cannot be unproblematically divided into discrete parts of speech in Late Archaic Chinese. Bisang’s contribution provides an intriguing insight into the problems of parts of speech as theoretical and typological entities, which was promoted as a rather central topic within construction grammar by Croft’s Radical Construction Grammar (2001). Chapter 4 also addresses the relationship between grammar and lexicon. Chapter 4 relates to Boas’ discussion on the connections between lexical semantics and constructions in Chapter 2, as well as to Leino & Östman’s ideas about pragmatic factors as explanatory tools in CxG in Chapter 3. Bisang starts from the observation that the lexicon does not determine the syntactic distribution of its items in the V[erb]-position. The assignment of a lexical item to the V-position is merely determined by pragmatic implicatures, stereotypes, which can be flouted for rhetorical purposes. In Late Archaic Chinese, the meaning of the lexemes in the V-position can be predicted from the interaction of the meaning of the construction and the lexical meaning of the word much more systematically than is the case with English, for example. Japanese is regularly cited as having a complex system of honorifics, both morphosyntactically and socially. In Chapter 5, Variations in Japanese honorification – deviations or a change in the making? Yoshiko Matsumoto approaches some observed variations in this area from the point of view of CxG. Variations and instability in the use of Japanese honorifics have often been regarded as grammatical deviations produced by speakers who lack the appropriate linguistic and social training. Examining attested discourse data, Matsumoto argues that the Japanese nonsubject honorific construction [o-Verbstem-suru] is undergoing a change and is becoming an addressee honorific. This change reflects a cognitive reorganization from the elements of the sentence to the speech context. Matsumoto shows that
Grammar as a Construction Site
the variations in the [o-Verbstem-suru] form can be more systematically explained from the constructional and frame semantic perspectives as a process of cognitive and intersubjective (cf. Traugott and Dasher 2002) reorganization of grammatical constructions motivated by contextual conditions of use and the speaker’s intention with respect to the addressee. Matsumoto’s account extends CxG towards an intersubjective dimension, which is further developed by Jan Lindström and Anne-Marie Londen in Chapter 6, Constructing reasoning: The complex conjunctions för att (causal), så att (consecutive) and men att (adversative) in Swedish conversations, the final chapter of the present volume. This chapter investigates three common formally complex conjunctions in spoken Swedish. These conjunctions are studied as constructions that enter in characteristic form–function relations in organizing the course of interaction. The authors use Conversation Analysis (CA, e.g. Sacks 1992 a, b; Goodwin & Heritage 1990; Heritage 1995) as a theoretical basis for this chapter, together with CxG. The combination of CxG with CA – also recently discussed by e.g. Fried & Östman (2005) – is highly valuable and fruitful for the development of both of these approaches, and there is a considerable interest in this combination both within CxG and CA. The reasons for this are clear. First, CxG and CA are remarkably compatible in some central aspects. Notably, both frameworks view both grammatical (in the strict sense) patterns and discourse practices as conventional models of action which emerge from language use and acquisition, and which are constantly re-negotiated by the participants of any discourse event. Second, CxG and CA complement each other in the sense that CxG has to a large degree focused on sentence structure and related matters, but not shown much interest in turn-taking practices and interaction in general, whereas CA takes the interactional character of language as its very starting point, but has not gone particularly deep into the (strictly speaking) syntactic structure of utterances. As Fried and Östman (2005: 1756) put it, “CxG and CA are compatible approaches that do different things and aspire to do different things, but they accomplish their goals in a complementary manner.”
3.
Main themes of the volume
Much of the traditional descriptive work on language diachrony has focused on the question of what has changed. A more elaborated question is that of how – i.e., how the process of change unfolds. What the diachronically oriented chapters of this volume wish to address are primarily questions of the why type: why is it that
Jaakko Leino
certain changes have taken place? Why does language change in the first place? And why does it change in the manner it does? The questions of the why type bring in notions like analogy, motivation, iconicity, and the isomorphism of formal and semantic structure. They also call for more elaborate answers to the what and how questions: Something has changed, but under what circumstances? In what types of context? Such issues, in turn, bring forth usage-based approaches to language, as well as the different phenomena related to discourse and context, etc. By addressing these types of questions, the authors of the present volume assume a point of view very much similar to that of grammaticalization theory, which has already suggested connections with CxG in recent years; notably, proponents of grammaticalization theory have turned towards a construction-based approach (cf. Traugott & Dasher 2002; Traugott 2003; Croft 2000). As pointed out above, one of the most striking properties of CxG, shown both in this volume and elsewhere, is a certain openness to influence from other approaches, a kind of compatibility which makes it possible to use CxG together with other theories. Many of the scholars who use CxG also use some other approach by its side: CxG is very commonly used as one part of a “mixed” framework which combines the construction-based ideas of CxG with some other fundamental ideas originating from some other framework. This is particularly clear in many of the papers published in Östman & Fried (2005): CxG has natural connections with theories like Conceptual Semantics (e.g. Jackendoff 1983, 1990; cf. Nikanne 2005), Word Grammar (Hudson 1984, 1990; cf. Holmes & Hudson 2005), and Cognitive Grammar (Langacker 1987, 1991, 2005; cf. J. Leino 2005), for example – the list could easily be extended. This openness is present in this volume as well. The chapters of this book include works which combine Construction Grammar with theories and approaches like Cognitive Grammar (notably Leino & Östman), Conversation Analysis and interactional linguistics (Lindström & Londen), and grammaticalization theory (Bisang, Matsumoto). Overall, the main focus of this volume is the internal organization of the language system (or grammar), and changes which take place in the system. The chapters of the book also touch upon issues relating to the intersection of diachrony – especially diachronic semantics and morphosyntax – and the use of various linguistic entities in text and discourse. The book approaches the organization of grammar and grammatical constructions both synchronically and diachronically. Synchronically, the authors present analyses of groups of interrelated constructions, and study the relationships and the division of labor between related constructions within a language. Diachronically, they bring forth analyses of changes in the grammar and the emergence of new constructions.
Grammar as a Construction Site
References Chomsky, Noam (1980). Rules and representations. Oxford: Basil Blackwell. Croft, William (2000) Explaining Language Change. An Evolutionary Approach. London: Longman. Croft, William (2001). Radical Construction Grammar. Syntactic Theory in Typological Perspective. Oxford: Oxford University Press. Croft, William (2005). Logical and typological arguments for Radical Construction Grammar. In Östman, Jan-Ola & Mirjam Fried (eds.), Construction Grammars. Cognitive grounding and theoretical extensions [Constructional approaches to language, 3] (273–314). Amsterdam: John Benjamins. Diewald, Gabriele (2006). Context types in grammaticalization as constructions. In Constructions special volume 1 – Constructions all over: Case studies and theoretical implications. (www.constructions-online.de) Fillmore, Charles J. (1988). The Mechanisms of “Construction Grammar”. In Proceedings from the 14th Annual meeting of the Berkeley Linguistics Society (35–55). Berkeley: Berkeley Linguistics Society. Fillmore, Charles J. & Paul Kay (1996). Construction Grammar. Unpublished manuscript. Stanford: CSLI. Fillmore, Charles & Paul Kay & Mary Catherine O’Connor (1988). Regularity and Idiomaticity in Grammatical Constructions: The Case of Let Alone. In Language, 64:3, 501–538. Fischer, Kerstin (2001). Pragmatic Methods for Construction Grammar. In Holmer, Arthur & Jan-Olof Svantesson & Erik Viberg (eds.), Proceedings of the 18th Scandinavian Conference of Linguistics [Travaux de l’Institut de Linguistique de Lund]. Lund: Lund University. Fried, Mirjam (2005). The stability of meaning-form associations across time. In Nejedlý, Petr & Miloslava Vajdlová (eds.) Verba et historia: Igoru Němcovi k 80. narozeninám (77–86). Praha: Ústav pro jazyk český AV ČR. Fried, Mirjam & Jan-Ola Östman (eds.) (2004a). Construction Grammar in a Cross-Language Perspective [Constructional Approaches to Language, 2]. Amsterdam: John Benjamins. Fried, Mirjam & Jan-Ola Östman (2004b). Construction Grammar: A thumbnail sketch. In Fried, Mirjam & Jan-Ola Östman (eds.), Construction Grammar in a Cross-Language Perspective [Constructional Approaches to Language, 2] (11–86). Amsterdam: John Benjamins. Fried, Mirjam & Jan-Ola Östman (2005). Construction Grammar and spoken language: The case of pragmatic particles. In Journal of Pragmatics, 37, 1752–1778. Fujii, Seiko (2001). Background knowledge and constructional meanings: English and Japanese concessive conditional clause-linking constructions. In Nemeth, Emiko (ed.) Pragmatics in 2000: Selected Papers from the 7th International Pragmatics Conference Vol. 2. Antwerp: International Pragmatics Association. Geeraerts, Dirk (1997). Diachronic Prototype Semantics. A Contribution to Historical Lexicology [Oxford Studies in Lexicography and Lexicology]. Oxford: Clarendon Press. Goldberg, Adele E. (1992). The inherent semantics of argument structure: The case of the English ditransitive construction. Cognitive Linguistics, 3:1, 37–74. Goldberg, Adele E. (1995). Constructions: A Construction Grammar approach to argument structure. Chicago: University of Chicago Press.
Jaakko Leino
Goldberg, Adele E. (2002). Surface generalizations: An alternative to alternations. Cognitive Linguistics, 13:4, 327–356. Goldberg, Adele E. (2006). Constructions at Work. The Nature on Generalization in Language. New York: Oxford University Press. Goodwin, Charles – Heritage, John (1990). Conversation Analysis. Annual Review of Anthropology, 19, 283–307. Gries, Stefan Th. (2003). Towards a corpus-based identification of prototypical instances of constructions. Annual Review of Cognitive Linguistics, 1, 1–27. Gries, Stefan Th. (2005). Syntactic priming: A corpus-based approach. Journal of Psycholinguistic Research, 34:4, 365–99. Günthner, Susanne & Wolfgang Imo (eds.) (2006). Konstruktionen in der Interaktion. Berlin: De Gruyter. Heritage, John (1995). Garfinkel and Ethnomethodology. Cambridge: Polity Press. Hilpert, Martin (2007). Germanic Future Constructions. A Usage-based Approach to Grammaticalization. Doctoral Dissertation, Rice University. Holmes, Jasper & Richard Hudson (2005). Constructions in Word Grammar. In Östman, JanOla & Mirjam Fried (eds.), Construction Grammars. Cognitive grounding and theoretical extensions [Constructional approaches to language, 3] (243–272). Amsterdam: John Benjamins. Hudson, Richard (1984). Word grammar. Oxford: Blackwell. Hudson, Richard (1991). English word grammar. Cambridge: Blackwell. Israel, Michael (1996). The way constructions grow. In Goldberg, Adele (ed.), Conceptual Structure, Discourse, and Language (217–230). Stanford: CSLI Publications. Jackendoff, Ray (1983). Semantics and Cognition. Cambridge: MIT Press. Jackendoff, Ray (1990). Semantic Structures. Cambridge: MIT Press. Johnson, Christopher R. (1999). Constructional grounding: The role of interpretational overlap in lexical and constructional acquisition. Ph.D. dissertation, University of California, Berkeley. Kauppinen, Anneli (1999). Acquisition of Finnish conditional verb forms in formulaic utterances. In Hiraga, Masako & Chris Sinha & Sherman Wilcox (eds.): Cultural, Psychological and Typological Issues in Cognitive Linguistics. Amsterdam: John Benjamins. Kay, Paul & Fillmore, Charles J. (1999). Grammatical Constructions and Linguistic Generalizations: The What’s X Doing Y? Construction. Language, 75:1, 1–33. Kemmer, Suzanne (2001). Causative Constructions and Cognitive Models: The English Make Causative. First Seoul International Conference on Discourse and Cognitive Linguistics: Perspectives for the 21st Century (803–846). Seoul: Discourse and Cognitive Linguistics Society of Korea. Lambrecht, Knud (1994). Information structure and sentence form. Topic, focus, and the mental representations of discourse referents. Cambridge: Cambridge University Press. Lambrecht, Knud (1996). On the formal and functional relationship between topics and vocatives. Evidence from French. In Goldberg, Adele (ed.), Conceptual Structure, Discourse, and Language (267–288). Stanford: CSLI Publications. Lambrecht, Knud (2001). A framework for the analysis of cleft constructions. Linguistics, 39: 3, 463–516.
Grammar as a Construction Site
Lambrecht, Knud & Kevin Lemoine (2004). On the interaction of information structure and formal structure in constructions: The case of French right-detached comme-N. In Fried, Mirjam & Jan-Ola Östman (eds.), Construction Grammar in a Cross-Language Perspective [Constructional Approaches to Language, 2] (157–199). Amsterdam: John Benjamins. Langacker, Ronald W. (1987). Foundations of Cognitive Grammar I: Theoretical Prerequisites. Stanford: Stanford University Press. Langacker, Ronald W. (1991). Foundations of Cognitive Grammar II: Descriptive Application. Stanford: Stanford University Press. Langacker, Ronald W. (2005). Integration, grammaticization, and constructional meaning. In Mirjam Fried & Hans Boas (eds.): Construction Grammar: Back to the Roots [Constructional approaches to language, 4] (157–189). Amsterdam: John Benjamins. Leino, Jaakko (2003). Antaa sen muuttua. Suomen kielen permissiivirakenne ja sen kehitys (Let it change. The Finnish permissive construction and its history). Helsinki: Finnish Literature Society. Leino, Jaakko (2005). Frames, profiles and constructions. Two collaborating CGs meet the Finnish Permissive Construction. In Östman, Jan-Ola & Mirjam Fried (eds.), Construction Grammars. Cognitive grounding and theoretical extensions [Constructional approaches to language, 3] (89–120). Amsterdam: John Benjamins. Leino, Jaakko & Jan-Ola Östman (2005). Constructions and variability. In Mirjam Fried & Hans Boas (eds.): Construction Grammar: Back to the Roots [Constructional approaches to language, 4] (191–213). Amsterdam: John Benjamins. Leino, Pentti & Ilona Herlin & Suvi Honkanen & Lari Kotilainen & Jaakko Leino (2001). Roolit ja rakenteet. Henkilöviitteinen allatiivi Biblian verbikonstruktioissa. (Roles and structures. Person-referring allative in the verbal constructions of the Finnish Bible from the year 1642.) Helsinki: Finnish Literature Society. Michaelis, Laura & Knud Lambrecht (1996). Toward a construction-based theory of language function: The case of nominal extraposition. Language, 72, 215–247. Nikanne, Urpo (2005). Constructions in Conceptual Semantics. In Östman, Jan-Ola & Mirjam Fried (eds.), Construction Grammars. Cognitive grounding and theoretical extensions [Constructional approaches to language, 3] (191–242). Amsterdam: John Benjamins. Östman, Jan-Ola (1999). Coherence through understanding through discourse patterns. Focus on news reports. In W. Bublitz et al. (eds.) Coherence in spoken and written discourse: How to create it and how to describe it (77–100). Amsterdam: John Benjamins. Östman, Jan-Ola (2005). Construction discourse: A prolegomenon. In Östman, Jan-Ola & Mirjam Fried (eds.), Construction Grammars. Cognitive grounding and theoretical extensions [Constructional approaches to language, 3] (121–144). Amsterdam: John Benjamins. Östman, Jan-Ola & Mirjam Fried (2004). Historical and intellectual background of Construction Grammar. In Fried, Mirjam & Jan-Ola Östman (eds.), Construction Grammar in a Cross-Language Perspective [Constructional Approaches to Language, 2] (1–10). Amsterdam: John Benjamins. Östman, Jan-Ola & Mirjam Fried (eds.) (2005). Construction Grammars. Cognitive grounding and theoretical extensions [Constructional approaches to language, 3]. Amsterdam: John Benjamins. Pinker, Steven (1994). The Language Instinct. The New Science of Language and Mind. London: Penguin Books.
10
Jaakko Leino
Rostila, Jouni (2006). Storage as a way to grammaticalization. Constructions 1/2006 (www.constructions-online.de). Sacks, Harvey (1992a, b). Lectures on conversation, vol. 1 & 2. Edited by Gail Jefferson. Oxford: Blackwell. Stefanowitsch, Anatol & Stefan Gries (2003). Collostructions: Investigating the interaction of words and constructions. International Journal of Corpus Linguistics, 8, 209–243. Sweetser, Eve (1990). From Etymology to Pragmatics. Cambridge: Cambridge University Press. Johnson, Christopher, R. (1999). Constructional grounding: The role of interpretational overlap in lexical and constructional acquisition. Ph.D. dissertation, University of California, Berkeley. Tomasello, Michael (2003). Constructing a language. A Usage-Based Theory of Language Acquisition. Cambridge: Harvard University Press. Traugott, Elizabeth Closs (2003). Constructions in Grammaticalization. In Joseph, Brian & Richard D. Janda (eds.) The Handbook of Historical Linguistics (624–647). Oxford: Blackwell. Traugott, Elizabeth Closs & Richard B. Dasher (2002). Regularity in semantic change. Cambridge: Cambridge University Press.
chapter 2
Resolving form-meaning discrepancies in Construction Grammar Hans C. Boas University of Texas at Austin
1.
Introduction
A central idea in Construction Grammar is that there is no strict division between the lexicon and syntax (Fillmore et al. 1988; Jurafsky 1992; Goldberg 1995). It has been argued that although grammatical constructions differ in their complexity, they are basically the same type of declaratively represented data structures that pair form with meaning (Goldberg 1995: 7). However, there has been relatively little work detailing the exact nature of the relationship between form and meaning in Construction Grammar. This is important when an analysis aims to posit different abstract argument structure constructions for licensing a single syntactic structure that may be interpreted in different ways. For example, Goldberg (1995) claims that the caused-motion and resultative interpretations of the sentences in (1) and (2) are ultimately due to the fact that independently existing meaningful constructions are capable of contributing additional arguments to the basic senses of verbs. (1) He sneezed the napkin off the table. (2) She drank him under the table.
(Goldberg 1995: 9) (Goldberg 1995: 157)
According to Goldberg, the caused-motion and resultative semantics associated with the constructions are encoded syntactically by a [NP V NP PP/AP] frame that typically does not occur with verbs such as those in (1) and (2). That is, in (2) the final interpretation of the combined verbal and constructional semantics is ‘X CAUSES Y TO MOVE Z by drinking’. This meaning is paired with a three-argument syntactic frame [NP V NP PP] that expresses the combined semantics of the verb and the construction. One advantage of this approach lies in the fact that it is not necessary to posit implausible verb senses based on intransitive verbs in order
12
Hans C. Boas
to account for the distribution of postverbal arguments in the examples above. Furthermore, because of the close relationship between form and meaning it becomes possible to infer the meaning of a sentence based on its form (syntactic bootstrapping, Landau & Gleitman 1985; Fisher et al. 1991; Gleitman 1992). While Goldberg’s constructional approach is quite successful at explaining the distribution of a large number of caused-motion and resultative constructions, it is not entirely clear how her abstract constructions would account for the distribution of postverbal arguments occurring in a syntactic pattern similar to that in (1) and (2). (3) You pushed a hole through the crazing. (BNC) (4) The army blew a hole in the barrier round the Americans’ fine embassy, … (BNC) (5) He suggests we knock a hole through the wall. (BNC)
The sentences in (3)–(5) display the same syntactic structure as the sentences in (1) and (2), that is, they exhibit a [NP V NP PP] frame. If one had never heard the verbs in (3)–(5) used in that syntactic pattern before, one might expect the meanings of (3)–(5) to be parallel to the meanings of (1) and (2). However, this is not the case. A comparison of the two example sets illustrates that it is not always possible to arrive at an unambiguous interpretation of sentences based on the meaning of a basic sense of a verb in combination with a syntactic frame that is not intransitive. Our examples show that there exist fundamental discrepancies between the relationship of a single syntactic frame [NP V NP PP] in combination with a basic sense of a verb and the range of different types of possible interpretations that are attributed to abstract constructions in Goldberg’s framework. The goal of this paper is to determine how such form-meaning discrepancies can be accounted for in Construction Grammar. Taking as a test case the ‘A hole through Y’ constructional pattern exemplified by sentences such as in (3)–(5), I will discuss in detail the role played by abstract Goldberg-style constructions in contributing different types of meanings to verbs belonging to different semantic classes. The insight emerging from this investigation is that form-meaning discrepancies can be accounted for when one pays closer attention to constructions that are much narrower in semantic scope. On this view, constructions exist at different levels of semantic specificity with respect to both the types of situations they may denote and how much lexical material they may contain.
. For a critical discussion of Goldberg’s approach to Construction Grammar, see Kay (1996, 2002), Nemoto (1998), and Boas (2002b, 2003a).
Resolving form-meaning discrepancies
Any analysis of form-meaning discrepancies in Construction Grammar necessitates an understanding of Frame Semantics, which proposes that a “word’s meaning can be understood only with reference to a structured background of experience, beliefs, or practices, constituting a kind of conceptual prerequisite for understanding the meaning” (Fillmore & Atkins 1992: 76). An example is the Theft Frame, which represents a scenario with different Frame Elements that can be regarded as instances of broader semantic roles such as agent, undergoer, instrument, etc. (cf. Boas 2005a). This frame represents a scenario with different core Frame Elements such as goods (anything that can be taken away), perpetrator (the person or other agent that takes the goods away), source (the initial location of the goods before they change location), and victim (the person that owns the goods before they are taken away). The frame description defines the relationships between Frame Elements of the same frame. For the Theft frame, the frame description states that a perpetrator takes goods that belong to a victim. To exemplify, the verb steal evokes the Theft frame in sentences such as Ben stole the book from Nancy. The needed background to interpret the sentence requires an understanding of illegal activities, property ownership, taking things, and a great deal more (Boas 2005a: 139). For a detailed overview of Frame Semantics, see Petruck (1996). Before I begin with my analysis it is necessary to briefly explain the nature of my data. One major source of data used in this paper comes from electronic corpora. The data source is indicated in parentheses following example sentences, e.g., the British National Corpus (BNC), an electronic corpus containing ca. 100 million words of contemporary spoken (10%) and written (90%) English (see Kennedy (1998: 50–53) and http://www.hcu.ox.ac.uk/). Another provider of data were Usenet group archives and websites on the Internet. These archives contain texts from e-mails and web pages written by people who are relatively free and creative in their use of language. The messages of different Usenet groups can now be accessed at http://www.google.com/groups. The web-page address or the Message-ID number follows Internet citations. Searching the web for specific morpho-syntactic patterns and novel uses of verbs is by now an established method for data collection. See, for example, the special issue of Computational Linguistics 29(3) on “The Web as Corpus” (in particular Kilgarriff and Grefenstette 2003). Finally, I collected acceptability judgments to determine whether sentences with novel uses of words sounded acceptable to native speakers of English. These data supplemented my corpus data. To this end, I followed a loose sampling technique (see Johnstone 2000: 92). Judgments about the acceptability of examples come from twenty undergraduate students at the University of Texas. The students filled out a two-page questionnaire containing two types of sentences: those discussed in this paper, and control sentences that did not contain the types of
13
14
Hans C. Boas
argument structure constructions discussed in this paper. Students were asked to indicate whether sentences sounded acceptable or not. When the great majority of informants found a sentence to be unacceptable, a sentence is marked by an asterisk (‘*’). The use of the asterisk does not necessarily indicate that the sentence is ungrammatical (see Chomsky’s (1965: 10–21) discussion of grammaticalness vs. acceptability), but that it sounds unacceptable to the majority of my student informants. The use of question marks (‘?’) indicates that informants differ with respect to their acceptability judgments. The use of double question marks (‘??’) indicates that a large number of speakers find the sentence unacceptable. The paper is structured as follows. Section two gives a more complete overview of the types of verbs that occur with what we will call for now the “A hole through Y” constructional pattern (henceforth AHTY). Section three takes a closer look at why Goldberg’s independently existing meaningful constructions are problematic when it comes to determining the distribution of verbs occurring with AHTY. Section four offers an alternative analysis that looks in more detail at the lexical semantic properties of verbs capable of occurring with AHTY, suggesting that AHTY is an idiomatic construction in its own right. Section five discusses the relationship between AHTY and other grammatical constructions. Finally, section six proposes how to go about resolving form-meaning discrepancies in Construction Grammar, summarizes our findings, and suggests directions for further research.
2.
Syntactic properties of verbs
Before going into a detailed analysis of the status of AHTY it is necessary to provide a more comprehensive overview of its distribution among different types of verbs. As pointed out in our discussion of examples (3)–(5), AHTY can only occur as a whole unit with certain types of verbs. Consider the following examples. (6) a. *You pushed a hole. b. ??You pushed through the crazing. c. You pushed a hole through the crazing. (BNC) (7) a. *He suggests we knock a hole. b. ??He suggests we knock through the wall. c. He suggests we knock a hole through the wall. (BNC) (8) a. *The food won’t burn a hole. b. ?The food won’t burn through you either. c. The food won’t burn a hole through you either. (www.gothicchicago.com/hunger.html)
Resolving form-meaning discrepancies
(9) a. b. c.
*The fowl projectile has blown a hole. ?The fowl projectile has blown through your chest. The fowl projectile has blown a hole through your chest. (www.amerinfoserv.com/maverick/doom.html)
(6)–(9) illustrate that verbs such as push, knock, burn, and blow may occur only with the whole phrase a hole through Y and not with parts of it. Verbs requiring the presence of a hole through X as a coherent unit in postverbal position will be called Class I verbs. One semantic property Class I verbs share with what will be called Class II verbs is the fact that they denote activities in which an entity that can be construed as an agent exerts energy. This is illustrated by the following examples with Class II verbs. (10) a. Using a hammer drill and carbide bit, drill a hole. b. Using a hammer drill and carbide bit, drill through the sill plate. c. Using a hammer drill and carbide bit, drill a hole through the sill plate. (www.ci.campell.ca.us/strctprep.html) (11) a. The fluid is drained out from under the retina by creating a hole. b. *The fluid is drained out from under the retina by creating through the whole part of the eye. c. The fluid is drained out from under the retina by creating a hole through the whole part of the eye. (www.eyecaresite.com/retina_s.html) (12) a. Once I can make a hole. b. *Once I can make through them. c. Once I can make a hole through them. (BNC) (13) a. Meanwhile, Amy successfully digs a hole. b. Meanwhile, Amy successfully digs through the oubliette, and escapes. c. Meanwhile, Amy successfully digs a hole through the oubliette, and escapes. (www.thex-files.com/epi308.html)
Another property shared by Class I and Class II verbs concerns the interpretation of sentences including AHTY, i.e. the activity denoted by the main verb results in the creation of a hole in some surface. Class I verbs differ crucially from Class II verbs in that the former require the presence of the entire phrase a hole through . Dowty (1991) points out that often there are no clear semantic boundaries between thematic roles such as agent, patient, theme, and instrument. Based on a review of different thematic role hierarchies, Dowty proposes that most predicates exhibit a clustering of so-called proto-agent and proto-patient properties (Dowty 1991: 572). As such, entities that do usually not exhibit agent-like properties (e.g. food in (8), or projectile in (9)) may be construed as agentlike participants in the proper contexts. (cf. Boas 2003a: 243)
15
16
Hans C. Boas
NP whereas the latter do not. Another interesting property of Class II verbs is that they differ from each other with respect to whether they allow omission of the patient argument. In examples (12) and (13) them and the oubliette represent surfaces that can be construed as patient arguments: in each case the surface is affected by the activity described by the main verb. In other words, whereas Class II verbs may occur only with a hole in postverbal position without through NP (cf. (10a)–(13a)), this option is disallowed by Class I verbs (cf. (6a)–(9a)). With this overview of the distribution of different verb classes occurring with AHTY, we now return to the question of how to account for their licensing.
3.
Form-meaning relations among abstract constructions
As briefly mentioned in the introduction, Goldberg’s approach to Construction Grammar (henceforth CxG) maintains that there are independently existing meaningful constructions capable of contributing arguments to a verb’s semantics. Following the notion of the linguistic sign (Saussure 1916), a construction is taken to be a pairing of a particular form with a specific meaning. Goldberg’s definition of a construction is as below. For alternative definitions of constructions, see Croft (2001: 17–21), Fried and Östman (2004: 18–23), and Goldberg (2006: 5–9). C is a CONSTRUCTION iffdef C is a form-meaning pair
such that some aspect of Fi or some aspect of Si is not strictly predictable from C’s component parts or from other previously established constructions. (Goldberg 1995: 4)
For example, Goldberg argues for an independently existing caused-motion construction, which is a pairing of the meaning ‘X CAUSES Y TO MOVE Z by V-ing’ with the corresponding syntactic frame [NP V NP PP]. When the caused-motion construction fuses with the meaning of the intransitive verb sneeze in (1) above, it contributes additional constructional roles to the meaning of the verb, thereby arriving at the final interpretation ‘X CAUSES Y TO MOVE Z by sneezing’. This meaning is expressed syntactically by the [NP V NP PP] frame that typically does not occur with the intransitive sneeze. Similarly in (2) above, the final interpretation of the combined verbal and constructional semantics is ‘X CAUSES Y TO MOVE Z by drinking’. Again, this meaning is paired with a three-argument syntactic frame [NP V NP PP]. Based on Goldberg’s assumption about the interconnection between form and meaning, one might want to assign the sentences in (3)–(5) a caused-motion interpretation ‘X CAUSES Y to MOVE Z by V-ing’ since this particular type
Resolving form-meaning discrepancies
of meaning is paired with the [NP V NP PP] syntactic frame. This view is supported by the fact that just as in (1) and (2) above, (3)–(5) contain activity verbs, which co-occur with three arguments, namely an agent, a patient, and a location. Further, just as in (1) and (2), the postverbal arguments in (3)–(5) are not prototypical object arguments of the verbs push, blow, and knock (cf. *You pushed a hole, *The army blew a hole, and *We knock a hole). However, comparing the interpretations of the sentences in (3)–(5) with the meaning of the caused-motion construction clearly shows that they do not share an ‘X CAUSES Y TO MOVE Z by V-ing’ meaning. This illustrates that there is no one-to-one mapping between form and meaning and that Goldberg’s caused-motion construction does not account for the distribution of AHTY in (3)–(13). An alternative explanation of the data in (3)–(13) could involve another independently existing construction postulated by Goldberg, namely the resultative construction, which – according to Goldberg – is a metaphorical extension of the caused-motion construction. This construction pairs a resultative semantics ‘X CAUSES Y TO BECOME Z by V-ing’ with a specific syntactic frame [NP V NP PP/AP] in order to license postverbal arguments in sentences such as Miriam ran herself to exhaustion or Michael talked himself blue in the face. This means that the resultative construction contributes constructional arguments to the intransitive senses of run and talk, resulting in a ‘X CAUSES Y TO BECOME Z by running/ talking’ interpretation that is represented at the syntactic level by the corresponding [NP V NP PP/AP] frame. But note that an explanation of the distribution of AHTY in (3)–(13) in terms of the resultative construction runs into similar problems as that observed for the caused-motion construction. That is, although the relevant sentences in (3)–(13) exhibit an [NP V NP PP] syntactic frame, their interpretations do not coincide with the ‘X CAUSES Y TO BECOME Z by V-ing’ semantics of the resultative construction. Our discussion of the data suggests two things. First, the current Goldbergstyle inventory of abstract grammatical constructions is not specific enough to deal with the types of data exemplified by (3)–(13). Second, the same syntactic form does not always entail the same interpretations. In other words, the main problem here lies in the fact that although our examples exhibit the same syntactic form [NP V NP PP] and a great deal of semantic overlap, their final interpretations seem to be radically different from the semantics associated with the . Clearly, a caused-motion interpretation would not be assigned to just any [NP V NP PP] syntactic frame as in sentences such as I saw a beautiful woman by the pool. (Thanks to Jaakko Leino who pointed this out to me.) Instead, one would only try to attempt such an interpretation with sentences in which a main verb occurs with a particular preposition (through, in, into, out of, etc.) that could be interpreted as implying caused-motion.
17
18
Hans C. Boas
caused-motion and resultative constructions. This problem calls for a solution that involves pairing the same syntactic form as that exhibited by these two constructions with a more specific type of meaning, a point to which we now turn.
4.
AHTY as an idiomatic construction
Based on Goldberg’s (1995: 4) definition of a construction (see above), it is necessary to attribute the licensing of postverbal arguments in (3)–(13) to a construction different from the caused-motion and resultative constructions. There are two reasons for this: First, the semantics of the two constructions do not apply to (3)–(13). Second, there does neither seem to be any form or meaning component in our grammar (i.e., Fi or some aspect of Si) nor any other construction that would make it possible to predict the distribution of postverbal arguments in (3)–(13) on a principled basis. The fact that the distribution of arguments in (3)–(13) cannot be attributed to other independently existing constructions in the language strongly suggests the existence of an specialized idiomatic construction. In what follows, I discuss in detail the lexical semantic properties of Class I and II verbs which in turn leads me to propose an independently existing AHTY construction capable of integrating the semantics of different verbs.
4.1 The relevance of verb classes Recall that there are in principle two different classes of verbs that occur with AHTY. Class I verbs require the presence of the entire phrase a hole through NP whereas Class II verbs may only occur with a hole while omitting the locative PP. Looking at the distribution of Class II verbs first, the question arises as to why these verbs do not require the presence of a locative PP. The solution involves looking at the implicit meaning of Class II verbs. That is, transitive verbs such as make, create, drill, and dig are not only all transitive verbs, but also contain a meaning component that expresses the making of an object. This meaning component is very similar to the meaning of a class of verbs described by Levin (1993). She observes that most of the verbs belonging to a class she calls “verbs of creation and transformation” take “as one argument an agent that creates or transforms an entity” (1993: 172). Furthermore, Levin notes, “these verbs take what are called ‘effected objects’ – objects brought into existence as a result of the action named by the verb” (1993: 173). Of the seven sub-classes postulated by Levin for the verbs of creation and transformation, members of the sub-class of “Create Verbs” (1993: 175–176) ex-
Resolving form-meaning discrepancies
hibit properties very similar to our Class II verbs. This similarity becomes obvious when we compare their syntactic and semantic properties. For example, Levin demonstrates that the syntactic properties of Create Verbs do not occur in the Material/Product Alternation (Dixon 1991), as (14) illustrates. (14) a. David constructed a house (out of/from bricks). b. *David constructed the bricks into a house.
(Levin 1993: 176)
(15) a. Miriam drilled a hole (through the table). b. *Miriam drilled the table into/from the hole. (16) a. Joe made a hole (through the wall). b. *Joe made a wall into/from the hole.
Class II verbs such as drill and make in (15) and (16) are syntactically similar to Levin’s Create Verbs in three ways: they are transitive, allow omission of the prepositional phrase, and do not occur in the Material/Product Alternation. However, note that members of the two classes are semantically different. That is, whereas the prepositional phrase occurring with Create Verbs in (14a) denotes the material (i.e., bricks) used to create the end product (i.e., a house), there is no such equivalent with Class II verbs. This difference has to do with the fact that the entity created by the activity denoted by Class II verbs – the “effected” object in Levin’s (1993: 173) terms – is not being created by putting together materials such as bricks, for example. Instead, the activity denoted by Class II verbs results in the creation of an opening (i.e., a hole) in the patient object that the agent manipulates. This semantic difference also manifests itself syntactically by the fact that not all of Levin’s Create Verbs are capable of appearing with AHTY (cf. *Collin coined a hole through the window, *Lila derived a hole through the door, and *Michael synthesized a hole through the paper). Besides discussing the Material/Product Alternation, Levin points out that Create Verbs are found neither in the Benefactive Alternation (Green 1974; Cattell 1984; Goldberg 1995) (cf. (17)) nor in the Causative Alternation (Jackendoff 1990; Croft 1998) (cf. (19)). The following examples illustrate that some Class II verbs (cf. (18) and (20)) share these syntactic properties with Levin’s Create Verbs. (17) a. David constructed a house for me. b. *David constructed me a house.
(Levin 1993: 176)
(18) a. Chris drilled a hole for me. b. Chris drilled me a hole. (19) a. David constructed the house. b. *The house constructed.
(Levin 1993: 176)
19
20 Hans C. Boas
(20) a. Chris drilled a hole. b. *The hole drilled.
Our comparison of Levin’s Create Verbs with Class II verbs shows two important things. First, due to the similar semantics they employ to denote the creation of an “effected object”, the two verb classes share the same type of syntactic distribution in a number of alternations. Second, the two verb classes are semantically different with respect to the types of effected objects that are the outcome of the activity denoted by the verb (physical object put together (e.g., a house) vs. creation of an opening in a surface (i.e., a hole)). This semantic difference has syntactic ramifications with respect to the types of verbs that may occur with AHTY. Whereas some of Levin’s Create Verbs such as form, produce, and construct, among others, may occur with AHTY, others such as coin, derive, and synthesize do typically not share this property. It is this semantic and syntactic difference that sets Class II verbs apart from Levin’s Create Verbs. With this overview we now turn to the question of how to account for the distribution of AHTY with Class II verbs.
4.2 AHTY with Class II verbs Since the semantic difference between the two verb classes is mirrored by a difference in syntactic distribution, I propose that Class II verbs form their own specific sub-class of what Levin (1993: 172) calls “Verbs of Creation and Transformation.” The postulation of a separate verb class is based on the distribution of a number of verbs with respect to a very specific syntactic phenomenon, i.e., their distribution with AHTY, and the type of meaning these verbs encode. This classification does not preclude members of this verb class to be classified differently when it comes to other types of constructions. As such, they have the same types of semantic and syntactic properties as Create Verbs discussed above, but differ crucially with respect to the nature of the effected object. As the distribution of postverbal arguments with Class II verbs is not attributable to more abstract constructions (see section three), I suggest that most Class II verbs are conventionally associated with . Ultimately, this difference is one of classification. That is, since only some of Levin’s Create Verbs occur with AHTY, there exists a sub-class within Levin’s Create Verbs. Thanks to Jaakko Leino for pointing this out. . For example, Levin (1993) postulates different verb classes based on the distribution of verbs in more than thirty different syntactic alternations. For an alternative classification of verbs on a frame semantic basis, see Baker & Ruppenhofer (2003) on how Levin’s verb classes match up with verb classes in FrameNet (see http://www.icsi.berkeley.edu/~framenet). See also Croft (2001).
Resolving form-meaning discrepancies
Figure 1. The ‘A hole through Y’ Construction (AHTY)
AHTY semantics. That is, they are associated with a conventionalized instantiation of a specific meaning ‘X CREATES AN OPENING IN Y BY V-ing’ in combination with a specific form ‘a hole through NP’. This conventionalized form-meaning pairing, the ‘A hole through Y’ Construction is illustrated in Figure 1. As the meaning of the ‘A hole through Y’ Construction is paired with a specific form, it becomes possible to automatically interpret any sentence containing the ‘A hole through Y’ Construction provided one knows the meanings of the matrix verb as well as the relevant event participants (i.e., agent (AG) and patient (PT)). However, when it comes to producing sentences including AHTY, matters become more complicated as we have seen in the previous sections. The fact that Class II verbs closely related in meaning do not all occur with the AHTY Construction has led us to assume that they exhibit different idiosyncratic subcategorization patters that can not be predicted on general grounds. I propose to encode the conventionalized meanings of Class II verbs in terms of mini-constructions that inherit the meaning and form from a generalized higher-level schema, namely the AHTY Construction. As such, the AHTY is a schema with a high level of abstraction (in the sense of Langacker (2000)) over a large number of low-level sub-schemas (individual mini-constructions representing the senses of Class II verbs) that all instantiate it. A mini-construction is a form-meaning pairing representing an individual sense of a verb (see Boas 2002a, 2003a). Mini-constructions are extremely specific with respect to the types of form-meaning pairings they represent. Compared with more abstract constructions, mini-constructions differ in their complexity but are in principle the same type of declaratively represented data structure as other types of construc. Note that the meaning part of AHTY contains ‘AN OPENING’ as a description of the type of effected object created by verbs occurring with AHTY. This general formulation is meant to also license examples including a gap or an opening instead of a hole in postverbal position. . Following a bottom-up methodology to semantic description that includes the influence of different types of context on the interpretation of a verb’s meaning (see Cruse 2000; Fillmore & Atkins 2000; and Boas 2003a/2003b), I follow a splitting approach to describing word meanings rather than a lumping approach. A splitting approach has the advantage that it is more finegrained when it comes to the distribution of arguments belonging to different senses of verbs (see Boas 2003a: 160–192).
21
22
Hans C. Boas
tions, exemplifying the notion that there is no strict division between the lexicon and syntax (cf. Fillmore et al. 1988; Jurafsky 1992; Goldberg 1995). Mini-constructions contain detailed semantic, pragmatic, and syntactic information about the types of event participants (semantic arguments) that may occur with a specific sense of a verb. Adapting the main ideas of Fillmore’s (1982, 1985) theory of Frame Semantics, we can capture the distribution of Class II verbs with simplified lexical entries such as the following: (21) a. makeCREATE-SUB : SEM: [ AG ___ OPENING (through PT) ] b. drillCREATE-SUB : SEM: [ AG ___ (OPENING) (through PT) ] (22) RESTR: AG: construable as exerting energy which directly affects PT OPENING: opening construable as part of PT which goes through the PT PT: construable as an object with a surface that is affected by energy emitted by AG
The mini-constructions representing the special creation sub-senses of make and drill in (21a) and (21b) contain semantic and syntactic information about the types of event participants they license. For example, the special sub-sense of drill in (21b) is capable of licensing three event participants. The semantic and pragmatic restrictions in (22) (which apply to both (21a) and (21b)) tell us that the agent licensed by this mini-construction has to be construable as exerting energy that directly affects a patient. The mini-construction also licenses postverbal event participants which denote openings construable as part of patient arguments. The third event participant is a patient, which has to be construable as an object with a surface that is affected by the energy emitted by the agent. As pointed out above, both drill and make allow omission of the patient argument that is embedded in an optional prepositional phrase headed by through (cf. They {made/dug} a hole). The fact that both make and drill allow omission of the patient suggests that the patient contained in the prepositional phrase headed by through is implicitly understood. That is, it is conventionalized world knowledge that whenever a hole is created, there is a surface (the patient) involved that undergoes the activity denoted by the main verb. As a result of the activity, a part of the patient (i.e. the surface) is removed, thereby creating a new entity (i.e. the hole). The use of parentheses indicates that the prepositional phrase headed by through containing the patient event participant is optional. Note that the full constructional semantics of the AHTY Construction ‘X CREATES AN OPENING IN Y by V-ing’ is only evoked in cases in which all three event participants are overtly realized. . Due to space limitations, interactions of mini-constructions such as those in (21) with abstract tense/aspect constructions resulting in examples such as *We made for five hours vs. We dug for five hours are not discussed here.
Resolving form-meaning discrepancies
The reasons for stating an independently existing AHTY construction might not be clear by looking at Class II verbs alone. That is, for Class II verbs, no independently existing AHTY construction is necessary in order to license all of its three event participants (as each conventionalized mini-constructions is already specified for the form-meaning pairing). Due to the high token frequency of Class II verbs, this conventionalized form-meaning pairing affects lexical storage. For example, out of 1112 verbs occurring in the BNC with a frequency of 10 or more, make ranks 9th on the verb frequency list of the BNC (see http://www.comp.lancs. ac.uk/ucrel/bncfreq/ flists.html), create ranks 104th, and dig ranks 592th. Bybee (1988) shows that token frequency has an effect on the way words are stored and processed, as words have varying lexical strengths according to their frequency of use. The fact that high-frequency words have stronger representations in memory makes them easier to access. As such, the AHTY Construction is a generalization arising from the high frequency of mini-constructions that are conventionalized with the AHTY’s form-meaning pairing. This suggests that conventionalized mini-constructions instantiating the AHTY Construction as well as the AHTY Construction itself form a network of constructional schemas, that is, a higherlevel generalization as well as its specific instantiations (for a first step towards a formalization of such networks, see Langacker 2000). For this reason, speakers of English will typically not have problems interpreting non-conventional instances of AHTY. However, the fact that speakers are capable of producing novel AHTY patterns with verbs that are not conventionalized with its form-meaning pairings necessitates the postulation of an independently existing AHTY construction capable of contributing additional arguments to a verb’s semantics (see Section 4.3 for more details). While there seems to be a large degree of conformity between the mini-constructions representing the special senses of make and drill, note that there is a crucial difference between the two mini-constructions in (21a) and (21b). That is, the former allows optional realization of only one event participant whereas the latter allows us to leave out two event participants. Compare the following examples. (23) a. *Lila made through the wall. b. Lila made a hole through the wall. (24) a. Abbey drilled through the wall. b. Abbey drilled a hole through the wall.
This difference in distribution shows that the two mini-constructions representing the two individual verb senses differ crucially as to the number of event participants they license when not instantiating the full AHTY Construction. Dif-
23
24
Hans C. Boas
ferences such as these necessitate the postulation of different mini-constructions such as in (21) representing the individual conventionalized senses of verbs and their specific semantic, pragmatic, and syntactic requirements. Our analysis has the advantage of licensing the full range of sentences occurring with specific subsenses of verbs by employing mini-constructions licensing different combinations of postverbal arguments (including AHTY). At the same time, we have seen that such a detailed lexico-semantic analysis is capable of ruling out unattested sentences. Having accounted for the licensing of AHTY and other combinations of postverbal arguments with Class II verbs, we now turn to the question of how AHTY is licensed with Class I verbs.
4.3 AHTY with Class I verbs Recall that Class I verbs differ crucially from Class II verbs in that the former typically do not occur with a hole as their direct object argument alone (cf. Brian pushed a car/*Brian pushed a hole, Brigid wore a blouse/*Brigid wore a hole), whereas the latter do. This syntactic difference is mirrored by the semantics of the two verb classes. Class II verbs such as drill, make, pierce, create, and dig all have a specific sub-type of a conventionalized creation sense referring to an effected object being created. This is not the case with Class I verbs such as push, knock, wear, and blow. I propose that the distribution of AHTY with Class I Verbs is licensed by unifying an existing conventionalized sense of a Class I Verb with the AHTY Construction. On this view, mini-constructions representing senses of Class I Verbs acquire new meanings because they are capable of being unified with the AHTY Construction. This means that a speaker utilizes her existing grammatical resources in order to create novel sentences. Figure 2 illustrates how the semantics of a Class I Verb is integrated into the AHTY Constructions in order to license sentences such as Joe knocked a hole through the wall. The semantic restrictions for the agent and patient of knock look as follows: AG: construable as exerting energy by striking with a sharp blow; PT: construable as exhibiting a surface. On the left side in Figure 2 we see the AHTY Construction. The mini-construction representing knock is found on the right in Figure 2. It specifies that the agent of this sense of knock must be construable as exerting energy by striking with a sharp blow. The dotted arrow leading from the verb knock to the verb slot of the AHTY Construction means that the semantics of the verb knock (which in this case only contains information about the nature of the agent and patient) is unified with the semantics of the AHTY Construction. Note that the unification of the mini-construction’s semantics only takes place when it is compatible with
Resolving form-meaning discrepancies
Figure 2. Integration of Class I Verbs into AHTY Construction
the semantic specifications regarding the interpretation of the agent and patient of the AHTY Construction. This integration leads to knock taking on the ‘X CREATES A HOLE THROUGH Y by V-ing’ semantics of the AHTY Construction, which in turn is mirrored at the syntactic level by the [NP [V a hole through NP]] frame, resulting in sentences such as Joe knocked a hole through the wall. If nothing else were said about the conditions under which mini-constructions can fuse with the AHTY Construction, our account would vastly over-generate, resulting in sentences such as *Lila sneezed a hole through the peace, *Collin pierced a hole through the ocean, or *The blanket blew a hole through the atmosphere. In order to avoid our account from generating unattested sentences, we need a set of constraints that restricts the integration of Class I Verbs into the AHTY Construction.
4.4 Partial productivity of AHTY The first constraint concerns the nature of the activity denoted by the mini-construction to be integrated into the AHTY Construction. It requires the agent of the activity-denoting verb to be construable as emitting energy (i.e., physical force). This constraint allows for examples such as (25a)–(25c) to be licensed while ruling out examples such as (26a)–(26c).10 . Most of the constraints discussed in this section assume a “normal” world knowledge and are thus subject to override by contextual background information. For example, in a science fiction story in which blankets have the capability to emit energy that allows them to slice up surfaces, sentences such as (26a) are acceptable. For a detailed discussion of the influence of world knowledge and contextual background information on the licensing of non-conventionalized sentences see Boas (2003a: 270–277) and Boas (submitted). 10. Note that the current account does not address figurative usages of AHTY as in Claire {glared/stared} a hole through Natasha. It appears that figurative usages such as these pair the same syntactic form with a meaning different from that of the AHTY Construction. The fact that the productivity of figurative usages is extremely limited (cf. *Claire {looked/saw/watched}
25
26 Hans C. Boas
(25) a. b. c.
Then Googol sprawled wretchedly on a couch, whispering of how a power axe had sliced a hole through the door … (BNC) The bullet blew a hole through the right ear. (www.ipsn.org/cicero1.html) He would draw the line over and over until he wore a hole through the paper. (www.schizophrenia-help.com/april.htm)
(26) a. *The blanket sliced a hole through the door. b. *The book blew a hole through my desk. c. *The peace wore a hole through her wall.
The second constraint restricting integration of Class I verbs with the AHTY Construction requires the energy emitted by the agent to physically affect the patient in such a way that it can be construed as being able to affect the physical integrity of the patient object.11 Examples such as (27a)–(27c) are permitted by this constraint whereas examples such as (28a)–(28c) are typically ruled out. (27) a. You pushed a hole through the crazing. (BNC) b. Benefits being (a) the acid doesn’t eat a hole through the can wall, and (b) tin salts aren’t particularly toxic. (www.intercorr.com/desire.htm) c. Kim burned a hole through the blanket. (BNC) (28) a. *Dawn sneezed a hole through the universe. b. *The water ate a hole through my glass. c. *Christian talked a hole through the chair.
The third constraint on the application of AHTY concerns the nature of the construction’s patient argument. It requires the physical properties of the patient to be construable as exhibiting a surface. This constraint rules out sentences (30a)– (30c) while allowing sentences such as (29a)–(29b).12 (29) a. I’ve burned a hole through Mars’s moon and singed fannies on Pluto. (www.toyraygun.com/books.html) a hole through Natasha) suggests that they are frozen idioms that do not form a basis for productive one-shot extensions based on integration with the AHTY Construction. 11. Note that the force-dynamic relations holding between the event participants need to be construable according to world knowledge. This means that in sentences such as (26) more fragile surface patients such as a window or the steam are acceptable patient arguments because they can be construed as being affected by the agent in such a way that the impact of the agent leads to the creation of a hole. 12. As with previous examples, unacceptable sentences may be rendered acceptable given proper contextual background information that may override the constraints restricting the integration of Class I verbs into the AHTY Construction (see Boas (submitted)).
Resolving form-meaning discrepancies
b. But when these same snails chew a hole through the leaf or stem of the plant to lay their eggs … (www.pondscapes.com/page147.html) (30) a. *Jen burned a hole through the air. b. *Julio blew a hole through the metal plate.
The final constraint restricting integration of Class I verbs into the AHTY Construction requires that the result of the activity denoted by the verb be construable as causing the creation of an opening through the entire patient. This constraint permits sentences such as those in (31), but not those in (32). The examples in (31) differ from those in (32) in that the verbs in the latter do not encode a creation sense because they lack the proper force dynamics. More specifically, when the verbs co-occur with a hole, they are typically not construable as encoding the AHTY semantics. (31) a. b. c.
Woolley took out his revolver and blew a hole through Hawthorn’s dispatch case. (BNC) Using a grapefruit knife push and twist a hole through each potato working crosswise. (BNC) Imagine their foot wearing a hole through the carpet. (offtheboss.com/nofear.htm)
(32) a. b.
A site must be running an NT-based firewall or must provide a hole through the firewall. (www.avolio.com/fw+vpns.html) Thermo-Pond is a unique patent pending pond heater that keeps a hole the ice in backyard ponds for just pennies a day. (www.bestfish.com/thermopnd.html)
Having discussed the constraints limiting the fusion of Class I Verbs with the AHTY Construction, the question remains as to how productive this construction really is. As we have seen, AHTY is not quite as productive as Goldberg’s abstract ditransitive construction (1995: 141–150) and way construction (1995: 199–218). This is because AHTY denotes a semantic space (i.e., the range of semantic possibilities (see Croft 2001)) that is much narrower in scope, pairing the ‘X CREATES A HOLE THROUGH Y by V-ing’ semantics with the [NP [V a hole through NP]] frame. However, within this limited semantic space, AHTY is quite productive as the following examples attest. (33) a. But Oswald doesn’t stop by this idea, he says there’s a shorter way than following the 2-dimensional tour and he eats a hole through the apple. (users.pandora.be/vannoppen/science4.htm)
27
28
Hans C. Boas
b. c. d. e. f. g.
They found that the bee uses its spiky, toothed mouth parts to chew a hole through the side of the corolla. (sci.agriculture.beekeeping, Message ID [email protected]) Perhaps they would’ve sizzled a hole through the gas ball, but would they have hit anything? (rec.music.christian, Message ID [email protected]) Dribbling a hole through the metal isn’t much different from ramming a war hammer spike through it. (rec.games.frp.dnd, Message ID [email protected]) I’ve also put grease fittings in the ball ends, by EDMing a hole through the outer race. (rec.motorcycles.dirt, Message ID [email protected]) I think Tomken was worried about rubbing a hole through it. (rec.autos.makers.jeep+willys, Message ID 85cud8$v9a$1@ tabloid.unh.edu) With patience and persistence, you can piss a hole through a rock. (www.vamp.org/comments-02-97.html)
Typically, the conventionalized senses of verbs like eat, chew, sizzle, dribble, EDM, and piss do not occur with the types of postverbal arguments as in (33a)–(33g). On the account proposed here, these verbs fuse with the AHTY Construction, which provides additional semantic arguments, thereby licensing the [NP [V a hole through NP]] frame. Note that all four constraints on the fusion of verbal and constructional semantics mentioned above are observed. This analysis has the advantage of not having to postulate separate verb senses requiring the realization of both postverbal arguments at the same time. Having accounted for the distribution of AHTY with both Class I and Class II verbs, we now turn to the following question: How does the AHTY Construction fit into the overall architecture of grammar?
5.
The architecture of (Construction) Grammar
One of the main tenets of Construction Grammar is that there exists no strict separation of the lexicon and syntax. This constructional view of language is summarized by Goldberg & Jackendoff (2004) as follows: (34) The Constructional View a. There is a cline of grammatical phenomena from the totally general to the totally idiosyncratic.
Resolving form-meaning discrepancies
b. c.
Everything on this cline is to be stated in a common format, from the most particular, such as individual words, to the most general, such as principles for verb position, with many subregularities in between. That is, there is no principled divide between ‘lexicon’ and ‘rules’. At the level of phrasal syntax, pieces of syntax connected to meaning in a conventionalized and partially idiosyncratic way are captured by constructions. (Goldberg & Jackendoff 2004: 532–533)
Given these assumptions regarding the separation between the lexicon and syntax the question arises as to how the AHTY Construction differs from other types of constructions and whether these differences can be accounted for systematically. Closely related to this issue is the question of how to represent constructional productivity and how to integrate it into an overall constructional model of language. Traditionally, productivity has been at the center of much research in morphology. Drawing on work by Aronoff (1976, 1983), Lieber (1981), Bybee (1995), and many others, Bauer (2001: 211) defines morphological productivity as follows: “The interaction between the potential of a morphological process to generate repetitive non-creative forms and the degree to which it is utilized in language use to yield new lexical items.” Although specific accounts differ in how they prefer to measure the productivity of morphological processes, most analyses agree that a process is productive if the conditions of its applicability do not require the listing of exceptions.13 In other words, the more constraints a construction exhibits on its applicability, the less productive it is.14 For example, within inflectional morphology the third person singular -s in English is said to be completely productive because it is suffixed without exception to the stem of a verb whenever a third person singular subject is present. Within Construction Grammar, Fillmore, Kay & O’Connor (1988) adopt the notion of productivity and extend it to describe and explain the distribution of linguistic units that consist of more than just one word. One example they discuss is the pattern [the X-er the Y-er]: “In spite of the fact that it is host to a large number of fixed expressions, the form has to be recognized as fully productive. Its member expressions are in principle not listable: unlimitedly many new ex13. Aronoff (1976: 36) measures productivity as follows: “The proportion between actual items and potential items generated by a morphological process.” Lieber (1981: 114–115) defines it as: “The number of words a morphological process may apply to.” For an in-depth overview of morphological productivity, see Bauer (2001). 14. This correlation is the same for any grammatical process, pattern, or rule in other frameworks, and is not specific to constructions within Construction Grammar. Compare, e.g., Chomsky’s (1995) Last Resort or Minimal Link Condition that constrain the application of certain checking relations of features.
29
30
Hans C. Boas
pressions can be constructed within its pattern, their meanings constructed by means of semantic principles specifically tied to this construction (1988: 507). Another pattern analyzed by Fillmore et al. (1988) is the let alone construction as in I didn’t get up in time to eat my lunch, let alone cook my breakfast (Fillmore et al. 1988: 531). The let alone construction differs from the [the X-er the Y-er] construction in that it is associated with a number of syntactic, semantic and pragmatic conventions that restrict its applicability. For example, a sentence like #He wasn’t even a commissioned officer, let alone a second lieutenant is typically judged odd because the rank of second lieutenant is the lowest commissioned rank, as Fillmore et al. (1988: 626) point out. As such, the sentence violates an entailment about scales that is implied in the use of the let alone construction. The comparison shows that inn contrast to the [the X-er the Y-er] construction, the let alone construction exhibits more restrictions on its applicability. This means that it is less productive. Fillmore et al. (1988) use these examples, among many others, to demonstrate that “those linguistic processes that are thought of as irregular cannot be accounted for by constructing lists of exceptions: the realm of idiomaticity in a language includes a great deal that is productive, highly structured, and worth of serious grammatical investigation” (1988: 534). Goldberg (1995) limits the productivity of argument structure constructions by positing detailed semantic constraints. The number and granularity of her semantic constraints differ depending on the construction. For example, to limit the ditransitive construction (He baked me a cake) from generating unattested sentences, Goldberg proposes the following constraints: (1) The subject must intend the transfer (1995: 145); (2) The second object has to be understood as a beneficiary, or a willing recipient (1995: 146).15 While these two constraints limit the ditransitive construction from unifying with certain verbs, they are fewer in number and far less specific than the types of constraints posited for resultative constructions (He wiped the table clean, He talked himself blue in the face (1995: 189)): (35) a. b. c. d.
The two-argument resultative construction must have an (animate) instigator argument. The action denoted by the verb must be interpreted as directly causing the change of state: no intermediary time intervals are possible. The resultative adjective must denote the endpoint of a scale. Resultative phrases cannot be headed by deverbal adjectives. (Goldberg 1995: 193)
15. These constraints rule out sentences such as *Joe threw the right fielder the ball he had intended the first baseman to catch (Goldberg 1995: 143) or *Sally burned Joe some rice (1995: 146).
Resolving form-meaning discrepancies
Figure 3. The productivity continuum
The fact that Goldberg’s resultative construction is subject to more fine-grained constraints than her ditransitive construction means that its range of applicability is smaller. In other words, Goldberg’s resultative construction is less productive than the ditransitive construction. In fact, Boas (2002c, 2003a) argues that the productivity of resultatives is even more restricted. Based on an investigation of more than 6000 resultative sentences in the BNC, Boas (2003a) demonstrates that Goldberg’s general semantic constraints are not precise enough to describe and predict the full range of English resultatives. This observation leads him to encode fine-grained semantic, pragmatic, and syntactic restrictions for resultatives in terms of so-called mini-constructions. On this view, each sense of a verb constitutes a form-meaning pairing (so-called mini-construction) specifying the conventionalized selection restrictions on postverbal arguments in resultative constructions. Non-conventionalized resultatives such as Joe sneezed the napkin off the table are licensed by analogical association on the basis of existing mini-constructions that conventionally encode this type of form-meaning pairing (e.g. Joe blew the napkin off the table) (see also Boas (submitted)). Boas’ (2003a) results suggest that the productivity of English resultatives is much more limited as each sense of a verb imposes its own particular selection restrictions on postverbal arguments. Our discussion of constructional productivity has so far resulted in an interesting inventory of grammatical constructions that differ in productivity. Taken together, they can be located along different parts of what I call the productivity continuum. On one end of the continuum, we find constructions such as the [the X-er the Y-er] construction (Fillmore et al. 1988) that are subject to very few constraints and are therefore very productive. On the other end of the continuum we find constructions that are extremely limited in their productivity such as what is commonly referred to as the resultative construction, but which is really a conglomerate of individual mini-constructions. Another example of extremely limited productivity is what Kay (2002) calls patterns of coining. Kay (2002) analyzes phrases such as dumb as an ox, green as grass, dark as night, easy as pie and cold as hell as instances of a pattern ‘A as NP [interpretation: extremely A]’. Based on an extensive review of data he shows that one cannot freely use the patterns to coin
31
32
Hans C. Boas
new expressions. He argues that this pattern is not very productive, which leads him to call it ‘patterns of coining.’ In between the opposite poles of the productivity continuum we find Goldberg’s (1995) ditransitive construction, which exhibits a small number of constraints. Finally, we return to the AHTY construction, which is located between the resultative and the ditransitive constructions on the productivity continuum. As our discussion in Section 4.4 has shown, there are a number of significant semantic constraints that restrict verbs from unifying with the AHTY construction. While these constraints are somewhat more fine-grained than those placed on Goldberg’s ditransitive construction, the AHTY construction nevertheless exhibits a fair amount of productivity (cf. examples (33a)–(33g)) that sets it apart from the less productive resultative construction and Kay’s (2002) [A as NP] pattern of coining. Clearly, future research is needed to investigate the relevance of constructional productivity and its role in Construction Grammar. The goals of this section have been more modest: to determine the relationship of the AHTY construction vis-à-vis other grammatical constructions by locating them at different points along the productivity continuum. These constructions demonstrate not only that productivity is a matter of degree. They also show that research in Construction Grammar must develop a more coherent set of procedures that allow us to measure productivity more precisely in order to arrive at a more integrated view of linguistic usage.
6.
Conclusions and outlook
This paper confirms that sameness in form does not always entail sameness in meaning. I first argued that the [NP V NP PP] frame associated with Goldberg’s caused-motion and resultative constructions does not yield the proper interpretation when it comes to specific types of sentences containing a hole through. This observation led me to search for a more specific construction, whose unique combination of different elements is not predictable on the basis of other constructions. In doing so, I started with an abstract Goldberg-type construction in order to see whether its semantics matched the [NP V NP PP] frame. I finally arrived at a construction that is semantically much more specific. In other words, we have pursued a bottom-up approach to arrive at the proper interpretation of AHTY. Postulating mini-constructions representing conventionalized form-meaning pairings of Class II verbs allowed us to arrive at an abstract AHTY Construction (a higher-level generalization over a number of sub-schemas) that is instantiated by all Class II verbs. For the interpretation and production of non-conventional-
Resolving form-meaning discrepancies
ized instances involving AHTY, we resorted to the AHTY construction supplying Class I verbs with additional constructional arguments. The picture emerging from our discussion is clear: when analyzing a syntactic surface pattern, we should first attempt to pair form with meaning at the most abstract level. Only when we do not arrive at any proper interpretation do we need to look into discovering other types of constructions that are semantically more specific. In doing so, we need to be careful to consider the relationship between independently existing meaningful constructions and individual senses of verbs represented in terms of mini-constructions. For example, a number of recent Construction Grammar analyses argue that Goldberg’s independently existing meaningful constructions are too powerful with respect to the range of possible argument expressions they license (Kay 1996/2002; Morita 1998; Nemoto 1998; Boas 2002b/2003a; and Iwata 2002). As we have seen in this paper, our analysis of AHTY does not run into this problem because the AHTY Construction is not only very specific in its semantic scope (i.e., the types of possible situations it may denote), but its productivity is limited by a number of construction-specific constraints that keep it from generating non-attested sentences. This paper has examined the nature of a very specific type of construction, which I dubbed the AHTY Construction, in order to show how form-meaning discrepancies may be dealt with in Construction Grammar. But there is still much work to be done. To begin with, we need to investigate the semantic relationship between the AHTY Construction and other types of constructional patterns that are closely related in meaning. For example, the interpretation of sentences such as Sascha dug a hole in the snow and Nicole poked a hole into the butter involves similar verbs as those found with AHTY, yet there are slight differences in meaning caused by the interpretation of the prepositional phrases. The question here is whether there might be a single higher level construction that subsumes the three individual constructional patterns including different types of prepositional phrases, thereby allowing for a compositional analysis with an open slot for different types of prepositions encoding different types of situations. Another topic to be investigated in more detail concerns the relation between constructional productivity and semantic specificity of constructions. That is, the AHTY Construction differs from other types of constructions such as the [A as NP] pattern of coining (Kay 2002) or the resultative construction as analyzed by Boas (2003a) in that it is far more productive. If it turns out that there is a direct correlation between a construction’s productivity and the types of semantic spaces it describes, we may well be a step closer to a better understanding of how to go about compiling a complete inventory of constructions of a language. This would include not only high-level argument structure constructions, but also the types of detailed
33
34
Hans C. Boas
mini-constructions discussed by Boas (2003a, 2005b), as well as intermediate level constructions as discussed in Figure 3 above.
References Aronoff, Mark (1976). Word Formation in Generative Grammar. Cambridge, MA: MIT Press. Aronoff, Mark (1983). “Potential words, actual words, productivity and frequency”. Proceedings of the 13th International Congress of Linguists. Baker, Collin F., & J. Ruppenhofer (2003). FrameNet’s Frames vs. Levin’s Verb Classes. Proceedings of the 28th Annual Meeting of the Berkeley Linguistics Society, 27–38. Bauer, Laurie (2001). Morphological Productivity. Cambridge: Cambridge University Press. Boas, Hans C. (2002a). Lexical or Syntactic Selection of Non-subcategorized Nominal Arguments? The case of German ‘Sätzchen’. In I. Rauch & G.F. Carr (Eds.), New Insights in Germanic Linguistics III (9–26). New York: Peter Lang. Boas, Hans C. (2002b). On Constructional Polysemy and Verbal Polysemy in Construction Grammar. Proceedings of the 2000 Western Conference on Linguistics, 12, 126–39. Boas, Hans C. (2002c). On the Role of Semantic Constraints in Resultative Constructions. In R. Rapp (Ed.), Linguistics on the way into the new millennium (34–44). Frankfurt: Peter Lang. Boas, Hans C. (2003a). A Constructional Approach to Resultatives. Stanford: CSLI Publications. Boas, Hans C. (2003b). A Lexical-constructional Approach to the Locative Alternation. Proceedings of the 2001 Western Conference on Linguistics, 13, 27–42. Boas, Hans C. (2005a). From Theory to Practice: Frame Semantics and the Design of FrameNet. In S. Langer & D. Schnorbusch (Eds.), Semantisches Wissen im Lexikon (129–160). Tübingen: Narr. Boas, Hans C. (2005b). Determining the Productivity of Resultatives: A Reply to Goldberg & Jackendoff. Language, 81.2, 448–464. Boas, Hans C. (submitted). Leaking Argument Structures in Construction Grammar”. In M. Fried & J.-O. Östman (Eds.), Pragmatics in Construction Grammar and Frame Semantics. Amsterdam/Philadelphia: Benjamins. Bybee, Joan L. (1988). The Emergent Lexicon. Papers from the 34th Regional Meeting of the Chicago Linguistic Society: The Panels, 421–435. Bybee, Joan L. (1995). Regular Morphology and the Lexicon. Language and Cognitive Processes, 10.5, 425–455. Cattell, Ray (1984). Syntax and Semantics 17: Composite Predicates in Russian and English. New York: Academic Press. Chomsky, Noam (1965). Aspects of the Theory of Syntax. Cambridge: MIT Press. Chomsky, Noam (1995). The Minimalist Program. Cambridge, MA: MIT Press. Croft, William (1998). Event Structure in Argument Linking. In M. Butt & W. Geuder (Eds.), The projection of arguments (21–64). Stanford: CSLI Publications. Croft, William (2001). Radical Construction Grammar. Oxford: Oxford University Press. Cruse, Alan (2000). Aspects of the Micro-structure of Word Meanings. In C. Laecock & Y. Ravin (Eds.), Polysemy (30–51). Oxford: Oxford University Press.
Resolving form-meaning discrepancies
Dixon, R.M.W. (1991). A New Approach to English Grammar, On Semantic Principles. Oxford: Oxford University Press. Dowty, David (1991). Thematic Proto-Roles and Argument Selection. Language, 67, 547–619. Fillmore, Charles J. (1982). Frame Semantics. In Linguistic Society of Korea (Ed.), Linguistics in the Morning Calm (111–38). Seoul: Hanshin. Fillmore, Charles J. (1985). Frames and the Semantics of Understanding. Quadernie di Semantica, 6.2, 222–254. Fillmore, Charles J. & B.T.S. Atkins (2000). Describing Polysemy: The Case of ‘Crawl’. In Y. Ravin & C. Laecock (Eds.), Polysemy (91–110). Oxford: Oxford University Press. Fillmore, Charles J., P. Kay & M.C. O’Connor (1988). Regularity and Idiomaticity in Grammatical Constructions: The case of let alone. Language 64, 501–38. Fisher, Cynthia, G. Hall, S. Rakowitz & L. Gleitman. 1991. When it is Better to Receive than to Give: Syntactic and Conceptual Constraints on Vocabulary Growth. IRCS Report 91–41. Philadelphia: University of Pennsylvania. Fried, Mirjam & Jan-Ola Östman (2004). Construction Grammar. A thumbnail sketch. In M. Fried & J.-O. Östman (Eds.), Construction Grammar in a Cross-Language Perspective (11– 86). Amsterdam/Philadelphia: Benjamins. Gleitman, Lila (1992). Presentation at the 15th International Congress of Linguists. Quebec City, Canada. Goldberg, Adele E. (1995). Constructions: A Construction Grammar Approach to Argument Structure. Chicago: University of Chicago Press. Goldberg, Adele E. (2001). Patient Arguments of Causative Verbs can be omitted: The Role of Information Structure in Argument Distribution. Language Sciences, 34/4–5, 503–24. Goldberg, Adele E. (2006): Constructions at Work. Oxford: Oxford University Press. Goldberg, Adele E. & Ray Jackendoff (2004). The English Resultative as a Family of Constructions. Language, 80.3, 532–568. Green, Georgia (1974). Semantics and Syntactic Regularity. Bloomington: Indiana University Press. Iwata, Seizi (2002). Does MANNER Count or Not? Manner-of-Motion Verbs Revisited. Linguistics, 40, 61–110. Jackendoff, Ray (1990). Semantic Structures. Cambridge: MIT Press. Johnstone, Barbara (2000). Qualitative Methods in Sociolinguistics. Oxford: Oxford University Press. Jurafsky, Dan (1992). An On-line Computational Model of Human Sentence Interpretation: A Theory of the Representation and Use of Linguistic Knowledge. Doctoral dissertation, University of California at Berkeley. Kay, Paul (1996). Argument Structure: Causative ABC Constructions. Manuscript. University of California at Berkeley. Kay, Paul (2002). Patterns of Coining. Plenary Talk at the Second International Conference on Construction Grammar, University of Helsinki, Finland, September 2002. Kennedy, Graeme (1998). An Introduction to Corpus Linguistics. London: Longman. Kilgarriff, Adam & Gregory Grefenstette (2003). Web As Corpus. Computational Linguistics, 29.3, 1–15. Landau, Barbara & Lila R. Gleitman (1985). Language and Experience: Evidence from the Blind Child. Cambridge, Mass.: Harvard University Press. Langacker, Ronald W. (2000). A Dynamic Usage-Based Model. In S. Kemmer & M. Barlow (Eds.), Usage-based models of language (1–63). Stanford: CSLI Publications.
35
36
Hans C. Boas
Levin, Beth (1993). English Verb Classes and Alternations. Chicago: University of Chicago Press. Lieber, Rochelle (1981). On the Organization of the Lexicon. Bloomington, IN: Indiana University Linguistics Club. Macfarland, Talke (1995). Cognate Objects and the Argument/Adjunct Distinction in English. Ph.D. dissertation, Northwestern University, Evanston, Illinois. Morita, Pat (1998). Some Notes on Prepositional Resultatives. Tsukuba English Studies, 17, 319–40. Nemoto, Noriko (1998). On the Polysemy of Ditransitive save: The Role of Frame Semantics in Construction Grammar. English Linguistics, 15, 219–42. Petruck, Miriam (1996). Frame Semantics. In J. Verschueren, J.-O. Östman, J. Blommaert & C. Bulcaen (Eds.), Handbook of Pragmatics (1–13). Amsterdam/Philadelphia: Benjamins. Saussure, Ferdinand de (1916). Cours de linguistique générale. Paris: Payet 1973.
chapter 3
Language change, variability, and functional load Finnish genericity from a constructional point of view Pentti Leino and Jan-Ola Östman University of Helsinki
1.
Overview
In this study we look in detail at the feasibility of explaining language change and variability in language with the help of the notion ‘grammatical construction’. The study shows that only by extending the constructional view of grammar into a full-fledged constructional model of language is it possible to account for diversity and language change in its entirety. A full understanding of the mechanisms involved in language change requires syntactico-semantic analyses to be integrated with discourse and areal information. In addition, language has to be approached as a system that is constantly changing. The analysis focuses on an emergent construction in Finnish where the second person singular form is being used as a generic form – cf. You just drive in English. The study shows that in the historical development of Finnish impersonal constructions, the direction of change has been influenced by two forces pulling in slightly different directions. Language change and variability are orchestrated through an intricate interplay of (i) (emergent) form-meaning constellations as constructions, and (ii) functionally motivated, conceptual discourse patterns. In an attempt to get closer to explaining language change, the study not only attends to the theoretical issues involved in representing language change, but it also addresses aspects of the sociology of language change. . We would like to thank Jaakko Leino, Ilona Herlin, Mirjam Fried and an anonymous referee for comments on earlier versions of this study.
38
Pentti Leino and Jan-Ola Östman
2.
Constructions and language change
We use the term ‘grammatical construction’ in the sense the term was initially used and has subsequently been developed by Charles Fillmore and his students within Construction Grammar (CxG; for overviews, see Fillmore 1998; Fried & Östman 2004). In addition to making use of the Construction Grammar framework, we also draw on insights from Cognitive Grammar (CgG; cf. in particular Langacker 2005; Barlow & Kemmer 2000). Crucially for our purposes, within both frameworks scholars have long recognized the need to make reference to, and to account for discourse-level phenomena (cf. Langacker 2001; Östman 2005; Fried & Östman 2005). Both frameworks have also recognized the need to account for variability and language change, although the mechanisms for language change in a constructional approach have so far received modest attention (but cf. J. Leino & Östman 2005). The study builds on a number of previous studies on grammatical change in Finnish, notably Herlin & Visapää (2005), P. Leino et al. (2001), and J. Leino (2003). We take the following four general points as our starting point in regard to language change. First, we align ourselves with the view that cognitive aspects like expansion, analogy, and subjectification are at the heart of language change, but we maintain that they only take us part of the way towards explaining language change. Secondly, both sentence-internal and sentence-external sources have to be taken into account; consequently, for a constructional approach to be viable, it has to acknowledge both constructions as (sentence-level) form-meaning constellations and constructions as functional, discourse-level patterns. Thirdly, context also has to be accounted for and among contextual causes for language change we specifically stress the importance of language contact and ensuing ideological and cultural issues, and the importance of the role of the individual. Finally, we note that the speed with which linguistic change takes place is not constant: e.g., all grammatical changes are not necessarily slower to take place than lexical changes; and with the recent advent of mass communication, linguistic change – on all ‘levels’ – can be very rapid indeed.
. Within CxG, the work by Knud Lambrecht has in certain respects gone beyond the sentence (cf. Lambrecht 1994; Michaelis & Lambrecht 1996), but the starting point for these studies has continued to be the single sentence, whereas in particular Östman (2005) has taken conventions in discourse as a starting point. Fillmore’s work on Frame Semantics (e.g., 1982) and within FrameNet (cf. Johnson, Fillmore et al. 2001) recognizes both sentence-level frames and more general, situational frames.
Finnish genericity
3.
The empirical issue
During the last couple of decades, there have been signs that a new form-meaning constellation is about to appear in spoken Finnish; and in the last couple of years, the acceleration of the emergence of this construction has been very rapid. A simple example of this structure is given in (1). Informally, we could say that what are known as ‘impersonal passives’ in Germanic languages have in this structure received a counterpart in Finnish. In English, an ‘impersonal passive’ would be an active construction with a generic subject like people, they, one, you. The example in (1) could – in the appropriate context – be translated as either ‘You just drive’ or as ‘One simply just drives’. (1) Sä vaan ajat. you-sg just drive-2sg
At first glance, we seem to be faced with a well-known form (the 2nd person singular pronoun as subject used together with active 2nd person singular verb morphology) that is (in some cases, ambiguously) being used with a new meaning; or, conversely, a familiar meaning (variously known in Finnish grammar as passive, impersonal, generic, indefinite-person, or avoidance of responsibility) is being dressed up in a different structure. However, since not just any combination of a 2nd person singular pronoun as subject with active 2nd person singular verb morphology can be given a passive/impersonal/generic/indefinite meaning, nor can any passive/etc. meaning be expressed with a 2nd person singular pronoun together with active 2nd person singular verb morphology, the emergent constellation is in need of a unified description and explanation. In this study we give a descriptive account of the emergent form-meaning constellation as a construction. Since the construction is a fairly recent one that is getting itself established in Finnish, it is crucial not only to account for its present conditions of usage and variations in form, but its historical development and the reasons for the emergence of the construction need to be explicated. In addition, the place of the ‘new’ emergent construction in the Finnish constructicon as a whole need to be described and explained.
. The constructicon in CxG is the inventory of all the abstract constructions that are needed to license (well-formed) constructs (i.e., expressions) in a language.
39
40 Pentti Leino and Jan-Ola Östman
4.
Conditions of usage
Seppänen (2000) shows that the generic use of the 2nd sg itself is nothing new in Finnish; however, the form-meaning constellation of the new construction is slightly different from the corresponding forms found in dialects and in 19th century texts. Finnish has three ways of indicating that an activity is performed by a personas-pronoun: the personal pronoun itself can be left out, in which case person is marked only with a suffix on the verb (2a); or, person marking is done both with a pronoun and with a verbal suffix (2b–c), in which case speakers have to choose which paradigm to follow (cf. J. Leino & Östman 2005), the standard paradigm (2b; of which 2a is an alternative in this paradigm), or what is seen as a more informal paradigm (2c). The three variants in (2) can all mean ‘You drive’ – the 2nd sg suffix is -t, and the pronoun is either the more formal sinä or the more informal sä. (2) a. Ajat. b. Sinä ajat. c. Sä ajat.
Whereas the traditional way to use 2nd sg forms to indicate an indefinite or generic agent in sayings and in the Finnish dialects in general, is in the fashion of (2a), the newly emerging structure strongly favors the 2c-type. Furthermore, whereas the traditional way of leaving the referent unspecified had a more generic meaning and referred to ‘anyone who does the activity specified’, the newly emerging structure has a more affective meaning with the speaker seemingly attempting to include the addressee in the speaker’s own, lived experience. Here we thus see that when a form (2nd sg) we encounter elsewhere in the grammar gets combined with a meaning (generic, indefinite) which we also encounter elsewhere, the output constellation does not necessarily constitute only the sum of the parts already established in grammar, but the new form-meaning combination may become conventionalized as a construction per se. Still, it is clear that the form of the new construction is related to – and inherits, in CxG terms – the garden-variety 2nd sg form; and the meaning of the new construction is related to – and possibly even inherits – a general indefinite-generic meaning. A constructional approach to grammar can adequately account for such shifts in the function of form-meaning constellations, and at the same time adhere to a general cognitive principle of categorization, according to which one form has one meaning – at some level. In the case at hand, we see that a particular variant . Cf. the naked -t forms in the following sayings: Niinkuin petaat myös makaat ‘You sleep according to how you make your bed’; Mitä teet, niin tervan kanssa ‘Whatever you do, use tar’.
Finnish genericity
of the possible forms (2c of 2) is picking out one sense of the available meaning network, and together the meaning and the form produce a unique conventionalized linguistic unit, a construction in terms of CxG.
5.
The inception, emergence, generalization, and acceptance of a construction
The construction exemplified in (1) has been variously called ‘the sä passive’, ‘the imperialistic passive’, and ‘the Häkkinen passive’. Since syntactically defined periphrastic (and morphological) passives have a very limited use in Finnish, it has been customary in Finnish grammar to talk about the indefinite -TA- constructions (cf. example (8) below) as ‘passives’. Thus, in Finnish, the semantic criterion for suppressing the agent has been the common denominator for such ‘passive’ constructions. Although syntactically speaking the majority of these are not passives, we will here continue the Finnish tradition, and refer to the construction under study (exemplified in (1)) as ‘the sä passive’. There are two potential sources for the sä passive in Finnish: one is languageinternal, the other is to be located in external contact influence. As we have seen, the prerequisites for this construction to develop have for long been available in spoken Finnish, in sayings, and in the Finnish dialects (which, in turn, may suggest the indefinite-person constructions in Russian as a source; cf. Kuusi 1977). However, as we have also seen, the sä passive is neither in form nor in meaning identical to the traditional Finnish pattern. Still, one source for the construction – what probably makes it accessible as a Finnish construction (albeit frowned upon by many as slang) – is the possibility that it could have developed as a languageinternal extension of the system. The second reason why it has been possible for this construction to develop is the influence of Germanic languages – or, more generally: the influence of IndoEuropean languages – on Finnish, and the influence of English in particular. In fact, Finnish has for long been said to be going through a gradual shift from structurally being an agglutinating language into more and more becoming an analytic, isolating language, including becoming more of a subject-prominent language. At present, the spoken language is more isolating, whereas the written, normative language is more agglutinating. The change in the direction of becoming more . The reasons for the latter two denominations will become clear later on. . For discussions, see Östman (1981), Shore (1986). Hakulinen & Karttunen (1973) deals with generic structures in Finnish in general and Leinonen & Östman (1983) gives a general overview of semantic factors affecting passive constructions.
41
42
Pentti Leino and Jan-Ola Östman
isolating is not only due to recent Anglo-American and EU trends, but has a long history, since Finland is part of the Circum-Baltic linguistic area, where most of the other languages are Indo-European languages. (Cf. Raukko & Östman 1994; Östman & Raukko 1995; Dahl & Koptjevskaja-Tamm 2001.) We suggest that the combinatory effect of these two prerequisites has laid the ground for the development of the sä passive as a construction in its own right in present-day Finnish. Further support for this view can be found in history. Finland was part of Sweden until 1809, and an autonomous Grand Duchy of Russia until 1917, when it declared its independence. Throughout its independence, Finland has had two national languages, Finnish and Swedish. However, the traditional indefinite-person construction in Swedish is not one involving the 2nd sg, but rather – as in German – a construction with man as the subject. Thus, there is clearly something new and ‘Anglo-American’ in the use of the sä passive, so much so that it has been termed the imperialistic passive, partly due to its presumed source as an outcome of AngloAmerican globalization. Irrespective of the details of when the sä passive first entered Finnish, it is clear that language contact as an external source played a role. Although Seppänen (2000) notes that the sä passive can sporadically be found in spoken Helsinki Finnish in the 1970s (cf. Paunonen 1995), it is not until very recently that the construction has caught on in earnest. The one source for the frequent use of the sä passive is claimed to be the Formula-1 driver and world champion Mika Häkkinen, who used the construction abundantly in sports interviews. The possibility that one influential individual can be the cause of a major linguistic change is nothing new, but it is still seen as an exceptional case. What is even more astonishing is the speed with which such a structural linguistic change can take place. The sä passive caught on in record time and was perpetuated in daily interviews with Häkkinen and others on TV and in the tabloids’ news and sports stories. In fact, the role of Mika Häkkinen is so pervasive that the construction has also been referred to as ‘the Häkkinen passive’. Example (3) is a typical example of a Häkkinen passive. (3) Kun sä ajat formulaa, sä et ajattele mitään muuta. when you drive formula you don’t think anything else ‘When you’re driving a Formula-1 car, you don’t think about anything else.’
One characteristic of the sä passive is that there is an element of displacement from the here-and-now associated with its use, to the extent that it is often construed as potentially non-factual, or even hypothetical: even more typical than the locution . Cf. e.g., Hinton (1979) on how a phonological change in Mojave and Northern Pai can be traced back to a journey that Chief Irataba once made.
Finnish genericity
‘when you drive’ in (3), is a construal of the form Jos sä ajat ... ‘if you drive’. If there is no explicit marker of spatial or temporal displacement, like the ‘when’ and ‘if ’ conjunctions, the sä-passive utterance typically contains an affective adjunct, like the vaan ‘simply’ in (1), or a discourse marker or pragmatic particle, to indicate the special stance of the expression. If there is no marker, as in (2c), it is the particular context which makes it clear that a generic interpretation is called for.
6.
The construction in relation to other constructions
The sä passive is a member of a family of constructions. There is an abundance of ways in which the agent can be left out or unspecified in Finnish, but these are seldom interchangeable – each has its own specific task and meaning. Such subjectless structures instantiate a general semantic domain which could be called an ‘Avoidance of responsibility pattern’, or more simply, as we will do here, the ‘Genericity pattern’. This denomination refers both to an extension of the kinds of situational frames which make up the backbone of Frame Semantics – in this case we are talking about abstract conceptual frames – and it refers to the kind of discourse patterns that have recently been suggested (Östman 2005) to play a role as establishing the discourse constraints for constructions in CxG. Our point here is that in addition to constructions as form-meaning constellations, we also see CxG as needing to make reference to functionally defined patterns.10 The traditional Finnish way of expressing Genericity is to use the 3rd sg verb form without a concomitant 3rd person pronoun11 as in (4). (For extensive dis-
. Notice in particular the following subjectless examples – all meaning ‘if the door opens/if one opens the door’ – which form a gradient scale of increasing ‘agent responsibility’ the further you move downwards in the list: Jos ovi avautuu, … (reflexive) Jos ovi avataan, … (indefinite person, traditionally known as the ‘passive’) Jos avaa oven, … (traditional generic) Jos sä avat oven, … (the sä passive) . The notion ‘avoiding responsibility’ is a traditional way of referring to the function of ‘passive’ constructions; unfortunately, it suggests an element of intentionality, which is not necessarily present as a condition of usage. Thus, we here prefer to use the term ‘genericity’, a nominalization of generic. 10. Östman (2006) refers to a construction as a form-meaning-function constellation. 11. In Finnish, although the personal pronouns are optional for verbs in the 1st and 2nd person, they are obligatory (except in cases of clear sentence-internal ellipsis) for verbs in the 3rd person in non-generic sentences.
43
44 Pentti Leino and Jan-Ola Östman
cussions of this construction, see Hakulinen & Karttunen 1973; Vilkuna 1992; Hakulinen et al. 2004: 1253–1300.) (4) Kun ajaa formulaa, ei ajattele mitään muuta. when drive-3sg formula not-3sg think anything else ‘When one is driving a Formula-1 car, one doesn’t think about anything else.’
This general Finnish Generic construction lacks an overt subject,12 although the subject is discernable in the verb morphology. The meaning of constructs as realizations of the Generic construction is one of indefinite agency which can get specified only through the cotext or in a situational context. Thus, in comparison to the Generic construction, the sä passive of (3) works in a somewhat opposite manner: It starts out, overtly, with a specific 2nd sg subject, and – in context – becomes indefinite, to indicate ‘anybody’. The example in (4) is typical of structures with zero overt subjects. Now, with the emergence of the sä passive and other indefinite-person constructions (cf. below) Finnish is gradually becoming a subject-prominent language. That is, the emergence of the sä passive is not an independent phenomenon; for instance, and as further support of subject prominence in Finnish, the 3rd person neuter pronoun se ‘it’ is developing into an expletive subject in Finnish. These empirical findings suggest that changes with considerable impact in a language are not random, but that there is typically some element of ‘drift’ crucially present. Two points need to be stressed here: First, the generic structure itself is nothing new in Finnish – the generic avoidance-of-responsibility meaning (cf. above), as tied to particular structures, is available in Finnish. What we see now is that Finnish accepts yet another way (in terms of the sä passive) of expressing genericity. This type of language change in terms of extension is well recorded in languages world-wide. The second point is that especially in bilingual contact situations, constructions – or what seem to be elements of constructions in one language – easily move from one language into another, thus creating variability in the target language. Whatever is the cause of language drift, it is not a priori possible to keep its influence within a particular language. This is a particularly pertinent point to be made for the Finnish situation, where Indo-European and Finno-Ugric languages (in particular, Swedish and Finnish) have been in close contact throughout his-
12. In Finnish grammar, the missing subject is called ‘a generic zero’.
Finnish genericity
tory. As Finnish is changing typologically, it has received material and the general inception for establishing new constructions from Germanic languages. The example in (5a) is close to the type we find in (4); however, here we have an impersonal verb, väsyttää ‘it tires (me)’, in the 3rd sg, meaning simply – if said by itself as in (5a) – ‘I’m tired’. When this is construed in a non-factual or otherwise modalized structure – a ‘when’ or ‘if ’ structure – it can be given a generic interpretation: (5b) can mean ‘If you are tired / If one is tired, ...’. We thus see that the boundary between the construction instantiated in (5b) and that instantiated in (5c), with the 2nd sg pronoun (sä in the partitive: sua) is not easy to draw. That is, since impersonal verbs always take 3rd sg morphology, only context can decide whether (5c) is an example of the Generic sä passive construction (cf. (4)), or whether it is, in fact, an example of an ordinary 2nd sg directional – cf. ‘If you’re tired (, please just lie down)’. (5) a. Väsyttää. b. Jos väsyttää, ... c. Jos sua väsyttää, ...
By the same line of reasoning, an ‘if ’-clause of the type exemplified in (6), meaning ‘if you drive’ (cf. 2c), can be seen as related to – as inheriting – either a default transitive construction (cf. 2a–b) or the sä passive. (6) Jos sä ajat, ... if you drive
A general characteristic of the fairly large set of impersonal constructions in Finnish is that they are not accompanied by a heavy functional load. They take on new tasks easily, and their respective meanings and functions waver within the group of impersonal constructions. Historically speaking, their behaviors have frequently required a reanalysis of the impersonal system as a whole. But what is constant is the centrality of the Genericity pattern in Finnish grammar. In addition to the sä passive, the construction exemplified in (7) has also fairly recently upset the system. This construction can only be realized with a first-personplural subject; it has the same verb morphology as what is often referred to as the ‘passive’ construction in Finnish13 (exemplified in (8)), but (7) has active meaning. (7) Me ajetaan formulaa. we-nom drive-TA-n formula-part ‘We’re driving a Formula-1 car / in a Formula-1 race.’ 13. Syntactically, however, this is not a passive in the sense that an active object or oblique is promoted to subject status; cf. Östman (1981).
45
46 Pentti Leino and Jan-Ola Östman
(8) Formulaa ajetaan (kesäisin ympäri maailmaa). formula-part drive-TA-n (in-summers around world) ‘In the summertime there are Formula-1 races all around the world.’, lit. ‘Formula is driven in the summertime around the world.’
The constructs in (7) and (8) need to be related to different constructions, albeit that they are related through certain characteristics they have in common. We will not deal with this issue here. What we should note, however, is that (7) and (8) are neutral with respect to factuality, whereas the sä passive is more non-factual (cf. the discussion above) and more specific in meaning, and thus differently motivated. The rise of constructions that can license constructs like (7) and (3) are different attempts to cope with responsibility of action: in both, responsibility and accountability are downplayed, but in linguistically different ways. In (7) the verb morphology is impersonal/‘passive’, to which a personal agent as subject is added: responsibility is accepted, but the activity to some extent takes place on its own.14 In (3) the verb morphology is active – somebody is responsible for the activity – but the agent as subject is a generic you, equivalent to a responsibility-avoiding one. Rather than taking ‘systems’ like the Genericity pattern for granted as such, we see the amoeba-like expansion and variation in constructions as motivated by changes (e.g., reshapings) in the conceptualizations of speakers and/or in the inherent variation in speech communities. Ensuing cognitive fluctuations are motivated by external, e.g., ideological, changes in society, either due to contact influence, a reshuffling and reorganization of genres and text types in a community, or a political-economic trend in society (e.g., globalization) or in a subculture. Chapter 7 summarizes the characteristics of the sä passive in Finnish.
7.
The sä passive as a construction
The bulleted list below specifies the characteristics of the sä passive that will have to be taken into account when describing, explaining, and formalizing the construction. Some of these have been mentioned or hinted at above, others are introduced here.
14. We say ‘to some extent’, because expressions like (7) have been further grammaticized and now occupy a regular slot in the verb morphology for first person plural in the paradigm for spoken Finnish (cf. J. Leino & Östman 2005).
• •
• •
•
•
•
Finnish genericity
The form of the 2nd person singular pronoun is sä, not sinä. The form sä is not used as an address term. We cannot interpret a relational sentence like Sä oot hullu ‘You are crazy’ as generic.15 (In addition, the use of the present tense here de-semanticizes the expression into a general typenaming event.) The verb is non-specific, i.e., it does not refer to any particular instance of an activity. The combination of a 2nd sg pronoun with a bare present tense does not occur too frequently. (Cf. example (1).) Even in English one does not typically say – with an indicative meaning – You are driving (and even less often, You drive); minimally, one would qualify such a statement in order to make it relevant, e.g. (said over the phone): So, you are, in fact, driving right now. The construction involves some degree of hypotheticality or displacement from the here-and-now. In the discussion above, we have seen the use of conjunctions like kun ‘when’ and jos ‘if ’, and particles like vaan, which all function as space builders in Fauconnier’s (1985) sense and thus specify the conditions for the generic interpretation; the space builders can be anything that takes you beyond the here-and-now. In Frame-Semantic terms, we might want to talk about general abstract sense-frames like ‘genericity’16 in addition to local, predication-related frames, and situational frames (as domains). In order to differentiate between the various Genericity constructions, we need attributes which will specify whether the referent of sä includes the speaker (with the values + [yes], – [no], and 0 [uncommitted]); and whether it includes the addressee (+, –, 0). The referent of the sä passive includes both the speaker and the addressee. In addition, we need attributes specifying the point of views of the speaker and of the addressee. Another way to look at non-factuality is to see it in terms of dialogical sense-making. Ojajärvi (2000) is responsible for calling the sä passive an imperialistic passive (cf. above), but for him the reason for the term was not primarily because of globalization via Anglo-Americanization, but the way the speaker-‘I’ behaves imperialistically towards his or her addressee, ‘speaking for’ Others in a very non-dialogical manner. The sä passive becomes
15. Here oot is the 2nd sg form of the copula olla ‘be’. Native speakers tend to disagree with respect to the extent to which such relational sentences can be interpreted generically inside an ‘if ’-clause. They are hard to construe as generic in cases like Kun sä ajat formulaa, sä oot hullu ‘When/If you are driving in a Formula-1 race, you are crazy’, but easier in cases like Jos sä oot tarpeeksi hullu, sä voit vaikka ajaa formulaa ‘If you are crazy enough, you may even drive in a Formula-1 race’. 16. Other such frames would be ‘ambiguity’, ‘ambivalence’, and domains like ‘color’.
47
48 Pentti Leino and Jan-Ola Östman
a manifestation of a competitive market-economy stance. And it does seem to be the case that the areas where the sä passive is most often used are competitive sports and business. This may remain a defining characteristic of the construction, or it may well be a temporary, emergent aspect of the function and use of the construction, which was needed in order to carve out a place for it in relation to other Genericity constructions, and which will gradually fade away. Be that as it may, in the formalization of the conditions of use for the construction in present-day Finnish, this aspect needs to be included. Thus, the sä passive here takes the point of view of the speaker but not the point of view of the addressee. The speaker of a sä passive asks you to step into his or her role and look at the issue under consideration from his or her point of view. These are the main characteristics of the sä passive that will have to be specified in attribute-value terms in a full formalization of the construction.
8.
Theoretical implications for language change
The way we propose to view the sä passive in Finnish as a construction offers interesting insights into the manner in which the reorganization of grammatical constructions work in general in language. We have indirectly illustrated how language change takes place according to what have traditionally been seen as sense developments or semantic developments, but which should rather be talked about in terms of cognitive principles: expansion, abstractification, and subjectification. In addition, we have specifically wanted to stress that there very often needs to be an external trigger for these processes of change to take place – at least in order to speed up a linguistic change. Whether such external sources for change can override the cognitive plausibility and predictability of linguistic change is unclear, since such overridings can easily be reanalyzed in cognitive terms. One of the most pervasive factors in language change is extension through analogy. In the present case, the sä passive has seemingly started out in a very specific context of non-factual modality (cf. the discussion in connection with example (3)), from where it has been extended in accordance with conversational implicatures (cf. the discussion in connection with the examples under (5)); finally, this extension has become conventionalized.17 17. Cf. the -TA- passive (8) and its extension into the 1st pl through its use as an imperative: Mennään kotiin ‘Let’s go home’. Here, historically, we presume that the agent referent has become de-individuated, which in turn has given rise to a generic interpretation.
Finnish genericity
But in addition to such internal causes, there has been a clear external source (connected to a particular (sub-)culture) to trigger the generalization of the construction. Changes in the concomitant frame set-up have influenced, and become incorporated into, the constructional make up. Here we have only mentioned in passing the importance of language contact for language change in relation to the emergence of the sä passive. The importance of this aspect cannot be overestimated; it affects all ‘levels’ of language and structuring.18 In the case of the sä passive, we have a situation where the language largely lacks an independent passive construction definable on syntactic grounds, but it clearly has the ‘semantic category’ passive (cf. the notion of indefinite person) – which we have here referred to as the Genericity pattern. At present we seem to be at a stage where the ‘passive form’ (-TA-) is increasingly being used with ‘active meaning’ (as in the Me ajetaan type of examples, cf. (7)), with one consequence being that other ‘active forms’ with generic meaning (cf. (5) above) have to reshuffle their area of application and use in order to make space for the newcomers. We are not saying that this is the reason why the sä passive has developed, but this circumstance is in concord with the emergence of the sä passive. We have also in this study wanted to stress the importance of the individual in language change. Mika Häkkinen and even before him another Formula-1 driver, Keke Rosberg, started to use the sä passive abundantly, almost as an idiosyncratic or sub-group trace, from which it spread into the news and sports magazines. However, although the sä passive might be said to have its roots in Häkkinen, it has become decontextualized with the effect that although Häkkinen is not driving in the Formula-1 anymore, and thus is not being interviewed as eagerly, this has not resulted in a decrease in the use of the sä passive.19 A further point we want to stress is the varying speed of language change. We are all familiar with slowly occurring linguistic changes. Here we have seen an example of a very rapid change. The extraordinary thing is not, however, that rapid changes occur – rapid changes are the norm in lexical borrowings. What is special with the sä passive is that a structural change which affects central aspects of Finnish grammar could take place so rapidly. The reasons, as we have seen, are twofold: the language already, internally, had the ingredients available for such a 18. For a recent study of the far-reaching implications of the historical coalescence of the accusative ‑m and the genitive -n in Finnish, see P. Leino et al. (2001: 491–508). In that case, the external cause is simply phonetic similarity. 19. One aspect that we have not dealt with here at all, but which is important from the point of view of language change is the polyphony at play in connection with the use of the sä passive. It may well be that one of the reasons why the construction caught on so rapidly was that it was used ironically by people in order to make fun of the way Häkkinen talks.
49
50
Pentti Leino and Jan-Ola Östman
change to take place, and the particular external source was so pervasive due to the public accessibility of Formula-1 drivers (on television and in newspapers, and through physical performances in Finland) and the exceptional role Finns have played in these races during the last decade. More generally, we want to stress that the speed of linguistic change is not constant – in particular, due to the ever increasing impact of mass communication, language change today takes much less time than during earlier periods of times.
9.
Language change in terms of constructions
The traditional way of seeing linguistic change is some variety of the following schema in terms of stages of gradual change: form 1 meaning 1
>
form 1 meaning 2
>
form 2 meaning 2
What we want to propose in place of this traditional format is a more global view, where what changes is a construction as a whole; that is, the form and the meaning do not change ‘independently’ of each other, one at a time. What we have is rather a process of the following form: [external force >]
X
→
Y (with marker; e.g. vaan)
For instance, and in relation to the sä passive: one might think that the meaning of ‘you’ would be stable, but even that changes; in fact, when Häkkinen says (3), he first and foremost uses sä ‘you’ to mean ‘I’. That is, 2nd sg in form means 1st sg. But describing what happens solely in terms of the meaning of one form changing into another meaning does not give the whole picture. Rather, what changes is the construction as a whole. In slightly technical terms, we can describe such changes as starting out with changes in the values assigned to certain attributes, and, more in particular, we can do this in terms of what J. Leino & Östman (2005) have elsewhere talked about as value pools. In terms of variability, values in such a pool can – or even tend to – get mixed up. This is why it is important to have them in a pool and not as separate entities in separate constructions. The emergence of the sä passive is not a solitary emergence but it affects the system as a whole. For instance, and in particular, it negatively affects the traditional, established use of the 3rd sg as the generic expression par excellence. The traditional Finnish way of saying (1) would have been (9). But because of the
Finnish genericity
emergence of the sä passive, the construction that licenses the kinds of constructs we find in (9) might now have to narrow its sphere of reference.20 (9) Sitä vaan ajaa. it-part just drive-3sg ‘One just drives.’
What we have in (9) has received a competitor in the seemingly more addresseedirected form of expression we find in the sä passive; whether the latter will ultimately replace expressions of type (9), or whether the two will divide up the sphere of application among them in some way is for the future to tell. From the point of view of general pragmatics, discourse, ideology, and culture, this change in the grammatical system might have repercussions for the culture as a whole – or the other way around, since clearly language and culture go hand in hand. The generic 3rd sg is felt to be symptomatic of the withdrawing, silent, and indirect Finnish culture, whereas the sä passive is felt to be arrogant and too direct, even in some sense Anglo-American. It is thus noteworthy that the sä passive has been able to become so entrenched in such a relatively short period of time in modernday Finnish despite a generally felt idea that the Finnish society is characterized by a strong prescriptive, Finnish language-ideological stance of ‘purism’. All in all, in the historical development of Finnish impersonal constructions, the direction of change has been influenced by two forces pulling in slightly different directions, but in directions similar enough to be reconcilable: (i) the formmeaning constellations as impersonal constructions; and (ii) a functionally motivated Genericity pattern. By acknowledging these as two interacting forces, we can better understand why some of the impersonal constructions have the same verbal morphology but different meanings and why some Genericity constructions, although they have very similar meanings, are formally very different. It is admittedly difficult to say where one construction stops and another starts, especially since the ingredients for both are available elsewhere in the language. A way to handle such issues has been described in terms of metaconstructions by J. Leino (2003; cf. also J. Leino & Östman 2005). The crucial point here is that related constructions are cognitively similar.
20. This begs an important question that we cannot go into here: What happens to a construction whose realm of applicability changes? Do constructions remain in language and only change their function or meaning, or domain of usage? For how long do they stay? In particular, is this what we mean, when we say that historically earlier stages are present in the language of today? Or do constructions simply fade out of language? Further, if a construction takes on a different sphere of reference, is it still the same construction?
51
52
Pentti Leino and Jan-Ola Östman
We have wanted to show that a full understanding of the mechanisms involved in language change requires an integration of syntactico-semantic analyses with discourse and areal information, and an acceptance of a view where ‘the system’ is constantly in flux. Specifically, we have wanted to show how language change and variability more generally are orchestrated through an intricate interplay between (in this case: two) forces. We have also shown that the re-organization of grammatical systems cannot be understood outside of systematic discourse considerations: constructional clusters are motivated not only by sharing narrowly grammatical properties (in terms of form and meaning) but also by their discourse function. In effect, we have thus also shown that a constructional approach to language change and variability has a number of advantages in comparison to other approaches. Our arguments to the effect that the cognitive organization and reorganization of language takes place in terms of constructions is not meant to suggest that organization and reorganization are two separate ‘stages’. Reorganization is definitional of language, since language is constantly changing: variability is just as central to language as is conventionalization as a code. The fact that ‘everything flows’ also needs to be taken into account cognitively.
References Barlow, Michael & Suzanne Kemmer (2000). Usage-based models of language. Stanford, CA: CSLI. Dahl, Östen & Maria Koptjevskaja-Tamm (Eds.) (2001). Circum-Baltic Languages: Volume 1: Past and Present; Volume 2: Grammar and Typology. Amsterdam & Philadelphia, PA: John Benjamins. Fauconnier, Gilles (1985). Mental Spaces. Aspects of Meaning Construction in Natural Language. MIT Press. Fillmore, Charles J. (1982). Frame semantics. In The Linguistic Society of Korea (Ed.), Linguistics in the morning calm (111–137). Seoul: Hanshin. Fillmore, Charles (1998). Grammatical Construction Theory and the familiar dichotomies. In Rainer Dietrich & Carl F. Graumann (Eds.), Language processing in social context (17–38). Amsterdam & New York, NY: North-Holland. Fried, Mirjam & Jan-Ola Östman (2004). Construction Grammar. A thumbnail sketch. In Mirjam Fried & Jan-Ola Östman (Eds.), Construction Grammar in a cross-language perspective [Constructional Approaches to Language, 2] (11–86). Amsterdam & Philadelphia, PA: John Benjamins. Fried, Mirjam & Jan-Ola Östman (2005). Construction Grammar and spoken language: The case of pragmatic particles. Journal of Pragmatics, 37(11), 1752–1778. Hakulinen, Auli & Lauri Karttunen (1973). Missing persons: On generic sentences in Finnish. CLS, 9, 157–171.
Finnish genericity
Hakulinen, Auli, Maria Vilkuna, Riitta Korhonen, Vesa Koivisto, Tarja Riitta Heinonen & Irja Alho (2004). Iso suomen kielioppi [SKS toimituksia 950]. Helsinki: Suomen Kirjallisuuden Seura. Herlin, Ilona & Laura Visapää (Eds.) (2005). Elävä kielioppi. Suomen infiniittisten rakenteiden dynamiikkaa [SKS toimituksia 1021]. Helsinki: Suomen Kirjallisuuden Seura. Hinton, Leanne (1979). Irataba’s gift: A closer look at the s > s > θ soundshift in Mojave and ˚ Northern Pai. Journal of California and Great Basin Anthropology – papers in Linguistics, 1, 3–37. Johnson, Christopher R., Charles J. Fillmore, Esther J. Wood, Josef Ruppenhofer, Margaret Urban, Miriam R.L. Petruck & Collin F. Baker (2001). The FrameNet Project: Tools for lexicon building [Version 0.7, January 2001]. Berkeley, CA: International Computer Science Institute. Kuusi, Matti (1977). Yksikön 2. persoona ims. sananlaskuston venäläistymisasteen mittarina. Neuvostoliittoinstituutin vuosikirja, 25. Lambrecht, Knud (1994). Information structure and sentence form. Cambridge, MA: Cambridge University Press. Langacker, Ronald (2001). Discourse in Cognitive Grammar. Cognitive Linguistics, 12, 143– 188. Langacker, Ronald W. (2005). Integration, grammaticization, and constructional meaning. In Mirjam Fried & Hans Boas (Eds.), Grammatical constructions: Back to the roots. [Constructional Approaches to Language, 4] (157–189). Amsterdam & Philadelphia, PA: John Benjamins. Leino, Jaakko (2003). Antaa sen muuttua. Suomen kielen permissiivirakenne ja sen kehitys. Helsinki: Suomalaisen Kirjallisuuden Seura. Leino, Jaakko & Jan-Ola Östman (2005). Constructions and variability. In Mirjam Fried & Hans Boas (Eds.), Grammatical constructions: Back to the roots. [Constructional Approaches to Language, 4] (191–213). Amsterdam & Philadelphia, PA: John Benjamins. Leino, Pentti, Ilona Herlin, Suvi Honkanen, Lari Kotilainen, Jaakko Leino & Maija Vilkkumaa (2001). Roolit ja rakenteet. Henkilöviitteinen allatiivi Biblian verbikonstruktioissa. [SKS toimituksia 813]. Helsinki: Suomalaisen Kirjallisuuden Seura. Leinonen, Marja & Jan-Ola Östman (1983). Passive patterns in Russian and Swedish. In Fred Karlsson (Ed.), Papers from the seventh Scandinavian conference of linguistics, I [Department of General Linguistics, Publications, 9] (175–198). University of Helsinki. Michaelis, Laura & Knud Lambrecht (1996). Toward a construction-based theory of language function: The case of nominal extraposition. Language, 72, 215–247. Ojajärvi, Jussi (2000). Imperialistinen sä-puhe – kun sä ajat formulaa. Kielikello, 3/2000, 19– 22. Östman, Jan-Ola (1981). The Finnish ‘passive’ and Relational Grammar. CLS, 17, 286–294. Östman, Jan-Ola (2005). Construction Discourse: A prolegomenon. In Jan-Ola Östman & Mirjam Fried (Eds.), Construction Grammars: Cognitive grounding and theoretical extensions. [Constructional Approaches to Language, 3] (121–144). Amsterdam & Philadelphia, PA: John Benjamins. Östman, Jan-Ola (2006). Ordstäv som en central del av språket. Om att få med “allt” i en grammatisk beskrivning [‘Wellerisms as a central part of language. On getting ‘everything’ into a grammatical description’]. In Per Ledin, et al. (Eds.), Svenskans beskrivning, 28, 389–401. Örebro university.
53
54
Pentti Leino and Jan-Ola Östman
Östman, Jan-Ola & Jarno Raukko (1995). The ‘pragmareal’ challenge to genetic language tree models. In Seppo Suhonen (Ed.), The Fenno-Baltic cultural area [Castrenianumin toimitteita, 49] (31–66). University of Helsinki & The Finno-Ugrian Society. Paunonen, Heikki (1995). Suomen kieli Helsingissä. Huomioita Helsingin puhekielen historiallisesta taustasta ja nykyvariaatiosta [‘Finnish in Helsinki. Notes on the background to, and the present-day variation of spoken Helsingian’]. Department of Finnish, University of Helsinki. Raukko, Jarno & Jan-Ola Östman (1994). Pragmaattinen näkökulma Itämeren kielialueeseen. [‘A pragmatic perspective on the areal linguistics of Baltic Europe’] [Department of General Linguistics, Publications, 25]. University of Helsinki. Seppänen, Eeva-Leena (2000). Sinä ja suomalaiset: Yksikön toisen persoonan yleistävästä käytöstä. Kielikello, 3/2000, 16–18. Shore, Susanna (1986). Onko suomessa passiivia [‘Is there a passive in Finnish?’] [Suomi, 133]. Helsinki: Suomen Kirjallisuuden Seura. Vilkuna, Maria (1992). Referenssi ja määräisyys suomenkielisten tekstien tulkinnassa [‘Reference and definiteness in the interpretation of Finnish texts’]. Helsinki: Suomen Kirjallisuuden Seura.
chapter 4
Precategoriality and argument structure in Late Archaic Chinese Walter Bisang University of Mainz
1.
Introduction – prerequisites for describing precategoriality in a Construction Grammar framework
1.1
Setting the stage
This paper aims to show that Late Archaic Chinese is a precategorial language and it will account for this typologically very remarkable fact by combining elements from different approaches to Construction Grammar with pragmatics in terms of stereotypical implicatures. The fact that such an account is possible proves the richness of the theoretical inventory provided within the different approaches to Construction Grammar that are available at the moment. In a precategorial language like Late Archaic Chinese, the lexicon does not determine the occurrence of its lexical items in the syntactic positions of nouns (N) and verbs (V) (cf. (8) for a definition). Thus, the interpretation of a lexical item in an utterance depends on the structural environment in which that item occurs. Construction Grammar with its pairing of form and meaning and with its assumption that constructions contribute their own meaning to a linguistic structure (Fillmore, Kay & O’Connor 1988; Goldberg 1995, 2005; Michaelis & Lambrecht 1996) excellently captures the interpretation of lexical items in their structural environment. For that purpose, it will be necessary on a more specific level to integrate semantic aspects of Goldberg’s (1995, 2005) approach with Croft’s (1991, 2000, 2001) conceptual space for parts of speech and his concept of word-class indicating constructions. Since Croft (1991, 2000, 2001) more equally integrates the semantic and the syntactic properties of a construction, his approach will be more prominent in this paper. His idea that the syntactic structure
56
Walter Bisang
of a construction is a “scaffolding” (Croft 2001: 238) for identifying the semantic components of a construction also serves as a link to the concept of coercion in terms of Michaelis (2004). The syntactic structure provides the pattern that has the power to coerce a lexical item into a particular function. The combination of pragmatics with Construction Grammar will be needed because pragmatics governs the assignment of lexical items to syntax (N-slot, V-slot) via stereotypical implicatures. The word-class indicating constructions to be analaysed will be the transitive and the intransitive argument structure constructions. In a wider context, the present paper aims to show that one may have to reckon with linguistic diversity at a very fundamental level, that is, diversity at the level of how such basic components of the architecture of language such as the lexicon, syntax, semantics/cognition, and pragmatics interact. If the relation between the lexicon and syntax is mediated through pragmatics, this is in direct contradiction to all linguistic theories that assume that the syntactic behavior of lexical items is determined by the lexicon. Before going into any details, the following examples may provide a first impression of the syntactic versatilıity of lexical items. A word such as měi, which is usually translated as ‘beautiful’, does not only occur in the function of a stative verb as in (1) and (2), it can also occur in the syntactic position of a transitive verb as in (3) or in the syntactic position of a noun as in (4): (1) měi is interpreted as a predicative stative verb (= adjective) (Lunyu 3.25): zˇı wèi Sháo jìn měi. master say Shao(style.of.music) perfectly beautiful ‘The master said that Shao-style music [i.e. the music of the cultural hero Shun] is perfectly beautiful.’ (2)
měi is interpreted as an attributive stative verb (= adjective) (Lunyu 9.12): yˇou měi yù yú sī. there.is beautiful jade/gem PREP here ‘There is a beautiful gem here.’
(3) měi is interpreted as a transitive verb with putative meaning (Zuo, Xiang 25): jiàn Táng Jiāng ér měi zhī. see/meet Tang Jiang and beautiful OBJ:3 ‘He saw Tang Jiang and thought her to be beautiful.’
Precategoriality in Late Archaic Chinese
(4)
měi is interpreted as a noun (Lunyu 6.14): Sòng Zhāo zhī měi Zhao.Duke.of.Song ATTR beauty ‘the beauty of Zhao, duke of Song’
In a similar way, the word xìn can be interpreted as an intransitive verb (5) or as a transitive verb (6) in the meaning of ‘be trustworthy’ and ‘believe someone to be trustworthy/consider someone as trustworthy’ or as a noun with the meaning of ‘trustworthiness, confidence’ (7). (5)
xìn is interpreted as an intransitive verb in the predicate position (Hanfeizi 47): xiū shì zhě wèi bì zhì, cultivated knight NML/TOP not.yet must be.wise zhì shì zhě wèi bì xìn. be.wise knight NML/TOP not.yet must V:trustworthy ‘A cultivated knight is not necessarily wise, a wise knight is not necessarily trustworthy.’
(6)
xìn is interpreted as a transitive verb (Meng 7A.34): rén jiē xìn zhī. man all V:believe OBJ:3 ‘People all believed him.’
(7) xìn is interpreted as a noun (Hanfeizi 18): [Context: If a ruler of men is not able to make the laws clear and thus control the authority of the great ministers] wú dào dé xiˇao rén zhī xìn yˇı. not.have way get/reach small man ATTR N:confidence PF ‘[he] has no way of achieving the confidence of the commoners.’
The rest of this subsection will be divided into three parts. Subsection 1.2 will outline the Construction Grammar approach to precategoriality adopted in this paper and Croft’s (1991, 2000, 2001) conceptual space for parts of speech. Subsection 1.3 will show how Croft’s markedness approach is related to pragmatic implicatures. The details of the overall structure of this paper will be given in Subsection 1.4. At this stage, it will suffice to say that Section 2 will be on argument structure and the interpretation of lexical items in the V-position. Section 3 will deal with pragmatics. It will present the stereotypical I-implicature operating in Late Archaic Chinese and it will show how the argument structure construction with its autonomous meaning became of central importance in philosophical debates about the relation between linguistic denotation and reality and the justi-
57
58
Walter Bisang
fication of an emperor’s government. Finally, Section 4 will present a conclusion and an outlook. To conclude this subsection, I would like to provide the reader with some more information on Late Archaic Chinese. The term “Late Archaic Chinese” refers to the Chinese of the period between the 5th and the 3rd centuries BC (Peyraube 1988). Most of the texts quoted in the present paper (Zuo, Lunyu, Meng, Zhuang, Hanfeizi, Gongsun Longzi) reflect the language of this period. This language also represents the model for the written language up to the beginning of the 20th century. A few quotations based on this model language are from the Han dynasty (206 BC to 220 AD; Shiji, Hanshu). The structure of Late Archaic Chinese is relatively consistent throughout the whole period with some variation between different texts. From the consistent system of information structuring (focus, topic, clefting) it is quite clear that the language as it is represented in the texts is colloquial (Bisang 1998: 22–32; also cf. Li 1996 and 1997, who comes to the same conclusion from different perspectives). Late Archaic Chinese represents the last part of a longer period called Old Chinese which refers to the Zhōu dynasty (11th to 3rd centuries BC).
1.2 Outline of a Construction Grammar approach to precategoriality – Croft’s conceptual space The high versatility of lexemes such as měi ‘be beautiful/beauty’ and xìn ‘believe/ belief ’ in examples (1) to (7) is due to precategoriality as presented in the following definition: (8) Definition of precategoriality: A language is precategorial if its lexical items are not determined in the lexicon with regard to the occurrence within a particular slot of a word-class indicating construction.
The above definition in (8) is based on the concept of construction and on certain syntactic positions or slots within it. The concept of construction adopted in this paper can be captured by comparing Goldberg (1995, 2005) and Croft (2001). Goldberg (2005) first points out that “All levels of grammatical analysis involve constructions” and then provides the following definition: (9) Goldberg’s (2005: 5) definition of constructions: learned pairings of form with semantic or discourse function, including morphemes or words, idioms, partially lexically filled and fully general phrasal patterns. ...
Precategoriality in Late Archaic Chinese
Any linguistic pattern is recognized as a construction as long as some aspect of its form or function is not strictly predictable from its component parts or from other constructions recognized to exist. In addition, patterns are stored as constructions even if they are fully predictable as long as they occur with sufficient frequency.
Croft’s definition of constructions runs as follows: (10) Croft’s (2001: 18) definition of constructions: Grammatical constructions in construction grammar, like the lexicon in other syntactic theories, consist of pairings of form and meaning that are at least partially arbitrary. Even the most general syntactic constructions have corresponding general rules of semantic interpretation. Thus, constructions are fundamentally symbolic units ... .
Even though both definitions treat constructions in terms of pairings of form and meaning, Goldberg (1995) looks at argument structure constructions from a syntax-independent semantic perspective (e.g. one and the same construction holds for active and passive), while Croft (2001) equally integrates the semantic side with the form side including syntax. His approach focuses on the analysis of “symbolic relations of a construction” that are defined as follows: [T]he relation between the construction as a whole and the complex semantic structure it symbolizes, and the relation between the elements of the syntactic structure and the corresponding components of the semantic structure. (Croft 2001: 238)
To describe the role of syntax in determining the meaning of a complex linguistic expression, Croft (2001: 238) adopts Langacker’s scaffolding metaphor, in which “component structures are seen as scaffolding erected for the construction of a complex expression” (Langacker 1987: 461). In Croft’s own words, “what appears to be the coding of syntactic relations is in fact scaffolding to help the hearer to identify which element of the construction symbolizes which component in the semantic structure of the construction” (Croft 2001: 238). The role of syntax in my account of precategoriality is crucial for the following two reasons (also cf. Subsection 2.1): – The syntactic function of a lexical item is defined by its position (slot) within a construction. – The semantics of a word is determined by a combination of its own lexical meaning plus the meaning contributed to it by the semantics of the syntactic slot within which it occurs or by the meaning of the construction as a whole.
59
60 Walter Bisang
In this context, I do not only prefer Croft (2001) because he generally looks more equally at syntax and semantics but also because his scaffolding approach immediately leads to the heart of what the syntactic slots of N and V are used for in Late Archaic Chinese. As will also be seen in Subsection 2.1, the syntactic slots within their argument structure construction are the only elements that provide the scaffolding for interpreting lexical items in terms of their ontological category (object, property, action) and in terms of other grammatical functions such as actor, undergoer, or predicate. In addition, the scaffolding function of syntax can be associated with the concept of “coercion” in terms of Michaelis (2004). Lexical items occurring in a given syntactic slot are coerced into the function associated with that slot (also cf. Subsection 2.1). Even though my approach is closer to Croft (2001) than to Goldberg (1995, 2005), her approach will be relevant for the general definition of the meaning of argument structure constructions (Subsection 2.1) and it will turn out to be crucial for the understanding of the productive transitive use of intransitive verbs which get an extra semantic role from the transitive argument structure construction (Subsection 2.2). The proximity of my approach to Croft (2001) is also evident from the concept of “word-class indicating construction” mentioned in my definition in (8). This concept is based on Croft’s (1991, 2000, 2001) typological view that universal properties of language, including parts of speech, are reflected in systematic patterns of variation. Even though I do not fully share Croft’s (2001) radical view of Construction Grammar, my account of precategoriality integrates Croft’s (1991, 2000, 2001) conceptual space for parts of speech which can make parts-of-speech distinctions visible through morphosyntactic structures. I will thus first introduce Croft’s conceptual space before returning to where I deviate from him, at the end of this subsection. A conceptual space is constructed on the basis of typological methods by examining cross-linguistic and language-internal distributional patterns of relevant constructions. The conceptual space presented here must be seen as a small part out of a much more complex space for language, and languages in general. In the horizontal dimension of Croft’s (1991, 2000, 2001) conceptual space for parts of speech as presented in Table 1 below, we find the pragmatic functions or propositional acts of reference, modification and predication (written in capital letters). The relevant constructions are those by which the pragmatic functions are encoded. The names of these constructions are given in italics in Table 1. The pragmatic functions located horizontally in the representation of the conceptual . For some other typological studies based on conceptual space (or semantic space or semantic map) cf. Kemmer (1993), Haspelmath (1997), Stassen (1997).
Precategoriality in Late Archaic Chinese
Table 1. Croft’s conceptual space for parts of speech and function-indicating morphosyntax (From Croft 1991: 67; Croft 2000: 89 and Croft 2001: 92) REFERENCE
MODIFICATION
PREDICATION
OBJECTS
object reference: unmarked nouns
object predication: predicate nominals
PROPERTIES
property reference: deadjectival nouns action reference: action nominals, complements, infinitives, gerunds
object modifier: genitive, adjectivalizations, PP’s on nouns property modifier: unmarked adjectives action modifier: participles, relative clauses
ACTIONS
property predication: predicate adjectives action predication: unmarked verbs
space are correlated vertically with the semantic classes (ontological categories) of those words which fill the relevant roles in the relevant constructions, i.e. objects, properties, and actions (also written in capital letters). The systematic pattern of variation that can be seen from cross-linguistic research is based on morphosyntactic markedness (cf. Subsection 1.3). Croft (1991) describes the markedness correlation between semantic classes and pragmatic functions as follows: There is an unmarked correlation between the semantic class of object and the function of reference so that a word denoting an object is unmarked in the function of reference but marked in other functions. A similar pattern holds between properties and modification and between actions and predication. (Croft 1991: 55)
The lexical items that denote objects and are unmarked in object reference constructions belong to the syntactic category labelled “nouns” in English. The syntactic categories of adjectives and verbs are defined in an analogous way. The present paper will only deal with those word-class indicating constructions which are relevant to argument structure, i.e., to property predication and to action predication. Since Late Archaic Chinese makes no distinction between the predication of properties and the predication of actions, it can be characterized as an adjectival-verb language in terms of Schachter (1985). The constructions relevant in this context are the intransitive argument structure construction and the transitive argument structure construction (for more details cf. Subsections 2.2 and 2.3). To conclude this section, I would like to point out in what sense my approach differs from Croft’s (2001) Radical Construction Grammar (RCG). Theories that reject the existence of universal syntactic categories like RCG fail to account for precategoriality at a universal level because they lack a general syntactic level and
61
62
Walter Bisang
thus cannot provide the tools needed for a general analysis of the relation between syntax and the lexicon. This is no problem for the discussion in this paper, since it is strictly based on the properties of language-specific constructions, i.e., argument structure constructions in Late Archaic Chinese with their specific positions for N and V. However, to keep my approach open to theories that are based on the assumption of universal syntactic categories, I do not subscribe to the languagespecific character of syntactic categories such as N and V. For the same reason, I opt for the integration of the relation between syntax and the lexicon as another prerequisite for a consistent and universal theory of parts of speech (cf. Sasse 1993a, b; Bisang forthcoming c.). Apart from this prerequisite, there are the three prerequisites of semantic criteria, pragmatic criteria, and criteria of morphosyntactic markedness as we find them in Croft (1991, 2000, 2001). In spite of my reservations, I think that it should be possible to integrate Croft’s (2001) conceptual space and the markedness patterns it evokes to theories based on the existence of universal syntactic categories even if one does not adopt Croft’s (2001) radical approach to Construction Grammar. For that purpose, it is necessary to look at the markedness of the lexemes occurring in the N-slot or in the V-slot under the pragmatically and semantically defined structures represented in Table 1. Even in a theory with universal syntactic categories the lexical items in the N-slot and the V-slot are subject to the markedness pattern predicted by Croft’s (1991, 2000, 2001) conceptual space for parts of speech. Finally, even the overall scaffolding function of syntax is not impaired because the existence of universal syntactic categories still can help to identify the semantic interpretation of a complex expression.
1.3
Markedness in terms of pragmatics (inference of stereotype)
The cross-linguistic pattern that varies within a conceptual space as described in Subsection 1.2 is a pattern of markedness. Thus, certain combinations of semantic classes (object, property, action) and pragmatic functions (reference, modification, predication) are more marked than others. Croft (1990: 2, 2003: 87–101, 110–121) presents the following criteria for markedness that are relevant at the level of the analysis of individual languages: (11) Criteria for markedness (adapted from Croft 1990: 92, also cf. Croft 2003: 87– 101, 110–121): a. structural: the number of morphemes to express marked and unmarked values.
Precategoriality in Late Archaic Chinese
b. behavioral: – inflectional: number of cross-cutting distinctions/phonemes the marked and unmarked values contain; – distributional: number of syntactic/phonological environments in which the marked and unmarked values occur. c. frequency: textual: number of occurrences of the marked and unmarked values in text.
In Late Archaic Chinese, at least the majority of words is not subject to changes in their morphological form across different domains of the conceptual space for parts of speech (cf. Table 1). Thus, there is no formal difference between objectdenoting words, property-denoting words, and action-denoting words when they occur in an argument-slot of an argument structure construction. In example (12), we find the property-denoting words guì ‘precious, high’ and jiàn ‘cheap, low’ in the actor position; example (13) shows words denoting actions such as jiàn ‘remonstrate’ and yán ‘speak, say’ in a nominal argument slot. (12)
Words denoting properties (Zhuang 17.35): guì jiàn yˇou shí wèi kěyˇı wèi precious/high cheap/low have time NEG can take.as cháng yě. be.constant EQ ‘High status and low status have their times, they cannot necessarily be taken to be constant.’
(13)
Words denoting actions as undergoers in the preverbal N-slot (Meng 4B.3): jiàn xíng, yán tīng ... remonstrate go speak/say hear ‘The remonstrations are followed [by the ministers], the words are heard, ...’
. There can be no doubt that there was morphology in the period of Old Chinese that immediately preceded Late Archaic Chinese between the 11th and the 6th centuries BC. But that morphology was not sensitive to N and V, i.e., morphemes were not necessarily associated with a fixed position in the N-slot or in the V-slot. Thus, the same morpheme could be interpreted as a noun with one root and as a verb with another root. What is remarkable is the fact that as soon as a root was marked by an additional morpheme, its word class became fixed (either N or V) (cf. Bisang forthcoming a, b for more information). In Late Archaic Chinese, morphology seems to have been considerably reduced. Even if there existed some words at that time that were subject to morphological change, their number does not seem to have been large enough to influence the overall pragmatic system of assigning lexical items to syntax.
63
64 Walter Bisang
As a consequence of what can be seen from (12) and (13), Croft’s (1990) structural criterion (11a) as well as his inflectional behavioral criterion (11b) are of no relevance to Late Archaic Chinese. What matters is a purely pragmatically determined markedness based on how object-denoting lexemes and action-denoting lexemes are stereotypically associated with nouns and verbs as represented by their syntactic positions (cf. sterotypical inferences in terms of Levinson 1987, 1991, 2000). In other words, the ontological category to which a word belongs triggers certain inferences concerning the constructional slots in which it can occur. A word such as ‘tree’, which denotes an object, is more likely to occur in an N-slot than in a V-slot. The stereotypical link between an object-denoting word, and an N-slot is thus pragmatically less marked than the link between the same word, and a V-slot which implies the flouting of a stereotypical inference. Needless to say, pragmatic markedness as discussed here can be seen as the basic factor which motivates the typological word-class indicating patterns discussed by Croft (1991, 2000, 2001). The fact that ontological categories like objects, properties and actions can trigger certain markedness patterns if associated with pragmatic processes like reference, modification and predication has its pragmatic foundation. Pragmatic markedness may not be without consequences for Croft’s criteria as listed in (11). In fact, structural and behavioral inflectional patterns of markedness can be motivated by pragmatics, since what triggers an interpretation complementary to the stereotype is often associated with choosing a more prolix form, i.e., a structurally more marked form (cf. I-implicatures vs. M-implicatures in terms of Levinson 1987, 1991, 2000). Although such pragmatically-based processes of grammaticalization are well attested cross-linguistically, this is not necessarily the case as can be seen from Late Archaic Chinese. Nevertheless, pragmatic markedness affects the distributional behavior (11b) and the text frequency (11c) of words in Late Archaic Chinese. The occcurrence of individual words in the V-slot and in the N-slot, and the frequency with which this is attested in texts seem to be correlated with the degree of stereotypicality with which a word is associated with these slots.
1.4 On the structure of the rest of this paper The overall structure of this paper is as follows. Section 2 will be dedicated to argument structure constructions. It will be shown how the specific meaning of a word in an utterance can be derived by computing its lexical meaning, its position in a verb slot or a noun slot and the meaning of the construction as a whole. The pragmatic and philosophical use of precategoriality will be the topic of Section 3. The first part of this Section (3.1) will be dedicated to pragmatics. It will deal with
Precategoriality in Late Archaic Chinese
the way in which stereotypical inferences determine the distribution of lexical items into syntax (i.e., into a noun slot or verb slot) and with the relation between the animacy hierarchy and stereotypical inferences. The second part (3.2) will show how the precategoriality of words in Late Archaic Chinese can be used for philosophical purposes concerning the truth status of the relation between a sign and the concept it refers to. Generally speaking, Sections 2 and 3 represent two different aspects of meaning. Section 3 deals with the verbal meaning of syntactically underspecified lexical items that are assigned to the syntactic slot for verbs by means of pragmatic inference. Section 2 shows how the concrete meaning of these lexical items in the verb slot can be calculated from the meaning contributed by the construction. The fourth and last section will present a conclusion and an outlook which compares precategoriality in Late Archaic Chinese with nouns used as verbs in English and a short discussion on the theoretical consequences of precategoriality in Late Archaic Chinese from a wider perspective.
2.
Argument structure constructions
2.1
Introduction: Basic concepts
There are two argument structure constructions that will be relevant for the discussion of precategoriality in this paper, the transitive and the intransitive argument structure constructions. The basic structure of these constructions is as follows: (14) Intransitive argument structure construction: NPS V or: V NPS (15) Transitive argument structure construction: NPA[ctor] V NPU[ndergoer] (with two overt arguments)
Intransitive verbs have one argument (NPS) position in front of the verb, transitive verbs have two argument positions. If both arguments of a transitive verb are overt the actor (NPA) precedes the verb and the undergoer (NPU) follows it. If there is no overt actor argument, an overt undergoer argument can occur in either position depending on constraints of information structure. If the undergoer is activated, it tends to occur preverbally, if not, it tends to occupy the postverbal position. The same seems to apply to the argument position of the intransitive construction. . As far as the position of the intransitive argument is concerned, I follow Li (1996: 115, fn. 11) who states: “Late Archaic Chinese already displays the tendency to place definite NPs preverbally and indefinite NP’s postverbally”. However, I have not been able to find contras-
65
66 Walter Bisang
From the perspective of the two reasons why the syntactic side of constructions matters (cf. Subsection 1.2), the structures in (14) and (15) are crucial for ascribing the syntactic functions associated with the N-slot (NPS, NPA, NPU) and the V-slot to lexical items in an utterance. The second reason based on the meanings provided by the syntactic slots first needs some general discussion of the meaning of argument structure constructions. The meaning contributed by the argument structure construction is associated with “scenes basic to human experience” (Goldberg 1995: 5). Thus, argument structure constructions represent “dynamic scenes: experientially grounded gestalts, such as that of someone volitionally transferring something to someone else, someone causing something to move or change state, ...” Goldberg (1995: 5). The meaning that is of particular relevance for analysing precategoriality in a Construction Grammar framework is causation. The relation between actor und undergoer in the transitive argument structure construction is generally captured by transitivity as defined by Hopper & Thompson (1980). The prototypical meaning of that construction is that of an actor volitionally making an undergoer do something or be something. In addition to the meaning of argument structure constructions, the syntactic slots also contribute their more specific meaning within the scenario provided by the argument structure construction. In this sense, they help to identify the semantic function of a lexical item in terms of scaffolding (cf. Subsection 1.2). If a lexical item occurs in the NPA-slot, its semantic interpretation will be that of an object in the function of an actor. If it occurs in the V-slot of a transitive argument structure construction it will be interpreted as a transitive verb, and so on. Thus, the syntactic slot-structure of a construction makes it possible to state whether a given lexical item is to be understood as an object, a property, or as an action and whether it is a predicate or an actor (A), an undergoer (U), or a single argument of an intransitive verb (S). The mechanism at work in the analysis of lexical items within a construction is called “coercion” in Construction Grammar (Michaelis 2004). Transferred to the situation in Late Archaic Chinese, this means that a lexical item occurring in a given slot of the argument structure construction is “coerced” into a particular semantic interpretation associated with that slot. The way in which coercion operates in Late Archaic Chinese will be shown in the next two subsections. Subsection 2.2 will present lexemes denoting states, tive constructions such as the following in Modern Standard Chinese: Rén lái-le. ‘The person has come.’ Lái-le rén. ‘A person has come’. For that reason, I refrain from presenting rigid rules concerning the position of NPS relative to the verb. In addition, further research may well reveal that the positional versatility of intransitive arguments is only possible with S-arguments in the function of undergoers (cf. unaccusative verbs) as in Modern Standard Chinese.
Precategoriality in Late Archaic Chinese
actions, and properties and their behavior in the V-slot of the transitive and the intransitive argument structure constructions. In Subsection 2.3, I shall look at lexemes denoting objects in the V-slots of the same two types of argument structure construction.
2.2 Intransitive verbs / Transitive verbs In most languages, there is a clear-cut distinction between verbs that can be used intransitively and those that can be used transitively (cf. the typologically thorough study by Nichols et al. 2004 on transitivizing and detransitivizing languages). If a verb can change its category, this is tied to some morphological changes such as transitivisation or detransitivisation. In Late Archaic Chinese, the situation is different. Almost any verb occurring in the V-slot of the intransitive construction can also occur in the V-slot of the transitive construction. One feels intuitively that verbs like shā ‘kill’ and shí ‘eat’ are transitive and that verbs like lái ‘come’ and kū ‘weep’ are intransitive; but when one looks for some sort of formal criterion on which to base this distinction, it turns out to be very elusive. Both types of verbs can take objects, including the formally marked objective pronoun zhī. (Norman 1988: 91)
This remarkable property of Late Archaic Chinese can be accounted for by an analysis in which the intransitive S-argument takes the function of an U-argument, while the A-argument is contributed by the construction. Such an analysis combines the relevance of syntactic structure with the ability of constructions to contribute argument structures in terms of Goldberg (1995, 2005: 38–43). The meaning of the overall construction is causative or putative. To understand this, it is necessary to distinguish two types of intransitive verbs, those with an argument that has control over the predicate (intr[+con]) and those with an argument that does not (intr[–con]). Verbs of the type (Vintr[–con]) are interpreted either causatively or putatively. Putative is a well-established term in Sinology that is derived from Latin putāre ‘consider’ and refers to lexical items in the V-slot that are interpreted in terms of ‘X considers Y to be ...’. Verbs of the type (Vintr[+dyn]) are only interpreted causatively. The causative/putative interpretation and its correlation with [±control] verbs is summarized in (16) and (17), based on (14) and (15). The general expression format of the notation is inspired by VanValin & LaPolla (1997) and Van Valin (2005). In addition to the CAUSE operator, I introduce the CONSIDER/TREAT operator that represents putative interpretation. . The two interpretations are not semantically independent. On a more general level of abstraction, the putative interpretation may well be described as being derived in some way from
67
68 Walter Bisang
(16) Causative interpretation: Vintr[–con]´ (NPS) → NPA [CAUSE [BECOME Vintr[–con]´ (NPU(S))]] Vintr[+con]´ (NPS) → NPA [CAUSE Vintr[+con]´ (NPU(S))] (17) Putative interpretation: Vintr[–con]´ (NPS) → NPA [CONSIDER/TREAT AS Vintr[–con]´ (NPU(S))]
Examples with Vintr[–con]: xiˇao ‘be small’, dà ‘be big’: (18)
With causative interpretation (Han Feizi 8.2): Bí dà kě xiˇao, xiˇao bù kě dà yě. nose big can small small NEG can big EQ ‘If a nose is big, one can make it smaller, if it is small, one cannot make it bigger.’
(19)
With putative interpretation (Meng 7 A 24.1): Kˇong-zˇı dēng Dōng Shān ér xiˇao Lˇu, Confucius ascend East Mountain and small Lu dēng Tài Shān ér xiˇao tiānxià. ascend Tai Shan and small beneath.the.heavens/the.world ‘Confucius ascended the Eastern Mountain and Lu appeared to him small [and he considered Lu to be small], he ascended the Tai Mountain and all beneath the heavens appeared to him small.’
Examples with Vintr[+con] lái ‘come’: (20)
Intransitive interpretation of lái ‘come’ (Lunyu 1.1): yˇou péng zì yuˇan fāng lái have friend P:from far direction come ‘to have friends who come from afar’
(21)
Causative interpretation of lái ‘come’ (Lunyu 16.1): jì lái zhī zé ān zhī. PFV come OBJ:3s then be.settled/content OBJ:3s [Context: If such a state of affairs exists, yet the people of far-off lands still do not submit, then the ruler must attract them by enhancing the prestige of his culture.] ‘And when he has made them come [= attracted them], he contents them.’
the causative interpretation. Given the frequency of the putative interpretation through all the texts of Late Archaic Chinese and given the cultural relevance of this concept (cf. Subsection 3.2), I prefer to treat both interpretations as equal in this language. Consequently, I also assume two operators for Late Archaic Chinese, i.e. a CAUSE operator and a CONSIDER operator.
Precategoriality in Late Archaic Chinese
Two more examples with Vintr[–con]: zhˇang ‘be old/venerable’, bái ‘be white’, and bìng ‘be ill’: (22)
With intransitive and putative interpretation (Meng 6 A 4) bˇı zhˇang ér wˇo zhˇang zhī, fēi yˇou zhˇang that.one elder and I elder OBJ:3 NEG.COP have elder yú wˇo yě, yóu bˇı bái ér wˇo bái zhī, PREP I EQ be.like that.one white and I white OBJ:3 cóng qí bái yú wài yě. according POSS:3 white PREP external EQ ‘That that [man there] is senior to me and that I treat him as senior has nothing to do with [the presence of] seniority in me, just as that [man] is white and I consider him as white according to his whiteness, which is external [to me].’
(23)
bìng ‘ill’ with intransitive and putative interpretation (Zhuang 23.33): rán qí bìng bìng zhě yóu wèi bìng yě. be.so POSS:3s be.ill illness NML be.like still.not V:be.ill EQ ‘The situation is such that he who considers his illness as an illness is still not quite ill.’
2.3 Lexemes denoting objects and their function in the V-slot Both argument structure constructions also take lexemes denoting objects in their V-position. The meaning of these utterances can be derived from combining the meaning of the cognitive subcategory to which the lexeme in the V-slot belongs with the meaning contributed by the construction, which basically remains the same as described in (16) and (17). The rest of this section is devoted to the presentation of the following six cognitive subcategories or lexical classes presented in (24) (there might be additional ones) and their function in the V-slots of the two argument structure constructions. (24) – – – – – –
lexemes denoting humans and relations among humans (family or official function) lexemes denoting instruments/man-made objects lexemes denoting sense organs lexemes denoting places and buildings lexemes denoting first and second person lexemes denoting numbers and measures
69
70 Walter Bisang
2.3.1 Lexemes denoting humans and relations among humans (family or official function) The interpretation of lexemes denoting humans and relations among humans (family or official function) in the V-slots of the intransitive and the transitive constructions follows the same principles as outlined in (16) and (17). The details are summarised in (25) below (INT stands for ‘used in the intransitive argument structure construction’, TR stands for ‘used in the transitive argument structure construction’): (25) N: person/function INT: a. NPS behaves like a (true) N, NPS is a (true) N b. NPS becomes a (true) N TR a. NPA CAUSE NPU[S] to Vintr (be/behave like a [true] N) b. NPA CONSIDER NPU[S] to Vintr (be/behave like a [true] N)
The syntactic basis of the following famous example from Lunyu is the intransitive argument structure construction. In both of its slots, i.e., in the argument-slot as well as in the V-slot, we find the same word denoting an object. The interpretation of this lexeme in the V-slot is ‘to behave like a true prince, minister, etc.’. In the second part of this example, the V-slot is indicated by the negation bù ‘not’. (26) INT: to behave like a (true) N (Lunyu 12.11): [Context: Duke Jˇıng of Qí asked Confucius about good government.Confucius replied: Jūn jūn, chén chén, fù N:prince V:behave.like.a.prince N:minister V:minister N:father fù, zˇı zˇı. Gōng yuē: shàn zāi! V:father N:son V:son duke say good EXCL xìn rú jūn bù jūn, chén bù believe/indeed if N:prince NEG V:prince N:minister NEG chén, fù bú fù, zˇı bù zˇı, V:minister N:father NEG V:father N:son NEG V:son sūi yˇou sù, wú dé ér shí zhū? even.if have millet I get and eat OBJ:3.Q ‘Let the prince behave like a prince, the minister like a minister, the father like a father and the son like a son. The duke said: How true! If, indeed, the prince does not behave like a prince, the minister does not behave like a minister, the father does not behave like a father and the son does not behave like a son, even if I have millet [i.e. food], shall I manage to eat it?’
Precategoriality in Late Archaic Chinese
In the following example, the lexeme wàng is used inchoatively in the sense of ‘to become a king’: (27)
wáng ‘N: king’/ wàng ‘V: be king’(Lüshi chunqiu 4.4): wáng zhě lè qí suˇo yˇı wàng. N:king TOP enjoy POSS:3 REL.NA take:INSTR V:be.king ‘Kings enjoy [the way] through which [they] became kings.’
In the transitive argument structure construction, the causative vs. putative interpretation seems to depend to a certain extent on the difference in social status between the subcategory denoted by the lexeme in the V-slot and the subcategory denoted by the lexeme in the actor-slot: (28) a. the lexeme in the actor-slot is socially higher than the lexeme in the V-slot: => causative interpretation b. the lexeme in the actor-slot is socially lower than the lexeme in the V-slot: => putative interpretation (the lower person considers the higher person to be a true higher person; the lower person treats the higher person like a higher person, i.e. with due reverence)
The following two examples constructed by Gassmann (1997: 74) are to illustrate this: (29) a. The lexeme in the actor-slot is socially higher than the lexeme in the V-slot: jūn chén zhī. N:prince V:make.minister OBJ:3s ‘The prince made him his minister/employed him as his minister.’ b. The lexeme in the actor-slot is socially lower than the lexeme in the V-slot: chén wàng wáng. minister V:consider.king N:king ‘The minister considers/treats the king as a real king [with reverence].’
. There is a tonal difference between wáng ‘king [N]’ and wàng ‘be a king [V]’. This tonal difference is due to the reduction of former morphological marking. The specific facts behind this process of reduction are accounted for by Haudricourt (1954), Wang Li (1958), Downer (1959). Also cf. Sagart (1999: 131–135) for a more recent assessment. On the overall relevance of morphology in Late Archaic Chinese, cf. Footnote 2.
71
72
Walter Bisang
In a constellation in which the higher person is the undergoer of a person-denoting lexeme occurring in the V-slot, we get the meaning of ‘to behave towards the higher position as N’, ‘to serve to the higher position in the function of N’ (cf. Gassmann 1997: 75). (30)
The higher person is object of a person-denoting lexeme in the V-slot (Zuo, Xiang 22.6): rán zé chén wáng hū? be.so then V:behave.like.a.minister king Q ‘Since this is so, will you serve the king as a minister?’
The meaning of the lexeme in the V-slot in example (30) may be represented as follows: (31) to behave like/be a (true) N towards NPU
If both nouns are of equal rank, we get either a putative (32) or a causative (33) reading: (32)
yˇou ‘friend’: consider/treat someone as a friend (Meng 5B.3): wú yú Yàn Bˇan yě, zé yˇou zhī yˇı. I PREP Yan Ban be thus V:friend OBJ:3 PF ‘What I am to Yan Ban, I treat him/consider him as a friend.’
(33)
yˇou ‘friend’: make someone one’s friend (Meng 5B.8): tiānxià zhī shàn shì, sī yˇou tiānxià zhī shàn shì. world ATTR good scholar this V:friend world ATTR good scholar ‘The best scholar in the world, he shall be/behave like a friend to the world’s good scholars.’ / ‘The best scholar in the world, he shall make the world’s good scholars his friends.’
In the case of qī ‘wife’ and nüˇ woman’, the interpretation of N is in the sense of ‘cause someone to be the wife of someone’, i.e., ‘give one’s daughter to someone to be married’: (34) (35)
qī ‘wife’ (Lunyu, 5.1): yˇı qí xiōng zhī zˇı qì zhī. take POSS:3 elder.brother ATTR child V:give.a.wife OBJ:3 ‘He gave him the daughter of his own elder brother for a wife.’ nüˇ ‘woman’: (Meng 4A.7): ˇ tì chū ér nü yú Wú. tear go.out and V:woman PREP Wu ‘[His] tears flowed forth while he gave his daughter to be married to [the prince of] Wu.’
Precategoriality in Late Archaic Chinese
2.3.2 Lexemes denoting instruments/man-made objects Apart from the causative interpretation to be expected from the generalisations in (16) and (17), lexemes denoting instruments/man-made objects in the V-slot of the transitive construction can also be understood in terms of application, i.e., in the sense of ‘actor applies N to undergoer’: (36) N: instrument INT: NPS is N/is used as N TR: a. NPA CAUSE NPU[S] to Vintr (be N): to use something/someone in the function of N b. NPA APPLY N on NPU[S]: to use the instrument on something/someone
The following examples illustrate the above interpretations of lexemes denoting instruments in the V-slot: (37)
qì ‘vessel, instrument’: NPS is N/is used as N (Lunyu 2.12): jūn zˇı bú qì. accomplished.scholar NEG V:be.an.instrument/vessel/implement ‘The accomplished scholar is not an instrument’ / ‘The gentleman is not an implement.’
(38) qì ‘vessel, instrument’: causative interpretation, to use something/someone in the function of N (Lunyu 13.25): jí qí shˇı rén yě qì zhī. reach/and POSS:3 employ man EQ/TOP V:instrument OBJ:3 ‘As for his [i.e. the prince’s/ruler’s] employment of people, he uses them as instruments [according to their capacities].’ (39) zhěn ‘pillow’: causative interpretation, to use something/someone in the function of N (Lunyu 7.15): qū gōng ér zhèn zhī. bend arm and V:pillow OBJ:3 ‘to bend one’s arm and make it/use it as a pillow’ (40) biān ‘whip’: applicative interpretation, to use the instrument on some thing/ someone (Zuo, Xi 23.4): gōng-zˇı nù yù biān zhī. prince be/get.angry want V:whip OBJ:3s ‘The prince got angry and wanted to whip him [apply the whip on him].’ . Since instruments lack intentionality, the behave-component at work with nouns denoting human relationships does not apply here. . There is again a change of tone from the nominal (zhěn) use to the verbal (zhèn) use of this word. For additional details, cf. Footnote 5.
73
74
Walter Bisang
Lexemes denoting garments can be interpreted in the causative sense of ‘to use NPU[S] as a garment’ (NPA CAUSE NPU[S] to Vintr [be N]). In a second interpretation, the lexeme denoting a garment is itself the undergoer of the action and we get the meaning of ‘to wear N’. In the case of the general word for garments, we get a tonal difference between its use in the N-slot (yī ‘garment’) and its occurrence in the V-slot (yì ‘wear [garments]’). Given its general meaning, yì is also used with the meaning of ‘to clothe oneself ’. (41) yī ‘garment’ in the causative interpretation of ‘to use NPU[S] as a garment/ wear [it]’, guān ‘hat’ in the interpretation of ‘to wear N’ (Meng 3A.4): Xˇu Zˇı, bì zhī bù ér hòu yì hū? ... Xu Zi must/necessarily weave cloth and after V:wear Q ‘Xu Zi weaves [his own] cloth and wears it, doesn‘t he?’ Xˇu Zˇı, yì hè. ... Xˇu Zˇı, guàn hū? Xu Zi wear haircloth Xu Zi V:hat Q ‘Xu Zi wears clothes of haircloth. Does Xu Zi wear a hat?’ (42)
yì ‘V: dress’ in the meaning of ‘to clothe oneself ’ (Zhuang 9): bˇı mín yˇou cháng xìng, zhī ér yì, that people have constant nature/course.of.life weave and dress gēng ér shí, shì wèi tóng dé. plough and eat this call common virtue/skill ‘The people have a constant course of life [proper to them]: to clothe themselves by weaving and feed themselves by ploughing. This is called their “common skills”.’
2.3.3 Lexemes denoting sense organs Lexemes denoting sense organs in the V-slot are interpreted in the same way as instruments, although the overall range of their functions seems to be restricted. Given the body-part relationship between the sense organ and the lexeme in the intransitive-argument slot (NPS), the latter is more active. Thus, we get the interpretation of ‘NPS DOES the action associated with N’. In the V-slot of the transitive construction, lexemes denoting sense organs are interpreted applicatively: (43) N: sense organs INT: NPS DOES the action associated with N TR: NPA APPLIES the action associated with N onto NPU
The following two examples illustrate the intransitive and the transitive interpretations of lexemes denoting sense organs in the V-slot:
Precategoriality in Late Archaic Chinese
(44)
mù ‘eye’: do/perform the action of the eyes (Zuo, Xuan 12): mù yú yuān jˇıng ér zhěng zhī. V:eye PREP dry well and save OBJ:3 ‘look into a dry well and save him [out of it].’
(45)
Apply the action of a given sense organ onto an object (hear/see) (Hanfeizi): jì ěr ér mù zhī yˇı. PFV ear and eye OBJ:3 PF ‘[he] heard and saw him.’
2.3.4 Lexemes denoting places/buildings Lexemes denoting places/buildings get the meaning ‘to do a stereotypical action at N’ in the intransitive constructions and ‘to make the object denoted by the lexeme’ as a special instance of causativity in the transitive construction: (46) N: place/building INT: NPS DOES what one stereotypically does at or to N TR: NPA CAUSE N to exist: to make N
The lexeme mén ‘gate, door’ is of particular interest. The two stereotypical actions associated with gates and doors are either ‘to guard’ (47) or ‘to besiege [the gate/door]’ (48): (47)
mén ‘door/gate’: do something specific at the gate (Zuo, Wen 15): ...yì rén mén yú Lì Qiū, jiē sˇı. one man V:door/gate PREP Li Qiu all/both die ‘[one man guarded the gate at Gou Mang], one man guarded the gate at Li Qiu, both died.’
(48)
mén ‘door/gate’: do something specific at the gate (Zuo, Xi 28): Jìn hóu wéi Cáo, mén yān duō sˇı. Jin marquis besiege Cao V:attack.the.gate PREP.OBJ:3 many die ‘The marquis of Jin besieged Cao, in an attack of one of its gates, many [of his soldiers] died.’
Another example illustrating the case of a stereotypical action done at the place denoted by the lexeme in the V-slot is the following: (49) guˇan ‘lodge, inn’: do something specific at or to the place/building (Meng 7B.30): Mèng Zˇı zhī Téng, guˇan yú shàng gōng. Mencius go Teng V:lodge PREP upper palace ‘When Mencius went to Teng he was lodged in the upper palace.’
75
76
Walter Bisang
Finally, example (50) illustrates the case in which the object denoted by the lexeme in the V-slot is created: (50)
cháo ‘nest’: make/create the place/building (Zhuang 1): jiāo-liáo cháo yú shēn lín. wren V:nest PREP deep forest ‘The wren builds its nest in the deep forest.’
2.3.5 Lexemes denoting first and second person Pronouns of the first and second person in the V-slot can have two interpretations. In a philosophical context they can be used with the meaning of ‘to consider X as 1st/2nd person’ as in example (51). This example will be further explained under (62) in Subsection 3.2. Since there are different pronouns for each person to express different degrees of politeness, the general interpretation of ‘NPA CONSIDER NPU to be N’ can be understood in the sense of ‘NPA ADDRESS NPU with the degree of politeness marked by PRON’ (cf. example (52)). In languages like French or German, these differences are expressed by verbal derivations from the relevant pronouns (French: tutoyer, vouvoyer; German: duzen, siezen). (51)
TR: NPA CONSIDERS NPU to be PRON (Zhuang, 6): qiě yě xiāng yˇu wú zhˇı ěr yˇı, yōng jù moreover mutually I/we OBJ:3 EXCL how zhī suˇo wèi wú zhī hū? know REL.NA call I/we OBJ:3 Q ‘Moreover, what we mutually consider to be us, how do we know what we call to consider it to be us?’
(52)
TR: NPA ADDRESS NPU with the degree of politeness marked by PRON: ēn yuàn xiāng ěr rú. favour disfavour RECIP you you ‘Favour and disfavour ‚are on a first-name basis‘ (i.e., they are very close).’ [The pronouns ěr and rú ‘you’ are both used in a rather informal context.]
2.3.6 Lexemes denoting numbers and measures Lexemes denoting numbers or quantities in the V-slot can be interpreted as follows: (53) INT: NPS is NUM/QUANT: to equal the quantity expressed by the numeral TR: a. make something n-times as much b. provide someone with a given quantity
Precategoriality in Late Archaic Chinese
Two out of these interpretations are illustrated by the following examples: (54)
To equal the quantity expressed by the numeral (Shiji Chushijia): jūn bù wˇu, bù gōng chéng, army NEG V:five NEG attack city bù shí bù wéi. NEG V:ten NEG besiege ‘If [the number of] armies is not five, [we] do not attack the city, if it is not ten, [we] do not besiege it.’
(55) Provide someone with a given quantity of ‘eleven penfuls [of animals]’ (Zuo, Ai 7): [context: Fan Yang of Jin was greedy, and threw aside all rules of propriety. He frightened our poor state with his greatness]: gù bì yì shí yì láo zhī. thus worn-out city ten one pen/corral OBJ:3 ‘Thus, [our] worn-out city gave him eleven pens [of animals].’
3.
The pragmatic use of precategoriality
The versatile potential of the lexicon of Late Archaic Chinese can be used for rhetorical purposes and for philosophical purposes. In Subsection 3.1, I shall try to show how the flouting of stereotypical inferences creates rhetorical effects. In Subsection 3.2, I shall briefly try to show how versatility or precategoriality is used for discussing the truth status of the relation between a sign and the element it refers to.
3.1
The pragmatic relation between lexicon and syntax
If a Late Archaic Chinese word can occur in an N-slot, it can also occur in a V-slot and vice versa. Some lexemes occur quite often nominally as well as verbally. Thus, lexemes such as měi ‘beautiful’ in (1)–(4) and xìn ‘believe/belief ’ in (5)–(7) seem to be quite versatile. Another word such as shì ‘N: matter, affair, work; V: serve, work’ is found with the same frequency in both functions in Mencius (cf. Kennedy 1964: 323). Other lexemes significantly prefer either the V-slot or the N-slot. The examples in Subsection 2.3 all prefer the N-slot, although they can be found in the V-slot as can be seen from the examples presented in that subsection. The reason for the fact that some lexemes are more versatile than others may be explained along the following line of argumentation.
77
78
Walter Bisang
Lexemes denoting concrete objects stereotypically imply the occurrence in an N-slot. Lexemes denoting abstract objects are more or less versatile. This correlation is expressed by the following formula, where ‘>’ means ‘implies stronger N-inference than’: (56) Concrete Objects > Abstract Objects
This explains why lexemes like the above shì ‘V:matter; N:serve’ equally occur in both functions. It also accounts for the relative versatility of lexemes like xìn ‘believe/belief ’ (examples (5)–(7), (63)) and měi ‘be beautiful, beauty’ in (4) and for the use of the following lexemes in the function of nominal abstracts: zhaˇ ng ‘old, age/seniority’ in (22), bái ‘white, whiteness’ in (22), and bìng ‘ill, illness’ in (23). If one further elaborates on (56), one may come up with the following version of the animacy hierarchy: (57) 1st/2nd person > proper names > human > nonhuman > abstracts
This hierarchy forms the basis of I-implicatures (Inference of stereotype) as defined by Levinson (1987, 1991, 2000). In terms of Levinson (2000: 115), “stereotypes are connotations associated with meanings, but not part of them, which nevertheless play a role in interpretation”. The higher the meaning of a lexeme is, i.e., the further to the left it is situated on the hierarchy (57), the more likely is the stereotypical implicature that it belongs to the cognitive category of object and has to fit into an N-slot. For that reason, lexemes which can denote abstractions such as měi ‘beauty’ or shì ‘matter, fact’ are more likely to occur in the V-slot apart from their occurrence in the N-slot. If the meaning of a lexeme is situated in a high position on the hierarchy in (57) the effect of flouting, i.e. the effect of putting this lexeme into the V-slot, is more powerful than with a lexeme in a lower position. Due to this difference, the flouting of higher meanings is used for rhetorical purposes. The positions to be discussed in this context are pronouns and proper names. Most instances of 2nd person pronouns occurring in the V-slot have to do with violation of politeness. Representatives of the nobility who do not behave properly and thus do not deserve to be addressed with the pronouns due to their status are addressed with the common pronouns ěr or rú ‘you comparable to German du’. The activity of not using the adequate pronoun in addressing a person of higher status is referred to by putting a lower pronoun into the V-slot followed by the lexeme denoting that person in the undergoer position.
. There are no 3rd person pronouns for actors in Late Archaic Chinese. Sometimes, we find demonstratives in this function.
Precategoriality in Late Archaic Chinese
Proper nouns or geographic terms in V-slots can be found in rhetorically highly marked situations. In example (58), the speaker of the sentence puts the proper name Wú wáng ‘King Wu’ into the V-slot. A large proportion of the meaning of that name can be derived directly from rule (25) on person-denoting lexemes in the transitive argument structure construction. Thus, the formula ‘NPA CONSIDER NPU[S] to Vintr (be/behave like a [true] N)’ yields the more concrete interpretation of ‘you CONSIDER me to be king Wu’. The part of mutual knowledge to derive the concrete meaning of (58) is that king Wu was murdered. With this historical background knowledge, the highly dramatic situational meaning of ‘Do you want to kill me?’ can easily be inferred. Example (59) refers to the rather remarkable fact that barbarians have the chance to be treated like Chinese. This is expressed by putting the terms of Zhōngguó ‘China’ and Yí ‘Yi-barbarians’ into the V-slot. The example is from Han Yu and belongs to the eighth century AD. Han Yu lived between 768 and 824. He was one of the most prominent and brilliant advocates of the »Old-Style« movement (gˇuwén movement), which tried to go back to the simplicity and to the expressive power of the literary style of the classical period represented by Late Archaic Chinese. By putting lexemes denoting two different people and cultures into the V-slot, he has recourse to the rhetorical effect produced by flouting a stereotypical implicature valid in the classical period and maybe still in use in his time. (58)
Zuo (Ding 10): Gōng Ruò yuē ěr yù Wú wáng wˇo hū? Gong Ruo say you want Wu king I Q ‘Gong Ruo said: “Do you want to deal with me as the King of Wu was dealt with?”’ [King Wu was murdered. +> ‘Do you want to kill me?’] (59)
Han Yu 6.42b: Zhūhóu yòng Yí lˇı, zé Yí zhī, Yí feudal.lords use N:Yi rites then V:Yi OBJ:3 V:Yi ér jìn yú Zhōngguó, zé Zhōngguó zhī. in.spite.of enter into N:China then V:China OBJ:3 ‘If the feudal lords followed the rites of the Yi, then [he = Confucius] treated them like Yi[–barbarians], if they joined China in spite of being Yi, [he] treated them like [civilized] Chinese.’
. The fact that Zhōngguó at the time of the classics rather meant ‘the countries of the center [zhōng ‘middle, centre’, guó ‘country’]’ does not change the basic facts mentioned above. The verbal use of this lexeme may be translated by ‘to treat someone as belonging to the countries of the centre’.
79
80 Walter Bisang
3.2 Precategoriality and the discussion of philosophical concepts The truth status of the relation between a sign and the element it refers to is of central importance in the philosophical and political debate at the times of Late Archaic Chinese. Many philosophical discussions centered on the question of whether a certain concept was correctly referred to by a given lexeme. Much of this discussion is subsumed under the term of zhèngmíng (rectification of the names), a term which was linked to the socio-political problem of whether a certain individual or a certain rite was rightly referred to by a given traditional term. Ultimately, zhèngmíng was concerned with the justification of an emperor and his dynasty. If he did not conform to the properties associated with his title and rank this was a reason for a change of government (for further discussion of this topic cf. Gassmann 1988). An excellent and very economical tool for discussing the truth status of a referential act is the transitive argument structure construction with its putative interpretation in the sense of ‘NPA CONSIDER NPU(S) to Vintr (be/behave like a [true] N)’ with inferences such as ‘NPA takes NPU(S) to be a (true) N’ or ‘NPA treats NPU(S) as a (true) N’. Some examples with the lexemes zhaˇ ng ‘be old; age’, bái ‘be white; whiteness’, and bìng ‘be ill; illness’ are given in examples (22) and (23) above. The following example is another instance of this style, where one and the same lexeme occurs twice, first in the V-slot and then in the N-slot: (60)
Lunyu (1.7): xián xián yì sè V:consider.virtuous N:what.is.virtuous/virtue slight colour/beauty ‘consider virtue to be virtue and slight beauty’
The following example deals with problems of assigning the correct designation, i.e. yáng ‘sheep’ and niú ‘ox’, to a selection of relevant objects. In this context, the lexemes denoting ‘sheep’ and ‘ox’ are used in the V-slot with the meaning of ‘consider something as a sheep/an ox’: (61) niú ‘N: ox/ V: consider as an ox’/ yáng ‘N. sheep/ V: consider as a sheep’ (Gongsun Longzi, cf. Graham 1957: 161): yáng yˇou jiˇao, niú yˇou jiˇao, niú zhī sheep have horn ox have horn V:consider.an.ox OBJ:3s ér yáng yě, yáng zhī although sheep EQ V:consider.a.sheep OBJ:3s ér niú yě, wèi kě. although ox EQ not.yet be.possible
Precategoriality in Late Archaic Chinese
‘Sheep have horns and oxen have horns. To consider [something] an ox although it is a sheep and to consider [something] a sheep although [it] is an ox is inadmissible.’ [General meaning: Although sheep and oxen share the property [+horn] they do not belong to the same species.]
The argument structure construction is even attested with the pronoun of the first person in example (51), repeated as (62): (62)
TR: NPA CONSIDERS NPU to be PRON (Zhuang, 6): qiě yě xiāng yˇu wú zhī ěr yˇı, yōng jù moreover mutually I/we OBJ:3 EXCL how zhī suˇo wèi wú zhī hū? know REL.NA call I/we OBJ:3 Q ‘Moreover, what we mutually consider to be us, how do we know what we call to consider it to be us?’
The transitive interpretation of wú ‘I’ in this example is overtly marked by the object pronoun of the third person (zhī) that immediately follows it. What is discussed is the problem of self-reference of a group of speakers about what constitutes themselves. What is it that makes us consider ourselves as being us? Thus, the sequence wú zhī [I/we OBJ:3] is analysed according to one of the general meanings of the transitive argument structure construction, i.e. ‘NPA CONSIDER NPU(S) to Vintr (be/behave like a [true] N)’. The last example to be discussed in this section certainly represents one of the more fascinating examples. It is from the philosopher Xunzi and starts with the sequence xìn xìn xìn yě analysed in (63) below (on the versatility of xìn, also cf. (5) to (7)): (63) xìn ‘N: trust, trustworthiness/ V: trust, consider sth. trustworthy’, yí ‘N: doubt, doubtfulness/ V: doubt, consider sth. doubtful’ (Xunzi, Graham 1964: 48): xìn xìn xìn yě, V: consider.trustworthy N:what.is.trustworthy N:trustworthiness EQ yí yí yì xìn yě. V:doubt N:what.is.doubtful also N:trustworthiness EQ ‘To consider as trustworthy what is trustworthy is trustworthiness, to doubt what is doubtful is also trustworthiness.’ (Graham’s 1964 translation: ‘Trusting the trustworthy is trustworthiness; doubting the doubtful is also trustworthiness.’)
81
82
Walter Bisang
As was pointed out by Graham (1964: 48) “even the syntax is doubtful” if we limit our view to the first four lexemes in (63) by themselves. Since the sequence of the first and the second line are structured in a parallel way, the structure of the first line can be analysed from the second line in which yí yí follows the VN-pattern presented in examples such as (60) above. Thus, the first xìn is in the V-slot followed by the second xìn in the N-slot. The whole sequence of VN is in the N1position of the equative construction marked by yě.10 Finally, the third xìn is in the N2-position of this construction. Of course, Xunzi could have used more overt indicators of how his sentence is to be understood. But this is the task of commentators such as Yang Liang whose comment is dated 818 AD in the preface. His paraphrase of (63) runs as follows: (64)
Comment by Yang Liang (from Graham 1964: 48): xìn kě xìn zhě, yí kě V:trust be.possible V:trust NML V:doubt be.possible yí zhě, yì suī bù tóng V:doubt NML attitude although NEG be.same jiē guī yú xìn yě. all/both go.back.to PREP N:trustworthiness EQ ‘Trust in what deserves trust and doubt of what deserves doubt, although as attitudes they are not the same, both belong to trustworthiness.’
Example (63) follows a general pattern of parallelism which was central to classical Chinese schools of logic. This structure of argumentation did not primarily aim at analyzing propositions but at laying them side by side and comparing them (cf. Graham 1964: 49). Consequently, Graham (1964) argues that the reason why Xunzi did not choose a more explicit form of expression was due to a wide-spread pattern of philosophical argumentation. In my view, this cannot be the whole story. If Xunzi only wanted to build up a couplet of parallel sentences in order to satisfy the rules of philosophical argumentation, he might as well have chosen a structure which allows to be grammatically more explicit in a parallel way. For that reason, I assume that the form in which (63) is presented is not only the result of how to argue philosophically, it is also the result of a rhetorical intention. A sequence such
10. Affirmative equational sentences have the structure NP1 NP2 yě. The following example is to illustrate this (on equative clauses and yě also cf. Gassmann 1980): (i) Meng (4.B.31) Zˇı-sī chén yě. Zi-si minister EQ ‘Zi-si was a minister.’
Precategoriality in Late Archaic Chinese
as xìn xìn xìn must have its rhetorical effect in the sense that it is a challenge to the reader/hearer who has to parse it, assuming that the writer/speaker somehow follows a principle such as Grice’s (1975) cooperative principle. The fact that a later commentator such as Yang Liang paraphrases this sequence may further corroborate its challenging character. The rhetorical effect may, to a considerable extent, be due to the versatile use of xìn in the V-slot as well as in the N-slot. Example (63) may thus be seen as another instance where versatility with regard to word class adds to rhetorical vividness and, maybe, causes provocation.
4.
Conclusion and outlook
The aim of this paper was to account for the precategorial character of Late Archaic Chinese by integrating different approaches of Construction Grammar with pragmatic implicatures. For that purpose, I tried to combine aspects of Croft’s (2001) and Goldberg’s (1995, 2005) approach to Construction Grammar. Goldberg’s (1995, 2005) approach provided the theoretical background that I needed for the semantic definition of the argument structure construction and for the description of how the transitive argument structure construction can contribute an additional argument that is not profiled by the lexical root. Croft’s (2001) approach brought in the perspective of form that I needed as a basis for defining the syntactic framework (or the “scaffolding” in terms of Croft 2001: 238) which is crucial for the analysis of utterances in Late Archaic Chinese. Within that syntactic framework, lexical items get “coerced” into their function of N or V along the lines of Michaelis (2004). The integration of stereotypical pragmatic inferences in terms of Levinson (2000) turned out to be necessary because they govern the relation between the lexicon and the syntactic slots of N and V. To conclude this section with an outlook, I would like to briefly discuss (i) how precategoriality in Late Archaic Chinese differs from the use of nouns as verbs in English and (ii) what the consequences of precategoriality are from a wider perspective of linguistic theory. If one looks at examples such as (65) and (66) one may get the impression that English is not that different from Late Archaic Chinese: (65) Croft (2001: 71): We schooled him in proper manners. (66) Clark & Clark (1979: 783): My sister Houdini’d her way out of the locked closet.
83
84
Walter Bisang
This first-glance impression of precategoriality in English is not quite adequate. In the case of examples such as (65), Croft (2001: 71) convincingly argues that it is not possible to find a general semantic definition or a pragmatic rule which derives the semantics of school in its verbal use from its more frequent nominal use. Thus, the verbal meaning of school can only be accounted for in terms of conventionalization and thus in terms of lexical meaning. In Late Archaic Chinese, there are rules such as (25) from which the meaning of an object-denoting lexeme in the V-slot can be derived. Examples such as Houdini’d in (66) differ from (65) because it is highly implausible to attribute their meaning to the lexicon. Thus, their meaning must be inferred from context. However, finding the unique denotation of such a verb depends on mutual knowledge in a very broad sense. Just about any kind of mutual knowledge can become relevant for uniquely determining the denotation of a noun surfacing as a verb. The famous example of Houdini’d in (66) is only possible if the speaker believes that s/he and the hearer both know about Houdini’s reputation as an escape artist. In Late Archaic Chinese, the set of possible interpretations is more restricted. In an example like (58), the meaning of the construction in combination with the meaning of the proper name Wú wáng ‘king Wu’ is ‘you CONSIDER me to be king Wu’ (cf. Subsection 3.1). The part of mutual knowledge needed for (58) to be communicatively felicitous is then reduced to the fact that king Wu was murdered. In English, a simple sentence of the type You Houdini me sounds strange. As is pointed out by Clark & Clark (1979: 797), this is due to the fact that this sentence does not provide enough overt specification. Thus, spontaneous denominal verbs such as Houdini’d need more overt information to become interpretable in English. In Late Archaic Chinese no such information is needed because the construction itself provides a lot more information. Thus, the analogue to [You would] Houdini me in Late Archaic Chinese is perfectly interpretable in the sense of ‘You treat me like Houdini’. From the above discussion of denominal verbs in English, one can see that they differ from precategoriality in Late Archaic Chinese. There is a considerable number of denominal verbs such as school in (65) which belong to the lexicon. Even if there is a certain productivity of using nouns as verbs in English as in (66) with Houdini’d, there is no uniform account of their meaning in terms of a small set of constructions such as the transitive and the intransitive argument structure constructions in Late Archaic Chinese. If precategoriality is possible in a language, this is a challenge to the large number of theories that take preclassification for syntactic categories in the lexicon for granted. While data from English exclude precategoriality, data from Late Archaic Chinese open up a different perspective in which precategoriality seems
Precategoriality in Late Archaic Chinese
to work. The concentration of many theories on data from English leads to the broader context of the uniformity hypothesis (Baker 1991: 538–539) and its challenge by Bresnan (1994). The basic idea of this principle is that the underlying structure of languages typologically distant from English should resemble English surface structure. Bresnan (1994) very adequately describes the consequences of this principle for typologists in the following quotation: As a result of the categorial uniformity thesis, to many descriptive linguists and typologists the ‘abstract’ underlying structures adopted in current UG look disconcertingly concrete, imputing to typologically distant languages a universal underlying structure derived from the morphological and surface structural cat(Bresnan 1994: 72–73) egories of English and familiar European languages.
As I pointed out at the very beginning of this paper, precategoriality is of considerable relevance for our assumptions concerning the architecture of language and the interactions of components such as the lexicon, syntax, semantics/cognition, and pragmatics. Given the above problems with the uniformity hypothesis, it seems to be necessary for linguists to look at linguistic variety in a more openminded way without any prejudice about how the structure of a language has to be. This is a difficult task but it seems to be the only way to get a clearer idea about the architecture of language and about the way its components interact. As I tried to show in this paper, a combination of different approaches within the framework of Construction Grammar is a promising way of describing typologically remarkable structures. Of course, the fact that it was necessary to take elements from different approaches also shows that there is no overarching allencompassing theory of Construction Grammar available at the moment. In my view, the only way to come a step closer to such a theory is by integrating typologically marked phenomena from the world’s languages.
List of abbreviations A ATTR COP EQ EXCL FUT GEN INSTR INT
Actor Argument Attributive Marker Copula Equative Marker (cf. Footnote 10) Exclamative Future Marker Genitive Instrumental Intransitive (Construction)
85
86 Walter Bisang
N NA NEG NML OBJ:3 PF PFV POSS:3 PREP Q RECIP REL REL.NA S TOP TR U V
Noun Non-Agentive Negation Nominalizer Object Pronoun of 3rd Person Perfect Perfective Possessive Pronoun 3rd Person Preposition Question Marker Reciprocality Marker Relative-Clause Marker Relative-Clause Marker with Non-Agentive Coreference Single Argument of Intransitive Argument Structure Construction Topic Marker Transitive Argument Structure Construction Undergoer Argument (cf. Van Valin & LaPolla 1997, Van Valin 2005) Verb
References Baker, Mark (1991). On some subject/object non-asymmetries in Mohawk. Natural Language and Linguistic Theory 9, 537–576. Bisang, Walter (1998). Grammaticalization and language contact, constructions and positions. In A. G. Ramat & P. Hopper (Eds.), The limits of grammaticalization (13–58). Amsterdam & Philadelphia: John Benjamins. Bisang, Walter (Forthcoming a). Precategoriality and syntax-based parts of speech – the case of Late Archaic Chinese. In U. Ansaldo (Ed.), Parts of speech. Amsterdam & Philadelphia: John Benjamins. Bisang, Walter (Forthcoming b). Underspecification and the noun/verb distinction: Late Archaic Chinese and Khmer. In A. Steube (Ed.), The discourse potential of underspecified structures: Event structures and information structures. Berlin: Akademie-Verlag. Bisang, Walter (Forthcoming c). Word classes. In J.-J. Song (Ed.), The Oxford handbook of Language Typology. Oxford: Oxford University Press. Bresnan, Joan (1994). Locative inversion and the architecture of universal grammar. Language 70, 72–131. Clark, Eve V. & Herbert H. Clark (1979). When nouns surface as verbs. Language 55, 767– 811. Croft, William A. (1990). Typology and universals. Cambridge: Cambridge University Press. Croft, William A. (1991). Syntactic categories and grammatical relations. Chicago: The University of Chicago Press.
Precategoriality in Late Archaic Chinese
Croft, William A. (2000). Parts of speech as typological universals and as language particular categories. In P. Vogel & B. Comrie (Eds.), Approaches to the typology of word classes (65–102). Berlin: Mouton de Gruyter. Croft, William A. (2001). Radical Construction Grammar. Syntactic theory in typological perspective. Oxford: Oxford University Press. Croft, William A. (2003). Typology and universals. 2nd edition. Cambridge: Cambridge University Press. Downer, G. B. (1959). Derivation by tone-change in Classical Chinese. Bulletin of the School of Oriental and African Studies 22, 258–290. Fillmore, Charles J., Paul Kay & Catherine O’Connor (1988). Regularity and idiomaticity in grammatical constructions: The case of let alone. Language 64, 501–538. Gassmannn, Robert H. (1980). Das grammatische Morphem yě. Berne & Frankfurt: Lang. Gassmann, Robert H. (1988). Cheng Ming. Richtigstellung der Bezeichnungen. Zu den Quellen eines Philosophems im Antiken China. Ein Beitrag zur Konfuzius-Forschung. Berne & Frankfurt: Lang. Gassmann, Robert H. (1997). Grundstrukturen der antikchinesischen Syntax. Eine erklärende Grammatik. Berne & Frankfurt: Lang. Goldberg, Adele E. (1995). A construction grammar approach to argument structure. Chicago & London: University of Chicago Press. Goldberg, Adele E. (2005). Constructions at work. Oxford: Oxford University Press. Graham, A. C. (1957). The composition of the Gongsuen Long Tzyy. Asia Major 11, 147–183. Graham, A. C. (1964). The logic of the Mohist Hsiao-ch’ü. T’oung Pao 51, 1–54. Grice, Paul H. (1975). Logic and conversation. In P. Cole & J. L. Morgan (Eds.), Speech acts (41–58). New York: Academic Press. Haspelmath, Martin (1997). Indefinite pronouns. Oxford: Oxford University Press. Haudricourt, A.-G. (1954). Comment reconstruire le chinois archaïque. Word 10, 351–364. Hopper, Paul J. & Sandra A Thompson (1980). Transitivity in grammar and discourse. Language 56, 251–299. Kennedy, George A. (1964). Selected works of George A. Kennedy. New Haven: Far Eastern Publications. Kemmer, Suzanne (1993). The middle voice. Amsterdam & Philadelphia: John Benjamins. Langacker, Ronald W. (1987). Foundations of Cognitive Grammar, Vol. 1: Theoretical prerequisites. Stanford: Stanford University Press. Levinson, Stephen C. (1987). Pragmatics and the grammar of anaphora: A partial pragmatic reduction of Binding and Control phenomena. Journal of Linguistics 23, 379–434. Levinson, Stephen C. (1991). Pragmatic reduction of the Binding Conditions revisited. Journal of Linguistics 27, 107–161. Levinson, Stephen C. (2000). Presumptive meanings. The theory of generalized conversational implicature. Cambridge, Massachusetts: MIT Press. Li, Charles N. (1996). A cryptic language with a minimal grammar: The Confucian Analects of Late Archaic Chinese. In E. Weigand & F. Hundsnurscher (Eds.), Lexical structures and language use (53–118). Tübingen: Niemeyer. Li, Charles N. (1997). On zero anaphora. In J. Bybee, J. Haiman & S. A. Thompson (Eds.), Essays on language function and language type (275–300). Amsterdam & Philadelphia: Benjamins. Michaelis, Laura (2004). Type shifting in construction grammar: An integrated approach to aspectual coercion. Cognitive Linguistics 15, 1–67.
87
88
Walter Bisang
Michaelis, Laura A. & Knud Lambrecht (1996). Towards a construction-based theory of language function: The case of nominal extraposition. Language 72, 215–247. Nichols, Johanna, David A. Peterson & Jonathan Barnes (2004). Transitivizing and detransitivizing languages. Linguistic Typology 8, 149–211. Norman, Jerry (1988). Chinese. Cambridge: Cambridge University Press. Peyraube, Alain (1988). Syntaxe diachronique du chinois (évolution des constructions datives du XIVe siècle av J.-C. au XVIIIe siècle). Paris: Collège de France, Institut des hautes études chinoises. Sagart, Laurent (1999). The roots of Old Chinese. Amsterdam & Philadelphia: John Benjamins. Sasse, Hans-Jürgen (1993a). Syntactic categories and subcategories. In J. Jacobs, A. von Stechow, W. Sternefeld & T. Vennemann (Eds.), Syntax (646–686). Berlin: Mouton de Gruyter. Sasse, Hans-Jürgen (1993b). Das Nomen – eine universale Kategorie?. Sprachtypologie und Universalienforschung 46, 187–221. Schachter, Paul (1985). Parts of speech systems. In T. Shopen (Ed.) Language typology and syntactic description, Vol. 1: Clause structure (3–61). Cambridge: Cambridge University Press. Stassen, Leon (1997). Intransitive predication. Oxford: Clarendon. Van Valin, Robert D. Jr. (2005). Exploring the syntax-semantics interface. Cambridge: Cambridge University Press. Van Valin, Robert D. & Randy LaPolla. (1997). Syntax. Structure, meaning and function. Cambridge: Cambridge University Press. Wang, Li (1958). Hanyueyu yanjiu [Research on Sino-Vietnamese]. Hanyu shilun wenji, 290– 410.
chapter 5
Variations in Japanese honorification – deviations or a change in the making?*
Yoshiko Matsumoto Stanford University
1.
Introduction
Japanese is known to have a morphosyntactically and socially complex system of honorifics. Variations and instability in the use of Japanese honorifics have often been regarded as grammatical deviations produced by speakers who lack the appropriate linguistic and social training. A recent example is the variation in the uses of the productive form of non-subject honorific construction, o-V-suru. However, the variations in the use of o-V-suru form can be more systematically explained from the constructional and frame semantics perspectives as the results of a process of cognitive and intersubjective (e.g. Traugott and Dasher 2002) reorganization of grammatical constructions that is motivated by contextual conditions of use and the speaker’s intention with respect to the addressee. Honorific forms of Japanese predicates, after Harada’s generative analysis (1975), have generally been classified into propositional honorifics and performative honorifics, otherwise known as referent honorifics and addressee honorifics. The use of referent honorifics is licensed by the speaker’s relation to the referents of the subject and/or non-subject of the sentence, and the relation between the referents of the subject and some non-subject of the sentence; while the * I am very grateful for those who gave me comments on various occasions in which I presented versions of this paper, especially the audience at the International Cognitive Linguistics Conference (2001), the Annual Meeting of the Linguistic Society of America (2002) and the participants of the Symposium on Functional Approaches to Japanese Grammar: Toward the Understanding of Human Languages, at University of Alberta, as well as the participants of the workshop “Phylogeny and Ontogeny of the Written Language” at Kyoto Univeristy (2005). I am also indebted to Elizabeth Traugott for her discussion and an anonymous reviewer of this paper for valuable comments. My gratitude also goes to Jaakko Leino, the editor of this volume.
90 Yoshiko Matsumoto
use of performative honorifics depends on the context of speech, notably on the speaker’s relation to the addressee. Referent honorifics are further subcategorized into subject honorifics and nonsubject honorifics depending on whether the triggering referent is expressed as the subject or a nonsubject of the sentence. These syntactically-based classifications are not free of problems (see Matsumoto 1997 for details), but they highlight the two major types of honorific constructions (referent honorifics and addressee honorifics), which can be used independently or in combination, i.e. a predicate may exhibit referent honorifics only, addressee honorifics only, neither, or both. This paper focuses on one of the referent honorific forms, the nonsubject honorific construction, which is in the general form o-V-suru (‘Honorific Prefix – Verb (stem) -do’) composed of an honorific prefix (HP) o, a verb stem, and the verb suru meaning ‘do’. This is one of the forms for which variations and changes in use have been most commonly noted in newspaper articles and in popular books on the “correct” use of honorifics. What is “honorified” in the conventional use of this construction is a nonsubject referent such as the direct object referent, while the subject referent is downgraded. For example, in (1), the teacher, the object referent of the honorific verb o-tasuke-sita ‘helped (NSH: Non-Subject Honorific)’ [sita is the past form of suru ‘do’], is elevated while the subject referent, Abe-san, is humbled. (2) is a similar non-honorific sentence, given as a contrast. In (2), neither the subject nor object referent is elevated or downgraded. (1) Abe-san ga sensei o o-tasuke-sita. Mr. Abe NOM teacher ACC HP-help-did ‘Mr. Abe helped (Non-Subject Honorific: NSH) the teacher.’ (2) Abe-san ga Oda-san o tasuketa. Mr. Abe NOM Mr. Oda ACC helped ‘Mr. Abe helped Mr. Oda.’
Recently, numerous uses of the o-V-suru form that deviate from this conventional description have been attested. One such example is given in (3), which was announced by a pilot on an international flight (Matsumoto 1997). (3)
ato go-zikan hodo de hizuke-henkoo-sen o more 5-hours about in international-dateline ACC o-mukae-itasi-masu. HP-meet-do.HUMBLE-ADD.HON ‘(We) will meet (NSH) the international dateline in about 5 hours.’ [pilot on an international flight]
Variations in Japanese honorification
The part in bold face, o-mukae-itasi-, is an example of a nonsubject honorific form in which -itasi- is the stem of itasu, a humble form of the verb suru ‘do’, and the underlined masu is an addressee honorific suffix. From the conventional description of nonsubject honorifics, the sentence would be anomalous because the object of mukae- ‘meet/reach’ – i.e. the‘international dateline’ – can hardly be the target of honorification. The only apparent potential target of honorific use in this discourse context is the addressees, i.e. the passengers on the airplane (for whom the addressee honorific masu is used as shown in the example), but they are not a participant of the described event. Another example that does not fit into the conventional use even more strikingly is given in (4). (4) ...soo site-itadaku to o-mati-suru o-zikan mo arimasen kara… so do-receive.HUM if HP-wait-do HP-time also not.exist.AH so ‘…if you could do so, there would be no time for you (= you don’t need) to wait (NSH) (for us), so…’ [a delivery service clerk to a customer]
In this example uttered by a delivery service clerk to a customer, the subject referent of the o-V-suru construction o-mati-suru is the customer, whereas the nonsubject referent is the speaker. This use is deviant from the normative perspective since the speaker, the referent of a nonsubject, is raised, while the subject referent, the customer, is downgraded. Examining variants such as (3) and (4) above, which are attested in spoken and written discourse, I argue in this paper that the nonsubject honorific construction in Japanese is undergoing a change in usage and is gradually becoming a performative honorific. This change in progress, I contend, reflects an intersubjective re-organization of the relevant participants of the grammatical construction o-V-suru from the participants in the described event to the participants in the speech context. I have previously observed (Matsumoto 1997) that, contrary to earlier syntactic and structural accounts, what is considered to be the conventional (or normative) use of the nonsubject honorific construction o-V-suru1 is subject to a pragmatic condition of benefit transfer between the two referents (subject and nonsubject), rather than being controlled by the grammatical roles. Building upon that observation and building also on Toshiki Tsujimura’s observation (1992) that speakers tend to present their actions as benefiting the addressee, I will argue that,
1. The history of the modern form o-V-suru is relatively recent. The form is considered to be established (i.e. accepted as a normative form of a ‘humble’ form) in the very beginning of the 20th century (e.g. Komatsu 1967).
91
92
Yoshiko Matsumoto
in a frame of a polite discourse, the targeted referents of the nonsubject honorific construction o-V-suru are re-organized from referents of the subject and the nonsubject in the described event to the two participants of the discourse, the speaker and the addressee. This type of shift from the domain of the sentence describing an event to the speech context has also provided a systematic explanation to various other linguistic phenomena. For example, in “speech act causality” discussed by Sweetser (e.g. 1990: 77) in such a sentence as What are you doing tonight, because there’s a good movie on, the because-clause gives the “cause” of the “speech act” embodied by the main clause, instead of the “real-world” causality. The use of a degree adverb tyotto ‘a little’ (originally meaning ‘dust’) in Japanese in a sentence such as Tyotto nakunarimasita mon de ‘It’s that (he) tyotto passed away’ qualifies the speech act of asserting the content of the sentence, rather than the proposition expressed by the main verb of the sentence, i.e. that he died (Matsumoto 1985, 2001). Other obvious cases of this semantic change and cline of grammaticalization are phenomena that have been discussed as examples of shifts from content to subjectivity to intersubjectivity, including similar drift in usage from referent to addressee honorifics in other Japanese predicate honorific constructions (Dasher 1995; Traugott and Dasher 2002). In order to elaborate on this point of semantic change, I will illustrate three main uses of the o-V-suru construction and demonstrate, modeling on ideas from Frame Semantics (Fillmore 1977, 1982, etc.) and Construction Grammar (Fillmore, Kay and O’Connor 1988; Goldberg 1995; Östman and Fried (eds.) 2005), that the currently observed increase in use of the construction as an addressee honorific is in fact predictable rather than unsystematic or deviant.
2.
O-V-suru construction: Use 1 – Nonsubject honorifics
Before I discuss the variations in usage, I will first provide the background of the current discussion by describing previous research on Japanese honorifics, especially in terms of what or who are considered to be the targets of honorification. I will then focus on the conditions for the use of o-V-suru construction, drawing attention to a pragmatic condition that is crucial to the explanation of the conventionally accepted usage. There have been numerous studies treating the system and the history of Japanese honorifics in the frameworks of traditional Japanese grammar and ethnomethodology (e.g., Yamada 1924; Tokieda 1941; Tsujimura 1967, 1968; Hayashi and Minami 1973–74; Kokuritsu Kokugo Kenkyujo (The National Language Research Institute) 1957, to name a few). In the generative framework, Harada’s (1976) seminal work brought to the fore syntactic aspects of honorification and
Variations in Japanese honorification
moved the discussion of honorifics into the arena of modern grammatical theories. Harada classified honorific forms into subject honorifics (SH), object honorifics (OH), and performative honorifics (PH). In Harada’s account, subject honorifics are used when the subject refers to a “person socially superior to the speaker (SSS)” while object honorifics are used when the indirect or direct object refers to an SSS. Subject honorifics and object honorifics are grouped into one main category “propositional honorifics” (also known as “referent honorifics”). Performative honorifics (also known as “addressee honorifics”) are distinguished by the fact that the presence of an SSS in the propositional content of the sentence is not required; they include different speech levels that vary with respect to the interlocutors and the situation of the speech. The straightforward syntactic description that Harada offered could be understood almost as a type of agreement phenomenon. Although such a description is problematical from the pragmatic and sociolinguistic points of view, it was valuable in pointing out that honorifics have grammatical consequences rather than simply being elegant lexical items. It has also been widely accepted as part of the basic understanding of the Japanese honorifics system and has led some linguists to consider SH as providing a useful test for subjecthood (e.g., Shibatani 1990). Harada’s account may be summarized graphically as in Figure 1 (in which the arrow denotes the direction of “social superiority”, in Harada’s terms). For object honorifics, however, Harada’s account has been regarded as less accurate (e.g. Kikuchi 1980; Kuno 1983; Martin 1975; McCawley 1993) in that the target NP need not be limited to a direct or indirect object but can in principle be any nonsubject. In this regard, Kuno’s (1987) terminology “nonsubject honorifics” captures more accurately the fact that such an honorific “elevates the referent of a nonsubject element and downgrades that of the subject.” This present paper also adopts Kuno’s terminology in referring to the o-V-suru construction as a nonsubject honorific construction. While Harada speaks of the honored party (which I shall call the “target of honorification”) as a person “socially superior to the speaker,” Kuno’s description is in terms of the speaker’s presentation of the
Figure 1. Harada’s Account of Honorifics in Japanese
93
94 Yoshiko Matsumoto
Figure 2. Kuno’s Account of Non-Subject Honorifics
relation between the referent of the subject and that of a non-subject. This is also more accurate in that it captures the subjective nature of honorific uses rather than implicating the predetermined social hierarchy. Apparently implicit in Kuno’s account, however, is the assumption that the speaker also is presented as in a lower position relative to the target of honorification. Kuno’s approach may be expressed graphically as in Figure 2. (The arrow in brackets pointing down next to the subject referent indicates the downgrade effect of the honorifics.) Kuno, however, does not explain which referent among the many possible semantic and grammatical roles can license a specific nonsubject honorific construction. In order to account for such unanswered questions, I re-examined the o-Vsuru construction elsewhere (Matsumoto 1997) and found noteworthy properties including the following: in the conventional use of the o-V-suru form, there is an implied relationship of benefit between the two (typically, human) participants, in which the exalted nonsubject referent is either the beneficiary or the source of benefit, depending on the predicate and on the context. The determination of which non-subject referent may license nonsubject honorifics depends, therefore, not only on the speaker’s decision to elevate one of the nonsubject referents but also on the consonance of the action or state that is described in the sentence with a pragmatic notion of benefactivity involving such nonsubject referent. The importance of benefactive relationship between the subject referent and the honorific target of the nonsubject referent is illustrated by the contrasts in the (a) and (b) sentences of examples (5) and (6). (5) a. Abe-san ga sensei o o-tasuke-sita. (= (1)) Mr. Abe NOM teacher ACC HP-help-did ‘Mr. Abe helped (Non-Subject Honorific: NSH) the teacher.’ b. # Abe-san ga sensei o o-korosi-sita. Mr. Abe NOM teacher ACC HP-kill-did ‘Mr. Abe killed (Non-Subject Honorific: NSH) the teacher.’ (6) a. Abe-san ga sensei kara hon o o-kari-sita. Mr. Abe NOM teacher DAT book ACC HP-borrow-did ‘Mr. Abe borrowed (Non-Subject Honorific: NSH) a book from the teacher.’
Variations in Japanese honorification
b. # Abe-san ga sensei kara hon o o-nusumi-sita. Mr. Abe NOM teacher DAT book ACC HP-steal-did ‘Mr. Abe stole (Non-Subject Honorific: NSH) a book from the teacher.’
The first of each pair, i.e., (5a) and (6a), contains a verb that is semantically benefactive, whereas verbs in the second example of the pair, i.e., (5b) and (6b), denote counter-benefactive actions. Example (5b) and (6b) with counter-benefactive expressions are normally not acceptable without a special context. This unacceptability or the rarity of use is also borne out in an Internet search, which I found not to yield such an instance, although this does not mean that sentences with o-korosi-suru are absolutely impossible. A similar Internet search readily finds examples with verbs of benefactive actions such as (7) and (8).2
(7)
Bannen Sonbun o o-tasuke-siteita koto kara mo, tyuugoku later years Sun-Wen ACC HP-help-did.PROG fact from also China to no kankei o omonjiteita koto ga wakarimasu. with GEN relationship ACC was-valuing fact NOM understand ‘The fact that [my grandfather] was valuing the relationship with China could also be seen from the fact that (he) was helping (NSH) Sun Wei in his later years.’ [excerpts from a monthly magazine Seiron]
(8) HP yoo sozai o o-kari-siteimasu. home page use materials ACC HP-borrow-do.PROG.AH ‘(We) have borrowed (NSH) materials for home pages.’ [a web page description]
The difference between (a) and (b) sentences in (5) and (6) can be explained by the fact that a benefactive relationship between a killer and the person killed as in (5b), for example, is very difficult to conceive of. Only in a situation such as euthanasia where the killing of the professor was intended to be for the benefit of the professor, or where it was done for the benefit of some other person to whom respect should be expressed, would (5b) be acceptable. A similar contrast is also observed between (6a) and (6b), although in (6a), the exalted target is the source of the benefit rather than the goal of the benefactive action. I should point out that (5b) and (6b) should be perfectly acceptable if nonsubject honorification were conditioned simply on the target of the action (the direct object referent and indirect object referent, respectively) denoting a socially superior person, as Harada’s account claims, irrespective of any benefactive relation 2. In contrast, there are many sentences with o-tasuke-suru and o-kari-suru. The web search of o-tasuke-simasu(HP-help-do.AH), for example, yielded 3,330 cases.
95
96 Yoshiko Matsumoto
between the subject referent and the target of honorification. Likewise, an analysis based on the case marking of the targeted NP, or one based on whether the NP is an argument of the verb (Kikuchi 1980; Mori 1993) cannot explain the pragmatic unacceptability of (b)-sentences in (5) and (6), since sensei is marked by the case marker o in (5b) and by kara in (6b), and is an argument of the verb, each of which is treated in the above-cited accounts as a sufficient condition for the acceptability of NSH. It is likely that the judgment as to whether a certain act is benefactive depends on the speaker’s subjective judgment, but the above examples provide clear cases for whether or not such subjective notion of benefaction is crucial.3 In examples (5) and (6), the target of honorification, i.e. a nonsubject, is a core argument of the main verb expressed as the direct and indirect object, respectively. It is worth noting, however, that the nonsubject target can be a non-argument as illustrated in (9), an example from Kuno (1987).
(9) Kopii-dai wa watasi ga tyokusetu kaikei ni o-harai- simasu. copying-fee TOP I NOM directly cashier DAT HP-pay- do.AH ‘I will pay (NSH) the copying fee directly to the cashier.’ [Example from Kuno 1987]
Kuno explains that sensei no kawari ni ‘instead of the teacher’ is implied (by the use of tyokusetsu ‘directly’) and that sensei ‘teacher’ is the target of respect. This can also be interpreted as representing a benefit relation between the subject referent, here the speaker, and the teacher, the exalted nonsubject referent. Similarly in (10), an example from a web page dedicated, according to the writer, to beginners of aromatherapy, the target of the honorification is an implicit nonargument referent, namely, the neophyte aromatherapists. (10)
… aroma terapii no kihontekina koto o o-kaki-sinakereba aroma therapy GEN basic thing ACC HP-write-do.NEG.COND toyuu simeikan ni karare,… COMP mission.sense BY be.driven ‘… being driven by the sense of mission that (I) should write (NSH) the basics of aromatherapy…’ [web page for aromatherapy novices]
These examples show that, regardless of the grammatical role that the target nonsubject takes, an elevated referent is understood to be in a benefit transfer relation with the subject referent in the o-V-suru construction. 3. It is conceivable that the property of benefit transfer between the subject and the target non-subject referents may be a later development in the usage of the o-V-suru form. This presents an interesting point from the perspective of semantic change and suggests further investigation of diachronic data.
Variations in Japanese honorification
In summary, the basic conditions for the conventional use of the non-subject honorific o-V-suru construction are that (i) the non-subject referent of the honorific verb is the target of the honorification while the subject referent as being relatively humbled, and (ii) there is a relationship of benefit transfer between those two referents in the described event. In this conventional use, referents of the subject and non-subject are not necessarily the speaker and the addressee. As discussed below, however, this last condition has been shifting.
3.
O-V-suru construction: Direction toward performative honorifics
3.1
Use 2: Subject Referent ≈ Speaker; Non-Subject Referent ≈ Addressee
Although Kuno did not explicitly provide a description of the discourse context for (9), it is likely that the sensei ‘teacher’ was the addressee. The same can be said for (10), since the aromatherapy novices are also the targeted readers (addressees) of the web page. In other words, the humbled subject referent is the speaker and the target of the benefit transfer and of honorification is the addressee. This is well-motivated since the most important and inescapable relation in a speech context is that of the speaker with the addressee. There are numerous instances in which this condition holds. (11) and (12) are attested examples in which the referent of the indirect object is the addressee and the subject referent is the speaker. Researchers such as Dasher (1995) consider the examples of this kind to be prototypical cases of the nonsubject honorific expressions. (11) anoo, tyotto o-tazune-simasu ga… um a.little HP-ask-do.AH but ‘(lit.) um, (I) have a question to ask (NSH) (you), …’ [at a department store] (12) zyaa atode ronbun o-okuri-simasu then later academic-article HP-send-do.AH ‘(I) will send (NSH) (you) (my) article later, then.’
[after a colloquium] There are also instances like (9) in which the (elevated) addressee is a nonargument of the verb in the construction, and therefore is less directly involved in the described event. (13) o-syooyu o syoosyoo o-ire-itasimasu. HP-soy sauce ACC a.little HP-pour in-do.HUM.AH ‘(I) pour in (NSH) a little soy sauce (to something that is being cooked).’ [cooking instructor in a TV cooking program; cited in Tsujimura 1992]
97
98 Yoshiko Matsumoto
(14) Nihon no kyuuka o o-tasuke-simasu. Japan GEN vacation ACC HP-help-do.AH ‘(The Association of National Park Resort Villages) will help (NSH) Japanese vacation.’ [Description of a web link to the Association of National Park Resort Villages] (15)
ato go-zikan hodo de hizuke-henkoo-sen o more 5-hours about in international dateline ACC o-mukae-itasimasu. HP-meet-do.HUM. AH ‘(We) will reach (NSH) the international dateline in about 5 hours.’ [pilot on an international flight]
(13) is an example cited by Tsujimura (1992: 478) as an utterance by a cooking instructor in a TV program. The semantic goal of pouring is the dish that the instructor (i.e. the speaker) is cooking, which would not be a usual trigger for honorifics. Tsujimura explains that the speaker used the form to imply that the pouring of soy sauce is done for the benefit of the addressees, the TV viewers. It is doubtful whether the viewers could be expressed as a nonsubject in the sentence and maintain the naturalness of speech (i.e. to say “I’m pouring in a little soy sauce for you” in this context would be unnatural) but such intention of benefit is implicit in the nature of the program in which the cooking demonstration is done for the benefit of the viewers and the speech context is normally intended to be refined and polite. Similarly in (14), an example from a web page, the prospective vacationers at the National Park Resort Villages will benefit from the link and are the target of the honorification, although they are the nonargument referents, not explicitly mentioned in the sentence. The sentence in (15), which repeats example (3), was an announcement made by the captain of an international flight. Under the analysis of the conventional use of the o-V-suru construction given in the last section, it should be unacceptable, since the only explicit nonsubject referent, the international dateline, is an unlikely target of benefit transfer and therefore of honorification. The only potential target of the honorification in the given context is the passengers, but their involvement in the described event, i.e. meeting the international dateline, is minimal and therefore the benefit of such an event to them is less evident than in the other two examples. For this reason, the sentence probably strikes normative users of the construction as even more odd than the other two examples. However, it is clear in these examples that the speaker is being humbled and the addressee (the reader) is elevated as the effect of the use of the o-V-suru construction in the
Variations in Japanese honorification
speech context, where the speaker wishes to be polite and to imply some sort of benefit relation with the addressee. This observation leads us to the third and currently controversial use of the construction.
3.2 Use 3: Addressee Honorific Use (Subject Referent ≈ Addressee) Examples (9) and (13)–(15) provided a glimpse of a direction of drift in the use of the o-V-suru construction in that the speaker and the addressee’s benefit relation is more prominent than that between the subject and nonsubject referents. In these examples, the speaker was denoted or assumed as the humbled subject referent, as the conventional use also predicts. However, there are currently attested instances in which the subject referent is the addressee. That is, the addressee, the intended target of honorification appears in the subject position, and the term nonsubject honorification is no longer descriptive. Examples such as (16)–(21) can easily be regarded as deviations from the normative point of view. On the other hand, similar usage was also observed by Hudson (1999), who pointed out the fact that o-V-suru forms are sometimes used as hyper-polite forms, and by the report of the Council of the National Language (1996), which noted that more than 40% of respondents in a survey judged examples such as these to be acceptable. This tendency is also supported by relatively common use (about 40 to over 200 cases depending on the expression) of such variations found on a recent (2003) web search. (16) Kotira de o-mati-site-kudasai. here LOC HP-wait-do-give.IMPERATIVE.AH ‘Please wait (NSH) here.’ [a receptionist to a guest] (17) O-motikaeri-simasu ka HP-take out-do.AH QUESTION PARTICLE ‘(lit.) Are (you) taking it out (NSH)? / Is this for takeout?’ [an attendant at a fast food store to a customer; cited in Asahi Newspaper] (18) …soo site-itadaku to o-mati-suru o-zikan mo arimasen kara… so do-receive.HUM if HP-wait-do HP-time also not.exist.AH so ‘…if you could do so, you would not have time to wait (NSH) around, so…’ [a delivery service clerk to a customer] (19) Nori o o-nose-site mesiagatte kudasai. dried.sea.weed ACC HP-place.on.top-do eat.SH give.me.AH ‘Please place (NSH) the dried sea weed on top before you eat (SH) (it).’ [waitress to diners at an upscale Japanese restaurant]
99
100 Yoshiko Matsumoto
(20) ...eto, o-kotae-dekiru han-i de kamawanai n desu ga… um HP-answer-do extent within fine. with NMLZ is.AH but‘ ‘… um, it’s OK if you could answer (NSH) to the extent that you feel comfortable…’ [a college student to an elderly person, prefacing her question about the interviewee’s personal background] (21)
OHP o o-tukai-suru n desu ne OHP ACC HP-use-do NMLZ is.AH Sentence Final Particle ‘(You) are going to use (NSH) an OHP, aren’t you?’ [a graduate student to a professor, an invited guest speaker at a graduate colloquium]
It is interesting to note that these examples were often attested in the context of the general relation of service provider to a client (i.e. (16)–(19)). In all examples (16)–(21), it still remains true that the o-V-suru construction is used in the context in which the speaker intends to be humble and polite to the addressee and the speaker and addressee are engaging in or implying a benefit transfer relation at the occasion of communicating the described event. In this sense, the pragmatic properties of the o-V-suru construction are still maintained.
4.
Conclusion
The differences among the three stages of the use of o-V-suru construction are schematically given below. The inner box represents the event described by the construction focusing on the two participants crucial in the construction, while the outer box represents the frame of the speech context, namely, the politeness context with the two participating parties, i.e. the speaker and the addressee. The first diagram illustrates the conventional use (Use 1) of o-V-suru construction as a nonsubject referent honorific, the second one (Use 2) shows the middle stage, so-to-speak, at which the subject referent of the described event is identified with the speaker and the addressee is identified with the explicit or implied target of benefit transfer in the event. In this sense, Use 2 can be seen as a special case of Use 1 in which there is particular attention to the speech event and its participants. The benefit transfer in this use is not simply relevant in the described event (i.e., the propositional content of the utterance), but also relevant at the level of speech context. In the third diagram (Use 3), however, the identification between the speaker and the subject referent, and between the nonsubject referent and the addressee is not maintained and the benefit transfer relation is re-analyzed to be relevant only between the speaker and the addressee, the two prominent par-
Variations in Japanese honorification 101
Politeness Speech Context
Figure 3. Use 1.
Politeness Speech Context
Figure 4. Use 2.
Politeness Speech Context
Figure 5. Use 3.
102 Yoshiko Matsumoto
ticipants of the speech context rather than of the described event. Therefore, this diagram represents what is in effect a performative or addressee honorific. The benefit is relevant at the level of communication (performance), not at the level of the propositional content. Although these three uses are concurrently found at present, abundant instances of the third use have been noticed most recently and are treated in how-to books as common mistakes of honorific uses. The currently observed tendency is that the profiled referents of the o-V-suru construction are cognitively shifting from the subject and nonsubject referents of the sentence describing the event to the two prominent participants of the discourse, the speaker and the addressee. Considering that the speaker’s relation to the addressee is the most relevant factor especially in a context of polite speech, such a direction of intersubjective development seems well motivated. Further, as Tsujimura (1992) suggests, it is understandable that speakers tend to present their actions as benefiting the addressees in a politeness context. The direction of diachronic change in honorifics from the referent to addressee honorifics has been pointed out by researchers including Tsujimura (1992, etc.), Dasher (1995), and Traugott and Dasher (2002). For example, the addressee honorific suffix -masu originated in a humble referent honorific. Interestingly, in his analysis of honorific grammaticalization, Dasher (1995) states that the o-V-suru form does not participate in this direction of change. Although the use of o-V-suru is not identical to the use of -masu in that it is not a suffix and that there should be benefit transfer relation between the speaker and the addressee, the data and analysis that I provided in this paper offer an addition to the previous findings by demonstrating that the general direction of development and change is in fact manifested in the o-V-suru construction when analyzed from the point of view of intersubjectification of the grammatical construction. The diagrams of the three uses of the o-V-suru construction may be identified with representations of the “cognitive frame” and “interactional frame” of the utterances (e.g. Fillmore 1982). Interactional frames represent the conceptualization of the discourse situation between the speaker and the addressee: from knowledge of deictic categories to knowledge of discourse genres. The interactional frame illustrated as the outer box in the diagram is a very general one– the politeness speech context. The inner box in the diagram illustrates the cognitive frame, which represents the event described by the utterance. The advantage of using both types of frames to describe the o-V-suru construction is the capability of illustrating the relation between the contextual participants and the event participants, and how these two interact, which is schematically indicated by the connecting lines between them. The diagrams may be made more formal to accommodate AVMs (Attribute Value Matrices) or made to have separate boxes for interactional frames. What is important here is that notions available in Frame
Variations in Japanese honorification 103
Semantics and Construction Grammar can well accommodate contextually dependent constructions, such as honorifics constructions, and their variations and diachronic changes, to be systematically accounted for. The findings in this light suggest that there is importance and great possibility in pushing forward Construction Grammar to the direction of extending its research to include more studies of grammaticalization.
References Bunka cho (Bureau of Culture) (1996). Kokugo Shingi-kai Hookokusho (Report of Council of the National Language) 20. Tokyo: Ministry of Finance. Dasher, Richard (1995). The Grammaticalization in the System of Japanese Predicate Honorifics. Unpublished Ph.D. dissertation. Stanford University. Fillmore, Charles J. (1982). Frame semantics. In Linguistic Society of Korea (Ed.), Linguistics in the Morning Calm (111–138). Seoul: Hanshin Publishing Co. Fillmore, Charles J. (1985). Frames and the semantics of understanding. Quanderni di Semantica VI. 2. 222–254. Fillmore, Charles, Paul Kay and Catherine O’Connor (1988). Regularity and Idiomaticity in Grammatical Constructions: The Case of let alone. Language 64: 501–538. Goldberg, Adele (1995). Constructions: A Construction Grammar Approach to Argument Structure. Chicago: University of Chicago Press. Harada, Shin-ichi (1976). Honorifics. In M. Shibatani (Ed.), Syntax and Semantics 5: Japanese Generative Grammar (499–561). New York: Academic Press. Hayashi, Shiro and Fujio Minami (Eds. ) (1973–1974). Keigo Kooza 1–10 (Lectures on Honorifics 1–10). Tokyo: Meiji Shoin. Hudson, Mutsuko Endo (1999). Teinei-hyoosiki tosite no kenzyoo-doosi (Japanese humbling verbs as politeness markers). In Yukiko Sasaki Alam (Ed.), Gengogaku to Nihongo Kyooiku (Linguistics and Japanese Language Education) (259–274). Tokyo: Kurosio Syuppan. Kikuchi, Yasuto (1980). Jooge-taiguu-hyoogen no kijutu (Description of expressions for the treatment of status distinctions). Kokugogaku (Japanese Linguistics) (39–54). Kokuritsu Kokugo Kenkyujo (The National Language Research Institute) (1957). Keigo to Keigo-ishiki (Honorifics and the Attitudes Toward Honorifics). Tokyo: Shueido Shuppan. Komatsu, Hisao (1967). ‘O – suru’ no seiritu (Establishment of ‘o – suru’). Kokugo to Kokubugaku (National Language and National Literature) (93–101). Kuno, Susumu (1983). Sin Nihon-bunpoo Kenkyuu (New Studies in Japanese Grammar). Tokyo: Taishukan. Kuno, Susumu (1987). Honorific marking in Japanese and the word formation hypothesis of causatives and passives. Studies in Language 11–1: 99–128. Langacker, Ronald W. (1990). Subjectification. Cognitive Linguistics 1: 5–38. Martin, Samuel (1975). A Reference Grammar of Japanese. New Haven: Yale University Press. Matsumoto, Yoshiko (1985). A sort of speech act qualification in Japanese: chotto. Journal of Asian Culture 9: 143–159. Matsumoto, Yoshiko (1997). The rise and fall of Japanese nonsubject honorifics: The case of ‘o-Verb-suru’. Journal of Pragmatics 28: 719–740.
104 Yoshiko Matsumoto
Matsumoto, Yoshiko (2001). Tyotto: Speech act qualification in Japanese revisited. Japanese Language and Literature 35. 1: 1–16. McCawley, James D (1993). A Japanese and Ainu linguistic feast (Review of Shibatani 1990). Journal of Linguistics 29: 469–484. Mori, Junko (1993). Some observations in humble expressions in Japanese: Distribution of o-V(stem) suru and V(causative) itadaku. In Soonja Choi (Ed.), Japanese/Korean Linguistics 3 (67–83). Stanford: Stanford Linguistics Association. Östman, Jan-Ola (2000). Construction Grammar meets the Postcard frame. Paper presented in the panel on “Pragmatic aspects of frame semantics and construction grammar,” at the 7th International Pragmatics Conference. Budapest, Hungary. Östman, Jan-Ola & Mirjam Fried (Eds.) (2005). Construction Grammars: Cognitive grounding and theoretical extensions [CAL 3]. Amsterdam: John Benjamins. Shibatani, Masayoshi (1990). The Languages of Japan. Cambridge: Cambridge University Press. Sweetser, Eve E. (1990). From Etymology to Pragmatics. Cambridge University Press. Tokieda, Motoki (1941). Kokugogaku Genron (The Theory of Japanese Linguistics). Tokyo: Iwanami Syoten. Traugott, Ellizabeth Closs (1989). On the rise of epistemic meanings in English: An example of subjectification in semantic change. Language 57: 33–65. Traugott, Ellizabeth Closs and Richard B. Dasher (2002). Regularity and Semantic Change. Cambridge: Cambridge University Press. Tsujimura, Toshiki (1967). Gendai no Keigo (Modern Honorifics). Tokyo: Kyobunsha. Tsujimura, Toshiki (1968). Keigo no Sitekikenkyuu (Historical Studies on Honorifics). Tokyo: Tokyo-do Shuppan. Tsujimura, Toshiki (1992). Keigo Ronkoo (Essays on Honorifics). Tokyo: Meiji Shoin. Yamada, Yoshio (1931/1924). Keigohoo no Kenkyuu (The studies of Honorification). Tokyo: Hobunkan.
chapter 6
Constructing reasoning The connectives för att (causal), så att (consecutive) and men att (adversative) in Swedish conversations Jan Lindström and Anne-Marie Londen Department of Scandinavian languages and literature / University of Helsinki
1.
Introduction
The purpose of this paper is to explore the constructional and interactional motivation of the sentence connectors för att ‘for (that), because’, så att ‘so (that)’ and men att ‘but (that)’ in Swedish with a special reference to their distribution and functions in conversational language. These connectives may be regarded as complex variants of corresponding simple forms så, för, and men. Indeed, one of our research questions is why the general complementizer att ‘that’ is frequently, especially in speech, combined with the simple forms even though it does not add any obvious functional value to the expression. We will argue that the frequency of the complex ‘that’-forms can be explained by a general sequentially linking role of att. Another question concerns the choice between the alternative paratactic and hypotactic construction methods available for för att and så att. Coordination is clearly favoured in conversational language, whereas the connectives are usually subordinating in the written standard. As regards men att, it does not occur in writing and it is overlooked in most grammars. We will show that semantic/pragmatic implications differentiate paratactically coded clausal units from hypotactic constructions. Moreover, a less categorical distinction between what is subordination and coordination is proposed. We will combine a grammatical and discoursal examination of empirically manifest linguistic data, inspired particularly by conversation analysis. This is a necessity, since linguistic forms are both abstract and concrete entities. A particular linguistic form is produced in a particular context of use with a more or less particular function. But a particular linguistic form is also an instantiation of an
106 Jan Lindström and Anne-Marie Londen
abstract form–function schema with a more or less stable function and potential to occur in contexts of use that may seem predictable. Form–function schemas tend to be systematically organized in the language, in that functionally related schemas share certain features in their concrete instantiations; i.e., the linguistic form coheres with linguistic meaning in a systematic, albeit complex, fashion. These matters constitute the essence of the theoretical framework of construction grammar (e.g. Michaelis & Lambrecht 1996; Fillmore et al. 1988; Fried & Östman 2004). We will hence use construction grammar (CxG) as an explanatory tool by which we attempt to relate our findings of concrete language use to a description of the general formal and pragmatic patterns, i.e. the grammar, associated with the connectives to be studied here. The analysis is predominantly synchronic, but also takes some diachronic developments into account to illuminate the cognitive (re)organization of these linguistic resources. Our point of departure is that by studying conversational language, we are not only studying clauses and sentences, but real utterances and actions that occur in a sequence of turns-at-talk. While it is important to account for the internal, in a linguistic sense traditional, syntax, we also need to account for at least certain systematic aspects of the sequential utterance structure (see Linell 2005, for such an analytical perspective). This approach is also compatible with the tenets of construction grammar and it provides relevant gains in a description of generalized constructions, which eventually have concrete instantiations on the level of sounds, constructs (including phrases, clauses, sentences), actions and action sequences. From a semantic point of view, causal and consecutive relations can be seen as mirror-imaged complements to each other. A causal clause states the reason for an event (or for a ‘result’ which is usually described in a prior clause) (1a), while a consecutive clause states the result of an event (which then has caused the result) (1b): (1) a. John got a raise because he works hard. b. John works hard so he got a raise.
It is thus motivated to combine the study of causal and consecutive connectives. The connective men att is in a sense adversative but has also a distinct consecutive function in discourse sequences, as will be shown later. All the three connectives also co-occur frequently in colloquial spoken Swedish. Så att has the greatest general frequency whereas för att has a richer variety of discourse functions, which is why it deserves extra attention in this study. Men att is a regional form, particularly common in spoken Swedish in Finland.
Constructing reasoning 107
The following excerpts (2)–(4) illustrate uses of the connectives that are characteristic of conversational language.
(2) GRIS:GML:4 ”Genetically boosted crops”. 01 H: då kan vi börja diskutera .hhh liksom eh: att f then we can begin to discuss like that 02 ja öhm att vi har dålit mä mat för de vi behöver yeah that we have a shortage of food therefore we need
→ 03 04
framställa to produce i Ryssland in Russia
mer mat. för att ↓eh↓ de finns ju masser (0.2) more food för att there are prt lots of å: (0.4) runtikring (0.2) där de k- kan odlas and around where you can cultivate
(3) GRIS:TT:3 ”A rock video”. 01 J: .hh >halvvägs igenom så märker att ja sitter halfway through (I) notice that I sit 02 tittar breve å e sär< va a: just ja watch beside and am like what yeah that’s right → 03 visst ja videon. .hh satt de de kan ju nte va of course the video så att that can prt not be 04 särskilt gott betyg åt själva videon particularly good grade for the video itself (4) HUSA:08 “Branded clothes”. 01 I: hu kan man se de¿ how can one see it 02 B: $på märkeskläder$ å, on branded clothes and 03 (0.4) 04 M: [jå yeah 05 B: [oftast har dom den där (.) salkkun i handen å $hh$ most often they have that briefcase in the hand and 06 I: å pärlhalsbande¿ and the pearl necklace 07 B: jå: pärlhalsbande å, (.) sen ibland finns dedä NMT yeah the pearl necklace and then sometimes there’s this NMT 08 ännu i fickan $å eh[eh$ also in the pocket and
108 Jan Lindström and Anne-Marie Londen
09 I: [jå yeah → 10 B: men att, märkeskläder de e- (.) de brukar nu alla ha. men att branded clothes that’s what they prt all usually have
All the focused connectives introduce a clause which is not syntactically or semantically subordinated to an assertion made in some previous clause; rather, they introduce an additional orientation to a prior assertion. In (2), the för att clause (line 3) accounts for what makes the speaker believe in the adequacy of what he has said in the prior utterance. Fragment (3) provides a case of a så att clause (line 3) that comments on the effect of the speaker’s prior argumentation; så att appears here in the reduced form satt. Finally, in fragment (4) the men att clause (line 10) recycles in a summarizing manner the claim that the speaker has made earlier in line 2, in some contrast to, or as a retreat from, what the speaker and the other party have said just previously (lines 5–8). We will later return to these examples and provide more in depth analyses. At this stage, we can note that the general communicative task of the connectives is to structure the global discursive activity of reasoning, to add evaluations and specifications of the speaker’s or other speakers’ utterances. Thus, the connectives constitute a vehicle by which a communicative project may be further elaborated and expanded in a sequence of utterances constructed on-line. The data for our analyses is ordered in three comparative categories: a mixed corpus of conversations recorded in Central and Western Sweden (GRIS), a corpus of informal Swedish conversations recorded in Helsinki, Finland (SAM), and interviews with Swedish-speaking students aged 17–20 in a number of Helsinki schools (HUSA). In other words, our material comprises two distinct major varieties of Swedish, the Swedish used in Sweden and the regional variant of Swedish used by the Swedish speaking minority in Finland. The quantities of words (tokens) in our material are seen in Table 1 in Section 2. It must be noted that the GRIS and SAM corpora are not explored exhaustively in the present quantitative survey; here we have included primarily conversations from an informal setting. The school interviews (HUSA) are a category of their own in certain respects, although these discussions also involve group conversations and a mode which is informal rather than formal.
. Finland was a part of the Swedish kingdom from ca 1200 until 1809. There is still today a minority of 300 000 Swedish speakers in Finland where the official languages are Finnish and Swedish. Finland Swedish is characteristically a more conservative, even archaic variant of Swedish and also to some degree influenced by the Finnish language in vocabulary and syntax.
Constructing reasoning 109
In the following analysis we will first account for the general syntactic properties associated with the connectives and their distribution in the data, then we will proceed by discussing the historical background and various functions of the common constituent part, the conjunction att. This will lead to a section where we provide more detailed analyses of the uses of the three connectives in conversational data and their organizational relevance in discourse. In these analytical chapters we will propose a CxG formalism based on an attribute-value matrix (AVM) to capture the connectives’ distinct discourse-constructional properties, particularly in order to contrast the different connectives and their uses with each other. Transcription symbols and abbreviations of grammatical terms are explained in a short section at the end of the paper. The English translations provided are half-idiomatic, that is, we have aimed at preserving the Swedish syntax as long as it does not complicate the understanding. If possible, we have aligned corresponding Swedish and English words under each other in the transcriptions. We have not used glossings in the conversational extracts, since our points are not morphological or lexical-semantic but pragmatic on the level of discourse and sequence organization.
2.
Syntactic composition and distribution
As regards the composition of the studied connectives, they are a combination of two words generally functioning as conjunctions: för + att, så + att, and men + att. With respect to the individual constituents, för and så may function either in a coordinating or a subordinating manner, while men is always coordinating. The shared element att is syntactically somewhat indeterminate; it is a subordinator (complementizer) mostly in written language, but it can also be used as a coordinator, especially in spoken language. The combination of för, så, men + att results in a word group which forms one unit, usually referred to as a causal (för att) or a consecutive (så att) “multi-word subjunction” in grammars (cf. SAG 2: 733). The connective men is labelled as an adversative conjunction, whereas the combination men att is scarcely mentioned anywhere in the Swedish grammatical tradition. The unity of the word combinations is reflected in the prosodic shape in which contraction frequently occurs resulting in forms like fö att, fratt, fatt or even fött (for för att) and satt (for så att), and occasionally mnatt (for men att). . The item att also appears in many other semi-lexicalized combinations with prepositions, adverbs and conjunctions, e.g. därför att ‘because’, (däri)genom att ‘thereby that’, fast(än) att ‘although that’, oavsett att ‘irrespective of that’, utan att ‘without that’, efter att ‘after that’. There are, however, considerable functional differences and at the very least differences in frequency between these att combinations and the very common lexicalized complex forms studied in this paper.
110 Jan Lindström and Anne-Marie Londen
Table 1. The occurrence of the connectives för att, så att and men att and the syntactic status of the clausal construction following the connectives in Swedish conversational data. OMS stands for ‘overt main clause syntax’, OSS for ‘overt subordinate clause syntax’ corpus
GRIS
variable total occ. 1/1000 words OMS occ. OMS % OSS occ. OSS % total of words
för att så att men att för att så att men att för att 45 50 6 89 105 60 220 1,3 1,4 0,2 1,7 2 1,2 2,2
SAM
så att men att 129 166 870 1,3 1,7 4,6
11 24% 2 4% 35000
35 27% 3 2%
8 16% – –
2 33% – –
28 31% 2 2% 52000
HUSA
20 19% – –
24 40% – –
44 20% 24 11% 100000
TOT
55 33% – –
227 26% 31 4% 187000
The syntactic function of för att and så att is conventionally described as subordinating. It has been reasoned that the first constituent part stands for the semantic relation – of cause (för) or of consequence (så) – and the second part att indicates the subordinating syntactic value (Thorell 1973: 180). However, it is sometimes briefly noted that för, så and men can be combined with att without a subordinating value in colloquial language (ibid. 185). The survey we have made of naturally occurring data shows that this, in fact, is almost a rule in conversational Swedish, as illustrated in Table 1. The indication of whether a clause is subordinated becomes not necessarily syntactically manifest in Swedish. We may speak about overt subordinate syntax (OSS in Table 1) when a clausal adverbial (e.g. a negation) occurs between the subject and the finite verb in a clause (5a); such an adverbial occurs after the finite verb in a typical main clause (5b) (see SAG 4: 5). (5) a. för att jag inte vill bo där. (causal subordinate clause) because I neg want to live there ‘because I don’t want to live there.’ b. Där vill jag inte bo. (declarative main clause) There want I neg to live ‘I don’t want to live there.’
Of course, the very existence of a clausal adverbial in a clause is a matter of the speaker’s choice. Such overtly hypotactic word order patterns are in a clear minority in the clauses prefaced by för att and så att in our data, and men att is always coordinating.
Constructing reasoning
Certain syntactic properties in the Swedish clause structure may be indications of overt main clause syntax (OMS in Table 1). This is the case when a clausal unit starts with a constituent other than the subject (this then also causes subject-verb inversion), like the adverbial där ‘there’ in (5b), and/or when a clausal adverbial occurs after the finite verb, like inte does in (5b). In nearly one third of the cases, the connectives för att, så att and men att are followed by a clausal unit that displays overt main clause syntax. This is a considerable amount, especially in relation to the low occurrence of corresponding overtly subordinate clauses. Hence, also many of the syntactically indeterminate clauses that are introduced by our connectives are probably paratactic, syntactically independent units. As regards the individual connectives, we may note that the distribution of men att has some special features. Firstly, there are no cases where the following unit would display subordinate syntax. The connective has thus in earlier commentaries been classified as an “idiomatic” word group where att has no functional motivation (Jörgensen 1976a: 23). Secondly, the frequency of men att in the data collected in Sweden is only a tenth of the frequency in the Swedish data collected in Finland. Despite these regional differences in the distribution, we include men att in this study because it is a very central element of spoken Finland Swedish, and not absent in the dialects in Sweden either. In this way we also want to repair a descriptive gap, since the comprehensive Swedish Academy Dictionary (SAOB), as well as the Swedish Academy Grammar (SAG 1999), curiously fail to note the existence of men att altogether.
3.
Functions of the connective att
As mentioned above, there has been some confusion about the motivation of the formation of the complex conjunctions where att is the second component. The traditional account has been to regard the role of att as purely syntactic in the combinations för, så and men + att. It has been argued that att marks the subsequent clausal combination as subordinate (cf. Teleman 1967: 169–171; Thorell 1973: 180; SAG 4: 539). Indeed, this has been suggested to be the function of att even in cases where the subsequent clausal combination does not
. The general frequencies for words and collocations presented in the statistics in Allwood 1999 accord with our calculations. The word combinations för att and så att (of which a number are only accidental collocations of an adverb and a conjunction or the inifinitival marker) had the frequency of 1,5 and 2,5 occurrences per 1000 words in the spoken standard Swedish database in Allwood, 1999, whereas the combination men att was not frequent enough to be listed in the statistics (occurrences below 0,2 per 1000 words were not listed).
111
112 Jan Lindström and Anne-Marie Londen
have any indications of hypotactic syntax (cf. Loman & Jörgensen 1971: 22, 25). Interestingly, Jörgensen (1976b: 236–237) questions the validity of a mechanical analysis of att as a subordinator when everything else in the shape of the clausal combination indicates syntactic independency. However, this leaves the problem that att in non-subordinated clausal combinations would seem more or less unmotivated, a pleonastic add-on. To tackle this analytical problem we have to begin by asking what att generally does in the language, and especially in conversational language. The space here allows us only to outline some of the most illuminating communicative characteristics of att. The word is assumed to originate from an Old Swedish demonstrative pronoun þat (þät), corresponding to modern Swedish det, and it is thus a cognate to modern English that and German daß. The development from the demonstrative to a conjunction has been described in the following fashion in SAOB (s.v. att konj.) and by Wessén (1956: 58). (6) a. Jag tror det, han kommer. I believe that, he comes. b. Jag tror, det (= att) han kommer. I believe, that he comes.
The demonstrative that originally occurred within a preceding clause (6a) has thus been reanalysed as belonging to and introducing a subsequent clause (6b). Indications of a probable transient stage of this development can be found in modern Icelandic where the demonstrative það ‘that’ can occasionally be more associated with what precedes it than with what follows it. Hence, það can have features of an argument to a preceding clause, as in (7), where it relates to the discourse marker veistu ‘you know, do you know’; in this case, það also constitutes a coherent prosodic unit with veistu rather than with the following pronoun ég ‘I’ that is incorporated in the inner clausal frame here. On the other hand, there is no great leap to an association of það with the subsequent sentence where it could be reanalysed as a complementizer (which is also possible in modern Icelandic). Example (7) is taken from the data in Wide 2002 (ex. 7.22). (7)
veistu you know til hennar to
það ég ætla nú að senda sérstakar kveðjur og bakklæti that, I intend to send special greetings and gratitude Kristínu Kristín
As suggested by (6b), one of the very basic functions of att is to introduce a clause that is subordinate to a mental predicate of saying, hearing, knowing, believing
Constructing reasoning 113
etc. In other words, att is a subordinating conjunction that introduces a narrative clause, often also with the explicit function of introducing an indirect quote (8) or a reference to mental activity (9) (see SAG 4: 540ff., 850ff.). (8) I går sades det att fabriken skulle slå igen. (SAG 4: 850) Yesterday said-pass it that factory.the would close down ‘It was said yesterday that the factory would close down.’ (9) Dom förstår ju inte att det är dyra lokaler. (SAG 4: 543) They understand prt neg that it is expensive rooms ‘They don’t really understand that it’s expensive rooms.’
There are characteristic uses of att which are more typical of spoken language and not well registered in grammars. Lundström (1939: 213, 216), who has studied the syntax of southern Finland Swedish dialects, cites examples of syntactically independent utterances prefaced by att. The task of the conjunction seems in these instances to be to introduce a concluding utterance (10) or a specifying addition of some kind (11). (10) Att nok är e skojjit, när man tänker po e. That prt is it funny when man thinks on it ‘(So) it’s certainly funny, when you think about it.’ (11) O senn flytta vi hit, att ja bodd där opp först. And then moved we here that I lived there up first ‘And then we moved here, (so) I lived up there first.’
The latter use is referred to as an anacoluthon by Lundström (p. 215f.), showing that the construction is regarded as sub-standard. Similar observations are made by Lehti-Eklund (2002) who notes a use of non-subordinating att clauses in conversational Swedish, especially frequent in Finland Swedish (a corresponding, more general Swedish use is exemplified in Anward 2003). Such paratactic utterances which are prefaced by att offer an elaboration of the content of a previous utterance, typically some kind of a repetition or a specification of what has been said. This may involve a commentary to what the speaker or another participant has said, the arrival at a conclusion in an argumentation, or the closing of a topic or of an activity. One example, taken from our data, is provided in (12), where the speaker (T) is talking about mice in her summer house; a concluding commentary to the speaker’s preceding contribution is given in line 7, in an utterance prefaced by att.
114 Jan Lindström and Anne-Marie Londen
(12) SAM:V2 ”Mice in the summer house”. 01 T: först va de in i en sån dän ask¿ å så va de in i at first it was within such a little case and then it was in 02 en plastpåse (.h) >å så så: ja att den dä plastpåsen< a plastic bag and then I saw that that plastic bag 03 va alldeles söndergna[°gad°. was totally gnawed in holes 04 E: [jå yeah 05 (0.3) 06 M: °.ju° yes → 07 T: att där ha°de di vari också° (so) that they had been there too
In such uses, att clearly operates on a discourse level, and not on a sentence organizing level. Linell (2005) relates “independent” uses of att to the phenomenon of turn continuations in dialogical sequences where att may signal that a spate of talk builds upon, is “parasitic” of, what has been said earlier by the speaker or by another participant. Connectives may thus mark a certain responsive relation to a prior utterance, and project the direction of a following one pragmatically. It must be noted that att clauses with a causal or consecutive function would be pragmatically odd as first elements in a complex sentence. This applies for paratactic att clauses (13a), consecutive så att clauses (13b), and causal (där)för att clauses (13c), as exemplified by SAG (4: 408): (13) a. *Att i morgon blir det tö, är jag helt övertygad om. that tomorrow become it thaw am I wholly convinced of ‘That it tomorrow will thaw, I am totally convinced of.’ b. *Så att han fick feber arbetade han med vedhuggningen. so that he got fever worked he with wood.chopping.the ‘So that he got fever, he worked with chopping wood.’ c. ?Därför att han tog ut sig fick han feber. therefore that he took out refl got he fever ‘Because he exerted himself he got fever.’
. With “continuations” Linell (2005) refers to turn extensions as described by Auer (1996) rather than to “increments” to turn constructional units which are described by Schegloff (1996, 2000).
Constructing reasoning 115
The oddness of such clausal constructions is probably a reflection of the fact that these types of att clauses make more sense as comments and expansions of a prior assertion, rather than as points of departure for a communicative move taking a new direction. This picture is very clear in the conversational data we have studied: there is no case of a för att, så att or men att clause that would be the first clausal part in a complex sentence combination with another clause. On the contrary, even those för, så, men + att clauses that stand as “independent” syntactic units and utterances can, in one way or other, be interpreted as subsequent units, either to a preceding turn, turn part, or a longer piece of discourse. Att, then, constitutes a formal link from a subsequent unit to a necessary antecedent unit. The historically original demonstrative meaning of att may still be seen to be in some relation to this function; indeed, demonstratives are one general source for the grammaticalization of relative or other subordinators (see Heine & Kuteva 2002: 113–116). Demonstratives refer characteristically to something that is supposed to be given in a discoursal or situational context, i.e. they point to something that is available or has been available for the interlocutors. Hence, Anward (2003: 81) notes that att introduces a known situation, something that has already been established in the interaction; cf. also Bolinger (1977: 11) on that in English: “If we think of that in its fundamental deictic or anaphoric use as a demonstrative, we see that it is appropriate when the clause in question does not represent a disconnected fact but something tied in with a previous matter to which that can point back”. In sum, the basic sequential structure of paratactic att constructions, like in (12), can be generalized in a schematic representation as shown in (14). (14) a. Make an assertion Plastpåsen va alldeles söndergnagad. ‘The plastic bag was totally gnawed in holes.’ b. Point back to (a) (=att) by expanding from (a) Att där hade di varit också. ‘That they had been there too.’
Step (a) then is an independent, possibly complete move that is only retrospectively commented on or enriched by what is brought in with step (b). The difference between paratactic and hypotactic construction is that in the latter, the . Note that (13a) would be acceptable if the att clause was left-dislocated: Att i morgon blir det tö, det är jag övertygad om. Left-dislocation is a construction type where the dislocated referent is presented as being more or less accessible, from the context or the preceding discourse (SAG 4: 448; Lindholm & Lindström 2004).
116 Jan Lindström and Anne-Marie Londen
steps (a) and (b) would be included in one. This would constitute “the whole assertion”, or a single predication: Jag tror att han inte kommer ‘I think that he does not come’. We suggest an analysis of the diverse uses of att and other related connectives as a case of constructional inheritance and cognitive (re)organization. Here, the schematic steps that were shown above constitute the basic shared constructional pattern. We will argue that a trade-off between semantic specification and sequential back-linking is, indeed, the general motivation of the compound forms för att, så att, and men att. The first part of the word group specifies the semantic relation, the latter att part points back in a discourse sequence. More specifically, we suggest that the backward linking and continuative function of att has a special import in conversational language where new utterances regularly build upon prior ones or are constructed as if developing from prior utterances in a sequence. Cohesion and coherence between sentences are, of course, important in written language as well, but it seems that this is achieved by other, lexically more explicit means than with the only vaguely demonstrative att linking. We will now proceed to further investigate and illustrate the use of the complex connectives in genuine conversational sequences. The connectives are dealt with in their respective sections in order to focus the description. The case analyses lead to abstracted representations of the “discourse construction grammar” of the connectives.
4.
Functions of för att in conversational sequences
The expression för att originates from the Old Swedish preposition firi (fore, for) ‘for’ which appeared in causal constructions together with the connective at and also with demonstrative elements (þy, þät, þär) resulting in word complexes like firi þy at, firi þät at, or þär firi at (Wessén 1956: 278–279). These lead to the present-day Swedish forms: the stylistically neutral därför att ‘therefore (that)’, the somewhat more colloquial för ‘for, because’, and the colloquial för att ‘because’ which is the subject matter of this section. Interestingly, there always seems to have been a tendency for causal clauses to take a fairly independent role in the sentence; historically causals have moved from being subordinated units into coordinated units. There are functionally specific variations of för att constructions which are not included in the present analysis. One of them is a purposive construction (with the general meaning ‘in order to’), which involves the futural auxiliaries ska(ll) ‘shall, will’ or skulle ‘should, would’ – depending on the tense in the whole
Constructing reasoning 117
clausal combination. Such purposive för att units are hypotactic (15a) or even non-finite (15b). (15) a. Då måste man vara där för att man inte skulle göra then must man be there för att man neg should do bort sig. away refl ‘Then you had to be there to make sure that you would not make a fool of yourself.’ b. Då måste man vara där för att inte göra bort sig. then must man be there för att neg do away refl ‘Then you had to be there in order not to make a fool of yourself.’
Another special constructional format is found in för att clauses that have features of clefted structures: Det är för att du inte har pengar ‘It is because you do not have any money’ (cf. SAG 4: 62). Probably due to the clefting format, these types of för att clauses have regularly hypotactic syntax (cf. Steensig 1998 on fordi ‘because’ in Danish). Because this type of för att constitutes a syntactic-pragmatic pattern of its own with the introductory format det är ‘it is’, rather than genuinely joining two clauses or sentences together, we leave the construction aside here. Studies of other languages, for instance Günthner (1996) on weil ‘because’ in German and Steensig (1998) on fordi ‘because’ in Danish, suggest that the grammatical distinction hypotaxis–parataxis in causals is matched by certain functional distinctions. One of the points in these studies is that hypotactic causals express a more direct, real-world (or fact-based) causality, while paratactic causals express epistemic (knowledge-based) or speech act (action-based) causality, i.e. they constitute an evaluation of a preceding assertion or a justification of a prior action (cf. Schiffrin 1987: 202; Sweetser 1990: 77). Indeed, hypotactic för att clauses are associated with a kind of “real-world” causality. Fragment (16), taken from our school interviews, is a case in point. The interviewer (I) points out that the majority of pupils continue to high-school (Sw. gymnasium), in an attempt to provoke the interviewee (P) who questions the meaningfulness of high-school studies. The subordinated för att clause in line 3 provides a causal link within a predication rather than an evaluation or a justification of the relevance of a prior predication.
. German weil is historically temporal in meaning, while fordi in Danish is related to modern Swedish conjunctions för and ty (both meaning ‘for, because’) combining both these elements, namely the original preposition for and the pronoun thy, a historic dative form of thet ‘that’ (ODS, s.v. fordi).
118 Jan Lindström and Anne-Marie Londen
(16) HUSA:10 “Continue to high-school?”. 01 I: nittiåtta procent fortsätter ninety eight per cent continue 02 P: jå (.) nu e de kanske de att dom kanske fortsätter yeah now it’s perhaps so that they perhaps continue → 03 för att dom int vet va dom annors sku göra för att they don’t know what they otherwise would do
Grammatically, the subordinated för att clause is an adverbial complement to the concluding matrix clause dom kanske fortsätter ‘they perhaps continue’. There is thus a constructional and semantic unity of the antecedent clause and the hypotactic för att clause; that is, the causal complement clause is needed in formulating a meaningful predication in this discourse context. Hence, we call this a predication-complementing causal construction. The structural and semantic unity is further reflected by the prosodic phrasing: the preceding clause and the subsequent för att clause are produced in one intonation phrase. Differences in function may be observed in cases where för att introduces a clause with main clause syntax. In fragment (17), first presented as example (2) above, syntactic independence is overtly indicated by the position of the epistemic adverbial ju ‘you see, of course’, which follows the finite verb finns ‘(there) is’ in the clause de finns ju masser ‘there is plenty of ’ in line 8. This is an arranged focus group conversation, in which five farmers discuss genetically manipulated food and cultivation. At this point in the conversation, one of the speakers (H) questions the need to boost crops by genetic manipulation, since, according to him, there is plenty of uncultivated farmland on the earth. (17) GRIS:GML:4 “Genetically boosted crops”. 01 H: om vi ser på de här me ma:t så- så e de ju nte: n if we look at this with food so so it’s prt not 02 vi kan ju börja diskutera de när vi ha: >eh< odlat we can prt begin to discuss it when we have cultivated 03 på varenda eh:: kvadratmeter [i: i värden. on every square meter in in the world 04 C: [.mm 05 C: mm[:¿ 06 H: [då kan vi börja diskutera .hhh liksom eh: att f then we can begin to discuss like that 07 ja öhm att vi har dålit mä mat för de vi behöver yeah that we have a shortage of food therefore we need
Constructing reasoning 119
→ 08 framställa mer mat. för att eh de finns ju masser (0.2) to produce more food för att there are prt lots of 09 i Ryssland å: [(0.4) [runtikring [(0.2) där de k- kan= in Russia and around where it can 10 C: [mm:, 11 A: [jo de e ju [ingen brist på. yes there is no shortage of that 12 C: [mm:, 13 H: = odlas [å som .hhh som i dag står under fäfot. be cultivated and which is today unused 14 C: [mm:.
H comes to a possible closure of his argumentation in line 8, indicated by terminating prosody, here symbolized with a period after the word mat ‘food’. He does not, however, halt here but continues immediately with a för att clause that makes an expansion of the turn-so-far. The function of the causal expansion is to evaluate and by this further motivate the speaker’s implicit claim in the prior discourse context, namely that “every square meter in the world” is not cultivated. Hence, the paratactic för att clause does not provide a cause that would partly constitute a certain predication, but presents a new predication that marks the cause of the speaker’s conclusion: his conviction that “there are lots (of uncultivated land) in Russia and around”. This use of the för att construction can thus be regarded as representing a more free-standing, discourse-evaluating causality; cf. epistemic causality in Sweetser’s (1990) terms. The use of the epistemic adverbial ju is similarly an evaluative device which suggests that the content of the causal utterance is based on common or shared knowledge between the participants. Grammatically, the för att clause is not a constituent in any prior clause but a syntactically independent addition, which is also evidenced by the independent prosodic phrasing. Line 4 of fragment (18) contains a further example of a paratactic för att clause which provides an evaluation of what the speaker has said previously. Different from (17), the causal expansion is produced with no obvious preceding or following prosodic gap. The fragment is a part of a long turn in a kindergarten board meeting, where the speaker is presenting different alternatives for getting funding for the activities in the kindergarten; note that för att in line 2 belongs to a purposive, non-finite construction as exemplified in (15b) above. (18) SAM:M1 “Funding for the activities”. 01 L: vi sku alltså behöva di där pengarna här i skarven we would thus need that money in this transition
120 Jan Lindström and Anne-Marie Londen
02 03 → 04 05
för att sen kunna leva (0.8) leva (0.6) lugnare (0.7) in order then to be able to live live more reassured just (0.5) december januari som allti blir såndäna månader just December January which are always such months °som° .hh å egentligen också februari för att de kommer ju which and actually also February för att there will be prt int (.) in pengar °förrän tidigast i mars° (---) no money before March at the earliest
The specification provided by the för att clause in line 4 follows a unit that itself has the character of an addition, introduced with the additive conjunction å ‘and’ (å egentligen också februari), that is, the modification of which months are economically tight. The causal unit could perhaps have been presented as a “real-world cause” (for instance, “February is tight because we get money only in March”), but the paratactic construction together with the epistemic adverbial ju signals that this is not done here. Instead, the argumentation follows the lines illustrated in (19). The speaker presents the cause (c) as a motivation for her adding February (b) to the original list of months (a); the motivation refers to the (shared) knowledge that funding is not received before March, but only after the tight months of December, January and February. (19) a. december, januari som alltid blir såndäna månader ‘December, January which will always be such months’ b. å egentligen också februari ‘and actually also February’ c. för att det kommer ju inte in pengar förrän tidigast i mars ‘because you see no money is coming in before March at the earliest’
The paratactic för att clause has thus the character of a parenthetical reminder addressed to the other parties in the meeting. The prosodic phrasing also sounds parenthetical, and low, which gives an impression that the causal unit is not unequivocally integrated in the preceding unit. As with (17) and (18), line 8 of (20) also contains an example of för att which introduces a paratactic explanative addition, but this time it is clearly inserted in the middle of an on-going narrative project. The speaker (M) describes the difficulties in having a say about in which room of her apartment an Internet socket should be installed. Lines 1–6 present M’s and her husband’s viewpoint, lines 10–12 the contrasting decision made by the institution that is taking care of the apartment. (The för att clause in line 1 is structurally ambiguous and not analyzed here.)
Constructing reasoning 121
(20) USAMGRAM 5:1 “The Internet socket”. 01 M: Sten å ja >höll på å< bråka me dom >för att<¿ (0.6) Sten and I were arguing with them because 02 han tyckte att man skulle få välja vicket ↑rum↑ he thought that you should be able to choose which room 03 man ville ha [ sladden ↓ in i↓. you want to have the cord in 04 L: [mm¿ 05 (0.8) 06 M: när dom nu ska göra (.) borra hål i väggen. when they now will make drill holes in the wall 07 (1.0) → 08 M: ↑>för att< vi har ju arbetsrum:↑ i de lilla rummet. för att we have prt the study in the little room 09 (0.6) 10 M: å då har (0.2) stiftelsen bestämt (för) alla ska and then the institution has decided that everybody will 11 få internet (0.6) uttagen ↓från↓ (0.2) från de get the internet sockets from from the 12 stora rummet. big room
Line 6 can be analyzed as a subordinate clausal (när=’when’) increment to the preceding turn-part, a retrospectively added part of the event description which is put forward in lines 1–3. The för att unit in line 8 is different; it is constructed with independent clausal syntax, and also stands on its own in the sequence, both prosodically and pragmatically. The function of this causal utterance is to provide an evident background – note again the presence of the epistemic ju – which makes it possible for the listeners to appreciate the point of the preceding as well as the following argumentation: knowing that the study is in “the small room” enables the listeners to understand that it is not practical to have the internet socket in the “big room”, which is said later (l. 11–12). Being in the middle of a larger communicative project, the för att utterance in line 8 has strong features of a parenthesis. It is an insertion which presents the cause of the speaker’s frustration and why she “knows better”. Hence, we can analyze this för att construction as a case of discourse-evaluating causal. It is worth noting, however, that the causal unit could also be understood in action-based terms within the narrative: the causal insertion justifies the speaker’s and her husband’s fuss about the placement of the Internet socket, which is initially brought in in line 1.
122 Jan Lindström and Anne-Marie Londen
An example of action-based causality proper is shown in (21). The speaker (V) is offering a colleague a copy of a certain paper during a meeting (“Do you want to have this one?”). We can easily see that the subsequent för att clause cannot be interpreted as a reason for the other party’s wish to have a copy of the paper. Rather, it indicates a reason for the offer that the speaker made (“this one I have no use for”). The predication in the för att clause thus justifies and enables the preceding action. (21) SAM:M1 “Copy of a paper”. → 01 V: vill du ha den här fö att den här b‑ gör ja want you have this here för att this here do I ‘do you want this one because this one I have’ 02 ingenting me om ºdu villº nothing with if you want ‘no use for, if you wish’
Like epistemic för att causals, action-justifying causals are constructed as paratactic clauses. The paratactic structure of the för att clause in (21) is overtly manifest in that the connective is not followed by the subject (ja[g]), as regularly in subordinated clauses, but by another constituent, here the nominal part of the prepositional phrase (med) den här ‘(with) this one’. We will now proceed to summarize and formalize our findings concerning conversational för att constructions, whereby we identify three subtypes: (a) predication-complementing (fact-based), (b) discourse-evaluating (epistemic), and (c) action-justifying (action-based) causals.
4.1 Predication-complementing för att It is apparent that the constructional difference between subordination and coordination basically reflects semantic and interactional differences. Subordinated för att clauses are per definitionem adverbial complements in a larger sentence. This structural fact is reflected in their semantic-pragmatic function in that the subordinated causal unit constitutes one joint assertion (or predication) with a prior matrix clause – i.e. the causal unit is constitutive to a fact to be described and thus exists in the same (or “real”) world as this fact. Such a predication causal construction could be represented in the schematic manner of Figure 1 with a constructional matrix of attributes and their realized values; a brief introduction into the basic AVM formalism is presented in Fried & Östman, 2004, and is not explained here as regards its basic parts.
Constructing reasoning 123
Proj: predication Seq: response Utype: TCU Cat: V+ Phon: IP
Cat: V+ Role: ↓head Sem: conclusion Lform: dom kanske fortsätter
Cat: V+ Syn: ↑advl Sem: cause [lxm: för] Link: + [lxm: att]sub Lform: dom inte vet vad dom annars skulle göra
Figure 1. The predication causal construction (subordinating för att) in Swedish exemplified by the linguistic form (Lform) dom kanske fortsätter för att dom inte vet vad dom annars skulle göra ‘they perhaps continue because they don’t know what they would otherwise do’.
We introduce here a number of attributes which enrich the description of the pragmatic domain of a construction: 1. The attribute project (proj) refers to different sub-types of communicative projects (for the term, see Linell 1998: 213). Possible values are predication on the sentence level (or something more specific like assertion, evaluation, justification, account, offer, request) and exposition beyond the sentence level (or something more specific like narrative, joke, advice). Note that this is more of an action-based category than attributes within the semantic domain (sem) which includes number, definiteness and semantic role. Here, we propose additional values such as cause, consequence, contrast for sentence meaning. 2. The attribute sequence (seq) refers to the interactional-sequential position and status of a communicative move. Possible values are antecedent/subsequent (i.e. a preceding move and a following move) or, where applicable, initiative/ response or yet more specifically first pair-part/second pair-part (i.e. in an adjacency pair like question/answer). These value pairs exclude each other. 3. The attribute unit-type (Utype) defines whether the spate of talk constitutes one turn-constructional unit (i.e. a possible, and minimal, turn, TCU) or whether several TCUs are involved (i.e. in constituting a multi-unit turn). Thus, possible values are TCU and DU (i.e. discourse unit) for a turn consisting of several TCUs. Only TCUs or DUs can be combined with the attribute seq(uence) – these are units that can take an independent slot in a dialogical exchange and constitute a turn.
. See Schegloff 1996, for a discussion of turn constructional units (TCUs); by the concept discourse unit (DU) we refer to Houtkoop & Mazeland, 1985.
124 Jan Lindström and Anne-Marie Londen
Note that we also define the following in the prosodic domain: the attribute phon may have the values IP/IPp. IP refers to an intonation phrase, typical of a prosodic sentence, whereas IPp refers to an intonation paragraph, typical of longer stretches of talk, say, when formulating a piece of advice. Some of the attributes may presuppose one another at least to some extent: a TCU is typically an IP, a DU is typically an IPp (and a more complex idea-unit than a predication). Furthermore, we propose that an utterance is ideally a TCU and an IP; hence, we do not propose a separate attribute which would refer to an utterance (nor an attribute like a turn, cf. point 3 above). Some comments to explicate the description of the predication causal construction in Figure 1 are due here. As a communicative project (proj), the antecedent matrix clause (head) and the subsequent embedded för att clause (which is an adverbial constituent) form together one predication in the categorial appearance of a (complex) sentence (V+). Hence, this is a possible TCU in interactional terms: such a unit is recognizable as a complete instantiation of a construction (i.e. a construct) and followed by potential speaker transition. A TCU that amounts to a predication-complementing causal construct can be used as a single move in an interactional sequence (seq); we choose the value response here because it fits in with the actual use of the construct that is analysed. But responsiveness is probably not an intrinsic characteristic of this construction type, thus the value for seq definition could be left undecided as well. From the phonological point of view (phon), the antecedent and subsequent clauses constitute together one coherent intonation phrase (IP). Semantically (sem), the antecedent clause presents a conclusion for which the subsequent causal clause is constitutive; the lexical causal marker (lxm) of cause between the clauses is för ‘for’. Finally, we note that a lexical link pointing back from the subsequent clause to the antecedent clause is present (+), namely (subordinating) att.
4.2 Discourse-evaluating för att In comparison to the predication-complementing construction, discourse-evaluating för att clauses are not constituents in another clause and thus not really units that would absolutely be necessary for formulating a certain predication. Instead, these clauses express a causal content which has to do with the “world of the discourse”. They have typically the character of an add-on to the prior utterance, e.g. providing an evaluation of or a motivation for a point that was made. These för att clauses communicate the reason for the speaker’s knowledge of or an attitude to a claim that was presented (cf. epistemic causality in Sweetser 1990: 77, 81). The
Constructing reasoning 125
Proj: exposition Seq: [] Utype: DU Phon: IPp
Proj: assertion Seq: antecedent #1 Utype: TCU Cat: V+ Phon: IP Sem: conclusion Lform: egentligen också februari
Proj: evaluation Seq: subsequent (response) of #1 Utype: TCU Cat: V+ Phon: IP Sem: cause [lxm: för] Link: + [lxm: att] Lform: de kommer ju int in pengar förrän tidigast i mars
Figure 2. The discourse causal construction (coordinating för att) in Swedish exemplified by the linguistic form (Lform) egentligen också februari, för att de kommer int in pengar förrän tidigast i mars ‘actually also February, because we won’t get any money before March at the earliest’.
epistemic force of discourse-evaluating för att causals is often strengthened by the presence of adverbial epistemic markers like ju ‘you see, of course’. In conversation analytical terms, discourse-evaluating för att clauses are a vehicle by which the intersubjectivity in an interaction is enhanced or restored – they are used when further clarifications and motivations are needed. We may thus speak about a discourse causal construction which we describe by the schematic representation in Figure 2. We note that the antecedent and subsequent turn-parts (or DU-parts) consist of syntactically independent units (no head attribute); these part-utterances are then possible TCUs, and they tend to constitute separate intonation phrases (IP). The part-utterances (TCUs) are included in a larger communicative project which we define as an expository discourse unit (DU) and which comprises more than one intonation phrase in an intonation paragraph (IPp). Semantically and pragmatically, the first sentence presents a conclusion, an assertion, or a claim of some kind, and the subsequent unit contains a subjective evaluation of the assertion and why it should be considered valid. This sentence has thus a responsive (sub)sequential function, i.e. it responds to the content that has been put forward in a preceding utterance. The regular sequential process for the production of paratactic discourseevaluating för att constructions could be further described in a three-step structure as shown in (22): (22) a. Make a claim (conclusion) b. Respond to (a) by saying what causes you to believe (a) c. Return to the main thread that follows from (a) or await a response
126 Jan Lindström and Anne-Marie Londen
A prototypical discourse causal move (b) can appear to be an afterthought or a parenthesis from the point of discourse organization, and is possibly followed by a return to the main thread in a narrative project (c), or alternatively followed by a turn transition that leads back to the conversational main thread.
4.3 Action-justifying för att Action-justifying för att clauses, like discourse-evaluating causals, constitute syntactically independent units, and they are not necessary for formulating a predication or an action. The task of action-justifying för att clauses is to provide a further motivation for why a certain action, like a request or an offer, was made by the speaker (cf. speech act causality in Sweetser 1990: 77, 81). Such an action causal construction, as illustrated in Figure 3, is a variation of the schematic picture of the discourse causal construction (cf. Figure 2). However, we note some differences. The general discourse pragmatic project is a first pair-part seeking a response. The antecedent TCU constitutes the action proper (an offer), whereas the subsequent utterance (the speaker’s self-responsive justification) is a retrospective move that clarifies the reason for the action made in the preceding utterance. Like for discourse causals, we propose in (23) a sequential schematic for the action causal construction: (23) a. Make a proposition (an offer or a request) b. Justify (enable) (a) by saying what causes you to do (a) c. Await responses
Proj: offer Seq: initiative Utype: DU Phon: IPp
Proj: offer Seq: antecedent (initiative) #1 Utype: TCU Cat: V+ Phon: IP Sem: interrogative Lform: vill du ha den här
Proj: justification Seq: subsequent (response) of #1 Utype: TCU Cat: V+ Phon: IP Sem: cause [lxm: för] Link: + [lxm: att] Lform: den här gör jag ingenting med
Figure 3. The action causal construction (coordinating för att) in Swedish exemplified by the linguistic form (Lform) vill du ha den här för att den här gör jag ingenting med ‘do you want to have this one, because I don’t do anything with this one’.
Constructing reasoning 127
Because step (a) in this type of sequence consists of an initiating, prompting action of some kind which is further justified by (b), it follows that these steps demand a responding action from the other party (c).
4.4 Summary of the causal constructions The three causal för att constructions – predicational, discoursal, actional – can according to the above representations be understood as belonging in different formal and functional categories. The difference does not lie in the choice of the lexical causal marker – since it is för att in all cases – but in the syntactic clausal structure associated with the marker. Hypotaxis carries the value of predicationcomplementing, real-world causality, and parataxis predominantly either the value of discourse-evaluating or action-justifying causality. Interestingly, Andersson (1975) points to this kind of trade-off between coordination and subordination in an earlier study on Swedish clause-types. However, the trade-off between hypotactic and paratactic causals is probably not definitive. Rather, the paratactic construction can be pragmatically ambiguous, and can also be interpreted in predication causal terms in an appropriate pragmatic context. The converse is not possible with hypotactic för att causals: they call for a predication-internal interpretation. This then effectively restricts their functional scope and frequency (see Scheutz 2001, for a similar point on German weil causals). Steensig (1998) suggests that paratactic fordi constructions in Danish, which can be related to corresponding uses of the Swedish för att, are sequentially implicative in contrast to hypotactic fordi constructions. That is, if the fordi utterance makes a pragmatically independent claim it is also something that is likely to be commented on in the subsequent interactional sequence. Indeed, many för att additions are made as if the speaker wanted to make it easier for the other parties to join in and appreciate what has been said, as in (17) above. This is even more obvious when action causality is involved (21) because för att justifies an initiative that already as such necessitates a response. The task of paratactic, especially discoursal and actional causals, then, is to contextualize the prior move, to enhance the chances of making an appropriate interpretation of what was said previously – and getting (an appropriate) response. It must be added that paratactic för att clauses are not all that independent in pragmatic terms. They make a contribution of their own, but they do not stand on their own altogether. Instead, there is a dependency in sequential terms, a dependency on an earlier move to which the causal adds an evaluation or a justification. The job of the connective att becomes thus clearly motivated, although not always syntactically subordinating. In the causal constructions, att points back to what
128 Jan Lindström and Anne-Marie Londen
was said previously, marking that the causal move is not a disconnected fact or utterance (cf. Bolinger 1977, on the motivation of that). Even though the causal clause is not a constituent in another clause, it is still produced within the scope of the pragmatic move that has just been put forward. We then could speak about syntactic independency and pragmatic dependency at the same time.
5.
Functions of så att in conversational sequences
The principal modern Swedish consecutive conjunctions are the simple så ‘so’ and the complex så att. The complex form originates from a clausal combination where the adverb sva ‘thus, this way’ originally formed a correlate to a subsequent, consecutive att clause. The adverb then moved to the subsequent clause and became amalgamated with att into the fixed combination så att (Wessén 1956: 259– 261). Like in the case of causals, there seems to be a tendency for the consecutive clause to stand out as an independent functional unit in clausal combinations and thus to receive a paratactic rather than a hypotactic coding. There are a great variety of consecutive structures involving the elements så and att, which may complicate the drawing of some lines of distinction. It seems that the different uses reflect the different stages in the historical development of the construction from the juxtaposition of the correlative adverb så and the conjunction att into a complex, lexicalized combination. For example, consecutive constructions may involve an adverbial så in a prior clause which is complemented with a subordinated att clause (24a) and (24b). These uses are not within the scope of our analysis because så and att belong to different clauses rather than form a joint conjunction. Occasionally så att may be used in the futural purposive construction (24c) instead of för att, compare (15a). (24) a. Det blåste så hårt att vi blev alldeles stelfrusna. (SAG 4: 632) It blew so hard that we became wholly stiff.frozen ‘It blew so hard that we became totally stiff and frozen.’ b. Han formulerade sig så att alla blev övertygade. (SAG 2: 668) he formulated refl so that everybody became convinced ‘He formulated himself so that everybody was convinced.’ c. Varenda liten bondgård byggdes om så att den skulle every.single little farm built-pass part so that it should ‘Every little farm was built so that it would be’ fungera modernt [...] (SAG 2: 738) function modernly ‘working in a modern way …’
Constructing reasoning 129
We leave all these constructional variants aside here, because they are either not true instances of conjunctional så att or represent special subcategories of the construction. This does not mean, however, that the constructions in (24) are unrelated to the general consecutive så att. The generally consecutive så att may introduce a hypotactic (25a) or a paratactic (25b) clause; the position of the adverbial (direkt, nog inte) singles out the syntactic difference in the examples: (25) a. Han ställde sig i dörren så att hon direkt kunde se he placed refl in door so that she directly could see ‘He placed himself in the doorway so that she could directly’ honom. (SAG 4: 632) him ‘see him.’ b. Karls cykel står inte där, så att han är nog inte Karl’s bike stands neg there so that he is prt neg hemma. (SAG 4: 943) at.home ‘Karl’s bike doesn’t stand there, so (that) he’s certainly not at home.’
The paratactic construction type is practically ubiquitous in our conversational data, the more specific variants in (24) excluded. A similar distribution for spoken Swedish is stated in Teleman 1967 (p. 171). Fragment (26), earlier represented as example (3), shows a clear-cut example of a syntactically independent så att clause functioning as a summary of the speaker’s turn so far. This example is from a programme on Swedish TV dealing with new music videos. One of the studio guests is asked to give his opinion about a video by Glenmark, a well known Swedish rock artist. Lines 9 and 11 are relevant. (26) GRIS:TT:3 ”A rock video”. 01 M: .hh eeh (0.1) John, 02 (0.7) 03 M: he hh he $va tycker ru om de här$, what do you think of this, 04 J: (a: hh hh) så svårt å koncentre:ra sej på, >ja satt å försökte so hard to concentrate on, I sat and tried 05 lisom< nu ska ja titta på de här å (s)ska ja verkligen .hh å like now I’m going to watch this and going to really and
130 Jan Lindström and Anne-Marie Londen
06 va e- va handlar den om va e re som händer å så ba’a .hh what is- what is it about what is happening and then just .hh 07 >halvvägs igenom så märker att ja sitter tittar breve halfway through I notice that I sit watch beside 08 å e sär< va a: just ja visst ja videon. .hh and am like what yeah that’s right of course the video → 09 satt de de kan ju nte va särskilt gott betyg åt så att that that can prt neg be particularly good grade for 10 själva videon å sen e re väl- .hh sen e re väl the video itself and then it is prt then it is prt ((appr. 14 lines omitted)) → 11 J: ((alarm goes off)) (.) ja: vet inte satt eh .hh I don’t know så att 12 men de kan väl bli en hit i och för sej de vet ja inte. but it could prt become a hit I suppose I don’t know
The så att clause in line 9 follows after John has been describing his reactions when watching the video. There is a juncture before this clause, marked by falling intonation; the speaker then extends the turn after an inhalation. The så att clause exhibits overt main clause syntax: the two clausal adverbials, the epistemic ju ‘of course’ and the negation (i)nte ‘not’, occur after the finite verb kan ‘can’. The conclusion introduced by the connective så att (in the reduced phonetic shape satt) stands out as concise and salient, while simultaneously being closely built upon the detailed, concrete description in the preceding discourse. The use of epistemic ju underlines the evaluative character of the summarizing comment: similar to paratactic för att clauses, non-subordinated så att clauses seem to have a subjective, discourse-evaluating rather than a predication-complementing relation to the claim they are referring to and expanding. This may be a natural result, since the speaker is engaged in the global activity of stating his opinions and giving judgements; in a way, he is also judging his own judgements. There is another instance of the connective så att (satt) later in the long turn (line 11). However, this time the projected construction is abandoned. After a hesitation marker (eh) and an inhalation, a new start is made, initiated by the simple connective men. It is a recurrent feature of ongoing talk that connectives are repeated, abandoned, or exchanged for others. This reflects the fact that connectives are an important resource in turn construction. They are a central means by which the turn may be further expanded and useful as a landmark for the logical status of the coming expansion. In each case the connectives are clearly carefully chosen to match what has been said and what will be said in a sequence.
Constructing reasoning 131
The somewhat hasty ending of the turn in (26), lines 11–12, is probably due to the time limit set for the contribution (cf. the alarm going off in line 11). Så att may introduce a new turn that points back to a previous sequence, thus constituting a sort of concluding, responsive extension of an exposition. This is the case in the following excerpt from an informal gathering with four elderly women (27). One of the participants (D) is telling the group about the serious illness of her brother-in-law. (27) GRIS:SÅI:NF ”Leukemia”. 01 D: å sen va heter det för nånting eh, så va de min svå:ger då:, and then how do you say it then it was my brother-in-law prt 02 som fick sin dom i (.) missommardan, (0.5) han hade mellan who got his verdict midsummerday, he had between 03 ett å ett å ett halvt år på sin höjd. (1.0) (blod) leukemi. one and one and a half year at the most. (blood)leukemia 04 B: hårda bud de. a tall order that 05 D: eller akut [(blodleukemi) or acute (blood leukemia) 06 C: [visst ä de of course it is 07 B: ja visst ä de de. yes of course it is so 08 C: .ja: yes → 09 D: så att eh de de e jobbit sådär å få de [sådär klart. så att it it is trying like that to learn it that definitely 10 C: [jo de e klart. yes that’s sure
As seen in the introductory part of D’s turn (lines 1–3), her brother-in-law is one among several persons she knows who is struck with illness; this “plentiness” is marked by list intonation as well. There are altogether four short responses from two co-participants dealing with D’s announcement. At this point D initiates a turn with the connective så att (line 9). In this turn she is summing up what has been said so far about the prognosis that has been given to her brother-in-law. In this consecutive utterance she then is evaluating the previous account from his perspective: it is hard to learn so definitely how long one has left to live. The nature of the conclusion is thus discourse-evaluating rather than predication-completing.
132 Jan Lindström and Anne-Marie Londen
The responsive side of this type of construction comes clearly to the fore in so called formulations where the recipient interprets the key content of the prior speaker’s turn in a summarizing manner (see Heritage & Watson, 1980). Such formulations can in Swedish be introduced with (paratactic) så att: (28) HUSA:11 ”Staying home until twenty”. 01 A: att dom sparar ju på att hålla en hemma lite. so they save prt by having you at home a little 02 I: 03 T: → 04 I:
mm men men alltså, but but so så att de e de som e pointen. så att it is that which is the point
Here A gives reasons why it can be profitable for the state if young people do not move from their parents before they are twenty. The interviewer (I) then uses a så att-prefaced turn to display her understanding of A’s reasoning, i.e. what was the point of A’s contribution. There is no overt indication of whether the consecutive clause is dependent or not, but the use of a cleft-construction (de e de som …) suggests a main clause status for the så att-prefaced unit. A slightly different example, this time with overt main clause syntax in the consecutive clause, is provided in (29). In this sequence, the parties at a meeting elect a secretary according to the agenda: Nanna is proposed and elected (her own candidate is Bettan). Regina, who is the chair, concludes the outcome of the election in a så att-prefaced formulation in line 6: “now we’ve chosen that”. (29) SAM:M2 “Electing a secretary for the meeting”. 01 R: okej å sen sekreterare okay and then secretary 02 E: Nanna 03 N: Bet$tan$ 04 R: nä: Nanna e bra= no Nanna is good 05 A: =Nanna understöds jå= Nanna is supported yeah → 06 R: =okej↑ (.) så att nu ha vi valt (0.5) de okay så att now we’ve chosen that
The example thus shows that a formulation can be used to summarize the gist of a collective action sequence, not only the content of a single prior contribution.
Constructing reasoning 133
In the next fragment (30), we see an example of the connective så att functioning as a kind of final tag, signalling a potential closure of the turn. It is noteworthy that ca 10 % of the consecutives we have studied are used in such utterance final position. (30) HUSA:34 “Swedish contacts”. 01 I: mm (0.2) $.jå$ men i övrit har du inte nåra (.) kontakt yeah but otherwise you don’t have any contact 02 nåra organisationskontakter. int me i någå föreningar any organizational contacts don’t belong to any associations 03 eller nå såndänt or anything like that 04 K: n- nä (0.3) nå okej alltså nu (.) nu e ja me i n- no well okay that is prt prt I belong to 05 föreningar men liksom (.) liksom idrottsföreningar å dom e associations but like like sports clubs and they are → 06 helt finska $så att$ completely Finnish så att 07 I: mm (0.2) inga närmare. du ha vari i Stockholm nån gång no closer. you have been to Stockholm some time
The excerpt is part of a school interview dealing with Finland Swedish students’ contacts with the Swedish language outside school. In line 4, Kenneth confirms the interviewer’s assumption that he does not have any contacts with (Swedish) organizations and does not belong to any (Swedish) associations. The confirmation token is simply ‘no’. Kenneth then proceeds to correct the (possible) implication that he does not belong to any associations at all. He explains that he does indeed belong to, for example, sports clubs and that they are completely Finnish. The turn ends with så att, said with smile voice (line 6). The conclusion that would follow from the preceding information seems to be treated by the speaker as so self-evident that it need not be said or, indeed, repeated. The interviewer signals his understanding, first by the response particle mm and then with a laconic formulation which gives the gist of Kenneth’s responses to the current question: “no closer” (Swedish contacts). He then immediately initiates a new sequence. Turn transition after the connective så att in line 6 is smooth: there are no signs of problems, there is for instance no audible pause. However, a connective which seems to be designed as a final tag, and consequently as a signal of turn closure, may be passed over by the speaker; that is, the speaker does not stop there but continues and produces a new ending (with other sequential implications). This is illustrated in the following excerpt (31) from a
134 Jan Lindström and Anne-Marie Londen
dinner conversation between a group of men. The host has (apparently) put a casserole on the table and there is some fuss about missing forks. Olle then asks (lines 1–2) whether there is something to go with the casserole, using a three-part list (the third component is a generalized completer, “or something”). (31) SAM:V1 “There is bread”. 01 O: ska man ha någå (.) potis ti de här också eller (.) shall we have some potatoes to this too or 02 ris eller nånting rice or something 03 L: hey don’t ask [me ((paper rustle)) 04 E: [>nå dehär< (.) de (.) bjuds nu ingenting annat well prt there will be nothing else served → 05 än de här så att de[där (.) bröd (.) bröd finns de than this så att prt bread bread there is 06 ?: [jaha↑ uh huh
The host’s response (line 4) to the question starts in slight overlap with Lasse’s response which is said in English: hey don’t ask me. That the host’s response will be a dispreferred one is signalled by the turn initial particles (nå dehär) and by the two micro pauses. The statement that nothing else than “this” will be offered is followed by the connective så att combined with the hesitation particle dedär, and this segment can be heard as a final tag (line 5). The (obvious) conclusion that the guests have to be satisfied with the casserole is left hanging in the air; cf. excerpt (30) above. One may note that there is a token of realizing (jaha) from one of the participants almost immediately after the host has uttered the connective så att. Following a micropause, however, a stressed noun phrase (“bread”) is added and then, after a micropause, the clause “bread there is” is produced. The host seems suddenly to realize that even if he cannot offer potatoes or rice, there is “something” available, this something being bread. Even though the remark that “there is bread” is not an obvious consequence of what precedes it, the new consecutive course of the turn makes sense as a result of the fact that there is no specifically prepared additional food – but bread, nevertheless. In excerpts (26)–(29) above we saw examples of the connective så att being used by the speaker to introduce an explicit, conclusive commentary on the content of the turn or of the sequence so far. In excerpts (30)–(31) on the other hand, så att does not initiate a new contribution, instead it seems to have turned into a kind of turn-final marker, leaving the conclusion unexpressed. It can be noted in this connection that Jefferson (1983) points out that “conjunctionals” (e.g. uh, well,
Constructing reasoning 135
and, but, because, so, or) often co-occur with overlapping talk in turn-transition. Conjunctionals seem to be occasionally indeterminate as regards the function of starting a new unit or marking so called pro-tem speakership. The latter means that a speaker is “available for speaker transition”, although in mid utterance, and aims not to continue if the other party takes the opportunity to start to speak (ibid. p. 6). In case of pro-tem speakership, there is thus the option to negotiate whether an utterance will be finished from, say, a connective position, or whether the turn goes to another party. If speaker transition occurs in such a mid utterance position, the connective forms a trail-off, i.e. a contribution that fades out in a position where it, e.g. syntactically, could have continued. As regards excerpt (31), one then could argue that the connective is exploited as a pro-tem marker. However, the possibly trailing-off så att is retrospectively transformed by the current speaker into a platform for a turn continuation that leads to a conclusion of some sort. The general picture of the uses of så att emerging from our conversational data is that the connective introduces a summary, explicitly stated or, in some cases, left unexpressed. The consecutive link is of a discourse-evaluating kind rather than of a predication-complementing kind; the link to preceding discourse is always discernible. At the same time the opportunity to say things slightly differently, maybe more saliently, maybe from a different speaker perspective, is opened up. The sequential process for such discourse-evaluating uses of så att could thus be outlined as in (32): (32) a. Make a description. b. Summarize or restate (a) by evaluating (a) from a personal perspective. c. Open the floor for new contributions by doing the summary.
Because a summary or a conclusion suggests the end of a line of reasoning, a så att utterance is always potentially indicative of upcoming turn transition (c). In the overwhelming majority of cases in the data, the clauses initiated by så att can be regarded as syntactically independent units. On the other hand, as they characteristically constitute expansions of what has preceded them in the discourse, these clauses cannot be marked as completely independent moves in the on-going talk. Thus, paratactic consecutive clauses are syntactically main clauses but have a pragmatically subordinate facet; cf. the discussion of main and subordinate clause-types in Andersson (1975: 57). In a similar manner as with för att constructions (cf. Figure 2), an attributevalue matrix for a discourse consecutive construction is made in Figure 4. The subsequent consecutive clause, constructed as a syntactically independent clause (a potential TCU), is nested in a larger discourse unit (DU) and stands in a responsive, commenting relation to what has preceded it.
136 Jan Lindström and Anne-Marie Londen
Proj: exposition Seq: initiative Utype: DU Phon: IPp
Proj: assertion Seq: antecedent #1 Utype: TCU Cat: V+ Phon: IP Sem: cause Lform: han hade mellan ett å ett å ett halvt år på sin höjd
Proj: evaluation Seq: subsequent (response) of #1 Utype: TCU Cat: V+ Phon: IP Sem: conclusion [lxm: så] Link: + [lxm: att] Lform: de e jobbit sådär å få de sådär klart
Figure 4. The discourse consecutive construction (coordinating så att) in Swedish exemplified by the linguistic form (Lform) han hade mellan ett å ett å ett halvt år på sin höjd, så att de e jobbit sådär å få de sådär klart ‘he had between one and one and a half year, so that it’s trying to learn it that clearly like’.
As with causals, we might want to identify an action-based variant of the consecutive construction (33). Instead of justifying a prior action, the så att utterance itself formulates an action (for example, an offer) for which the preceding context serves as a background, i.e. as its enablement or justification: (33) SAM:M1 ”The bookkeeping documentation”. → 01 L: ja har boken där bo:rta↓ [>så att vi kan ju= I have the book over there så att we can prt 02 N: [aha uhhuh 03 L: =titta< efteråt på den °om ni vill ta en titt° look afterwards at it if you wish to take a look
Here L, the chair of a meeting, first states that she has her bookkeeping documentation at hand, which is a fact that enables her subsequent offer that the other parties may take a look at the documentation (boken ‘the book’) if they wish. Interestingly, this initiative – for offers are typically initiating moves – is constructed as a response to the prior, enabling account within the same turn. The effect of an initiative then is ascribed to the turn as a whole, while the turn part that spells out the offer has a turn-internal responsive feature. The AVM representation of an action consecutive construction would then look as shown in Figure 5. In (33) it is not totally clear if the antecedent account and the subsequent offer should be analysed as one TCU or a DU consisting of two TCUs, since the intonation pattern is rather indeterminate, albeit somewhat falling after the element där borta ‘over there’ (line 1). However, the preceeding account and the subsequent offer are constructed as independent syntactic units, which marks a loose relation
Constructing reasoning 137
Proj: offer Seq: initiative Utype: DU Phon: IPp
Proj: account Seq: antecedent of #1 Utype: TCU Cat: V+ Phon: IP Sem: cause Lform: ja har boken där borta
Proj: offer Seq: subsequent of #1 Utype: TCU Cat: V+ Phon: IP Sem: conclusion [lxm: så] Link: + [lxm: att] Lform: vi kan ju titta efteråt på den
Figure 5. The action consecutive construction (coordinating så att) in Swedish exemplified by the linguistic form (Lform) ja har boken där borta, så att vi kan ju titta efteråt på den ‘I have the book over there, so that we can look at it afterwards’.
between them: the offer of looking at the book constitutes a response to the reference to the book’s availability and not a necessary consequence of it. Note also that the recipient (N) produces an acknowledgement token (aha, line 2) after the completion of the antecedent, which could suggest that he identifies it as a single, possibly complete move, i.e. as a possible TCU. Finally, when a dependent syntactic construction with så att is used – and it rarely is in conversational Swedish – we have a predication consecutive construction in which the consequence is a necessary condition for the formulation of a meaningful predication. In (34) we reiterate the example from SAG (4: 632), previously presented in (25a). (34) Han ställde sig i dörren så att hon direkt kunde se honom. He placed himself in the doorway so that she could directly see him.
The consecutive clause in (34) is not a summary or a restatement of some aspect of the prior claim, nor is it, of course, a separate speech act. Instead, the consecutive clause enriches, as an adverbial constituent, the description of the action communicated by the finite verb phrase ställde sig i dörren ‘placed himself in the doorway’; thus, the conclusion functions as predication-complementing. The AVM description is rendered in Figure 6. The typical pattern for så att consecutives in conversational Swedish is that a speaker first delivers an opinion, an evaluation, or an account, which then constitutes the basis or motivation for a subsequent concluding and summarizing turnunit which can also contain subjective evaluative aspects. The concluding move is done in a separate, commenting predication, which then calls for paratactic rather than hypotactic syntactic coding. Such concluding moves, like in the case of causals, are sequentially implicative in their own right (cf. Steensig 1998). This can
138 Jan Lindström and Anne-Marie Londen
Proj: predication Seq: [ ] Unit: TCU Cat: V+ Phon: IP
Cat: V+ Role: ↓head Sem: cause Lform: Han ställde sig i dörren
Cat: V+ Syn: ↑advl Sem: consequence [lxm: så] Link: + [lxm: att]sub Lform: så att hon direkt kunde se honom
Figure 6. The predication consecutive construction (subordinating så att) in Swedish exemplified by the linguistic form (Lform) Han ställde sig i dörren så att hon direkt kunde se honom ‘He placed himself in the doorway so that she could directly see him’.
be an important motivation for the abundancy of syntactically and semantically independent consecutives in conversational language. The recipient is expected to comment on (and accept) the gist of the prior argumentation, rather than taking a stand on the whole argumentation, which could be a more demanding task.
6.
Functions of men att in conversational sequences
There are no clear etymological accounts of the origins of the connective men att. The constituent part men ‘but’ goes back to Old Swedish män (men, man) and its original meaning has been ‘only’. From an adverb with a restrictive meaning, men apparently developed first into a concessive and then into an adversative conjunction. Possibly due to the bleached restrictive adverbial meaning, men has been used “pleonastically”, together with the adverbial allenast ‘only’, combining with a following att, i.e. men allenast att ... ‘only with the exception that, only with the restriction that’ (SAOB, s.v. men konj.). The meaning of this complex expression is not far from that of men att in our present-day data, and it is intriguing to observe that if the middle element allenast was dropped it would, indeed, leave us with men att. As shown previously in Table 1, the frequency of men att is ten times higher in Finland Swedish conversations if compared to a corresponding Central Swedish material. The regional nature of the expression is noted in a number of studies. Ber. Allén (1965) who has edited a collection of private letters from 1639–1655 makes two references to män att (men att). He notes the Middle Low German parallel form men dat which Schiller & Lübben (1875–81, s.v. men) compare with modern German aber ‘but’ and translate with ‘except for, only’ (Allén 1965: 124 Note 9, and p. 127 Note 9). Allén also suggests a functional similarity between män att and the expression allenast att, giving the translation utom att ‘except that’ (p. 136 Note 6).
Constructing reasoning 139
groth (1928) labels men att as non-standard, the dictionary of Swedish dialects in Finland (FO s.v. att) registers the combination men att in different Swedish regions in Finland, Saari (1975: 92–93) notes it as a characteristic of Swedish speakers in Helsinki, and Lindström & Londen (2001: 107–108) find that men att is common both in the Helsinki region and in the north eastern coastal region Ostrobothnia. The connective men att has been regarded as an idiomatic word group where att is a pleonastic, functionally unmotivated addition (cf. Jörgensen 1976a, FO, s.v. att). It is certainly tempting to come to this conclusion, since one cannot even attribute a subordinating functional value – which would add to the adversative semantic value of men – to att in any instances of men att. Neither is men att used in standard written Swedish, unlike för att and så att. However, we will argue that men att is as motivated as a compound conjunction as the previously discussed combinations för att and så att. Instead of saying that att is a pleonastic addition to men, we claim that there is a functional trade-off between the constituents men and att, men giving the semantic specification and att pointing back to prior discourse. In the excerpts above, we have seen instances of för att and så att in different positions in the turn. The most frequent use of the connectives, including men att, is as an introducing particle in a new TCU within the turn, that is, the connective projects a semi-independent extension of the turn and indicates how this projected spate of talk is relating to preceding discourse. The connective men att may be looked upon as a kind of contrastive counterpart to så att in that it introduces a conclusion that is in some contrast to the viewpoint(s) presented in the immediately preceding discourse but then sums up something that was stated earlier. This is markedly the case in the following fragment (35) from one of the school interviews, an excerpt that may well be compared to excerpt (26) above with the connective så att. At issue in this part of an interview with two girls is where and what to study after high-school (“Hanken” is a common name for the Swedish school of economics). As the turn in question is quite long, some lines have been omitted. (35) HUSA:03 ”Study at Hanken”. 01 I: ni vill int studera på Hanken you don’t want to study at Hanken 02 (1.0) 03 J: nä ja ha no int (0.3) ja ha no int tänkt studera no I have prt not I have prt not thought of studying 04 på Han$ken$ (hh) (0.7) enda som ja vet att de e at Hanken only (thing) that I know that it is 05 språ- att språkundervisningen e helt bra där (0.3) lang- that the language teaching is quite good there ((appr. 6 lines omitted))
140 Jan Lindström and Anne-Marie Londen
→ 06 07 08 I:
men att (0.5) men att ja ha no int planera de sådä: (hh) men att men att I have prt not planned it like that (1.0) mm
There is no immediate response to the interviewer’s (I) initial assumption that the girls do not want to study at Hanken, then J takes the turn (line 3). After her negative, confirming answer, she says that she has not “thought of ” studying at Hanken. After a break she goes on to explain that the only thing she knows about Hanken is that the language teaching is good, then proceeds to give reasons for why one could, in fact, study at Hanken. After this concessive turn-part, she introduces a TCU with the (repeated) connective men att (line 6). In this TCU she then dissociates from the viewpoints presented in the immediately preceding concession and recycles her original, direct answer to the interviewer’s question. This reprise is done by a slight rephrasing (partly due, of course, to the different position in the turn) and is highlighted in (36) where (a) represents the first TCU in J’s turn, (b) the last TCU. (36) a. b.
nä ja ha no int tänkt studera på Hanken no I have not thought of studying at Hanken men att ja ha no int planera de sådä: men att I have not planned it like that
The paratactic construction, apparent in the postverbal position of the adverbs no(g) and inte, is clearly manifest in the men att clause. One could say that the discourse produced after the direct answer in (35) shows the speaker’s awareness of there being other viewpoints or arguments. However, these arguments are pushed into the background by the men att utterance, men marking a contrast of some kind and att marking a linkage backwards to something which has been made accessible earlier. The conclusion that is put forward in this way stands out as the speaker’s “first and last word” on the matter. In excerpt (37), also used as example (4), the Hanken school is under consideration once more. Before the fragment starts, B, one of the girls interviewed, has claimed that those studying at Hanken are easy to recognize. The interviewer then asks how one is able to “see” this (line 1). The connective men att appears in line 10 and it introduces a new turn; cf. the use of så att in (27) above. (37) HUSA:08 “Branded clothes”. 01 I: hu kan man se de¿ how can you see it
Constructing reasoning 141
02 B: $på märkeskläder$ å, on branded clothes and 03 (0.4) 04 M: [jå yeah 05 B: [oftast har dom den där (.) salkkun i handen å $hh$ most often they have that briefcase in the hand and 06 I: å pärlhalsbande¿ and the pearl necklace 07 B: jå: pärlhalsbande å, (.) sen ibland finns dedä NMT yeah the pearl necklace and then sometimes there’s this NMT 08 ännu i fickan $å eh[eh$ also in the pocket and 09 I: [jå yeah → 10 B: men att, märkeskläder de e- (.) de brukar nu alla ha. men att branded clothes that’s what they all usually have
B gives a prompt answer to the interviewer’s question, using first a TCU consisting of a prepositional phrase: “on branded clothes” (line 2). She proceeds in line 5 to mention one more distinctive attribute, “the briefcase” (i.e. the adapted Finnish word salkkun), using a clausal TCU. The intonation and the utterance final conjunction å ‘and’ project that more will follow. At this point the interviewer comes in, providing another (feminine) attribute, å pärlhalsbande ‘and the pearl necklace’. Her suggestion is accepted by B (line 7) who then goes on to produce her own (presumably planned) addition to the list, a new cellphone (“NMT”), again using a clausal TCU. After an acknowledgement token from the interviewer, B initiates a new turn with men att (line 10). The connective marks a return to and a recycling of the speaker’s first, phrasal TCU from line 2, which contained the first item on the list of typical Hanken-student attributes. In the utterance prefaced by men att, the noun phrase “branded clothes” appears as a left dislocation in a clausal TCU, which reinforces the prominent, accessible status of this attribute (cf. Note 5 above on dislocations). Furthermore, an assessment is made in this utterance. The speaker states that they “all” usually wear branded clothes – a feature which is presented as a generic one (note the generic modal verb brukar ‘are used to [do]’). This modal feature of the utterance is also signalled by the epistemic nu ‘at least’. Compare this to what was said in connection with the two other attributes:
142 Jan Lindström and Anne-Marie Londen
they “most often” carry a briefcase (line 5) and they “sometimes” also have a new cellphone in their pocket (lines 7–8). Thus, the men att utterance presents the gist of the whole preceding sequence: “branded clothes” is restated as the most general and the most prominent attribute among those listed as characteristic of people studying at Hanken. As we have pointed out in (36), the connective men att has systematic uses in interactional sequences. This systematicity can be presented more accurately in a three-part formula as in (38): (38) a. make an assertion b. acknowledge alternative viewpoints c. men att: recycle the gist of (a) and by doing this, skip what was said in (b)
Interestingly, this formula is compatible with the discourse structure of show concessions as described by Antaki & Wetherell (1999). These also have a three-part structure consisting of “proposition, concession and reassertion”. The strategic aim of show concessions is to strengthen the speaker’s own position at the expense of a counter-argument. Also in prototypical men att-sequences, the conceding part (b) can be seen as a way of foreseeing possible counter-arguments to (a) and pre-emptying them before they even have been brought in by another party. We could thus conclude that the connective men att is a central vehicle by which show concession sequences are being carried out – and completed in phase (c) – in (Finland) Swedish; for English, Antaki & Wetherell (1999: 9) note the use of contrastive conjunctions like but and nevertheless. It is likewise clear that all the three moves in this kind of a sequence (38) are more or less independent in their nature although tied to the same communicative project. Thus, utterances with men att are always constructed as main clauses. Semantically, a typical men att utterance is both adversative and consecutive in its responsive quality; it contrasts with the immediately preceding concession but concludes the whole sequence by reasserting its most salient starting point. We will thus term such uses of men att as a discourse adversative-consecutive construction. Its AVM is represented in Figure 7. Note that the prototypically occurring conceding segment – i.e. (b) in (38) – is represented by [...] between the opening antecedent (a) and the closing subsequent (c); the phase (b) is thus not described due to space limitations. We will finally illustrate some relations as well as differences between så att and men att with an analysis of an excerpt that contains several instances of both of these connectives. In the rather short fragment (39), from a conversation between four women friends who meet regularly at each others’ homes, we find two occurrences of men att and three occurrences of så att all together produced by
Constructing reasoning 143
Proj: exposition Seq: response Utype: DU Phon: IPp
Proj: assertion #1 Seq: antecedent Utype: TCU Cat: V+ Phon: IP Sem: [ ] Lform: ja ha no int tänkt studera på Hanken
[...]
Proj: reassertion of #1 Seq: (closing) subsequent Utype: TCU Cat: V+ Phon: IP Sem: contrast [lxm: men] Link: + [lxm: att] Lform: ja ha no int planera de sådär
Figure 7. The discourse adversative-consecutive construction (coordinating men att) in Swedish exemplified by the linguistic form (Lform) ja ha no int tänkt studera på Hanken (...) men att ja ha no int planera de sådär ‘I have not thought of studying at Hanken (...) but I have not planned it like that’.
two different speakers, Asta (A) and Eva (E). Before the fragment starts, Eva has been talking about a dinner with Italian food to which she had treated her children and their families in her summer house some time ago. The connective men att appears in lines 2 and 12, and så att in lines 5, 11 and 14. (39) SAM:V2 ”Wine for dinner”. 01 A: hade du vin °till° did you have wine too → 02 E: jå ja hade vin men att di va ju, yeah I had wine men att they came prt 03 (0.4) 04 A: me bil= by car → 05 E: =~hälften me bil~ så att, half of them by car så att 06 (A): °mm° 07 (0.3) 08 E: de (.) blir aldri så my[cke drucket. (.hh) it will never be so much consumed 09 A: [jå yeah 10 (1.0) → 11 E: så att (.) (h) $de$ räckte me en flaska.= så att it was enough with one bottle
144 Jan Lindström and Anne-Marie Londen
→ 12 A: =ja just men att ändå↑ att ma[n har yes right men att still that man have 13 ?: [°jå° yeah → 14 A: >så att de [lite hör< <°liksom till°>. så att it kind of is like a part of it 15 E: [jå yeah 16 (.) 17 ( ): °.ja° yes
As regards the occurrence of men att in line 2, one may note, firstly, that Eva’s answer to Asta’s question is a full form answer, and as such it has features of a concession, probably implying problems that need to be sorted out (cf. Lerner 1995, on the markedness of “complete-sentence answers”). Secondly, the unit introduced by men att is begun immediately after the direct answer: there is no pause, no disfluency at this point, and it can also be heard as prosodically continuous from the prior clausal unit. This differs from the more prototypical uses in (35, 37) where men att clearly introduces a new TCU and an independent intonation phrase. In this case the men att clause does not connect back to something explicitly stated earlier, instead it draws on contextual knowledge shared by the participants. The fact that the guests had come by car seems to be taken as more or less given information by both Eva and Asta. Eva uses the epistemic marker ju ’of course, you see’ in the unit introduced by men att (line 2) and Asta is able to fill in the verb phrase with the adverbial “by car” (line 4) before Eva has verbalized this, apparently projected information with which she in a sense questions the relevance of having served some wine. The accessability of this information is also signalled by the prosody of the segment hälften me bil ‘half of them by car’ in the latter part of Eva’s utterance (line 5). It manifests a kind of stylization (marked with tilde ~) that seems to signal ‘nothing special’ or ‘nothing new’ (cf. Ogden, Hakulinen & Tainio 2005, on this prosodic feature in Finnish). The first occurrence of så att appears in Eva’s utterance in line 5. Using the modified repetition “half of them by car”, Eva makes Asta’s completion, found in line 4, more precise. She then immediately initiates a new move with så att. After a quiet response particle and a slight pause, the projected conclusion – that perhaps could have been left hanging in the air like in (30) – is presented. The conclusion exhibits overt main clause syntax (the negating adverb aldri(g) ‘never’ is placed after the finite auxiliary verb) and it is in the format of a generalizing remark. A second concluding comment initiated with så att follows after a pause (line 11).
Constructing reasoning 145
This comment is referring specifically to the family dinner: it reports the amount of wine consumed and it has the verb in the past tense. The second conclusion could thus also be seen as a modified resumption of Eva’s original, conceding reply that she indeed served some wine. In this sense, the utterance functions as a closure of the sequence as far as Eva is concerned; cf. Schiffrin’s (1987: 198) analysis of series of so-prefaced utterances. Asta comes in after Eva’s second så att utterance, which, being doubly concluding, effectively signals a possible turn transition. Asta’s turn (lines 12 and 14) may be said to bear some structural similarities to Eva’s first turn: it contains both a men att construction and a så att clause. In lines 12 and 14, she gives a summary of the whole sequence that she herself initiated. This turn is initiated with men att, contrasting with the preceding, and is developed with så att giving a conclusion. What this turn (which contains a few hedges) seems to do is to establish the general point that wine is an essential part of a dinner with Italian food: it is a return to the point implicated in the question in line 1. The use of men att is significant at this point. It serves as a resource by which Asta is able to treat the account by Eva as a kind of a concessive counter-argument, whereas what Asta is saying in her conclusion restates what really counts: wine is an essential part of an Italian dinner. The sequence as well as the whole topic ‘Eva’s family dinner’ are thereby effectively brought to an end. To summarize our observations regarding the short but complex fragment “Wine for dinner”: there is quite a lot of reasoning and negotiation going on between the interlocutors in this dyadic interaction, and the connectives men att and så att have an important role in the structuring of the discourse and the fitting in of the slightly shifting viewpoints of the two speakers. Så att introduces comments that summarize aspects of an immediately preceding utterance. Men att stands for a certain cyclicity in the reasoning; it introduces utterances which resume points that have been made earlier in the discourse and which thus contrast with the immediately preceding discourse context.
7.
Summary and conclusions
Our analysis of the Swedish complex connectives för att, så att and men att has led us to many important insights about the cognitive and interactional organization of turn construction. We have compelling evidence for our assumption that the general connective att has a motivated role when attached to the simple conjunctions för, så, and men. This role is in accordance with the function of simple att, which, in the most general terms, is to point back to a preceding discourse source and respond to this and expand from this. What is naturally left for the elements för, så
146 Jan Lindström and Anne-Marie Londen
and men is to indicate more clearly the semantic relationship – causal, consecutive, adversative – between the assertions and actions that are joined together. One of our central results is that the choice between hypotactic or paratactic coding of the causal and consecutive clauses studied here makes a discourse functional difference. Hypotaxis is clearly reserved for predication-complementing (fact-based) causal or consecutive relations, while parataxis signals principally a looser relation between a prior statement and the following explaining or concluding move. Paratactic för att causals enhance the adequacy of what was said previsously by explaining what causes the speaker to believe what was said. They can also justify an action, like an offer, which the speaker has made. Så att conclusions spell out an inference that could be made on the basis of what was said previously or they present the gist of a preceding sequence, thus bringing it to a potential closure. Furthermore, they can verbalize an action as resulting from the account which has been made in a prior utterance. The scope of men att is usually larger than a single TCU – because it prototypically reiterates and thus refers to something that has not to do with the immediately prior turn or turn-part. Another reflection of the relatively independent discourse function of paratactic för att, så att, men att expansions is that they normally constitute an intonation phrase – and TCU – of their own. As mentioned earlier, the formal and functional trade-off between paratactic and hypotactic causal and consecutive constructions is probably not absolute. The paratactic variants can occasionally stand in a more direct content relation to the antecedent, while the hypotactic variants are restricted only to predication-completing (“factual”, or “propositional”) causal or consecutive semantic relations. There are thus other conceivable reasons for why parataxis in general is favoured in spoken Swedish. We propose that the explanation lies in the fact that the nature of producing “sentences” in spoken language is incremental, i.e. the formally continuous units of language are smaller and stand in serial rather than in hierarchical relations to each other. In this kind of “add to the prior and add again”, not thoroughly planned language production, paratactic, syntactically non-embedded unit types are a natural resource of communication and unit linkage (see Linell 2005; on the incrementality of spoken language, cf. also Chafe 1988). The tendency to use more compact idea-units also enables speakers to communicate, say, explanations and conclusions, as commentaries that allow for the inclusion of shifting personal viewpoints rather than as mere facts about the world. By doing this, speakers make room for their personal involvement, and more importantly, they refer to and evoke interpersonal involvement in the dynamic, collective flow of social interaction (cf. Chafe & Danielewicz 1987; on involvement in spoken language). In presenting the gist of an argumentation in a separate TCU, the speaker makes it possible for the recipients to comment precisely on this sum-
Constructing reasoning 147
mary, which could be the basis for forming a mutual understanding of the part of an argumentation that really counts. It can be generally noted that the connectives studied here are an intrinsic part of the global interactional activity of reasoning: putting forward arguments, expanding them, giving motivations for them, retreating from them, returning to them and summarizing them. The units introduced by the connectives are thus always “subsequents”, i.e. they can be understood to stand in a responsive relation to something that was said earlier. In their paratactic function, the connectives seem to expand communicative projects (discourse units) rather than clausal or sententical packages – that is, they introduce units that occur outside the realms of “a sentence” or a single “idea”. But the responsive side of the units reflects another, pragmatic kind of dependency they possess. This dependency is not of a hierarchical, grammatical kind, but of a sequential kind – a matter of discourse organization. Hence, we propose that clausal connections in conversational language cannot be described simply in terms of either syntactic (and semantic) independency or dependency (cf. also Givón, 1990, who proposes different levels of subordination). A relevant difference could be drawn between predication subordination and discourse subordination. The former is typical of syntactically dependent clauses, which are a heavily used device in written language and well described in chapters on subordination in traditional grammars. The latter is typical of the causal and consecutive clausal utterances of the type which we have studied in conversational language. The clause (or the action that is put forward by the clause) is not a syntactically or semantically necessary complement (e.g. an adverbial specifier) but rather a continuation or an addition to an utterance which could have stood alone in its sequential context. This continuation formulates an own predication, quite often an epistemically modified commentary in a causal or concluding guise. But because the commentary per definitionem presupposes something prior which can be the object of the commentary, it is in a pragmatic, albeit not syntactic, sense dependent on a discoursal “antecedent”. It is our contention that the general “complementizer” att is carrying this pragmatically “subordinating” signal value when attached to för, så, and men in paratactic constructions. Although not being a subordinator proper in these uses (but rather a discourse marker), att nevertheless explicates that the unfolding predication is dependent on some source in the prior discourse context. By saying this, we wish to challenge the traditional view of att as a pleonastic, unmotivated element, although it superficially seems that it could be dropped without a real loss of meaning and only the simple forms för, så and men be used. But the possibility to leave out a certain grammatical element does not mean that the element is void of function when it is produced in a construct(ion). It is, for
148 Jan Lindström and Anne-Marie Londen
example, a general convention to leave out prepositions as well as subordinating conjunctions in certain contexts: Jag är rädd (för) att skrämma iväg honom ‘I am afraid (of) to scare him away’, Jag tycker (att) det är bra ‘I think (that) it is good’, Filmen (som) vi såg var spännande ‘The film (that) we saw was exciting’. Rather, the functional specification of an expression increases when it is constructed with maximal structural completeness. The use of att together with för, så, and men will thus foreground the discoursal and pragmatic dependency of a clause. If the marker is left unpronounced, this dependency relation is understood as a consequence of the causal, consecutive and adversative connective features of för, så, and men. These matters are quite complex and there is no adequate way of describing the differences between predication and discourse subordination within traditional grammar. Here the attribute-value matrix provided by the framework of construction grammar has proved to be a powerful tool. It allows us to describe the clausal units in a constructional representation that goes beyond the sentence level. With the help of matrix formalizations we have been able to identify the key (or indeed, minimal) differences, not only between paratactic and hypotactic constructional variants but also between sub-variant uses of causal and consecutive constructions: predication-complementing causal and consecutive constructions (hypotactic) as well as discourse-evaluating and action-justifying causal and consecutive constructions (always paratactic). Indeed, the formalism can be extended to capture the essence of the “show concession”, i.e. the discourse adversative-consecutive construction, which involves the connective men att in its concluding phase. The interaction-sequential formalism which was used in the present analysis is not, however, definite in any way but only a starting point for the quest of an empirically grounded discourse construction grammar.
Data GRIS. The kernel corpus of Swedish conversations made available within the project Samtalsspråkets grammatik (Grammar in conversation: a study of Swedish); TT:3 is one episode of the TV-program Tryck till, transcribed by Jenny Öqvist (also the English translation), ISK, Linköping University; GML:4 is a focus group discussion between five farmers and a moderator, transcribed by Henric Bagerius, Tema K, Linköping University; SÅ:INF is a conversation between four elderly women, transcribed by Karin Ridell, FUMS, Uppsala University. See http://www. tema.liu.se/tema-k/gris/.
Constructing reasoning 149
HUSA. A corpus of group interviews recorded in Swedish schools in Helsinki (2– 4 adolescents and a young interviewer per group); collected in the project Språk och attityder bland helsingforssvenska ungdomar (Language and attitudes among Helsinki Swedish adolescents). Transcribed by Charlotta af Hällström, Department of Scandinavian languages and literature, University of Helsinki. SAM. A corpus of Helsinki Swedish conversations collected in the project Svenska samtal i Helsingfors (Swedish conversations in Helsinki). V1 is a conversation between six men in their thirties, V2 is a conversation between four elderly women, M1 is a board meeting in a kindergarten, M2 is a board meeting in a youth organization. Transcribed by Ylva Forsblom-Nyberg and Hanna Lehti-Eklund, Department of Scandinavian languages and literature, University of Helsinki.
References Allén, Sture (1965). Johan Ekeblads brev till brodern Claes Ekeblad 1639–1655. [Nordistica Gothoburgensia 2]. Göteborg: Acta Universitatis Gothoburgensis. Allwood, Jens (Ed.) (1999). Talspråksfrekvenser. Second edition. Göteborg: Göteborgs universitet. Andersson, Lars-Gunnar (1975). Form and function of subordinate clauses. [Gothenburg monographs in linguistics 1]. Göteborg. Antaki, Charles & Margaret Wetherell (1999). Show concessions. Discourse Studies, 1, 2–27. Anward, Jan (2003). Att. Språk och stil, 13, 65–85. Auer, Peter (1996). On the prosody and syntax of turn-continuations. In E. Couper-Kuhlen & M. Selting (Eds.), Prosody in conversation [Studies in Interactional Sociolinguistics, 12] (57–100). Cambridge: Cambridge University Press. Bergroth, Hugo (1928). Finlandssvenska. Handledning till undvikande av provinsialismer i tal och skrift. Helsingfors: Schildts. Bolinger, Dwight (1977). Meaning and Form. London & New York: Longman. Chafe, Wallace (1988). Linking intonation units in spoken English. In J. Haiman & S.A. Thompson (Eds.), Clause combining in grammar and discourse [Typological Studies in Language, 18] (1–27). Amsterdam & Philadelphia: Benjamins. Chafe, Wallace & Jane Danielewicz (1987). Properties of spoken and written language. In R. Horowitz & S.J. Samuels (Eds.), Comprehending oral and written language (83–113). New York: Academic Press. Fillmore, Charles, Paul Kay & Mary Catherine O’Connor (1988). Regularity and idiomaticity in grammatical constructions: the case of let alone. Language, 64, 501–538. FO = Ordbok över Finlands svenska Folkmål (1976–). Helsingfors: Forskningscentralen för de inhemska språken. Fried, Mirjam & Jan-Ola Östman (2004). Construction grammar. A thumbnail sketch. In M. Fried & J.-O. Östman (Eds.), Construction grammar in a cross-language perspective (11– 86). Amsterdam & Philadelphia: John Benjamins.
150 Jan Lindström and Anne-Marie Londen
Givón, Talmy (1990). Syntax: A functional-typological introduction. Volume 2. Amsterdam & Philadelphia: John Benjamins. Günthner, Susanne (1996). From subordination to coordination? Verb-second position in German adverbial clauses. Pragmatics, 6, 323–370. Heine, Bernd & Tanja Kuteva (2002). World lexicon of grammaticalization. Cambridge: Cambridge University Press. Heritage, John & D.R. Watson (1980). Aspects of the properties of formulations in natural conversations: Some instances analysed. Semiotica, 30–3/4, 245–262. Houtkoop, Hanneke & Harrie Mazeland (1985). Turns and discourse units in everyday conversation. Journal of Pragmatics, 9, 595–619. Jefferson, Gail (1983). On a failed hypothesis: ‘Conjunctionals’ as overlap-vulnerable. Tilburg papers in language and literature, 28. Department of Language and Literature, Tilburg University. Jörgensen, Nils (1976a). Meningsbyggnaden i talad svenska. Lund: Studentlitteratur. Jörgensen, Nils. (1976b). Satsfogning – några preliminära funderingar. In Nordiska studier i filologi och lingvistik, Festschrift to Gösta Holm, July 8, 1976 (234–238). Lehti-Eklund, Hanna (2002). Om att som diskursmarkör. Språk och Stil, 11, 81–118. Lerner, Gene (1995). Turn design and the organization of participation in instructional activities. Discourse Processes, 19, 111–131. Lindholm, Camilla & Jan Lindström (2004). Dislokationer och fristående topiker. Två annexkonstruktioner i talad svenska. In B. Melander, U. Melander Marttala, C. Nyström, M. Thelander & C. Östman (Eds.), Svenskans beskrivning 26 (196–207). Uppsala. Lindström, Jan & Anne-Marie Londen (2001). men att de öppnades ju nog en ny värld för mej. Kombinationen men att – en ovårdad och onödig finlandssvensk variant? In M. Nordman, C. Laurén, S. Björklund & M. Koskela (Eds.), Svenskan i Finland 6 (104–116). Vasa: Proceedings of the University of Vaasa. Linell, Per (1998). Approaching dialogue. Talk, interaction and contexts in dialogical perspectives. [Impact: Studies in Language and Society]. Amsterdam & Philadelphia: Benjamins. Linell, Per (2005). En dialogisk grammatik? In J. Anward & B. Nordberg (Eds.), Samtal och grammatik. Studier i svenskt samtalsspråk (231–328). Lund: Studentlitteratur. Loman, Bengt & Nils Jörgensen (1971). Manual för analys och beskrivning av makrosyntagmer. Lund: Studentlitteratur. Lundström, Gudrun (1939). Studier i nyländsk syntax. Stockholm: Norstedt & Söner. Michaelis, Laura & Knud Lambrecht (1996). Toward a construction-based theory of language function: The case of nominal extraposition. Language, 72, 215–247. ODS = Ordbog over det danske sprog (1923). København: Gyldendal. Ogden, Richard, Auli Hakulinen & Liisa Tainio (2004). Indexing ‘no news’ with stylization in Finnish. In E. Couper-Kuhlen & C.E. Ford (Eds.), Sound patterns in interaction (299–334). Amsterdam & Philadelphia: John Benjamins. Saari, Mirja (1975). Talsvenska. En sociolingvistisk studie över syntaktiska drag i intervjusvar [Studier i nordisk filologi 60]. Helsingfors: Svenska litteratursällskapet i Finland. SAG = Teleman, Ulf, Staffan Hellberg & Erik Andersson (1999). Svenska Akademiens grammatik. Volumes 1–4. Stockholm: Svenska Akademien. SAOB = Ordbok över svenska språket (1898–). Lund: Svenska Akademien. Schegloff, Emanuel A. (1996). Turn organization: One intersection of grammar and interaction. In E. Ochs, E.A. Schegloff & S.A. Thompson (Eds.), Interaction and grammar [Studies in Interactional Sociolinguistics 13] (52–133). Cambridge: Cambridge University Press.
Constructing reasoning 151
Schegloff, Emanuel A. (2000). On turn’s possible completion, more or less: Increments and trailoffs. Ms, paper prepared for the EuroConference on Interactional Linguistics, Spa, Belgium, September 2000. Scheutz, Hannes (2001). On causal clause combining. The case of weil in spoken German. In M. Selting & E. Couper-Kuhlen (Eds.), Studies in Interactional Linguistics [Studies in Discourse and Grammar 10] (111–139). Amsterdam & Philadelphia: John Benjamins. Schiffrin, Deborah (1987). Discourse markers. Cambridge: Cambridge University Press. Schiller, Karl & A. Lübben (1875–81). Mittelniederdeutsches Wörterbuch, 1–6. Bremen: J Kühtmann’s. Steensig, Jakob (1998). Om fordi i forskellige sætningstyper i dansk talesprog. In K. Kristensen (Ed.), Selskab for Nordisk Filologi i København. Årsberetning 1996–1997 (179–192). København: Selskab for Nordisk Filologi. Steensig, Jakob (2001). Sprog i virkeligheden. Bidrag til en interaktionel lingvistik. Aarhus: Aarhus universitetsforlag. Sweetser, Eve E. (1990). From etymology to pragmatics. Metaphorical and cultural aspects of semantic structure. Cambridge: Cambridge University Press. Teleman, Ulf (1967). Bisatser i talad svenska. In G. Holm (Ed.), Svenskt talspråk. Fem studier. Stockholm: Almqvist & Wiksell. Thorell, Olof (1973). Svensk grammatik. Stockholm: Esselte Studium. Wessén, Elias (1956). Svensk språkhistoria III. Grundlinjer till en historisk syntax [Stockholm studies in Scandinavian philology 19]. Stockholm: Almqvist & Wicksell. Wide, Camilla (2002). Perfect in dialogue. Form and functional potential of the vera búinn að + inf. construction in contemporary Icelandic [PIC Monographs 3.] Helsinki.
Transcription symbols and grammatical abbreviations [ ] = . , ¿ ↑ ↓ word wo:rd $word$ >word< <word> ºwordº (word) wo-
a point of overlap onset a point at which two overlapping utterances both end a single continous utterance or two ”latching” utterances a falling intonation contour a continuing intonation contour a somewhat rising intonation contour prosodic up-step prosodic down-step a stressed syllable a stretching of a sound smile voice, possibly accompanied with a laughter compressed or rushed talk slower talk or drawl quiet or soft voice uncertain transcription a hearable cut-off
152 Jan Lindström and Anne-Marie Londen
~word~ .word ( ) hh .hh (.)
a stylized prosodic contour a word pronounced with a hearable inbreath no hearing or speaker identification a hearable out-breath a hearable in-breath a micropause, less than 2/10 of a second
(0.5) ((laughing)) (---) ?: man
a pause measured in tenths of a second transcriber’s comments omitted continuing talk uncertain speaker identification
neg pass prt refl
Sw. generic indefinite pronoun (cf. ‘one’ in English) negative operator passive form of the verb ‘particle’; most often an epistemic marker reflexive pronoun
Subject index
A abstractification 48 address term 47 adjacency pair 123 agent responsibility (avoidance) 39, 43, 46 agglutinating language 41 anacoluthon 113 analogy 6, 38, 48 analytic language 41 animacy hierarchy 65, 78 architecture of language 28–29, 56, 85 areal 37, 42 Circum-Baltic area 42 argument structure 55 argument structure constructions 2, 11, 14, 30, 33, 65 attribute-value 48, 50, 102, 109, 122, 135 B benefactive 94–96 benefactive alternation 19 benefit transfer 91, 96–98, 100, 102 British National Corpus 13 C causation 66 causative 67–68, 71–75 causative alternation 19 coercion 56, 60, 66, 83 cognition 52, 56, 85 cognitive 5, 38, 40, 46, 48, 51, 69, 78, 89, 101ff. cognitive frame 102 Cognitive Grammar 6, 38
cognitive organization 1, 52, 89, 106, 116, 145 communicative project 108, 121, 124, 125, 123 Conceptual Semantics 6 conceptual space 55, 57, 58, 60–63 conceptualization 46, 102 constructicon 39 construction, definition 16 constructional pattern 12, 14, 33, 116 contact see language contact context 38 conventionalization 52, 84 conventionalized instantiation 21 conventionalized world knowledge 22, 25, 32 Conversation Analysis 5, 6, 105 conversational implicature 48 cooperative principle 83 corpus studies 2 D detransitivizing languages 67 dialect 40–41, 111, 113 dialogic 47, 114, 123 discourse 2–6, 37, 38, 51–52, 91–92, 97, 102, 106, 109, 114–119, 124ff., 139 discourse marker 43, 47, 112, 147 discourse patterns 3, 37, 38, 43 Genericity 43, 45, 46, 49, 51 discourse subordination 147–148 discourse unit 123, 135, 147
drift 44 E emergence of constructions 40–41, 44, 48–51 epistemic causality 117, 119, 124 event participant 21–23, 26, 102 expansion 38, 44, 46, 48 F force dynamics 26, 27 form-meaning discrepancy 12–14, 33 formulation 132 frame 3, 13, 16–17, 38, 43, 47, 91, 100, 102 frame Element 13 FrameNet 20, 38n Frame Semantics 5, 13, 20, 22, 38n, 47, 89, 92, 102–103 frequency 23, 63, 64, 106, 109, 111, 138 functional load 4, 45 G generic 3, 37ff., 50–51 generic zero 44 genericity pattern see discourse patterns globalization 42, 46, 47, 50 grammaticalness 14 grammaticalization 2, 6, 64, 92, 102, 103, 115 H honorifics (Japanese) addressee 4, 90–93, 100, 102 non-subject 4, 89ff., 94ff. performative 90ff., 93, 97
154 Subject index
propositional 89, 93 referent 89ff., 93, 100 subject 89–95, 97, 99 I iconicity 6 Ideology 38, 46, 51 idiomatic construction 14, 18ff. idiosyncratic subcategorization 21 implicature 4, 48, 55–57, 64, 78–79, 83 increment 114, 121 individual in language change 38, 42 inference 64 information structure 2, 65 inheritance 21, 40, 45, 116 integration of verbs 25–27 interaction 5, 102, 105, 115, 122ff., 142ff. interactional frame 101 intersubjective 5, 89, 91, 101 intonation paragraph 124, 125 intonation phrase 118, 124, 125, 144, 146 intransitive argument structure constructions 65, 70 irony 49 isolating language 41 L language acquisition 2 language contact 38, 42, 44–46, 49 left dislocation 115, 141 lexicon 3, 4, 11, 22, 28, 29, 55ff. linguistic sign 16, 65, 77, 80 M markedness 61ff. material/product alternation 18 metaconstruction 51 mini-construction 3, 21, 22, 23, 24, 31, 33, 34
morphosyntactic markedness 61 motivation in grammar 6, 45, 105, 111, 116, 128, 138 N non-factual 42–43, 45, 48 O Otherness 47 P parts of speech 4, 55ff. passive Finnish sä passive 37–52 Finnish Häkkinen passive 41 Finnish imperialistic passive 41 point of view 48 polyphony 49 pragmatic particle 43, 47 pragmatics 2, 22, 24, 30, 31, 51, 55, 64, 77ff., 91–94, 96, 100, 105ff. precategoriality 55, 58, 85 preclassification for syntactic categories 84 predication subordination 147 pro tem speakership 135 productivity 25, 27, 29–33 productivity continuum 31 prototypical patient 17 putative 56, 67–72, 80 R Radical Construction Grammar 2, 4, 60–62 real-world causality 117 relation between lexicon and syntax 3, 4, 11, 22, 28–29, 56, 77 reshaping 46 responsibility see agent responsibility S scaffolding 56, 59, 60, 62, 66, 83
semantic change 92, 96 semantic restrictions 24, 33 semantic scope 12, 33 semantic space 27, 33, 60 semantic specificity 12, 33 show concession 142, 148 space builder 47 speech act 92, 137 speech act causality 92, 117, 126 speech context 91f, 98–102 speed of language change 38, 48–50 stereotype 4, 62, 64, 78 stereotypical inference 64–65, 77, 83 sub-culture 49 subject-prominent language 41, 44 subjectification 38, 48 subjective 94, 96, 130, 137 subjectivity 92 syntactic frame 3, 11, 12, 16–17, 32 T trail-off 135 transitive argument structure constructions 65, 70, 71 transitivity 66 transitivizing languages 67 turn constructional unit 114, 123 U uniformity hypothesis 85 V variability 3, 37, 38, 44, 46, 50, 52 verb class 18 verbs of creation 16, 18–20, 24 W word-class indicating construction 55–56, 58, 60–61, 64 Word Grammar 6
Index of constructions
Finnish Impersonal 37, 39, 41, 45, 46 Indefinite-person 39, 46, 49 Generic sä (the sä passive) 37–52 Genericity, general 44, 47–48, 50–51 Transitive 45 Japanese o-V-suru 89–94, 96–102 Swedish action causal 126 action consecutive 136 discourse adversative-consecutive 142 discourse causal 125 discourse consecutive 135 Indefinite-person 42 predication causal 122 predication consecutive 137 Russian Indefinite-person 41
In the series Constructional Approaches to Language the following titles have been published thus far or are scheduled for publication: 5 4 3 2 1
Leino, Jaakko (ed.): Constructional Reorganization. 2008. vi, 155 pp. Fried, Mirjam and Hans C. Boas (eds.): Grammatical Constructions. Back to the roots. 2005. viii, 246 pp. Östman, Jan-Ola and Mirjam Fried (eds.): Construction Grammars. Cognitive grounding and theoretical extensions. 2005. viii, 325 pp. Fried, Mirjam and Jan-Ola Östman (eds.): Construction Grammar in a Cross-Language Perspective. 2004. vi, 209 pp. Kuno, Susumu and Ken-ichi Takami: Functional Constraints in Grammar. On the unergative– unaccusative distinction. 2004. ix, 242 pp.
