Evidentiality and Epistemic Modality in Spanish (Semi-)Auxiliaries
≥
Applications of Cognitive Linguistics 5
Editors...
38 downloads
822 Views
1MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
Evidentiality and Epistemic Modality in Spanish (Semi-)Auxiliaries
≥
Applications of Cognitive Linguistics 5
Editors Gitte Kristiansen Michel Achard Rene´ Dirven Francisco J. Ruiz de Mendoza Iba´n˜ez
Mouton de Gruyter Berlin · New York
Evidentiality and Epistemic Modality in Spanish (Semi-)Auxiliaries A Cognitive-Functional Approach
by Bert Cornillie
Mouton de Gruyter Berlin · New York
Mouton de Gruyter (formerly Mouton, The Hague) is a Division of Walter de Gruyter GmbH & Co. KG, Berlin
앝 Printed on acid-free paper 앪
which falls within the guidelines of the ANSI to ensure permanence and durability.
Library of Congress Cataloging-in-Publication Data Cornillie, Bert, 1975⫺ Evidentiality and epistemic modality in Spanish (semi-)auxiliaries : a cognitive-functional approach / by Bert Cornillie. p. cm. ⫺ (Applications of cognitive linguistics ; 5) Includes bibliographical references and index. ISBN 978-3-11-018611-6 (hardcover : alk. paper) 1. Spanish language ⫺ Auxiliary verbs. 2. Spanish language ⫺ Semantics. 3. Spanish language ⫺ Modality. I. Title. II. Series. PC4315.A8C67 2007 4651.1⫺dc22 2007042931
Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available in the Internet at http://dnb.d-nb.de.
ISBN 978-3-11-018611-6 ISSN 1861-4078 쑔 Copyright 2007 by Walter de Gruyter GmbH & Co. KG, D-10785 Berlin All rights reserved, including those of translation into foreign languages. No part of this book may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Printed in Germany
Acknowledgements Although writing a book inevitably isolates the author from family and friends, it also strengthens professional, friendship and family ties. Let us for once look on the bright side of life: thanks to the great support of so many people I have been able to finish this book. First and foremost I would like to thank my supervisor, Professor Nicole Delbecque, and my co-supervisor, Professor Ricardo Maldonado. Their enthusiasm, warmheartedness, as well as their theoretical and descriptive sharpness have been very stimulating for me and resulted in considerable improvements of the earlier draft versions of the chapters in the dissertation that has led to this book. I especially owe a large debt of gratitude to Nicole Delbecque from the University of Leuven for taking up the challenge of supervising someone who came from literary and cultural discourse studies, which was quite a risk. Her patience, trust and sustained support made the task of writing more feasible. I also thank the University of Leuven for my predoc position as teaching assistant and my postdoctoral mandate from its Research Council. I want to thank Ricardo Maldonado wholeheartedly for his rigorous supervision during my stay at the Universidad Autónoma de Querétaro and the UNAM in the fall and winter of 2001-2002. Being a splendid host he taught me a lot about linguistics as well as about life outside academia. The five month stay in Mexico, which was funded by the Fund for Scientific Research – Flanders (Fonds voor Wetenschappelijk Onderzoek – Vlaanderen), was a crucial step forward to the final stage of my dissertation. I also thank Professors Salvador Gutiérrez Ordóñez, Kristin Davidse and Béatrice Lamiroy who agreed to be members of the jury of my dissertation. Kristin Davidse gave me invaluable, discrete support (and warnings) during both the doctoral and the postdoctoral project. In addition, her meticulous corrections were of great help. Special thanks to Béatrice Lamiroy for commenting drafts of my dissertation. Her criticisms were a guideline for a series of important improvements. I also want to thank Professor Nadia Lie for the nice collaboration we have had ever since 1995, for her encouragement and faith in the success of my research project(s). In addition to the above-mentioned people, a series of skilled linguists with life-long experience helped me find my way in the daunting field of linguistics. I am greatly indebted to Karel van den Eynde, who formed me in distributional, structural linguistics. I will never forget the time he spent
vi
Acknowledgements
discussing analyses and arguments with me. Pieter Seuren deserves my gratitude for sharing his insights on top - down linguistic models and for his unconditional support. I am indebted to Jan Nuyts for squeezing time from his tight agenda to look together at a list of evidential and epistemic constructions. I especially thank him for his kind job offer at the Center of Grammar, Cognition and Typology of the University of Antwerp, where I worked half time during two years within the framework of the Concerted Research Project (GOA) on Mood and Modality. Thanks are due to Johan van der Auwera for the many nice discussions and his helpful remarks. I want to thank Arie Verhagen for his encouragement and stimulating discussions. Many thanks to Péter Pelyvás for refining my knowledge of the grounding system, to Angela DiTullio for remarks on my preprints and to Concepción Company for insisting on the need for a diachronic analysis of semi-auxiliaries and for commenting on my papers. I want to thank René Dirven for his encouragement, his trust and wise advice as well as for his comments during the process of converting my dissertation into this book. I am grateful to John-Charles Smith (Oxford), Carita Paradis (Lund), Zlatka Guentchéva (CNRS - Paris) and Claus Pusch (Freiburg) for their kind invitations to present parts of my work at their respective departments. I also want to thank the following new-generation scholars for stimulating discussions on draft versions or e-mail questions: Elisa Benavent Payá, Frank Brisard, Kris Buyse, Costas Canakis, Barbara De Cock, Alex Demeulenaere, Helder De Schutter, Luna Filipovic, Marián Gómez, Patrick Goethals, Hilde Hanegreefs, Petar Kehayov, Peter Lauwers, Alexander Loengarov, Tanja Mortelmans, Carmen Núñez, Álvaro Octavio de Toledo y Huerta, Geert Sterck, Ana Stulic, Marina Terkourafi, José Tummers, Katleen Van den Steen, Freek Van de Velde, and last but not least, JeanChristophe Verstraete. Dagmar Divjak and Karen Lahousse merit a special mention for their tireless support and energy. I will never be able to thank them enough for all the times they read through one of my texts, for their quick answers to my many doubts or questions and, crucially, for the unforgettable moments of deep solidarity. Dagmar, Karen, your moral and intellectual support have been invaluable. I also thank my many other colleagues, friends and ex-colleagues at the departments of Linguistics of Leuven and Antwerp and for their daily kind words in the university corridors. Also thanks to Gitte Kristiansen, Barbara De Cock, Birgit Sievert and Wolfgang Konwitschny for helping me with the preparation of the manuscript.
Acknowledgements
vii
Without their knowing, many hard-working people I met in the South made the completion of this book a categorical imperative: the young vendors in Chiapas, the Quechua farmers around Apu Ausangate, the much too young porters in the Annapurna region, the Bedouin youth in Wadi Rum, the Palestinian students of Birzeit University, the people from Sulaymaniyah University, the eloquent Iranian students in Hamedan, Kermanshah and Isfahan, the youth leaders from Batman and Diyarbakir who struggle for education in their Kurdish mothertongue. The awareness that the moments of despair I went through could not stand the comparison with the life conditions of these people, stimulated me to keep working. This doctoral project and the postdoctoral project have been conducted with the quiet but solid support from my home base in Bekegem. Spending a day in this small West-Flemish village always made me return to Leuven with renewed strength. From the deepest of my heart I want to thank my parents for preserving my primary fascination with things and the relations between them. Many thanks to my brother Jan and sister Tine for the instant communication, illustrative of our strong connection, despite the many different places we have been living in. I also want to thank my parents-in-law for their encouragement and the nice weekends in Donk. With me in all this was my wife Sibien. She has not only been a shining alternative for the grey faculty building, but is also a driving force in imposing a well-balanced combination between quality life and academic subculture. Coming home meant tasting the real life, with the funny and less funny stories about school, transport and family. I can’t thank her enough for her moral support, understanding and independence. Together with our Merlijn and Reine, we will now catch up with the biking and hiking weekends. Leuven, September 2007
Contents Acknowledgements
v
Chapter 1: Background, objectives and methodology
1
1. Previous accounts of parecer 2. Previous accounts of poder, deber and tener que 3. Previous accounts of amenazar and prometer 4. Broad and narrow definitions of evidentiality 5. Corpus research 6. Cognitive-functional account 7. Outline of the book
3 5 6 6 7 9 13
Chapter 2: Evidentiality with parecer and resultar
15
1. Frequency distribution 2. From source of information to (inter)subjectivity 2.1. Evidentiality: a typology 2.2. Speaker commitment and reliability 2.3. (Inter)subjectivity 3. A three-dimensional analysis of parecer 3.1. Evidential types 3.2. Speaker commitment and reliability 3.3. (Inter)subjectity 4. A three-dimensional analysis of resultar 4.1. Evidential types 4.2. Speaker commitment and reliability 4.3. (Inter)subjectivity 5. Conclusions
16 18 18 20 23 25 25 37 42 48 49 52 53 55
Chapter 3: Degrees of transparency and subjectification with parecer and resultar
57
1. Transparency 1.1. Definition of transparency 1.2. Transparent parecer vs. non-transparent resultar 2. Transparency and focal prominence 2.1. Focal prominence 2.2. Parecer + infinitive and parece que
57 57 59 63 63 64
x
Contents
2.3. Resultar + infinitive and resulta que 2.4. Left-dislocation 3. Subjectification 3.1. Diachronic subjectification 3.2. Synchronic subjectification 4. Conclusions
67 68 69 69 79 83
Chapter 4: Evidentiality and modality with prometer and amenazar 85 1. Previous research lines on promise and threaten 1.1. Epistemicity, inferentiality and evaluative orientations 1.2. Different epistemic values 1.3. Transparency and subjectification 2. The lexical origins of the speaker commitment 2.1. Frequency distribution 2.2. Subject commitment 2.3. The different impact of conditional clauses 2.4. Illocutionary force 3. The modal orientation of amenazar and prometer 3.1. Working hypothesis 3.2. Prediction with amenazar 3.3. Refinement of the non-realization criterion 3.4. Prediction with prometer 3.5. Theoretical implications 4. The evaluative orientation of amenazar and prometer 4.1. Working hypothesis 4.2. Evaluation with prometer 4.3. Evaluation with amenazar 5. Conclusions
86 87 90 92 96 96 98 101 102 104 104 104 106 111 115 116 116 117 119 120
Chapter 5: Partial transparency with modal prometer and amenazar
122
1. Restrictions on the infinitive 1.1. Working hypothesis and methodology 1.2. Global distribution 1.3. Infinitives with amenazar 1.4. Infinitives with prometer 1.5. Infinitives absent in the corpus 2. Restrictions on the subject 2.1. Working hypothesis and methodology 2.2. Subjects of lexical amenazar and prometer
122 123 123 126 128 130 135 136 137
Contents
xi
2.3. Subjects of modal amenazar 2.4. Subjects of modal prometer 3. Conclusions
138 142 148
Chapter 6: Subjectification of prometer and amenazar
150
1. A diachronic approach to promise and threaten 1.1. An overview 1.2. Metaphorical extension 1.3. Metonymic extension 2. A diachronic analysis of amenazar and prometer 2.1. Amenazar 2.2. Prometer 3. Conceptual subjectification 3.1. Attenuation 3.2. Subjectification 3.3. Focal prominence 4. Conclusions
150 150 153 157 158 158 168 174 175 176 178 179
Chapter 7: Evidentiality, modality and transparency with poder, deber and tener que
181
1. Frequency distribution 2. Poder: preference for epistemic modality 3. Deber and tener que: on inferentiality and hearsay 3.1. Deber 3.2. Tener que 3.3. Deduction and induction 3.4. Non-speaker oriented evidentiality 4. Deber: preference for inductive inferences 4.1. General tendencies 4.2. Inductive inference 4.3. Epistemic modality 4.4. Deductive inference 4.5. Reportive readings 5. Tener que: preference for deductive inferences 5.1. General tendencies 5.2. Deductive inference 5.3. Inductive inference 5.4. Epistemic modality 5.5. Reportive readings 5.6. Discussion
182 184 187 187 189 191 195 199 199 200 201 202 204 206 206 207 208 210 211 212
xii
Contents
6. Transparency 6.1. Classification 6.2. Infinitives and subjects with poder 6.3. Infinitives and subjects with deber 6.4. Infinitives and subjects with tener que 6.5. Summary 7. Conclusions
213 213 214 216 219 221 222
Chapter 8: Epistemic and evidential modals: modal grounding and subjectification
224
1. Diachronic subjectification vs. modal grounding 1.1. Diachronic subjectification 1.2. Modal grounding 2. On formal and conceptual aspects of grounding 2.1. Grounding predications and subjectification 2.2. Content and profile 2.3. Dynamic evolutionary model 3. Aspectual and modal marking 3.1. Methodology 3.2. Infinitive and Spanish modals 4. The shift from deontic/dynamic to epistemic modality 4.1. Force dynamics and metaphor 4.2. Reference-point 4.3. Subjectification and interplay of forces 4.4. Different epistemic and evidential readings 5. Conclusions
225 225 230 234 234 239 240 241 242 242 244 244 246 248 250 251
Chapter 9: General conclusions
253
1. Evidential and epistemic qualifications 2. Linguistic representation 3. Further research
253 255 257
Notes
262
Chapter 1 Chapter 2 Chapter 3 Chapter 4 Chapter 5 Chapter 6
262 266 269 271 274 277
Contents
xiii
Chapter 7 Chapter 8
279 282
References
286
Author index Subject index
307 310
Chapter 1 Background, objectives and methodology
Language is a means of social communication by which speakers not only transmit information but also attempt to define their own position toward what they are communicating. Speakers often comment on the propositional content so as to guide the interlocutor(s) to a viable interpretation of what is said or to attenuate the statement in order not to undermine their own position. Since a speaker wants to avoid the risk of losing face, (s)he often suggests that the proposition uttered is based on a certain mode of knowledge and also indicates how certain they are about the truth of the proposition. These two dimensions of communication go by specific terms in modern linguistics: evidentiality refers to the “encoding of the speaker’s (type of) grounds for making a speech act” (Faller 2002: 2), while epistemic modality refers to the “evaluation of the chances that a certain hypothetical state of affairs under consideration (or some aspect of it) will occur, is occurring or has occurred in a possible world” (Nuyts 2001b: 21). In recent years, the concepts of evidentiality and epistemic modality have become increasingly important in the field of functional linguistics. Like the participant structure of the state of affairs, the speaker’s attitude toward the proposition is nowadays seen as a dimension of language in its own right. The study of sources of information and modes of knowing (evidentiality) and the analysis of the speaker’s judgments of the likelihood of the proposition (epistemic modality) are two ways of dealing with the speaker’s stance in discourse. Other ways of addressing stance consist in examining the illocutionary force present in the utterance, investigating the differences between declarative sentence types and interrogative ones or comparing different types of clause combining. This book aims to give a systematic and comprehensive corpus-based analysis of a series of (semi-)auxiliary verbs that are attributed evidential and epistemic modal readings in the literature.1 The verbs under examination can be divided into three groups, namely: (i) the evidential semi-auxiliaries parecer ‘to seem’ (1) and resultar ‘to appear/turn out’ (2) (1)
Más bien se han floreado mutuamente y prometido respaldo casi eterno. Esto parece confirmar que Fujimori y Hermoza están es-
2 Background, objectives and methodology trechando sus vínculos y debilitando a Montesinos (Notic: Perú: Caretas) ‘Rather they have cheered each other and promised almost eternal support. This seems to confirm that Fujimori and Hermoza are strengthening their ties and weakening Montesinos.’ (2)
La escuela laica que escogimos para Philip resulta estar llena. (Notic: España: ABC) ‘The secular school that we chose for Philip turns out to be full.’
(ii) the modal constructions with semi-auxiliary amenazar ‘to threaten’ (3) and prometer ‘to promise’ (4) (3)
El importante encuentro amenaza ser ensombrecido por medidas de presión de varios sectores, por lo que el gobierno dispuso la intervención de las Fuerzas Armadas. (Notic: Bolivia: ERBOL: 04/15/96) ‘The important meeting threatens to be overshadowed by pressure measures from several sectors, as a consequence of which the government ordered the intervention of the Armed Forces.’
(4)
Este invierno promete ser llovedor. -Llovedor, patrón, promete ser este invierno. -Se ve mucha chicharra en el monte. Y es la señal. (Barrios, Eduardo. Gran señor y rajadiablos) ‘This winter promises to be rainy. Rainy, boss, this winter promises to be. One notices lots of crickets in the mountains. And this is the signal.’
(iii) the modal auxiliaries poder ‘may/can’ (5), deber ‘must’ (6) and tener que ‘to have to/should’ (7) (5)
... una de las cosas que quizás... eh.... actualmente más me ha llamado la atención es... eh.... eh... la importancia que le está dando el cine a lo estético. Quizás puede ser por la competencia que hay con otros medios de comunicación de masas. (Habla Culta: Santiago: M8) ‘... one of the things that at the moment strikes me most is the importance that the film industry attributes to the esthetic. This may be due to the competition with other mass media.’
Previous accounts of parecer
3
(6)
Bueno yo... yo oí en el radio de que uno de los rehenes... había logrado escaparse por una ventana pero antes había tratado de suicidarse ¿no? lo cual me hace creer que bueno que el pobre debe de haber estado sufriendo horrores. (Habla Culta: Caracas: M15) ‘Well I.. I heard on the radio that one of the hostages succeeded in escaping through a window, but that before, he had tried to commit suicide, hadn’t he?, which makes me think, well, that the poor guy must have been suffering tremendously.’
(7)
... Claro que te digo que si las llevó a un buen joyero a España el joyero le tiene que haber dicho que eran muy baratas porque eran azulísimas, Claudia. (Habla Culta: Mexico: M5) ‘Sure, I am telling you that if he brought them to a jeweler in Spain, the jeweler must have told him that they were very cheap because they were very blue, Claudia.’
All seven verbs account for the speaker’s relation to reality and emphasize either that the speaker has evidence at his/her disposal to report on the event expressed in the infinitive (e.g. parecer and resultar), or that (s)he has assessed the likelihood that the proposition can be true (e.g. poder) or must be true (e.g. deber and tener que).For a structural analysis of these verbs, I refer to Cornillie (2004b: chapters 1–4) and Cornillie (2005d). This volume is concerned with the question whether evidentiality and epistemic modality apply to each of the three groups of verbs. The cognitivefunctional analysis of the corpus data also accounts for the linguistic expression of these qualifications both from a diachronic and a synchronic point of view. The aim of this volume is to give a unified analysis of the seven verbs, in contrast with the Spanish linguistic tradition in which, as will be shown in the following sections, these verbs have been described only partially and most often as distinct phenomena.2 1. Previous accounts of parecer The verb parecer has received considerable scholarly attention, whereas resultar has hardly been discussed in the literature. Hernanz (1982: 235– 237, 1999: 2232–2233), Porroche (1990: 129–136), Fernández Leborans and Díaz Bautista (1991) and Fernández Leborans (1999: 2450–2453) have all dealt with the relation between the construction type and the meaning of parecer (8), albeit without relying on corpus research.3 To a greater or lesser extent, these authors support the idea that the “factivity” expressed in
4
Background, objectives and methodology
parecer + adjetive/infinitive (8a) is lower than that conveyed by parece que (8b). (8)
a. Las condiciones del tratado parecen irritar a los diplomáticos. ‘The conditions of the treaty seem to irritate the diplomats’ b. Parece que las condiciones del tratado irritan a los diplomáticos. ‘It seems that the conditions of the treaty irritate the diplomats.’
The Spanish literature usually does not refer to parecer as an evidential verb, but Fernández de Castro (1999: 198–199) mentions the inferential nature of parecer + infinitive.4 Only recently has the constructional potential of parecer been interpreted in clearly evidential terms.5 Bermúdez (2002) argues that parecer + infinitive (8a) conveys a speaker-oriented inferential statement, while the parece que construction in (8b) can also report someone else’s opinion. Table 1. Non-factual readings (Fernández de Castro 1999: 328, my translation)6 Modalization: How does the speaker commit him/herself to the proposition?
Structuring value
Epistemic deductive commitment
Epistemic perceptual commitment
Verbal periphrases
Semantic value
strong
tener que (haber que) + inf
external necessity
middle
deber (de) + inf
Obligation
weak
poder + inf
Possibility
parecer + inf
Resemblance
Furthermore, Fernández de Castro’s (1999: 198) proposal relates the inference from external signs with parecer to the ‘deductive’ inference triggered by epistemic poder, deber and tener que. This leads Fernández de Castro
Previous accounts of poder, deber and tener que
5
(1999: 328) to discern different ‘modal’ readings in terms of different speaker commitment, as summarized in Table 1. Fernández de Castro (1999: 328) distinguishes between an ‘epistemic deductive commitment’ conveyed by modals and an ‘epistemic perceptual commitment’ in parecer.7 The question is whether both labels ‘perceptual’ and ‘deductive’ can be associated with the label ‘epistemic’. In this volume, I will show that they belong to the two different functional categories of evidentiality and epistemic modality respectively. 2. Previous accounts of poder, deber and tener que The semantics and pragmatics of modal verbs have received a great deal of attention over the last decades, but in the field of Spanish linguistics the semantic-pragmatic description has been somewhat overshadowed by the syntactic quest for periphrastic constructions (cf. Gómez Torrego 1999). Yet, Olbertz (1998) and Fernández de Castro (1999), among others, offer a detailed semantic description of the Spanish modal verbs in addition to the classification of periphrases. Other authors, such as Silva-Corvalán (1995) and González Vázquez (2000), opt for a semantic explanation of the modal readings.8 Each of these analyses present the different degrees of speaker commitment conveyed by the three verbs as a central topic. A drawback of these analyses is that, despite the use of corpus examples, they do not pay attention to differences in frequency to describe tendencies in the use of these verbs. In recent research on evidentiality, Dendale (1994) and Nuyts (2001b) argue that the (inferential) evidentiality of devoir and must precludes these verbs from being real epistemic modifiers. By contrast, authors like De Haan (1999) contend that a modal such as must is mainly epistemic since it cannot refer to another source of information than the speaker. The evidential dimension of modal verbs is only summarily dealt with in Spanish linguistics (cf. Silva-Corvalán 1995; Olbertz 1998; Gómez Torrego 1999), even though many authors mention aspects of evidential readings. Fernández de Castro (1999: 328), for example, refers to the modal verbs in terms of “epistemic deductive commitment” for all three verbs without explaining the different degree of speaker commitment. Hence, an analysis that examines the relation between the evidential qualification and the degree of speaker commitment is needed.
6
Background, objectives and methodology
3. Previous accounts of amenazar and prometer In the Spanish linguistic literature the modal reading of amenazar and prometer has hardly received any attention.9 Fernández de Castro’s (1999: 328) table, reproduced as Table 1, does not mention these verbs. Hernanz (1999: 2234) approaches them from a syntactic point of view. She states that when prometer and amenazar are devoid of their control reading, their syntax is similar to that of parecer, but she does not take into account their evidential and epistemic modal dimension.10 Vázquez-Laslop (2001), by contrast, describes some syntactic and semantic characteristics of “epistemic” prometer, but does not deal with amenazar. In order to deal with these gaps in the literature, I have focused on the strength of the prediction and the kind of evaluation expressed by the modal readings of both amenazar and prometer (cf. Cornillie 2004). The Spanish grammatical tradition has not provided a specific terminology for prometer and amenazar. Prometer and amenazar are most often referred to as non-lexical or raising counterparts of control verbs (cf. Hernanz 1999; Demonte p.c.; Cornillie 2005c). Vázquez-Laslop (2001) uses the term “epistemic” for non-lexical prometer, just like other authors do for the English, Dutch and French equivalents of amenazar and prometer (cf. Traugott 1993, 1997; Verhagen 1995, 1996; Rooryck 1997, 2000).11 However, since these verbs rely on inference and express a prediction, they are not specifically epistemic. 4. Broad and narrow definitions of evidentiality Evidentiality is understood in both a narrow and a broad sense in the literature. According to Bybee’s (1985: 184) ‘narrow sense’ definition, “evidentials may be generally defined as markers that indicate something about the source of the information in the proposition”.12 Chafe (1986: 262), for his part, uses the term “evidentiality” in its broadest sense, i.e. without restricting it to the expression of “evidence” only. According to him, what gives coherence to the heterogeneous group of evidential expressions is that they all “involve attitudes toward knowledge”. This broad definition of evidentiality leads Chafe (1986: 262) to put evidential and epistemic modal qualifications under the same cover term “evidentiality”, including expressions of belief such as I think, I guess, I suppose; inductive expressions such as must, seem, obvious, evidently; sensory evidence like I see/hear/feel her coming, she looks like she’s asleep, he sounds like he’s mad; hearsay evidence such as have been said
Corpus research
7
to, it seems, it is supposed to be, apparently; deductive expressions like he or she should take longer to respond, Adults presumably are capable, hedges such as sort of, about; expectations like of course, oddly enough, in fact, actually, even, but, however and nevertheless. These examples illustrate that in English, just like in Spanish, evidentiality and epistemic modality are most often lexically represented.13 Many non-European languages, by contrast, have special affixes, clitics and particles that mark evidentiality and modality grammatically (see Chafe and Nichols 1986; Willett 1988; Palmer 2001; or more recently Aikhenvald and Dixon 2003; Aikhenvald 2004). However, this does not mean that evidentiality lacks a grammatical status in English or Spanish. For example, Chafe (1986) and Anderson (1986), among others, consider auxiliaries like English must as typically evidential. I will use a more narrow definition of evidentiality, viz. reference to knowledge with special attention to how it differs from epistemic modality. This narrow definition also implies special focus on grammatical markers. Although the importance of evidentiality and epistemic modality lies in its function rather than in its form, I will not address lexical evidential or modal expressions such as mental state or evidential verbs, e.g. creo que ‘I think’, esto indica que ‘this indicates that’, or evidential/modal adjectives and adverbs. I will restrict the analysis to grammaticalized evidential and modal verbs, i.e. verbs that most often combine with an infinitive. 5. Corpus research In this study, as indicated in the introduction of this chapter, I will elaborate a unified approach to the three groups of verbs just discussed as evidential and modal expressions. I will do this on a solid empirical basis, viz. the actual use of these expressions as attested in corpus data, in which their evidential and modal meanings are clearly contextualized. My corpus-based analysis of the verbs will be both quantitative and qualitative.14 Most of my data come from the on-line Corpus del español, compiled by Mark Davies,15 more specifically from the 20th century subcorpus, which contains approximately 20 million words. This subcorpus includes three types of discourse: fiction prose, oral discourse and non-fiction prose (journalistic and encyclopaedic texts). All three parts consist of more or less the same amount of words. As for the geographical distribution, the corpus contains both peninsular and Latin American Spanish.16
8
Background, objectives and methodology
With the data extractions of all seven verbs taken together, my database consisted of 13790 synchronic examples. Table 2 gives the exact number of examples for the individual verbs.17 Table 2. The database fiction prose
parecer resultar amenazar prometer poder deber tener que
320 544
864
non-fiction prose
oral discourse
Total
2315 1327 301 191 1395 535 190
856 221 120 246 3237 631 1361
3171 1548 741 981 4632 1166 1551
6254
6672
13790
The verb forms of the corpus examples were coded for register, complementation (nominal complement, que-clause, compound or simple infinitive, etc.), tense, participant structure (in the case of lexical prometer and amenazar), main clause and subordinate clause position. I also labeled the particular reading of the verbs that show up in the examples and set up a semantic classification of their subject and infinitive. The diachronic analyses of parecer, resultar, amenazar, prometer and tener que were based on the diachronic sections of the Corpus del español. These analyses were mainly qualitative in nature. That is, I did not set up a specific database to measure the frequency of the specific construction types, but I was able to manually review the most relevant verb forms thanks to the online service of the Corpus del español, which presents the different morphological forms and which, on further request, yields the instances of each verb form in groups of one hundred examples. In addition to corpus material, I also relied on native speaker judgments for my qualitative analysis of evidentiality and epistemic modality. Corpora provide information about the constructions attested for a particular verb, but native speaker intuitions can be used to decide whether the reading of a particular construction differs from that of other constructions. In order to obtain natural speaker intuitions, I gave appropriate contextualizations of the constructions. Moreover, I checked whether the native speaker was not too much influenced by a particular regional variety. Therefore I submitted
Cognitive-functional account
9
the constructions discussed in this volume to both Spanish and Latin American speakers.18 6. Cognitive-functional account Even though restricting the analysis to seven verbs (parecer, resultar, amenazar, prometer, poder, deber and tener que) is not functionally motivated, the relation between evidentiality and epistemic modality certainly is. These two qualifications accompany the transmission of information and describe the speaker’s view of this process. Communication not only has speaker and hearer encoding and decoding information as a central function of language,19 but also involves speaker-oriented processes such as modalization and subjectification of what is said, which are not less functional.20 Within the framework of Relevance Theory,21 Ifantidou (2001) offers a solid account of a broad group of “evidentials” (including discourse markers, epistemic modifiers, etc.).22 Evidentials are seen as part of the speaker’s intention to make his/her utterance relevant enough to be worth the hearer’s attention.23 Hence, the speaker “must expect at least some of the assumptions expressed and implied by her utterance to be strong enough (i.e. evidenced enough) to achieve the intended effects” (Ifantidou 2001: 196). Ifantidou (2001) reserves a double function for the evidentials: they indicate both the source of information and the speaker’s degree of commitment to or certainty about the proposition expressed. The notion of speaker commitment leads Ifantidou (2001) to differentiate between evidentials and ‘hearsay particles’, because the latter lack such a speaker commitment. I argue that such a division is not needed if one takes into account the different functions of evidential expressions and (epistemic) commitment expressions. Moreover, the two qualifications need not be intended by the speaker, since in the Relevance Theory framework, the hearer’s role is crucial in the interpretation of the utterance anyway. Ifantidou states that: when no specific linguistic guidance is given, the circumstances of an utterance will provide clues to the intended degree of commitment. If the circumstances make highly accessible a particular hypothesis about the speaker’s degree of commitment, and if that hypothesis leads us to a manifestly satisfactory interpretation, then this is the only interpretation satisfying the hearer’s expectations of relevance, and all other interpretations are disallowed. (Ifantidou 2001: 196)
10
Background, objectives and methodology
It can be assumed that the speaker him/herself does not always aim at maximal relevance of the source of information. Since the speaker can also take advantage of vagueness of the source of information, the hearer seems to play a crucial role in processing the evidentially marked proposition in terms of reliability.24 Rather than dealing with inferential speaker – hearer interaction,25 my analysis focuses on the inferential or non-inferential basis of the evidential and modal statements. That the relation between evidentiality and epistemic modality is still a highly debated issue is probably due to the fact that the tradition of western linguistics has seen them as largely overlapping concepts. American structuralists Boas and Sapir explicitly link reference to sources of information (i.e. evidentiality in the narrow sense) with reference to certainty of knowledge (i.e. epistemic modality). From the 1980s on, the focus has been increasingly on the specific nature of evidentiality (cf. Chafe and Nichols 1986; Willett 1988). Yet, new terms and new ideas often lead to a new kind of terminological confusion.26 It is therefore not surprising that Dendale and Tasmowski (2001: 340) consider the relationship between the semantic domains of evidentiality and epistemic modality as “one of the main problems in the research area of evidentiality”. The relation between evidentiality and modality can be one of disjunction, one of inclusion or one of overlap (Dendale and Tasmowski 2001: 341). There are several reasons for differentiating between evidentiality and epistemic modality (see De Haan 1999; Nuyts 2001b: 35–36, 341–343; Faller 2002: 79–120). The two notions are conceptually different in that evidentiality “deals with the evidence the speaker has for his or her statement”, while epistemic modality “evaluates the speaker’s statement and assigns it a commitment value” [emphasis is mine] (De Haan 1999: 98). There are also clear cases in which only one of them is present. For example, the evidential hearsay construction it is said that does not refer to the speaker’s epistemic assessment of reality at all. In constructions with both evidential and epistemic values, it is, thus, necessary to differentiate between the mode of knowing and the speaker’s assessment of the proposition. There might be a primary preference for one of the two readings, from which the other reading can then be derived, but there is no reason why we should “lump [them] together on an a priori basis” (Nuyts 2001b: 27). Some authors consider the relation between evidentiality and epistemic modality as one of inclusion, mainly depending on the author’s definition of what evidentials are. On the one hand, it is often mentioned that the epistemic qualification can be derived from the evidential marking. Palmer (1986: 64–67, 2001: 36–37) gives the example of the evidentials in Tuyuca, a language spoken in Colombia and Brazil. Tuyuca has a series of different
Cognitive-functional account
11
suffixes for visual, non-visual, apparent, secondhand, and assumed information. According to Palmer, these five values can be readily interpreted as values on an epistemic scale indicating decreasing probability.27 On the other hand, Plungian (2001: 354) argues that the evidential value is always inherently present in the epistemic meaning: “while an evidential supplement can always be seen in an epistemic marker, the opposite does not always hold: not all evidential markers are modal in that they do not all necessarily imply an epistemic judgment”. Plungian’s comment paves the way for acknowledging some overlap between evidentiality and epistemic modality. Van der Auwera and Plungian (1998) do not see the relation between evidentiality and epistemic modality as one of inclusion, but contend that there is a specific domain in which evidentiality and epistemic modality meet. The epistemic necessity and inferential evidentiality expressed by the modal verb must is the best known example of this overlapping domain. With must, inference from direct evidence or reasoning leads the speaker to assume that his/her assessment of reality is true. By contrast, other modal verbs such as may expressing likelihood or possibility are not considered for membership of the overlapping domain. This book endorses the idea that evidentiality and epistemic modality are functionally different but related concepts which allow for the kind of inclusion proposed by Plungian (2001). A crucial question to be addressed here is how the evidential qualification28 (if at all) can influence the speaker’s epistemic assessment of the likelihood that the proposition is true.29 In other words, is evidentiality related to the evaluation of or to the commitment to the proposition? By applying this to the evidential and modal auxiliaries under examination, the following hypothesis suggests itself: (i) the verbs parecer and resultar only relate to evidentiality and do not involve epistemic modality (pace Palmer 1986). (ii) non-lexical amenazar and prometer convey both a modal qualification with a peripheral epistemic/evidential dimension. (iii) the modal verbs poder, deber and tener que yield different epistemic evaluations which correlate with different modes of knowing. In addition to the functional dimension of evidentiality and epistemic modality introduced so far, it has to be analyzed how human cognition shapes different linguistic expressions that carry evidential and epistemic qualifications. In recent years, the influence of human conceptualization on linguistic performance has become fairly uncontroversial among cognitive linguists. It is generally assumed that there exist “strong and reliable parallels between the structural and semantic properties of verbs” and that “the syntax of verbs is a quite regular, although complex projection from their
12
Background, objectives and methodology
semantics” (Fischer et al. 1991: 331). Thus, the analysis of semi-auxiliaries is motivated by the special attention the cognitive-linguistic literature draws to the difference between non-finite and finite verbs. Givón (1993: 2) argues for an “isomorphic” or “iconic” relation between the meaning of the finite verb and the syntax of its complement clause, which he calls “one of the best examples of a correlation between form and function in grammar”.30 Applied to evidential and epistemic verbs, one can argue that the more speaker-oriented the view of the event the more extensive will be “the syntactic integration of the two propositions into a single clause” (Givón 1993: 2). In other words, speaker-oriented readings show up more easily in finite verb + infinitive constructions than in the ones with a que-clause. Moreover, the syntactic integration can be seen as the result of a cognitivelydriven grammaticalization process by which linguistic forms receive an increasingly more abstract, and more speaker-oriented, meaning (cf. Traugott 1989; Traugott and Dasher 2002; Narrog 2005a,b). These verbs most often manifest increased transparency; the fading away of their lexical properties enhances the grammaticalization and subjectification of the finite verb forms. Given that a verb + infinitive construction is more grammaticalized than a verb + que-clause, my hypothesis is that a different degree of grammaticalization corresponds to a difference in scope of the evidential or modal qualification. A synchronic cognitive-linguistic view of the difference between verb + infinitive and verb + que-clause constructions takes into account semantic nuances related to these specific constructions. The speaker is assumed to be able to choose certain linguistic expressions in contrast to other ones, since “the lexical and grammatical conventions of a language provide an array of alternative expressions for coding and conceptualization: expressions are often functionally equivalent but nonetheless different in meaning by virtue of the contrasting images they convey” (Langacker 1987: 111).31 Nuyts (2001b: 4) argues that a form-to-function approach is problematic, since not all linguistic expressions are functionally motivated: structures have their own “inherent properties and once created (for some purpose) start to live a life of their own”.32 The relation between linguistic and conceptual representation indeed is an old yet unresolved point of discussion in the field of linguistics and the philosophy of language. It can be assumed, however, that, in spite of the remaining black-box problem of the human mind (Nuyts 1992, 2001b: 11–12), linguistic representation allows the analyst to detect paths of conceptualization, in that a linguistic form encodes and transmits conceptual information (cf. Nuyts and Pederson 1997).33
Outline of the book 13
The strength of Cognitive Grammar lies in the way it accounts for meaning nuances between apparently similar linguistic expressions: the construal of a linguistic expression is a representation of an interplay of prominent and less prominent units of meaning. I subscribe to Nuyts’ criticism of the form-to-function approach when it comes to analyzing very different linguistic expressions or, more specifically, verbs that lack alternations in the complementation. But, as I will show, it is plausible to assume that the existence of different constructions of one and the same evidential or modal verb is meaningful. Such an alternation points to different communicative strategies based on the foregrounding or backgrounding of certain aspects of the construction’s meaning. Cognitive Grammar is useful to the extent that it accounts for the difference between the infinitival process and the que-clause in terms of “grounding”, i.e. the reference to “fundamental notions such as time, reality and referential identity” (Langacker 2003: 7). That an ungrounded infinitival process needs grounding by the preceding finite verb, while the grounded finite verb in the que-clause does not, will be a central element of my cognitive-functional account. 7. Outline of the book The structure of the book is organized around the three groups of verbs. In Chapter 2, I deal with the difference between evidentiality and epistemic speaker commitment with parecer and resultar, on the one hand, and the alternation between the que-clause and the infinitive with these verbs on the other hand. I argue that the concept of speaker commitment does not always hold for evidential parecer and resultar, and I show that the subjectivity of parecer + infinitive vs. the possible intersubjectivity of parece que is a satisfactory functional alternative. Chapter 3 deals with how the different evidential readings of parecer and resultar are linguistically represented. There I present the results of corpus research confirming that parecer, unlike resultar, can be considered transparent and hence, readily allows combination with an infinitive. I argue that, in addition to the transparency, the prominent anteposed subject and the ungrounded nature of the infinitival process lead to a dynamic, subjective conceptualization of the parecer + infinitive construction. I also show that parecer + infinitive involves diachronic and synchronic subjectification, which is a further refinement of the non-factive reading of this construction as proposed in the literature.
14
Background, objectives and methodology
In Chapter 4, I first point out that the literature on promise and threaten does not clearly link the inferential (evidential) dimension of these verbs and the prediction and evaluation they express. I then analyze the different speaker commitment with amenazar and prometer in terms of their different lexical-semantics and illocutionary force. This analysis is complemented with corpus data corroborating stronger prediction with prometer than with amenazar. In Chapter 5, I will give a corpus-based description of the restrictions on the subjects and the infinitives of modal amenazar and prometer. I will argue that the preferred type of infinitive points to different viewpoints on the state of affairs. In addition to the corpus data, I will also discuss a series of infinitives that do not combine with prometer and amenazar at all. The infinitives that block a modal reading most often activate the agentive patterns of prometer and amenazar. Chapter 6 is concerned with the diachronic and synchronic shift from lexical amenazar and prometer to their modal readings. The diachronic description of the different infinitives that follow modal amenazar and prometer will be shown to explain Verhagen’s (2000) hypothesis that a modal ‘threat’ emerges earlier than a modal ‘promise’. The last chapters of the book are dedicated to the evidential and epistemic qualifications expressed by poder, deber and tener que. In Chapter 7, I illustrate the inferential modal system in greater detail. The resulting overview indicates that a modal judgment can be based on different modes of knowing: (i) inferential readings based on deduction, (ii) inferential readings based on induction, (iii) non-inferential speaker-oriented readings, (iv) non-speaker-oriented, reportive readings. I show that the low speaker commitment of poder is basically related to the usual absence of inference, while the stronger commitment with deber and tener que is related to inductive and deductive inferences. In the second part of this chapter I examine on the basis of corpus data the kind of subjects and infinitives the three verbs combine with in their different readings and I analyze the relative transparency of the modal verbs. Finally, in Chapter 8, I will discuss the diachronic and synchronic subjectification of poder, deber and tener que. I show that in view of their reference-point function and their subjectification, the special modal grounding function of both epistemic and evidential modals is conceptually different from their deontic and dynamic readings. The attenuation of the deontic force further refines the traditional accounts of deber and tener que which refer to obligation.
Chapter 2 Evidentiality with parecer and resultar
In this chapter, I give a functional account of the alternation between parecer + infinitive (1a) and parece que (1b). In addition, I will also discuss the left-dislocation (1c), the parenthetical (1d) and the me/le parece que (1e) constructions. (1)
a. Las condiciones del tratado parecen irritar a los diplomáticos. ‘The conditions of the treaty seem to irritate the diplomats.’ b. Parece (ser) que las condiciones del tratado irritan a los diplomáticos. ‘It seems (to be) that the conditions of the treaty irritate the diplomats.’ c. Las condiciones del tratado parece que irritan a los diplomáticos. ‘The conditions of the treaty, it seems that they irritate the diplomats.’ d. Las condiciones del tratado, parece, irritan a los diplomáticos. ‘The conditions of the treaty, it seems, irritate the diplomats.’ e. Me parece que las condiciones del tratado los irritan. ‘It seems to me that the conditions of the treaty irritate them.’
The Spanish linguistic literature attributes a “non-factive” value to the proposition introduced by parecer + infinitive (1a), while parece que (1b) is said to be related to a “near-factive” reading (cf. Fernández Leborans 1999: 2453; Porroche 1990: 129–135). Yet, no explanation is given for these different degrees of factivity. The functional account focuses on the actual use of the expressions and is centered around the following dimensions: (source-)evidentiality, speaker commitment/reliability and (inter-) subjectivity. I will also analyze whether the analysis can be extended to the above-mentioned constructions in (1c), (1d) and (1e). At the end of the chapter, I will apply the same analysis to the resultar constructions. The chapter is organized as follows. In Section (1.), I will give a descriptive overview of the frequency distribution of the different complements that combine with parecer and resultar. In Section (2.), I will scrutinize the sources of information or modes of knowing used for an evidential statement and explain how they (do not) relate to reliability and intersubjec-
16 Evidentiality with parecer and resultar tivity. In Sections (3.) and (4.) these dimensions are analyzed in the various constructions with parecer and resultar. 1. Frequency distribution The evidential verb parecer combines with an infinitive or a que-clause, with an adjectival or a nominal complement, and sometimes also stands alone. The corpus sample helps us determine the most frequent uses of this verb in written and spoken Spanish. As shown in Table 3, it is in the clausal combinations (infinitive, que-clause) that big differences appear between written and oral discourse.34 Table 3. Distribution of parecer constructions parecer
Queme/le Infi- Adjectival Nominal None Other 36 Total 35 parece nitive attribute attribute clause que
Written %
271 11.7
42 1.8
797 34.43
562 24.28
207 8.94
218 9.42
170 7.34
2315 100
Oral %
184 21.50
208 24.30
11 1.28
208 24.30
73 8.53
79 9.23
91 10.63
856 100
More than 45% of the oral parecer examples are followed by a que-clause; more than half of these constructions have an overt conceptualizer (me ‘to me’ and le ‘to him’).37 Less than 2% of the occurrences in the spoken corpus have an infinitive, whereas this construction constitutes more than 34% of the written parecer examples. In other words, when looking at the distribution of the three clausal constructions alone, in the written language the infinitive (72%) outnumbers parece que (24.4%) and me parece que (3.6%), while the reverse is true for the spoken discourse: parecer + infinitive only represents 3% of all clausal constructions. Adjectival attributes occur in 24% of the attested spoken and written examples, while nominal and zero complements appear in about 9% of cases. The distribution of resultar differs considerably from that of parecer, as shown in Table 4.38
Frequency distribution
17
Table 4. Distribution of resultar constructions resultar
Que – clause
Written %
40 3.01
74 5.58
1026 77.32
Oral %
133 60.18
1 0.45
57 25.79
None
Other39
Total
111 8.36
6 0.45
70 5.28
1327 100
9 4.07
13 5.88
8 3.61
221 100
Infinitive Adjectival Nominal attribute attribute
The resultar construction with a que-clause is typical of oral discourse (60.18%) and it is twenty times more frequent than in written discourse. It stands for 99% of the resultar + clause constructions. In the written corpus, by contrast, the infinitival construction barely reaches the five percent level (5.58%), while resultar + que-clause is even less frequent. The infinitive and the que-clause correspond to 65.5% resp. 34.5% of the clausal types in written Spanish. As for the other constructions, in written discourse adjectival attributes are three times as frequent as in oral discourse, and represent more than three quarters of the examples (77.32%). Nominal attributes occur twice as often in written discourse as in oral discourse, yet they remain under the ten percent level (8.36%), just like the other categories. It does not suffice to account for the que-clause and the infinitive constructions in terms of the raising mechanism (cf. Hernanz 1999). The significant differences between oral and written language point to a special functional meaning of the two constructions. One can thus suppose that hearer-speaker interaction in spoken language clearly favors the use of a que-clause with parecer. The analysis in this chapter aims to clarify this functional variation in evidential terms. Moreover, the distribution of evidential parecer and resultar differs considerably. In written discourse, the infinitival construction represents more than one third of all parecer constructions, while it remains rather marginal with resultar. By contrast, in oral discourse the que-clause is significantly more frequent with resultar than with parecer. This distribution could point to different types of qualifications with parecer and resultar and different tendencies in the conceptualization of the event expressed in the clause following them.
18 Evidentiality with parecer and resultar 2. From source of information to (inter)subjectivity 2.1. Evidentiality: a typology Over the last two decades several typologies of evidentiality40 have been set up, of which I discuss the ones proposed by Chafe (1986) and Willett (1988).41 I also briefly comment on Aikhenvald (2004). As shown in Figure 1, Chafe (1986) distinguishes between several sources of knowledge – (i) no source, (ii) evidence, (iii) language and (iv) hypothesis – which he connects to four modes of knowing: (i) belief, (ii) induction, (iii) hearsay and (iv) deduction.
Figure 1. Chafe’s (1986: 263) evidential schema
According to Chafe, belief is most often expressed by mental state predicates such as I think, I guess, I suppose, while evidential verbs as must and seem or adjectives like obvious involve induction. Phrases like have been said to, is supposed to, apparently, and it seems refer to hearsay information. Finally, modal auxiliaries such as should, could or would and adverbs like presumably involve deduction. Willett’s (1988) cross-linguistic typology of evidentiality distinguishes between direct –or attested– information, on the one hand, and indirect information, on the other. The direct source of information can be visual, auditory or contain other sensory information. The indirect type of information is further subdivided into reported information (secondhand, thirdhand, folklore) and speaker-oriented inferential statements (from reasoning or results).
From source of information to (inter)subjectivity Direct Attested visual auditory other sensory
19
Indirect Reported secondhand thirdhand folklore
Inference results reasoning 42
Figure 2. Willett’s (1988) taxonomy of Sources of Information
Two of Chafe’s (1986) modes of knowing, namely induction and hearsay, correspond to the notions in bold presented in Willett’s (1988) taxonomy (attested, reported, inference): induction can be equated with attested and inferential evidence, while hearsay corresponds to reported evidence. The main difference between Chafe (1986) and Willett (1988) is the direct link the former establishes between “evidence” and “induction”. The fact that, in Chafe’s model, induction includes both direct evidence and inference implies that all types of evidence lend themselves for induction. From this point of view, direct visual or sensory evidence and inference are not as opposed as suggested by Willett (1988). Furthermore, Chafe’s attention to evidence seemingly excludes inference on the basis of language or reasoning, which, nevertheless, is an important inferential type. Hence, for induction to be a real mode of knowing, it should refer to the speaker’s processing of different sources of information. A rather different approach is the one proposed by Aikhenvald (2004). From a typological point of view, she analyzes to what extent languages combine different evidential values. Some languages combine only two types of markers (firsthand and non-firsthand, or reported vs ‘everything else’), whereas other languages combine up to five different evidential types (cf. Aikhenvald 2004: 25–36). In doing so, Aikhenvald not only restricts her analysis to traditional source-evidentiality, but also strongly condemns the use of the term “evidential” for lexical expressions that do not belong to the obligatory grammatical paradigms of a language.43 From her point of view, the lexical evidential expressions can only be considered “evidentiality strategies”. Since the analyses presented here rely on a functional definition of evidentiality, this volume does not follow this restriction. Moreover, an integrated account of both grammatical and lexical evidentiality can contribute to a better understanding of the whole domain (cf. Squartini in press).44 As for the the evidential types of Spanish parecer and resultar, my analysis takes the benefit of Chafe (1986) and Willett (1988). In line with Chafe, I use “Hearsay” to refer to indirect, reported information and use the term “Belief” for the mode of knowing for which no source of information is available, i.e. when there is no induction from evidence.
20 Evidentiality with parecer and resultar Thus, “Belief” expressions are not strictly speaking evidential: they can either convey an epistemic qualification or be a hedge with a subjectifier function. In line with Willett (1988) I take the term “Inference” to refer to the speaker’s indirect evidence with parecer, which is either based on some kind of “attested evidence” or on reasoning. Since direct evidence without inference is expressed by perception verbs in Spanish, this source of information will not be addressed here. I will test the different evidential types by means of a series of concrete contextualizations. First, I will combine a parecer or resultar statement with the comment pero yo no lo veo así ‘but I don’t see it that way’. The infelicitous combination illustrates that it is impossible to dissociate the speaker and the source of information. In this case, the parecer expression has a speaker-oriented inferential basis. Second, from Nuyts (2001b: 71– 72) I borrow the (translated) questions ¿Quién lo dice? ‘Who says that?’ and ¿Tú crees? ‘Do you think so?’, which differentiate between hearsay information and inferential information or belief. Third, more distant hearsay can be detected by means of a felicitous o eso dicen ‘at least that is what is said’ comment. 2.2. Speaker commitment and reliability When a speaker utters a qualification, he is or is not committed to it. If the speaker endorses the qualification, it is used performatively, entailing speaker commitment. In the cases without speaker commitment, we have a descriptive use. From such a point of view, commitment is directed to qualifications. In this book, speaker commitment is understood in another way, namely as commitment to the state of affairs expressed. With epistemic qualifications the degree of speaker commitment refers to the idea that the modalized proposition is likely to be or become true, while the commitment associated with evidential qualifications has to do with the speaker’s appraisal of the knowledge used. Here the ‘evidential commitment’ differs from the commitment to the state of affairs. This observation sheds new light on the literature regarding whether speaker commitment depends on the evidential types or not. Palmer (1986: 54, 2001: 52) argues that the degree of speaker commitment and the probability of the statement depend on the type of information used and Nuyts (2001b: 344) mentions that “evidentials bring the naming of sources to bear on the state of affairs in terms of degrees or types of speaker commitment”.46 Chafe (1986); Floyd (1996); Fitneva (2001); and Plungian (2001) show that the link between source-evidentiality and
From source of information to (inter)subjectivity
21
speaker commitment is not always warranted. For example, the value of reported knowledge is also associated to the authority attributed to the source of information, so that the speaker commitment can differ from one reported voice to another. Reported knowledge seen on television differs from information gathered in a bar. Floyd (1996: 935) argues that the personal commitment of the speaker attributed to reportives also depends on whether the speaker judges the relationship between the previous conceptualizer and the designated scenario to be “direct” or not. In other words, the evidential type as such does not determine the speaker commitment. All this invites us to examine whether the evidential expressions stand for different degrees of reliability rather than conveying different degrees of speaker commitment to the state of affairs. It has the advantage of being more in line with the values and judgments of knowledge that inherently play a role in the evidential qualification. It is not clear, however, whether there is a generally accepted scale of reliability. Although direct evidence can be considered more reliable than any other source, it is the question whether (subjective) inference can systematically be seen as more reliable than hearsay. Note that different linguistic expressions of the same mode of knowing or evidential type can differ as to their reliability. Chafe (1986: 263), for example, differentiates between guess and suppose, the former being less reliable than the latter. Different degrees of reliability seem thus to depend on a different processing of the knowledge involved. The speaker attributes more value to his or her own knowledge with suppose than with guess. Let us give an example from Spanish deber ‘must’ and parecer to illustrate how the concept of reliability can be understood. In (2) and (3), I differentiate between two evidential types: inferential deber ‘must’, and evidential parece ser que ‘it seems to be that’, which has either an inferential or a hearsay reading. The modal verb deber ‘must’, in (2), indicates that the speaker thinks (s)he is in a position to believe that his/her interlocutor has a lot of color words at his/her disposal. Although the deber expression in (2) is generally seen as conveying strong speaker commitment, it does not necessarily score high on the reliability scale. In the example below, deber can be attributed an intermediate degree of reliability. The state of affairs is conjectured or inferred; hence, the qualification has a psychological basis rather than an empirical one. (2)
A. …en esa misma gama de los platinados. B. -Ud. debe tener un vocabulario interesante al respecto a colores de pelos ¿Qué colores conoce? Uno es negro, otro es platinado. (Habla Culta: Havana)
22 Evidentiality with parecer and resultar ‘A. ... in the same range of the platinum blonds. B. You must have an interesting vocabulary with regard to the colors of hair. Which colors do you know? One is black, the other is platinum gold.’ The question is whether the relative reliability of the deber statement is also linked to its inferentiality. Rather, the degree of reliability depends on the speaker’s processing of the knowledge. Given the variety of reliability judgments with one and the same evidential type, it need not be linked to inferentiality. In this context, a combination of different kinds of information (impeding an exclusively speaker-oriented statement) may be a recipe for the reliability of the qualification. It can also be assumed that the speaker negotiates the degree of reliability by using a specific expression. In this respect, the lexical stem of obligation certainly contributes to higher reliability. The expression itself is then meaningful in that it guides the hearer to a certain interpretation by excluding other ones. The statement with parece ser que ‘it seems to be that’ in (3) is at first sight more attenuated one than the one with deber. (3)
¿Cómo pagan hoy los servicios secretos de todo el mundo a sus agentes? ¿Quiénes pagan mejor? - Señor Marina: Bueno, parece ser que la CIA es la que paga mejor. Yo diría que la KGB paga bastante menos.... (España Oral: CDEB033A) ‘How do the world’s secret services pay their agents? Who is paying the best? Mister Marina: Well, it seems to be that the CIA pays the best. I would say that the KGB pays far less.’
In (3), the statement says that there is (or that the speaker has) information that the CIA pays most. The evidential qualification can either involve inference or be based on hearsay, but the reliability of the qualification does not differ. With parece ser que, the speaker’s processing of personally acquired information or the processing of reported information, results in reference to an established, reliable state of affairs. Since parece ser que does not involve strictly personal conjectures, it does not convey speaker commitment, but its reliability turns out to be higher than with deber. For the two above-examined constructions, the high/low degree of speaker commitment correlates with low/high degree of reliability. The argument is that with in (2) and (3), not the inference but the speaker’s processing of evidence determines the degree of reliability. But how then should we understand the strong speaker commitment with inferential deber? The evidence as such does not lend itself for more or less commit-
From source of information to (inter)subjectivity
23
ment; it is the speaker’s subjective appraisal of the evidence that makes more or less commitment possible. In Chapter 7, I will show that it is probably due to a specific mental operation that leads to the evidential qualification. Now, the question is how the degree of speaker commitment and the reliability of the different constructions can be measured. This is not evident. For example, with a concessive clause such as aunque puedo equivocarme ‘though I may be mistaken’, which refers to the speaker’s uncertainty about his/her own perception, the clause can be seen as an attenuation of a statement that inspires reliability and/or strong commitment. Yet, it can also be an attenuation of a statement that is already weak. Moreover, it remains to be seen whether the reverse holds: does an infelicitous combination of a main clause and a concessive clause correspond to lower reliability or speaker commitment of the statement in the main clause? The infelicity could also be due to other factors. As a consequence, I will examine each example very carefully and take into account the context as much as possible. 2.3. (Inter)subjectivity The third central notion of my analysis is (inter)subjectivity. The term of intersubjectivity was first used in modern linguistics by Benveniste (1958), who regards the speaker-addressee interaction as the fundamental condition for linguistic communication. According to the French linguist, intersubjectivity refers to the fact that in communication each participant is a speaking subject who is aware of the other participant as speaking subject (Benveniste 1958: 258–260). Traugott (2003) further develops Benveniste’s (1958) concept of intersubjectivity along the line of conversational interaction between speaker and hearer. Intersubjectivity, she argues, is the “explicit expression of the speaker's attention to the “self” of the addressee in both an epistemic sense (paying attention to their presumed attitudes to the content of what is said), and in a more social sense (paying attention to their “face” or “image needs” associated with social stance and identity), whether or not there is mutual understanding” (Traugott 2003: 128). Verhagen (2005) further refines Benveniste’s line of research on intersubjectivity by paying special attention to the construal relationship between subject(s) of conceptualization and object of conceptualization.47 In Verhagen’s (2005: 6) view, the role of the different speech participants as cognizers is encompassing and pervasive: “the addressee is always engaging in cognitive coordination with
24 Evidentiality with parecer and resultar some subject of conceptualization” and “ a speaker [...] is committed to the assumption that her utterancer is in principle interpretable by someone else sharing the knowledge of certain conventions”. This volume relies on a less ambitious, i.e more applied, definition of intersubjectivity in terms of shared interest in the statement emitted. Subjectivity and intersubjectivity “emerge from an interaction between what an author does – including action intended to be perceived and designed as such and action not so intended – and an audience interpretation of all available information” (Schiffrin 1990: 142). Schiffrin’s (1990) definition of intersubjectivity goes in the direction of Nuyts’ (2001a,b) understanding of the notion which is adopted in this book. As for the (inter)subjectivity of epistemic expressions, Nuyts argues that: The alternatives within this dimension [of evidentiality] could [...] be phrased as follows: does the speaker suggest that (s)he alone knows the evidence and draws a conclusion from it?; or does (s)he indicate that the evidence is known to (or accessible by) a larger group of people who share the conclusion based on it? In the former case the speaker assumes strictly personal responsibility for the epistemic qualification, in the latter case (s)he assumes a shared responsibility for it (although (s)he remains coresponsible too, of course). (Nuyts 2001b: 34)
Nuyts’ (2001a,b) use of (inter)subjectivity moves from audience interpretation one step further to the shared responsibility of the epistemic judgments (in Dutch and German). Nuyts’ (2001a) proposal consists in separating the analysis of the speaker’s assessment of reality from the question as to whether the speaker shares the evidence,48 or assumes “the hearer’s knowledge” about the evidence used for the epistemic statement (cf. Nuyts 2001b: 37). Shared evidence – or the assumption of shared evidence – leads to an intersubjective view of the state of affairs expressed by the speaker. I subscribe to the correlation proposed by Nuyts (2001a,b) between subjectivity based on non-shared evidence and intersubjectivity based on evidence which is assumed to be shared. Since (inter)subjectivity is sometimes employed in terms of shared or non-shared access to information (cf. Bermúdez 2002), the link between the kind of evidence and the speaker’s statement should be looked at in more detail. Consider the following example: (4)
From today’s newspaper it seems that in the Iraqi town of Falluja a new Saddam is in the making.
A three-dimensional analysis of parecer
25
Although in (4) the access to the source of information, i.e. newspaper article(s), can hardly be considered restricted to the speaker, the speaker’s conclusion is not necessarily shared with other people. In other words, not all readers of the newspaper need to endorse the inferential statement. Provided that the evidence can be better specified, so as to untangle the confusion between information and evidence, the proposal of (inter)subjectivity can be appplied. Evidence is then only a small and deliberately limited part of the information available to a broader group of people, i.e. the part that involves the very basis of the evidential statement. Thus, the shared or non-shared status of the information should be differentiated from the shared (intersubjective) or non-shared (subjective) reference to the specific evidence used for the statement. 3. A three-dimensional analysis of parecer 3.1. Evidential types 3.1.1. Parecer + infinitive In (5), the speaker communicates that (s)he has the impression that people are reluctant to stand up against the power of television. The evidential reading of parecer + infinitive is based on inference from direct evidence or reasoning. (5)
... y la música pervierte en la misma forma como un vocabulario vulgar. Exactamente, es lo mismo. Enc. – Sí. Es que em... la gente parece tener miedo al enfrentar esos poderes de la televisión... (Habla Culta: Caracas: M12) ‘and the music is perverting in the same way as popular language. Exactly, it is the same. A: Yes, in fact... people seem to be reluctant to stand up to these powers of broadcasting.’
The tests in (6b) detail the kind of inference at stake; they also confirm that this construction does not yield a hearsay reading.49 For the sake of convenience, I repeat the central clause of the example before giving the contextualization.
26 Evidentiality with parecer and resultar (6)
a. ... la gente parece tener miedo al enfrentar esos poderes... b. A: * pero yo no lo veo así. ‘but I don’t see it that way.’ B: *¿Quién lo dice? ‘Who says that?’ B: ¿Tú crees? ‘Do you think so?’
The example in (6a) cannot be followed by the adversative clause pero yo no lo veo así ‘but I don’t see it that way’ (6b), by which the speaker would contradict the preceding proposition on the basis of his/her own understanding of reality. A hearsay question by the interlocutor, e.g. ¿quién lo dice? ‘who says that?’ (6b) does not felicitously correspond to the statement in (6a), either. This contrasts with the fact that the statement can easily be followed by the speaker-oriented question ¿tú crees? ‘do you think so?’. The infelicitous contextualizations of (6b) illustrate that parecer + infinitive is not based on hearsay: no secondhand evidence is involved without the speaker’s own processing. In conclusion, parecer + infinitive has an inferential reading, while it does not refer to hearsay. 3.1.2. Parece que In contrast with parecer + infinitive, the parece que construction can rely on two sources of information. Parece que (7a) can rely on hearsay or on an inference from evidence, as illustrated by the contextualization in (7b). (7)
a. A: Y ahora, después, conociendo los entretelones del asunto, ... hubo que tener una explicación del problema porque parece que el secretario general le quiso hacer favores a una dama y la eligió de secretaria y trabajaba poco en secretaría; en cambio se dedicaba a otras cosas. (Habla Culta: Santiago: M49) ‘And now, afterwards, knowing the details of the case, ...he had to have an explanation of the problem, because it seems that the secretary-general wanted to do favors to a lady and elected her as a secretary and she worked little in the secretariat; instead she devoted herself to other things.’ a. …porque parece que el secretario general le quiso hacer favores a una dama… b. A: pero yo no lo veo así. B: ¿Quién lo dice? B: ¿Tú crees?
‘but I don’t see it that way.’ ‘Who says that?’ ‘Do you think so?’
A three-dimensional analysis of parecer
27
The parece que construction in (7a) is not as strictly speaker-oriented as the infinitival one: the adversative clause pero yo no lo veo así in (7b) can readily contradict the evidential statement made by (7a). Furthermore, both the hearsay question ¿Quién lo dice? and the question ¿Tú crees? in (7b) are completely unproblematic. These questions show that the parece que construction can rely on hearsay as well as on inference. 3.1.3. Me/le parece que The me/le parece construction has a dative,50 which functions as the conceptualizer,51 i.e. it expresses the speaker’s or another participant’s belief or opinion. (8)
a. INF A. -No creo que haya tenido ningún problema especial nuestra familia... INF.B. -Claro. INF.A. -... y me parece que ese número aparece como un número digamos- muy normal. INF.B. -¡Qué curioso! Igual que mi madre; son seis también. (Habla Culta: Santiago: M40) ‘A: I don’t think that our family has ever had a special problem. B: True. A: and I think that this number looks like a, let’s say, very normal number. B: How strange! Just like my mother’s [family], they are also six.’ a’. INF A: Fue una vez sola al oculista, nada más. Y ya después no volvió a salir. Y no se acostumbra a la casa – le cuesta acostumbrarse. Muchas veces le parece que está en la otra casa, que está en Charcas. INF.B. -Extraña. INF.A. -Extraña mucho, sí.. (Habla Culta: Buenos Aires: M27 B) ‘He went to the ophthalmologist only once. And after [this time] he never went out again. And he doesn’t get used to the house, it is hard to get used. Many times he thinks that he is in another house, that he is in Charcas. B: he is homesick A: yes he is very much homesick.’ a. …me/le parece que ese número aparece como un número – digamos – muy normal…
28 Evidentiality with parecer and resultar b. A: * pero {yo/él} no lo {veo/ve} así. ‘but {I don’t/he doesn’t} see it that way.’ B: * ¿Quién lo dice? ‘Who says that?’ B: ¿Tú crees?/¿El cree? ‘{Do you/does he} think so?’ Just like parecer + infinitive, the me/le parece que construction (8a) does not combine with the adversative clause pero {yo no lo veo/él no lo ve} así (8b), i.e. refutation by the conceptualizer is not possible. This shows that the conceptualizer overtly attached to the verb parecer coincides with the source of information – whether it be the speaker in me, or a third entity referred to by le, as in (8a’). The opinion can but need not be based on reasoning, since it can also rely on direct evidence. The infelicitous question ¿quién lo dice? in (8b) shows that the statement is not congruent with hearsay. Thus, although the speaker is not necessarily the conceptualizer, the addition of a conceptualizer blocks a clear hearsay reading of the parece que construction. The mode of knowing with me/le parece que is close to the “belief” type because it goes beyond the inference from evidence. Unlike the inference, which is necessarily drawn at the moment of enunciation, the “belief” type is usually based on the conceptualizer’s general knowledge. In (9), the speaker uses me parece que for different means. The construction in the first line is inferential, i.e. the speaker has some evidence regarding María Isabel’s departure. The other constructions do not rely on specific evidence, but emphasize the speaker’s subjective version of the state of affairs. Nuyts (2004) considers them “subjectifiers” which belong to a different functional category than evidentiality. (9)
María Isabel se fue. Y... eh... y me parece que hace como – no sé un lustro ... muchísimo tiempo y son dos años pero me parece muchísimo. Y me parece que ya María Isabel debe haber cambiado que ya me va a costar mucho trabajo... por... para más la bestia no escribió ni una sola carta así que – me parece que va a ser otra persona y como yo la quiero mucho me da mucha pena. (Habla Culta: Buenos Aires: M21 B) ‘María Isabel went away... eh... and I think it was, I don’t know 5 years ago... a lot of time and two years went by, but it seems very long ago. I think that María Isabel must have changed, which will cost a special effort...moreover, the brute didn’t write me not a single letter, thus – I think that it will be another person and since I love her so much it grieves me a lot.’
A three-dimensional analysis of parecer
29
In sum, the me/le parece que construction relies on reasoning or on personally verified evidence, but can also be a mere subjectifier without evidential qualification whatsoever.52 This is in clear contrast with the parece ser que construction discussed in the following subsection. 3.1.4. Parece ser que The parece ser que construction is a special case. By means of this construction, the speaker lets the hearer know that (s)he has evidence to prove that the assertion is valid.53 According to Maldonado (p.c.) the link between ser and the que-clause is related to the nominal phrase un hecho ‘a fact’: (10)
Es un hecho que > parece ser un hecho que > parece ser que It is a fact that > it seems to be a fact that > It seems to be that
The statement with parece ser que (11) can be inferential or convey hearsay evidence. (11)
a. ¿Cómo pagan hoy los servicios secretos de todo el mundo a sus agentes? ¿Quiénes pagan mejor? - Señor Marina: Bueno, parece ser que la CIA es la que paga mejor. Yo diría que la KGB paga bastante menos.... (España Oral: CDEB033A) ‘How do the world’s secret services pay their agents? Who is paying the best? Mister Marina: Well, it seems to be that the CIA pays the best. I would say that the KGB pays far less.’ b. A: pero yo no lo veo así. B: ¿Quién lo dice? B: ¿Tú crees?
‘but I don’t see it that way.’ ‘Who says that? ‘Do you think so?’
The inferential reading of parece ser que differs from that of parece ser in that the inference is not provoked by visual or auditory evidence. The qualification then yields a certain inferred version of reality, which the speaker him/herself does not necessarily subscribe to, as shown in (12). (12)
a. El fondo es este, la cultura, los derechos de los pueblos indígenas, de los pueblos indígenas de todo el país, no solo de los de Chiapas. Parece ser que los indígenas solo sirven como escenario,
30 Evidentiality with parecer and resultar como asunto de escenografía, pues, del poder, y así no debe ser. (Entrevista (PAN), Martínez Cázares, 19-10-1998) ‘The essence is this, the culture, the rights of the indigenous peoples, of the peoples of the whole country, not only those in Chiapas. It seems to be that the indigenous only serve as a platform, as an affair of decor of power, it must be different.’ b. A: pero yo no lo veo así. B: ¿Quién lo dice? B: ¿Tú crees?
‘but I don’t see it that way.’ ‘Who says that? ‘Do you think so?’
In line with other inferentials, a parece ser que construction can combine with the speaker-oriented ¿tú crees? question in (11b) and (12b). By means of the ¿tú crees? question the interlocutor shows the desire to know whether the speaker agrees with this version of perceived reality, rather than questioning the speaker’s own statement. The speaker can also contrast the evidential statement in (11a) and (12a) with the adversative pero yo no lo veo así, e.g. (11b) and (12b). In these examples the speaker is clearly not a firsthand witness of the payments by the CIA or the political neglect of the indigenous peoples. By means of the parece ser que construction (s)he distances him/herself from his/her source of information. Finally, the interlocutor can ask the ¿quién lo dice? question, which shows that (s)he can assume that the speaker’s statement relies on hearsay information or that the inference is not his/her own opinion about the state of affairs. The readings of parece que and parece ser que display a different combination of hearsay and inference. With parece que, both hearsay information and various types of inferred knowledge can be used, whereas inferences are restricted with parece ser que. In order to avoid inference from visual or auditory evidence, the speaker can insert ser ‘to be’ in between parecer and que. This distinction is corroborated in (13), where the visual and auditory evidence is mentioned explicitly by means of solo les dejan tocar la guitarra ‘they only let them play the guitar’. (13)
a. Parece (ser) que los indígenas solo sirven como escenario. ‘It seems to be that the indigenous only serve as a platform.’ b. Parece que los indígenas solo sirven como escenario, es que solo les dejan tocar la guitarra. ‘It seems that the indigenous only serve as a platform, they only let them play the guitar.’
A three-dimensional analysis of parecer
31
c. ??Parece ser que los indígenas solo sirven como escenario, es que solo les dejan tocar la guitarra. ‘It seems to be that the indigenous only serve as a platform, they only let them play the guitar.’ The evidential statement by means of parece que can easily be followed by solo les dejan tocar la guitarra ‘they only let them play the guitar’ (13a). By contrast, this explicit reference to visual or auditory evidence does not readily combine with the parece ser que construction (13b). The inference from direct evidence as witnessed by solo les dejan tocar la guitarra ‘they only let them play the guitar’ enhances the speaker-orientedness of the expression, which turns out to be too concrete for the statement with parece ser que, except in ironic contexts. An additional test confirms that parece ser que involves a more diffuse view of the state of affairs. The contrast between the felicitous addition of the conceptualizer in (14a) and its infelicitous counterpart in (14b) shows that a distant or preferably implicit source of information is used for the parece ser que construction. (14)
a. {Según él, / le} parece que los indígenas solo sirven como escenario. ‘(According) to him, it seems that the indigenous only serve as a platform.’ b. {?? Según él, / * le} parece ser que los indígenas solo sirven como escenario. ‘(According) to him, it seems to be that the indigenous only serve as a platform.’
The addition of an overt third person conceptualizer (él or le), fully compatible with parece que, is infelicitous with parece ser que (hence the asterisk in (14b)). This shows that the latter construction tends to be in contrast with a concrete, overt source of information. The speaker avoids reference to known conceptualizers and relegates the legitimacy of the proposition to general knowledge. 3.1.5. Parece que with a left dislocated subject Parece que and parece ser que not only appear in sentence-initial position, but can also follow a left-dislocated subject, as in (15a). The question is, however, whether the specific sentence position should be considered as an
32 Evidentiality with parecer and resultar evidential construction type in its own right. As I will show, the answer is negative. (15)
a. INF.A. -Y fíjate que tiene dos alumnas casadas. INF.B. -Sí. INF.A. -Pero las hermanas casadas parece que son, por más que ella le cuente, no serán del tipo de la madre. (Habla Culta: Buenos Aires: M27 B) ‘A. Well look, he has two married sisters. B Yes. A. But the married sisters, it seems that they are, whatever she may say, they will not be the mother’s type.’ b. A: ?? pero yo no lo veo así. B: ¿Quién lo dice? B: ¿Tú crees?
‘but I don’t see it that way.’ ‘Who says that? ‘Do you think so?’
The first contextualization in (15b) by means of pero yo no lo veo así shows that the speaker cannot easily distance him/herself from the statement made in (15a), which argues for a more speaker-oriented reading than the parece que construction in sentence-initial position.54 In contrast, ¿quién lo dice? and ¿tú crees? questions, which test hearsay and inference respectively, are not impossible. This suggests that the left-dislocated construction position cannot be explained in terms of sources of information: for a statement such as the one in (15) the speaker can either rely on inference or on hearsay. The infelicitous pero yo no lo veo así is then due to the emphasis entailed by the left-dislocation of the subject or the object. The speaker subscribes to the qualification more clearly than with a sentenceinitial parece que construction. When ser is inserted into a construction with a left-dislocated subject, as in (16a), the evidential reading of parecer is more diffuse. As the contrast between the first question in (15b) and the first question in (16b) shows, the left-dislocation of the subject in (16a) easily leads to a hearsay reading. (16)
a. O sea, que tiene que pasar un proceso de selección hasta que encuentra uno - Tiene que - mira, mira a mi favor tengo lo siguiente, los sifones parece ser que es una materia que - necesita una licencia especial para fabricar porque es una - una cosa explosiva. (Habla Culta: Madrid: M14) ‘Thus, he has to go through a selection process till he finds someone. He has to – look, look positive for me is the following, traps it
A three-dimensional analysis of parecer
33
seems to be is a matter that requires a special license to be produced because it is something explosive.’ b. A: pero yo no lo veo así. B: ¿Quién lo dice? B: ¿Tú crees?
‘but I don’t see it that way.’ ‘Who says that? ‘Do you think so?’
The difference between parece que and parece ser que can be further emphasized by adding o eso dicen or ‘that is what is said’ to the evidential statement, as in (17a–b). This kind of tag stresses that the information used for the statement is likely to have an unspecified, distant status; hence, it allows us to verify whether the left-dislocated construction is compatible with such a reading. (17)
a. ?? Los sifones parece que es una materia que necesita una licencia especial, o eso dicen. ‘Traps it seems is a matter that requires a special licence, or this is at least what is said.’ b. ? Los sifones parece ser que es una materia que necesita una licencia especial, o eso dicen. ‘Traps it seems to be is a matter that requires a special licence or this is at least what is said.’
The tests in (17) show that the hearsay tag o eso dicen ‘that is what is said’ conflicts with both parece and parece ser with a left-dislocated subject. This can be explained by the fact that the left-dislocation brings the source of information closer to the speaker. The acceptability is somewhat lower for parece que than for parece ser que with a left-dislocated subject construction, however. Furthermore, in the corpus, left-dislocation patterns with first-person conceptualizers, e.g. (18a): no second- or third-person conceptualizers were found. The difference in frequency between a left-dislocation construction with me as in (18a) and the one with le in (18b) indicates that leftdislocation readily allows for more speaker-oriented evidence. (18)
a. INF.A. Yo no sé si esto sería el, el Ateneo antes, o... INF.B. -El Ateneo no. INF.A. -¿No? Yo me hago un lío. INF.B. -El Ateneo me parece que pilla más para arriba. (Habla Culta: Madrid: M16)
34 Evidentiality with parecer and resultar ‘A: I don’t know whether this would be, the Society, or... B: The Society, no, A: No? I am mixing it up, B: The Society it seems that it is more upwards.’ b. ? El Ateneo le parece que pilla más para arriba. In conclusion, a parece que construction with a left-dislocated subject yields a less distant, more speaker-oriented evidential reading than a sentence-initial parece que construction. Although hearsay is certainly plausible, left-dislocation tends to foreground the speaker’s interpretation of hearsay. 3.1.6. Parenthetical parecer The parenthetical use of parecer, as exemplified in (19), is another construction that has more to do with a specific position in the utterance than with a specific evidential type. My corpus research indicates that the most frequently encountered parenthetical in written discourse is al parecer ‘seemingly’, e.g. (19a) and (19b), which alternates with según parece ‘by all appearances’ (19c). Furthermore, me parece ‘it seems to me’ (19d) and por lo que parece ‘by all appearances’ (19e) are most often found in oral discourse. (19)
a. Cuando quiso reaccionar ya el pasillo estaba copado por hombres encapuchados y armados con revólveres y cuchillos, estos últimos fabricados, al parecer, en las propias celdas de los reclusos. Uno de ellos esgrimía con furia un bate de béisbol... (Notic: Col: Semana: 833) ‘When he wanted to react, in the corridor he was already surrounded by masked men armed with guns and knives, the last ones [were] fabricated, it seems, in the very cells of the prisoners. One of them furiously threw a baseball bat.’ b. A comienzos de este año, varios paramilitares se entregaron a las autoridades del Magdalena, muy cerca de Santa Marta. Al parecer, según fuentes consultadas por Semana, esos hombres entregaron a la Fiscalía una serie de documentos. (Notic: Col: Semana: 826) ‘At the beginning of this year, several paramilitary surrendered to the authorities of Magdalena next to Santa María. Seemingly, ac-
A three-dimensional analysis of parecer
35
cording to sources consulted by Semana, these men submitted a series of documents to the court of justice.’ c. sí, hacen precisiones que rubrica la sola firma del presidente general de la compañía. En esta misma forma es nombrado un director, a disgusto, según parece, de los Cano, ya para la fecha, accionistas más que minoritarios. (Notic: Col: Semana: 830) ‘yes, they make some precisions that carry the sole hallmark of the secretary general of the company. In the same way, he is appointed director, despite the reluctance, it seems, of the Canos, [who are] now shareholders in a more than minority position.’ d. ... escribía y la señora le dictaba al dactilógrafo. Trabajamos como seis... siete días. Eran tres días para contestar eso pero había un día San Pedro creo por ahí en julio... julio me parece... eh... yo pedí prórroga. (Habla Culta: Santiago: M24) ‘... he wrote and the lady dictated him. They worked six ... seven days. We had three days to reply to it, but it was also Saint Peter’s Day, I think there in July ... July it seems to me... eh. I requested respite.’ e. INF -Mm, ah, bueno, habrá pendientes, habrá laderas que sean boscosas, pero – en general Auvernia es un país muy pobre. ENC. -¡Ah, sí? Por lo que parece, no. (Habla Culta: Buenos Aires: M13 A) ‘INF Mm, ah, good, there will be mountain slopes, there will be hillsides that are forested, but – in general, Auvernia is a very poor country. ENC: - oh, yes? By the look of it, not [really].’ In the above examples the speaker uses the parenthetical form to stress that (s)he has evidence to claim that there is some basis for the statement uttered. However, as can be seen from the foregoing translations, the different parenthetical constructions do not rely on the same source of information. Indeed, the al parecer constructions in (19a) and (19b) are based on reported information. In (19b), the source is mentioned by means of según fuentes consultadas por Semana, ‘according to sources consulted by Semana’. Since Semana is the Colombian weekly in which the article on the paramilitary is published, the journalist refers to him/herself getting information from others: the inference with the parenthetical al parecer is thus clearly based on gathered information.
36 Evidentiality with parecer and resultar Other parentheticals do not necessarily involve secondhand information. For example, the inferential statement with según parece in (19c) can be based on visual or auditory evidence. The speaker interprets the situation surrounding the appointment of the new director and infers that the change of boss did not fit the strategy of the Canos. The other two parentheticals license inferences that do not involve other parties. The me parece parenthetical in (19d) is an expression with a subjectifier function similar to the belief mode of knowing. This is corroborated by fact that creo que ‘I think’ precedes parenthetical me parece. The parenthetical por lo que parece in (19e), for its part, has an inferential value which comes from direct evidence only. Here the speaker is involved in the visual or sensory perception and comes to a conclusion on the basis of what (s)he perceives. In contrast to examples (19b) and (19d) where the stimuli, i.e. las fuentes consultadas and creo que, are overtly mentioned, in (19e) the source of the inferential statement does not show up explicitly.55 In sum, parenthetical statements can be provoked by several types of concrete stimuli: reported information, inference from reasoning or visual/auditory evidence, and belief. This illustrates that parentheticals cannot be considered a construction type with a specific mode of knowing.56 3.1.7. Summary The parecer constructions correlate with various combinations of evidential values. The main finding is that the parecer + infinitive construction does not allow for hearsay readings, while the parece que construction can yield hearsay and inferential readings. Table 5 summarizes the relation between evidentiality and the distribution of parecer. Table 5: Evidential types with parecer Belief
Inference
Hearsay
parecer + infinitive Me/le parece que
me parece
parecer + que-clause parece ser + que-clause según parece al parecer
A three-dimensional analysis of parecer
37
In the next section, I will examine whether the different sources of information lead to different degrees of speaker commitment and reliability to the proposition. 3.2. Speaker commitment and reliability Fernández Leborans (1999: 2453) observes that parece que conveys real speaker commitment to the proposition, whereas parecer + infinitive lacks such a commitment. In previous Spanish analyses, the speaker commitment is linked to the factivity of the proposition (cf. Porroche 1990). Against the background of the evidentiality-based analysis presented in the preceding section, I will now discuss different degrees of speaker commitment and reliability. In the literature on evidentiality, it has been argued that the speaker is more committed to the proposition when (s)he emits an inferential statement than when relying on hearsay (see Palmer 1986: 54). If we apply this assumption to the parecer constructions under consideration, the parecer + infinitive construction should convey a stronger speaker commitment than parece que. In the Spanish linguistic tradition, however, it is generally assumed that the speaker is less committed to the event expressed in the infinitive than to that of the inflected que-clause. Thus, the Spanish evidential verbs seem to go against the oft-mentioned assumption that inference conveys a stronger speaker commitment than other sources of information such as hearsay. I will show that, although the various parecer constructions manifest different degrees of speaker commitment, this commitment is to be seen independently of their source of information. Moreover, the concept of reliability may prove more useful if it comes to measuring the strength of the qualification. The examples presented in the section below show that, on the one hand, inference as such does not lead to a specific, fixed type of speaker commitment, while, on the other hand, speaker commitment is not at stake in hearsay statements, since this source of information “neither automatically precludes belief nor automatically presumes doubt” (Floyd 1996: 931).57 This section is organized as follows. I first analyze parecer + infinitive and the me/le parece que construction (3.2.1.). In (3.2.2.), I deal with the parece que construction, while (3.2.3.) is dedicated to parentheticals and left-dislocation.
38 Evidentiality with parecer and resultar 3.2.1. Parecer + inf. and me/le parece que In (20a), parecer + infinitive is inferentially driven, while the subjectifier in (20b) renders the conceptualizer’s opinion based on pure knowledge or on inference. When I add the concessive clause, the statements remain intact, as shown by the contrast between (20a) and (20b). (20)
a. La gente parece tener miedo al enfrentar esos poderes de la televisión..., aunque puedo equivocarme. ‘The people seem to be reluctant to stand up to these powers of broadcasting, though I may be mistaken.’ b. Me parece que ese número aparece como un número -digamosmuy normal, aunque puedo equivocarme. ‘It seems to me that this number looks like a, let’s say, very normal number, though I may be mistaken.’
The fact that the concessive statement aunque puedo equivocarme can (further) attenuate the statement, as illustrated in (20a), points to some commitment in parecer + infinitive. Although the speaker may be committed, this commitment cannot be equated with certainty about the likelihood of the proposition. The inferential knowledge just indicates that the speaker has evidence to presuppose that the presentation of the state of affairs is reliable. Yet, the reliability is not high: the speaker processes his/her own impression and adds that it may be a wrong presentation. The me parece que statement in (20b) does not rely on evidence, but has a subjectifier function which, rather than qualifying the state of affairs, emphasizes the reliability of the proposition.58 3.2.2. Parece que The parece que construction in (21) can be based on inference from direct evidence and on hearsay. Parece que cannot readily be attenuated by the concessive clause, which is probably due to the possible involvement of more information than the speaker alone. Hence, the reliability of the statement is rather high. On the other hand, if the concessive clause is perceived as felicitous, the doubt may be directed to the transmission of the information rather than to the speaker’s inference.
A three-dimensional analysis of parecer
(21)
39
a. ... hubo que tener una explicación del problema porque parece que el secretario general le quiso hacer favores a una dama, aunque puedo equivocarme. ‘... he had to have an explanation of the problem, because it seems that the secretary-general wanted to do favors to a lady, though I may be mistaken.’
The hesitation about this construction entails two comments on speaker commitment. First, in line with the literature that links evidentiality with speaker commitment, one could say that an inferential reading of a parece que construction leads to a stronger speaker commitment than the hearsay reading of the same construction. Second, the possible hearsay reading can be considered alien to any consideration of commitment or self-confidence. Then, only the inferential reading of parece que can be investigated. These alternative possibilities reveal the specific problem regarding the speaker commitment. It would seem that only one of the two, namely the inferential value based on direct evidence, directly corresponds to some kind of speaker commitment. However, if this commitment cannot be compared with that of other sources of information, it cannot provide us with a valid account of the different constructions of parecer. Reliability is a good alternative to speaker commitment since it allows for a unified account of the reading attached to the construction. It does not make sense to differentiate between two different degrees of commitment based on their respective evidential readings. Therefore, a unity reading that is influenced by both its hearsay dimension and its inferential dimension is a plausible idea. The same observation holds for the statement parece ser que ‘it seems to be that’ in (22a). As the questionable acceptability of (22b) suggests, the strength of the commitment to the judgment is irrelevant here. The reliability of the utterance is high, probably because it is based on generally available knowledge, be it processed in terms of hearsay or as inference. (22)
a. Parece ser que los indígenas solo sirven como escenario... ‘It seems to be that the indigenous only serve as a platform.’ b. ?? Parece ser que los indígenas solo sirven como escenario, aunque puedo equivocarme. ‘It seems to be that the indigenous only serve as a platform, though I may be mistaken.’
In sum, the parece que construction cannot readily be analyzed in terms of speaker commitment based on the source of information. This is mainly
40 Evidentiality with parecer and resultar due to the possible hearsay reading, which annuls any kind of speaker evaluation. 3.2.3. Parentheticals and left-dislocation The parentheticals are a special case: they may be seen as an explicit expression of speaker commitment, since, by their very position in the sentence, they never render the core statement of the utterance. Given that a variety of expressions fit in the parenthetical position, as illustrated in (2.1.6.), it is not surprising that different types of speaker commitment are involved. Although the position of the parenthetical in the utterance may also play a role, this issue exceeds the scope of the present volume.59 As shown in (23), some parenthetical constructions conflict with the concessive subclause, while others do not. (23)
a. ... pero había un día San Pedro creo por ahí en julio... julio me parece... eh... yo pedí prórroga, aunque puedo equivocarme. ‘...but it was also Saint Peter’s Day, I think there in July ... July it seems to me.. eh. I requested respite, though I may be mistaken.’ b. ?? En esta misma forma es nombrado un director, a disgusto, según parece, de los Cano, aunque puedo equivocarme. ‘In the same way, he is appointed director, despite the reluctancy, it seems, of the Canos, though I may be mistaken.’ c. ?? ... estaba copado por hombres encapuchados y armados con revólveres y cuchillos, estos últimos fabricados, parece ser, en las propias celdas de los reclusos, aunque puedo equivocarme. ‘When he wanted to react, in the corridor he was already surrounded by masked men armed with guns and knives, the last ones [were] fabricated, it seems to be, in the very cells of the prisoners, though I may be mistaken.’
The different degrees of felicitousness in (23) suggest different degrees of speaker commitment with parentheticals, depending on the presence or absence of a conceptualizer. The subjectifier me parece, in (23a), can yield a strong commitment, while the distant según parece (23b) and parece ser (23c) do not. In the latter examples, the speaker seemingly does not consider to be responsible for the evidence, which (s)he delegates to generally known reality.
A three-dimensional analysis of parecer
41
On the other hand, it is the reference to the known reality that renders the statement more reliable. The speaker embeds the evidence in a broader, non-personal view of the state of affairs, which increases the realiability of the qualification. The construction with a left-dislocated subject in (24) is one of the few examples in which speaker commitment and reliability go hand in hand. Not the evidence, but the specific foregrounding of the speaker’s view then motivates the commitment. (24)
Pero las hermanas casadas parece que son, por más que ella le cuente, no serán del tipo de la madre, aunque puedo equivocarme. ‘But the married sisters, it seems that they are, whatever she may say, they will not be the mother’s type, though I may be mistaken.’
In sum, parentheticals allow for various degrees of speaker commitment and a parecer construction with a left-dislocated subject displays a stronger speaker commitment than the sentence-initial parece que construction. 3.2.4. Summary The idea that speaker commitment to the propositional content is directly linked to the evidential type is less straightforward than assumed in the literature (cf. Palmer 1986: 54), where the hearsay information is generally considered as involving low speaker commitment, in contrast with the strong speaker commitment of inference. Table 6. Speaker commitment with parecer Weak speaker commitment
Strong speaker commitment
Parecer + infinitive Me/le parece que Parecer + que-clause Parece ser que Left-dislocation + parece que Según parece
me parece
42 Evidentiality with parecer and resultar I have shown that the inferential statement does not prove to stand always for a strong speaker commitment. Moreover, the hearsay readings cannot automatically be considered as unreliable. It has become clear that the test with the concessive clause does not yield unequivocal results about the speaker commitment. Reliability has been proposed as an alternative, but should be elaborated more in order to account for the “non-factivity” or “near-factivity” of the propositional content introduced by parecer + infinitive and parece que, respectively (cf. Porroche 1990; Hernanz 1982). In Section (3.3.) the notion of (inter)subjectivity will allow us to underpin the intuitions about reliability. 3.3. (Inter)subjectivity A statement is subjective when it is restricted to the speaker’s realm, while a statement is called intersubjective when the speaker (assumes (s)he) shares it with other people (cf. Nuyts 2001b: 34–37). In contrast with Bermúdez (2002), who argues that parecer + infinitive is intersubjective, I show that it is parece que that can be considered intersubjective in the majority of cases. This section is organized as follows. I first discuss the difference between subjective parecer + infinitive and (inter)subjective parece que constructions (3.3.1.). Then I refute Bermúdez’ (2002) arguments (3.3.2.), before I look at the parentheticals and the left-dislocation constructions (3.3.3.). 3.3.1. Parecer + infinitive vs. parece que I argue that parecer + infinitive stands for a subjective evidential statement, while parecer + que-clause, which lacks an overt conceptualizer, can express an intersubjective statement. In (25) a person has traveled abroad, which seems to have a good impact on him/her. (25)
Decidí respirar un poco y viajar fuera del país por unas semanas. Parece que el viaje me hizo mucho bien; según ellos, tenía mejor semblante. (Notic: Perú: Caretas: 1451) ‘I decided to breathe a bit and travel abroad for a couple of weeks. It seems that the journey did good to me, according to them, I had a better face.’
A three-dimensional analysis of parecer
43
In (25), the evidential statement conveyed by parece que is shared with other people, which is corroborated by según ellos ‘according to them’. This explicit mention of the shared evidence leads me to conclude that the statement is intersubjective. In direct speaker-hearer interaction, however, parece que can also yield a subjective statement, e.g. (26): (26)
Parece que no me entiendes. (Habla Culta: Colombia: M1) ‘It seems that you don’t understand me.’
Although the statement in (26) is not likely to be shared, all speech participants have access to the evidence, which explains the reliability associated with both the subjective and the intersubjective readings of parece que. By contrast, the infinitival constructions in (27) always imply a subjective statement. (27)
a. La segunda ley es la que tendrá que decidir cómo reestructurar los pasivos de las empresas con el Estado. Es decir, la carne y cómo cortarla. Todo parece indicar que el gobierno se inclina por aceptar únicamente las deudas tributarias. (Notic: Perú Caretas: 1427) ‘The second law is that one that will have to decide how to restructure the assets and liabilities of the companies with the State. That is, the meat and how to cut it. Everything seems to indicate that the government is inclined to only accept the debt subjected to tax.’ b. En estas corrientes se basa la polémica de hoy, en un contexto donde la pedagogía parece atravesar por una crisis. (Encicl: Pedagogía) ‘In these streams is based today’s controversy, and in a context where the pedagogics seem to go through a crisis.’
In (27a), the author makes her/his point by using the expression todo parece indicar ‘everything seems to indicate that’. The interlocutors or other people are not responsible for the inference made by the speaker/writer. This is also the case of (27b). The access to the source of information may be shared with other people, but the statement that the pedagogical sciences are facing a crisis is a personal consideration about his/her understanding of pedagogical sciences. I argue that the (inter)subjective status of evidence and statement is a good descriptive alternative for the abstract notion of “near-factivity” used in previous accounts of parecer. When a statement is supported by more
44 Evidentiality with parecer and resultar people it can be seen as receiving a higher factive value and as being more reliable. By contrast, when the speaker is solely responsible for the evidential qualification with parecer its factivity is not warranted, and its reliability is therefore deemed to be low. Note that parece que sometimes balances between a subjective and an intersubjective statement, which correlates with Hernanz’ (1982) intuitions regarding the double reading of this construction.60 The replacement of (non-)factivity by (inter)subjectivity implies a revision of Bermúdez’ (2002) account. 3.3.2. Bermúdez (2002) on (inter)subjectivity In this section, I discuss four of Bermúdez’ examples in the paragraphs marked by A., B., C. and D., and refute the way he analyzes them. A. As shown in (2.2.3.), the me/le parece que construction is considerably different from the parece que construction, in that the former is a subjectivity marker that emphasizes the conceptualizer’s belief, while the latter can yield both an inferential or a hearsay reading. Bermúdez (2002) bases his subjectivity account of parece que on these two constructions without differentiating between them. This leads him to argue that a mí me parece que saben ‘it seems to me that’ in (28) is subjective, as opposed to intersubjective parecen no saber ‘they seem not to know’. (28)
… Sí, los personajes parecen no saber nada, pero al mismo yo al mismo tiempo [sic] a mí me parece que saben mucho. (from Bermúdez 2002: 27) ‘... Yes, the characters don’t seem to know anything, but at the same I at the same time to me it seems that they know a lot.’
Bermúdez (2002: 27) is of course right in attributing the a mí me parece que construction a subjective reading based on non-shared information. The dative obviously annuls the intersubjective dimension of the construction: this overt conceptualizer, be it me ‘me’, te ‘you’ or le ‘him/her’, renders impossible a statement shared between speaker and other people. Unlike what Bermúdez argues, the parecer + infinitive construction in (28) does not give any indication as to the accessibility of the evidence to other people than the speaker. Second, the statement expressed by means of parecer + infinitive can be both subjective as to the evidence, and at the same time based on generally accessible information. Therefore I can argue that the evidential statement expressed by parecer + infinitive is not shared
A three-dimensional analysis of parecer
45
in (28): it is the speaker who comments on the capacity of the subject personajes ‘characters’ and their apparent lack of knowledge. B. I now contrast the overt presence of the conceptualizer in the parece que construction (29a) with the restrictions on mentioning a specific conceptualizer in parecer + infinitive (29b). Bermúdez (2002) argues that the infelicitous conceptualizer in (29b) confirms the intersubjectivity of parecer + infinitive. (29)
a. Me parece que Laura tiene frío. (from Bermúdez 2002: 27) ‘It seems to me that Laura is cold.’ b. ?? Laura me parece tener frío. ‘Laura seems to me to be cold.’ c. * Laura le parece tener frío. ‘Laura seems to him/her to be cold.’
With regard to (29b), it is worth noting that, though rather uncommon, me parecer + infinitive is not excluded. Mentioning a second or a third person conceptualizer, however, as in (29c), would be hardly acceptable. Unlike Bermúdez, I consider the restriction to the first person conceptualizer (me) a confirmation of the subjectivity present in the infinitival construction. The unproblematic adding of an overt conceptualizer me to the parece que construction in (29a), for its part, can be understood as a procedure to block the intersubjective reading associated with such a construction. The mere fact that not one but several conceptualizers can be added to a parece que construction illustrates that the evidential statement expressed by parece que cannot only rely on several sources of information, it can also draw on support from various speech participants. By contrast, the me/le parece que construction only refers to the conceptualizer’s belief state, at the exclusion of other conceptualizers. C. The account given in B. can also be extended to the me parece ser construction in Bermúdez’ example taken from the dialogue between Euthyphro and Socrates, repeated in (30). The explicit mention of me underlines the subjective nature of the statement. (30)
Eutifrón. – A mí me parece, Socrates, que esta chanza va muy bien con lo que ambos decimos. Porque no soy yo solo el que inspiro la necesidad de que estas hipótesis merodeen alrededor de nosotros y no permanezcan fijas. Eres tú precisamente quien me pareces ser Dédalo, puesto que si de mí dependiese haría lo possible para que permaneciesen. (from Bermúdez 2002: 27)
46 Evidentiality with parecer and resultar ‘Euthyphro. – It seems to me, Socrates, that this joke agrees fairly well with what both of us are saying. Because it is not only me who observes the necessity that these hypotheses be circling around us and do not remain fixed. It is precisely you who seems to me to be Daedalus, since, if it depended on me, I would have done everything to keep them here.’ From a pragmatic point of view, the disagreement with Socrates might be an indication that this construction is unlikely to be intersubjective. D. Bermúdez’ (p.c.) examples given in (31) can be accounted for in the same line. (31)
a. ¿Sabes qué? Parece que Luis está enamorado. ‘Do you know? It seems that Luis is in love.’ b. ??¿Sabes qué? Luis parece estar enamorado. ‘Do you know? Luis seems to be in love.’
In (31a), the parece que construction stating that Luis is in love can felicitously follow the ¿sabes qué? question, while this is not the case for the infinitival one. If the speaker asks the interlocutor whether the latter already knows about the news, it is not difficult to imagine that the evidence used for the evidential statement is shared with other people. But the parece que construction in (31a) can also be a subjective statement by a speaker who has just discovered that Luis is in love. This explanation is in favor of the possibility of combining subjectivity and intersubjectivity. That parecer + infinitive is not congruent with the ¿sabes qué? question is then to be seen against the background of the intersubjective statement this question entails. In sum, in contrast with what Bermúdez claims, there are several reasons to see the parecer + infinitive constructions as expressing subjective statements, while parece que constructions can include an intersubjective dimension, i.e. can convey a shared statement. 3.3.3. Left-dislocation and parentheticals Let us now examine whether the left-dislocation construction (32) and the parentheticals (33) can also be analyzed in terms of subjectivity and intersubjectivity.
A three-dimensional analysis of parecer
(32)
47
INF.A. -Y fíjate que tiene dos alumnas casadas. INF.B. -Sí. INF.A. -Pero las hermanas casadas parece que son, por más que ella le cuente, no serán del tipo de la madre. (Habla Culta: Buenos Aires: M27 B) ‘A: Well look, he has two married students. B: Yes. A: But the married sisters, it seems, whatever she may say, they will not be the mother’s type.’
A parecer construction with a left-dislocated subject, as in (32), is oriented towards the interlocutor: the background knowledge concerning the topical subject las hermanas casadas ‘the married sisters’ is shared by the speaker and the hearer(s). By adding parece que son... serán ‘it seems that they are... will be’, the speaker focuses on his/her involvement in the evidential statement, however, without restricting it to a strictly subjective one. Again, it is clear that the left-dislocation construction cannot be considered to have a specific evidential dimension which contrasts with other constructions. The same is true for the parenthetical constructions, as in (33). (33)
a. Sí, hacen precisiones que rubrica la sola firma del presidente general de la compañía. En esta misma forma es nombrado un director, a disgusto, según parece, de los Cano, ya para la fecha, accionistas más que minoritarios. (Notic: Col: Semana: 830) ‘yes, they make some precisions that carry the sole hallmark of the secretary general of the company. In the same way, he is appointed director, despite the reluctance, it seems, of the Canos, [who are] now shareholders in a more than minority position.’ b. ... escribía y la señora le dictaba al dactilógrafo. Trabajamos como seis... siete días. Eran tres días para contestar eso pero había un día San Pedro creo por ahí en julio... julio me parece... eh... yo pedí prórroga. (Habla Culta: Santiago M 24) ‘... he wrote and the woman dictated him. They worked six ... seven days. We had three days to reply to it, but it was also Saint Peter’s Day, I think there in July ... July it seems to me.. eh. I requested respite.’
The availability of various parenthetical constructions suggests that it is not surprising that the parenthetical as such is neither subjective nor intersubjective.61 In most of the cases, however, the parenthetical stands for a subjective statement e.g. (33a) and (33b).
48 Evidentiality with parecer and resultar 3.3.4. Summary The foregoing sections have shown that the difference between the parece que and the parecer + infinitive constructions can be grasped in terms of (inter)subjectivity. In Table 7, the (inter)subjective values of the different constructions are summarized. Table 7. (Inter)subjectivity with parecer Subjective statement
Intersubjective statement
Parecer + infinitive Me/le parece que Parecer +que-clause Left-dislocation + parece que Me parece Según parece
An (inter)subjectivity account pays more attention to the different dimensions of the evidential reading of parecer and takes into account the speaker-hearer interaction. Moreover, it avoids the problematic association of source-evidentiality and epistemic commitment. 4. A three-dimensional analysis of resultar Since the literature has hardly dealt with resultar, the present analysis will explore the semantic structure of the verb. Hernanz (1982) discusses adjectival (copular) constructions, without discussing resultar + infinitive, and Combé (1981) examines resultar with a left-dislocated subject, as an equivalent of parecer. In this section, I apply the criteria regarding evidentiality, speaker commitment and (inter)subjectivity to evidential resultar. I show that resultar considerably differs from parecer as regards these three dimensions. Before we proceed, let us first recall the distribution of resultar. In contrast to parecer, resultar cannot combine with me ‘me’ (34a) and does not have a parenthetical use, as shown in (34b). As far as the infinitive in (34c) is concerned, only copular ser/estar ‘to be’, the existential quantifier haber ‘to be’ or verbs of possession such as tener ‘to have’ fit in the infinitive slot
A three-dimensional analysis of resultar
49
of resultar.62 The resulta ser que construction (34d) and the left-dislocated construction (34e) are generally felicitous with resultar. Finally, the queclause construction (34f) is the most frequent one with resultar. (34)
a. *Me resulta que las condiciones del tratado les irritan. ‘It turns out to me that the conditions of the treaty irritate them.’ b. * Las condiciones del tratado, resulta, les irritan. ‘The conditions of the treaty, it turns out, irritate them.’ c. *Las condiciones del tratado resultan irritarles. ‘The conditions of the treaty turn out to irritate them.’ d. Resulta ser que las condiciones del tratado les irritan. ‘It turns out to be that the conditions of the treaty irritate them.’ e. Las condiciones del tratado resulta que les irritan. ‘The conditions of the treaty, it turns out that they irritate them.’ f. Resulta que las condiciones del tratado les irritan. ‘It turns out that the conditions of the treaty irritate them.’
Below I will show that resultar, unlike parecer, has only one central reading. Since the distribution of resultar is more restricted than that of parecer, I will discuss the different constructions together. This allows me to directly group the information under the three following headings: evidential types (4.1.), speaker commitment/reliability (4.2.) and (inter)subjectivity (4.3.). 4.1. Evidential types The evidential reading of resultar can be based on hearsay or on inference from direct evidence. The following tests show, however, that the speaker is not involved in the conclusion expressed by resultar to the same extent as (s)he is with parecer. I first test the inferential reading of resultar. The combination of a resultar construction with an adversative stance from the part of the speaker, pero yo no lo veo así ‘but I don’t see it that way’ in (35), (36) and (37), proves to be odd. (35)
a. Hasta aquí todo correcto. Pero más allá de este dato se observa que mientras la Colonia Güell mide precisamente 44 metros de alto, esta cifra resulta ser la cuarta parte de la Sagrada Familia. (Notic: España: ABC)
50 Evidentiality with parecer and resultar ‘Up till here everything is correct. But beyond this fact one observes that while the Güell column measures precisely 44 meters, this figure turns out to be the fourth part of the Sagrada Familia.’ b. A: ?? pero yo no lo veo así. B: ? ¿Quién lo dice? B: ¿Tú crees? B: ?? o eso dicen.
‘but I don’t see it that way.’ ‘Who says that?’ ‘Do you think so?’ ‘Or that is what they say.’
By means of resultar + infinitive, the speaker compares the size of the Güell column with that of the Sagrada Familia. The statement in (35) cannot be followed by pero yo no lo veo así ‘but I don’t see it like that’, which points to a speaker-oriented reading. The acceptability of the ¿tú lo crees? question points to a clearly inferential dimension with resultar + infinitive. The hearsay question ¿quién lo dice?, though rare, is not impossible. But the hearsay cannot be detached from the speaker’s own view of reality, as shown by the infelicitous addition of o eso dicen ‘or that is what they say’. The resulta que construction in (36) is somewhat different. Although this construction can also rely on both inference and hearsay, the latter source of information comes to mind more easily than with resultar + infinitive. Therefore, the eso dicen comment is not incompatible with the statement expressed by resulta que. (36)
a. La crisis de los teatros suele explicarse con una realidad que, aunque sea palmaria, no interesa: ocurre en toda Europa; pero resulta que la europeización nos quita las buenas costumbres sin darnos otras a cambio. (Notic: España: ABC) ‘The crisis of the theatres is commonly explained by means of a reality that, albeit obvious, is not interesting: it happens all over Europe; but it turns out that the Europeanization deprives us of the good habits, without replacing them with new ones’ b. A: ?? pero yo no lo veo así. B: ¿Quién lo dice? B: ¿Tú crees? B: eso dicen.
‘but I don’t see it that way.’ ‘Who says that? ‘Do you think so?’ ‘That is what they say.’
Finally, resultar with a left-dislocated subject in (37) differs from the previous construction in that the hearsay question (¿quién lo dice?) is judged inappropriate, just like the eso dicen comment. In other words, the type of evidence appears to be more speaker-oriented.
A three-dimensional analysis of resultar
(37)
51
a. ... un huaco retrato de bastante mala factura, encima de un escritorio metálico color gris. [...] Yo, distraída como ando, pensé, “ah, es una nueva requisa en el aeropuerto, ojalá sea de la colección de mi tía Chabuquita Larco y se lo devuelvan a la pobre”. Pero no, hija, el ceramio resulta que hablaba y además, pucha, de temas de actualidad. (Notic: Perú: Caretas: 1468) ‘a badly painted archaeological object, above a grey metal table. I thought, absent-minded as I have been lately: “ah, this is a new confiscation in the airport, hopefully it is from my aunt’s collection and they will give it back to the unfortunate”. But no, my friend, [the figure on] the bottle it turns out that it was talking, moreover, Jesus, about the topics currently discussed in the media.’ b. A: ?? pero yo no lo veo así. B: ?? ¿Quién lo dice? B: ¿Tú crees? B: * eso dicen.
‘but I don’t see it that way.’ ‘Who says that? ‘Do you think so?’ ‘That is what they say.’
Left-dislocation constructions based on hearsay should not be excluded. When a football fan who was unable to attend the match hears from his friend that Barcelona was defeated, he can still tell a third friend the following: (38)
El Barça resulta que ha perdido el partido de esta tarde. ‘FC Barcelona, it turns out, has lost the match of this afernoon.’
Thus, the resultar construction with a left-dislocated subject also allows the speaker to distance him/herself from the statement.63 In sum, it is shown that the results of the pero yo no lo veo así test differ from resultar to parecer. The fact that all constructions reject this contextualization leads me to conclude that in resultar’s readings the speaker’s perception of reality cannot be seen in contrast with the general perception. Yet, it has been shown that this does not automatically imply that resultar generally stands for a subjective, speaker-oriented view of the state of affairs. Hearsay information is more easily encountered in a resulta que construction than in resultar + infinitive or in resulta que with a left-dislocated subject.
52 Evidentiality with parecer and resultar Table 8. Evidential types with resultar Inference
Hearsay
Resultar + infinitive Resultar + que-clause Left dislocation + resulta que
4.2. Speaker commitment and reliability In the previous section I have shown that the speaker him/herself does not specifically interfere in the construction. The verb resultar expresses a relatively high reliability, without therefore involving the speaker’s own commitment. In Section (3.2.), I have argued that for attenuation to make sense, the statement should be felt to be strong enough. As the different markings suggest (viz. star, one and double interrogation mark), the degree of infelicity varies according to the resultar construction used. (39)
a. ... esta cifra resulta ser la cuarta parte de la Sagrada Familia, aunque puedo equivocarme. ‘this figure turns out to be the fourth part of the Sagrada Familia, though I may be mistaken.’ b. *... resulta que la europeización nos quita las buenas costumbres, aunque puedo equivocarme. ‘it turns out that the Europeanization deprives us of the good habits, though I may be mistaken.’ c. ... el ceramio resulta que hablaba y además, pucha, de temas de actualidad, aunque puedo equivocarme. ‘[the figure on] the bottle it turns out that it was talking, moreover, jeezes, about the topics currently discussed in the media, though I may be mistaken.’
Only in the resulta que construction the specific speaker commitment does not apply, which correlates with the general opinion its represents. In (39a) and (39c), the concessive clause with resultar can either display a rather strong commitment of the speaker, or attenuate it.
A three-dimensional analysis of resultar
53
Table 9. Speaker commitment with resultar Weak speaker commitment
Strong speaker commitment
Resultar + infinitive Resulta + que-clause Left dislocation + resulta que
4.3. (Inter)subjectivity Intersubjectivity is crucial for the analysis of resultar. With parece que, it is possible to change from an intersubjective to a subjective statement. This possibility is far more restricted with resultar: the lack of overt conceptualizers in (40), on the one hand, and strong restrictions on infinitival and parenthetical constructions, on the other, have to do with the intersubjective status of the statement expressed by resultar. (40)
a. {*me/ Ø} resulta que la europeización nos quita las buenas costumbres. ‘it turns out to me that the Europeanization deprives us of the good habits.’ b. {*según él/ Ø} resulta que la europeización nos quita las buenas costumbres. ‘According to him it turns out that the Europeanization deprives us of the good habits.’
The presence of an overt conceptualizer indicates that the evidential statement can be delegated to a specific entity, be it the speaker or other participants. The fact that the resulta que construction cannot be attributed to any specific conceptualizer then indicates that this construction stands for an intersubjective evidential statement.64 I have shown for parecer that the infinitival constructions are most suitable for subjective readings without intersubjective dimension. In the same way I argue that the restrictions on the resultar + infinitive construction correlate with the inherent intersubjectivity of this verb in present-day Spanish. The few resultar + infinitive constructions form an exception: they stand for a more subjective statement, as is illustrated by the example
54 Evidentiality with parecer and resultar with resultar + infinitive preceded by creo ‘I believe/think’ found on the Internet, as in (41). (41)
Y yo noto que se avergüenza por ello (creo que resulta ser la mejor parte del relato), dice que le molesta no poder controlarlo........siempre ha podido hacerlo, por lo menos con mujeres..... ‘And I see that he is ashamed because of it (I believe this turns out to be the best part of the story), he says that he doesn’t care that he cannot control himself ... he has always been able to do so, at least with women.’
In (41), the speaker gives his/her personal view of the state of affairs expressed in the copular resultar construction. The use of the hedge creo que ‘I think’ shows that the speaker does not share the statement with other people. Hence, the combination of the mental state verb creer ‘to think/believe’ and the resultar + infinitive construction in (41) supports the attribution of subjectivity to resultar + infinitive. The fact that in the corpus and on the Internet, I have not found a single resulta que example with creo que corroborates that this construction usually expresses an intersubjective statement. Unlike parecer, sentence-initial resulta que cannot render either an intersubjective or a subjective statement, but is limited to an intersubjective one based on shared evidence. This is also the case in (42), where the inference from direct evidence can be shared with other people. (42)
... un huaco retrato de bastante mala factura, encima de un escritorio metálico color gris. [...] Yo, distraída como ando, pensé, “ah, es una nueva requisa en el aeropuerto, ojalá sea de la colección de mi tía Chabuquita Larco y se lo devuelvan a la pobre”. Pero no, hija, el ceramio resulta que hablaba y además, pucha, de temas de actualidad.. (Notic: Perú: Caretas: 1468) ‘a badly painted archeological object, above a grey metal table. I thought, absent as I am lately: “ah, this is a new confiscation in the airport, hopefully it is from my aunt’s collection and they will give it back to the unfortunate”. But no, my friend, [the figure on] the bottle it turns out that it was talking, moreover, jeezes, about the topics currently discussed in the media.’
Thus, subjectivity is not excluded from the construction with a leftdislocated subject.65 In fact, this construction occupies a position between
A three-dimensional analysis of resultar
55
the resulta que and the resultar + infinitive construction on the (inter)subjectivity scale. Table 10. (Inter)subjectivity with resultar Subjectivity
Intersubjectivity
Resultar + infinitive Resulta + que-clause Left-dislocation + resulta que
5. Conclusions In this chapter, it has become clear that the evidential dimension of the purely evidential verbs does not necessarily correlate with a certain degree of speaker commitment to the proposition. This difference has invited me to invoke the notion of reliability and to apply the criterion of (inter)subjectivity of the evidence and the statement to parecer and resultar as a refinement of the previous descriptions. The distinctions between the parecer + infinitive and the parece que constructions described by Hernanz (1982) and Porroche (1990) are now further underpinned on an evidential basis. By means of a series of functional tests I have shown that the parece que construction involves a broader array of evidential types than parecer + infinitive. The tests confirm the claim that the latter construction conveys a more speaker-oriented evidential qualification. Speaker commitment has been a difficult dimension to pin down. The fact that hearsay readings of parecer cannot be analyzed in terms of speaker commitment attenuates the oft-made assumption that inference or belief entails stronger speaker commitment than hearsay. The relation between subjectifiers such as me parece que and speaker commitment shows that epistemic-like expressions can more easily be accounted for in terms of speaker commitment than truly evidential ones. Reliability is a step in the good direction, but needs the concept of (inter)subjectivity. Since intersubjective statements are more broadly supported than subjective ones, the former are more reliable and hence, considered to be expressing “near-factivity”, while, based on the speaker’s inference only, the latter lead to a “non-factive” interpretation. The parece
56 Evidentiality with parecer and resultar que construction can yield either subjective or intersubjective statements, but the access to the evidence is always intersubjective. By contrast, parecer + infinitive conveys only subjective statements. The different distribution of the readings is a more refined representation of the difference between the “near-factivity” and “non-factivity” related to parecer. All the above mentioned characteristics have also been tested for resultar. I have argued that the distribution of resultar is restricted due to its strong intersubjectivity. The sources of information and the speaker commitment pattern differently with the infinitival construction and leftdislocation, on the one hand, and the sentence-initial resulta que-clause on the other. In the next chapter, I will show how the linguistic expressions of parecer and resultar + infinitive lend themselves for conveying subjectivity. I will especially deal with the prominence of the subject, the ungrounded infinitival process and the process of subjectification.
Chapter 3 Degrees of transparency and subjectification with parecer and resultar
In the previous chapter I have accounted for the parecer and resultar + infinitive constructions on the basis of their subjective inferential readings. In this chapter I will show that these constructions are more frequent with parecer than with resultar. The different frequency distribution will be explained against the background of the different conceptual structure. Moreover, it will be linked to the transparency and subjectification of parecer, in contrast to the lack of these with resultar. The chapter is organized in the following fashion. I first show that the transparency of parecer gives prominence to the grammatical subject, which in turn, allows for a dynamic construal of parecer’s infinitive (cf. Langacker 2000). Second, I illustrate that parecer displays more advanced subjectification than resultar. Finally, I present a diachronic and synchronic account of the subjectification of parecer and resultar. 1. Transparency In this section I apply the notion of transparency (1.1.) to the corpus data of parecer and resultar (1.2.). I will discuss the degree of transparency on the basis of the semantics of the infinitives and the finite verbs in the queclauses that follow parecer and resultar. 1.1. Definition of transparency In Cognitive Grammar the concept of transparency has become increasingly popular since the mid-nineties, but formal analyses used the term some time before (cf. Gaatone 1976, 1995; Lamiroy 1995, 1998). Langacker (2000: 342)66 defines the notion of transparency as follows: the fact that any element which can occur in the appropriate position in the subordinate clause can likewise occur in “raised” position in the main clause. The main clause itself imposes no constraints on the raised NP – its
58 Degrees of transparency and subjectification structural motivation comes from its role in the subordinate clause. (Langacker 2000: 342)
Gaatone (1976, 1995), for his part, defines transparency in terms of the free selection of the subject of the finite verb. This author offers two main criteria to verify the absence of selection restrictions: on the one hand, the grammatical subject of a transparent verb is constrained by the infinitive that follows, on the other, transparent verbs are compatible with impersonal verbs.67 The following French verbs or verb phrases are transparent according to Gaatone (1995): aller ‘to be going’, (s’)arrêter ‘to stop’, s’avérer ‘to appear’, avoir beau ‘to do in vain’, avoir failli ‘just not happen’, avoir l’air ‘to look like’, avoir manqué ‘to fail to’, cesser ‘to stop’, commencer ‘to begin’, continuer ‘to continue’, devoir ‘must’, être censé ‘to be expected’, être en passe ‘be on the verge of’, être en train ‘to be V-ing’, être en voie ‘to be V-ing’, être loin ‘not to be in a position’, être près ‘to be close’, être supposé ‘to be supposed’, être sur le point ‘to be on the verge’, finir ‘to finish’, n’en pas finir ‘not stop’, ne pas laisser ‘not let + inf’ , ne pas manquer ‘not to be absent’, menacer ‘to threaten’, paraître ‘to appear’, pouvoir ‘can/may’, se révéler ‘to prove to’, risquer ‘to risk’, ne saurait ‘to be unable to’, sembler ‘to seem’, tarder ‘to keep waiting to’, tendre ‘to tend to’, se trouver ‘to stand to’, venir ‘to come’.
Langacker’s (2000: 349) overview of transparent verbs and adjectives presented in (a), (b) and (c) contains English verbs similar to the French ones mentioned in Gaatone (1995): (a) may, will, must, can, be, have, do, be used to; (b) sure, certain, liable, (un)likely, apt, bound, destined, supposed, set, gonna, about; (c) seem, appear, tend, chance, turn out, prove, happen, promise, threaten, fail, get, begin, start, come, continue, cease, keep (on), persist, quit, stop, end up.
Since evidential verbs paraître, sembler, avoir l’air and se révéler show up in Gaatone’s list and seem, appear, turn out show up in Langacker’s list, it can be expected that parecer and resultar are also transparent to some extent. Moreover, Lamiroy (1995, 1998) argues that some of the so-called transparent verbs face restrictions on the infinitive. For example, verbs such as avoir beau ‘to do in vain’ or être supposé ‘to be supposed’ which Gaatone (1995) considers transparent cannot combine with meteorological infinitives (cf. Lamiroy 1995: 280–283).68 As far as my analysis is concerned, parecer does not seem to face any of the semantic restrictions men-
Transparency
59
tioned by Lamiroy (1995), while resultar only admits a small group of infinitives and most often combines with a que-clause. Langacker’s CG approach to transparency pays considerable attention to the prominent role of the preverbal subject of the transparent verb, which contrasts with the subject in the finite que-clause (see Section 2.). It is thus interesting to analyze whether all types of infinitive allow for a preverbal position of their subject. In my analysis, I combine Langacker’s version of transparency with Lamiroy’s (1995, 1998) proposal to examine the infinitives. Since parecer, unlike resultar, is generally taken as fully transparent, I hypothesize that parecer + infinitive combines with all kinds of verbs and that the verbs in infinitival position and those in the que-clause are semantically similar. 1.2. Transparent parecer vs. non-transparent resultar For the analysis below, I differentiate between nine verb types which combine with parecer: (i) action verbs (action), (ii) aspectual verbs (aspect), (iii) change-of-state verbs (change), (iv) communication verbs (communication), (v) copular verbs and other verbs referring to attribution (attribution), (vi) modal verbs (modal), (vii) mental activity verbs (mental), (viii) perception verbs (perception) and (ix) verbs expressing psychological reaction (psychological). The columns of Table 11 represent the different verb types, discussed in the previous section, while the rows refer to the absolute and the proportional figures of the verb types in the written and the oral corpora. Table 11. Semantic classification of the infinitives following parecer type attribut action com aspect change modal mental percep psych Tot writ 313 176 117 18 96 1 49 14 9 797 % 39.27 22.08 14.68 2.26 12.05 0.13 6.15 1.76 1.13 100 oral 5 2 0 0 1 0 0 2 1 11 % 45.45 18.18 9.09 18.18 9.09 100
Table 11 shows a clear preference for attributive predicates (written: 39.27%, oral 45.45%).69 The verbs expressing action (around 22%) and communication (14%) rank second and third, closely followed by the change-of-state verbs (12%).70 The other categories range from marginal (mental activity 6%) to very rare (the rest). Those infinitives requiring an agentive subject, such as the action or communication verbs, are slightly less frequent than non-agentive attributive and change of state verbs.
60 Degrees of transparency and subjectification Let us now examine the corpus data on the finite verbs following parece que. I have labeled the verbs found in the corpus with parece que in the same way I did for the analysis of the infinitives, so that we can compare the semantic distribution of parecer’s infinitives with the semantics of the finite verbs following parece que. Table 12. Semantic classification of the finite verb following parece que type attribut action com Aspect change modal mental percep psych writ 106 42 11 21 43 11 27 1 3 % 39.11 15.50 4.06 7.75 15.87 4.06 9.96 0.37 1.11 oral 107 27 5 8 17 5 7 3 3 % 58.15 14.67 2.72 4.35 9.24 2.72 3.80 1.63 1.63
Tot 271 100 184 100
Table 12 shows that the attributive verbs are far more frequent than the other categories (written: 39.11%, oral: 58.5%).71 The verbs expressing action (15.5% resp. 14.7%) and change-of-state verbs (15.9% resp. 9.3%) form the second and the third group. When one compares the written data in Table 11 and Table 12, one observes that the attributive verbs are most frequent in both the infinitives and the que-clause (both 39%). The other categories only differ slightly: with parece que action verbs and communication verbs are less frequent than in the infinitive, while aspectual and modal verbs are more frequent. 72 These small differences indicate that, despite the advanced transparency, parecer admits a certain variation in the que-clause and in the infinitive.73 Interestingly, verbs expressing communication or action are more subjectoriented and show up more often in the infinitival position. Thus, the verbs that appear slightly more often in the infinitive have an agentive subject, which gives the latter more prominence (see Section 2.). By contrast, the speaker-oriented modals appear more easily in the que-clause. Yet, the corpus data do not show serious restrictions on the semantic category of the verb with the infinitive nor the que-clause. Since their distribution is similar to a large extent, it can be concluded that parecer is transparent. The me/le parece que construction in (1) is different from both parecer + infinitive and parece que because of its overt reference to the conceptualizer of the event expressed in the que-clause. (1)
INF A. -No creo que haya tenido ningún problema especial nuestra familia... INF.B. -Claro. INF.A. -... y me parece que ese número aparece como un número digamos- muy normal. INF.B. -¡Qué curioso! Igual que mi madre; son seis también..
Transparency
61
(Habla Culta: Santiago: M40) ‘A: I don’t think that our family has ever had a special problem. B: True. A: and I think that this number looks like a, let’s say, very normal number. B: How strange! Just like my mother’s [family], they are also six.’ The me/le parece que construction is not part of the alternation of infinitive and que-clause constructions with parecer. As shown in Table 13, this kind of construction is much more frequent in the oral corpus than in the written one.74 Table 13. Semantic classification of the finite verb following me/le parece que type attribut action Com aspect change modal mental percep psych writ 31 0 1 0 3 2 4 0 1 % 79.49 2.56 7.69 5.13 10.26 2.56 oral 148 18 8 5 8 15 4 1 1 % 71.15 8.65 3.85 2.40 3.85 7.21 1.92 0.48 0.48
Tot 39 100 208 100
The overwhelming majority (more than 70%) of the verbs that show up with me/le parece que belong to the attributive type. The second and third group are the action verbs (8.6%) and the modal verbs (7.2%). In comparison with parecer + infinitive and parece que, the action verbs are less frequent and the modal verbs more frequent. All verb types are possible with me/le parece que; hence, in this construction too, parecer proves to be a transparent verb. Yet, quantitatively there appears to be a preferential link with attributive, modal and other qualifications. One can thus conclude that there is a shift to less actionoriented and more speaker-oriented verbs in me/le parece que constructions. Let us now compare these results with those of resultar. As shown in Table 14 and Table 15, the main difference with parecer is that the infinitival construction is restrained to attributive predicates, e.g. (2). (2)
El mensaje de Albert Camus siempre mira hacia adelante, deja la puerta abierta a la esperanza, resulta ser en el fondo una llamada, si no al optimismo, si al coraje y a la acción. (Notic: España: ABC) ‘The message of Albert Camus always looks ahead, leaves the door to hope open, as a matter of fact [it] turns out to be a call, if not for optimism, at least for courage and action.’
Table 14 gives an overview of the verbs that combine with resulta que. 75
62 Degrees of transparency and subjectification Table 14. Semantic classification of the finite verb following resulta que type attribut action Com aspect change modal mental percep psych writ 18 11 1 0 6 2 1 0 1 % 45 27.5 2.5 15 5 2.5 2.5 oral 51 39 7 2 9 5 7 4 1 % 38.35 29.32 5.26 1.50 6.77 3.76 5.26 3.01 0.75
Tot 40 100 133 100
In both corpora, attributive predicates rank highest (45% resp. 38.4%). The action verbs (27% resp. 29%), and the change-of-state verbs (15% resp. 6.8%) are the second and third most important categories. In written discourse, the frequency (45%) of attributive predicates with resulta que slightly differs from that of parecer + infinitive (39%) and parece que (39%). On the other hand, it is worth stressing that resultar combines far more often with action verbs. As shown in Table 15, the case of the infinitive with resultar is different. Table 15. Semantic classification of the infinitives following resultar type attribut action com aspect change Modal mental percep psych writ 74 0 0 0 0 0 0 0 0 % 100 oral 1 0 0 0 0 0 0 0 0 % 100 -
Tot 74 100 1 100
All 75 infinitives of resultar found in the corpus belong to the attributive type. Hence, resultar is not transparent since it cannot combine with the infinitival form of action or change-of-state verbs. Resultar itself does not have an agentive structure similar to other non-transparent verbs, however; the verb is still attached to its basic resultative, copular reading.76 In sum, parecer is transparent because (i) it combines with all verb types and (ii) the inflected and infinitival forms that combine with it display a similar semantics. This does not mean that there is no variation, however. Action and communication verbs appear more readily in a parecer + infinitive construction, while aspectual and modal verbs do not. When a conceptualizer is added, the attributive predicates are preferred. The verb resultar is not transparent, since at the present stage of the language only attributive predicates can show up in the infinitive.
Transparency and focal prominence
63
2. Transparency and focal prominence Since transparent parecer can combine with all kinds of infinitives it is also able to select a thematic subject (cf. Gutiérrez Ordóñez 1986, Porroche 1990, Cabeza Pereiro 1997). In this section I show that the role of parecer’s grammatical subject can be further accounted for in relation to the nonfinite status of the process (cf. Achard 1998, Langacker 2000). I will present a dynamic interpretation of both the thematic subject and the infinitival process. I first briefly present Langacker’s view of the “focal” subject (2.1.). Then I deal with parecer + infinitive and que-clause (2.2.) and examine the same constructions for resultar (2.3.). In the last subsection, I address the focal prominence of the left-dislocated subject (2.4.). 2.1. Focal prominence According to Langacker (2000: 327–329), the main difference between the infinitive (18a) and the that-clause (18b) constructions consists in the choice of focal participants.77 (3)
a. Don is likely to leave. (taken from Langacker 2000: 326) b. That Don will leave is likely.
The raising construction in (3a) is characterized by the prominence of Don as the most salient participant in the process expressed by the infinitive. According to Langacker, this constellation emphasizes the “participant’s role in the overall relationship”, while in (3b) it is the process as a whole that has focal prominence. Importantly, Langacker (2000: 329) suggests that Don’s volition is “critical” in the raised construction of to be likely, while it is more neuter in the that-clause. He seems to suggest that Don’s intention is more prominent in (3a) than in (3b). Invoking the volition of the subject may be valid for the to be likely constructions but cannot be generalized for verbs like seem or parecer. If volition correlates with “near-factivity” of the event communicated in the infinitive versus “non-factivity” of that in the that-clause, this interpretation of the subject’s role would go against all intuitions about seem/parecer + infinitive described in the literature, and the evidential account presented in the previous chapter. Thus, the focal prominence of the subject is not necessarily critical for the realization of the event. Instead, I argue that the thematic subject of the
64 Degrees of transparency and subjectification Spanish evidential verbs plays a special role in the speaker’s conceptualization of the infinitival process. The focal prominence or thematic role of the subject evokes possible worlds, which leads the speaker to compare the described event with other possible outcomes.78 2.2. Parecer + infinitive and parece que In this section I argue that a dynamic interpretation of the thematic subject with focal prominence can further underpin the (inter)subjectivity account given in the previous chapter. Let us look at the parecer constructions in (4).79 (4)
a. El rumbo menos conservador designado por su nuevo director, Gérard Mortier, parece haber retraído a muchos aficionados “de toda la vida”. (Notic: España: ABC Bravo Julio) ‘The less conservative turn fixed by its new director, Gerard Mortier, seems to have drawn back a lot of long-life aficionados.’ b. Parece que el rumbo menos conservador designado por su nuevo director, Gérard Mortier, ha retraído a muchos aficionados “de toda la vida”.
The subject rumbo ‘turn’ in (4a) is thematic and receives focal prominence as the most salient part of the utterance, while in (4b) the whole que-clause is the salient part of the utterance. Now, the speaker associates the characteristics of the subject el rumbo ‘the turn’ in (4a), by virtue of its thematic function, with his/her viewing of the infinitival process. Thus, the characteristics associated with the subject in (4a) play their role in the speaker’s construal of the non-finite process expressed by the infinitive. Unlike the finite verb in the que-clause in (4b), the infinitive is not “grounded” as an instantiated type of a process, i.e. is not situated with respect to the reality acknowledged by the hearer and the speaker (cf. Langacker 1991). The ungrounded status of the infinitival form enhances the possibility of subjectively elaborating the infinitival process from the point of view of the prominent subject. This subjective elaboration consists in comparing the central event of the statement with other similar processes in relation to the knowledge the speaker has about the subject. This subjective comparison leads to a less exact view of the infinitival process, which in the literature has been considered “non-factive”.
Transparency and focal prominence
65
The subject of the que-clause in (4b), by contrast, lacks focal prominence and does not invite a dynamic construal, since this subject is embedded and agrees with a grounded, i.e. tensed, process. In the que-clause, tense and mood (i.e. the ground) situate the process with regard to the speech participants (the speaker and the hearer). Although the statement is not necessarily shared, the speaker refers to a grounded process that is accessible to the speech participants. In other words, the speaker comments on a grounded process that may be ‘already there’ for the hearer. This contrasts with the dynamic conceptualization of parecer + infinitive. Yet, the question arises what to do with the Spanish evidential constructions with a covert subject as in (5)? (5)
Pareció haber escuchado la advertencia del general Lébed. (Notic: Arg: Cronista: 676_EUE) ‘He seemed to have listened to General Lebed’s warning.’
This type of construction has an implicit subject which is present in the previous discourse. The possibility of omitting the subject implies that it remains activated in the flow of discourse. Hence, there is a conceptual basis for further processing the subject. At the same time, however, the absence of an overt subject in (5) enhances the prominence of the infinitival processes haber escuchado ‘have listened’. Now, how can subjectless constructions such as the meteorological ones in (6) be dynamically construed without the presence of an element bearing focal prominence? The infinitival process in (6a), which receives more prominence, is dynamically construed without reference to elements other than the process itself. Here the use of an impersonal infinitive is possible because no concrete temporal grounding is needed for the event expressed. Yet, it is a case in point that impersonal verbs tend to favor sentence-initial parece que such as in (6b). (6)
a. Parece llover mucho últimamente. ‘It seems to be raining a lot lately.’ b. Parece que está lloviendo/llueve mucho. ‘It seems that it is raining/it rains a lot.’
I will now show that the focal prominence of the subject and the ungrounded infinitival process correlate with the notion of subjectivity presented in the previous chapter. I exemplify this by contrasting the intersubjective parece que construction in (7a) with the subjective infinitival con-
66 Degrees of transparency and subjectification struction in (7b). In this example the speaker/writer says that it seemed that the last aficionados of the opera were about to disappear. (7)
a. ¿…o ha degradado la ópera al hacerla popular? Si este género no se hace más popular, morirá. No hace demasiado tiempo parecía que iban a desaparecer los últimos aficionados y que la ópera perecería. (Notic: España: ABC) ‘... or has opera been degraded by making it more popular? If this genre does not become more popular, it will die out. Not so long ago it looked as if the last aficionados would disappear and that opera would meet its end.’ b. Los últimos aficionados parecían { ?ir a/Ø} desaparecer.
In (7a), the que-clause following parecía ‘it seemed’ is both prominent and grounded with respect to the speech participants. Overt grounding of the state of affairs expressed in the que-clause favors an intersubjective statement, since it is presented as an event that is already conceptualized. This is in line with the assumption that the embedded subject of the finite verb does not invite for a dynamic construal of the state of affairs. As shown in the previous chapter, the parecer + infinitive construction can only express a subjective statement. Since the dynamic conceptualization of the subject and the infinitival process is not driven by other participants than the speaker, it correlates with a subjective, speaker-oriented statement. Moreover, the fact that the aspectual periphrasis ir a ‘to be going to’ fits less well in the infinitive construction in (7b) corroborates the idea that the speaker’s conceptualization of the infinitive also includes the subjective construal of other speaker-oriented dimensions such as aspect and modality. In sum, the ungrounded infinitive correlates with a special role to the subject. The focal prominence of the overt subject that precedes parecer influences the conceptualization of the state of affairs as a specification of the subjective point of view. With or without preposed subject, parecer expresses a subjective view of the ungrounded infinitival process, although with a focal subject the infinitival process is less prominent. In this respect, parecer differs from resultar, as will be shown in the next section.
Transparency and focal prominence
67
2.3. Resultar + infinitive and resulta que The issue of focal prominence with resultar is related with its (inter)subjectivity. As shown in section (1.2.) the que-clause construction, in (8), has the near monopoly on the uses of resultar, while this verb only combines with a small number of attributive infinitives, as illustrated in (9a) and (9b). (8)
Las autoridades están desconcertadas con el asesinato del abogado Eduardo Umaña. Resulta que después del crimen el gobierno ofreció una recompensa de 500 millones de pesos. (Notic: Col: Semana: 837) ‘The authorities are dismayed by the assassination of the lawyer Eduardo Umaña. It turns out that after the crime the government offered a reward of 500 million pesos.’
(9)
a. El mensaje de Albert Camus siempre mira hacia adelante, deja la puerta abierta a la esperanza, resulta ser en el fondo una llamada, si no al optimismo, si al coraje y a la acción. (Notic: España: ABC) ‘The message of Albert Camus always looks ahead, leaves the door to hope open, as a matter of fact [it] turns out to be a call, if not for optimism, at least for courage and action.’ b. El mensaje de Camus resulta tener rasgos de una llamada. ‘The message of Camus turns out to have the characteristics of a call.’ c. *El mensaje de Camus resulta basarse en una llamada. ‘The message of Camus turns out to be based on a call.’
The que-clause in (8) is a fully objective proposition; its subject does not lead to a dynamic conceptualization. It is clear that this piece of information is picked up by the speaker and that no other resulting views are taken into account. The examples of resultar + infinitive in (9) tend to express subjectivity. Resultar can combine with the copula ser ‘to be’ (9a) and the attributivelike verb phrase tener rasgos de ‘to have the characteristics of’ (9b), but not with the more action-oriented basarse ‘to base oneself on’ (9c). Thus, the infinitives that combine with resultar cannot refer to actions and other evolving processes. Only attributive predicates qualify for a dynamic, i.e.
68 Degrees of transparency and subjectification subjectively elaborated, conceptualization. In other words, with resultar, the possibility of considering other resulting views is restricted to the categorization of attributes. The impossibility of a focally prominent subject with other infinitives leads me to conclude that, at this stage of Spanish, resultar blocks most infinitival processes that involve different resulting views. In sum, resultar + infinitive implies a shift to subjectively construing the ungrounded process. So far the grammatical subject of resultar can only have focal prominence when the verb combines with attributive predicates. This implies that with verbs other than attributive ones, the view over the whole state of affairs is more important than specific prominence for one of the parts of this process. 2.4. Left-dislocation Let us now look at left-dislocation in line with the above-proposed link between focal prominence and subjectivity.80 A left-dislocated subject is thematic, but the que-clause does not give rise to the same subjective construal as the infinitive, as in the examples (10) and (11). (10)
INF.A. -Y fíjate que tiene dos alumnas casadas. INF.B. -Sí. INF.A. -Pero las hermanas casadas parece que son, por más que ella le cuente, no serán del tipo de la madre. (Habla Culta: Buenos Aires: M27 B) ‘A. Well look, he has two married students. B. Yes. A. But the married sisters, it seems that they are, whatever she may say, they will not be the mother’s type.’
(11)
... un huaco retrato de bastante mala factura, encima de un escritorio metálico color gris. [...] Yo, distraída como ando, pensé, “ah, es una nueva requisa en el aeropuerto, ojalá sea de la colección de mi tía Chabuquita Larco y se lo devuelvan a la pobre”. Pero no, hija, el ceramio resulta que hablaba y además, pucha, de temas de actualidad. (Notic: Perú: Caretas: 1468) ‘a badly painted archeological object, above a grey metal table . I thought, absent as I am lately: “ah, this is a new confiscation in the airport, hopefully it is from my aunt’s collection and they will give it back to the unfortunate”. But no, my friend, [the figure on] the
Subjectification
69
bottle it turns out that it was talking, moreover, Jesus, about the topics currently discussed in the media.’ In (10) and (11), hermanas casadas ‘married sisters’ and ceramio ‘bottle’ precede a grounded que-clause, and gramatically agree with the finite verb. The main difference with an infinitival process lies in the fact that with leftdislocation the process expressed in the que-clause cannot be further subjectively elaborated because of the grounded subclause. Nevertheless, the focal prominence of the subject does not exclude a subjective, speakeroriented construal of the predicate which parecer and resultar combine with. 3. Subjectification So far I have not taken into account the dynamics of language change in progress. In this section, I will account for the subjectivity of parecer and resultar + infinitive in terms of both the diachronic evolution and the synchronic semantic shift referred to as subjectification. Subjectification is the process whereby a certain part of speech receives an increasingly subjective reading, which can be understood as a more speaker-oriented reading (cf. Traugott 1989, 1995, 2003) or as the conceptualizer’s subjective viewing of the linguistic expression (Langacker 1990, 1998, 2000, 2006). I first conduct a Traugottian diachronic analysis of the subjectification with parecer and resultar (3.1.). Then I show that the Langackerian notion of synchronic subjectification accounts for the subjective reading of the infinitival constructions in terms of transparency and lack of overt conceptualizers (3.2.). 3.1. Diachronic subjectification 3.1.1. Traugott’s (1989) proposal Traugott (1989: 31) defines subjectification as “the historical pragmaticsemantic process whereby meanings become increasingly based in the speaker’s subjective belief state, or attitude toward what is said”. Traugott argues that, for subjectification to take place, there should be some kind of (metonymic) pragmatic strengthening (through inferencing). Traugott’s version of subjectification addresses meaning change in terms of certain
70 Degrees of transparency and subjectification aspects of the linguistic expression inviting the speaker to widen his/her scope on what is said. Traugott and Dasher (2002: 97) repeat that subjectification is “associative and metonymical” to the speaker/writer’s act of communication, most especially to the speaker/writer’s attitude to the proposition. Both Traugott (1989) and Traugott and Dasher (2002) pay special attention to the speaker/writer - hearer/reader interaction in subjectification. Traugott argues that the temporal conjunction since, which originally only meant ‘after(ward)’ or ‘from that time on’ invited the inference of a more subjective, causal interpretation (cf. Traugott 1989: 34). The shift from temporal since to causal since implies a change from a concrete relation in the state of affairs to an abstract view of the state of affairs. In other words there is a change from scope over the predication to scope over the proposition. Let us now apply this type of subjectification to parecer and resultar. From the point of view of diachronic subjectification (cf. Traugott 1989), the hypothesis is that parecer and resultar have undergone a diachronic evolution from concrete readings to more abstract ones. A detailed description of parecer’s and resultar’s concrete and abstract readings is hence needed in order to verify this process. I comment on diachronic examples from the online Corpus del español (13th century – 19th century). I first present parecer (3.1.2.) and then pass on to describe resultar (3.1.3.). 3.1.2. A diachronic view of parecer + infinitive The corpus data I present below point to an evolution of parecer from a verb of physical appearance including movement to a speaker-oriented evidential auxiliary that can combine with infinitives. My data also illustrate that, while the first infinitives in the 14th century were only copular ones, the later ones include all types of verbs. Bolinger (1991 [1972]) describes parecer with examples from 13thcentury Old Spanish, but does not discuss later examples. Below I refer to Bolinger’s comments where needed, and complement his analysis with examples from the Corpus del español from the 13th to the 16th century.81 In the 13th century, parecer is often used as a verb denoting appearance or location, which can be understood both physically or metaphorically. The parecer examples below refer to the physical appearance ‘to appear, to show up’, e.g. (12a) and (12b), or the presentation of historical facts in historiography (12c). This appearance or location reading could be considered a concrete reading of parecer.
Subjectification
(12)
71
a. ... siempre se mueuen so ell orizon & non parece ninguno dellos. (Alfonso X. Libros del saber de astronomía. 13th century) ‘They always move under the horizon and no one of them shows up.’ b. De que me aduxiste en tan alto puyal. De qui toda la tierra pareçe fastal mar. (Berceo, Gonzalo de. Milagros de Nuestra Sennora. 13th century) ‘That you took me in so high a ‘place’. From here the whole world appears up till the sea.’ c. Rey de Leon el qual cercaron los moros en la cueua de Covadonga que es en Asturias de ouiedo & por quien dios mos?? muy grand ??? en aquel logar segund pareçe en esta estoria delas cosas que *acaescieron en su tiempo. (Alfonso X. Estoria de España II. 13th century) ‘The King of Leon who was surrounded by the Moors in the cave of Covadonga which is in Asturias de Oviedo and for whom God showed very big thing at this place, as it appears in this history of the things that happened at his time.’
Importantly, my 13th-century corpus data also contain nominal (13a) and adjectival (13b) attributes, highlighting the copular function of parecer. In these constructions, the writer compares the situation with a miracle (13a), states that the fruit looks red (13b) or that the moon seems bigger (13c). In the latter case, the writer indicates that the reality is different. The speaker does not express certainty about his/her perception. This implies that the subjective impression reading is also one of the “concrete” readings from the first stages of the Spanish language. (13)
a. E esto dize la ystoria porque parece milagro enla tierra que estaua yerma & seca & quemada dela calentura del sol tornar pesquera grande & muy abastada por el agua dela mar. (Anonymous. Gran conquista de Ultramar. 13th century) ‘And the history tells this because it seems a miracle to find a lake provided with water from the sea in the wild, dry region that was burnt by the hot sun.’ b. pero [esta fruta] es fuert & dura de quebrantar, quando la quebrantan; fallan la de dentro aspera & de color uerde & clara pero
72 Degrees of transparency and subjectification de fuera parece uermeia; mas no con tan grand claridat. (Alfonso X. Lapidario. 13th century) ‘But this fruit is strong and hard to break, when they break it; they break the one that is rough from within and green-colored and clear, but from outside it seems red, but not with great clarity.’ c. ... sauet que esta estrella que nos dezimos luna por esta rrazon paresçe mayor delas otras por que esta mas çerca de nosotros e dela tierra por que esta enel primero çirculo e esta planeta ha el cuerpo rredondo. (San Isidoro. Semejanza del mundo. 13th century) ‘Know that this star that we call moon for this reason seems [to be] the bigger than the other ones because it is closer to us and to the earth, because it is in the first circle and this planet has a round body.’ Bolinger (1991: 30) argues that these copular expressions are the result of a comparative parecer construction with como ‘like’, as in (14a), which he paraphrases in terms of ‘it looked like’ or ‘it looked as if’. According to Bolinger, this como-construction itself points to a widely productive adverbial combination with parecer in Old Spanish. (14)
a. et paresçia como era uermeia. ‘And it looked like it was red.’ b. assi como oy en dia paresce. ‘so like it appears today.’ (Alfonso X. Estoria de España. 13th century, taken from Bolinger 1991: 29–30)
My examples in (13) illustrate that constructions with or without como are used at the same time. Anyhow, the combination of the ‘appearance’ reading of parecer is extended by way of the ‘look like’ reading of adjectival and nominal complements toward that of a “mere linking verb” (Bolinger 1991: 31). The parecer + infinitive as used in present-day Spanish is not attested in the 13th century. Bolinger (1991: 34) mentions that of his 700 parecer examples from Old Spanish only three combine with an infinitive, which then necessarily was preceded by the preposition de, as in (15). (15)
a. ... parescçe grieue de guerrear Calatrava a aquellos que la uinien combater. (Alfonso X. Estoria de España. 13th century) ‘It seems necessary to struggle [for] Calatrava against those that come to fight.’
Subjectification
73
b. ... quiero y dexar cient moros et cient moras, ca parescrie mal de leuar moros nin mora. (Alfonso X. Estoria de España. 13th century) ‘I want to leave a hundred male and a hundred female Moors, because it would look bad to carry Moors.’ c. ... nin paresçe bien de apartar se una de tantas. (Alfonso X. Estoria general. 13th century, taken from Bolinger 1999: 34) ‘Nor does it look well to split off one from a lot of other ones.’ In these constructions, the infinitive has a subject-like function and, thus, parecer cannot be considered the “linking verb” it is nowadays. In fact, it is not clear whether the parecer + infinitive construction really originates from this construction. It stands to reason that parecer + infinitive has arisen as a variant of the que-clause construction. This construction is attested from the first stages of the Spanish language on. Bolinger (1991: 32) argues that parece que has a strong adverbial dimension: bien ‘well’ or manifiestamente ‘certainly’, as in (16a), often show up with parece que. Yet, in the Corpus del español, from the 13th century onwards I found constructions without adverbs such as me parece que (16b) and parece que (16c). (16)
a. Manifiestament[r]e paresçio que assi fue como el juido dixo. (Alfonso X. General Estoria. 13th century, taken from Bolinger 1991: 32) ‘It obviously appeared that it happened like the jew [had] said.’ b. ... a todos los altos hombres que eran ay: & dixo les assi. Amigos este combatir no me parece que nos aprouecha: ca a ellos no podemos fazer daño. (Anonymous. Gran conquista de Ultramar. 13th century) ‘To all these important people who were there: and I told them like this. Friends, this fighting does not seem to me that it serves us: because these guys we cannot hurt them.’ c. El huesped fue & mucho alegre quando entendio estas palauras [...] pues paresçe que aquell cauallero era ally quando que vos fuistes librado dela mala prision çertas dixo el Rey uos dezides verdat que el era ally. (Anonymous. Cuento de Tristán de Leonís. 14th century) ‘The host was really happy when he heard these words, well it seems that this knight was there when you were liberated from the
74 Degrees of transparency and subjectification bad prison. Right, said the King, you speak the truth that he was there.’ Bolinger (1991: 37) states that by the end of the 16th century “the personalization of parecer with infinitive had accomplished”, without further detailing his assumption.82 In what follows, I will discuss Bolinger’s hypothesis regarding the shift to the infinitive and then comment on my own 14thcentury corpus data of parecer + infinitive. Bolinger (1991: 38) suggests that the infinitive constructions with parecer made their way through the accidental formal resemblance between a sentence-initial parece que construction (17a) and a left-dislocation one (17b). (17)
a. Parece que Etna arde siempre. ‘It seems that the Etna is always glowing.’ b. Etna parece que arde siempre. (taken from Bolinger 1991: 38) ‘The Etna it seems that it is always glowing.’ c. Etna parece arder siempre. ‘The Etna seems always to be glowing.’
Bolinger states that a construction like the one in (17a) only has to displace its thing-person subject in order to accomplish the first shift toward an infinitival construction. Importantly, Bolinger (1990: 38) mentions that not all subjects move as easily to the left-dislocated position as a third person subject, as shown in the constructions with first personal yo in (18). The infelicitous left-dislocation, Bolinger contends, could account for the infelicitous combination of a first person subject with parecer + infinitive. (18)
a. Parece que yo no soy así. ‘It seems that I am not like that.’ b. * Yo parece que no soy así. ‘Me it seems that I am not like that.’ [lit.] c. ?? Yo no parezco ser así. ‘I don’t seem to be like that.’
Drawing on Bolinger’s data it is possible to argue for a narrowing of the perception from a broader, intersubjective view with parece que to a speaker-oriented, subjective view with parecer + infinitive. The fact that the speaker has difficulties to view him/herself then implies that the expression is subjective.
Subjectification
75
Yet, my own examples from the Corpus del español show that an extension of the copular construction is more plausible. I found a few examples of parecer + ser towards the end of the 14th century. In (19), the writer has recourse to parecer with the infinitive ser verde ‘to be green’ to tell that the viewed reality is sometimes false: the reeds look fresh, but in reality they are not. The infinitival construction may thus be seen as an extension of the adjectival attribute uermeia ‘red’ exemplified in (13b). (19)
... por ende la maldat le aconpaña sienpre, perdida la bondat. Al junco e al carrizo el tal es conparado, que paresçe ser verde e todo es desecado: fuera tiene frescura, con que ha engañado la vista de los omnes que lo tengan provado. (López de Ayala, Canciller Pedro. Libro Rimado de Palacio. 1380) ‘Finally the badness always accompanies, once the goodness is lost. One can compare it with reeds, which seems to be green and [but] all is dried out: from outside it has a fresh look, with which it has misled the view of the people that have tasted it.’
Interestingly, in the 15th century, verbs other than ser start to combine with parecer and display a greater lexical variety from then on. In (20a), parecer combines with the infinitive mostrar ‘to demonstrate’, deleitarse ‘to enjoy’ (20b), convenir ‘to agree’ (20c) and llegar ‘to arrive/reach’ (20d), among other verbs. (20)
a. E avn podemos esto considerar acatando el vocablo que lo paresce mostrar / ca dezimos piedad commo si dixiesemos patriedad que es por respecto alos padres & patria. (Cartagena, Alfonso de. Oracional de Fernán Pérez de Guzmán. 1487-03-26) And we even can consider it respecting the word that seems to demonstrate it, because we say devotion as if we said patriotism, which is [the equivalent] of respect for the fathers and the fatherland.’ b. ... y aquellos que en tus largos vicios estan lançados havnque paresce deleytarse no es possible que la anima en vicios corronpida puede alegre biuir. (Flores, Juan de. Triunfo de amor. 15th century) ‘... and those [of you] who are chucked in your lavish vices, although you seem to enjoy it, it is impossible that a soul can live joyfully when it is corrupted in perversion.’
76 Degrees of transparency and subjectification c. ... habent insidiatrix sanguini. i. quien asecha para matar & parece conuenir mas con la letra que se sigue. (Fernández de Santaella, Rodrigo. Vocabulario eclesiástico. 15th century) ‘..habent insidiatrix sanguini.. i. who sets a pitfall to kill & seems to agree more with the letter that he follows.’ d. Donde despues del pecado del primero padre ninguno onbre pudo llegar por que es cercado del biuo fuego que sube tan alto que parece llegar cerca del cielo. (Valera, Diego de. Crónica de España. 1482) ‘Where after the sin of the first father no one could arrive because it is surrounded by heavy fire that goes so high that it seems to reach the sky.’ In the 16th century, examples are found of parecer with estar + gerund (21). From the 16th century on, the diversity of infinitives that combine with parecer steadily became larger, which has made parecer an increasingly more transparent verb. (21)
De la otra parte acá del hondo valle, el río por su anchura más s'estiende. De la una parte d'él verán alisos, que de lexos paresce estar cayendo sobre una tabla d'agua tan hermosa, tan clara, que la sombra y arboleda. (Montemayor, Jorge de. Obra selecta. 1540) ‘From that side there of the deep valley, the river widens. From this side you will see alders, that from far seem to be falling over a water table, which is as beautiful, as clear as the shadow and the trees.’
The foregoing examples show that the ‘complement’ of parecer undergoes an extension from adjectival or nominal attributes toward infinitives. The parecer reading of appearance seems to have disappeared and is nowadays expressed by aparecer ‘to show up’. Subjectification is at stake since parecer is “increasingly based in the speaker’s subjective belief state” (Traugott 1989: 31). The evolution presented above does not show, however, that the meaning of parecer radically changes: there is only a narrowing down of meanings that were already there from the beginning. My corpus data indicate that parecer does not undergo major metonymic changes: the only possible one is the shift from a physical appearance reading to a subjective impression reading. Traugott’s subjectification applies to parecer to the extent that one of the initial meanings is now spread over different linguistic expressions with parecer. It is observed that the verb becomes more
Subjectification
77
transparent without acquiring a new meaning. In present-day Spanish, parecer broadens the scope over the predication in that it can also include various types of infinitival processes. Needless to say that further research should refine the above-described tendencies. 3.1.3. A diachronic view of resultar + infinitive Resultar has a resultative meaning which hardly differs from its origins, which are situated in the 15th century, as shown in example (22). (22)
El escrivano diz: “Señor, diez ducados resultó de esta armada en que yo contribuí la ochava parte; mandadme dar la ochava parte de estos diez docados”. (Colón, Cristobal. Textos y documentos completos de Cristobal Colón. 15th century) The clerk says: “Sir, ten ducates was the result of this war fleet to which I contributed the eighth part; make me receive the eighth part of these ten ducates”.’
It is in the 17th century when one observes examples of resultar with estar ‘to be’ and with ser ‘to be’.83 In the utterance in (23), resultar expresses the result of a lawsuit against Bernardo, and less the speaker’s observation about the state of affairs. (23)
BERNARDO: Presto venís a leer mi sentencia. SECRETARIO: Estos son cargos. Cargos Primeramente resulta estar culpado don Bernardo de Cabrera en no haber agradecido a su majestad el haberle hecho Conde de Val, Almirante de la Mar. (Amescua, Antonio. La adversa fortuna de Don Bernardo de Cabrera. 15741644) ‘Bernardo: I ask you, come and read my verdict. Secretary: These are the allegations. First Don Bernardo de Cabrera turns out to be accused of not having thanked his majesty for having awarded him Count of Val and Sea admiral.’
In (24), the concrete result is less central than the general assessment of the state of affairs as viewed by the speaker and probably also by other people. (24)
... acto positivo de nobleza el no sujetarse a leyes tan santas, ordenadas con acuerdo del más prudente, más docto y más grave sena-
78 Degrees of transparency and subjectification do del mundo: de que resulta ser menor el fruto que de ellas se consigue, que el daño de habituarse el pueblo a la transgresión de leyes justas. (Fernández Navarrete, Pedro. Conservación de Monarquías y Discursos Políticos. 1626). ‘... a positive act of nobility [is] not respecting the laws so holy, and commanded with the agreement of the most cautious, most learned and most serious senate of the world: of which the benefit that they get from it turns out to be smaller than the damage of the people getting accustomed not to respect just laws.’ This is also the case in the 19th century example in (25), which yields a future reading. (25)
Si el indianete ese resulta ser lo que aparenta, y, andando los días, te apunta deseos de casarse contigo, por mí no lo dejes. (Pereda, José María de. La Puchera. 1870) If this Indian boy turns out to be what he makes believe, and, after a few days, lets you know desires to marry with you, for me don’t refuse.’
These examples suggest that resultar experiences a slow evolution toward a more speaker-oriented reading, which only breaks through in the 19th century, though it was already initiated in the 15th century. 3.1.4. Discussion The admittedly small set of data presented in the previous sections suggests that the subjectification of parecer and resultar is a century-long process. However, the case of parecer shows that speaker-oriented readings may have emerged pretty early in association with adjectival attributes. The extension to all kinds of infinitives follows in the 15th century, apparently without sensible meaning change. Further research should spell out whether parecer + infinitive also shifts from an evidential reading to a more epistemic reading, in line with what Traugott (1989) points out for the modal verbs. Resultar’s evolution towards more transparency is a more recent development than that of parecer. Resultar + infinitive testifies to a semantic change from the resultative reading to a more speaker-oriented reading, which only appears in the 19th century.. Bolinger’s (1991) hypothesis about the left-dislocation may be problematic, although it would further underpin what I have been arguing in the
Subjectification
79
present and the previous chapter. The left-dislocation construction as intermediate stage between parece que and parecer + infinitive can explain how the speaker-oriented reading acquired an infinitive construction. The fact that, unlike present-day Spanish, at the beginning only copular infinitives combined with parecer illustrates that this speaker-oriented reading based on inference was initially rather reduced. Given the great diversity of infinitives found in the present-day corpus, one can conclude that the subjective reading of parecer + infinitive has become conventionalized over the centuries. 3.2. Synchronic subjectification In this section, I examine whether Langacker’s synchronic version of subjectification (3.2.1.) can be applied it to parecer (3.2.2.) and resultar (3.2.3.). 3.2.1. Objective construal vs. subjective construal Langacker’s version of subjectification is commonly understood as the shift towards a more subjective construal, which is driven by “cognitive operations inherent in its conception” (Langacker 2003: 5). One of these operations is “mental scanning”, i.e. the viewing of a scene, which occurs along a continuum with two extremes: the maximally objective construal, when the linguistic expression is the object of conception – all participants being “onstage” –, and the maximally subjective construal, when the construction itself is the subject of conception – the participants remaining “offstage”. In other words, an element of the linguistic expression that is not overtly present, but covertly construed, is more subjective. For example, Langacker (1990: 23) regards be going to in (26) as subjective because the earthquake is not directly controlling the event by means of be going, but rather the offstage speaker is foretelling the event’s occurrence. (26)
An earthquake is going to destroy that town.
That aspectual verbs such as go to are subjective touches the heart of Langacker’s model. The mental scanning of the go to-construction uses the inherent motion pattern along temporal lines. Since the switch from motion to temporality or ‘fictive motion’ is subjectively construed, it is a manifestation of subjectification. Thus, it is clear that Langacker does not attempt
80 Degrees of transparency and subjectification to consider subjectification in terms of pragmatic speaker-hearer interaction and does not subscribe to pragmatic strengthening as a process of semantic change.84 More recently, Langacker has shifted towards a new version of subjectification that is closer to Traugott’s view of subjectification as metonymic change. “Subjectification is the ‘laying bare’ of conceptual operations which are immanent in the original lexical meanings and in that sense constitute their deepest properties” (Langacker 1998: 88). Since parecer and resultar + infinitive do not substantially differ from their lexical meanings, I will continue to use the definition of subjectification in terms of absence of the “onstage” conceptualizers. I first discuss parecer (3.2.2.) and then comment on resultar (3.2.3.). 3.2.2. Conceptualizers with parecer The onstage, i.e. overtly mentioned, conceptualizer is most often the speaker, but can also be the hearer or a third entity responsible for the viewing of the process. For example, the verb seem can have a conceptualizer like me in (27a), while evidential promise and threaten cannot. (27)
a. There {seems/appears} (to me) to be a problem. (from Langacker 2000: 256) b. This promises {*to me/ Ø } to be a good argument. c. This threatens {*to me/ Ø } to be a bad argument.
In (27a) and (27b), the conceptualizer necessarily remains offstage and unprofiled, which implies that the conceptualizer is construed subjectively. According to Langacker, the absence of such a participant implies that all attention is directed to the infinitival process, which is modally or aspectually qualified. The overt prepositional object to me situates the seem construction objectively in the speaker’s realm. In Langacker’s (2000: 256) terms, the indirect object me in (27a) is an “onstage conceptualizer”, i.e. an explicit expression of the viewer. From a conceptualist point of view, the overt expression of to me makes the construal more objective, and prevents him/her from subjectively construing the conceptualizer of the infinitival process. Most often the viewer is only potential, and remains “offstage”. In this respect, the subjectification consists in the subjective construal of an “offstage” conceptualizer that plays a role in the viewing of the linguistic expression and, hence, of the state of affairs expressed.85 Note that, whereas
Subjectification
81
to me is possible, to him or to her is excluded. This illustrates that there are restrictions on the type of participant that can be onstage. I now apply these theoretical considerations to parecer and resultar. The subjective import of evidential verbs like parecer or resultar can be analyzed in terms of conceptualizers that remain “offstage”. In Langacker’s example (27) the conceptualizer precedes the copular infinitive be, but other more lexically diverse infinitives are also possible. In Spanish, by contrast, the “onstage” conceptualizer is only available in a construction like the one with a copular infinitive in (28a), which is the only example I found in the corpus. With other infinitives, parecer + infinitive cannot be accompanied by an “onstage” conceptualizer, as shown in (28b).85 (28)
a. ... en tercer lugar quisiera indicarle lo que me parece ser la... problemática más decisiva en la cual la iglesia esta envuelta en estos momentos. (Habla Culta: San Juan: M12) ‘Thirdly, I would like to point out what seems to me to be the most decisive problems that the Church is facing at the moment.’ b. *... quisiera indicarle lo que me parece poner trabas al desarrollo económico. ‘I would like to point out what seems to me to block the economic development.’
Lexically more elaborate infinitives such as the action verb poner trabas ‘to block’ cannot combine with an “onstage” conceptualizer as easily as the copular verbs. In Langackerian terms, the absence of an “onstage” conceptualizer and the greater variety of infinitives that combine with parecer point to increased subjectification. Langacker (2000, 2003) repeatedly mentions that the conceptualizer refers to the speaker or the hearer involved in the communication. In (29a), I show that it sometimes turns out to be difficult to introduce a second person conceptualizer in the parecer + ser construction by means of a clitic form. And (29b) testifies to the sometimes questionable acceptability of the third person clitic. (29)
a. ? ¿La problemática te parece ser decisiva? ‘Do the problems seem to be decisive to you?’ b. ? Las condiciones del tratado le parecen ser duras. ‘The conditions of the treaty seem to him/her to be hard.’
In (29), it is illustrated that with parece ser, the onstage conceptualizer does not naturally appear in the third person, nor in the second person. Yet there
82 Degrees of transparency and subjectification are special cases in which these constructions can be used: for example, the emphatic reaction of disbelief or surprise in the case of (29a) or the “echo” comments made by a journalist for (29b).86 These pragmatically marked uses corroborate that when the onstage conceptualizer appears, it is predominantly centered on the speaker. Moreover, they show that the infinitive construction is inherently speaker-oriented, whether the speaker be onstage or offstage. When the conceptualizer is offstage, the speaker construes the scene (and its own intervention) more subjectively.87 3.2.3. Conceptualizers with resultar In line with the differences discussed so far, parecer and resultar differ as to the position of “onstage” conceptualizers. When resultar is followed by an infinitive (30a) or a que-clause (30b), the verb cannot combine with me, while with an adjectival attribute, this is no problem, as shown in (30c). (30)
a. Bueno... y entonces hacen arreglos de ese tipo, y por supuesto... e... a veces {*me / Ø} resultan ser buenos libros, que... que son... útiles, pero, normalmente, otros... en otros casos, no. (Habla Culta: Caracas: M7) ‘Well, and then they make arrangement of this type, and of course, sometimes it turns out to be good books, which are useful, but normally, in other cases, they are not.’ b. {*me / Ø} resulta que son buenos libros. c. Los libros {me / Ø} resultan buenos. ‘The books prove interesting to me.’
The question is for what reason the conceptualizer cannot be overtly mentioned. One could argue that resultar’s intersubjectivity blocks “onstage” conceptualizers. Yet, the subjective infinitival construction also needs an “offstage” conceptualizer. Moreover, the conceptualizer of resulta que constructions is diffuse and does not readily connect to a particular entity. Does this mean that the absence of an “onstage” conceptualizer leads to a subjective construal of the infinitival process following resultar? The answer is affirmative. Not the offstage conceptualizer as such, but the shift from a construction with a finite verb to an infinitival one entails a subjective conceptualizer associated with the speaker. Since this subjective reading is not the most natural reading of resultar, the infinitival construction is rather scarcely used. This account confirms the observations made on resultar so far. Hence, with verbs that do not witness an onstage/offstage alter-
Conclusions
83
nation of the conceptualizer, the offstage conceptualizer is only crucial when it is related to the difference between constructions with a que-clause and those with an infinitive. In sum, the notion of subjectification is useful to explain the degree to which the conceptualizer of an evidential statement is subjectively construed. Neither parecer nor resultar can appear with an overt conceptualizer in all infinitival constructions, albeit for different reasons. Parecer readily yields a subjective construal of the infinitival process, whereas with resultar only a couple of verbs allow for a subjectively construed conceptualizer in an infinitival construction. 4. Conclusions As for the transparency, my corpus data have shown that the infinitives that may follow parecer correspond more or less to the same semantic types as the finite verbs in parecer’s que-clause. This is not at all the case for resultar, which only combines with a very limited group of infinitives. The focal prominence of the subject is a result of the verb’s transparency. My account of the prominence is a dynamic interpretation of the thematic subject of parecer described by Porroche (1990) and Cabeza Pereiro (1997), in that the grammatical subject has been argued to play a crucial role in the subjective conceptualization of the event expressed. The ungrounded nature of the infinitival process makes possible a subjective comparison with other similar views related to the elements associated with the subject. The interrelatedness of the non-finite process and the focal position of the subject has proven to be in line with the earlier comments on subjectivity. That is, an intersubjective statement does not give rise to the dynamic mechanism of subjective elaboration. This argument further unravels the oft-mentioned opposition between near-factive and non-factive readings of parecer. Although evidential parecer and resultar do not convey epistemic modality, the subjective comparison with other processes explains then why parecer + infinitive is perceived as less factive than parece que. As for resultar, it is not surprising that the predominantly intersubjective statements do not readily show up in infinitival constructions. There are considerable differences between Traugottian and Langackerian subjectification. Yet, for both models, the greater variety of infinitives involves more subjectification. From a diachronic point of view, the shift from copular complements to all verb types implies an extension of the subjective reading. From a synchronic point of view, the lack of conceptu-
84 Degrees of transparency and subjectification alizers with verbs other than the copular ones leads to a subjective construal of the conceptualizer. In sum, although the analysis of evidence and statements in terms of (inter)subjectivity is not at issue in Cognitive Grammar, one can keep the (inter)subjectivity distinction as a level of interpretation in the conceptualization of evidential expressions, on the one hand, and at the same time cognitively underpin the specific complementation of the linguistic expression of evidentiality, on the other.
Chapter 4 Evidentiality and modality with prometer and amenazar
Another pair of verbs with evidential and epistemic dimensions is prometer and amenazar. Rather than pure likelihood, these verbs express a prediction made on the basis of explicit or implicit evidence. The central argument of this chapter is that prometer conveys a stronger prediction about the event expressed by the infinitive than amenazar. Applied to (1) and (2), the speaker considers it more obvious that winter should be rainy than that the important meeting should be overshadowed by pressure measures. (1)
Este invierno promete ser llovedor. -Llovedor, patrón, promete ser este iinvierno. -Se ve mucha chicharra en el monte. Y es la señal. (Barrios, Eduardo. Gran señor y rajadiablos) ‘This winter promises to be rainy. Rainy, boss, this winter promises to be. One notices lots of crickets in the mountains. And this is the signal.’
(2)
El importante encuentro amenaza ser ensombrecido por medidas de presión de varios sectores, por lo que el gobierno dispuso la intervención de las Fuerzas Armadas. (Notic: Bolivia: ERBOL: 0415-96) ‘The important meeting threatens to be overshadowed by pressure measures from several sectors, as a consequence the government ordered the intervention of the Armed Forces.’
This different kind of prediction can be explained by the different commitment to the event conveyed by lexical prometer and amenazar, here hallar una solución ‘to find a solution’ (3) and dejar la coalición ‘to leave the coalition’ (4). (3)
Los vecinos recorrieron todas las calles del barrio con gran alboroto, hasta que se presentaron autoridades del Partido Comunista, que prometieron hallar una solución a la grave crisis. (Cuba: CubaNet: Junio 11, 1998)
86
Modality with prometer and amenazar
‘The local residents went through all the streets of the district with much racket, till officials of the Communist Party appeared [and] promised to find a solution for the serious crisis.’ (4)
El MBL [Movimiento Bolivia Libre] vuelve a amenazar con dejar la coalición. (Notic: Bolivia: ERBOL: 06/13/96) ‘The MBL once again threatens to leave the coalition.’
In addition to the predictive dimension, there is also an evaluative dimension, which in (1) and (2) consists in an orientation towards respectively a positive and negative evaluation of the event expressed by the infinitive. It is generally assumed that the positive characteristics of a promise and the negative features associated with a threat lead to a positive versus a negative evaluation of the likely event. I will examine the evaluation conveyed by these two verbs in greater detail and show that this assumption is borne out by amenazar but not by prometer. The chapter is organized as follows. The first section presents a series of research lines dealing with promise and threaten, which will be linked to modal prometer and amenazar. In Section (2.), I give an overview of the frequency distribution of modal and lexical prometer and amenazar and describe the interaction of the participants and the illocutionary force (or the absence of it) with the latter readings. The description of the lexical readings will lay the basis for the examination of the prediction hypothesis in Section (3.). In Section (4.), I test the hypothesis that prometer allows for both positively and negatively oriented complements, while amenazar is restricted to negatively oriented ones. 1. Previous research lines on promise and threaten In the field of Spanish linguistics, little attention has been paid to the semantic, pragmatic and conceptual characteristics of modal amenazar and prometer. Hernanz (1999) briefly tackles the issue of subject raising with amenazar and prometer, which she considers an equivalent of the raising mechanism with parecer. Howeve, her formal description does not take into account the semantics of these verbs. More recently, Vázquez-Laslop (2001: 246), has analyzed the difference between lexical prometer, which she refers to as a “deontic control verb”, and “epistemic” prometer, “which shows some features of a raising verb”. Vázquez-Laslop (2001: 246–255) mainly deals with the semantic and syntactic constraints on “epistemic” prometer without commenting on amenazar. In the cognitive-functional
Previous research lines
87
literature, modal promise and threaten have been widely discussed in terms of the development of epistemic meaning (cf. the seminal papers by Traugott 1989, 1997; and by Verhagen 1995, 1996). But Rooryck (1997, 2000), who works in the formal tradition, also links the French equivalents of promise and threaten to different kinds of epistemic modality. Several authors deal with promise and threaten in terms of subjectification. I will discuss these research lines below. 1.1. Epistemicity, inferentiality and evaluative orientations Promise and threaten express some kind of epistemic modality, but the label “epistemic” is not sufficient for their semantic analysis in view of this predictive dimension and this possible evidential import. A question that imposes itself is how the evidence from the world out-there relates to the speaker’s (autonomous) assessment of the state of affairs and the resulting prediction expressed by the semi-auxiliaries. Moreover, the verbs under consideration convey the speaker’s appraisal of the probable state of affairs in positive or negative terms. The relation between the evidential dimension and these two qualifications needs to be investigated further. Traugott (1997) discusses the likelihood and the evaluation expressed by “epistemic” promise and threaten. On the one hand, Traugott (1997: 190) argues that their “epistemic” reading has an inferential dimension since the speaker draws a conclusion on the basis of the reality outside the speaker’s realm. She mentions that promise and threaten share some properties with “inferential” epistemic modals like should and ought to, which Coates (1983: 19) paraphrases as: “from the evidence available I tentatively assume that …”. On the other hand, Traugott (1997) states that promise and threaten also resemble “non-inferential” epistemic modals such as may, might and could, which contain judgments about the likelihood of the state of affairs, situated in the speaker’s subjective realm and correspond to “I think it is likely” (Traugott 1997: 190). As Traugott (1997) rightly points out, the inference can be based on “appearance” (direct evidence) as well as on “expectations and knowledge” (reasoning), as in (5a) and (5b). (5)
a. The Capitol promised to be a large and handsome building, judging from the part about two thirds already above ground. (1795 Twin [ARCHER], taken from Traugott 1997: 188) b. The hapless, aggrieved house-husband threatens to become as rigid and unexamined a comic invention as the grotesquely intru-
88
Modality with prometer and amenazar
sive mother-in-law once was. (1992 Independent [ARCHER], taken from Traugott 1997: 194) In the promise example (5a), the speaker explicitly describes how this halfbuilt building appears to him/her, and bases his/her statement upon it. With threaten in (5b), the comic role of the aggrieved husband is likely to replace the intrusive mother-in-law role. The examples in (5) illustrate that the line between non-inferential reasoning and inference based on reasoning is thin. With promise in (5a) the evaluative comment on the building is clearly based on direct evidence and may stand for both likelihood and evaluation. Reasoning is not absent, however, since any inference involves mental processing that parallels reasoning. The comment in (5b) on the negative evolution of the comic roles is not directly based on direct evidence and, hence, emphasizes the importance of reasoning. The specific status of the non-inferential dimension raises several questions: is non-inferential likelihood implicated from the evidential type of inference? In other words, is the inferential dimension primary with respect to the likelihood expressed or is the non-inferential likelihood inherently present as an independent qualification conveyed by promise/threaten? Traugott (1997) seems to be aware of the problem when she states that neither the (positive or negative) evaluation nor the likelihood can be seen as either inferential or non-inferential. This leads me to conclude that the inferential (evidential) and the non-inferential (epistemic modal) qualifications are distinct from one another. In other words, even if the source of information is clearly inferential (either from direct evidence or from reasoning), the strength of the speaker’s prediction does not necessarily depend on the inferential input. Verhagen (1995, 1996, 2000) argues that Dutch and English beloven/promise and dreigen/threaten have an “argumentational orientation”, which is based on an inferential orientation, on the one hand, and an evaluative orientation, on the other. These orientations apply to both “descriptive” (lexical) and “epistemic” (subjective) readings of the verbs under consideration. Verhagen (2000) states that the lexical and the subjective readings have the same argumentational orientation, which, according to him, involves for both verbs an increased strength of the assumption or expectation. Yet, Verhagen’s own (2000: 200–202) examples do not unambiguously support this argument, as shown in (6).
Previous research lines
(6)
89
a. A: Is Peter coming to the party? (Verhagen 2000: 200-202) B: Sure, he promised. b. A: Are they really going to freeze the budget? B: Well, they did threaten to do so.
In the promise example (6a), the speaker’s reporting that someone promised to be present leads to the expectation that the person will actually be there. Although there is a certain strength of expectation, the threaten example in (6b) does not create the same strength of expectation as that expressed by promise. Nevertheless, Verhagen (1996: 802, 2000: 202) states that the inferential orientation of dreigen ‘to threaten’ is the same as that of Dutch beloven ‘to promise’ and English promise: “[the verb] also function[s] to strengthen the belief or expectation that may be formulated in the complement”. The evaluative orientation of promise/beloven is concerned with the favorable view of the conclusion suggested in the utterances. Verhagen sees the evaluation as a separate dimension, apart from inferential knowledge. Verhagen (1996: 802–804) argues that the evaluative orientation of “epistemic” beloven ‘promise’ is not as strong as the inferential orientation, because beloven can also combine with the negatively oriented attribute tijdrovend ‘time-consuming’, as in (7). The fact that beloven is not restricted to positively oriented complements, shows that the usually positive evaluation can be downplayed to emphasize the likelihood of the event expressed. (7)
[Er wacht hem een nieuwe taak,] die tijdrovend belooft te zijn. (Verhagen 1996: 797) ‘A new task awaits him, which promises to be time-consuming.’
Verhagen (2000: 202) contends that with dreigen ‘threaten’, however, there is no difference of strength between the inferential and the evaluative orientation: “Dreigen and threaten orient the interpreter towards the conclusion that the proposition expressed in the complement may become true, just like beloven and promise do, but this time this conclusion is evaluated negatively, suggesting that something should be done about it; the latter is what makes these verbs different from beloven and promise”. Verhagen’s evaluative orientation offers both useful and problematic elements for the analysis of amenazar and prometer. On the one hand, in distinguishing between an inferential and an evaluative orientation, Verhagen implicitly suggests that the evaluation is never based on inference, whereas the likelihood is seen as exclusively inferential. This view is prob-
90
Modality with prometer and amenazar
lematic, however, since the likelihood should then always depend on the evidential type. Moreover, it is not excluded that the positive or negative view of the event expressed in the evaluation correlates with the speaker’s assumptions based on inference from external evidence. On the other hand, by refraining from attributing an inferential status to the evaluative orientation in “epistemic” promise and threaten, Verhagen (1996) opens the path to an evaluation of the state of affairs that relies on the speaker’s own assessment only. When the evaluation is (partly) detached from inference based on direct evidence, promise and threaten and their equivalents cannot keep a monopoly on positive and negative connotations respectively. That is, the speaker has more flexibility to assess the state of affairs in positive, negative or neutral terms, separately from inferential knowledge. 1.2. Different epistemic values Rooryck (2000) differentiates between an evaluative, a temporal and a modal dimension in French promettre ‘to promise’ and menacer ‘to threaten’.88 As for the evaluative and the temporal dimension, Rooryck (2000) argues that the verb promettre is closely related to the inchoative aller ‘to be going to’ “carrying an additional positive connotation”, while menacer “seems to be basically a variant of risquer ‘to risk’, with pejorative import” (Rooryck 2000: 43; see also Diewald 2004 and Reis 2005 for similar observations on aspect). The temporal dimension of promettre and menacer is concerned with a state of affairs projected into the future, just like the constructions with aller and risquer.89 Importantly, Rooryck links the two verbs to different types of modality, claiming that promettre conveys epistemic necessity, while menacer expresses epistemic possibility, as summarized in Table 16. Table 16. Epistemic necessity vs. possibility (Rooryck 2000: 44) Objective epistemic necessity
Objective epistemic possibility
Future
Aller/promettre
Risquer/menacer
Past
Venir de
Avoir failli
Previous research lines
91
Rooryck (2000: 44) illustrates the opposition between epistemic necessity and epistemic possibility with the two examples presented in (8) and (9). In the menacer example (8), the house may – but need not necessarily – collapse, which is illustrated by the infelicitous reformulation with the adjective “necessary” in (8b). Rooryck argues that with menacer certain but not all conditions are present for the collapse of the house to happen, as seen in (8b). (8)
a. Cette maison menace de s’écrouler. (Rooryck 2000: 44) ‘This house threatens to collapse.’ b. It is possible/*? necessary that this house will collapse. ‘Certain conditions are present for this house to collapse.’
Rooryck (2000: 44) mentions that the possibility expressed in (8b) is in line with the possibility attached to risquer, i.e. the expression of uncertainty. For promettre, by contrast, the possibility reading is too weak, as shown in (9b). (9)
a. Cette maison promet d’être un havre de paix. ‘This house promises to be a haven of peace’ b. It is *? possible/necessary (inevitable) that this house will be a haven of peace ‘All conditions are present for this house to be a haven of peace.’
On the basis of the test in (9), Rooryck claims that it is impossible to say that there is only a chance of the house becoming a haven of peace. The assumption that with promettre all conditions are present for the house to be a peaceful haven leads Rooryck to argue that promettre conveys a necessity reading.90 In this context, Vázquez-Laslop’s (2001: 246–47) comment on Spanish “epistemic” prometer in (10) is helpful. Unlike Rooryck (2000), this author argues that the state of affairs expressed by prometer is “quite likely” (10b) rather than “highly likely” (10c) to come true. (10)
a. El señor del Rosal ha gastado hasta la camisa, pero la boda que ha planeado promete ser de lo más espectacular. ‘Mr. del Rosal has spent down to the shirt on his back, but the wedding he has planned promises to be most spectacular.’ b. It is quite likely/it is expected that the wedding Mr. del Rosal has planned will be most spectacular. c. ? It is highly likely for the wedding to be most spectacular.
92
Modality with prometer and amenazar
My corpus data presented in Section (3.) will also challenge Rooryck’s (2000) distinction between epistemic necessity and possibility.91 Menacer/amenazar and promettre/prometer certainly differ as to the predictive value they express. Although the reformulations of the constructions in (8) and (9) in terms of different conditions may be useful, epistemic necessity is too strong for promettre and its equivalents in other languages. Instead of having recourse to necessity, the relative strength of the prediction will suffice to account for prometer. Whereas Rooryck (2000) does not offer an explanation for the difference between promettre and menacer, Vliegen (2006) does for Dutch and German dreigen/drohen and beloven/versprechen. Vliegen’s main argument is that the different likelihood of realization of the event expressed in the infinitive is due to their different lexical meaning. The likelihood derives from (i) the commitment of the subject in the lexical reading of the verbs, and (ii) the presence or absence of illocutionary force in the verbs’ lexical readings.92 The subject of lexical beloven/versprechen is more committed to the realization of the event expressed in the infinitive than that of dreigen/drohen. The proposal to take subject commitment and illocutionary force into account is an alternative to accounts that link epistemic modality to source-evidentiality. 1.3. Transparency and subjectification Transparent verbs are characterized by their loss of valency and their shift away from the lexical content. The syntactic tradition has mainly focused on the former, while cognitive linguistics most often addresses the latter. Nowadays transparency is increasingly referred to as a continuum with two endpoints, i.e. completely transparent verbs and fully lexical ones, that are opposed to each other both semantically and formally (cf. Langacker 1995, Lamiroy 1995: 278). Lamiroy (1995, 1998) draws attention to the restrictions on the type of infinitive that certain verbs manifest. In her analysis of aspectual verbs, Lamiroy (1995: 283) states that if such a verb can combine with a static verb (verbe d’état), it also can do so with an action verb (verbe d’action), as opposed to Ruwet’s (1983) observation that raising verbs seem to prefer static verbs. It can be assumed that so-called raising verbs such as modal menacer and promettre react differently to an action verb than the inchoative se mettre à ‘to begin’. As for the shift in the lexical content discussed by cognitive-functional linguists,93 transparency is seen as the final stage of the attenuation of objective characteristics associated with a certain verb. Langacker establishes
Previous research lines
93
a clear correlation between the attenuation of subject control, the raising construction and the final transparency of the verb, which is an alternative to the so-called raising of the subject.94 Langacker (2000) argues that modal promise has reached the “extreme situation where attenuation of subject control results in transparency”. Yet, Langacker does not give any empirical underpinning of the alleged total absence of restrictions on the subject or infinitive with promise and threaten. In my analysis of amenazar and prometer, I will be more cautious about the transparency of these verbs and examine the different infinitives that can combine with the modal uses of these verbs. In Chapter 5, I will show that these verbs are subject to restrictions and are, thus, still on the way to complete transparency. The increasing transparency of these verbs can be seen as illustrative of the ongoing subjectification process. The analysis of subjectification starts from the assumption that the shift from the lexical to the “epistemic” readings of promise and threaten is not unmotivated. In the literature there are roughly speaking two concepts of subjectification. On the one hand, there is Traugott’s (1989) notion of diachronic subjectification. On the other hand, there is Langacker’s conceptual and synchronic subjectification concept. For the case of promise/beloven and threaten/dreigen, Verhagen (1995: 121–126, 2000: 203–206) also proposes to differentiate between “character subjectivity” and “speaker-hearer subjectivity” which guide different paths of subjectification. For Traugott, modal promise and threaten are the results of a “unidirectional” process of grammaticalization.95 Traugott (1997) differentiates between three stages for both promise and threaten. From the very beginning, the verb is a performative and commissive verb which expresses commitment to the interlocutor (cf. Traugott 1997: 186). The second stage starts in the sixteenth century, when both verbs allow for a “non-intentional, noncommissive, epistemic” reading in their transitive use with a nominal complement. In examples such as the one in (11), Traugott states that the verbs mean ‘predict’. (11)
... the title of this Paper promising some Experiments about the Production of Electricity, I must not omit to recite.... (1675-1676 Boyle. Electricity and Magnetism 20–21, from Traugott 1997: 187)
The third stage, beginning in the eighteenth century, is characterized by the “non-intentional and epistemic” meanings of promise and threaten with infinitival complements, as in the present-day example in (12).
94 (12)
Modality with prometer and amenazar
The route promised to throw a variety of weather at us. (1992 Flying [HECTOR], taken from Traugott 1997: 189)
In contemporary English, the speaker views the proposition as likely, and evaluates it positively or negatively. Traugott argues that the “epistemic” readings of promise and threaten with an infinitive are a rather recent phenomenon in English (end of the 19th century). In Chapter 6, I will look more closely at the subjectification process of both English promise/threaten and Spanish prometer/amenazar. Traugott’s analysis of promise and threaten is a good basis for analyzing amenazar and prometer from a diachronic point of view. A drawback of Traugott’s (1997) proposal is that promise and threaten are ascribed the same path of evolution, which is questioned by Verhagen (2000). Verhagen’s (1995, 2000) view of subjectification emphasizes the speaker − hearer interaction in the conventionalization of meaning. Although Verhagen (2000) posits the same epistemic value for promise/beloven and threaten/dreigen, his diachronic account, by contrast, points to important differences between the two verb pairs. Verhagen shows that “epistemic” dreigen ‘to threaten’ is older than beloven ‘to promise’ and that these readings appear much earlier than the 19th century, as suggested by Traugott (1993, 1997). Interestingly, in present-day Dutch, “epistemic” dreigen + infinitive is much more frequent than lexical dreigen + infinitive. Moreover, modal dreigen shows up considerably more often than modal beloven.96 According to Verhagen (2000: 203), the synchronic difference in frequency correlates with a diachronic difference. He mentions that while “epistemic” dreigen already appears in the 16th century, as in (13a), beloven begins to appear in the 19th century. The oldest example in the Woordenboek der Nederlandse Taal is from 1793, quoted by Verhagen (2000: 203) and printed in (13b).97 (13)
a. Het schip, twelck dreycht te sinken... (1566) ‘the ship, which threatens to sink.’ b. Dat zelfde meisjen, 't welk iets beloofde te worden.(1793) ‘That same girl, who promised to become something.’
Verhagen (2000: 204) argues that the difference between beloven and dreigen can be explained by their different answers to the following question: “To whom may we attribute “responsibility,” so to speak, for the choice of the words beloven and dreigen on a particular occasion of use?” Verhagen (2000) argues that with beloven/promise the speaker holds the reported
Previous research lines
95
subject responsible for the choice of the word promise, in contrast with a reported threat, which does not allow for such an interpretation. Since threaten cannot be used performatively, the producer of a threat sentence cannot be taken to be reporting the person’s wording of the threat. In other words, “categorizing an event as a threat is definitely the responsibility of the producer, whereas categorizing it as a promise may be taken as simply echoing the subject of the promise” (Verhagen 2000: 204). According to Verhagen, this responsibility of the speaker makes threaten more speakerhearer subjective “from the start”, while promise has remained charactersubjective for a longer period. That dreigen outnumbers beloven is then due to the high frequency of speaker-hearer subjectivity and the low frequency of character subjectivity. This proposal is challenging and could indeed be a valid refinement of Traugott’s account. In my own analysis, I will check whether modal amenazar + infinitive appears earlier than modal prometer + infinitive. Finally, a third type of subjectification presented here is Langacker’s conceptual subjectification. While Traugott’s understanding in terms of pragmatic strengthening is exclusively diachronic, Langacker’s subjectification through attenuation is mainly synchronic. 98 Subjectification stands for “a semantic shift or extension in which an entity originally construed objectively comes to receive a more subjective construal” (Langacker 1991: 215). As Langacker (2000: 298) points out, with verbs like promise and threaten subjectification is no(t) (longer) to be understood in terms of replacement of one situation by another. He follows Verhagen’s (1995: 114) argument that: ... what is called objective use does not differ from subjective use in that subjectivity (of the relevant kind) is lacking, but rather in that it exhibits a certain descriptive objectivity that is absent from the subjective use. I will try to argue that the role of the objective sense of “promise”, etc. in connected discourse in fact incorporates certain features that are definitional of the subjective sense. (Verhagen 1995: 114)
Thus, it stands to reason that the subjective component is immanent in the objective conception and becomes more prominent when the objective construal shifts to a subjective one. The increased subjectivity goes hand in hand with the attenuation of the objective features of the verb. Grammatical constructions undergoing subjectification involve attenuation when the control exerted by an agent decreases (Langacker 2000: 297). Like so many nouns in –ation, attenuation stands for a gradual process that can be represented as a continuum from low to advanced attenuation. An extreme form of attenuation is complete transparency of a grammatical form. The above
96
Modality with prometer and amenazar
presented concept of attenuation, as a motor of conceptual subjectification is an alternative synchronic account of the shift from lexical (“objective”) readings to non-lexical (“subjective”) readings. In Chapter 6, I will analyze the attenuation and subjectification of amenazar and prometer in detail. 2. The lexical origins of the speaker commitment This section discusses the lexical patterns that form a basis for analyzing the speaker commitment of prometer and amenazar’s modal readings. In lexical constructions with prometer, the subject expresses maximal commitment and the action only takes place if the subject wants it. With amenazar, by contrast, the subject does not have a maximal intention to act, due to the fact that the speaker him/herself usually also tends to avoid the accomplishment of the threat. The absence or presence of illocutionary force is an interesting extension of the different commitment to the infinitival process with amenazar and prometer. Below, I first give an overview of the frequency distribution of the two verbs (2.1.). Then I comment on the general properties of subject commitment (2.2.) and test them on the basis of the underlying conditional relations (2.3). Finally, illocutionary force will be presented as an extension of the commitment pattern (2.4.). 2.1. Frequency distribution Table 17 highlights a considerable frequency difference in the proportion of modal and lexical prometer and amenazar constructions with an infinitival complement. Table 17. Infinitival complements with amenazar and prometer. Amenazar
Prometer
Modal
124
% 47.51
34
% 9.86
Lexical
137
52.49
311
90.14
Total
261
100
346
100
The lexical origins of speaker commitment
97
Proportionally speaking, the modal readings of amenazar are more than four times as frequent as the ones of prometer. In other words, almost half of the examples of amenazar with an infinitive have a modal reading, while only 10% of the infinitival prometer constructions do so. A more detailed view is presented in Tables 18 and 19, where I take into account the diversity of the complements of amenazar and prometer. Table 18. Distribution of amenazar constructions Amenazar Fiction Modal % Lexical % Nonfiction Modal % Lexical % Oral Modal % Lexical %
Infinitive
Oblique
85 97.7 61 26.18
Direct objects
queclause
Zero
total
NP
Prep NP
Clitic
0 40 17.17
1 1.15 50 21.46
0 20 8.58
0 23 9.87
0 17 7.30
1 1.15 22 9.44
87 100 233 100
30 71.43 59 22.78
0 34 13.13
7 16.67 48 18.53
0 57 22.01
0 23 8.88
0 2 0.77
5 11.9 36 13.90
42 100 259 100
9 75 17 15.74
0 18 16.67
1 8.33 21 19.44
0 19 17.59
0 22 20.37
0 1 0.93
2 16.66 10 9.26
12 100 108 100
What stands out in Table 18 is the overwhelming majority of infinitival complements in modal amenazar constructions (97.7%, 71.43% and 75%). If we look at the distribution of modal and lexical amenazar + infinitive constructions, there is a big difference between fiction prose, on the one hand, and non-fiction prose and oral speech, on the other. Whereas more than half of these constructions are modal in fiction prose, this is the case for only one third of the constructions in non-fiction and spoken language.
98
Modality with prometer and amenazar
Table 19. Distribution of prometer constructions Prometer
Fiction Modal % Lexical % Nonfiction Modal % Lexical % Oral Modal % Lexical %
Direct objects Nominal Pronominal
Infinitive
queclause
18 40.91 202 40.40
0 97 19.40
13 29.55 97 19.40
13 54.17 59 35.12
0 19 11.31
3 23.08 49 21.12
0 54 23.28
Adjective
Zero
Total
2 4.55 50 10.00
7 15.91 0 -
4 9.09 32 6.40
44 100 500 100
4 16.67 72 42.86
0 10 5.95
2 8.33 0 -
5 20.83 7 4.17
24 100 168 100
3 23.08 32 13.79
1 7.69 56 24.14
4 30.77 0 -
2 15.38 40 17.24
13 100 232 100
Table 19 shows that in the distribution of both lexical and modal prometer complements other than the infinitive play an important role. Except for the fact that que-clauses are impossible and adjectival constructions are possible, the modal prometer constructions with a (pro)nominal complement or without complement seem to be very much in line with the lexical constructions. 2.2. Subject commitment I will now argue that the relation between the subject and the event denoted by the infinitival complement is different in lexical prometer (14) and amenazar (15). (14)
El general Dwight D. Eisenhower, comandante en jefe de las fuerzas aliadas, prometió al pueblo italiano una paz honrosa si los italianos retiraban su ayuda a los alemanes. (Encicl: Italia) ‘General Dwight D. Eisenhower, Commander in Chief of the Allied Forces, promised the Italian people an honorable peace if the Italians withdrew their support for the Germans.’
The lexical origins of speaker commitment
(15)
99
Los partidos MBL y UCS amenazan con abandonar la coalición si el caso no es remitido a la justicia ordinaria, para que ésta sea la instancia encargada de castigar a los criminales. (Notic: Bolivia: ERBOL: 06/14/96) ‘The parties MBL and UCS threaten to leave the coalition if the case is not referred to the ordinary court, so that this instance is in charge of punishing the criminals.’
The subject of lexical prometer is a central, exclusively human or humanoid entity, which assumes responsibility for the thing or the event promised, e.g. una paz honrosa ‘an honorable peace’ in (14). Hence with prometer, the promisor, here general Eisenhower, conveys some kind of commitment to the realization of the promised peace expressed in the complement. The general wants to arrange an honorable exit strategy for the Italians if they withdraw their support from the Germans. As illustrated by this example, in the default case, the act of promising implies that the promisor and the promisee, i.e. the (explicit or implicit) indirect object or oblique, are positively oriented towards realization of the promised act. Although one can also promise not to do something, the promisor generally intends to realize something that the promisee likes. In line with Searle (1969: 63), the utterance of the promise “counts as the undertaking of an obligation to do [the future act] A”.99 Thus, a promise differs considerably from a threat, as Searle (1969: 58) points out: One crucial distinction between promises, on the one hand, and threats, on the other, is that a promise is a pledge to do something for you, not to you; but a threat is a pledge to do something to you, not for you. A promise is defective if the thing promised is something the promisee does not want done; and it is further defective if the promisor does not believe the promisee wants it done, since a non-defective promise must be intended as a promise and not as a threat or warning. (Searle 1969: 58)
This quote points to different behavior of the subjects of lexical prometer (14) and lexical amenazar (15). The subject or addresser of amenazar is not as committed as the promisor of prometer, since the former verb is used to introduce an action that is not undertaken for its own sake, but is considered a means for achieving something else. The addresser only proceeds to the realization of the threat if the requirements expressed are not fulfilled.100 With amenazar, there is no agreement between the addressee and the addresser on the accomplishment of the event threatened. This is an obvious consequence of the fact that the addresser feels obliged to do something that the addressee does not like.
100
Modality with prometer and amenazar
In (15), it is not the implicit addressee of the coalition partner but the covert subject in the conditional clause, i.e. those persons who have to refer the case to the ordinary court, who have the final responsibility for the possible departure of the two political parties announced in the infinitive. The warning expressed in the amenazar statement makes the threat conditional and its realization less likely. By contrast, the utterance of the promise makes the promisor oblige him/herself to attempt to realize the promised event, which is in line with the relative strength of the prediction uttered. With example (16), Verhagen (1996: 801) indicates that the commitment of the subject of prometer leads to an increased argumentational force that points to the realization of the event. (16)
Speaker A: Do you think John will be coming to the party? Speaker B: Well - he promised.
Verhagen (2000) also argues that both “descriptive” and “epistemic” threaten have the force of an argument in favor of the prediction that the event expressed in the infinitive will become true. Interestingly Verhagen (2000: 202) founds his point upon the “descriptive”, i.e. lexical, examples reprinted in (17). (17)
a. Speaker A: Are they really going to freeze the budget? Speaker B: Well, they did threaten to do so. b. He threatened to come, so be prepared: he could appear any minute. c. He threatened to fire us, but things might not be as black as they look.
In my view, these examples do not unambiguously argue in favor of the realization of the event expressed by the infinitive. Examples (17a) and (17c) may also suggest the possible non-realization of the event expressed by the infinitive following lexical threaten. When a speaker says that something might not be as it looks, (s)he predicts that it may not become true. The threaten construction in (17b) is less explicit as to non-realization, but the use of could also implies that it could not. Non-accomplishment of the event introduced by threaten is thus still possible. In conclusion, the different commitment of the subject to the realization of the event is an important factor explaining why the modal readings of prometer and amenazar convey different kinds of predictions.
The lexical origins of speaker commitment
101
2.3. The different impact of conditional clauses In this section I further underpin the difference between lexical prometer and amenazar on the basis of adding a clause introduced by a condición de que ‘provided that’, as in (18) and (19). While si in (14) and (15) is ambiguous in that it can yield both a temporal and a conditional reading, the conjunction a condición de que ‘provided that’ only yields a conditional reading. The conditional clause with a condición de que ‘provided that’ in (18) and (19) will help us deepen the insights into prometer and amenazar gathered so far. (18)
El general Dwight D. Eisenhower, comandante en jefe de las fuerzas aliadas, prometió al pueblo italiano una paz honrosa a condición de que los italianos retiraran su ayuda a los alemanes. ‘General Dwight D. Eisenhower, Commander in Chief of the Allied Forces, promised the Italian people a honorable peace provided that the Italians withdrew their support for the Germans.’
(19)
a. ?? Los partidos MBL y UCS amenazan con abandonar la coalición a condición de que el caso no sea remitido a la justicia ordinaria, para que ésta sea la instancia encargada de castigar a los criminales. b. Los partidos MBL y UCS no amenazan con abandonar la coalición a condición de que el caso sea remitido a la justicia ordinaria, para que ésta sea la instancia encargada de castigar a los criminales. ‘The parties MBL and UCS {Ø / do not} threaten to leave the coalition (but) only if the case is {not / Ø} referred to the ordinary court, so that this instance is in charge of punishing the criminals.’
With prometer in (18), the replacement of si ‘if’ by a condición de que ‘provided that’ does not entail a reformulation of the construction. By contrast, with amenazar the negation is not felicitous within the scope of the conditional clause, as shown in (19a). This leads to the logical reformulation in (19b) in which the subject stops threatening on condition that another action is undertaken. When a condición de que ‘provided that’ is used with amenazar, the negation of the subclause in (19) has to move to the main clause, while this is not the case for prometer in (18). If the exclusively conditional conjunction a condición de que is used, the positive reading of the conditional subclause cancels the threat of the amenazar clause. This corroborates the point that amenazar has an underly-
102
Modality with prometer and amenazar
ing implication that the threat can be averted. With prometer, by contrast, the promise will be carried out if the criteria are fulfilled. In the next section I will look in more detail at the illocutionary force of promise and threat. 2.4. Illocutionary force Illocutionary force refers to the strength with which the speaker performs what he communicates in the utterance. When the speaker says prometo ‘I promise’, i.e. using the first person and the present tense of prometer, (s)he produces a speech act in which the act of promising conveys direct commitment to what (s)he says. Below I reformulate the examples given at the beginning of the chapter which contain the promises by the communist authorities (20a–b) and the threat to the coalition partners (21a–b) to leave the Bolivian government. Examples (20) and (21) show that when the addresser is involved in communicative interaction with an addressee, prometer and amenazar display a different illocutionary force. (20)
a. “Vamos a encontrar una solución a la grave crisis” les prometieron autoridades del Partido Comunista. ‘ “We will find a solution for the serious crisis”, officials of the CP promised them.’ b. “Voy a encontrar una solución a la grave crisis”, te lo prometo. ‘ “I will find a solution for the serious crisis”, I promise you (it).’ c. “Mi padre te va a denunciar”, te lo prometo. (Maldonado p.c.) ‘ “My father will inform against you”, I promise you (it).’
The distributional potential of prometer, illustrated in (20), shows that the promise expressed can be either descriptive, “without involving speaker commitment to it at the moment of speaking” (Nuyts 2001b: 39), as in (20a), or performative, when the speaker’s commitment is involved at the moment of speaking, as in (20b) and (20c). The latter forms have illocutionary force and are, hence, uttered with the highest degree of performativity, as shown by the clitic doubling in (20b)101 and by the use of a negatively oriented proposition (20c). In (20c), someone promises that (s)he will inflict harm to the interlocutor, which boils down to a threat. In such a construction, an interpretive shift takes place from promising something positive to predicting and ascertaining something negative.
The lexical origins of speaker commitment
103
Thus, in contrast to the normal lexical prometer constructions, the emphatic prometo construction with illocutionary force can have either a positive or a negative result for the interlocutor, as shown in (20c). This observation corroborates Searle’s (1969: 58) comments on the English constructions I promise and I hereby promise, which according to him, “are among the strongest illocutionary force indicating devices of commitment provided by the English language”. This is not the case with amenazar, as shown in (21), where the illocutionary force is reduced. (21)
a. Te amenazo con dejar la coalición. ‘I hereby threaten you with leaving the coalition.’ b. “Vamos a dejar la coalición” el MBL les amenazó. ‘ “We will leave the coalition” the MBL threatened them.’ c. *“Voy a dejar la coalición” te lo amenazo. ‘ “I will leave the coalition”, I threaten you (it).’ d. *“Voy a ayudarte” te amenazo. ‘ “I will help you”, I threaten you (it).’
A threat cannot be expressed with the same illocutionary force as a promise. In (21a), the infinitival amenazar construction has illocutionary force directly threatening the addressee. In (21b–d) amenazar combines with direct speech, but has only a performative value in direct speaker-hearer interaction similar to (21a). The ungrammaticality of the utterances in (21c) and (21d) shows that amenazar does not allow for an emphatic assertion directed to the interlocutor.102 In line with the absence of character subjectivity with lexical amenazar given in the previous section, the speaker is prevented from playing a prominent role in the illocutionary commitment to the proposition. Finally, the fact that a positive threat, such as the one in (21d), is excluded illustrates that amenazar does not manifest the shift from a clearly negative expectation to a broader axiological domain. In sum, illocutionary force is another feature of lexical uses that can account for the modal uses. The illocutionary force of lexical prometer explains why the modal use which developed from it, makes a strong prediction. By contrast, lexical amenazar lacks this illocutionary force and, hence, has not developed a modal use making a strong prediction with different evaluative orientations.
104
Modality with prometer and amenazar
3. The modal orientation of amenazar and prometer In this section, I will present corpus evidence showing that speaker commitment differs in modal prometer and amenazar. I will first present the hypothesis to be tested in this section (3.1.). I will discuss the quantitative results of the data analysis of amenazar in section (3.2.), and those of prometer in section (3.3.). 3.1. Working hypothesis The working hypothesis of this section is that the low speaker commitment with modal amenazar correlates strongly with non-realization of the event expressed by the infinitive, while the strong speaker commitment with modal prometer correlates with the realization of the event (cf. Vliegen 2006). The tendency towards realization vs. non-realization illustrates that the verbs make different predictions. To a great extent, my data confirm Vliegen’s (2006) point that, unlike promise-like verbs, threaten-like verbs can more easily be linked to non-realization of the event expressed in the infinitive. However, as shown in Cornillie (2004: 11–17), there are restrictions on the tense of these forms: non-realization is exclusively found in past contexts. Furthermore, an exception should be made for meteorological expressions. 3.2. Prediction with amenazar The analysis presented in this section will confirm that there is low speaker commitment with amenazar, which is in line with Rooryck’s (2000: 43–44) analysis of menacer in terms of epistemic possibility. We will begin with a series of amenazar examples that illustrate how the event expressed by the infinitive behaves with regard to (non-)realization. In a considerable number of examples the (complete) realization of the event is clearly blocked, as in examples (22) and (23). In (22), a porción de leche ‘portion of milk’ is about to boil over. Before this happens, the woman removes the fire that heats up the milk and by doing so prevents the accomplishment of the event described in the infinitive from taking place. (22)
Al lado de la mesa en el suelo y dentro de un brasero de latón, borboteaba una tetera y amenazaba subirse, dentro de un jarro de
The modal orientation
105
hierro enlozado, una porción de leche. La mujer removió el fuego, puso unas tazas y unos platillos. (Rojas, Manuel. Hijo de ladrón) ‘Next to the table on the ground and in a tin brazier, a tea can was bubbling and in an iron pitcher a portion of milk threatened to boil over. The woman removed the fire, placed cups and a couple of saucers.’ In (23), something tense (algo tenso) is about to break down. In the subsequent utterance, however, it is literally mentioned that nothing happened. (23)
En la obscuridad blanquean las camisas de los trabajadores y en el aire hay algo tenso que amenaza romperse de un momento a otro. Nada llegó a romperse, sin embargo. (Rojas, Manuel. Hijo de ladrón) ‘In the darkness the workers’ shirts appeared white and in the air there is something tense that threatens to break down from one moment to another. Nothing happened to break down, however.’
As the preceding examples show, contextual clues often offer conclusive evidence. This does not mean, however, that the exact number of cases with non-realization can be predicted conclusively. Indeed, some contexts of amenazar remain undecided as to the realization of the event expressed by the infinitive, as in example (24) for instance. Here, the protagonist seems to be unable to hold his false teeth in his mouth properly. (24)
Natalicio, Gondra, Coronel Jara, Eligio Ayala, todo lo que podía recordar y cada nombre desataba un torrente de amarga verborragia que amenazaba llevarse la dentadura postiza que bailaba enloquecida y parecía salirse una y otra vez. (Halley Mora, Mario. Manuscrito alucinado: las mujeres de Manuel). ‘Natalicio, Gondra, Coronel Jara, Eligio Ayala, all yhat he could remember and each name unleashed a stream of bitter wordmongering that threatened to remove his false teeth which foolishly danced and seemed to pop out several times.’
I have analyzed an example such as (24) as “open” with regard to the nonrealization or realization of the event expressed in the infinitival complement in conformity with the lack of concrete evidence.
106
Modality with prometer and amenazar
Table 20. Realization or non-realization of the event expressed by the infinitive modal
nonrealization
open perspective
realization
total
amenazar %
33 26.61
91 73.39
0 0
124 100
As shown in Table 20, in up to 26% of the modal constructions with amenazar, there is evidence that the infinitive is ultimately not realized, while the realization of more than 73% of the infinitives following amenazar is left open. Importantly, no examples were found in which the realization of the event was explicitly communicated.103 From Table 20 we can conclude that amenazar does not express strong speaker commitment to the realization of the proposition. Note that all the above examples with ultimate non-realization of the event are in the past tense or in the historic present tense. In narrative discourse, a past form of amenazar seems to acquire a sort of aspectual reading with the meaning ‘to be about to + inf’ (cf. Diewald 2004; Reis 2005; Heine and Miyashita 2007). As a matter of fact, amenazar introducing an eventually non-realized event is a means of putting emphasis on the possibility of this event in the succession of actions or events in the narrative discourse (Nuyts p.c.).104 The fact that almost all non-realization examples of amenazar are in the past tense is not without importance. The apparent tense restriction invites us to verify to what extent the possibility of a nonrealization reading is also available with the present tense constructions of amenazar. This will be the topic of the next section. 3.3. Refinement of the non-realization criterion This subsection examines whether modal amenazar constructions express another type of speaker commitment when they occur in the present tense. I first discuss the general tense distribution (3.3.1.). Then, I show that in the present tense amenazar can also convey modal non-realization of the event expressed in the infinitive (3.3.2.). Finally, I mention that some amenazar constructions tend to express likely realization of the event (3.3.3.).
The modal orientation
107
3.3.1. Tense distribution Let us first look at the distribution of the present and past tenses in all examples with modal amenazar in my data-base, represented in Table 21.105 Table 21. Modal amenazar: distribution of tense present imp.
imp. past
simple past
Gerund past (past) subjunc.
Total
non-realization %
2 6.06
20 60.61
7 21.21
1 3.03
3 9.09
33 100
open perspective %
40 43.96
46 50.55
1 1.10
4 4.40
0 0
91 100
As shown in Table 21, only two occurrences of modal amenazar with a non-realized event are in the present tense, which is only 6% of all expressions suggesting non-realization, and only 5% of all present tense examples of modal amenazar (2/40). The past tense constructions thus constitute the overwhelming majority of examples in which the event is not realized.106 At the same time more than 55% of the constructions with an open perspective are also in the past tense. Threfore, the question imposes itself whether there is a correlation between present tense and strong prediction of realization of the event expressed in the infinitive. In other words, can the examples in the present tense – which are not related to the historic present – somehow escape the tendency to non-realization of the event they describe? 3.3.2. Possible non-realization in present tense constructions Although nearly all amenazar examples in the corpus are in the past, modal amenazar in the present can also convey possible non-realization. The possible non-realization with amenazar in the present tense seems to be related to the speaker’s view of the characteristics of the subject. With amenazar, the overt mention of the subject enables a changing speaker’s view of the state of affairs.107 In (25), amenazar appears in the present tense and communicates that the existing tensions (fricciones y tensiones) have the potential of leading to mutiny. However, the speaker does not say that (s)he believes that there
108
Modality with prometer and amenazar
will be disorder, but rather expresses the presupposition that ongoing tensions may lead to more disorder. These tensions could also disappear, however, in which case there will be no mutiny. (25)
Las fricciones y tensiones que los afligen en la actualidad amenazan con llevarlos a un motín. Pero somos, lastimosamente, una minoría y poco caso se hace de nuestros consejos y recomendaciones. (González Real, Osvaldo. El Mesías que no fue y otros cuentos) ‘The frictions and tensions that plague them currently threaten to lead them to mutiny. But we are, unfortunately, a minority and they don’t listen to our advice and recommendations.’
In example (26), aventuras mal orientadas como estas amenazan ‘misguided adventures like these threaten’ implies that more of the same activities could have serious political repercussions for the governor in question. (26)
... la FNCA denunció el reciente viaje del gobernador republicano de Illinois, George Ryan, y afirmaron que ‘aventuras mal orientadas como estas amenazan con tener serias repercusiones políticas para el Partido Republicano. (Fidel Castro, 01/11/99) ‘... the FNCA denounced the recent trip of the Illinois Republican governor, George Ryan, and they affirmed that ‘misguided adventures like these threaten to have serious political repercussions for the Republican Party.’
Still, it is not certain that the governor’s first trip to Cuba will have any consequences at all. The trip of the governor might not lead to serious repercussions for the Republican party after all. So, in the end the utterance keeps an open perspective on the event, which includes both nonrealization and realization. In (27), the infrastructural problems with electricity and water supply have the potential of inflaming the citizens. Once again it is said that when this situation gets worse, people’s patience will come to an end: that is, the event described by the infinitive will be eventuate. However, if things improve, the event will not be realized. (27)
Eso, unido a la escasez crónica del fluido eléctrico y a las esporádicas interrupciones en el suministro de agua, amenazan con acabar con la paciencia de los ciudadanos. (Notic: Col: Semana: 820)
The modal orientation
109
‘This, together with the chronic scarcity of electric power and with the sporadic interruptions of the water supply, threaten to consume the patience of the citizens.’ So far I have shown that modal amenazar constructions that have an overt subject convey the potential of non-realization of the event. They, thus, express low speaker commitment and convey a weak prediction. The amenazar constructions that lack such a subject are discussed in the next subsection. 3.3.3. Possible realization in meteorological utterances Modal amenazar also combines with complements expressing meteorological processes as in (28). In such a construction without overt subject, the prediction with amenazar cannot be based on different phases of the potential evolution of the state of affairs. (28)
a. Amenaza con llover. ‘It threatens to rain.’ b. El día amenaza con ser muy deprimente. ‘This day threatens to be very depressing.’
In these examples, the speaker is more committed to the proposition and probable realization is presented as the most natural outcome. Since the utterances in (28) lack an underlying potential that can prevent the realization of the event, nothing can influence this view of the realization. Meteorological verbs or subjects like el día ‘the day’ do not invite the speaker to subjectively elaborate on the preceding context or the evolution sketched in the utterance. Statements with amenazar in the present tense as in (28) rely on inference from direct evidence. This type of inference implies that the relevance of the event expressed in the infinitive is high for the speaker and the interlocutor. The likelihood that the speaker’s prediction corresponds to reality is much higher than with past constructions or with present constructions containing a referential subject. By contrast, the amenazar constructions in (26) and (27), which have a potential towards non-realization, involve inference from reasoning. In the present tense modal constructions with amenazar, the difference between inference from direct evidence and inference from reasoning seems to correlate with an open perspective versus a possible non-realization reading, respectively. This might indicate that, in
110
Modality with prometer and amenazar
some cases, the type of inference leads to a specific predictive judgment and subsequent speaker commitment. Further research on this topic is needed, since this may not hold for all modal amenazar forms in the past tense. The following four Internet examples of amenazaba con llover show that (i) the four statements are based on inference from direct evidence, and (ii) all outcomes are possible, realization (29–30) as well as non-realization of the rain (31–32). (29)
Más tarde, sobre las dos de la tarde, quedamos en la peña con todo el mundo y empezamos a preparar la paella. El tiempo había cambiado, estaba nublado y amenazaba con llover, cosa que no tardó mucho. (http://www.sebulcor.com/lapaella.htm) ‘Later on, around 2 PM, we met in the bar with all the other people and we started preparing the paella. The weather had changed, it was cloudy and it threatened to rain, something that did not last.’
(30)
El día jueves cuando salimos de viaje el tiempo amenazaba con llover y llegamos con lluvia. (http://www.punksunidos.com.ar/salud/amanita2.htm) ‘Thursday when we left for a journey the weather threatened to rain and we arrived with rain.’
In (29) and (30) the source of information is inference from direct evidence, because the respective speakers are present in these utterances. Importantly, other past tense examples of inference from direct evidence clearly include the observation that rainfall in the end did not take place. In (31), a jobless person from Argentina tells that it threatened to rain, but he also mentions that in the end it did not. In (32) the concessive clause communicates that, although showers were imminent, the rain did not eventuate. (31)
Muchas de las personas que pasaron la noche frente al Nación provenían del conurbano bonaerense y llegaron hasta el lugar con mantas y ropa para hacer frente al frío de la madrugada. "Amenazaba con llover, pero por suerte no llovió", indicó Rodolfo, un desempleado que ocupaba el lugar 133 en la fila. (El día, Argentina, 28-03-2002) ‘Many persons which remained [in the line] the whole night in front of the Nation [building] came from the Buenos Aires urban quarters and arrived at the place with blankets and clothes to brave
The modal orientation
111
the early morning cold. “It threatened to rain, but fortunately it didn’t rain”, Rodolfo, an unemployed who occupied place 133 in the row, pointed out.’ (32)
Era de noche y el cielo amenazaba con llover aunque no lo hacía, pero la noche se encontraba completamente oscura y nublada... (http://www.fortunecity.es/ilustrado/inanimado/180/gmagicas/Lauri er3.htm) ‘It was night and the sky threatened to rain, although it didn’t, but the night was completely dark and cloudy.’
From the preceding past tense examples, it has become clear that inference from direct evidence is not a criterion that can account for all constructions with expected realization. Although the inferential type may be crucial in modal amenazar constructions in the present tense, this is not so in past tense constructions. 3.4. Prediction with prometer In this section I will discuss prometer in terms of realization and nonrealization of the event expressed in the infinitive. This section is subdivided into two subsections. In (3.4.1.), I first discuss the tense distribution in examples with prometer, before dealing more analytically with the corpus examples in (3.4.2.). 3.4.1. Tense distribution Table 22 adds the figures of prometer to those of amenazar. Since in my data I did not find any clear cases of non-realization of the event expressed by the infinitive of prometer, we can conclude that, unlike amenazar, prometer always leaves open the future, which, as we will see, correlates with stronger speaker commitment to the prediction made.108
112
Modality with prometer and amenazar
Table 22. (Non-)realization of the event expressed by the infinitive modal
nonrealization
open perspective
realization
Total
amenazar %
33 26.61
91 73.39
0 0
124 100
prometer %
0 0
34 100
0 0
34 100
Since tense has proved a crucial indication of the strength of the prediction with some amenazar constructions, it is also important to look at the distribution of tense with prometer. Table 23. Modal prometer: distribution of tense present imper- past imper- subjunctive fect fect (past)
total
open perspective
23
10
1
34
%
67.65
29.41
2.94
100
If we compare Table 23 with Table 21 for amenazar it becomes clear that the tense distribution with modal prometer is the inverse of that with amenazar: two thirds of the non-lexical prometer constructions are in the present tense. In spite of the fairly limited number of examples at my disposal, some correlations can be observed. (i) The fact that the illocutionary force of lexical prometer only shows up in the present tense may favor a modal construction in the present tense. (ii) The present tense may be evidence for the assumption that prometer is preferably based on inference from direct evidence. (iii) The preference for the present tense allows for pragmatic strengthening of the future dimension. This future dimension of the present tense enables strong prediction.
The modal orientation
113
3.4.2. Possible realization in present and past tense constructions In this section I will examine the likelihood of realization of the event with prometer and compare it with the insights gathered for amenazar. In (33), the past form of prometer describes the likelihood of the continuation (prolongarse) of a state of affairs defined as alegría ‘joy’. (33)
Los sones del arpa, las guitarras y el bandoneón daban el condimento especial a una alegría que prometía prolongarse por lo menos hasta el alba del día siguiente. (Dimas Aranda, Santiago Vida, ficción y cantos) ‘The harp sounds, the guitars and the bandoneon gave a special flavor to a [atmosphere of] joy that promised to be continued at least till dawn the next day.’
The belief that the joyful atmosphere will continue is not influenced by potential changes of the state of affairs. The characteristics associated with the subject do not suggest for any change in the state of affairs. The same reading is possible for (34), where all evidence available points to the continuation (persistir) of the rain. Although there is no suggestion of nonrealization with prometer, there is no guarantee that it will certainly continue to rain either. (34)
Traían sólo unas pocas botellas, pues una mayor parte la habrían ocultado en algún lugar del bosque. Empezaba a caer una lluvia fina que prometía persistir y levantada la barbacoa se trasladó el fogón al amparo del techo de tacuapí. (Rivarola Matto, José María, Follaje en los ojos: Los confinados del Alto Paraná) ‘They only had a couple of bottles with them, because they had hidden the major part of these in some place in the forest. The drizzle began to fall, which promised to go on and once the barbecue came to an end, the burner was passed under the reed roof.’
In the present tense construction in (35), the walls are expected to be transformed (transformarse) into something special (algo especial), which can be many different things. The important element in this construction is that the speaker can no longer balance hedge between possible non-realization or realization, as is the case with amenazar. With prometer the speaker can only express commitment to the prediction of realization of the event. This is also the case with (36).
114
Modality with prometer and amenazar
(35)
Más allá, casi colgando de un barranco, un atisbo de otra casa. Son cuatro paredes que se levantan y prometen transformarse en algo especial, un atelier quizás. (Karlik, Sara, Nocturno para errantes eternos) ‘Further away, almost hanging on a gorge, a glimpse of another house. It is four walls that stand up and promise to become something special, a workshop perhaps.’
(36)
La muestra, que promete ser apasionante, empieza en el XIX, con el alemán Leibl o el norteamericano Eakins, Maillol, Ensor, Munch, Khnopff y otros simbolistas. (Notic: España: ABC) ‘The exposition, which promises to be fascinating, begins in the 19th century, with the German Leibl or the North-American Eakins, Maillol, Ensor, Munch, Khnopff and other symbolists.’
The above discussed prometer constructions do not support Rooryck’s (2000: 43–44) proposal to handle prometer in terms of “epistemic necessity”. Instead, they express prediction. In (37), I will use the tests that Rooryck (2000: 44) applies to promettre and menacer (see Section 1.2). As we will see, the notion of prediction will prove helpful to nuance the distinction between possibility and necessity. The (b) reformulations are adapted from Rooryck’s (2000: 43–44) analysis, while those in (c) are my own reformulations. (37)
a. Los sones del arpa, las guitarras y el bandoneón daban el condimento especial a una alegría que prometía prolongarse por lo menos hasta el alba del día siguiente. (Dimas Aranda, Santiago Vida, ficción y cantos) ‘The harp sounds, the guitars and the bandoneon gave a special flavor to a [atmosphere of] joy that promised to be continued at least to dawn the next day.’ b. It is *? possible/necessary (inevitable) that the joy will continue at least to dawn the next day. ‘All conditions are present for the joy to continue at least till dawn the next day.’ c. It is expected/can be predicted/possible/likely/ *necessary (inevitable) that the joy will continue at least till dawn the next day. ‘All conditions are present for the joy to continue at least to dawn the next day.’
The modal orientation
115
Although the alegría example in (37) displays all conditions for assuming that the joy continues at least till dawn the next day, this does not mean that the event will necessarily or inevitably take place.109 Against Rooryck’s rejection of a “possible”-analysis of the promettre constructions (37b), I hold that possibility seems to be always implied in prometer.110 Since “possibility” as such does not suffice to describe prometer, I also refer to the predictions or expectations about the realization of the event expressed in the infinitive. In other words, in (37), it is said that all conditions are present for assuming that the joy will continue: the continuation is in line with the expectations. Since the speaker cannot foresee every possible impediment to the realization of the event evoked in the infinitive, the epistemic necessity reading proves to be too strong. In sum, with modal prometer all conditions are present for believing that the event expressed in the infinitive will come true. The speaker commitment is stronger than that of amenazar. Yet, this does not mean that the event will necessarily be realized, as the term “epistemic necessity” suggests. Instead, prometer conveys a kind of prediction about the possibility of the event, which is often seen as an equivalent of “high probability/likelihood”.111 3.5. Theoretical implications In view of the above-presented analysis, I would like to make some comments on terminological and theoretical issues raised by van der Auwera and Plungian’s view of the overlap between evidentiality and epistemic modality (1998). The corpus data illustrate that epistemic modality is present in both amenazar and prometer. Since these verbs are also considered evidential in nature, they belong to the overlap domain between evidentiality and modality. Van der Auwera and Plungian (1998: 85–86) limit this overlap domain to epistemic necessity and inferential evidentiality: The claim is only that the inferential reading amounts to epistemic modality and more particularly necessity: for both categories we are dealing with the certainty of a judgment relative to other judgments. From this point of view it also causes no surprise that inferential evidentials often receive an English translation with epistemic must. Inferential evidentiality is thus regarded as an overlap category between modality and evidentiality. [see Table 24]
116
Modality with prometer and amenazar
Table 24. Inferential evidentiality = epistemic necessity Necessity ...
Deontic necessity
Epistemic necessity = Inferential evidentiality
Quotative evidentiality
...
Evidentiality
They rightly state that the inferential reading should not exclude other evidential readings (e.g. the Turkish evidential -miş has inferential, reportative and mirative readings,112 cf. Aksu-Koç and Slobin 1986: 159). However, van der Auwera and Plungian’s (1998: 85) restriction to epistemic necessity does not seem warranted against the background of the analysis presented above. The case of modal prometer and amenazar shows that the overlap category between modality and evidentiality should not be exclusively situated in the domain of necessity but can also include prediction. In the next section, I will examine whether the lexical patterns of amenazar and prometer also influence the evaluative orientation of their modal counterparts. 4. The evaluative orientation of amenazar and prometer Amenazar and prometer also have a diametrically opposed evaluative orientation viz. negative versus positive. In this section I will investigate to what extent the modal readings follow this pattern. In the first subsection I present the working hypothesis (4.1.). The other subsections describe the negatively oriented complements of prometer (4.2.) and the positively oriented ones of amenazar (4.3.). 4.1. Working hypothesis In its lexical readings, prometer generally patterns with a positively evaluated event expressed in the infinitival complement, while amenazar combines with an infinitival process whose realization is to be avoided. Negatively evaluated complements with prometer and positively oriented ones with amenazar even seem to be semantically incompatible. It is often argued that promise by default stands for a positive speaker evaluation, whereas threaten leads to a negatively oriented evaluation.
The evaluative orientation
117
However, Verhagen (1995: 117) mentions both negatively and positively oriented complements for “epistemic” beloven ‘to promise’. Verhagen (1996: 801–802) states that a negative construction often renders a kind of ironical statement that proves the opposite of what is said, while the “epistemic” reading of dreigen ‘to threaten’ “necessarily indicates a negative evaluation”. Verhagen (2000: 202) argues that the negative evaluation of dreigen and threaten “makes these verbs different from beloven and promise”. Although these characterizations are not wrong, a more fine-grained account of the opposite orientation of the complements of both prometer and amenazar should be based on corpus data and be related to dominant characteristics of the lexical reading. As shown in Section (2.4), due to its illocutionary force, prometer can combine with negatively oriented events, while this does not hold for amenazar. The working hypothesis of this section, then, is as follows: the modal reading of prometer explains why prometer can emphasize the prediction of a negative event despite its default positive connotation. With prometer, the speaker foregrounds the prediction and backgrounds the evaluation of the state of affairs. I hypothesize that the inability of lexical amenazar to introduce a positive event leads to negatively oriented complements only. The weaker prediction attached to amenazar leaves more room for the course of the events, i.e. the natural evolution of reality. The inability to transcend this evolution is expected to corroborate amenazar’s reluctance to combine with positively oriented complements. 4.2. Evaluation with prometer The corpus examples of prometer below allow us to check whether the speech act function of prometer is stronger than the positive evaluation. As shown in (38), modal prometer can also precede a negatively oriented complement. (38)
Tomé aquel desayuno y me dirigí a encarar mi trabajo de mecánico. El sol prometía ser inmisericorde, y no me resultó extraño que Nila apareciera de nuevo trayendo para mi uso un sombrero pirí ancho. (Halley Mora, Mario. Los habitantes del abismo) ‘I had that breakfast and I went to start my job as a mechanic. The sun promised to be merciless, and I wasn’t surprised that Nila reappeared bringing me a very wide sombrero.’
118
Modality with prometer and amenazar
In (38), the modal reading of prometer has a mixed effect: although the evaluation by means of inmisericorde ‘merciless’ can hardly be perceived as positive, the whole construal is not necessarily negative. Thus, the kind of prediction expressed transcends a positive or a negative evaluation. The construction in (39) confirms that prometer readily combines with negatively oriented complements without therefore inducing a negative evaluation. (39)
Con un merengue hip-hop del grupo dominicano Sandy & Papo, Telecom lanzó el primer golpe en la que promete ser la más agresiva contienda comercial de los próximos años: la lucha por el mercado de la larga distancia. (Notic: Col: Semana: 840) ‘With a hip hop merengue by the Dominican group Sandy & Papo, Telecom launched the first assault of what promises to be the most aggressive trade war of the coming years: the struggle for the longdistance market.’
In (39), the aggressive battle for a good commercial position on the market is not seen as something negative, although the complement is not inherently positive at all. This is also the case with (40). (40)
Lleras Restrepo tuvo que dirigir las controvertidas elecciones del 19 de abril de 1970. Su libro, que promete ser explosivo por las nuevas revelaciones que hace, tuvo sin embargo varios obstáculos. (Notic: Col: Semana: 831) ‘Lleras Restrepo had to coordinate the controversial elections of April 19th 1970. His book, which promises to be explosive because of its new revelations, nevertheless had several obstacles.’
The adjective explosivo ‘explosive’ carries, in principle, a negative evaluation by nature. Yet, in (40) it is used in a metaphorical way, thus overriding incompatibilities between the default positive evaluation of prometer and the negative complement. This figurative use cannot be the main explanation, however. The strong prediction plays an important role: the inference is strengthened by the explicit mention of por las nuevas revelaciones que hace ‘because of the new revelations it makes’. In sum, the speaker’s ability to express his/her own evaluation of the likely state of affairs is well developed in prometer, since the speaker can emit both negative and positive statements. The prediction is even more prominent than the evaluation (cf. Verhagen 1996: 802). The speaker commitment also attenuates the speaker-orientedness of the negative orienta-
The evaluative orientation
119
tion. Both elements are linked: prometer’s tendency towards realization reduces the impact of the positively or the negatively oriented complements.113 4.3. Evaluation with amenazar I will now verify whether the lack of a speech act function impedes transcending the negative evaluation with amenazar. Below I attempt to describe the rationale behind amenazar examples that combine with positive complements, such as in (41). (41)
… las nubes multiplican su volumen, ganan peso y amenazan con quebrar la superficie de la tierra liberando a las criaturas del abismo. (Ledrado, Arturo. Viceversa) ‘the clouds increase their volume, become heavier and threaten to break the surface of the ground liberating the creatures from the abyss.’
In (41), the speaker describes the evolution of an event that is about to happen. In this literary example, the clouds will probably break the surface of the ground into pieces. Such an evolution is neither positive nor negative. In fact, the outcome of this contact between the clouds and the soil is the liberation of the creatures from the abyss, which could be something positive. Examples like these show that with amenazar too, the prediction can overrule the orientation of the evaluation. Although in many modal examples with amenazar the lexical semantics of threat remain present, the utterance in (41) suggests that amenazar may be experiencing a process of attenuation of the threat. There is a tendency to probable realization when the evaluation is backgrounded. In an utterance such as (42), the speaker does not have a welldetermined position with regard to the evaluation usually associated with amenazar. (42)
El sol se ocultaba lentamente detrás del cerro San Francisco, dos mbiguás nadaban en pareja hacia el sur. La correntada era fuerte y una tenue brisa amenazaba con convertirse en un viento más ligero. (Gayoso Manzur, Milia. Ronda en las olas) ‘The sun slowly disappeared behind the San Francisco hill, two mbiguá-birds swam together southwards. The stream was strong and a persistent breeze threatened to become a lighter wind.’
120
Modality with prometer and amenazar
In (42), the change from a breeze to a lighter wind cannot be viewed as negative. In this example, prometer could replace amenazar, while keeping the same evaluation. The two verbs differ in temporal perspective, however. Amenazar usually stands for (short-term) immediacy: the speaker/ writer emphasizes that the change is about to happen. Prometer, by contrast, can involve (long-term) futurity. In other words, amenazar without evaluation appears to have an increasingly more pronounced aspectual dimension of immediacy. In the corpus, the examples of amenazar with positive complements are less frequent than the ones of prometer with negative complements, despite the fact that modal amenazar + infinitive appears four times as much as prometer. Although in certain readings it yields a more aspectual (inchoative) reading, amenazar still very much has a negative evaluation associated with it. 5. Conclusions In search of a better explanation for the different types of speaker commitment and prediction conveyed by modal prometer and amenazar, we have found that the illocutionary force and subject commitment present in their lexical readings form an important explanatory path. In lexical prometer constructions the subject expresses maximal commitment to the realization of the event expressed by the infinitive, while with amenazar the subject is not held to a commitment to act. The absence or presence of illocutionary force is an important extension of these different degrees of commitment. On the basis of this description two hypotheses were postulated: (i) the prediction of the event expressed by the infinitive is weaker with modal amenazar than with prometer (first hypothesis); (ii) the evaluation is more speaker-oriented with prometer than with amenazar (second hypothesis). The first hypothesis is confirmed to a large extent. The modal values of amenazar and prometer correspond to low versus intermediate speaker commitment, respectively. Although the majority of examples with amenazar have past tense, in the majority of cases the present tense forms of modal amenazar can also have non-realization. In some performative cases, inference through direct evidence leads to stronger speaker commitment. Yet, this is no guarantee for a general correlation between source of information and speaker commitment. The analysis given in this section refines Vliegen’s (2006) claim about the likelihood of non-realization in that not all dreigen/amenazar constructions express the same value. My corpus data also show that Rooryck’s
Conclusions
121
label of “epistemic necessity” is too strong a term to refer to the prediction expressed by prometer. The combination of prediction and evidentiality refines van der Auwera and Plungian’s (1998) overlap category of inferential evidentiality and epistemic necessity. The second hypothesis tested in this chapter was concerned with the evaluation expressed by prometer and amenazar. The analysis has shown that prometer readily combines with negatively oriented complements, while amenazar is reluctant to combine with a positively oriented complement. But the corpus examples of amenazar introducing positively oriented ones point to ongoing change from lexical to modal amenazar. In the next two chapters I will detail the semantic change of prometer and amenazar in terms of transparency (Chapter 5) and subjectification (Chapter 6).
Chapter 5 Partial transparency with modal prometer and amenazar
Lexical prometer and amenazar are often followed by verbs that depict action. Their subjects that utter a promise or a threat are then agentive. The subjects of modal prometer and amenazar, by contrast, are not the utterers of a direct promise or a direct threat.114 These modal constructions also necessarily lack an explicit interlocutor or addressee of the promise or threat typical of the lexical readings. Because of the lack of an agentive role for the subject and the lack of an addressee, modal prometer and amenazar can be considered more “transparent” than their lexical counterparts. Langacker (2000: 315) argues that modal promise represents the “extreme situation where attenuation of subject control results in transparency”. It is not clear, however, whether interferences related to some kind of subject control are completely absent. A more detailed analysis of the infinitival complements and the subjects of the modal constructions allows the analyst to examine to what extent they can be considered transparent. In this chapter I will show that modal amenazar and prometer still face restrictions that stem from the lexical readings. The chapter is organized as follows. Section (1.) examines in detail the restrictions on the infinitives that follow modal prometer and amenazar. In section (2.) I describe the restrictions on the subject of these verbs. 1. Restrictions on the infinitive In this section, I will present the results of corpus research on the infinitives selected by modal amenazar and prometer. In Cornillie (2004: 17–24), it was shown that the infinitives following modal amenazar and prometer belong to a different semantic type. The analysis of the infinitives presented below elaborates on this difference and corroborates Lamiroy’s (1995, 1998) conclusion that full transparency is rare. In (1.1.), I first present the hypothesis and the proposed classification of the infinitives and then give a general overview of the distribution of the infinitives (1.2.). Next, I analyze the infinitival complements of amenazar (1.3.) and those of prometer (1.4.). Finally, in (1.5.), I will discuss the in-
Restrictions on the infinitive 123
finitives that cannot combine with amenazar and prometer. Subsection (1.6.) summarizes the main points of this section. 1.1. Working hypothesis and methodology The central hypothesis about transparency with amenazar and prometer is that although the subject of modal amenazar and prometer is expected to be constrained by the infinitive (cf. Gaatone 1995), the two verbs have a preference for combining with specific types of infinitive while they reject other ones (cf. Lamiroy 1995, 1998). Due to the competition with the agentive lexical readings, modal prometer is expected to combine with a more limited group of infinitives than modal amenazar.115 For the analysis of the infinitival complements below, I have distinguished between four semantic categories of verbs, namely, (i) copulas and other attributive verbs, (ii) aspectual verbs, (iii) verbs expressing a change of state and (iv) action verbs.116 I link the different infinitives with amenazar and prometer to the internal and external viewpoints they express respectively. The fourfold categorization allows me to differentiate between change-of-state and action verbs that point to an internal viewpoint with amenazar, on the one hand, and attributive predicates suggesting an external viewpoint with prometer, on the other. Unlike with my analysis of parecer (Chapter 3, Section 1.3.), I will list the infinitives that occurred in the data in the endnotes, as they may form the basis for further research. 1.2. Global distribution Let us first look at the infinitives found in the lexical versus the modal constructions of amenazar and prometer. For the examination of the lexical readings I will complement the four verb types proposed in the previous section with verbs of communication, e.g. hablar ‘to talk’, decir ‘to say’ or contar ‘to tell’, and with mental state verbs, e.g. esperar ‘to wait/hope’ and respetar ‘to respect’. As shown in Table 25, more than 70% of the infinitives with lexical amenazar and prometer are action verbs, such as ayudar ‘to help’ in (1). (1)
¿Cómo llegamos? El camionero prometió ayudar pero puso sus condiciones. Nada mejor pude encontrar. (Dimas Aranda, Santiago. La pesadilla)
124 Partial transparency with modal prometer and amenazar ‘How do we get there? The truck driver promised to help, but he imposed his conditions. I could not find any better.’ Communication verbs and mental activity verbs combine more or less equally frequently with lexical amenazar and lexical prometer. The most striking difference between the two verbs lies in the fact that change-ofstate verbs are slightly more frequent with amenazar (A), while the attributive verbs are more frequent with prometer (P).117 Aspectual verbs are infrequent with both amenazar and prometer. Table 25. Infinitives with lexical amenazar (A) and prometer (B)
A Present % Past % P Present % Past %
action
changeof-state
attribution
aspect
communication
mental activity
total
30 73.17 72 75.79
5 12.20 3 3.16
0 1 1.05
0 2 2.11
4 9.76 14 14.74
2 4.88 3 3.16
41 100 95 100
70 71.43 160 76.56
1 1.02 6 2.87
6 6.12 8 3.83
2 2.04 4 1.91
13 13.27 22 10.53
6 6.12 9 4.31
98 100 212 100
A comparison of the infinitives following lexical amenazar and prometer with those of their modal counterparts yields interesting results. Table 26 indicates that communication and mental state verbs, which always need a human subject, do not show up in the modal constructions. Furthermore, modal amenazar and prometer tend to combine with different infinitives. Verbs expressing action and change of state have the lion’s share of the infinitives with modal amenazar. In the case of prometer, by contrast, the most extensive group is formed by attributive verbs. The percentage of action verbs is much lower in the modal constructions with prometer than in those with amenazar.118
Restrictions on the infinitive 125 Table 26. Infinitives with modal amenazar and prometer119
Amenazar % Prometer %
action
change-ofstate
attribution
aspectual
total
61 49.19 9 26.47
57 45.96 5 14.7
2 1.62 17 50
4 3.23 3 8.83
124 100 34 100
Attributive verbs express a static and categorizing view of events, while action verbs and change-of-state verbs inherently represent a dynamic process themselves. The fact that modal prometer prefers attributive verbs (50%), while amenazar barely combines with this verb type (1.62%) points to restrictions on prometer. The latter verb seems to experience a shift from strong agentivity in the lexical construction to necessarily low agentivity in the modal one. By contrast, amenazar still regularly combines with dynamic processes in modal constructions. This observation corresponds to the analysis of the subject commitment presented in the previous chapter (Section 2.1.) and will receive a diachronic underpinning in Chapter 6 (Section 2.). The fact that amenazar, unlike prometer, patterns predominantly with action and change-of-state verbs could be a basis for arguing that modal amenazar is more transparent than modal prometer. However, since one could also say that prometer is more transparent by avoiding the positive/negative polarization in the evaluation, it does not seem justified to unequivocally apply the concept of transparency to amenazar and prometer. Therefore I propose to deal with the restricted transparency of modal amenazar and prometer in terms of the different viewpoints construed by the verbs, namely the internal and external viewpoint (cf. Cornillie 2004: 17–20). In the following sections, I will argue that the attributive infinitives with prometer suggest that the speaker/conceptualizer adopts an external viewpoint on the event. An external viewpoint allows the speaker/conceptualizer to comment on an event from a vantage point outside the state of affairs.120 As for amenazar, the predominance of action and change-of-state verbs point to an internal viewpoint on the event expressed in the infinitival complement. An internal viewpoint is limited to the ongoing evolution presented in the utterance and does not license the speaker to view the state of affairs from another vantage point than the process itself. Since with
126 Partial transparency with modal prometer and amenazar amenazar a broader viewpoint is usually not available, the modal reading often displays an aspectual dimension. 1.3. Infinitives with amenazar Modal amenazar most often introduces dynamic activities or events. Interestingly, the majority of these infinitives follow a past form of amenazar, while the aspectual and attributive verbs are equally distributed over the past and the present tense, as shown in Table 27. It is clear that rather than absolute restrictions to a specific type of infinitive, non-lexical amenazar displays collocation preferences for certain infinitives. Table 27. Amenazar: infinitives in modal constructions121
present past
action
change-of-state
aspectual
attribution
24 37
15 42
2 2
1 1
The action verbs can be exemplified by the infinitive destruir ‘to destroy’ in (2). Despite the agentive structure of the infinitive, this construction does not yield an agentive reading. This may but need not necessarily be related to the non-agentive subject mercadería ‘products’ (cf. Section 2.). (2)
De ese modo, la mercadería extranjera subsidiada copó el 40% del mercado argentino y amenaza con destruir nuestra industria nacional. (Notic: Arg: Cronista: 696_laet) ‘In that way, the foreign subsidized products took 40 % of the Argentinian market and threaten to destroy our national industry.’
This amenazar construction conveys a further elaboration of an ongoing process. In (2), the event expressed in the infinitival complement is seen as the final stage of the ongoing deterioration of Argentina’s economic output: the destruction of its industry (destruir nuestra industria). The realization of the event is seen as an endpoint, which is conceptualized from the viewpoint of other stages of the ongoing evolution. Other action verbs that follow modal amenazar are listed in the notes.122 These examples can be given an analogous analysis. The second most frequent group of infinitives with modal amenazar are the change-of-state verbs, such as romperse ‘to break down’ in (3).
Restrictions on the infinitive 127
(3)
En la obscuridad blanquean las camisas de los trabajadores y en el aire hay algo tenso que amenaza romperse de un momento a otro. Nada llegó a romperse, sin embargo. (Rojas, Manuel. Hijo de ladrón) ‘In the darkness the workers’ shirts appear white and in the air there is something tense that threatens to break down from one moment to another. Nothing happened to break down, however.’
Note that these verbs frequently occur in a middle voice construction with se. Such a middle form reflects a major focus on the event itself to the detriment of the agentivity of the subject, due to the reflection of the subject entity in the se-form (cf. Maldonado 1999). These constructions most often imply a sudden change, which can be interpreted as the culminating stage of an evolution. The subject algo tenso ‘something tense’, in (3), for instance, has the potential of leading to several states of affairs: the tension is increasing and awaits a climax. Other infinitives expressing change of state are mentioned in the notes.123 Most aspectual verbs with amenazar found in the corpus express an evolution in which the subject entity is involved or affected.124 In (4), for example, the buildings (edificios) are already in a bad state and risk to deteriorate further.125 (4)
Nuestra tribu se estableció en el territorio que parecía más seguro: el parque México. Ahí estaríamos lejos de los edificios que amenazaban con seguir cayendo. Ahí reiniciamos la vida como pudimos. (Cadena, Agustín, Los Tiquiliches) ‘Our tribe settled in the territory that seemed most secure: the Mexico park. There we would be far away from the buildings that threatened to tumble down further. There we restarted our life as best as we could.’
Finally, there is the attributive relation between the subject and the complement of amenazar. Interestingly, the attributive complements in some way or another also express a change of state, as in (5).126 (5)
El proyecto, sin embargo, amenaza con ser más largo y difícil de lo que se esperaba. (Notic: España: ABC author: Serbeto Enrique) The project, however, threatens to be longer and more difficult than what we expected.’
128 Partial transparency with modal prometer and amenazar The project mentioned in (5) runs the risk of being too long and too difficult. The explicit mention of más largo implies that the view of a state of affairs yielded by amenazar is not a static one. In conclusion, there do not seem to be many restrictions on the infinitival complement of amenazar: action and change-of-state verbs are most frequent, while aspectual and attributive verbs form only a small proportion. The complements of amenazar have the following characteristic in common: in one way or another, they express some kind of change that constitutes an elaboration of the state of affairs from an internal viewpoint. This internal viewpoint of modal amenazar correlates with the lexical reading of amenazar. The threat is only realized when there is no other way out. In the same way the event denoted by the infinitives that follow modal amenazar is seen as a possible endpoint after a series of preceding steps. This view emphasizes the course of the action and is, hence, dynamic. In the next section, I will show that modal prometer does not construe such an internal, dynamic perspective. 1.4. Infinitives with prometer The four infinitive types observed for amenazar are also available with prometer, albeit in a different proportion, as shown in Table 28. Table 28. Prometer: infinitives in modal constructions127
Present Past
Action
change-ofstate
attribution
aspectual
8 1
5 2
12 5
0 3
I will start the discussion of prometer’s infinitives with the action verbs. Example (6) shows that despite the agentive lexical reading of the action verb generar ‘to generate/engender’ prometer yields a modal reading. Note that in order to avoid a lexical reading of prometer this action infinitive combines preferably with an inanimate subject.128 (6)
La Asobancaria planea revelar un estudio de los logros y retos de la política social en el país que en el contexto de la actual contienda política promete generar un debate interesante. (Notic: Col: Semana: 840)
Restrictions on the infinitive 129
‘The Asobancaria plans to unveil a report of the achievements and the challenges of the social policy of the country, which in the present political combat promises to generate an interesting debate.’ The infinitival complement generar un debate ‘to provoke a debate’ in (16) is the probable result of a study (un estudio) that is yet to come. Hence, it is not the result of an internal evolution or the elaboration of previous stages of a given evolution. The debate in (6) is conceptualized in relation to the context of a battle for the market position (el contexto de la actual contienda), and is, hence, viewed from an external viewpoint. Other examples of action verbs with prometer are listed in the notes.129 As far as the attributive infinitival complements are concerned, the majority of the examples of prometer lead to a categorization of the subject. In (7), ser llovedor ‘be rainy’ categorizes the invierno ‘winter’. Interestingly, given that the overall conclusion in (7) can only take place when winter is over, an external viewpoint is plausible. (7)
Este invierno promete ser llovedor. -Llovedor, patrón, promete ser este invierno. -Se ve mucha chicharra en el monte. Y es la señal. (Barrios, Eduardo. Gran señor y rajadiablos) ‘This winter promises to be rainy. Rainy, boss, this winter promises to be. One notices lots of crickets in the mountains. And this is the signal.’
In the corpus I have found nominal attributes such as cosaco ‘Cossack’ and categorizing adjectives such as explosivo ‘explosive’, formidable ‘fantastic’ or inmisericorde ‘merciless’. Other attributes are listed in the notes.130 One could argue that the change-of-state and aspectual verbs following prometer in (8) and (9) are counterexamples to the external viewpoint analysis presented in the preceding paragraphs. Indeed, these examples sketch an ongoing evolution that can also be viewed from an internal point of view. Yet, elements such as algo especial ‘something special’ in (8) or por lo menos ‘at least’ in (9) point to a clear presence of the speaker/conceptualizer. This leads me to conclude that with most change-of-state and aspectual verbs the speaker/conceptualizer analyzes and categorizes the events from an external point of view too.131 (8)
Más allá, casi colgando de un barranco, un atisbo de otra casa. Son cuatro paredes que se levantan y prometen transformarse en algo especial, un atelier quizás. (Karlik, Sara. Nocturno para errantes eternos)
130 Partial transparency with modal prometer and amenazar ‘Farther away, almost hanging in a gorge, a glimpse of another house. It is four walls that stand up and promise to become something special, a workroom perhaps.’ (9)
Los sones del arpa, las guitarras y el bandoneón daban el condimento especial a una alegría que prometía prolongarse por lo menos hasta el alba del día siguiente. (Dimas Aranda, Santiago Vida, ficción y cantos) ‘The harp sounds, the guitars and the bandoneon gave a special flavor to a [atmosphere of] joy that promised to be continued at least till dawn the next day.’
In sum, with prometer the attributive infinitives are preferred to those expressing an action or a change of state. Interestingly, many of the infinitives that combine with modal prometer, especially the attributive ones, have a strong categorizing function that allows the speaker/conceptualizer to comment from an external viewpoint. So far I have not focused on restrictions on the modal constructions. Hence the following section deals with the infinitives that cannot combine with modal prometer and amenazar. 1.5. Infinitives absent in the corpus In this section I examine whether the verbs under consideration can combine with a series of infinitives which Lamiroy (1995: 282–284, 1998: 292) uses in order to examine the gradual transparency of temporal and aspectual verbs.132 Among others, these infinitives include meteorological verbs, deontic verbs such as falloir ‘hacer falta/to need’, impersonal verbs like s’agir ‘tratarse/to be about’ and middle voiced infinitives such as se résoudre ‘resolverse/to be arranged’. 1.5.1. Impersonal and middle voiced infinitives Let us first look at meteorological verbs such as the (usually) negatively oriented llover ‘to rain’ or the positively oriented hacer bueno ‘to be nice weather’ in (10). (10)
a. Hoy promete hacer bueno. ‘The weather promises to be nice today.’
Restrictions on the infinitive 131
b. Hoy promete ? llover/ ??nevar. ‘Today it promises to rain/snow.’ [lit.] c. ? Amenaza con hacer bueno. ‘It threatens to be nice weather.’ [lit.] Modal prometer does not combine as well with llover ‘to rain’ and nevar ‘to snow’ in (10b) as with hacer bueno ‘to be nice weather’ in (10a). However, in a context of continuous drought, llover ‘to rain’ can be a perfect infinitival complement of modal prometer. For nevar ‘to snow’ it is harder to imagine a specific context in which a modal reading is possible. As shown in the previous chapter (Section 4.), it is also difficult to find a positively oriented complement with amenazar. In example (10c) the combination of hacer bueno and amenazar sounds odd. Only if good weather is undesired for the speaker, this combination can be felicitous. Since this is not very common, the example in (10c) has a certain ironic tone. Other impersonal infinitives which are the grammatical subject of the construction cannot combine with prometer or amenazar, such as tratarse ‘to be about’ in (11). (11)
*Promete/*amenaza con tratarse de un asunto muy serio. ‘It promises/threatens to be about a really serious topic.’
However, the infinitive of the impersonal tratarse ‘to be about’ is not prevented from following modal amenazar in the internet example (12). This is not caused by the difference between prometer and amenazar, however, but has to do with the subject assignment in the construction. Unlike the construction in (11), not the infinitive following modal amenazar and prometer, i.e. tratarse, but another infinitive, i.e. pensar, is the grammatical subject. (12)
En otros términos el problema de la responsabilidad médica y el seguro es un problema realmente complejo. Pensar en resolverlo con una simple norma que atribuye un derecho a ser indemnizado mediante el seguro obligatorio, amenaza con tratarse de una solución demasiado simple para un problema muy, pero muy complejo. (www.gacetajuridica.com.pe/legal_express/ legal32.pdf) ‘In other words the problem of medical responsibility and insurance is a really complex problem. Thinking to solve it with a simple norm that attributes the right to be compensated to obligatory insurance threatens to be too simple a solution of a very very complex problem.’
132 Partial transparency with modal prometer and amenazar The same construction type is possible with prometer and resolverse ‘to solve’ in (13). In this example, esta contienda ‘this battle’ is the subject of the middle voice construction with resolverse ‘to solve’. (13)
Las pugnas son de tal magnitud que los gobiernos no consiguen arbitrar entre las posiciones en pugna. Fiat versus Ford, Techint contra Usinor, son algunos de los protagonistas de esta contienda que promete resolverse ‘a sangre y fuego’. (http://www.po.org.ar/po/po646/para.htm) ‘The conflicts are so big that the governments do not manage to arbitrate between the conflicting positions. Fiat versus Ford, Techint against Usinor, these are some of the protagonists in this battle which promises to be solved harshly.’
1.5.2. Double infinitive constructions Amenazar and prometer do not combine with double infinitive constructions, such as with the impersonal infinitives hacer falta ‘to be needed’ and ser necesario ‘to be necessary’ in (14). (14)
a. [*Promete/*amenaza con] hacer falta llegar a un acuerdo. ‘It promises/threatens to be needed to reach an agreement.’ b. [*Promete/*amenaza con] ser necesario llegar a un acuerdo. ‘It promises/threatens to be necessary to reach an agreement.’
These infinitives yield a deontic reading, which does not pattern with the modal prediction reading. The same holds for deontic modal infinitives such as tener que ‘have to’ in (15). (15)
La prensa [*promete/*amenaza con] tener que publicar artículos interesantes. ‘The press [promises/threatens] to have to publish interesting articles.’
The deontic reading of tener que conflicts with the modal reading of prometer and amenazar. Infinitives that involve too strong an interplay of forces seem to be incompatible with the specific profile given by prometer and amenazar. In such constructions the deontic forces activate the original lexical patterns of the verbs. This shows that the attenuation of prometer/promise is not in its final stage yet (pace Langacker 2000).
Restrictions on the infinitive 133
Interestingly, with dynamic poder ‘can/to be able’ it depends on the second infinitive whether the prometer construction can receive a modal reading or not. In (16a), the verb prometer receives a lexical reading: the journalists promise to be able to respect the principles of the monarchy, while in (16b) a modal reading is more natural: here the speaker refers to the ability to work in a good atmosphere. This is also the case in (16c). (16)
a. La prensa promete poder respetar a los principios de la monarquía. ‘The press promises to be able to respect the principles of the monarchy.’ b. La prensa promete poder funcionar en un buen ambiente. ‘The press promises to be able to function in a good atmosphere.’ c. Así que lo bueno es que el arbitraje internacional sí presenta un modelo análogo que promete poder funcionar en el VMDR. (www.alfa-redi.org) So that the good thing is that the international arbitrage really presents an analogous model that promises to be able to function in the WMDR.’
On the Internet, I have found only a few examples of amenazar with dynamic poder, such as the one in (17). (17)
a. La baja productividad del arroz amenaza con no poder satisfacer las demandas de una población creciente. (www.canalsolidario.org/web/noticias/ver_noticia/asp?id_noticia= 2167) ‘The low rice productivity threatens not to be able to satisfy the needs of a growing population.’
Note that, unlike dynamic poder, the deontic reading of poder does not cooccur with modal amenazar and prometer. In a deontic poder construction the speaker or a third entity grant the subject the permission to realize an action. Just like deontic tener que, deontic poder is incompatible with the modal profile construed by amenazar and prometer. Against the background of the infelicitous combination with deontic verbs, I will try to determine which other verbs are precluded from combining with modal amenazar and prometer because of conflicting force structure.
134 Partial transparency with modal prometer and amenazar 1.5.3. Action and communication verbs In this section, I will discuss action and communication verbs which assign a special role to the addressee.133 The absence of overt participants with prometer is probably due to a strategy to omit interference with the agentive lexical semantics of prometer, in which the promisor and the promisee may be felt to stand in a closer relationship than the addresser and the addressee associated with amenazar. The elicited examples in (18), however, appear to offer evidence against this intuition. (18)
a. [El colega/el nuevo contacto/el nuevo libro] promete ayudarte a llegar lejos. ‘[The colleague/the new contact/the new book] promises to help you to further your career.’ b. [El colega/la falta de contactos/el nuevo libro] amenaza con dejarte atrás. ‘[The colleague/lack of contacts/the new book] threatens to leave you behind.’
In (18a) prometer can receive a modal reading and still combine with the agentive infinitive ayudarte ‘to help you’ provided that the subject is not human or humanoid. In (18a), prometer with a human subject such as el colega is understood as a lexical verb. By contrast, prometer readily yields a modal reading with a subject like el nuevo libro ‘the new book’, though this is less evident with el nuevo contacto ‘the new contact’, which can either be a human or a non-human entity with a human agent implied. The modal readings of amenazar in (18b) differ from those of prometer in (18a) in that they combine with agentive infinitives even when the subject is human. Although the subject colega ‘colleague’ can certainly trigger a lexical reading of amenazar, the modal reading does not pose any problem. The amenaza dejarte construction with the falta de contactos and the nuevo libro ‘new book’ by default yields a modal reading. All of this points to the importance of the subject, which will be dealt with in Section (2.). Some agentive infinitives which are compatible with lexical amenazar and prometer cannot combine at all with the non-lexical counterparts when they have a human subject: verbs inflicting harm to the interlocutor such as castigar ‘to punish’ and pegar ‘to hit’ are directly associated with the lexical reading of amenazar and, as shown before, action verbs such as ayudarte ‘to help you’ or hacerlo ‘to do it’ which involve a direct interlocutor cannot combine with modal prometer.
Restrictions on the subject 135
Finally, the combination of modal amenazar and prometer with infinitives expressing communication did not occur in the corpus. This is probably to do with the intuition that the preferred subjects of the modal constructions are non-human. However, communication verbs should not be excluded for amenazar. Example (19) shows that non-human subjects (with an agent implied) trigger different readings of prometer and amenazar. In (19a) the authors of el texto ‘the text’ have the intention of doing something with the text, which entails that it is not immediately understood modally. By contrast, the amenazar construction in (19b) overrides the potentially intentional reading of the subject and, hence, is spontaneously read modally. With a human subject and a communication infinitive, finite amenazar has a lexical reading, as shown in (19c). (19)
a. La versión final del texto promete [decir/repetir] lo mismo. ‘The final version of the text promises to [say/ repeat] the same.’ b. La versión final del texto amenaza con [decir/repetir] lo mismo. ‘The final version of the text threatens to [say/ repeat] the same.’ c. El profesor amenaza con [decir/repetir] lo mismo. ‘The professor threatens to [say/ repeat] the same.’
In sum, several infinitives do not readily co-occur with modal amenazar and prometer. Verbs depicting a strong deontic interplay of forces cannot follow modal amenazar and prometer. Some meteorological constructions sound odd in combination with these verbs, and impersonal subject infinitives are not felicitous either. However, middle constructions of tratarse and resolverse with a preverbal subject do combine with modal prometer and amenazar. Finally, amenazar can combine with a communication verb, provided the subject is not human. 2. Restrictions on the subject In this section I will analyze to what extent lexical patterns impose restrictions on the subject of modal prometer and amenazar. Concretely, I will determine which characteristics of the subject prevent amenazar and prometer from being fully transparent. This section is organized as follows. I first present the working hypothesis and comment on the methodology (2.1.). I then give a general overview of the subjects in the lexical constructions (2.2.). After this general presentation, the subjects of modal amenazar (2.3.) and those of prometer (2.4.) are discussed.
136 Partial transparency with modal prometer and amenazar 2.1. Working hypothesis and methodology It is generally assumed that a verb is transparent when the grammatical subject is constrained by the infinitive (cf. Gaatone 1995). Although the grammatical subject of modal amenazar and prometer agrees thematically with the infinitive, the finite verb forms still have some kind of thematic relation with the subject. This relation is not agentive, however. As shown in the previous chapter, lexical prometer prototypically has a human subject who can see to the intended accomplishment of the infinitive, while amenazar does not need such an agentive subject. Although both verbs express a kind of intentional commitment in their lexical readings, the intention is stronger with prometer than with amenazar. The working hypothesis of this section is that certain subject types which are frequent with lexical prometer and amenazar also appear with their modal counterparts. If this hypothesis is true, it confirms that both the formal claim regarding the lack of thematic role assignment and Langacker’s claim concerning full attenuation are too strong. In this section I also test whether the subject of modal prometer is submitted to more restrictions than that of modal amenazar. The semantic description of the subject carried out in the next section is based on the classification of subject and oblique entity proposed by Delbecque (2003: 383) for passives, which differentiates between 9 semantic categories, which I reproduce in Table 29. Table 29. Classification of the subject 1. Individual or person 2. Concrete object 3. Event, action, process 4. State of affairs 5. Concept, notion, idea 6. Place, spatial entity 7. Semiotic entity 8. Collective entity 9. State of mind
O. Paz, presidente ‘president’ cedro ‘cedar’, dinero ‘money’ crimen ‘crime’, reparto ‘distribution’, guerra ‘war’, intento ‘intention’ bienestar ‘welfare’, hecho ‘fact’, situación ‘situation’, coyuntura ‘economic climate’ tema ‘theme’, condición ‘condition’ Buenos Aires, Occidente ‘Occident’, celda ‘cell’, isla ‘island’, puerto ‘port’ texto ‘text’, libro ‘book’, debate ‘debate’, código ‘code’, película ‘film’ Institution ‘institution’, grupo ‘group’, comunidad ‘community’, Nación ‘Nation’ emoción ‘emotion’, nostalgia ‘nostalgia’
Restrictions on the subject 137
This classification was applied to the corpus data of amenazar and prometer. For the sake of completeness, I also added the “zero” subject type, which corresponds to constructions which lack an explicit subject. The next section then presents a general overview of the subjects that combine with all uses of amenazar and prometer. 2.2. The subjects of lexical amenazar and prometer The figures presented in Table 30 confirm that the prototypical subject of lexical prometer is a person or an animate individual (54%). Since the “zero” subjects (31.27%) mostly refer to individuals, the amount of animate individuals is actually still higher. The percentage of human subjects with amenazar is lower. Almost 30% of the subjects of amenazar are animate, and 24% of the subjects remain implicit. Moreover, unlike prometer, concrete objects, collective nouns, states of affairs and processes together constitute about 40% of all subjects with lexical amenazar. Table 30. Semantic classification of the subject135 Subjects
amenazar
%
prometer
%
Individual Object Process State of affairs Notion Location Semiotic Collective Mind
218 84 68 81
29.38 11.32 9.16 10.92
534 27 25 11
54.21 2.74 2.54 1.12
10 19 14 6 10
1.35 2.56 1.89 0.81 1.35
6 6 15 16 1
0.61 0.61 1.52 1.62 0.10
Zero134
181
24.39
308
31.27
741
100
985
100
In the subsections below, I will connect this information to the analysis of the subjects of the modal constructions. The previous section has shown that agentive patterns in the subject can be activated by agentive infinitives, so I will deal with the subjects in combination with the types of infinitive.
138 Partial transparency with modal prometer and amenazar Recall that I have subdivided the infinitival complements of prometer and amenazar into the four semantic groups used in section (2.): action verbs (with an agent), aspectual verbs, attributive verbs and change-of-state verbs (without an agent). 2.3. Subjects of modal amenazar All types of subject are possible with amenazar, as shown in Table 31. Although individuals and collective entities are not frequent at all, they are by no means excluded. Yet, the table most clearly confirms that processes, states of affairs and objects occupy the most prominent position among the subjects of modal amenazar. Table 31 presents the subject types in the rows and the infinitive types in the first four columns. This ordering allows me to relate the type of subject to the type of infinitive. The fifth column gives the total of the subjects, while the sixth one gives the percentages. Table 31. Subjects of modal amenazar Subject
action
Individual Object Process State of affairs Notion Location Semiotic Collective Mind
3 22 18 7
Zero
2
change of state
16 14 17
attribution
1 1
2 1 3 3
aspect
Total
1 1
3 39 34 25
2.42 31.45 27.42 20.16
3 4 4 5 3
2.42 3.23 3.23 4.03 2.42
4
3.23
124
100
1 3 2 1 2
1 1 1
Restrictions on the subject 139
2.3.1. Non-human subjects In the modal amenazar constructions, human individuals are rather infrequent subjects, while a subject such as proyecto ‘project’ expressing a process or an event is more prototypical. (20)
El proyecto, sin embargo, amenaza con ser más largo y difícil de lo que se esperaba. (Notic: España: ABC author: Serbeto Enrique) The project, however, threatens to be longer and more difficult than what we expected.’
The figures presented in Table 31 bring to the fore that the action or change-of-state verbs most frequently have subjects expressing a concrete object, a process or a state of affairs. Such subjects do not activate agentivity in the construction. A typical example of a subject denoting a “concrete object” with an infinitival complement expressing action is arbustos ‘bushes’ in (21): in this example the bushes cannot be viewed as endowed with intentions. (21)
Los neumáticos se enterraban en el fango, pero Santana se las agenciaba para sacarlas y seguir. Los arbustos espinosos amenazaban con detener su guagua, pero que va: Santana era un chofer de mucha experiencia. (de la Paz, Jorge Luis. La guagua) ‘The tires sank in the mud, but Santana managed to pull them out and to go on. The prickly bushes threatened to stop his car, but did not as so. Santana was a very experienced driver.’
In (22), the subject un cambio de gobierno ‘a governmental change’ denotes a process and combines with an action verb romper ‘to break’. Although the subject here can be seen as the result of human action, this combination does not block a modal reading. (22)
El periodo más difícil se inició en 1974, cuando Turquía intervino para apoyar a los turcochipriotas después de que un cambio de gobierno amenazara con romper el equilibrio constitucional a favor de aquellos grecochipriotas que deseaban la unión. (Encicl: Organización de las Naciones Unidas) ‘The most difficult period started in 1974, when Turkey intervened to support the Turkish Cypriots after a governmental change threatened to alter the constitutional balance in favor of those Greek Cypriots who wanted the union [with Greece].’
140 Partial transparency with modal prometer and amenazar State of affairs subjects such as escasez ‘scarcity’ and interrupciones ‘interruptions’ in (23) readily combine with amenazar. In (23), the lack of an agentive pattern in the subject cancels any potential agentive relation between the subject and the infinitival complement acabar con ‘to make an end to’. (23)
Eso, unido a la escasez crónica del fluido eléctrico y a las esporádicas interrupciones en el suministro de agua, amenazan con acabar con la paciencia de los ciudadanos. (Notic: Col: Semana: 820) ‘This, together with the chronic scarcity of electric power and with the sporadic interruptions of the water supply, threaten to consume the patience of the citizens.’
2.3.2. Human subjects Let us now look at the human subjects and subjects which somehow involve the participation of a human agent. The group of human and humanoid subjects ranges from individuals to less individual entities. I first discuss the subjects denoting individual persons. In (24), for example, the subject niños indisciplinados ‘undisciplined children’ does not include intentional commitment to the change-of-state complement convertirse en delincuentes ‘becoming delinquents’, which is not agentive at all. (24)
El director tiende a ver en aquéllos a un grupo de niños indisciplinados que amenazan convertirse en delincuentes juveniles. (Notic: Arg: Cronista: 645_LA) ‘The director tends to see them as a group of undisciplined children that threaten to become young delinquents.’
However, it appears that intentional commitment is not crucial for determining whether the reading of amenazar is to be considered lexical or modal. In (25), for example, aquellos ‘those people’ refer to the mighty, who do not physically threaten to annihilate the world, but only do so in the subjective interpretation of the speaker. (25)
... tenemos frente a ese colosal imperio y la infinita fuerza que poseen aquellos que están haciendo retroceder al mundo y amenazan con llevarlo al exterminio, cuyas ideas, cuyos conceptos y cuyas mentiras hay que destruir. (Fidel Castro. 03/07/98)
Restrictions on the subject 141
‘... we have in front of us a gigantic empire and the infinite force in the hand of those who are deteriorating the world and threaten to bring it to an end, whose ideas, whose concepts and whose lies should be destroyed.’ Moreover, although the powerful certainly can annihilate the world, one does not expect that they have any interest in doing so. Hence, the speaker and the hearer can subjectively view this apocalyptic approach. The same happens in (26), where the speaker or the central figure is part of a human mass. Although it is formed by human beings, the mass itself is not an intentional entity undertaking any action. (26)
Me dispuse a satisfacer su pedido, mientras luchaba -con mayor o menor éxito- contra la marea humana que amenazaba aplastarme (mi finalidad no era de ningún modo morir por asfixia, como en los terribles ómnibus alemanes de exterminio). (González Real, Osvaldo. Anticipación y reflexión) ‘I was ready to meet his request, while I was struggling – more or less successfully – against the human sea that threatened to crush me (my objective was by no means to die from suffocation, as in the terrible German extermination coaches).’
Finally, there are also collective subjects with a humanoid component. The instituciones ‘institutions’ in (27), for example, imply an agent that runs the complex management of these institutions and takes the decisions. Although these agents can display intentional commitment to the accomplishment of the process expressed by the infinitive, in (27), they do not automatically trigger a lexical reading of amenazar. (27)
El retraso agita más todavía el pugilato político entre las diversas instituciones que componen el Consorcio Madrid Capital Cultural, y que amenazan con quebrar la celebración. Los ciudadanos serán, como siempre, al final, los únicos perjudicados. (Bravo, Julio. Abc) ‘The delay increases even more the political fight between the different institutions that form the Madrid Capital Cultural Consortium, and that threaten to crack the celebration. In the end the citizens will, like always, be the only victims.’
There are two possible readings of amenazan con quebrar in (27). On the one hand, the subject of the amenazar construction can be understood as
142 Partial transparency with modal prometer and amenazar uttering a real threat. In this case, the delay leads those responsible for the institutions to seriously consider cancelling the celebration. The combination of the collective entity instituciones and the highly agentive action verb quebrar la celebración ‘to cancel the celebration’ makes this reading possible. On the other hand, focus on the delay and the subsequent political problems entail that the behavior of the different institutions will disturb the celebration. This implies that the celebration itself will take place. In other words, a modal reading is also possible. Note that this double reading might be favored by the specific position of amenazar in the relative clause which is preceded by the conjunction y ‘and’. Indeed, constructions with human or humanoid subjects, as in (26) and (27), do not display the tendency to put the modal verb in the subordinate clause. This is illustrated by the agentive main clause examples in (28). (28)
a. Las diversas instituciones que componen el Consorcio Madrid Capital Cultural amenazan con quebrar la celebración. ‘The different institutions that form the Madrid Capital Cultural Consortium threaten to cancel the celebration.’ b. El retraso y el pugilato político entre las diversas instituciones que componen el Consorcio Madrid Capital Cultural amenazan con quebrar la celebración. ‘The delay and the political fight between the different institutions that form the Madrid Capital Cultural consortium threaten to crack the celebration.’
In (28a), the finite form of amenazar has been moved to the main clause, which almost imposes a lexical reading on the institutions’ threat. By contrast, in the main clause in (28b) the non-human subjects el retraso ‘the delay’ and el pugilato político ‘the political fight’ do not block a modal reading.136 2.4. Subjects of modal prometer In this section I will analyze whether modal prometer combines less often with agentive human subjects than amenazar. At first sight, this is certainly the case: the corpus data in Table 32 show that prometer, in contrast with modal amenazar, does not have subjects that denote individuals (nor locations), while collective subjects are possible, though not frequent.
Restrictions on the subject 143 Table 32. Subjects of modal prometer Subject
Individual Object Process State of affairs Notion Location Semiotic Collective Mind
action
change of state
attribution
aspect
1 4
3
4 6 2
1
4
1
2 1 1
None
1 1 1 1
total
%
0 9 10 2
0.00 26.47 29.41 5.88
7 0 2 2 1
20.59 0.00 5.88 5.88 2.94
1
2.94 100
Table 32 indicates that the bulk of the subjects with prometer refer to processes, objects and notions (day, month and season). It also shows that action verbs prefer processes to individuals, objects or collective nouns. The attributive predicates, which do not involve any action at all, combine with most subject types. This distribution emphasizes that the combination with specific infinitives downplays the potential agentivity of the subject. These corpus data seem to corroborate the hypothesis that agentive subjects do not favor a modal reading of prometer. The following sections detail thes distribution of subjects found with modal prometer. 2.4.1. Collective nouns As noted before, in my data human individuals do not figure among the subjects of modal prometer. By contrast, the corpus contains a couple of collective nouns in which human beings are involved such as operadores de larga distancia ‘long distance operators’. In (29), the operators are expected to become the biggest advertising investors on the market, they do not make any promise themselves. (29)
Los operadores de larga distancia prometen convertirse en los mayores inversionistas publicitarios pues el ejemplo de otros paí-
144 Partial transparency with modal prometer and amenazar ses contempla una inversión entre 15 y 30 por ciento. (Notic: Col: Semana: 840) ‘The long distance operators promise to become the biggest advertising investors because the example of other countries points to an inversion between 15 and 30 %.’ In (30), the subject ronda ‘round/turn’ refers to a kind of temporal and physical organization with a specific aim, e.g. delivering services or security; ronda can thus be understood as a group of persons and may convey intentional commitment to avoiding major problems. Nevertheless, no agentivity emanates from the rondas in the present example, prometer has a modal reading here. (30)
En enero de 1994, 76 familias regresaron a Purus. Aquí han reconstruido su comunidad y se han organizado en rondas que prometen mantener alejados a los terroristas de su pueblo. (Notic: Perú: Caretas: 1433) ‘In January 1994, 76 families returned to Purus. Here they reconstructed their community and they have organized themselves in rounds which promise to keep at a distance the terrorists of their village.’
Other subjects are more clearly the product of human beings. El remake in (31), for example, obviously needs a producer, who is mentioned here: Umberto Polar. The question thus needs to be considered whether the act of recuperar la armonía ‘to recover the harmony’ reflects an intention of the producer or a mere by-product of the remake perceived by the speaker. Given the descriptive subclause at the end of the example, the modal reading seems most plausible. However, more information about Umberto Polar’s plans could alter this and lead to a lexical “promise” reading, instead. (31)
Se grabó en Buenos Aires pero se urdió desde Lima. El remake producido por Umberto Polar para el 5, promete recuperar la armonía de ese barrio protagónico que fue el primero en figurar en un melodrama. (Notic: Perú: Caretas: 1431) ‘It was recorded in Buenos Aires but it was planned from Lima. The remake produced by Umberto Polar for the 5, promises to recover the harmony of this prominent barrio which was the first to be used in a melodrama.’
Restrictions on the subject 145
In contrast with the lack of knowledge about the agent’s plans in (31), in (32) it is explicitly mentioned that the infinitival process will probably be realized a través de los últimos avances tecnológicos ‘through the latest technological advances’. Moreover, the subject cinetismo refers to an artistic movement, which is naturally constituted by producers. Since it is hard to pin down the responsibility for this movement to a couple of agents, prometer does not directly entail an agentive lexical reading. (32)
Otros artistas de nuestro continente muestran que el cinetismo no ha agotado aún todas sus posibilidades y promete a través de los últimos avances tecnológicos llevar todavía más lejos sus originales propuestas visuales. (González Real, Osvaldo. Anticipación y reflexión) ‘Other artists from our continent show that the “cinetismo” movement has not yet exploited all its possibilities and promises to further develop its original visual proposals through the latest technological advances.’
This example shows that knowledge about the plans of the artistic movement does not as such impose a lexical reading. In other words, the combination of responsibility and agentivity appears to be the most central element of the lexical reading. It is important to stress that the two dimensions have to be combined, since agentivity itself is not sufficient, as shown in (33). In the example below, the agentive infinitive repetir ‘to repeat’ is the subject. Interestingly, the modal reading is still possible since there is no intention attached to the act of repeating the allusions. (33)
Al fin, risueño a su turno, interrogó: - ¿Alguna picardía murmuraba esta mala lengua? Como repetir las alusiones prometiera remover el escándalo, Felipe Toledo intervino: - Basta, que no viene. (Barrios, Eduardo. Gran señor y rajadiablos) ‘Finally, smiling at his turn, he interrogated: - some slyness mumbled this gossip? As if repeating these allusions promised to take away the row, Felipe Toledo intervened: - It is alright, he will not come.’
The discussion of the above-mentioned examples has shown a complex interplay of different elements that co-determine whether prometer yields a modal or a lexical reading. For the humanoid subjects, intention, agentivity and responsibility are the most important ones. This analysis is in line with the observation that the shift from lexical to modal readings in promise-like
146 Partial transparency with modal prometer and amenazar verbs is blocked when the verb combines with a humanoid subject that is directly associated with agentivity and responsibility. Intention has proved to be less important than the former two factors. 2.4.2. Human subjects According to Vliegen (2006), animate subjects cannot combine with modal promise/beloven/versprechen. Although such subjects are absent in my data, I argue for a more gradual shift from lexical to modal readings in Spanish. Let us first look at some evidence from Germanic languages. Vliegen (2006) assumes that an “epistemic” reading of promise/beloven/versprechen is blocked when its subject is “capable of intentional action”. Even if the subject is not a concrete agent, strong intention, Vliegen argues, would make a modal reading of prometer impossible. In this context, Vliegen mentions that in German and Dutch a prometer construction invoking an experiencer is “practically excluded in the epistemic reading”. I reproduce his examples in (34). (34)
a. Eva verspricht den Tag zu geniessen. (from Vliegen 2006) b. Eva belooft van de dag te genieten. (from Vliegen 2006) ‘Eva promises to enjoy the day.’
In Spanish, the subject of modal prometer does not seem to adhere to the same restrictions, however. In (35), an animate subject, i.e. the girl Ana, with aprovechar ‘to enjoy/exploit’, i.e. an infinitive equivalent to the Dutch and German ones in (34), does not block a modal reading. The girl Ana will probably go to Buenos Aires to have a tango course. From the context, the reader can deduce that the event expressed in the infinitive is likely to be realized in one way or another. (35)
Alguien le pregunta a Ana si baila tango. Mientras contesta que “solo un poco”, se va poniendo de pie: baila. Baila con gracia, erguida, con estampa, atenta, con el codo derecho un poco levantado y tratando de no mover la cabeza como en la salsa. Promete aprovechar la oportunidad que brinda Buenos Aires de aprender a bailar tango. (http://protagonistas.terra.com.co/protagonistas2/dia_a_dia/22-072003/nota98653.html) ‘Someone asks Ana if she dances the tango. While she answers “only a little”, she stands up: she can dance. She dances elegantly,
Restrictions on the subject 147
upright, with rhythm, attentively, with the right elbow a little bit in the air and trying not to move the head as in salsa. She promises to make the most of the opportunity that Buenos Aires gives [her] to learn to dance the tango.’ Yet, in (35), the volition or agentivity of the human individual could simply be less prominent due to the fact that the subject Ana is not explicitly mentioned in the utterance (Gergel p.c.). This means that lexically elaborate animate subjects are not expected to appear in modal readings of prometer with a human subject and an agentive infinitive. When the explicit subject Ana is added, the modal reading indeed becomes less acceptable. The intention to learn the tango strengthens the tendency towards a lexical reading. (36)
Baila con gracia, erguida, con estampa, atenta, con el codo derecho un poco levantado y tratando de no mover la cabeza como en la salsa. Ana promete aprovechar la oportunidad que brinda Buenos Aires de aprender a bailar tango. ‘She dances elegantly, upright, with rhythm, attentive, with the right elbow a little bit in the air and trying not to move the head as in salsa. Ana promises to make the most of the opportunity that Buenos Aires gives [her] to learn dancing tango.’
Interestingly, the infinitival complement aprovechar la oportunidad ‘to make the most of the opportunity’ can only be associated with a human being. Nevertheless, without an explicit subject, prometer can have a modal reading when it combines with this complement. This is rather surprising, since the modal reading of prometer is usually not available in constructions with a human subject and an agentive infinitive, as shown with salir enseguida ‘to leave/come out immediately’ in (37). In (37a), there is no indication that a modal reading of the person’s leaving might correspond to the intention of the speaker, while in (37b) the prometer construction referring to the release of the book can only be modal. (37)
a. [Promete/amenzaza con] salir enseguida. ‘(s)he promises/threatens to leave immediately.’ b. El libro promete salir enseguida. ‘The book promises to come out immediately.’
As shown before, prometer can most easily background the agentivity of the human subject when the infinitive does not depict an action or a change
148 Partial transparency with modal prometer and amenazar of state. This is clearly the case in sentences like (38), in which prometer combines with a copular complement. (38)
Emilio Antonio es un talento musical que promete ser un gran artista en el futuro. ‘Emilio Antonio is a musical talent who promises to be a great artist in the future.’
Emilio Antonio does not interfere in the modal prometer construction, although logically speaking, he certainly could. Although Emilio Antonio will no doubt have the intention of becoming a great artist, in these kind of utterances, the future does not lead to a lexical reading. An infinitive like debatir ‘to debate’, e.g. (39), by contrast, does not readily trigger a modal reading. (39)
En sesión adelantada por el 9 de Julio, los ediles prometen debatir esta tarde un dictamen elaborado por las comisiones de Planeamiento Urbano, Ecología y Vivienda. (Notic: Arg: Cronista: 669_ELCO) ‘In the advanced session instead of July 9th, the councillors promise to debate this afternoon a proposal worked out by the Urban Planning, Ecology and Housing commissions.’
The construction with los ediles ‘the councillors’ is generally considered agentive and, hence, does not readily lead to a modal reading. However, the temporal marker en sesión adelantada por el 9 de Julio ‘in the advanced session instead of July 9th’ establishes a positive horizon of expectations regarding the accomplishment of the infinitive which paves the way for a modal reading. In such a construction, the inference of probable realization of the event expressed can background the impact of the animate subject. This is the basis for a future modal reading. 3. Conclusions The detailed corpus-based analysis of the infinitive and the subject has shown that amenazar and prometer are not completely transparent. First, the type of infinitive is linked with a specific viewpoint expressed by modal amenazar and prometer. The frequent change-of-state and action infinitives with amenazar correlate with an internal viewpoint on the state of affairs, while the frequent association of modal prometer with attributive
Conclusions 149
infinitives points to an external point of view. Furthermore, a small group of deontic and communication verbs do not co-occur with modal amenazar and prometer. In general, the difference between lexical and modal prometer is larger than between lexical and modal amenazar. The strong intentions that lexical prometer conveys make the possibility of modal constructions smaller; this is also reflected in the small number of modal prometer in the corpus. Recall that only 9.9% of the constructions with infinitive examples have such a reading. By contrast, since lexical amenazar lacks strong intentions, this reading is not so different from a modal reading. This results in a higher frequency of modal readings: up to 47.5% of the amenazar constructions with an infinitive have a modal reading. Moreover, the higher frequency also leads to a greater variation of infinitival complements. The analysis of the subjects points to a great variety of non-human and human subjects with both amenazar and prometer. Here too, amenazar turns out to be less influenced by agentive patterns. Illustrative of this is the possibility of having a modal reading with a human subject and an agentive infinitive. One can thus conclude that, as far as the attenuation of agentivity is concerned, amenazar is more transparent than prometer. The case of prometer/promise and amenazar/threaten confirms Langacker’s (2000: 358) idea that full transparency and full non-participation are “the endpoint in a continuous spectrum of possibilities”. However, prometer and amenazar have not reached this endpoint yet. In the next chapter, I will go into more detail about the shift from lexical to modal readings and present various stages of the process of subjectification with amenazar and prometer.
Chapter 6 Subjectification of prometer and amenazar
The synchronic analyses presented in the previous chapters have shown that the two readings of amenazar and prometer are “no accidental homonyms” (Verhagen 1996: 801). This chapter offers an explanation for the shift between the two readings in terms of subjectification. On the one hand, I will review previous metaphor and metonymy-based analyses of promise/beloven and threaten/dreigen and apply them to amenazar and prometer (Sections 1 and 2). On the other hand, I will check to what extent Langacker’s synchronic subjectification is in line with the results of the diachronic analysis. I will also examine whether it is a viable account of the diachronic evolution itself (Section 3). 1. A diachronic approach to promise and threaten This section lays the basis for the description of the diachronic corpus data of Spanish prometer and amenazar conducted in Section (2.). First I present Traugott’s (1997) description of the historical evolution of promise and threaten (1.1.). Next, I argue that it is hard to explain the shift from lexical to modal readings in terms of metaphorical extension (1.2), and propose that metonymic change is a more appropriate explanation (1.3.). 1.1. An overview Traugott (1997) differentiates between three different stages for the development of the epistemic meaning of promise and threaten. These three stages, however, are not directly associated with the three general tendencies (T) of pragmatic strengthening, viz., T I: shift from an external to an internal situation, T II: from an external/internal to textual and metalinguistic situation, T III: subjectification. I first present the different stages and then link them up with the first and the third tendency. Stage I of promise is situated in the Middle English period, when the verb enters the English language as a borrowing from French. From its first use in the English language, promise has been a performative and commissive verb which expresses commitment to the interlocutor (cf. Traugott
A diachronic approach
151
1997: 186). Moreover, the subject is necessarily animate, i.e. (s)he is able to perform an illocutionary speech act. The first stage of threaten goes back to Old English. Just as in present-day English this verb does not necessarily convey commitment or performative speech acts. The second stage, Traugott argues, starts in the 16th century, when both verbs allow for a “non-intentional, non-commissive, epistemic” reading in their transitive use with a nominal complement. Traugott (1997: 187–188) mentions that the verbs in (1) mean ‘to predict’. (1)
a. Yf any man all this can gett, shall he haue the greatest felicitie, shall he fynde her [felicity] in these [honor, glory, pleasure] that we haue shewed you, promise more than they giue?” (1593 Queen Elizabeth, transl. of Boethius, 57 [DHCE])137 b. ... the title of this Paper promising some Experiments about the Production of Electricity, I must not omit to recite.... [capitals original] (1675-1676 Boyle. Electricity and Magnetism 20–21 [DHCE]) c. As the morning promised a fair day we set out, but the storm coming up again we were obliged to come to. (1784 Muhl [ARCHER]) d. The croupous symptoms had appeared early and gone on rapidly so that suffocation was threatened. (1864 Spencer [ARCHER])138
Traugott (1997: 198) claims that the change to a ‘predict’ reading is motivated by the fact that the “pragmatic inference of futurity implied by promise and threaten at Stage I came to be semanticized in contexts where no proposition is syntactically expressed, i.e. in contexts with direct objects”. Although this is certainly true for promise because of the frequent combination with nominal complements, Traugott does not support this claim for threaten with corpus evidence. The fact that promise combines frequently with indefinite complements at Stage II (or with no complement at all) leads to an underspecification of the commitment expressed by these verbs and “presumably allowed for the demotion of the semantics of commitment” (Traugott 1997: 198). Importantly, Traugott points out that other verbs of the commissive class, like guarantee and pledge, which only preselect definite objects, have not made the shift to an epistemic reading of the future event. Furthermore, the inanimate subjects in (1) point to an extension towards a more abstract reading, although they still ‘control’ the finite predicate. The control relation between the subject and the finite verb leads Traugott to conclude that Stage II is an intermediate step between Stage I, with full control, and Stage
152
Subjectification of prometer and amenazar
III, without control. Stage III, beginning in the 18th century, is characterized by the “nonintentional and epistemic” meanings of promise and threaten with infinitival complements. The speaker views the proposition as likely, and evaluates it positively or negatively. Traugott (1997: 188) mentions that earlier examples of promise contain more often inchoative be-complements (2a) than “stative copula and activity verbs” (2b), the latter being very frequent in present-day English. Note that the inchoative event reading may stem from the specific context in which the construction is used; example (2a), for instance, describes a building under construction. Traugott (1997: 189) states that “it is not clear that [threaten] was first limited to inchoative becomplements, although the early examples involve change-of-state verbs”, such as in (2c). (2)
a. The Capitol promised to be a large and handsome building, judging from the part about two thirds already above ground. (1795 Twin [ARCHER]) b. The route promised to throw a variety of weather at us. (1992 Flying [HECTOR]) c. I am sometimes frightened with the dangers that threaten to diminish it [my estate] (1780 Mirror No81 [OED])
According to Traugott (1997: 198), Stage III is motivated by the general process whereby “epistemic meanings gradually shift from more “objective” possibility based in general beliefs and attitudes to more “subjective” possibility in the individual speaker’s belief or attitude”. Thus, in Stage III the evaluation of the likelihood of the state of affairs is directly related to the speaker. Although the differentiation between the three stages is an inspiring model to follow, Traugott’s (1997) analysis also has certain drawbacks: (i) promise and threaten are basically treated as verbs with the same historical evolution. It is striking that Traugott gives far fewer examples of threaten than of promise. In my own analysis I will clearly differentiate between amenazar and prometer and verify whether the stages of their evolution differ (cf. Verhagen 2000). (ii) The historical changes are temporally separated from each other over a long period of time. Yet, there is no reason why, once the verb allows for a modal reading of promise/threaten + nominal complement (Stage II), an infinitival construction (Stage III) should only show up three centuries later. My account of Spanish will show that these forms show up in the same period. (iii) Traugott (1988) extensively discusses the cognitive notions “metaphor” and “metonymy” in relation to
A diachronic approach
153
the semantic-pragmatic tendencies (I, II, and III), but Traugott (1997) does not directly apply them to the subjectification of promise and threaten. In my analysis of amenazar and prometer I will use slightly modified notions of metaphor and metonymy. In the following subsection, I will first examine in further detail whether the three historical stages presented for promise and threaten correspond to the tendencies which Traugott considers as manifestations of a general principle of grammaticalization and subjectification. I will show that the historical stages mainly correspond to the first two tendencies and that it is less easy to pin down the third tendency to one specific stage. I argue that metaphorical change is involved in the extension to nominal complements (first tendency) (1.2.), while metonymic change seems to account well for the other stages (1.3.). After this theoretical discussion I will analyze the rise of modal readings in amenazar and prometer. 1.2. Metaphorical extension Below I show that the first semantic-pragmatic tendency, i.e. the shift from an external to an internal situation, applies to both promise and threaten, but cannot readily be accounted for in purely metaphorical terms. 1.2.1. Principles of metaphorization In their lexical readings, promise and threaten express a concrete commitment to a thing or an action (complement), and connect the possible consequences of the realization of the complement with an interlocutor or/and other participants. In their modal readings, by contrast, promise and threaten have undergone a shift to a cognitive, perceptual and evaluative domain. In the grammaticalization literature, this shift from external to internal situations has largely been claimed to be metaphorical in nature. Metaphorization occurs when a thing or process present in the speakerhearer interaction (the source domain)139 is specified in terms of another thing or process which is not present in the context (the target domain) (cf. Traugott 1988: 413). For example, a metaphor such as life is a journey connects two different domains by introducing the notion of journey (and the different phases it implies) to characterize the concept of life discussed in the conversation. Now, it is the question to what extent this mapping also applies to the shift in meaning of verbs such as promise and threaten.
154
Subjectification of prometer and amenazar
1.2.2. Forces and counterforces The alternation of two readings of a verb often invites one to assume direct mapping between the domain expressed by the first reading to that of the second reading. In this way, Sweetser (1990) deals with the English deontic and epistemic modal verbs in terms of metaphor.140 (3)
a. John may go. (deontic reading) b. John may be there. (epistemic reading)
According to Sweetser, in the deontic reading of may (3a) there is a sociophysical force of the imposer on the imposee: the imposer cancels the barrier the imposee could face. Sweetser (1990) argues that through a metaphorical shift this deontic force constellation leads to the epistemic modal reading of may in (3b). In (3a), Sweetser (1990) interprets the utterance as “John is not barred from going”, whereas in (3b), it is the speaker who is not barred by his/her premises “from the conclusion that John is there” (cf. Chapter 8 for a more detailed discussion of Sweetser’s (1990) metaphorical model). Such a metaphorical shift is not directly applicable to promise and threaten. When threaten has an interlocutor, the subject communicates to the addressee/participant that the threat will materialize if (s)he crosses a virtual barrier. When there is no interlocutor involved, there is no barrier, however. By contrast, promise always implies an interlocutor/participant but does not impose real barriers on him/her.141 If one assumes that the force relations in lexical promise and threaten are metaphorically reinterpreted as force directed to the proposition in the modal reading, one cannot rely on the subject and the interlocutor of promise and threaten. Moreover, the specific force structure in the lexical reading consists of more than barriers alone. As said in Chapter 4 (Section 2), the subject of promise usually promises to do something for another person or entity, while with threaten, the subject only reluctantly proceeds to act towards another person or entity if the conditions for avoiding the threat are not fulfilled. The different subject intentions and the possible counteracting forces also play a role in promise and threaten (cf. Pelyvás (1996) for the English modals). With promise, the possible counteracting force (or combination of forces) is not located in the promisee, who is the beneficiary, but can only correspond to the promisor. The existence of counterforce is also hinted at by Searle (1969: 58–63), when he mentions that an explicit promise always implies a special effort. As for the lexical readings of threaten, by contrast, the interlocutor/participant is the default counteracting force
A diachronic approach
155
which attempts to resist or prevent the threat from being accomplished. This interplay of forces does not yield a view of the force structure that can satisfactorily serve as a basis for metaphorical extension. In sum, the force structure is far more complex than it looks at first sight. The metaphorical shift in the alternation of the lexical and non-lexical readings of promise and threaten in present-day English may be all too simple to account for the diachronic evolution. It is at least striking that Traugott’s (1997) diachronic analysis of promise and threaten does not refer to metaphor. 1.2.3. Personification as a restricted metaphor Verhagen (2000: 204–205) states that if metaphor plays a causal role in the change of promise and threaten, it is not “mappings from the real-world domain to the domain of reasoning, but rather more ordinary ones like personification”. On the basis of the personification example with beloven ‘to promise’ reprinted in (4a), Verhagen (2000: 205) argues that a modal reading of a nominal construction of beloven is available considerably earlier than the first infinitival one, in (4b). In (4a), the body is not committed to the soul, but provides evidence for a particular expectation about the soul, although it may be false. In (4b), this reading is extended to constructions with an infinitival complement. (4)
a. Een schoon lichaem belooft wel een schoone Ziele; maar het aenghezicht belieght oock wel het herte. (WNT/ De Brune, Bank. 1, 14, 1657) ‘A beautiful body may promise a beautiful Soul; but the face may well deceive the heart.’ b. Dat zelfde meisjen, 't welk iets beloofde te worden. (1793) ‘That same girl, who promised to become something.’
The personification in (4a) is in line with Traugott’s argument that the first shift to Stage II involves a process of semanticization of the “pragmatic inference of futurity implied by promise and threaten” in utterances with a direct object, i.e. contexts where no proposition is syntactically expressed (Traugott 1997: 198).142 It is worthwhile verifying, however, whether the shift from an intentional reading to a non-intentional reading, as in (5), can be explained in metaphorical terms.
156 (5)
Subjectification of prometer and amenazar
As the morning promised a fair day we set out, but the storm coming up again we were obliged to come to. (1784 Muhl [ARCHER])
The example in (5) shows that, although the shift from Stage I to Stage II is metaphorical to a certain extent, the very shift between “domains”, generally required in prototypical metaphors, is lacking. At Stage II, according to Traugott, the inanimate subject of the subjective construction, here morning, predicts something without uttering a promise or a threat. Verhagen (2000: 206), for his part, states that metaphorical uses seem to have preceded the “use as a modal with an infinitival complement”.143 Moreover, he argues for a different evolution of “speaker-hearer subjective” dreigen/threaten (6a) and “character subjective” beloven and promise (6b), based on the latter’s illocutionary force. (6)
a. He threatened to come home early. b. He promised to come home early. (from Verhagen 2000: 204)
Verhagen (2000) contends that by using the verb promise the speaker holds the reported subject responsible for the choice of the word promise, as opposed to the speaker’s interpretation of a reported threat. The specific responsibility of the speaker makes threaten more speaker-hearer subjective “from the start”, which then implies that dreigen/threaten shows up earlier in a modal reading than promise, the latter remaining character subjective for a longer period (cf. Verhagen 2000). Yet, it is not clear how the criterion of illocutionary force combines with the personified uses of beloven/promise and dreigen/threaten, since the intermediate stage of personification does not involve illocutionary force. It would seem that the issue of speaker-hearer subjectivity faces a problem similar to that of mapping between domains, namely that it does not take into account the intermediate stages of the evolution. My analysis will suggest another path focused on the similarity between the early nominal and infinitival complements. Modal amenazar + nominal complement appear together with modal amenazar + infinitive in the 15th and 16th century. By contrast, prometer + nominal complement can predict before the construction extends to combine also with infinitival complements. Before I present this central argument, I first discuss the differences between promise and threaten in terms of the metonymic shift from lexical to non-lexical readings.
A diachronic approach
157
1.3. Metonymic extension Although Traugott (1997) does not explicitly mention the metonymic reinterpretation of promise and threaten, metonymy has become increasingly important in Traugott’s work. Traugott and Dasher (2002: 29) argue that metonymy is probably more basic to language and cognition than metaphor. This assumption is based on the idea that a potential metaphor “must be understood or perspectivized metonymically for the metaphor to be possible” (Barcelona 2000: 31). The observation that promise and threaten differ as to the importance of elements such as the subject or the counteracting force may support the argument that local, metonymic changes are at the basis of the process of subjectification. The underspecification of the commitment of the subject (the shift from Stage I to Stage II) is illustrative of this metonymic change towards the new “epistemic” readings. The metonymic shift towards the personification of the subject of promise and threaten could be based on the loss of the addressee. Recall that in the basic lexical readings the two verbs have a different relation to the addressee. The threat often, but not always, involves a concrete addressee, while the promise is always directed to an interlocutor (be it explicitly or implicitly). Since the addressee is not always present in threaten, this verb lends itself more easily to a subjective view of the threat (as the relative frequency of modal dreigen/drohen and amenazar shows, cf. Vliegen (2006) and Chapter 5, Section 2.1.). The threat is not directed to any particular entity but is seen as a more general threat of an imminent situation that had better be avoided. Since with promise the promise made by the subject can fade away only in a construction without an interlocutor, I hypothesize that promise does not as easily lend itself to a prediction reading which is more distant from its original lexical reading. The question now is whether there is a second metonymic shift from Stage II to Stage III. That is, can the so-called “raising” with promise and threaten and the gradual shift from “objective” possibility (general beliefs) to a more “subjective” possibility (individual speaker’s belief) (cf. Traugott 1997: 198) be accounted for in terms of metonymic change? If such a change takes place, this would imply that the subjective possibility is a part of the objective possibility. This seems reasonable since a generally shared opinion usually also includes the opinion of the speaker. In the terminology used in this volume, the metonymic shift is actually a shift from an intersubjective commitment toward a subjective commitment to the judgment. In this case, there is a process of narrowing the perspective from a broadly held view of the state of affairs to a strictly speaker-oriented view. Stage III clearly corresponds with Tendency III, where the meanings “tend to be-
158
Subjectification of prometer and amenazar
come increasingly based in the speaker’s subjective belief state/attitude toward the proposition” (Traugott 1989: 34–35; Traugott and Dasher 2002: 95). Still, it is important to verify whether “objective” possibility and “subjective possibility” really correspond to the constructions with a nominal complement and those with an infinitival complement, respectively. In other words, does the objective-subjective opposition imply that within the infinitives expressing “subjective possibility” further specifications are impeded? The diachronic corpus data will show that the types of infinitive with Spanish amenazar and prometer undergo a broadening of the set of constructions, which hence leads to a more frequent speaker-oriented view of the state of affairs. In sum, a metonymy approach to promise and threaten avoids the problematic side of the metaphorical approach in terms of domains and forces. Moreover, it focuses on smaller and more local changes in the course of the evolution from lexical to subjective verbs. 2. A diachronic analysis of amenazar and prometer In this section, I verify to what extent the historical evolution of amenazar and prometer correlates with the history of promise and threaten proposed by Traugott.144 The analysis below is based on a thorough scrutiny of the historic sections of Corpus del español (http://www.corpusdelespanol.org): I have systematically looked at the most frequent verb forms of amenazar and prometer between the 13th and 19th century. For the present description I consider the 13th century as an early stage of Spanish, although it could be argued that previous centuries constitute the earliest centuries. I have doublechecked the corpus data to make sure that the observations made below are not built on loose sand. Yet, since the analysis is not based on a detailed quantitative description, further corpus research will certainly be able to refine the point made below taking into account the frequency of the different construction types and the different readings.145 I first discuss amenazar (2.1) and then deal with prometer (2.2). 2.1. Amenazar In this section I will argue that at the end of the 15th century the first occurrences of modal amenazar + infinitive co-occur with amenazar + nominal complement. I first comment on the modal readings of the nominal con-
A diachronic analysis 159
structions (2.1.1), before I pass on to present the first occurrences of modal amenazar + infinitive (2.1.2). Finally, I describe the modal uses in the 18th and the 19th century (2.1.3). 2.1.1. The modal readings with amenazar At the end of the 15th century, Antonio de Nebrija published his Dictionarium latino-hispanicum (Salamanca, 1492), in which he translates the Latin words portentosus and prodigium by means of amenazar, as shown in (7). In these uses, amenazar does not have an intentional subject, nor an addressee. Interestingly, the subject milagro is positive, and is compensated by the negatively oriented nominal complement mal ‘evil’. (7)
Portentum. Portentosus. Prodigium.
milagro que significa mal miracle that-COMPL signify.3SG bad-OBJ ‘the miracle that signifies something bad.’ lo que assi amenaza mal it that-COMPL this way threaten.3SG bad-OBJ ‘which is a bad portent.’ milagro que amenaza mal miracle that-COMPL threaten.3SG bad-OBJ ‘miracle that portends evil.’
Two years earlier, in his Universal vocabulario de latín en romance (Sevilla, 1490), Alfonso de Palencia gave more or less the same comment on portentum when discussing a series of expressions such as antedezir ‘to foretell’ and mostrar ‘to show’, as in (8), which can be called evidential. (8)
monstruum es quasi monestrum: que amonesta auer de venir algo siniestro. ‘that warns that something bad is to happen.’ & prodigium lo antedize: ‘foretells it.’ & portentum lo amenaza: ‘gives a sign of it.’ & ostentum lo muestra: ‘shows it.’
As far as the semantics are concerned, these first occurrences of non-lexical amenazar have a clearly evidential dimension: they refer to the direct evidence the speaker uses to make his/her statement. These examples of humanist lexicographical description show that before the 16th century, amenazar is associated with the prediction of bad things that could happen.
160
Subjectification of prometer and amenazar
The use of the clitic lo ‘it’ and the nominal complement mal show that de Palencia (1490) and Nebrija (1492) consider non-lexical amenazar a transitive verb that assigns an object function to the announced event. In Alfonso de Palencia’s (1490) work, amenazar appears together with the verb amonestar ‘to warn’, which also occurs in other 15th-century texts, as shown in the lexical (9a) and evidential (9b) constructions below. In (9a), the subject of amenazar directs a threat to a person, denoted by le ‘to him/her’, and warns him/her, while in (9b), the two verbs convey a more evidential meaning.146 (9)
a. Enlas otras tres coplas seguientes declara que sera guerra y hanbre y mortandad. Enlas otras quatro coplas que se syguen le amenaza e amonesta que faga oraçion y confesion. (Coplas y glosas de Mingo Revulgo Burgos, 1485) ‘In the other three following coplas he declares that there will be war, famine and death. In the other four coplas that follow he threatens and warns him/her that (s)he should pray and confess.’ b. ... como casa envegeçida cuyo çimjento se acuesta que amenaza y amonesta con señales su cayda sy asy la nuestra vida es continuo amenazada por que sera salteada de muerte tan comedida. (Cancionero de Salvá. Anonymous, 15th century) ... as an old house whose mortar falls down and so it threatens collapse and warns with symptoms, like this our life is continuously threatened because it will suddenly be overridden by simple death.’
In (9a), the combination of amenazar and amonestar shows that the words of the subclause are definitely the words uttered by the human subject of amenazar and amonestar. The illocutionary force of amonestar in (9a) gives amenazar a more performative reading, leading to a more subjectoriented reading than the one in (9b). In the latter example, the possibility of the house’s collapse is phrased by means of the reference by amenazar and amonestar to concrete indications that the mortar is not strong enough. In (9b), su cayda ‘its collapse’ is the object of amenazar and amonestar, and the complement con señales ‘with signals’ is associated with both amonestar and amenazar, which clearly lays the basis for an evidential reading based on inference from external evidence. The potential character of amenazar is shown in the prepositional construction in (10), where sospechar ‘to suspect’ and amenazar are used in the same context.
A diachronic analysis 161
(10)
... & la crisi[s] siempre sospecharas que siempre amenaza de recayda. (Bernardo de Gordonio. Anon. translation. Prognostica. Las pronósticas. 1495) ‘... and the crisis you will always suspect that it always threatens with fall.’
Although these evidential constructions lack an addressee, they do not involve real personification, since amenazar is not restricted to non-human subjects. An alternative explanation is that a metonymic shift excludes a potential addressee/participant and runs parallel with a shift from an external commitment to an internal, mental commitment: in (9b) and (10) it is the speaker who judges that there is a risk that something negative might happen. Interestingly, the nominal complement of amenazar is frequently caída ‘fall’ or ruina ‘ruin’, as exemplified by the humanist dictionaries like the one of Alfonso de Palencia (1492).147 The amenaza ruina construction is a fixed expression in present-day Spanish. In the next subsection, I will argue that the infinitives that show up with evidential amenazar are similar to these nouns. 2.1.2. Modal amenazar + infinitive: first examples In the early stages of Spanish, amenazar combines with infinitives, queclauses, nominal complements or clitics. In (11), I give an example of the 13th century lexical amenazar with an infinitive.148 (11)
… mato a ponpeo & le tajo la cabeça. Et fizo el esto a ponpeo & esto mjsmo mas que lo non quiera dios mesura el de ty & esto te amenaza de fazer. (Alfonso X. General estoria. 1272-1284) ‘... he assassinated Pompeus and decapitated him. And he did this to Pompeus and although God does not want it, he plans the same for you, he threatens to do this to you.’
In addition to the lexical reading, I now show that amenazar + infinitive also develops a modal dimension in the 15th century. The caer examples presented below indicate that the first occurrences of modal amenazar + infinitive show up more or less around the same period as the above discussed modal constructions. Example (12) comes from the translated encyclopaedia Propiedades de las cosas (1492) and describes the profession of the dentist. When the encyclopedist says that the tooth threatens to fall out,
162
Subjectification of prometer and amenazar
he describes a potential outcome, a likely state of affairs that is not considered a positive evolution. (12)
... desçeca & consume los humores podridos que corrompen las rayzes de los dientes & las enzias. refirma el diente que se anda & amenaza de se caer. (Bartholomaeus Anglicus; Vicente de Burgos (translator). Liber de proprietatibus rerum (1230-1240). Propiedades de las cosas. Toulouse. 1494) ‘[this] dissects and depletes the rotten pieces that corrupt the roots of the teeth and molars. Repair the tooth that moves and threatens to fall.’
The construction in (12) is related to the amenaza caída construction discussed in the previous section. Since the step from a change-of-state complement caída to a change-of-state infinitive is minimal in comparison with shifting from a lexical to a subjective threat, this example suggests that modal amenazar + infinitive develops together with the evidential reading of the nominal construction. The absence of an addressee in both the infinitival construction in (12) and the nominal ones mentioned before is an ideal basis for a modal reading. Although example (12) comes from a translated text (from the medieval Franciscan encyclopedist Bartholomaeus Anglicus), it clearly exemplifies modal amenazar with a change-of-state verb as its complement. Moreover, this type of construction persists: at the end of the 16th century Lope de Vega’s utterance in (13) contains the same infinitive caer ‘to fall’ in a construction with amenazar.149 Since modal amenazar + caer also appears in the 16th century (13), it is clear that example (12) is not a hapax.150 (13)
… Este tronco anciano, que ya amenaza a caer de la muerte en el mar cano. (Vega, Lope de. El vaso de elección San Pablo. 1598) ‘This old trunk, which already threatens to fall dead in the grey sea.’
As for the shift from lexical to non-lexical readings, the question now arises whether these constructions involve a metaphorical extension from the physical threat. The answer is seems to be negative. Although diente ‘tooth’ and tronco ‘trunk’ could embody a personification of a threat and something could be undertaken to avoid it, they themselves do not convey any threat in (12) and (13). In what follows, I will show that the amenaza de se caer construction is representative for other similar constructions at the end of the 15th century.
A diachronic analysis 163
As exemplified in (14) and (15), The Crónica de Aragón (1499) contains occurrences that present competition between lexical and modal readings of amenazar, which illustrate the same subjective potential, although in a less explicit way. Given the presence of action verbs in the infinitival complement, these examples display more competition between the lexical and modal readings. (14)
... assi apoderas la poderosa Castilla. que assi domas la nunca domada prouincia. O espada mas que tajante y tan vencedora que assi vences la que siempre amenaza vencer: y queda tanto y tantas vezes por ti vencida. (Gauberto Fabricio de Vagad. Crónica de Aragon. 1499) ‘... Like this you empower the powerful Castilia, that this way you dominate the never dominated province. O sword so sharp and so victorious that this way you conquer that [country that] always threatens to win: and that so many times has been conquered by you.’
In (14), Castilla implies a whole group of potentially human agents, which foregrounds the physical threat. In the opinion of this Aragonese writer the threatening collective entity Castilla ‘Castilia’ retains some kind of intention to win. By contrast, a modal reading could represent Castilla as a country that always awaits the moment to win. In (15), the death of King James is not an intentional subject that threatens to harm Spain. (15)
... començo de anocheçer nuestra España: y a sentir el agrauio que la muerte le amenazaua fazer: que bien como su vida [del rey don Jayme] era quasi publica muerte delos enemigos de dios. (Gauberto Fabricio de Vagad. Crónica de Aragon. 1499). ‘... Our Spain started a dark period: and so to feel the injury/hurt that death threatened her to do, which just like his life [of King James] was almost like a public killing of the enemies of god.’
This utterance relates the harm that the death of the king has possibly inflicted upon the Spanish people (referred to by means of the clitic le). In this construction, if muerte ‘death’ is personified, there is a metaphorical extension from the subject associated with the action of threatening to the allegorical ‘Death’. Since the death presented here rather refers to a sad event with a great impact, however, the amenazar construction is more likely to receive a modal reading.
164
Subjectification of prometer and amenazar
In conclusion, although the modal dimension of these examples from the end of the 15th century is not as clear-cut as the amenaza de se caer occurrence in (12), they nevertheless show that amenazar could have a modal reading in a construction with other infinitival complements before the beginning of the 16th century. I also give two examples from the 16th century. In (16) a sad suspicion threatens to dominate the speaker’s mind. Since this suspicion is not an intentional threatening entity, it does not lead to a metaphorical extension in terms of personification. (16)
…. Ha tomado del alma mía gobierno un triste recelar, que con espanto amenaza hacer mi mal eterno. (Cetina, Gutierre de. Obra selecta. 1536) ‘Has taken over my soul a sad suspicion, that dreadfully threatens to harm me my life long.’
The amenazar construction in (16) favors a modal reading since the state of mind does not correspond to the agentive pattern of hacer ‘to do’. The same non-human subject without agentive structure is found in (17). Here the combination of the subject, i.e. enfermedades ‘diseases’ and the agentive infinitive cortar ‘to cut’ renders a non-lexical reading possible. Literally, the state of someone’s health cannot cut, and thus can hardly be in itself the motive of a threat directed to the participants. (17)
… especialmente cuando las enfermedades suceden en tal coyuntura que les deshacen las trazas de sus oficios, o los amenazan con cortarles la tela de la vida cuando estaban en lo mejor della. (Puente, Luis de la. Vida del P. Baltasar Álvarez. 1589) ‘… especially when the diseases go that way that they ruin the career or threaten to cut the web of life [lit: to them] when they are in the best part of it.’
The infinitives and the subjects of the constructions presented so far deserve some further comments.151 The infinitives caer ‘to fall’, vencer ‘to overwin’ and fazer ’to do’, hacer ‘to do’ and cortar ‘to cut’ point to a dynamic process involving a change of state or an action. Thus, the first examples of modal amenazar + infinitive do not express a static or categorizing view of the state of affairs, but express change-of-state. There is an important difference: the amenaza de se caer construction does not manifest a competition between lexical and non-lexical readings, while the ones with fazer/hacer and cortar do. As for the subjects of these constructions, the 15th century diente ‘tooth’, espada ‘sword’ or muerte ‘death’ are con-
A diachronic analysis 165
crete objects or concrete events that often but not always allow for personifications. I will now show that the range of infinitives and subjects broadens in the 18th and the 19th century. 2.1.3. Modal amenazar + infinitive: 18th and 19th century In this section I will show that, while the initial infinitives with modal amenazar most often involved action or change-of-state, a considerable amount of copular infinitives appear three centuries later. From the second half of the 18th century on, the subject often relates to an abstract state of affairs, as in (18), where a situation is expected to go on for a while. (18)
Las golondrinas anidan a su placer en los desvanes, y las ratas trotan tranquilamente en los sótanos. Semejante estado amenaza prolongarse hasta que los viejos muros cedan a su propia pesadumbre. (Barreras, Antonio El espadachín: narración histórica del motín de Madrid en 1766) ‘The swallows nest in the attic and the rats run untroubled in the basement. Such a situation threatens to go on till the old walls give in to their own sorrow.’
Modal amenazar + infinitive boom in the 19th century, which leads to a considerable diversification of the subjects and the infinitives that combine with the verb, and which co-exist with the earlier constructions (cf. layering, Hopper 1991). In the 19th century, just like in present-day Spanish, there are still personifications of real threats. In (19), for example, the personified sea threatens to devour the mountain. (19)
El mar embravecido amenaza engullirse y tragarse entero este monte, pero la soberbiade sus olas queda confundida por la dureza de sus peñas, y se convierte en espuma. (Francisco Palau, Beato. Mis relaciones con la Iglesia. 1842) ‘The wild sea threatens to devour and to swallow the whole mountain, but the pride of its waves is confounded by the hardness of the cliffs, and becomes foam.’
In the 19th century other modal amenazar constructions without personification are increasingly more frequent. Despite the personification potential of the subjects, the infinitive convertirse ‘to become’ in (20) and (21) does not provoke a metaphorized threat reading of amenazar. For example, not-
166
Subjectification of prometer and amenazar
withstanding the fact that the statue of Melancholy could give rise to a personification rather easily, in (20) it does not invite a metaphorical extension on the basis of personification, since there is no intentional relation between the subject and convertirse ‘to become’. The same observation holds for the demon in (21) who threatens to become a real monster. (20)
Ophelia, deshojando flores, simulaba la estatua de la Melancolía cuando llega a ese punto en que amenaza convertirse en desesperación y en catástrofe. (Sawa, Alejandro. La mujer de todo el mundo. 1886) ‘Ophelia, stripping leaves of flowers, simulated the statue of Melancholy when it reaches the point that it threatens to become despair and catastrophe.’
(21)
Hoy día, muchachos y muchachas son víctimas del excesivo aprendizaje de lecciones, y en número creciente, de año en año, van cayendo en las garras del demonio de los exámenes, que amenaza convertirse en el más implacable monstruo que el mundo haya conocido jamás en la realidad ni en la leyenda. (Giner de los Ríos, Francisco. Escritos sobre la universidad española. 1877) ‘Today, boys and girls are victims of the excessive rehearsal of lessons, and year after year in increasing number, they are falling in the claws of the exams demon, which threatens to become the most inexorable monster that the world has ever known in reality and in legend[s].’
The change-of-state infinitive convertirse ‘to become’ expresses a qualification rather than an action. More generally, in the 19th century, modal amenazar manifests a shift from combining with an action-oriented infinitive towards a larger categorizing view of the preceding subject. This shift towards a more speaker-oriented view is illustrated by the following examples of amenazar with a copular infinitive, i.e. ser la mayor energía ‘to be the major energy’ (22), other attributives such as faltar ‘to be lacking’ (23) or hacerse universal ‘to become universal’ (24), to quote but a few. Since these qualifications still have a dimension of change, they cannot be seen as exclusively categorizing. (22)
El mundo contemporáneo ha comenzado a entrever la aurora de esta fuerza que amenaza con ser la mayor energía social de nuestro tiempo. (Giner de los Ríos, Francisco. Ensayo. Selección. 1877)
A diachronic analysis 167
‘The contemporary world has started to hint at the daybreak of this force that threatens to be the major social energy of our time.’ (23)
Ya ha avisado del peligro de la seca el jefe delacueducto; ya ha rogado el mayor de la ciudad que economicen los vecinos el agua que amenaza faltarles. (Martí, José. En los Estados Unidos. Escenas norteamericanas. 1874) ‘The boss of the aqueduct has warned about the drought; the mayor of the city has asked that the citizens save water that threatens to be lacking for them.’
(24)
...grandes diplomáticos y personajes de diversas naciones, celosos de su popularidad, que amenaza hacerse universal como su gloria. (Castro, Rosalía de. El caballero de las botas azules. 1861) ‘... important diplomats and persons from different nations, jealous of each other’s popularity, which threatens to become as universal as their glory.’
This emergence of infinitives that express a qualification suggests a broadening of the distribution of modal amenazar in comparison with the constructions found in the 15th and the 16th century.152 Traugott (1997:198) describes the shift to the infinitival constructions with English threaten and promise as motivated by the general process whereby “epistemic meanings gradually shift from more “objective” possibility based in general beliefs and attitudes to more “subjective” possibility in the individual speaker’s belief or attitude”. The fact that, from the 19th century on, the infinitives of Spanish amenazar more easily express qualifications suggests that this gradual shift can be refined. Within the infinitival use, there appears to be an evolution in Spanish from a restriction to change-of-state and action verbs toward an extension to attributives.153 This shift does not imply that the verb as such becomes more speakeroriented, since there is layering with older readings of modal amenazar. Moreover, this co-existence of readings can already be observed in 16thcentury Spanish. Furthermore, Spanish amenazar has acquired a “subjective possibility” reading much earlier than English threaten in Traugott’s (1997) diachronic description. In this respect, the analysis elaborated above contributes to Verhagen’s (2000) suggestion that just as Dutch “epistemic” dreigen appeared already in the 16th century, English “epistemic” threaten + infinitive might have appeared earlier than the 18th century. My account of the infinitives patterning with amenazar can thus be a basis for further research on Traugott’s (1997: 189) observation that “it is
168
Subjectification of prometer and amenazar
not clear that epistemic threaten was first limited to inchoative becomplements, although the early examples do involve change of state verbs”. My analysis has shown that the modal reading of amenazar first shows up with non-copular change-of-state verbs and is found in combination with copular verbs only at a later stage. In (2.2) I will show that the first infinitives of modal amenazar clearly differ from those with modal prometer. 2.2. Prometer Let us now look at the development of the modal reading of prometer from the lexical commitment reading with possible illocutionary force. I first discuss the first shift from the lexical readings of prometer to modal readings (2.2.1). Then, I present the first modal readings of prometer + infinitive (2.2.2). Finally, I briefly review the greater variety of infinitives in the 19th century (2.2.3). 2.2.1. The rise of modal readings with prometer In the early stages of Spanish, prometer and amenazar considerably differ as to the use of the infinitive. For example, in the 13th and 14th century prometer is much more frequent with an infinitive than amenazar. In (25), Alfonso X describes a person who promises to keep his faith; the infinitive conveys the promised event. (25)
... la promission que cada vno faze por si: o la que fazen sus padrinos por el quando resçibe el baptismo que reniega del diablo & de todas sus obras & promete de guardar la fe de nuestro señor ihesu xpisto & los mandamientos dela fe catolica: & por esta razon. (Alfonso X. Siete partidas. 1251-1265) ‘...the promise that each one does for him: or the one that his godfathers do for him when he is presented for baptism that forsakes the devil and all his works and promises to keep the faith in our lord Jesus Christ and all orders of the Catholic faith and for this reason.’
The speech act function of prometer involving illocutionary force is present from the early stages of Spanish in both the que-clause construction (26) and the infinitival one (27).
A diachronic analysis 169
(26)
... yo prometo que si entrare en España que vaya a Santiago. (Alfonso X. Siete partidas. 1251-1265) ‘... I promise that when I enter in Spain, I will go to Santiago.’
(27)
... te ruego que me otorgues mi madre & mi mugier. & mis fijas que tienes alla. & prometo te yo de te dar quantos thesoros e en tierra de Mayda. (Alfonso X. General Estoria IV. 1280). ‘I request you that you give me back my mother, my wife and my daughters who you have there and I promise you to give you all the treasury that I have in the country of Mayda.’
In the Corpus del español, the modal readings of prometer are less frequent at the end of the 15th century than those of amenazar. In (28), prometer shows signs of an incipient evidential reading in terms of personification of the malas aficiones ‘vices’ and the pensamientos ‘thoughts’ that promise sweetness, which in the end turns out to be false. (28)
... pues si con igual hazaña qual sin dubda todo ombre podra fazer sacudiendo de si malas aficiones: & pensamientos torpes que al principio prometen dulçura & ala fin paren tristeza & disfamia. (Fernando del Pulgar. Claros varones de Castilla. 1500) ‘... well if with such a heroic deed without doubt everyone will be able to shake off vices and awkward thoughts that at first promise sweetness but in the end seem like sadness and a bad name.’
Although the subjects malas aficiones ‘vices’ and pensamientos ‘thoughts’ are not strictly non-human, they are non-intentional since they do not commit themselves to do something for a participant. Yet, the subject of the preceding clause, i.e. todo ombre ‘everyone’, may but need not be the experiencer. In (28), it is the speaker who associates the complement dulçura ‘sweetness’ with the subjects on the basis of concrete evidence from the speaker’s general experience. The same reading is found in Garcilaso de la Vega’s utterance reproduced in (29). (29)
“Las minas prometen tanta riqueza, que, a pocos años que se labren, valdrá más el hierro que la plata”. (Garcilaso de la Vega, El Inca. Comentarios reales. 1578) ‘The mines promise so much wealth, that, if they are exploited for a couple of years, iron will be worth more than silver.’
170
Subjectification of prometer and amenazar
Garcilaso expresses his impression that the mines have a great economic potential. Here too, the subject is non-intentional and non-commital and the whole proposition is related to the speaker’s appraisal of the state of affairs. In the two preceding examples no direct evidence is given; instead, these constructions indicate that the subjects have a potential to realize what is denoted by the complement. Verhagen (2000) argues that due to the illocutionary force beloven/ promise + infinitive acquire their “epistemic” reading later than dreigen/ threaten. However, beloven/promise constructions with a nominal complement (personifications amongst others) do not have illocutionary force. If the “epistemic” reading of the construction with an infinitival complement comes from the nominal construction, illocutionary force can only play a minor role in the development of the “epistemic” reading of prometer constructions with a nominal complement. In the following section I will argue that illocutionary force has no real influence on the development of the modal reading of prometer + infinitive.154 2.2.2. Modal prometer + infinitive: first examples The first examples of modal prometer + infinitive are found at the end of the 16th century. In (30) Lope de Vega says that the detainee was firm and steadfast and that he promised to be a very hard diamond. (30)
TEOSINDO Ya queda preso. LEOVIGILDO ¿Y queda reducido? ORMINDO Antes, firme y constante, promete ser durísimo diamante. LEOVIGILDO Pues ablandarle quiero con la inocente sangre de un cordero. (Vega, Lope de. La mayor corona. 1598) ‘T: He is already caught; L: And has he already been subjugated? O: On the contrary, he was firm and steadfast, he promises to be a very hard diamond; L: Thus, I want to spank him with the innocent blood of a sheep.’
The question is whether the combination of prometer and the copula ser ‘to be’ is meant to reflect the words of the detainee or those of the speaker. In the latter case, the speaker promises not to say anything and to remain firm; then the construction would be agentive. Such an interpretation is not plausible, however, since the utterance with prometer would be a strange an-
A diachronic analysis 171
swer of a detainee to the question concerning his own state. The prometer example in (31) comes from a text by Vázquez de Espinosa published in 1600. In this text, the studies (estudios) of a college (colegio) are presented as very principled which leads the writer to conclude that the institution itself is likely to be among the best. In (30) and (31), unlike the nominal prometer constructions, inference from direct evidence is at issue. (31)
... hay otro convento de Nuestra Señora de Guadalupe en la otra parte de la ciudad, colegio dedicado para estudios, sus principios son tan grandes que prometen ser uno de los mejores de las Indias y que puede competir con los aventajados de Europa. (Vázquez de Espinosa, Antonio. Compendio y descripción de las Indias Occidentales. 1600) ‘There is another convent of Our Mary of Guadalupe in the other part of the city, a college dedicated to studies, its principles are that high that [the college] promises to be one of the best of the Indias and one that can compete with the most outstanding ones of Europe.’
(31) is an example of the so-called accusativus cum infinitivo. The writer omits el colegio ‘the college’ between the finite verb form prometen ‘they promise’ and the infinitive ser ‘to be’. In (31), prometer nevertheless receives a modal reading. Now, how can the later development of modal prometer + infinitive be accounted for? I argue that the later appearance of the modal reading with this construction can be explained by the different complements of amenazar and prometer. As shown in (2.1), modal amenazar combines with change-of-state nouns such as ruina ‘ruin’ and caída ‘fall’ and change-of-state verbs such as caer ‘to fall’. By contrast, the nominal complements of prometer denote abstract qualifications such as dulçura ‘sweetness’ and riqueza ‘wealth’ that do not lend themselves to infinitival extensions. Moreover, with these abstract nouns there remains a link between what is promised and an undetermined promisee. For example, the minas ‘mines’ promise wealth to someone, while the threat of an imminent event such as a fall does not have an addressee. In other words, with prometer the shift from a clear object status of the nominal complement to a nearmodal one of the infinitive is more complex than with amenazar, which easily alternates change-of-state nouns and verbs. Rather than a difference in illocutionary force, the type of nouns and infinitives could be the real
172
Subjectification of prometer and amenazar
reason for the slower evolution towards modal readings of prometer + infinitive. Traugott (1997: 188), for her part, argues that the early examples of promise are limited to “inchoative events with the verb be”. In Spanish, by contrast, although the first complements of prometer are copular, it is not clear whether they correspond to inchoative events. The ser-complements in the above-mentioned examples express rather an attributive relation with the subject than an inchoative event. The detainee in (30) is already there and his being firm and steadfast is known, and the college in (31) is already imparting education. These subjects are thus not participating in an inchoative event: the future dimension is related to the prediction, which can turn out to be true or false in the future. The attributive nature of the infinitives does not imply that they are restricted to copular verbs only. At the beginning of the 17th century, Salas Barbadillo (1608) uses modal prometer both in an infinitival construction (32) and in a que-clause construction (33). In the first construction, prometer is followed by durar ‘to last’ and its subject is discurso ‘speech’. This subject cannot be seen as an intentional subject referring to a human agent. Here, the speaker makes a judgment about the probability that the discourse will take some time. (32)
No es razón que las canas del señor Cosme Laurencio estén tanto tiempo en pie como este discurso que nos aguarda promete durar, que su presencia persuade respeto, y la necesidad que tenemos della le debe estimación. (Salas Barbadillo, Alonso Jerónimo de. El sagaz estacio, marido examinado. 1608) ‘There is no reason why the hackles of Mr. Cosme Laurencio should constantly raised, as this speech that awaits us promises to last, as his presence invokes respect, and the need that we have of it, deserves consideration for him.’
The same author uses modal prometer also with an addressee le ‘to him’ and a que-clause. This construction indicates that what his heart tells him about his proven courage is a basis to assume that he will probably return home victoriously. (33)
... por Flandes va el mancebo generoso, al templo de la Fama peregrino. Fianzas de su brazo valeroso le ha dado el corazón, que le promete que ha de volver a España vitorioso. (Salas Barbadillo, Alonso Jerónimo de. El sagaz estacio, marido examinado. 1608) ‘Through Flanders goes the generous boy, to the sanctuary of the
A diachronic analysis 173
Fama he goes on a pilgrimage. His heart has assured him about his courage, [his heart] that promises/tells him that he will return to Spain victoriously.’ The construction in (33) cannot be considered purely lexical. The subject clause is inanimate and the subordinate clause does not correspond to a clear promise. This example indicates that at the beginning of the 17th century, modal prometer appears not only in constructions with an infinitive, but also with a que-clause and an addressee, e.g. le ‘to him’. Hence, this non-intentional reading of prometer seems to be well integrated at that stage of the language. Example (34) is a confirmation of this pattern. (34)
.... estudia y deja el rústico ejercicio, que las letras prometen ensalzar tu nombre y fama. (Molina, Tirso de. La Elección por la virtud. 1634) ‘...study and leave the rural practice, that the writings promise to glorify his name and fame.’
The (34) example from Tirso de Molina confirms that modal prometer is established in the Spanish language in the first half of the 17th century. Moreover, the infinitive ensalzar ‘to glorify’ illustrates that there is an incipient extension of its use to other infinitives than the attributive ones. 2.2.3.
Modal prometer + infinitive: 18th and 19th century
From the second half of the 18th century on, a greater diversity of infinitives appears: we find ocupar ‘to occupy’ (35), adquirir ‘to acquire’ (36) and resarcir ‘to indemnify, to forgive’ (37), to mention but a few. The nonintentional subjects of these constructions block a lexical reading of prometer. (35)
Este objeto, cuando no estuviese contenido en las instrucciones del viaje, debía [estar] por su importancia y por la utilidad que promete ocupar muy principalmente la atención de usted. (Jovellanos, Gaspar Melchor de. Correspondencia. 1778) ‘This object, if it were not present in the travel instructions, it should be because of its importance and its usefulness which promises to particularly occupy your attention.’
174
Subjectification of prometer and amenazar
(36)
... ha buscado novia a toda prisa, y se ha casado con la hija de un rico labrador de aquí, sana, frescota, colorada como las amapolas, y que promete adquirir, en breve, un volumen y una densidad superiores a los de su suegra doña Casilda. (Valera, Juan. Pepita Jiménez. 1864) ‘.. he has looked for a girlfriend in haste, and he has married a daughter of a rich farmer from here, sane, fresh and with the coloring of poppies, and she promises to acquire a volume and a stoutness superior to those of his mother in law Miss Casilda.’
(37)
Esta maldición amedrenta a su mala ventura y la suerte se le muestra favorable. ¡Qué gusto! un bello juego promete resarcirle sus pasados afanes y pérdidas. (Montengón, Pedro. Eusebio. 1784) This curse terrifies his bad chance and luck is favorable to him. That’s good! A beautiful game promises to redeem his previous aspirations and losses.’
This greater variety of infinitives leads me to argue that at the beginning of the 20th century modal prometer underwent a shift from a combination with copular verbs or aspectual verbs such as durar to a combination with action verbs and change-of-state verbs. This evolution goes in the opposite direction from the one described for amenazar. Thus, the shift to more “subjective possibility”, as proposed by Traugott (1997), should only be considered in relation with the semantics of the verbs. Prometer is less restricted with copular complements and the shift towards a “subjective possibility” reading is made clear by the broadening of the group of infinitives, as Traugott (1997) describes for English. 3. Conceptual subjectification In this section, I will propose that the diachronic change is in line with the synchronic conceptual interpretation of subjectification. In the first subsection, I will discuss the link between attenuation and subjectification (3.1.). Then, I will analyze amenazar and prometer with regard to attenuation (3.2.). In the last subsection, I will link up the notion of attenuation with focal prominence (3.3.).
Conceptual subjectification 175
3.1. Attenuation Langacker (2000: 297) argues that the subjectification of grammatical constructions involves “attenuation” when the control exerted by an agentive subject decreases. An extreme form of attenuation is the complete transparency of a grammatical form. Since amenazar and prometer are not completely transparent, as was shown in Chapter 5, the attenuation encountered in amenazar and prometer may be partial too. According to Langacker (2000: 301), attenuation of a linguistic expression can be observed with respect to at least the following four parameters: (1) change of status (actual or potential), (2) change of focus (particular elements as focus of attention, profile), (3) shift in domain (physical interaction > social or experiential one), (4) change in locus of activity or potency (from a focused onstage participant to an offstage one, the addressee). Since chapter 4 has extensively dealt with the likelihood of the proposition, I will not return to the actual or potential status here. In the previous section, I have shown that from a diachronic point of view an analysis in terms of shift of domains is not warranted for prometer and amenazar. Therefore, I mainly discuss the change of focus (1) and the locus of potency (4). Let us first consider Langacker’s comments on the “objective” and “subjective” readings of promise.155 In [a lexical promise construction], the subject has to be capable of volition, planning, and communicative interaction. By contrast, [the subjective construction] merely expresses the speaker’s assessment based on an overall appreciation of the current situation; hence there are no restrictions on the subject. The overall pattern documented here suggests that the so-called “raising” constructions are best seen as the limiting case of “equi” constructions, representing the extreme situation where attenuation of subject control (or its counterpart in the object-raising constructions) results in transparency. (Langacker 2000: 315)
The attenuation of the thematic structure and the so-called raising of the subject goes in the direction of certain generative accounts of the link between raising and non-lexical readings (see Rooryck 2000: 44–47).156 The nominal complements discussed in Section (2.) indicate, however, that it is not justified to postulate a necessary correspondence between epistemic value and syntactic verb raising, which is in line with Traugott (1997: 191). Attenuation has to be seen as a dimension of the subjectification that prometer and amenazar undergo. Since there are clear differences between amenazar and prometer, it is to be expected that their subjectification will also differ. Langacker mentions promise and threaten in the same context
176
Subjectification of prometer and amenazar
and thereby seems to suggest that one can give the same account for both verbs. Below I will show that at least the Spanish equivalents need a different account. 3.2. Subjectification Subjectification with amenazar and prometer boils down to the shift from an objective construal of the scene to a subjective construal of the relation expressed by these verbs. An analysis of prometer within Langacker’s Cognitive Grammar framework goes as follows: the act of promising expressed in lexical prometer functions as a trajector, and the accomplishment of this promise is the landmark. The intention of the subject is seen as potency directed toward realizing the event expressed in the infinitive. The conceptualizer traces a mental path from the trajector to the landmark, which situates the infinitival event “downstream in the flow of time relative to some reference point” (Langacker 2000: 302). Importantly, in Langacker’s view the trajector of subjective prometer does not change with respect to that of objective prometer. What is different from the objective scene is the profile and the locus of potency associated with this trajector. With subjective prometer, the trajector has a diminished role in the profiled relationship, since it is the landmark event, i.e. the infinitival process, that is profiled. Attenuation involves a realignment of the locus of potency from the subject of the finite verb to the speaker. From the above, it is clear that within Cognitive Grammar the shift from an objective to a subjective construal of prometer does not imply any change of the semantic structure, but is focused on the realignment of the profiled relationship and the locus of potency. This process is the subjectification of the ‘objective’ construal. This version of subjectification in terms of gradual attenuation does not conflict with what we have seen in the prometer constructions with metaphorical personification in the 15th and the 16th century. The relation between the subjects malas aficiones ‘vices’ and pensamientos ‘thoughts’ and the finite verb in (39) is already attenuated, since the subjects do not involve intention or volition but indicate that they present themselves as sweet or nice. The experiencer is diffuse: the subject of the preceding clause todo ombre ‘everyone’ can but need not be the experiencer. (39)
... pues si con igual hazañaqual sin dubda todo ombre podra fazer sacudiendo de si malas aficiones: & pensamientos torpes que al
Conceptual subjectification 177
principio prometen dulçura & ala fin paren tristeza & disfamia. (Fernando del Pulgar. Claros varones de Castilla. 1500) ‘... well if with such a heroic deed without doubt everyone will be able to shake off vices and awkward thoughts that at first promise sweetness but in the end seem like sadness and a bad name.’ In a construction such as in (39), there is still potency in the trajector/subject and, hence, the landmark is not exclusively profiled. With the infinitival construction in (40), by contrast, the locus of potency is realigned from the subject to the speaker. The infinitival landmark process is profiled and receives all attention. (40)
La muestra, que promete ser apasionante, empieza en el XIX, con el alemán Leibl o el norteamericano Eakins, Maillol, Ensor, Munch, Khnopff y otros simbolistas. (Notic: España: ABC) ‘The exposition, that promises to be fascinating, begins in the 19th century, with the German Leibl or the North-American Eakins, Maillol, Ensor, Munch, Khnopff and other symbolists.’
Although the process of attenuation also holds for amenazar, the trajector landmark relation may be different. As said in Chapter 5, the intention (the trajector) of amenazar + infinitive is directed at change in the addressee/participant. Thus, the realization of the infinitival process is not the final landmark, but a focus of attention between the trajector and the landmark of amenazar, as opposed to prometer whose trajector - landmark relation is direct. The real landmark of amenazar is then the process intended by the threat. Since the trajector - landmark relation differs for amenazar and prometer, the attenuation and subjectification manifested by them are also expected to be different. In constructions in which the interlocutor is not present, the complement of amenazar can become the landmark. The absence of an interlocutor means that in such a construction the potency of the subject is less crucial. Moreover, since such a subject is less profiled, a construction without addressee paves the way for a more subjective reading. In the first occurrences of subjective amenazar the absence of the interlocutor/participant reduces the potency of the subject. Look at amenazar in (41): here the potency of the subject crisis is reduced due to the absence of an addressee, i.e. the lack of a real landmark.
178 (41)
Subjectification of prometer and amenazar
... & la crisis siempre sospecharas que siempre amenaza de recayda. (Bernardo de Gordonio. Anonymous translation. Prognostica. Las pronósticas. 1495) ‘... and the crisis you will always suspect that it always threatens with regression.’
In the infinitival construction in (42), el diente ‘the tooth’ is not the locus of potency since it does not stand for any threat.157 Yet the locus of potency cannot unambiguously be linked with the speaker either, but has become more ‘diffuse’. Just as in the case of prometer, the attenuation of potency causes the infinitival process to be the most profiled element of the construction. (42)
... desçeca & consume los humores podridos que corrompen las rayzes de los dientes & las enzias. refirma el diente que se anda & amenaza de se caer. (Propiedades de las cosas. Toulouse. 1494) ‘dissect and deplete the rotten pieces that corrupt the roots of the teeth and molars. Repair the tooth that moves and threatens to fall.’
In sum, in the subjective construal of prometer, the locus of potency is very strongly associated with the speaker, while with amenazar the potency is never fully associated with the speaker but becomes more diffuse. This implies that the subject may retain some potency as long as it is not directed to an interlocutor. Hence, amenazar does not need complete attenuation to yield a subjective construal. With prometer, by contrast, the locus of potency cannot be the subject on any account, in that it would coincide with the control of the subject. The attenuation of prometer is slower, since the conflict with its agentive patterns is always present. 3.3. Focal prominence In this section I will briefly deal with the focal prominence of the subject with subjective amenazar and prometer. Although in earlier stages both verbs had modal readings in combination with a que-clause, in present-day Spanish, a modal que-clause construction is not possible with amenazar and prometer. Nowadays the subject is usually overtly expressed, but can also remain covert. In Chapter 4, I have shown that the subject of modal amenazar plays a role in the conceptualization of the prediction reading.158 There I argued that the possible nonrealization of the event expressed in the complement can be seen in relation
Conclusions
179
to the potential of change attached to the subject. This has not been observed for prometer. In the previous section, I have shown that the locus of potency is realigned with the speaker in the case of prometer, while this is not so clear for amenazar. The specific role of the subject with amenazar has to be interpreted in this context. The potency becomes diffuse, which does not prevent the subject from playing a role in principle. With a prominent subject, the inferential statement can be influenced by the characteristics related to the subject and point to different outcomes, including nonrealization of the event expressed by the infinitive. By contrast, when such a subject is absent, there is full attention to the realization of the prediction. 4. Conclusions Lexical and modal readings of amenazar and prometer (and their English counterparts) are no accidental homonyms. The above sections have also shown that Traugott’s (1989) version of subjectification and Langacker’s (1990, 2000) are different but complementary for the description of the shift from lexical to modal readings of amenazar and prometer. In this chapter I have also reviewed the role of metaphor and metonymy in the semantic change that promise and threaten undergo. Metonymy can account for the different shifts from lexical to modal readings, while metaphor in terms of “mapping between domains” cannot (cf. Sweetser 1990). Verhagen’s (2000) proposal of (metaphorical) personification is useful to analyze promise/prometer constructions with a nominal complement. And the metonymic shifts with amenazar lead to greater flexibility in the emergence of modal readings. The fact that, unlike prometer, the interlocutor is not a necessary part of the amenazar construction leads more easily to metonymic shifts than with prometer. Traugott’s diachronic account of promise and threaten in three historical stages runs parallel to the shift in Spanish prometer and amenazar. My analysis has shown that the different stages of the Spanish verbs are closer to each other than those proposed for English. At the end of the 15th century, the nominal and infinitival constructions with modal amenazar may even have arisen simultaneously, albeit with a restricted number of infinitives. Moreover, the link between the nominal complements and the infinitival complements, on the one hand, and the metonymic changes, on the other, have proven to be a stronger motive of change than that of illocutionary force (cf. Verhagen 2000).
180 Subjectification of prometer and amenazar As for Langackerian subjectification, the attenuation of the subjects leads to a change in the locus of potency, which gives a special profile to the complement of the constructions. I have shown that this kind of subjectification is not in contrast with the diachronic approach. Although Langacker (2000) does not reject the idea of mapping between domains, Langackerian subjectification takes into account the two endpoints of the continuum, namely the maximally objective construal and the maximally subjective construal, and, more importantly, the continuum itself.
Chapter 7 Evidentiality, modality and transparency with poder, deber and tener que
In the Hispanic literature a gradience is observed from poder over deber to tener que in terms of weak, intermediate and strong speaker commitment (cf. Fernández de Castro 1999, Gómez Torrego 1999). Since in the generallinguistic literature there is no agreement on the origin of speaker commitment, it is worth examining whether the evidential dimension is related to the different degrees of commitment conveyed by the modals. Unlike truly evidential parecer/resultar and modal amenazar/prometer, the Spanish modals poder, deber and tener que can – but need not exclusively – express the speaker/conceptualizer’s judgment of the likelihood that the proposition is true. In this chapter, I will argue that the epistemic qualification and the speaker commitment of poder, deber and tener que are to some extent related to the evidential modes of knowing involved.159 This chapter thus deals with the overlap domain of inferential evidentiality and epistemic modality (cf. van der Auwera and Plungian 1998). Although the readings of deber and tener que can vary according to the type of inference that takes place, the corpus data from the Habla Culta show that each of the three epistemic/evidential modals has a prototypical reading: (i) an epistemic judgment with poder usually does not involve inferences and corresponds to a weak speaker commitment, (ii) a statement with deber is most often based on inductive inference and stands for an intermediate commitment, (iii) the statements with tener que are most often driven by deductive inference and convey strong commitment. In this chapter I also discuss the relative transparency of the three different epistemic and evidential modal verbs. I want to examine whether they combine with the same kinds of subject and infinitive. The chapter is organized as follows. I first briefly discuss the frequency distribution of the modal verbs in the spoken and written corpora (1.). Then I examine the possible relation between inference and poder (2.), and deal with the different readings of deber (3.) and tener que (4.) attested in the corpus. In the last section I examine the transparency of the modals on the basis of the corpus results.
182 Evidentiality and transparency with modals 1. Frequency distribution In the overwhelming majority of cases, poder is construed with an infinitive. Its readings can be roughly subdivided in three types of modality: dynamic modality, deontic modality and epistemic modality. Yet, since there are a lot of bridging contexts where it not clear whether an expression has one or the other reading, additional columns are provided. Dynamic modality is usually characterized in terms of the capacity or ability of the subject-participant or the situation. Second, deontic modality is traditionally defined in terms of “permission” and “obligation” (Palmer 1986: 96– 97). Deontic modality stands for the relation between the speaker (or another addresser) and the participant(s) or addressee(s), who may but need not be mentioned in the utterance. Third, epistemic modality stands for the assessment, typically but not necessarily by the speaker, of the likelihood of the event expressed. Epistemic modality can be either performative or descriptive. Table 33. Readings of poder. Dynamic
Epistemic Deontic
Dyn Epis
Deont Epi
Deont Dyn
Total
written %
620 44.57
419 30.12
228 16.39
108 7.76
6 0.43
10 0.72
1391 100
oral %
1942 60.18
330 10.23
597 18.50
263 8.15
20 0.62
75 2.32
3227 100
The dynamic readings of poder clearly outnumber the epistemic and deontic readings in both the oral and the written corpus. Interestingly, the epistemic readings are more frequent in the written corpus than in the oral one, while the deontic readings occur with a similar frequency. The modal verb deber, almost exclusively construed with infinitives, basically expresses two types of meanings: a deontic one and an evidential one. Table 34 presents the distribution of deber. Two elements deserve special attention: (i) the overwhelming majority of deber examples are deontic in nature, and (ii) the evidential reading is more frequent in the spoken corpus.
Frequency distribution
183
Table 34. Readings of deber Deber
Deontic
Evidential
Deontic Evidential
Total
written %
495 92.52
36 6.73
4 0.75
535 100
oral %
467 74.13
153 24.29
10 1.59
630 100
The results presented in Table 34 invite us to further analyze the relation between the type of meaning and the spoken or written nature of the corpus. Since non-deontic deber has an evidential dimension, it is expected to be related to the speaker’s immediate environment. Such an interaction between the speaker and the surrounding reality is most common in spoken discourse. The case of tener que ‘have to’ is similar in this respect, as shown in Table 35. Table 35. Readings of tener que Deontic
Evidential
Deontic Evidential
Alethic
Total
written %
189 100
0 0
0 0
0 0
189 100
oral %
1260 94.52
60 4.50
9 0.68
4 0.30
1333 100
The epistemic/evidential readings of tener que are only observed in the oral corpus and are less frequent than those of deber. Although tener que confirms the tendency observed for deber that epistemic modals with a strong evidential dimension are more common in spoken discourse, the distribution of tener que above all points to a strong deontic dimension. Recall that the purely epistemic modal readings of poder are far more frequently attested in the written corpus. I will take this difference as a basis for the cognitive-functional account of the modals and the evidential and epistemic qualifications they convey.
184 Evidentiality and transparency with modals 2. Poder: preference for epistemic modality The epistemic modal poder is more directly speaker-oriented than the other modals: “the fact that the speaker makes use of some kind of evidence upon which her appreciation is based does not seem to belong to the immediate semantic scope of the item in question” (Mortelmans 2001: 133).160 The epistemic verb poder/may is most often defined in terms of autonomous assessment of the possibility that the proposition corresponds to reality. Fernández de Castro (1999: 158)161 gives an elaborate semantic description of poder, to which he applies Lyons’ (1977) distinction between subjective and objective epistemic modality:162 (1)
Juan puede venir. ‘John may come.’ A: I think that John is coming/will come. B: That John comes is a possible fact.
According to Fernández de Castro (1999: 164–166), the A interpretation is “subjective”, while the B interpretation is “objective”. The author argues that the latter interpretation can also be considered “neutral dynamic modality”, since the circumstances allow that John comes (cf. Palmer 1979: 176). Fernández de Castro (1999: 171) claims that the context determines the epistemic, dynamic or deontic uses of poder, suggesting that there is only one meaning of poder with several contextual varieties. In the same vein, Silva-Corvalán (1995: 73) adopts a “mono-semantic” approach and distinguishes between an invariant meaning, on the one hand, and a contextualized meaning on the other. As for the epistemic readings of poder, Silva-Corvalán discerns “root possibility” (2a) and “epistemic possibility” (2b) on a contextual basis: (2)
a. Ya puede tener todos los millones que quiera [la persona sigue siendo buena, si es buena por naturaleza]. (from Silvá-Corvalán 1995: 82) ‘(S)he may have all the millions (s)he wants [the person continues to be good, if (s)he’s good by nature].’ b. Que si hoy se cierra el plazo, dentro de tres meses ya puede ser el examen. ‘if the deadline is today, then the exam may be within three months.’
Epistemic poder
185
According to the author, both readings in (2) are concerned with the speaker’s assessment of the likelihood that the content of a proposition is or may become true. Silva-Corvalán (1995: 82) sets up an ideal context for these two readings. The clearest context of “root possibility” (2a) includes (i) a non-agentive, non-volitional, non-specific subject, (ii) a stative infinitive (e.g. ser ‘to be’, estar ‘to be’, cf. the class of attributive verbs mentioned before), (iii) no evidence of previous actualization of the proposition, and (iv) a non-past time orientation. The truly epistemic modal reading of poder (2b), Silva-Corvalán (1995: 82) argues, involves “the speaker’s logical inference and lack of confidence about the possibility of actualization of the proposition”. The best context for such a reading is one with (i) inanimate, non-specific subjects, or obligatory subjectless sentences, (ii) stative verbs, and (iii) a negated infinitive. Although Silva-Corvalán refers to deduction and inference, she does not consider them part of an evidential qualification in poder. More concretely, she argues that lack of evidentiality is crucial for epistemic possibility (2b), as opposed to root possibility (2a): “If evidentiality exists, an example may then be interpreted as conveying root possibility” (1995: 83). The “non-past orientation” of the “root possibility” may conflict with its alleged evidential dimension, since evidential readings can also draw on past events. For example, Fernández de Castro (1999: 168) mentions that the epistemic reading of poder with the passivized infinitive ser engañado ‘be cheated’ in (3a) is inferential in nature.163 (3)
a. Pedro puede ser engañado por Juan. ‘Peter may be cheated by John.’ b. Pedro puede ser engañado por Juan. Es que no se ven última mente. ‘Peter may have been cheated by John. Therefore they don’t see each other lately.’
The fact that es que no se ven últimamente ‘they don’t see each other lately’ can be added to this clause, as in (3b), without loss of the epistemic possibility reading, shows that inferentiatility is not incompatible with epistemic possibility. Yet, the link between absence of evidentiality and prominence of epistemic possibility paves the way for considering weak speaker commitment in terms of lack of evidential qualifications. In the rest of the chapter, I will contrast this with the intermediate and strong commitment of deber and tener que. Let us have a further look at poder. In corpus example (4), the conclusion expressed by “root-epistemic” poder is based on an inference from the
186 Evidentiality and transparency with modals preceding context, which indicates that the evidential dimension is not necessarily absent in poder. (4)
... por ejemplo una amiga se la pasa todo el santo día en la casa... después cuando ya están establecidos en el matrimonio etcétera ¿no? Eso creo que puede traer unos problemas hasta más graves que los anteriores ¿no?. Bueno... (Habla culta: Caracas: M16) ‘For example a friend stays the whole day at home... then when they are settled in marriage etc., you know? I think that this [situation] may/can bring problems that are more serious than the earlier ones, you know? Well…’
In (4) the speaker communicates that on the basis of his/her own knowledge (s)he subscribes to the possibility that problems may arise. The poder construction is preceded by the mental state verb creer ‘to believe’. As shown in (5), however, there need not be a speaker-oriented hedge for an inferential reading to be possible. (5)
ENC: Que alguna persona dijiste.... INF B: ... que en operación de la garganta o quién sabe qué, quedó sordo. INF A: Todas esas cosas pueden ser muy peligrosas para el oído. (Habla Culta: México: M18) ‘ENC: that a person, you said. INF B: which in a throat or whatever operation, became deaf. INF A: All these things may be very dangerous for the ears. ’
Most often the conclusion expressed by poder is based on the speaker’s own belief and, hence, is not the result of additional reasoning. The most common autonomous epistemic value corresponds to the assessment of the likelihood that the proposition could be true. The particle quizás ‘maybe’ in (6) is illustrative of the low speaker commitment of the poder construction. (6)
... una de las cosas que quizás... eh.... actualmente más me ha llamado la atención es... eh.... eh... la importancia que le está dando el cine a lo estético. Quizás puede ser por la competencia que hay con otros medios de comunicación de masas. (Habla Culta: Santiago: M8) ... one of the things that at the moment strikes me most is the importance that the film industry attributes to the esthetic. This may be due to the competition with other mass media.’
Inferential deber and tener que
187
The epistemic reading of poder usually does not rely on the speaker’s additional reasoning on the basis of evidence. This fact makes it possible for the epistemic modal to also refer to another conceptualizer, as in (7). (7)
A veces se me ... se me ha dicho que posiblemente el que una persona se ... exprese con bastante concisión puede ser influencia del uso del inglés yo ..v.. efectivamente [ he estu ... he estu ] he estudiado con mucho cuidado el idioma inglés. (Habla culta: San Juan: M20) ‘Sometimes I have been said that possibly the fact that a person expresses him/herself with a lot of precision may be an influence from using English. Indeed I have carefully studied the English language.’
In (7) the speaker reports on someone else’s assessment of the possible influence of English on their mother tongue. Thus, poder is not restricted to the speaker’s own assessment, but can also express the point of view of other conceptualizers. In sum, the inferential dimension is not crucial to the epistemic reading of poder, which is most often based on the conceptualizer’s autonomous assessment, although poder can be inferential or refer to other conceptualizers than the speaker. 3. Deber and tener que: on inferentiality and hearsay The modal verbs deber and tener que have so far not been accounted for in terms of inductive and deductive inferentiality or reportive evidentiality . In this section I will show how my analysis can fill this gap. 3.1. Deber In the recent linguistic literature, the relation between “deontic” and “epistemic” uses of English must and its counterparts has been widely discussed (cf. Bybee et al. 1994; Goossens 1999, 2000). In the Spanish linguistic tradition, the similarities between the two readings have not gone unnoticed either. Gómez Torrego (1999: 3348), for example, links the “likelihood” (probabilidad) reading of deber (de) in (8a) with an “obligation” (obligación) reading.164
188 Evidentiality and transparency with modals (8)
a. Mi padre debe de estar en casa en estos momentos. ‘My father must be at home at the moment.’ b. Mi padre puede estar en casa en estos momentos. ‘My father may be at home at the moment.’
Gómez Torrego (1999: 3348) notes that an utterance with deber (8a) expresses a hypothesis based on “external or pragmatic circumstances”, while no external evidence is needed with poder (8b). With deber, there is a situation that makes the speaker think that his father is at home. Since the proposition is only true under the “condition” that nothing irregular happens, a “margin of doubt” remains in this deber expression. Fernández de Castro (1999) and Silva-Corvalán (1995) refer to the obligation patterns in a more systematic way. Fernández de Castro (1999: 179– 180) argues that what is communicated by means of deber (de) is the “conviction that there exists a cause-effect relation” between the circumstances and the statement, which he calls an “illocutionary act different from a simple assertion, since it involves a “qualification of commitment” by the speaker to the content of the message” [my translation]. Fernández de Castro (1999: 180) also observes that modal adverbs expressing high likelihood like seguramente ‘certainly’ and muy posiblemente ‘very possibly’ can paraphrase deber constructions, while epistemic poder corresponds to lower likelihood, as expressed by adverbs such as quizás ‘maybe’ or talvez ‘perhaps’. Unlike the Anglo-Saxon tradition, Silva-Corvalán (1995: 86), for her part, relates the different readings of deber to “epistemic possibility”, which differs from poder in that deber expresses the speaker’s confidence in the realization of the proposition. The author considers deber “more strictly modal in its semantics” than poder in that it is “in all contexts non-factual”. According to Silva-Corvalán (1995: 87), this semantic difference “accounts for the fact that one of the Contextualized Meanings of deber is EP (epistemic possibility), but not RP (root possibility), which is related to ability and evidentiality”. While deber does indeed not have a root possibility reading, it certainly involves evidentiality, as Olbertz (1998) points out. Olbertz (1998) argues from a Functional Grammar point of view that deber is the only periphrasis expressing inferential modality. This type of modality is different from epistemic modality with respect to “the source of the speaker’s commitment to the truth of the proposition” (Olbertz 1998: 410). While in the case of epistemic modality, the source is the speaker with his/her personal convictions and beliefs, in the case of inferential modality, the speaker relies on evidence from which (s)he infers that the proposition must be true. Thus, as far as deber is concerned, it is justified to
Inductive inferences with deber
189
assume that the mode of knowing and the speaker’s commitment go hand in hand. 3.2. Tener que Fernández de Castro (1999: 190) argues that tener que scores highest on the commitment scale. Interestingly, his corpus data indicate that the “maximal” speaker commitment of tener que to the statement receives contextual support from the presence of elements such as por fuerza ‘by force’ and estoy segura ‘I am sure’, as in (9a). Other examples by Fernández de Castro (1999: 191) show that mental predicates such as sospechar ‘to suspect’ (9b) express too weak a speaker commitment to combine with tener que, in contrast with creer ‘to think’, which easily combines with tener que (9c). (9)
a. Vamos, que no le gusta [una determinada ciudad]. Pues también tendrá que tener cosas bonitas, hombre, estoy segura; monumentos artísticos, palacios [...]. ‘Thus, he doesn’t like it [a certain city]. Well it should also have beautiful things, you know, I am sure; artistic monuments, palaces.’ b. ? La polícia sospecha que tuvo que tratarse de un atentado. ‘The police suspects that there must have been an attack’ c. Sí, creo que tiene que haber todavía una bolsa por lo menos. ‘Yes, I think that there must be at least one bag.’
Olbertz (1998: 396) considers tener que as a periphrasis that can express both “event-oriented epistemic modality” and “proposition-oriented epistemic necessity” but no “inferential modality”. Event-oriented epistemic modality indicates that the occurrence of the state of qffairs (SoA) is “certain”, as opposed to a “possible” SoA (poder + infinitive) and a “highly probable” SoA (deber + infinitive).165 Olbertz (1998: 396) argues that the source of the judgment is “general knowledge of the world and/or of the situation, which is not made explicit”. This leads her to argue that tener que does not express inferential modality. Although the evidence is not overtly mentioned in a tener que construction, the statement can be based on personal evidence. Moreover, personal beliefs may not be the best criterion, since the belief of the speaker is central in both deber and tener que whether it is based on explicit evidence or not. I will therefore discuss Olbertz’ (1998) account of tener que in further detail. Olbertz (1998: 413) argues that when using tener que + infinitive, the speaker commits him/herself to the truth of the proposition. By contrast,
190 Evidentiality and transparency with modals with deber, Olbertz argues, the speaker’s commitment to the truth of the proposition depends on his/her data: “if the data turn out to be wrong, the speaker is no longer committed to the truth of the proposition”. This leads Olbertz (1998) to contend that tener que “expresses unconditioned modal necessity, whereas deber expresses conditioned modal necessity”. Olbertz (1998) bases her argument on the different distribution of negation, which I reproduce in (10). The infelicitous negation with tener que in (10a) is argued to be illustrative of the fact that “strong truth commitment can be expressed in positive terms only” (Olbertz 1998: 405). In (10b) the negation falls within the scope of deber and refers to the proposition: the weaker commitment does not block the negation.166 (10)
a. * No tiene que ser fácil para ti. ‘it does not have to be easy for you.’ b. Tú claro, de estas cosas no debes estar muy al tanto. ‘You, of course, are not very much acquainted with these things, I suppose.’
Interestingly, Silva-Corvalán (1995: 90–91) suggests that the restriction on negation with tener que is due to the fact that the “possibility” reading of tener que is a more recent development. This chapter will show that both deber and tener que express an inferential reading. The observations described in the literature are a good basis for the investigation into the evidential dimension of deber and tener que. Several questions arise. First, if tener que is associated with special contextual support (hedges), what kind of hedges does deber combine with? My corpus data will show that deber occurs more often with hedges such as creo que ‘I think’ than tener que. Second, Gómez Torrego (1999) seems to suggest that deber and tener que are always based on external evidence. Yet, I will show that these verbs can also have purely epistemic readings which are not based on external (direct) evidence. Third, the observation that the “epistemic” reading of tener que could be more recent than that of poder and deber opens an interesting line of research. Against the background of the assumed shift from deontic to epistemic readings, there may be a correlation between the commitment and the evidentiality of tener que, on the one hand, and its recent development from deontic readings, on the other. Fourth, the link between the evidence used and the inferential statement is not clearly defined in the Spanish linguistic literature. Olbertz’ (1998) refusal to attribute inferential modality to tener que and her argument that deber is “conditioned by the source of information” could be imputed to a restricted view of inference, viz. based only on direct evidence. On the
Inductive inferences with deber
191
other hand, it can indicate that these verbs do not have one but several evidential readings, i.e. both inferential and non-inferential ones. I will argue that inference from reasoning is crucial in both deber and tener que and that they tend to be based on a non-logical (inductive) versus a logical (deductive) type of inference. I will therefore review the logic behind inference from reasoning in (3.3.). Furthermore, I will discuss the existence of noninferential evidential readings in (3.4.). 3.3. Deduction and induction In order to improve our knowledge of inferential mechanism, I will first discuss Dendale’s (1994) and Kronning’s (1996, 2001) logic-based approach to French devoir. Dendale (1994, 1999, 2000) argues for an inferential mechanism for all readings of evidential devoir,167 while Kronning (1996, 2001) has a three-dimensional view of this verb in terms of deontic, alethic and epistemic readings. In this section, I will mainly discuss the former view, which is based on deduction and induction. The most frequent form of deduction is the syllogism, i.e. the deductive argument in which a conclusion (q) is inferred from two premises (p), a major and a minor premise. The major premise stands for an implication and guarantees the shift from the minor premise to the conclusion. The modus ponens of deduction is the following: (11)
p —> q p ... q
With devoir it is rare to find two explicit premises. The most frequent explicit premise is the minor one, as in (12):168 (12)
On ne doit pas connaître cette maladie-là, vu que c’est une maladie toute récente, existentielle en plus de ça.(Queneau L439. 26) ‘One must ignore this disease, given that it is a very recent disease, and in addition an existential one.’
In (12) the minor premise is made explicit by means of a causal subordinate clause introduced by vu que ‘given that’, while the major premise consists in assuming that important diseases are generally known. The combination of the two premises provides a solid ground for the conclusion communicated by means of devoir. If the minor premise is not explicitly mentioned
192 Evidentiality and transparency with modals in the utterance, it is more difficult to pinpoint the premises of the inference. For example, for the conclusion ‘she must be ill’ in (13b), the speaker may rely on a whole series of possible premises to which the hearer does not have access. (13)
a. Caroline a mauvaise mine. ‘Caroline looks bad.’ b. Elle doit être malade. (taken from Dendale 1994: 27) ‘She must be ill.’
The process of creating information follows three stages: (i) premises, (ii) inference and (iii) evaluation of the conclusion(s). It is a “parcours mental avec recherche et confrontation d’hypothèses” (Dendale 1994: 33–34). The validity of the deductive inference depends on the validity of the premises: To say that a deductive argument is valid is to say that it is not possible for its conclusion to be false when its premises are true. Therefore, a deductive argument is valid when, if its premises are true, its conclusion must be true. Deductive arguments that fail to guarantee the truth of their conclusions, if their premises are true, are thus invalid. (Copi and Cohen 2003: 16)
Although the context often does not contain explicit premises, it is the speaker who evaluates the different premises and decides on the final conclusion. Since the premises are often unclear, speaker and hearer never have a guarantee of complete validity of the conclusion. Belief in one’s own premises can be strong, however, and the conclusion is often presented as a logical result of the speaker’s reasoning.169 Dendale’s (1994) account of devoir is stated in terms of deduction rather than abduction (see also Dendale and De Mulder 1996). The latter type of inference involves a process of finding premises, which is the basis for an interpretive reconstruction of causes and intentions (Peirce 1965: 28, 374).170 What is at stake, for Dendale (1994: 30), is a logical implication of the consequent from the antecedent which is not necessarily based on a unidirectional cause-effect relation, as shown in (14d). (14)
a. Jean a beaucoup travaillé. ‘John has worked a lot.’ b. Jean est fatigué. ‘John is tired.’ c. Jean doit être fatigué parce qu’il a beaucoup travaillé. ‘John must be tired because he has worked a lot.’ d. Jean doit avoir beaucoup travaillé car il est fatigué. ‘John must have worked a lot because he is tired.’
Inductive inferences with deber
193
Note, however, that the utterance in (14a) is based on an abstract logical relation, while the one in (14b) involves some kind of perception, which is also hinted at by the conjunction car.171 This may point to different kinds of deductive readings, or question the exclusively deductive basis of devoir in (14b). Kronning (1996, 2001) offers an alternative to the overall mechanism of logical deduction, without rejecting, however, the deductive inferential dimension of devoir proposed by Dendale (1994). Kronning’s (1996, 2001) account distinguishes between “alethic modality”, which expresses all nondeontic forms of necessity (15a) and “epistemic modality” which refers to likelihood (15b). (15)
a. Tout ce à quoi on réfère doit exister. Appelons cela l’axiome d’existence. (Searle 1969:121, taken from Kronning 2001: 68) ‘All what is referred to must exist. Let us call this the existence axiom.’ b. Le ciel se découvre. Le temps doit être en train de s’améliorer. ‘The sky is clearing. The weather must be getting better.’
The statement in (15b) is based on premises in praesentia (cf. Dendale 1999: 21), while the one in (16) relies on premises in absentia. (16)
D’après les prévisions météo, le temps doit s’améliorer demain. (from Kronning 2001: 76) ‘According to forecasts, the weather will get better tomorrow.’
Kronning (1994: 288, 2001: 76) points out that the premises in praesentia are generally perceptual data on which the speaker bases his/her statement via an inferential process. Squartini (2004: 876) argues that the premises in absentia in (16) come from an external source, which is reported by the speaker. Note that premises in absentia do not necessarily refer to indirect sources of information, since reasoning can also be a basis for speakeroriented inference.172 Prototypically, deductive inference is based on the speaker’s own reasoning; but, the logical relation is also found in inferences based on explicit evidence and reported hearsay. Moreover, when the logical relation is backgrounded, a purely epistemic reading may become possible. In addition to the logical inferences based on deduction, there are socalled non-logical inferences, namely induction, mathematical calculation and estimation (cf. Dendale 1994). Induction can be split up into three stages. The first stage is to understand the observation of the evidence. The
194 Evidentiality and transparency with modals second is to form a hypothesis that attempts to describe the information in relation to the person's general knowledge. The resulting conclusion goes beyond the initial information by incorporating one's general knowledge. The third step is to evaluate the validity of the conclusion that was reached. Unlike deductive inferences, inductive inferences yield conclusions that increase the semantic information over and above that found in the initial premises. Dendale (1994: 32) gives the following example of inductive inference with devoir: (17)
Les Allemands de l’Est doivent avoir beaucoup souffert si on en juge par ce que cette famille est-allemande a subi. ‘The East-Germans must have suffered a lot if you see what this East-German family has undergone.’
In (17) the account of the family’s situation is used as a premise for inductive reasoning, which leads to an assumption regarding the whole of East Germany. In the case of inductive inferences, we cannot be sure that the conclusion is a logical result of the premises, but we may presuppose that it is reliable. The speaker adds an interpretive dimension to the evidence used, which is likely to be true given the information available in the world out-there. Yet, unlike a deductive conclusion, an inductive conclusion does not suggest that the event is a necessary outcome according to the speaker’s knowledge. Dendale (1994: 37) argues that the different evidential values, i.e. the logical or non-logical type of operations chosen by the speaker, are aimed to provide the interlocutor with the “elements needed for [the latter’s] evaluation of the degree of reliability of the information”. This evaluation is subjective in principle and susceptible to variation since it depends to a large extent on the expectations and norms the interlocutor subscribes to. By contrast, the degree of speaker commitment is stable: it is stronger with logical inferences than with non-logical ones. The analysis presented in Sections 4 and 5 will account for the different degrees of speaker commitment in terms of the different inferential types. The division between logical and non-logical inferences with devoir highlights the fact that the latter involve creative, inductive reasoning, which leads to lower speaker commitment than deductive inference. Deductive and inductive inferences with devoir are not always based on a factual statement but can also rely on information that comes from other people. Now, after the discussion above we can go a step further. It stands to reason that in some cases the inference itself is made by other people and reported by the speaker (cf. Squartini 2004).173 In this case, despite the modal value
Inductive inferences with deber
195
of devoir, the speaker does not necessarily support the statement made elsewhere. This possibility illustrates that the speaker can be foregrounded or backgrounded, which points to specific discourse strategies. Dendale (1994) pays a lot of attention to the logical relations involved in the inference. But little is said about how the logical mapping works in discourse. For example, how does the deductive modal combine with speaker-oriented hedges such as creo que ‘I think’? What is the difference with inductive inferences? In Section 3., I will show that hedges like creo que ‘I think’ or supongo que ‘I suppose’ more often accompany inductive than deductive readings. These hedges, I will argue, are in turn an indication of the reduced role of inference itself, in favor of a purely epistemic reading. 3.4. Non-speaker oriented evidentiality The relation between epistemic modality, inferential evidentiality and reported knowledge has extensively been discussed for the French, Dutch, German and English equivalents of deber. Below I will look at how previous studies deal with the speaker commitment involved (or the absence of it). Lyons (1982: 109) distinguishes between “objective” and “subjective” readings of must to account for the ambiguity in (18). (18)
You must be very careful. a. You are required to be very careful. b. I require you to be very careful. c. It is obvious from evidence that you are very careful. d. I conclude that you are very careful.
The two “epistemic” interpretations of English must correspond to a more evidential, “objective” reading (18c) and a more epistemic, “subjective” one (18d).174 The evidence mentioned in (18c) could be further subcategorized into different types of evidence: the evidence can be shared (intersubjective) or non-shared (subjective). For English must, it is generally assumed that the evidence is directly related to the speaker, and is, hence, subjective. Although this is certainly the case for English must, De Haan (1999) argues that this need not be so for Dutch moeten ‘must’. This author observes a reportive evidential reading that is different from the so-called epistemic reading. The relation between reportive and epistemic readings is not uncontroversial, however. Generally speaking there are two views on
196 Evidentiality and transparency with modals this issue: (i) the reportive reading is different but derived from an initial inferential statement, as in De Haan (1999), (ii) the reportive reading is different from the inferential reading and is not derived from the latter, as argued by Traugott (1989), Mortelmans (2001) and Squartini (2004). De Haan (1999) argues for a strict distinction between evidentiality and epistemic modality. He does not consider English must to be a full-fledged evidential verb, and therefore gives it the label “epistemic”: “its epistemic necessity is based on some kind of evidence the speaker has for the statement”. According to De Haan (1999: 86), the existence of evidence alone is not a sufficient condition to take must for a fully grammaticalized evidential, since the same evidence can be used with the epistemic modal may, as in (19). (19)
a. John is at home, because the light is on. b. John may/must be at home, because the light is on.
De Haan (1999) contends that in Dutch, unlike English, the evidence used for a modal statement of moeten may be indirect, that is, taken from another person’s evaluation, as in (20). (20)
Het moet een goede film zijn. ‘It must be/ it is said to be a good movie.’
According to De Haan, Dutch moeten is developing into an evidential by means of conversational implicature. The hearer picks up a proposition like (20) and can decide that the “presence of evidence for the speaker’s statement is more important than the evaluation” of the likelihood that the proposition is true. Then, the hearer (the future speaker) “throws out the evaluation part and keeps the evidence part” (De Haan 1999: 87). De Haan (1999) emphasizes that the relation between the inferential/epistemic reading and the evidential one is not a “logical entailment”, and argues that this is the reason why this change does not occur in many languages. Interestingly, when the speaker throws out the evaluation restricting himself to reporting the evidence, there is no commitment involved. This once again confirms the link between speaker commitment and epistemic judgment of the proposition. Other reportive uses do not convey speaker commitment either. Squartini (2004) analyzes Romance devoir/devere/deber and their ability to function as a reportive marker. Squartini (2004) considers the reportive reading as intermediate between the deontic and the epistemic reading, just like ‘alethic modality’ (cf. Kronning 2001: 75–76). As noted above, Kronning’s
Inductive inferences with deber
197
(2001: 76) example (21a) illustrates that devoir can report on a source external to the speaker. The reportive reading of devoir in (21a) combines with a future dimension: the proposition is expected to become true. This futural use of devoir is also possible in the hypothetical protasis of a conditional clause (21b) and the future in the past reading (21c). (21)
a. D’après les prévisions météo, le temps doit s’améliorer demain. ‘According to the forecasts the weather will get better tomorrow.’ b. Si je ne dois pas vous voir bientôt, je vous enverrai l’objet par la poste (from Fleischman 1982: 148) ‘If I don’t see you again soon, I’ll send the thing to you by mail.’ c. Je savais qu’il devait venir. (from Müller 1964: 84) ‘I knew that he would come.’
Interestingly, Squartini (2004: 889) shows that Spanish does not admit the latter two uses: (22)
a. * Si debía/debiera/debe llover, podríamos ir en coche. ‘If it rains we could go by car.’ b. * Esa actitud debía/debería influir en los demás. ‘This attitude was to influence the others.’
As for the reportive function, Squartini (2004: 887–890) states that, unlike devoir, Spanish deber only allows for a “non-factual reportive” function. More concretely, he observes that deber can be a reportive marker with tenses expressing a “lower degree of assertiveness”, such as the imperfect (23a) and the conditional (23b) in Olbertz’s (1998: 395) examples. (23)
a. Debía haber llegado a casa hace una hora. ‘He should have arrived home one hour ago.’ b. No deberías estar en Madrid hasta dentro de quince minutos por lo menos. ‘You should not be in Madrid any earlier than at least fifteen minutes from now.’
It is not clear, however, whether it is the reportive dimension that is at stake in the examples in (23). In (23), counterfactual deber is reportive in that it reports on the promise to arrive (23a) or the agreement to meet (23b) at a certain place at a certain moment. Such a report does not contain an evaluation of the likelihood, and is very close to the deontic values of deber. More
198 Evidentiality and transparency with modals fundamentally, the question arises whether a counterfactual can have a reportive reading.175 Traugott (1989: 41–42) also comments on the relation between evidentiality and epistemic modality when she discusses the development of English should from sculan. Originally this verb had a deontic reading, expressing financial and moral obligation – like Spanish deber –, but soon developed a reportive or, more specifically, a quotative function, as in (24). (24)
Mænige sæden the hit geseon sceoldan. (from Traugott 1989: 41) ‘Many said who supposedly had seen it.’ [that it seen should]
What is interesting about this construction is that it shows that in some cases the reportive reading may have begun to be used before the first inferential one appeared. In English should there is thus a path from deontic modality over an evidential reading based on hearsay to an epistemic, more speaker-oriented reading. Mortelmans (2001), for her part, shows that the elaborate German modal system has a strongly evidential but weakly epistemic modal sollen ‘must/should’, as opposed to the weakly evidential but strongly epistemic readings of müssen ‘must’.176 The quotative (i.e. reportive) use of sollen expresses a meaning based on a skeptical interpretation on the part of the speaker, suggesting that the source of information is not very reliable. The lack of real commitment is emphasized by the speaker’s ability to distance him/herself from the reported state of affairs by means of speaker-oriented particles, such as ja ‘after all’, sogar ‘even’, or selbst ‘even’ expressing surprise (Mortelmans 2001: 138). According to Mortelmans (2001), epistemic inferential müssen, by contrast, involves strong subjective commitment, which is illustrated, amongst others, by its combination with ‘harmonizing’ modal adverbs such as sicher ‘surely’ and bestimmt ‘definitely’. Importantly, müssen is only rarely accompanied by speaker-oriented hedges. I will analyze this in further detail in Sections 4. and 5. 4. Deber: preference for inductive inferences In this section I will show that deber yields both inferential readings and non-inferential readings. After the presentation of the general tendencies (4.1.), I will deal with the inductive inferential (4.2.), epistemic (4.3.), deductive inferential (4.4.) and reportive (4.5.) readings found in the corpus.
Inductive inferences with deber
199
4.1. General tendencies A qualitative analysis of the corpus data indicates that deber most often applies to past events; and inferences are an ideal means to construe a past event. In the corpus these inferences are more often based on reasoning than on direct evidence. In addition to inferential readings, the corpus also contains epistemic readings of deber. Coates (1983: 46, 138) mentions that the English equivalent of deber, must, is accompanied by speaker-oriented hedges such as I think, I mean and I suppose in one quarter of the cases. Goossens (1999), for his part, describes a shift from an objective-inferential to a more subjectiveepistemic component in must. In the same vein, I will show that Spanish deber undergoes an evolution towards a more subjective-epistemic reading. In the next sections I will distinguish between four readings according to the mode of knowing involved. As shown in Table 36, the inferential reading of deber is most often based on inductive reasoning. Second, in a considerable number of non-deontic constructions deber also has a purely epistemic reading, i.e. the speaker is the only source. Third, deductive reasoning is not excluded with deber. Fourth, according to the corpus data, deber has not (yet) developed a reportive function, in which the information comes from other people.177 Table 36. Modes of knowing with non-deontic deber Inductive inferential
Epistemic
Deductive inferential
Reportive
Total
deber
90
54
8
0
152
%
59.21
35.53
5.26
-
100
I will deal with these four readings in the four sections below. 4.2. Inductive inference Inductive (non-logical) inferences constitute the most frequent group of non-deontic deber readings (59.2%). They are based on the speaker’s creative associations provoked by some kind of evidence. Interestingly, such an inferential judgment does not correspond to the highest speaker commitment, since it leaves open possible associations by other people. In (25), the
200 Evidentiality and transparency with modals speaker’s inductive reasoning is made explicit in the sentence preceding deber. (25)
Bueno yo... yo oí en el radio de que uno de los rehenes... había logrado escaparse por una ventana pero antes había tratado de suicidarse ¿no? lo cual me hace creer que bueno que el pobre debe de haber estado sufriendo horrores. (Habla Culta: Caracas: M15) ‘Well I.. I heard on the radio that one of the hostages succeeded in escaping through a window, but that he had tried to commit suicide before, hadn’t he?, which makes me think, well, that the poor guy must have been suffering tremendously.’
This example testifies to the three stages of induction mentioned in the previous section. First, the speaker takes in the information heard (on the radio): one of the hostages had tried to escape after a suicide attempt. Second, the information is connected to the speaker’s general knowledge and makes him/her form the hypothesis that kidnappings are generally painful and that no one attempts to commit suicide without serious reason. Third, the resulting conclusion that has incorporated this knowledge goes beyond the initial information: debe de haber estado sufriendo ‘he must have been suffering’. The speaker’s choice of deber indicates that this unique conclusion has been validated as much as possible, but is not completely certain. The inductive inferences abundantly appear in past tense constructions, as exemplified in (26), where the conceptualizer is searching for a valid interpretation of the state of affairs (s)he is confronted with. (26)
Sí. Y después cuando pasamos a mi cuarto vio la cama. Dijo: ah ¿esa es la cama de la niña? Dije: no ahí duermo yo. No dije nosotros porque... ¿no? dije: ahí duermo yo. Y se quedó callada. Ella debió pensar una cama tan grande tan grandotota para esta señora; bueno no sé qué pensaría pero algo pensó porque se quedó callada. (Habla Culta: Bogotá: M42) ‘Yes. And afterwards when we passed to the room she saw the bed. She said: ah this is your daughter’s bed? I said: No, I sleep there. I didn’t say “we” because ... you know? I said I sleep there. She remained silent. She must have thought such a large bed for this woman; in fact I don’t know what she thought but she thought something, because she remained silent.’
In (26), the conceptualizer observes a woman who remains silent when she sees the bed. This visual perception leads him/her to hypothesize what is
Inductive inferences with deber
201
going on in the head of the woman. By using deber, the conceptualizer is elaborating further on the available information in an inductive way. 4.3. Epistemic modality In 35.5% of the cases deber has a reading that is less inferential and more epistemic in nature. This rather high number of non-inferential readings points to an increasingly important epistemic qualification with evidential deber. Since this epistemic reading is a byproduct of the inductive inferential reading, it stands for the same intermediate speaker commitment to the epistemic judgment. Epistemic modality is often enhanced by means of subjective hedges such as creo que ‘I think’ or supongo que ‘I suppose’, which highlight the speaker’s belief state. In the spoken corpus data, these hedges appear in more than one fifth of the evidential/epistemic readings of deber. These hedges often make for a purely epistemic reading replacing the explicit reference to an inductive inferential process, as in (27). (27)
INF: No son... no es como para enorgullecerse los políticos de Chile. No sé cómo será en otras partes. Me imagino que en otras partes debe ser peor todavía; pero por lo menos en Chile no es para... pa' eno... enorgullecerse. (Habla Culta: Santiago: M4) ‘INF: No they are, you cannot be proud about the politicians in Chile. I don’t know how it is elsewhere. I imagine that elsewhere it must be even worse; but at least in Chile it is not something to be proud about.’
Since, in (27), the hedge me imagino que ‘I imagine that’ is a substitute for evidence to draw upon, debe ser has a more epistemic reading. Yet, not all constructions with speaker-oriented hedges automatically receive an epistemic reading without an inferential dimension, as shown in (28). (28)
INF A: hay más incentivo intelectual ¿ah?, yo creo aquí en Santiago por lo menos. INF B: Claro. ENC: Esto es lo que piensa la gente. INF A: Y yo creo que debe ser cierto ¿ah? en cierta manera.... INF B: Sí es cierto es cierto.... INF A: porque aquí está todo. (Habla Culta: Santiago: M6)
202 Evidentiality and transparency with modals ‘INF A: There are more intellectual incentives, ah, at least here in Santiago, I think INF B: definitely, ENC: This is what the people think, INF A: I think that it must be true, in some way INF B: yes, it is true, it is true. INF A: Because here you have everything.’ Example (28) shows that the borderlines between autonomous epistemic assessment of likelihood and inductive inferences are sometimes fuzzy. In the first line, speaker A makes an epistemic statement with yo creo ‘I think’. Then, creo que debe ser ‘I think that it must be’ emphasizes that the general opinion about the intellectual life in Santiago de Chile is likely to be true. Finally, the speaker communicates the evidence (s)he had in mind. On the one hand, the deber statement can be based on the speaker’s autonomous assessment of the likelihood. On the other, given the evidence, this statement can be seen as based on inductive inference. Note that these two modes of knowing do not imply a different degree of commitment to the likelihood: both stand for intermediate commitment, without absolute certainty. The excerpt in (28) contains two more elements that I still have to comment on: (i) one could argue that a causal subclause such as porque aquí está todo ‘because you have everything here’ also invites logical deduction, (ii) the presence of esto es lo que piensa la gente ‘this is what the people think’ right before debe ser cierto raises the question of a possible reportive reading of deber. I will first discuss logical deduction with deber (4.4.), and then turn to the reportive readings (4.5.). 4.4. Deductive inference In the 5.2% of occurrences of deber invoking deductive inferencing, the speaker expresses a stronger commitment to the likelihood that the proposition is true, as shown in (29) and (30). (29)
Me acuerdo que yo escribí por el periódico decía: Si existe una manera venezolana de tocar piano.... ya... debe haber desaparecido porque yo no conozco esa manera. (Habla Culta: Caracas: M2) ‘I remember that I wrote for the newspaper, I said: If there exists a Venezuelan way of playing the piano… it must already have disappeared, because I don’t know this way [of playing the piano].’
In (29), the speaker logically deduces that, in case there existed something like a Venezuelan way of playing the piano, it must have disappeared. The minor premise is that the speaker does not know this way of playing piano,
Inductive inferences with deber
203
the major premise is that (s)he knows everything about pianos. In the same vein, in (30) speaker B strengthens speaker A’s creo que ‘I think’ statement by means of deber. (30)
INF A: Creo que es gas. INF B: Debe ser gas sí porque este edificio es nuevo. INF A: Sí claro el gas es barato [...] gas. INF B: No ahora todas - las calderas últimas son de gas. (Habla Culta: Buenos Aires: M22) ‘INF A: I think it is gas. INF B: It must be gas yes because this is a new building. INF A: Yes, gas is cheap INF B No, all new boilers work with gas.’
From the fact that it is a new building, speaker B deduces that the source of energy used must be gas. The minor premise is that this is a new building, and the major premise is that new buildings use gas. Here the deductive character of the inference entails a stronger commitment to the likelihood than the one conveyed by inductive inferential or epistemic belief readings. Note that the deductive readings of deber seem to come close to the deontic readings of the verb. In (30), it is as if gas were obligatory in all new buildings. While these deductive readings tend to express necessity, they downplay the deontic obligation pattern.178 The potentially deontic dimension of deber can be neutralized by a subjective hedge such as creo que, as exemplified in (31). (31)
¿y qué has sabido de Guillermo? pasando a otro tema. Pues de Guillermo... estoy bravísima con él porque no me ha escrito. Creo que mañana me debe llegar una carta. Yo le escribí hace unos ocho días. Yo no puedo decir que le escribí una carta. Le escribí fue un regaño. Espero que reaccione y en la sem... por ahí mañana o pasado mañana reciba una carta. (Habla Culta: Bogotá: M32) ‘And any news about Wilhelm? Turning to another topic. Well Guillermo… I am angry with him because he didn’t write me. I think that tomorrow a letter should reach me. I wrote him a week ago. I cannot say that I wrote him a letter. What I wrote was a reprimand. I hope that he reacts and this week tomorrow or the day after tomorrow I receive a letter.’
In (31), the speaker expects a letter from her friend Guillermo. She thinks that he will reply to her last letter. In fact, the speaker considers the expected arrival of Guillermo’s letter as a logical consequence of her own
204 Evidentiality and transparency with modals letter. Yet, since logical deduction would be too strong in this case, the speaker weakens her inferential statement by means of creo que ‘I believe that’. Thus, rather than enhancing the likelihood of the deber statement, the hedge creo que reduces the degree of speaker commitment to the claim. 4.5. Reportive readings It remains to be investigated whether Spanish does not admit a reportive reading of deber. From the corpus data it appears that deber hardly allows the speaker to transmit someone else’s inference. Recall that French devoir can have a reportive function in a meteorological construction such as (32a). In order to verify whether Spanish deber can have the same function, I have done an Internet search on según las previsiones meteorológicas ‘according to the forecasts’. The search engine Google has found only one example of deber similar to the French, which is reproduced in (32b). By contrast, there are numerous examples of poder in the same context, as in (32c). (32)
a. D’après les prévisions météo, le temps doit s’améliorer demain. (from Kronning 2001:76) ‘According to the forecasts, the weather will get better tomorrow.’ b. Según las previsiones meteorológicas debe haber un cambio a mejor para el fin de semana. (http://usuarios.lycos.es/anjana/carta019.html) ‘According to the forecasts, there will/must be a change to better weather for the next weekend.’ c. Según las previsiones meteorológicas, pueden caer lluvias de 30 litros en una hora en las provincias de Girona, Tarragona y Barcelona. (El Mundo, 15-10-2003) ‘According to the forecasts, there may be showers of 30 liters per hour in the provinces of Girona, Tarragona and Barcelona.’
I have not come across similar examples in the spoken corpus. Moreover, other constructions such as debe ser que in (33) do not qualify for a reportive function in Spanish. (33)
Inf. A.–... y parece que... Inf. B.–... esta.... matica de café, que le dicen después también Inf. A.–... que se murió... Inf. B.–... aja...
Inductive inferences with deber
205
Inf. A.–... entonces debe ser que Tordito... que le sucedió una cosa mala también. Inf. B.– Sí, exacto. Inf. A.– Sí, hizo un mal negocio, o algo así. (Habla Culta: Caracas: M25) ‘Inf. A: ‘and it seems that’, Inf. B: this settlement, that they said afterwards; Inf. A; that he died; Inf B: aja...; Inf. A: well it must be that Tordito suffered something bad too; Inf. B: Yes exactly; Inf. A: Yes, he had a bad business problem, or something like that.’ The judgment expressed by debe ser reflects the speaker’s inference, not his/her report of an inference made earlier by another person. It is not possible to interpret it as ‘it is said that’, despite the que-clause with a finite verb in the past, which in other contexts favors such an interpretation. On the Internet, however, I have found a few deber examples that go in the direction of adding a reportive dimension to the overall reading. (34)
No he leído Pan, debe ser un libro hermoso, y puedes envíarmelo. Tengo hambre de buena literatura; mejor dicho, de hermosa literatura. (Internet example, Carta de Ricardo Flores Magón. 23 de mayo 1922) ‘I have not read Pan... it [must be/is said to be] a beautiful book, and can you send it to me[?]. I am hungry for good literature; in other words, beautiful literature.’
In (34), deber keeps its original evidential reading. Yet, the revolutionary captive who writes this letter could also be echoing the positive opinion concerning the book stated by somebody else. In this case, there is no real speaker commitment to the epistemic judgment that the book is possibly beautiful. Though not impossible, such an incipient hearsay reading is not at all productive at the present stage. In (35), the positive judgment is endorsed by the speaker; it could be a reported judgment, although it does not solely derive from reported information. (35)
Con respecto al de las TTSS 2, aun no lo tengo, pero te tengo que decir que por lo que me dice y manda Señor Oni debe ser un libro maravilloso. Con toda la información sobre las grandes bestias de este espantoso lugar y las historias de cada uno de ellos. Creo que es muy bueno también. (http://www24.brinkster.com/sogetsu/vertema.asp?tema=521&pag)
206 Evidentiality and transparency with modals ‘With respect to the one of TTSS2 I don’t have it yet, but I have to tell you that according to what Mr Oni says and sends me, it must be a marvellous book. With all information about the big monsters of this terrible place and the stories of each one of them. I think it is very good too.’ In (35), the inference may be made by Mr. Oni. The book reported to be commented on by Mr. Oni, is then presented to the speaker as a splendid work. The speaker him/herself may also be responsible for the inference. The endorsement by means of creo que... también corroborates the double reading: on the one hand, this statement can express the speaker’s support of someone else’s inference, on the other hand, it can be a confirmation of the speaker’s own inference, which in any case coincides with Mr. Oni’s opinion. In sum, reportive readings are not regularly available with deber, although in some cases the verb is on the verge of reporting someone else’s inference. 5. Tener que: preference for deductive inferences In the literature tener que is attributed the strongest speaker commitment and the highest likelihood (cf. Gómez Torrego 1999; Fernández de Castro 1999). I will link this commitment to the prominent position of deductive inferences. I will first present the general tendencies of the use of tener que (5.1.), before I pass on to describe the different readings in sections (5.2.), (5.3.), (5.4.) and (5.5.). 5.1. General tendencies The most important mode of knowing with tener que is deductive reasoning, while inference from direct evidence is extremely rare. Table 37 gives an overview of the evidential/epistemic readings of tener que. More than half of the occurrences involve deductive inferences, while epistemic and inductive inferential readings each cover about one quarter.
Deductive inferences with tener que
207
Table 37. Modes of knowing with non-deontic tener que Deductive inferential
Epistemic
Inductive inferential
Reportive
Total
tener que
31
15
14
0
60
%
51.67
25.00
23.33
-
100
The fact that tener que is a more recent development than the other modals (Silva-Corvalán 1995) leads me to assume that tener que has not undergone the same evolution towards a speaker-oriented epistemic reading. Olbertz (1998), for her part, argues that tener que does not belong to inferential modality but expresses the speaker’s “unconditioned” belief. I will show that this is untenable in view of the different modes of knowing involved in this modal. 5.2. Deductive inference In 51.7% of the cases, tener que has a deductive inferential reading, which involves a logical relation between the premises and the conclusion. Deductive inferential tener que is often accompanied by a conditional, causal or temporal clause with an explicit minor premise, as in (36). (36)
Pero si [la música] tiene ciento dos años y todavía se conoce y se... y se toca tiene que tener calidad ¿verdad?. ... exacto.... y el músico que lo compuso tenía que ser de calidad. (Habla Culta: Caracas: M7) ‘But insofar as the music is one hundred and two years old and it is still known and it is played, it must be of quality, true? ... exactly ... and the musician who wrote it must be of quality.’
The speaker argues that insofar as century-old music is still played, one can conclude that it meets high quality standards. The minor premise is explicitly mentioned in the si-clause, while the major premise is that only good things last. In (37), deductive inferential tener que combines with a causal porque-clause. The inferential statement is introduced by a si-clause situated on a higher syntactic level.
208 Evidentiality and transparency with modals (37)
... Claro que te digo que si las llevó a un buen joyero a España el joyero le tiene que haber dicho que eran muy baratas porque eran azulísimas, Claudia. (Habla Culta: Mexico: M5) ‘Sure, I am telling you that if he brought them to a jeweler in Spain, the jeweler must have said to him that they were very cheap because they were very blue, Claudia.’
In (37) the speaker formulates a deductive conclusion on the basis of information about the jeweler and the jewels themselves. Other examples, such as the one in (38), relate a porque clause to the context preceding the tener que statement. (38)
Lucaks no puede por su fe marxista este ... permitir ni la más mínima ... ataque o cosa contra la historia para él eso es lo más importante. Bueno pues este ... a él le tiene que gustar más un realista porque el realista ... se agarra a la realidad y la realidad en última instancia es la historia. (Habla Culta: San Juan: M4) ‘Because of his marxist belief, Lucaks cannot tolerate the slightest attack or thing against history, which is the most important thing for him. Well, he.. he must prefer a realist because the realist clings to reality and reality in the end is history.’
There is a direct logical relation between the tener que statement and the evidence given in the causal subclause introduced by porque ‘because’. In sum, all examples presented here express a strong commitment to the likelihood that the proposition is true. These likelihood readings are determined by the logical relation between the evidence and the statement. Interestingly, this is an alternative to Olbertz’ (1998) claim that tener que is unconditioned by inferences. 5.3. Inductive inference When the speaker’s own creative reasoning plays a role in reaching a conclusion, inductive inference wins out from logical deduction. The corpus data of tener que show that inductive inference is at issue in more than 23% of the readings. In these readings, the degree of speaker commitment expressed is somewhat lower than that of the deductive inferential ones.
Deductive inferences with tener que
(39)
209
Yo de agricultura no entiendo nada oye pero por los resultados se ve que Chile tiene que tener unos problemas tremendos. (Habla Culta: Santiago: M9) ‘I don’t know anything about agriculture, hey, but through the results one observes that Chile must have a couple of tremendous problems.’
In (39), the state of affairs the speaker is observing around him/her leads to the conclusion that Chile must have a lot of problems. A similar view holds for (40). (40)
Es que por desgracia yo no... yo descuento que incluso la inmensa mayoría del pueblo norteamericano sea enemigo de esas estupideces. Claro tiene que serlo porque yo no me explico que una inmensa mayoría esté contagiada de locura no creo en eso; al contrario creo que la comunidad en general es el depósito de la sensatez. (Habla Culta: Bogotá: M2) ‘Unfortunately I don’t, I conclude that the overwhelming majority of the North-American people is against these stupidities. Yes, it must be so, because I can’t understand that an overwhelming majority would be infected by madness, I don’t believe in that; on the contrary I think that society in general is the safeguard of common sense.’
In (40) the speaker’s reasoning is explicitly mentioned by means of descontar ‘to conclude’, explicarse ‘to understand’ and creer ‘to think’. Rather than a logical relation, it is the speaker’s creative reasoning that determines the conclusion. The contextual elements show that the commitment of the speaker is lower with inductive inferences than with the purely deductive ones. In (41), the speaker’s reasoning with tener que is marked by means of subjective hedges, which help avoid an exclusively deductive view of what is said. Instead, the speaker is put at the center of the creative process of inductive inferencing. (41)
ENC: Dime una cosa: en términos generales así, la obra... por qué esta obra... está situada en la guerra civil ¿ah? INF A: Mm... mm. ENC: Ahora si se publicó en España yo me imagino que esta obra tiene que ser... eh... si se le pone color político ¿no es cierto?.... INF A: Sí.
210 Evidentiality and transparency with modals ENC: ... tiene que ser a favor del régimen actual. (Habla Culta: Santiago: M2) ‘ENC: Tell me something: in general terms, this work ... why this work is situated in the civil war ah? INF: Mm.. mm. ENC: Well if it was published in Spain I imagine that this work must be .. eh .. if one gives it a political color ... isn’t it? INF A: yes INF it must be in favor of the present regime.’ The speaker elaborates on his/her own knowledge about Spanish politics and infers that the author of the books is writing in favor of the present regime. The hedge weakens the strength of the conclusion and eliminates two other readings: (i) it blocks the possibility of a deductive inferential reading, and (ii) turns an otherwise deontic reading of obligation into an inductive one. It is important to stress that with hedges such as me imagino que or creo que, the reading of tener que does not shift to an epistemic reading, but to an inductive one. This contrasts with the deber constructions which most often shift directly to an epistemic reading. This supports a gradient view of evidentiality and epistemic modality from tener que over deber to epistemic poder. In sum, around a quarter of the non-deontic tener que constructions have an evidential reading based on inductive inference. As a result, speaker commitment is weaker than with deductive inference. 5.4. Epistemic modality For another 25% of the cases, tener que yields an epistemic reading without a clearly inferential dimension, which Olbertz (1998: 410) calls a “belief” reading.179 Rather than an overall “belief” reading, tener que just conveys an epistemic reading as a byproduct of the original deductive inferential reading. But the epistemic belief readings in (42) and (43) can by no means be regarded as dominant. (42)
... esa tremenda disciplina que hay en el ejército y que tú tienes... tienes que haberla sentido en la armada; eso de obedecer ciegamente... (Habla Culta: Santiago: M16) ‘This terrible discipline that rules in the army and that you must have felt in the war fleet; this blindly obeying.’
Deductive inferences with tener que
211
In (42) the speaker assumes that the interlocutor has undergone the strict discipline without relying on explicit evidence or additional reasoning. The speaker is simply committed to his own belief. The same pattern can be observed in (43): (43)
ya voy a comenzar los ensayos del... de mi última pieza que se llama Todo lo contrario; es... una pieza sobre la reversión petrolera la [sic] tiene que... que haber preocupado en... en... en mayor o menor medida a todos los venezolanos... ¿no? que... que tienen uso de razón o sea que piensan. (Habla Culta: Caracas: M20) ‘I will start the rehearsals of my last play which is called Todo lo contrario; it is a play about the recession of the oil sector which must have alarmed to a greater or lesser degree all Venezuelans... you know? all those who use common sense, in other words those who think.’
In (43), the speaker takes it for granted that people with common sense must have been alarmed by the recession in the seventies. The statement is not linked up with any evidence from the present context. Furthermore, there is no risk of a competing deontic reading (which would have been formulated by means of the counterfactual conditional debería haber preocupado). In sum, it has become clear that tener que can also yield an epistemic reading that only refers to the speaker’s or the conceptualizer’s autonomous assessment of the proposition. In this reading, the degree of commitment is similar to that of a deductive inferential reading. 5.5. Reportive readings In line with deber, the tener que statement can be reported only if it is introduced by a verb that explicitly refers to other people’s assessment of the state of affairs, as shown by se ha concluido ‘it has been concluded’ in the Internet example (44). (44)
A partir del supuesto que los Reyes Magos procedieran del oriente, se ha concluido que su punto de origen tiene que haber sido Babilonia.(http://www.iac.es/galeria/mrk/Magos.htm) ‘Starting from the assumption that the Magic Kings came from the East, it has been concluded that their origin must have been Babylonia.’
212 Evidentiality and transparency with modals Tener que is not used for reporting hearsay knowledge. No examples of the type made up in (45) were found in the corpus or on the Internet. (45)
?
Según las previsiones meteorológicas tiene que haber un cambio a mejor para el fin de semana. ‘According to forecasts, there will/must be a change to better weather for the next weekend.’
The construction in (45) yields a deontic reading which conveys deontic force emanating from an unspecified entity. This is a general characteristic of tener que: if the speaker is not involved, the construction easily shifts to a deontic reading which most often expresses obligation emanating from another entity than the speaker, in this case an external force. We can thus conclude that tener que has not yet reached the same stage in the evolution as deber. 5.6. Discussion In the previous sections I have argued that the inferential processes of deber and tener que result in a stronger epistemic commitment. The low degree of speaker commitment expressed by epistemic poder can be related to the fact that this verb most often does not rely on inferential processes. This leads me to conclude that unlike what happens with the evidential verbs dealt with in the other parts of this volume, evidentiality does play a role in the epistemic judgments made by modal verbs. Moreover, the evidential qualification determines the strength of the epistemic qualification. In the previous sections each of the three epistemic/evidential modal verbs has been argued to have a core reading. I have shown that deber most often correlates with inductive inference and tener que with deductive inference. Accordingly, these three different sources for the statement entail the weak, intermediate and strong speaker commitment usually associated with poder, deber and tener que respectively. The hedges used in one fifth of the evidential deber constructions suggest that this verb experiences a tendency towards construing a more epistemic view of the state of affairs. The hedges used with tener que reveal a reading based on inductive instead of deductive inferencing. This is in line with the gradient shift from deductive inferential modality over inductive inferential modality to epistemic modality. The gradience from poder over deber to tener que can also be linked to a different distribution of subjects and infinitives. It is generally assumed
Transparency 213
that poder, deber and tener que are fully transparent. Although there are no serious restrictions on the subjects and the infinitives of these verbs, the distribution of these verbs still may display different tendencies. Starting from the observation that poder has developed most strongly into an epistemic verb, one can assume that it has become an independent epistemic form that combines with the widest array of verbs. According to Silva-Corvalán (1995), the ideal distribution of the epistemic readings of poder includes stative verbs and agentless subjects. Yet, rather than dwelling on the ideal context, it is useful to take into account the real use of the different verbs. I expect, on the one hand, that poder more readily combines with action or change-of-state verbs than deber and tener que and, on the other hand, I hypothesize that poder combines with a wider group of subjects than the other two verbs. The different distribution of the subject and the infinitive will point to different degrees of transparency. 6. Transparency In this section, I will compare the distribution of epistemic/evidential constructions with that of the dynamic/deontic ones. The section is structured as follows: I will first present the classification criteria used for the calculations presented in the tables (6.1.). Then I will discuss the quantitative data on the infinitives and the grammatical subjects of poder (6.2.). Next I will examine deber (6.3.) and tener que (6.4.). 6.1. Classification The classification of the infinitives is in line with the labels used for the analysis of parecer, amenazar and prometer: (i) action verbs (action), (ii) aspectual verbs (aspect), (iii) change-of-state verbs (change), (iv) communication verbs (communication), (v) copular verbs and other attributive verbs (attribution), (vi) modal verbs (modal), (vii) mental activity verbs (mental), (viii) perception verbs (perception) and (ix) verbs expressing psychological reaction (psychological).180 As for the subjects, I follow the same classification used for amenazar and prometer (cf. Delbecque 2003: 383). Besides the absolute frequency of the combinations, the tables present (i) the relative frequency of the infinitive mentioned in the first column with the four readings and (ii) the relative frequency of the infinitive in
214 Evidentiality and transparency with modals comparison with other infinitives that combine with the reading mentioned in the heading.181 6.2. Infinitives and subjects with poder The epistemic constructions with poder, on the one hand and the dynamic and deontic constructions, on the other, have a different distribution. Epistemic poder combines most with attributive infinitives (51.32%), while this is not the case with the dynamic (6.48%) and deontic readings (9.22%). From another point of view, nearly half of the attributive infinitives (49.02%) entail an epistemic reading of poder, while in 23.18% and 10.61% of the cases an attributive infinitive leads to a dynamic or a deontic reading respectively. Table 38. Infinitives with poder epistemic
%
Action
201
8.58 (i) 29.39 (ii)
Aspect
24
Change
56
Communication Attribution Modal
12
25.26 3.51 42.75 8.19 2.29 1.75 49.02 51.32 0.00 0.00 6.23 2.63 6.06 0.88 20.00 0.58
Mental activity Perception Psychological reaction None
351 0 18 6 4 12
5.41 1.75
dyna -mic epist 162 9 25 17 123 0 22 0 3 9
%
dynamic
%
deontic
%
6.91 43.78
1489
63.55 58.16
491
20.96 59.59
9.47 2.43 19.08 6.76 3.25 4.59 17.18 33.24 0.00 0.00 7.61 5.95 0.00 0.00 15.00 0.81
46
48.42 1.80 28.24 1.45 70.55 14.41 23.18 6.48 0.00 0.00 73.70 8.32 76.77 2.97 45.00 0.35
16
16.84 1.94 9.92 1.58 23.90 15.17 10.61 9.22 0.00 0.00 12.46 4.37 17.17 2.06 20.00 0.49
4.05 2.43
37 369 166 0 213 76 9 155
69.82 6.05
13 125 76 0 36 17 4 46
20.72 5.58
Transparency 215
In the dynamic and deontic poder constructions, by contrast, almost 60% of the infinitives express an action. Interestingly, in more than 63% of the cases, an action verb leads to a dynamic reading of poder, while such a verb entails a deontic reading in more than 20% of the occurrences. Other infinitives that easily provoke a dynamic reading are the aspectual ones (48%), the communication verbs (70.55%), and the mental activity (73.7%) and perception (76.77%) verbs. The absence of an infinitive also favors a dynamic reading (69.82%). Interestingly, not one group of infinitives leads to a deontic reading of poder in more than 25% of the cases. By contrast, in addition to the attributive infinitives (49.02%), aspectual infinitives (25.26%) and change-of-state verbs (42.75%), such as caer ‘to fall’, readily combine with epistemic poder. The above description confirms Silva-Corvalán’s claim (1995) that stative (attributive) infinitives favor an epistemic reading with poder, although it is shown that change-of-state verbs frequently yield such a reading too. On the other hand, epistemic poder combines with an action verb in nearly 30% of the cases, but such an infinitive only leads to an epistemic reading of poder in less than 10% of the cases. Let us now look at the distribution of the grammatical subjects with poder presented in Table 39.182 Some 35% of all subjects with epistemic poder are human individuals such as gente ‘people’, which is less than half the percentage observed for dynamic (79%) and deontic (83.7%) readings. Of all individual subjects, 63.8% and 21.79% lead to a dynamic and a deontic reading of poder respectively, while only 7.73% entail an epistemic reading. The collective subjects more often invite a dynamic reading (50.59%). By contrast, the subjects expressing a process, e.g. obras ‘works’, or a state of affairs, e.g. problema ‘problem’, are far more frequent with epistemic poder than with dynamic and deontic poder. Of all subjects denoting a state of affairs, 59.44% provoke an epistemic reading; for concrete objects and processes this is the case in 44.83% and 47.53% of the cases. Note that 53.29% of the poder constructions that lack an overt subject have a dynamic reading, while only 26.09% receive an epistemic reading. The low number of covert subjects in deontic constructions (9.12%) has to do with the easily identifiable second person singular of numerous deontic forms. The speaker then directly addresses the obligation to the interlocutor. Third person singular covert subjects, by contrast, often refer to what is said before the modal statement is uttered.
216 Evidentiality and transparency with modals Table 39. Grammatical subjects with poder epistemic
%
Individual
245
Concrete object Process
26
7.73 35.77 44.83 3.80 47.53 15.47 59.44 12.41 29.41 0.73 30.30 1.46 40.00 2.34 20.59 5.11 16.67 0.15 38.24 1.90 26.09 20.88
106
State of affairs Notion
85
Location
10
Semiotic
16
Collective
35
State of mind Nonspecified None
1
5
13 143
dynamic epist. 212 9 37 16 2 5 8 12 2 6 62
%
dyna mic
%
deontic
%
6.69 57.14 15.52 2.43 16.59 9.97 11.19 4.31 11.76 0.54 15.15 1.35 20.00 2.16 7.06 3.23 33.33 0.54 17.65 1.62 11.31 16.71
2023
63.80 78.96 24.14 0.55 27.80 2.42 25.17 1.41 35.29 0.23 51.52 0.66 32.50 0.51 50.59 3.36 50.00 0.12 26.47 0.35 53.47 11.44
691
21.79 83.76 15.52 1.09 8.07 2.18 4.20 0.73 23.53 0.48 3.03 0.12 7.50 0.36 21.76 4.48 0.00 0.00 17.65 0.73 9.12 6.06
14 62 36 6 17 13 86 3 9 293
9 18 6 4 1 3 37 0 6 50
In sum, Silva-Corvalán’s (1995) ideal context is corroborated to a large extent. Epistemic poder combines less often with action verbs and human individual subjects than deontic and dynamic poder. This is also confirmed by the fact that, as far as the type of infinitive and the subject are concerned, poder constructions that combine an epistemic and a dynamic reading occupy an intermediate position between epistemic and deontic/dynamic readings. 6.3. Infinitives and subjects with deber I will now examine the infinitives and the subjects that combine with the evidential and the deontic deber constructions. We will start with the corpus results of the infinitives summarized in Table 40.183 Table 40 highlights that all infinitives combine more with deontic deber than with evidential deber. The highest probability of a deontic reading is
Transparency 217
found with action, e.g. dar ‘to give’, and communication verbs (both 96%). By contrast, more than 30% of the attributive infinitives lead to an epistemic reading of deber. Note that more than 70% of the infinitives that combine with evidential deber are attributive ones like ser ‘to be’. Table 40. Infinitives with deber
Action
evidential/ epistemic 18
Aspect
4
Change
11
Communication
2
Attribution
136
Modal
0
Mental activity
14
Perception
2
Psychological reaction None
2 0
%
Deontic
%
3.91 9.52 18.18 2.12 18.64 5.82 3.45 1.06 31.26 71.96 0.00 0.00 15.38 7.41 16.67 1.06 28.57 1.06 0.00 0.00
442
96.09 45.95 81.82 1.87 81.36 4.99 96.55 5.82 68.74 31.08 0.00 0.00 84.62 8.00 83.33 1.04 71.43 0.52 100.00 0.73
18 48 56 299 0 77 10 5 7
More than 68.74% of the attributive infinitives correlate with a deontic reading; many of these forms are passive ones. I will now examine whether the discrepancy between attributive verbs and action verbs is also reflected in the kinds of subjects that are found with deber. Table 41 gives us an overview.184 Although 37% of the grammatical subjects of evidential deber are individuals, less than 14% of all individual subjects entail an evidential reading, in contrast with 86% for deontic deber. Subjects that denote a process or a result, such as cumplimiento ‘completion’, most frequently yield a deontic reading of deber (96.2%) and constitute 18% of all subjects with this reading. This high frequency can be related to the frequent passive infinitives. Note that up to 38% of the evidential deber constructions do not have an overt subject. The absence of an overt subject leads to an evidential reading
218 Evidentiality and transparency with modals in more than half of the cases (53.68%). Such utterances often refer to the whole state of affairs. Table 41. Grammatical subjects with deber
Individual
evidential/epistemic 70
Concrete object Process
4
State of affairs Notion
6 0
Location
2
Semiotic
7
Collective
5
State of mind
1
NonSpecified None
14
7
73
%
deontic
%
13.83 37.04 4.40 2.12 3.80 3.70 15.00 3.17 0.00 0.00 12.50 1.06 23.33 3.70 5.68 2.65 11.11 0.53 53.85 7.41 53.68 38.62
436
86.17 45.32 95.60 9.04 96.20 18.40 85.00 3.53 100.00 2.60 87.50 1.46 76.67 2.39 94.32 8.63 88.89 0.83 46.15 1.25 46.32 6.55
87 177 34 25 14 23 83 8 12 63
In conclusion, although evidential/epistemic deber combines with an attributive verb in more than 70% of the cases, this does not mean that an attributive infinitive automatically entails an evidential reading. Moreover, the percentage of human subjects of evidential and deontic deber does not differ a lot. What differs is the fact that evidential deber appears more often in constructions without an overt subject than with a human individual overtly expressed subject. Notice that these two characteristics are more prominent with evidential deber than with epistemic poder constructions. This can be explained by the speaker’s strategy to avoid agentive patterns typical of the deontic reading.
Transparency 219
6.4. Infinitives and subjects with tener que The distribution of the infinitives found with tener que is presented in Table 42.185 Table 42. Infinitives with tener que
Action
evidential/ epistemic 4
Aspect
0
Change
6
Communication
2
Attribution
43
Modal
0
Mental activity
4
Perception
0
Psychological reaction None
1 0
%
deontic
%
0.46 6.67 0.00 0.00 8.82 10.00 2.22 3. 33 14.88 71.67 0.00 0.00 3.33 6.67 0.00 0.00 7.69 1.67 0.00 0.00
868
99.54 59.90 100.00 2.83 91.18 4.28 97.78 6.07 85.12 16.98 0.00 0.00 96.19 6.97 100.00 1.38 92.31 0.83 100.00 0.76
41 62 88 246 0 101 20 12 11
All types of infinitives combine much more with deontic tener que than with evidential tener que. Less than 15% of the attributive infinitives occur in an evidential reading. On the other hand, when we look at the most frequent infinitives found with tener que, the difference between the action infinitives (60%), e.g. elaborar ‘to elaborate’, combining with a deontic reading and the attributive infinitives (71.67%), e.g. haber ‘to be’, combining with an evidential reading is greater than with deber. Evidential tener que does not easily combine with action verbs (0.46%), perception and aspectual verbs (0%), while the attributive infinitives are as frequent as with evidential deber (71%). Let us now examine whether similar differences are found with the subjects of tener que, the distribution of which is summarized in Table 43.186 The deontic construction with tener que has most often a human subject (66%), and 97.75% of the human subjects pattern with a deontic reading.
220 Evidentiality and transparency with modals While only 2.25% of the human subjects crop up in an evidential construction, subjects like tú ‘you’, nevertheless make up 36% of all evidential constructions. Although the relative frequency of covert subjects is high with evidential verbs (36.7%), most of these subjects are found in a deontic construction (88.04%). Table 43. Grammatical subjects with tener que
Individual
evidential/epistemic 22
Concrete object
5
Process
1
State of affairs
1
Notion
1
Location
3
Semiotic
3
Collective
1
State of mind
0
Pronoun
1
None
22
%
deontic
%
2.25 36.67 7.58 8.33 0.60 1.67 7.69 1.67 10.00 1.67 27.27 5.00 15.00 5.00 2.22 1.67 0.00 0.00 7.69 1.67 11.96 36.67
956
97.75 65.96 92.42 4.21 99.40 11.46 92.31 0.83 90.00 0.62 72.73 0.55 85.00 1.17 97.78 3.04 100.00 0.14 92.31 0.83 88.04 11.18
61 166 12 9 8 17 44 2 12 162
In sum, just like deber, evidential tener que predominantly combines with attributive verbs, although such an infinitive is no guarantee for an evidential reading. This leads me to conclude that the deontic reading allows for all kinds of combinations with tener que.
Transparency 221
6.5. Summary Table 44 summarizes the most important elements of the above presented distributions and allows us to compare the different verbs. The percentages are given in the same order as in the previous tables. Table 44. Summary of poder, deber, tener que Percentages
Epistemic poder Evidential deber Evidential tener que Deontic poder Deontic deber Deontic tener que Dynamic poder
action verb
attributive verb
individual subject
% 8.58 29.34 3.91 9.52 0.46 6.67 20.96 59.52 96.09 45.95 99.54 59.90 63.55 58.12
% 49.02 51.24 31.26 71.96 14.88 71.67 10.61 9.21 68.74 31.08 85.12 16.98 23.18 6.48
% 7.73 35.77 13.83 37.04 2.25 36.67 21.79 83.76 86.17 45.32 97.75 65.96 63.80 78.96
object SoA process % 50.6 31.68 7.70 8.99 5.29 11.67 9.26 4.00 92.26 30.97 94.71 16.51 25.70 4.38
None
% 26.09 20.88 53.68 38.62 11.96 36.67 9.12 6.06 46.32 6.55 88.04 11.18 35.29 11.44
First, epistemic poder combines with a higher number of infinitives expressing an action and a lower number of attributive infinitives than evidential deber and tener que. Action verbs almost always pattern with a deontic reading of deber and tener que, but the attributive infinitives with deber and tener que do not automatically imply an evidential reading. Interestingly, the 31.26% of attributive infinitives with deber indicate that this verb is more advanced in downplaying the deontic patterns. In general, both action verbs and attributive predicates are increasingly infrequent with epistemic poder (8.58%/49.02%), evidential deber (3.91%/31.26%) and evidential tener que (0.46%/14.88%) in that order. The figures in Table 44 corroborate that poder more easily conveys an epistemic reading than deber and tener que express an evidential one. My hypothesis is that action verbs are less problematic for poder due to the preference of dynamic poder, and hence not of deontic poder, to combine with action verbs. A dynamic construction lacks the agentive patterns associated with deontic modality, which facilitates a possibility reading. In this
222 Evidentiality and transparency with modals context it is not surprising that when poder combines with an attributive infinitive, it has an epistemic reading in half of the cases. The distribution of the subjects shows that when poder combines with a subject denoting an object, a state of affairs or a process, it receives an epistemic reading in half of the cases. This is not at all the case for evidential deber and tener que where these subjects combine with the deontic counterparts of the verbs in more than 90% of the cases (92.26% and 94.71%, respectively). This shows that poder is more transparent than deber and tener que. Evidential deber and tener que more often lack an overt subject than epistemic poder does. Although with deber the absence of an overt subject leads to an evidential reading in more than half of the cases, no other observations can be made on the basis of the information available. 7. Conclusions The use of modal poder, deber and tener que presents different proportions of evidentiality and epistemic modality. All three verbs can involve inferential evidentiality, albeit to a different extent: while it is less crucial with poder, the evidential dimension of deber and tener que are more developed. The analysis presented above is a refinement of the previous analyses in three different ways. First, the description of the non-deontic modal readings in terms of their most important mode of knowing has underpinned the gradience of weak, intermediate and strong speaker commitment from poder over deber to tener que, as suggested by Fernández de Castro (1999). The corpus data show that these three modals have a prototypical non-deontic reading: (i) an epistemic judgment with poder corresponds to a weak commitment, (ii) an inductive inferential statement with deber conveys intermediate commitment, (iii) a deductive statement with tener que yields strong speaker commitment. Second, I have shown that the speaker commitment is not as fixed as suggested in the Spanish literature. The different modes of knowledge, I have argued, lead to different kinds of speaker commitment. Small shifts in commitment occur when the epistemic or evidential modal does not rely on its most habitual mode of knowing. For example, tener que based on inductive inference yields a weaker speaker commitment than tener que based on deductive inference. Third, the possibility of purely epistemic readings of deber and tener que illustrates that these verbs do not necessarily convey evidential read-
Conclusions
223
ings. The distinction between epistemic modals and evidential modals is thus not always as strict as assumed in the literature (cf. Anderson 1986, Nuyts 2001b). The existence of these epistemic readings correlates with the gradience from the less epistemic readings of tener que to the more epistemic ones of poder. Interestingly, this gradience is also reflected in the different distribution of subjects and infinitives with the three modal verbs.
Chapter 8 Epistemic and evidential modals: modal grounding and subjectification
In this chapter I will argue that the epistemic and evidential modals poder ‘can/may’, deber ‘must’ and tener que ‘have to’ realize modal grounding of the infinitival process. This specific form of modal grounding comes about through the diachronic and synchronic subjectification of the dynamic and deontic readings of these verbs. In considering Spanish epistemic/evidential modals as “grounding predications”, I extend Langacker’s (1990) notion of “grounding predication”, because Spanish modals include temporal and modal grounding at the same time. Unlike English modal auxiliaries, the Spanish modals have tense and mood inflection, and, hence, do not belong to a well-defined morphosyntactic category of modal auxiliaries. Yet, although the morphosyntactic restrictions on the Spanish epistemic modals are less far-reaching than those on the English modal auxiliaries, the Spanish epistemic modals present a specific, restricted distribution. In line with Cornillie (2003, 2005a, 2006), I will argue that, although formally speaking the Spanish epistemic and evidential modals cannot be considered real grounding predications, their modal grounding function and their conceptual content are very similar to the function and the content of the English ones. As for the narrowing of Langacker’s concept of grounding predications, I follow Pelyvás’ (1996) and Nuyts’ (2002) argument that grounding predications are limited to epistemic/evidential modals, and hence exclude the deontic modals. The chapter is structured in the following way. First, I will briefly present an historical view of the Spanish modals in terms of diachronic subjectification. Then I will discuss whether grounding should be approached in a purely grammatical or in a cognitive-functional way. In Section (2.) I will explain the link between grounding predications and the different types of subjectification proposed by Langacker. I also tackle the difference between the notions of content and profile, and relate them to Langacker’s dynamic evolutionary model. In Section (3.) the unavailability of a ‘modal’ infinitive will be proposed as a major criterion for epistemic modality. In Section (4.) I will comment on the debate concerning the shift from deontic to epistemic modality and propose to apply the concepts of reference-point and subjectification. The difference in subjectification accounts for the different types of epistemic modality expressed by the Spanish modals.
Diachronic subjectification
225
1. Diachronic subjectification vs. modal grounding In this section I will first discuss the diachronic subjectification of the Spanish modals (1.1.) before addressing the question of modal grounding and how it differs from temporal grounding (1.2.). I will not give a detailed diachronic account of poder and deber but confine myself to presenting new data on the historical evolution of tener que. 1.1. Diachronic subjectification 1.1.1. From deontic to epistemic poder and deber In many typologically diverse languages the epistemic meanings of modals have developed from their dynamic and deontic counterparts (Bybee et al. 1994: 202–204).187 Dynamic capacity or deontic permission and obligation are then reinterpreted in terms of weak vs. strong speaker commitment to the truth of the proposition. The shift from deontic readings to epistemic ones implies that deontic meanings are diachronically prior to epistemic readings. Yet, since modal deber and poder are not ‘homegrown’ but originate in the Latin verbs debere and posse, the genesis of epistemic meanings might have taken place before the Spanish verb forms started to be used (cf. Olbertz 1998: 246). Yllera (1980: 138–140) mentions that in the first Spanish texts, the epistemic readings of deber and poder are infrequent. Epistemic poder is observed from the 13th century onwards, while epistemic deber is not regularly used until the 15th century. Olbertz (1998: 246) refers to the Latin ancestors of the modals, namely debere and posse, which have fulfilled epistemic functions from the earliest texts on (cf. Bolkestein 1980: 123– 127). Interestingly, Olbertz (1998: 246) suggests that the reason why epistemic examples in Spanish texts are less frequent than those in Latin lies in the type of texts these constructions appear in. Many Latin texts discuss philosophical problems, while the medieval Spanish texts are most often narratives.188 Further corpus research should corroborate whether the epistemic modal readings of deber and poder come from their deontic or dynamic counterparts, but the high frequency of the latter readings in presentday Spanish in comparison with the low frequency of the epistemic ones makes this assumption highly plausible.189
226
Modal grounding and subjectification
1.1.2. The development of tener que Since the Spanish deontic and epistemic modal readings of tener que are homegrown, it is useful to consider this verb in terms of grammaticalization and subjectification. Recall that the “epistemic” reading of tener que is a recent development that deserves closer scrutiny.190 I will mainly argue that tener que has developed as an alternative for tener de. This view constitutes a revision of the explanation that the tener que construction has developed from constructions with non-finite relative clauses with or without antecedent, as in (1) (cf. Mori 1974; Olbertz 1998). I give two examples from Yllera (1980: 100,117) 191 quoted in Olbertz (1998: 250–252): (1)
a. en aquestol’ avemos mucho que gradeçer. (Fernán González. around 1250) ‘in that respect we have much to owe him.’ b. non tenemos que comer. (Alfonso XI. around 1350) ‘we do not have (anything) to eat.’
In the literature it is often stated that the development of deontic tener que supposes two shifts. First, the tener que construction is a result of the shift from possessive haber ‘to have’ in (1a) to possessive tener ‘to have’ in (1b). A second, more complicated change involves a shift from a situation in which some person has something (tener) to a situation in which something should be done (tener que). Since this is not a straightforward shift, I will first detail other aspects of tener que below. Olbertz (1998: 250) mentions that the predecessors of tener que are haber que ‘to have to’ and haber de ‘to have to/will’. The hundreds of occurrences of haber de in the Corpus del español show that this verb is by far the most common one between the 13th century and the second half of the 19th century. Moreover, this verb is polyfunctional in that it has both deontic, futural and epistemic/inferential readings. The deontic reading of haber de is not necessarily the first one. Since the future form in Romance is based on habere (cf. cantare habeo > cantaré ‘I will sing’), it is plausible that the futural reading of haber de is prior to the deontic reading. In any case, the inferential reading is not the basic reading of haber de. The three readings have in common that the actions or the propositions are not actual but potential. That is, the speaker and the hearer rely on the idea that the future will show whether the action is really accomplished or the proposition has become true.
Diachronic subjectification
227
The haber que construction does not have these three different readings, but seems to be restricted to the deontic reading. A review of the Corpus del español yields only a few examples of deontic haber que. In the 13th century example (2a), the que-complement of haber has an object function. The corpus also indicates that from the 14th century on haber que is only used in impersonal constructions, as in (2b). Ha que + infinitive is the form that precedes hay que + infinitive.192 This construction is the result of grammaticalization accompanied by subjectification: a possessive reading of a transitive construction shifts to a deontic reading of a modal construction. Interestingly, the deontic haber que construction in (2b) is performative in nature, and hence, displays speaker commitment. (2)
a. Ca al que en njnguna culpa non cahe non le han que perdonar. (Alfonso X. General estoria V. 13th century) ‘Because to the one who has no guilt they have nothing to absolve him from.’ b. ... del quarto año del Regnado deste Rey don Ordoño non ha que contar que […] si non tanto que murio en este año el papa Basilio & fue puesto ensu lugar. (Pedro Afonso de Barcelos. anonymous transl. Crónica de 1344. 14th century) ‘... about the fourth year of the Rey don Ordoño’s government one should only tell that pope Basilius died and was carried to the grave.’
The low frequency of haber que is a case in point. It shows that the haber construction with de is preferred to the one with que and that the construction with de also serves to express various readings. Now, it is not justified to postulate that tener que originates from haber que. First of all, tener que is a more recent development than haber que: although haber que + infinitive is restricted to the third person impersonal construction after the 14th century, tener que is not attested before the end of the 15th century. The fact that the tener de construction expresses a deontic reading considerably earlier than tener que supports the analysis. Moreover, an account based on tener de can refine the traditional explanation in terms of the shift of the relative clause tener trabajos que hacer to tener que hacer trabajos. The deontic readings with tener were present from the first stages of Spanish (cf. Yllera 1980: 110–116; Delport 2004). The deontic readings of tener de are first found in the 13th century, as shown in Alfonso X’ writings in (3).
228
Modal grounding and subjectification
(3)
… quiso otrosi quela parte dela labor dela yglesia que fuese en poder del obispo demandar en que cosas segaste & esto es porque el tiene de dar cuenta adios dello. (Alfonso X. Siete partidas. 13th century) ‘He also wanted that part of the work of the Church was in the hands of the bishop to ask in which things you cut and this is because he has to report on it to God.’
In this construction, tener de expresses a descriptive deontic reading in which the obligation does not stem from the speaker but from a third entity. Other tener constructions are semi-transitive. They combine the preposition de with the conjunction que, thus conveying a kind of partitive reading, as exemplified in (4).193 (4)
si el procurador fuere rebelde que non sea restituydo el señor del pleito maguer que diga que el procurador non tiene de que pagar. (Alfonso X. Siete partidas. 13th century) ‘if the attorney opposes the fact that the lawyer be refunded although he says that the attorney does not have of to pay.’ (i.e. “something for which to pay”)
From a traditional point of view, one could say that the (semi-)transitive construction in (4) refers to an intermediate stage between tener de and tener que. Such a view is only part of the story, however, since the first occurrences of tener que in the Corpus del español do not point to a (semi-) transitive relation between the que-complement and tener, as shown in (5a) and (5b). (5)
a. ... no como dize sant isidoro dela orden del a b c. que la a es primera. la b segunda la c tercera por que desta orden no tiene que hazer el gramático. (Antonio de Nebrija. Gramática castellana. 1492) ‘... not like Saint Isidoro in alphabetical order, where a is the first, b the second, and c the third since the grammarian should not work in this order.’ b. ... para hacer el monumento no tienen que desvelarse los frailes, ni para qué buscar paños, ni tapices, ni otros atavíos. (Mendieta, Jerónimo de. Historia eclesiástica indiana. 1564) ‘… to celebrate the monument, the brothers must not stay awake, neither do they have to look for clothes, carpets nor other finery.’
Diachronic subjectification
229
In the example from Nebrija’s grammar (5a), tener que has a deontic reading. The que-clause cannot be seen as a relative clause that refers to an overt or covert antecedent. This is also the case in (5b), where the intransitive infinitive desvelarse ‘to stay up’ receives a deontic perspective. These two forms are characterized by their performativity. Unlike the examples of tener de, they express direct speaker commitment to the proposed event. Further research is needed, but one could hypothesize that the shift from tener de to tener que involves speaker-oriented subjectification. This does not mean, however, that tener que cannot have descriptive readings anymore. The data from the Corpus del español indicate that until the 16th century tener de is more frequent than tener que. In the 17th century both constructions are used quite regularly. From the 18th century onwards, tener que is more frequent than tener de. In the 19th century, tener de is almost extinct. More refined corpus research should confirm this general survey of the data. The epistemic/evidential readings of tener de/que appear later than the deontic ones. Corpus data show that the tener de construction yields an epistemic/evidential reading in the 16th century, as in (6), while with tener que this reading only shows up in the second half of the 19th century, as illustrated in (7). (6)
… hay disposición y suelo para hacer una muy buena y gran ciudad, y según sus calidades y trato y contratación, yo creo que tiene de ser antes de mucho tiempo muy populosa y estimada. (Motolinía, Toribio de. Historia de los indios de la Nueva España. 1543) ‘There is provision and territory to construct a very good and big city, and according to its qualities, agreement and contract, I think that before long [the city] must be very densely populated and respected.’
(7)
El resultado de la equivocación deshecha tiene que ser desastroso. (Arenal, Concepción. Observaciones sobre la educación física, intelectual y moral de Herbert Spencer. 1856) ‘The result of this sad mistake must be disastrous.’
In sum, this diachronic overview has shown that the deontic reading of tener que appeared earlier than its epistemic/evidential counterpart. Despite the Latin origins of poder and deber, it is justified to state that in Spanish the deontic and dynamic readings precede the epistemic ones. The deontic readings of tener de crop up in the 13th century and perhaps appeared even earlier. In the Corpus del español I have not found deontic tener que con-
230
Modal grounding and subjectification
structions before the end of the 15th century. If it really had been a variety of the possessive, transitive constructions with a que-complement, the tener que construction would have been used earlier. In this respect, the description based on tener de above can be considered an alternative account of the development of tener que. No doubt more systematic quantitative research should further underpin the general description proposed here. 1.1.3. Diachronic vs. synchronic accounts The diachronic analysis presented in the previous section has confirmed that the epistemic modals show up later than the other modal readings. The result of the diachronic evolution is a wide array of different coexisting modal uses in Spanish and other languages. It is also useful to account for the different coexisting dynamic, deontic and epistemic modal readings of these modals in present-day Spanish. In the remainder of the chapter I follow Langacker’s model of subjectivity insofar as it helps us account for the synchronic uses of the modals in language in general and in Spanish in particular. I will argue that the epistemic modals are a semantic extension of the non-epistemic modals through conceptual subjectification. I will show that the Spanish epistemic and evidential modals have a reference-point function and can be considered modal grounding predications. In the next section I will address the question of temporal and modal grounding. 1.2. Modal grounding In this section I will argue that the tense marking of the Spanish modals does not have to be considered a problem for the modal grounding these verbs realize. I first discuss the difference between temporal and modal grounding (1.2.1.). Then I present Nuyts’ and Pelyvás’ views on modal grounding (1.2.2.). Finally I formulate my own position on the matter (1.2.3.). 1.2.1. Temporal vs. modal grounding Within the framework of Cognitive Grammar, the grounding of linguistic expressions occupies a prominent position: it specifies a designated entity with regard to “fundamental notions such as time, reality and referential
Diachronic subjectification
231
identity” (Langacker 2003: 7).194 Importantly, Cognitive Grammar only considers certain forms as real grounding devices: they are “a small set of highly grammaticized elements, one of which has to be chosen as the final step in forming a full nominal or finite clause” (Langacker 2002: 29). Elements with a clear grounding function for English verbal constructions are, on the one hand, tense and mood markers, and, on the other hand, modals. When tense markers or modals ground processes, they provide the process considered with a putative address with respect to reality. In English, indicative marking―just like the -s plural for nouns―provides grounding for the process considered, while modals do not have tense nor mood, as shown in (8). (8)
a. Peter already goes to school. b. Peter must [*s/Ø] go to school.
This opposition leads Langacker to argue that, as far as grounding is concerned, modal auxiliaries and tense are mutually exclusive. Since the untensed English modal auxiliaries combine modal and temporal grounding, they are the prototype of grounding predications. The English modal auxiliaries may, can and must are present tense forms. But the past tense forms might and could do not necessarily convey a reported statement about a situation in the past: they often express a future reality (9a) or a present statement about the past (9b) (the examples come from Radden 1999: 265). Unlike may and can, deontic must does not have such a past form without tense inflection; instead, the past form of have to is used, as shown in (9e). (9)
a. She may wake up – she might wake up. b. She may be back – she might be back. c. I can run the marathon in less than four hours. d. I could run the marathon in less than four hours in my younger days. e. He must come – he had to come.
The difference between may, can and must and their past forms, on the one hand, and had to on the other is that the former allow for performative uses, i.e. with speaker commitment to the statement, while the latter is necessarily descriptive, i.e. without such a commitment (cf. Verstraete 2001, Nuyts 2001b). A further differentiation of the uses of might and could is possible: might is most often used performatively, while could often stands for descriptive dynamic and deontic uses.
232
Modal grounding and subjectification
In the Romance languages modal verbs carry tense, mood, and person marking. One can therefore argue that the grounding function is still provided by these markers, even with modals. Mood is a pervasive grounding element in Spanish, since the subjunctive is a specific means of subjective grounding (cf. Achard 1998, 2002; Lunn 1989, 1995). In (10a), epistemic poder receives indicative marking, while the finite verb pensar in poder’s que-clause has subjunctive mood in (10b). (10)
a. Estos alumnos tienen más bajos puntajes... Ya ¿Sí o no?. Ya. Tú lo dices frente a eso; sí puede ser y además gente que no le interesa yo creo además. ¿ah?.... Claro. (Habla Culta: Santiago: M2) ‘These pupils have lower scores. Well. Yes or no? Well. You say it in comparison with this. Yes it may be and, moreover, people who are not interested, I believe. ah? Definitely.’ b. Me da la impresión... eres muy individualista o sea tienes una personalidad muy propia... puede que pienses así ¿ya?; pero hay otras personas que no son así. (Habla Culta: Santiago: M9) ‘I have the impression... you are very individualistic, namely you have a very particular personality... it may be that you think that way, okay?, but there are other persons who are not like that.’
Real modal grounding predications are performative in nature, while the descriptive modal expressions combine temporal grounding and modal grounding. In other words, the Spanish present tense is an unmarked form of the performative epistemic modal expressions, as opposed to the past forms that specifically locate an epistemic statement in the past.195 In this respect, a distinction can be made between temporal grounding and modal grounding. In Cognitive Grammar terms, both Spanish and English performative epistemic modals profile the infinitival process, the latter being the grounded entity, “rather than the grounding relation” represented by tense and person marking (Langacker 1990: 27). If performativity rather than tense is the major criterion for modal grounding predications, other linguistic expressions may also qualify for realizing modal grounding. Nuyts (2002) and Pelyvás (1996), who work within cognitive-functional linguistics, have already widened the concept of grounding predication to (epistemic) modal grounding. Below, I will briefly comment on these proposals.
Diachronic subjectification
233
1.2.2. Widening the concept of grounding From a cognitive-pragmatic point of view, Nuyts (2001, 2002) argues for a grounding system that goes beyond the highly grammaticalized linguistic forms: he considers mental state verbs such as to think (10a) and modal adjectival and adverbial predicates such as probable (10b) and probably (10c) also as means of epistemic grounding. (10)
a. I think they have run out of fuel. (from Nuyts 2001b: 107) b. It is probable that they have run out of fuel. c. They have probably run out of fuel.
Rather than emphasizing the grounding of a proposition with regard to the speech event, Nuyts (2002) focuses on epistemic grounding which expresses the speaker’s assessment of the proposition’s likelihood. Nuyts’ (2002) proposal regarding modal grounding is very much in line with Traugott’s account in terms of “pragmatic strengthening”, in that it mainly links grounding with the speaker’s belief state toward the proposition. Interestingly, a rather similar view of modal grounding has been proposed within the framework of Cognitive Grammar. Pelyvás (1996: 188) claims that the symbolic relationship of epistemic grounding “manifests itself in function rather than in form”, which leads him to view cognitive, i.e. mental state, predicates and sentence adverbs as (epistemic) grounding predications.196 Hence, in Pelyvás’ view, the tense marking of the cognitive/mental state predicates does not prevent them from being considered grounding predications. Importantly, Pelyvás (1996: 171) also argues that the tense criterion for grounding predications is “too inclusive” because it does not take into account the different conceptual nature of epistemic and deontic modals. According to the author only epistemic modals deserve the label of “grounding predication”. Pelyvás bases his argument on the shifting focus of attention from the participants associated with the deontic interplay of forces, such as speaker/imposer and the subject/doer (narrow scope), to the epistemic view of the process as a whole (wide scope), which leaves the prominent interplay of forces intact. I will come back to this topic in Section (4.). 1.2.3. Application to Spanish The views of modal grounding discussed in (1.2.2.) are particularly pertinent to the analysis of the tensed epistemic modals in Spanish. I follow
234
Modal grounding and subjectification
Langacker’s argument that the tenseless modal auxiliaries are the real grounding predications, but, as said in (1.2.1.), I will minimize the temporal grounding in the present tense constructions in favor of the modal grounding. Although tense remains the principal temporal grounding element in Spanish, the modal grounding in the present tense constructions deemphasizes the temporal grounding of the verbs.197 The conceptual dimension of Langacker’s theory, developed to account for both temporal and epistemic grounding, may well serve to analyze the epistemic grounding conveyed by the Spanish epistemic modals. 2. On formal and conceptual aspects of grounding Below I will discuss the evolution of the English modals towards grounding predications (2.1.), and deal with the content and the profile of the modals (2.2.). I will argue that the conceptual content of epistemic modals fits better in the dynamic evolutionary model than the content of the deontic modals (2.3.). 2.1. Grounding predications and subjectification From the emergent English modals to the present-day modals, the ground has shifted from being objectively construed, i.e. onstage, to being subjectively construed, i.e. offstage (cf. Langacker 1990, 1991). I first discuss the two types of subjectification proposed by Langacker (1990) (2.1.1.), and then apply them to Spanish (2.1.2.). 2.1.1. Two types of subjectification According to Langacker (1990, 1991) the subjectification process of the English modals is associated with two kinds of losses or replacements: (i) the potency originally located in the subject is increasingly realigned with a more diffuse source of potency and (ii) the inflection and past participles are lost and are replaced by subjective grounding by the conceptualizer/speaker. The English modals have developed from main verbs with a syntactic subject (trajector) being the onstage locus of potency. For example, Old English magan,198 which led to may, expressed the strength or physical ability of the subject (cf. Radden (1999: 272) for a complete list of the Old English forms).199
On formal and conceptual aspects of grounding
235
The first stage of subjectification (type I) comes about when the directed potency is realigned from the objective axis to the subjective axis. The fact that the objectively construed subject of deontic have to in (11) is no longer the locus of potency implies that the potency relation is to some extent subjectively construed. The objectively construed subject, here he in (11), is “the trajector of the landmark process” (Langacker 1990: 26). (11)
He had to go to the bathroom. (Radden 1999: 267)
In the traditional CG-view, the tense marking ensures that the ground itself is still objectively construed. With type I subjectification, the modal relation is “put onstage and is profiled” (Langacker 1991: 271), as the bold line shows in Figure 3.
Figure 3. Periphrastic grounded predications (Langacker 1990:26)
It can be inferred from Langacker’s (1990, 1991) analysis that, in addition to have to, the tensed forms of English “periphrastic” grounded modals like need to and be able to with inflection have reached this first stage of subjectification. The second type of subjectification (type II) stands for the highest subjectivity and is only expressed by English modal auxiliaries such as must, can and may, as in (12). Langacker considers these verbs grounding predications. (12)
He may be finishing his paper.
As with the periphrastic modals have to and need to in Figure 3, the potency in these modal auxiliaries is not associated with the subject. The difference between the English periphrastic modals and the modal auxiliaries lies in the fact that the former are said to profile the modal relationship, while with the latter the infinitival process is profiled, as shown in Figure 4.
236
Modal grounding and subjectification
Figure 4. Grounding predications (Langacker 1990: 26)
Langacker’s combination of formal and conceptual facets of grounding makes it hard to apply the two types of subjectification to other European languages. In the next section I will apply the proposed system to Spanish and will also discuss a series of problems. 2.1.2. Application to Spanish Unlike English, it is not clear whether Spanish also has a proto-modal type such as English magan. The fact that Spanish deber and poder come from verbs that, as noted in (1.1.1.), were already modal in Latin might imply that Spanish does not have such originally non-modal forms. The case of tener que is not an example of objective potency either. One could state that the dynamic reading of poder, as in (13), is closest to the modal reading with an onstage locus of potency. (13)
... en aquellos momentos tenía en mi compañía o en la compañía en la cual yo trabajo que es Producciones Cisne un número de actores que podrían representar bien aquella obra... había podido conseguir una niña ..v.. que podía interpretar muy bien el papel de Helen Keller un papel muy difícil. (Habla Culta: San Juan: M17) ‘At that time I had in my company or in the company in which I work, Producciones Cisne, a number of actors that would be able to perform this work well.. I had been able to get a girl that could perform the role of Helen Keller, a very difficult role, very well.’
The first type of subjectification of periphrastic grounded modals readily applies to the deontic modal reading of poder in (14). The locus of potency is no longer the subject, here ustedes, which implies that the potency relation is to some extent subjectively construed.200 (14)
“No, padre, usted se equivoca; yo soy mexicana, me casé con un señor de origen árabe... me llevó a vivir a Chile. Cuando él murió
On formal and conceptual aspects of grounding
237
yo heredé este lugar; pero ustedes pueden entrar: no se les cobrará nada”. (Habla Culta: Mexico: M15) “No, father, you are mistaken; I am Mexican, I got married with a man of Arabic descent... he took me to live in Chile. When he died, I inherited this place; but you may come in: you will not be charged anything.’ A first problem is that Langacker (1990) considers all tensed modals as necessarily belonging to the category of periphrastic grounded modals (Type I Subjectification) (cf. Langacker’s (1990: 27) comments on German sollen ‘to have to’ or mögen ‘can/may’, to name but two). From this point of view, all Spanish modals should be considered periphrastic grounded predications. As argued before, the present tense does not alter the view of the modal expression, however. Moreover, tense reflects the viewpoint in time from which the speaker assesses the likelihood of the proposition, without therefore situating the infinitival process in the past, present or future. The compound infinitive in (15) illustrates that haber further anchors the infinitival process in the past independently from the temporal grounding of the finite verb, which has subjunctive mood. In my view, a performative construction such as the one in (15) illustrates that modal grounding predications are possible in Spanish. (15)
... eran españoles, principalmente refugiados universitarios, personas que han pasado a América como consecuencia de la Guerra Civil Española. Por esa razón quisiera señalar que... el influjo que pueda haber tenido en mí lo español ..v.. es más bien a través de lecturas a través de ... relaciones personales con estos emigrados y estos profesores y menos a través de contactos directos. (Habla Culta: San Juan: M20) ‘They were Spaniards, mainly university refugees, people who came to America as a consequence of the Spanish Civil War. For that reason I would like to point out that the influence that the Spanish [things] may have had on me ... is more through readings and through personal relations with these emigrants and these professors and less through direct contacts.’
Thus, the present tense of the grounded modals, be it in the indicative or the subjunctive mood, does not prevent the epistemic modal from modally grounding the infinitival process. A second problem in the application of Langacker’s (1990: 28) account to Spanish is the argument that the formal differentiation between grounded
238
Modal grounding and subjectification
and grounding predications should not be confused with the distinction between deontic and epistemic readings. Yet, in Spanish the tense and person marking has a different impact on an epistemic modal relation than on a deontic modal relation.201 The temporal and person grounding of deontic modals is much broader than that with epistemic modals. There is a link between the tense of the deontic relation and the compound infinitive, as shown in (16). (16)
Pensaba que por ejemplo vos tenés ya todo dispuesto ahí tenés una mesa vestida ¿no? Sí. Podías haber puesto el grabador ahí abajo ¿no? y entonces avisar a todos los que llegaran que efectivamente todo lo que dijeran se iba a grabar.... (Habla Culta: Buenos Aires: M24) ‘I thought that for example you have everything, there you have a dressed table. Right? Yes. You could have put the recorder there below, right? and then inform all the people that arrive that everything they said would be recorded.’
It is almost impossible to formulate an epistemic construction with second person marking and imperfect past tense marking. The modal construction in (16) directs a force at the interlocutor who is onstage. Thus, a distinction should be made between applying (deontic) force oriented to another participant and applying force to the epistemic view of the infinitival process. In (17), the poder construction in the third person yields different readings in the past and in the present. In (17a), the past construction can yield an epistemic, deontic or a dynamic reading, while the present construction can only yield an epistemic reading. (17)
a. Podía haber puesto el grabador ahí abajo. ‘(S)he could have put the recorder there below.’ b. Puede haber puesto el grabador ahí abajo. ‘(S)he may have put the recorder there below.’
The reduced distribution of the epistemic modal grounding is also exemplified by the puede que construction in (18), which cannot yield dynamic and deontic readings. (18)
Me da la impresión... eres muy individualista o sea tienes una personalidad muy propia... puede que pienses así ¿ya?; pero hay otras personas que no son así. (Habla Culta: Santiago: M9)
On formal and conceptual aspects of grounding
239
‘I have the impression... you are very individualistic, namely you have a very particular personality... it may be that you think that way, okay?, but there are other persons who are not like that.’ The puede form in (18) cannot be put in the past form and can, thus, be considered as the prototypical grammaticalized and performative grounding predication in Spanish. Since the que-clause necessarily takes subjunctive mood, the process expressed in the que-clause is subjectively construed and grounded by the modal construction. In sum, when one differentiates between temporal grounding and modal grounding, only the epistemic modals appear to function as real grounding predications that profile the infinitival process, which is then a grounded head. Thus, an epistemic modal like poder, although not a real grounding predication in the Langackerian sense, has the same function as its English counterparts.202 In the next section I will show that the content of epistemic modals differs from that of the deontic ones. 2.2. Content and profile Langacker (1990:14) argues that the difference between type I Subjectification, i.e. grounded predications (e.g. have to), and type II Subjectification, i.e. grounding predications (e.g. must), has to do with the distinction between “base (content) and profile (designation)”. Langacker (1990: 14–15) contends that it is not the conceptual content which determines the final shape of the construal but its designation or profile. In the analysis of English and German modals, the difference between content and profile implies that “must designates (or profiles) the landmark process, whereas müssen ‘to have to’ by definition profiles the modal relation of necessity” (Mortelmans 2002: 393). Thus, in Langacker’s (1990: 14) view both deontic and epistemic must “evoke as [their] base a grounding relationship, but within that base [they] profile only the grounded entity (a schematically-characterized thing or process)”. However, the idea that deontic and epistemic modals give a similar profile to the grounded entity (infinitive) is problematic, since a deontic modal statement implies that the speaker him/herself can interfere in the process, while this is impossible with epistemic modals. In other words, since deontic force and its participants are more salient than epistemic modal ‘force’, the content of deontic and epistemic modality should be different. In this context, it is appropriate to view epistemic modality on a conceptual level, more or less dissociated from the type of linguistic expressions
240
Modal grounding and subjectification
available in a particular language (cf. Nuyts 2001b; Pelyvás 1996, 2001, 2003). Since the base or conceptual content is not language specific, I argue that, in Spanish, just like in German and in English, epistemic modality assigns a special profile to the process as a whole, while deontic modality profiles the force relation between participants involved in this process. This view is also supported by Langacker’s dynamic evolutionary model. 2.3. Dynamic evolutionary model The argument that Spanish epistemic modals are more suitable for modal grounding is in line with the dynamic evolutionary model (see Figure 5), which contains the notions of potential and projected reality, i.e. the speaker’s knowledge of the possible (future) evolution of reality (see Langacker 1991).203 On the one hand, potential reality includes all future events that can potentially take place from the moment of speaking onward, or includes all views on the past that are potentially true. On the other hand, projected reality stands for developments in the future that the speaker considers very likely to take place.
Figure 5. Dynamic evolutionary model (Langacker 1991: 277)
In most European languages, epistemic modals readily express potential and projected reality. Spanish modals, just like the English ones, allow the speaker to have an encompassing view of the state of affairs, which is more potential than actual. The projected reality, for its part, should be viewed as a kind of potential reality that is seen as more certain. The projected reality is prototypically expressed by the future tense. Applied to the Spanish modals, it is plausible to assume that poder expresses potential but not projected reality, while deber and tener que stand for both potential and projected reality. Since these verbs do not express the same degree of likelihood, the potential dimension may determine the reading to a greater or a
Aspectual and modal marking
241
lesser extent. Moreover, the projected reality does not necessarily include the future. The compound infinitives with deber and tener que show that past events also combine with projected reality. In many cases the projected reality has an evidential dimension. As shown in the previous chapter, the view of reality can be “projected” both on the basis of inferential processes and on the basis of the speaker’s epistemic assessment of the likelihood. The strongest projection of reality is based on logical inferences, while the readings without inferences have developed a stronger potential reality dimension. The proposed dynamic elaboration of reality adequately accounts for epistemic modality, but is less appropriate for dynamic and deontic modality. The dynamic reading expressing capability stands for a present (known) reality, since it focuses on a characterization of the subject. Nevertheless, the dynamic reading also indicates that it can be further developed in the future. The deontic reading is part of the present reality, since the speaker or another participant interferes at the moment of enunciation. The imposer appeals to the addressee to realize the desired event or action. Although the outcome is projected and remains potential, the force relation is not. As observed in the previous chapter (Section 3.), some inferential constructions of tener que have an inherent deontic dimension, such as in (19). (19)
… compran el periódico y lo primero que leen es la página deportiva. Enc. A eso era que te quería yo llevar. Tiene que haber un interés permanente cuando nuestros periódicos todos tienen una sección de deportes. (Habla Culta: San Juan: M11) ‘They buy the newspaper and the first thing they read is the sports page. Enc: To this point I wanted to come. There must be a permanent interest when all our newspapers have a sports section.’
While the deontic overtones can often be excluded from epistemic deber, this is not possible with the epistemic reading of tener que, where the locus of potency is not the speaker but a more diffuse entity.204 3. Aspectual and modal marking In this section I first present the methodology of the analysis (3.1.), before I discuss the impossibility of construing epistemic modals as infinitives (3.2.).
242
Modal grounding and subjectification
3.1. Methodology Aspectual marking will allow me to structurally underpin Pelyvás’ (1996, 2001, 2006) argument that only epistemic modals qualify for epistemic grounding. Although other verbs could also be used, inchoatives such as empezar ‘to begin’ or the future ir a ‘to be going to’ have a double advantage: (i) they allow us to examine whether the (synthetic) tense marking can be broadened to (analytic) aspectual marking. If the modal cannot receive any other explicit temporal marker, then the impact of the tense marking on the modal profile is minimized altogether; (ii) it is assumed that verbs assigning epistemic modal grounding cannot be further profiled themselves. On the one hand, the availability of a profile is then limited to processes which involve possible participants. On the other hand, the lack of supplementary marking illustrates the special relationship between the modal and the grounded entity. Since the infinitive following the aspectual is viewed as a process, only deontic modals can be the target of conception, whereas epistemic ones only represent the subject of conception. 3.2. Infinitive and Spanish modals In (20), the inchoative verb ir a ‘to be going to’ can readily precede the infinitival form of poder, which then receives an ability and a permission reading. (20)
Buenos Aires está atestada de autos y si la gente va a poder usar la bicicleta en los trayectos cortos y medianos, es posible que reduzcamos el caos en el tráfico. (Noticias: La Nación. 19/05/1997) ‘Buenos Aires is full of cars and if people are going to be able to use the bike for small and medium length distances, it is possible that we reduce the traffic chaos.’
In this construction, both the deontic (permission) and the dynamic (ability) reading have prominent involvement of participants. The permission in (20) is based on a letting schema with the beneficiary la gente ‘the people’, while the focus of the ability reading is centered on the capacities of the subject la gente ‘the people’ to use the bike. By no means could epistemic poder show up in a construction with ir a ‘be going to’. Importantly, deber ‘must’ and tener que ‘have to’ in (21) cannot express epistemic modality when these verbs are preceded by an aspectual verb.
Aspectual and modal marking
(21)
243
a.... la gente va a poder usar la bicicleta. [*epistemic/dynamic] ‘people are going to be able to use the bike.’ a....* la gente va a deber usar la bicicleta. [*epistemic] ‘*people are going to must use the bike.’ b.... la gente va a tener que usar la bicicleta. [*epistemic/deontic] ‘people are going to have to use the bike.’
The epistemic modals poder, deber and tener que in (21) cannot explicitly be further grounded themselves, i.e. they cannot be profiled by another verb. This is very much in line with the assumption that an epistemic modal verb stands for the subjective grounding of aspect, as a part of the projected or potential reality. Silva-Corvalán (1995: 69) mentions that deontic deber can be preceded by ir a, as shown in (22a). Although I have not found deontic deber in an infinitive position in the corpus nor on the Internet, Silva-Corvalán’s native speaker judgment indicates that a deontic reading of an aspectually marked infinitive is in principle not considered problematic. This example corroborates my claim.205 (22)
Juan va a deber cruzar el río. ‘John’s going to have to cross the river.’
In infinitival constructions tener que seems to be preferred to deber. In corpus example (23), tener que is preceded by the inchoative vamos a ‘we are going to’. While tener que can have both a speaker-oriented and a participant-oriented deontic reading, in combination with an aspectual profile tener que seems to shift to a more speaker-oriented reading. (23)
....entonces me lastimé las dos rodillas.. ¡Qué barbaridad! ¿Y ahora ya no haces ningún deporte? Pues no pero lo vamos a tener que hacer ahora también por los chamacos para que sigan el mismo paso que nosotros porque es una cosa muy.... Mi suegro por ejemplo ahorita corre tres kilómetros. (Habla Culta: Mexico: M5) ‘Then I hurt both my knees. What a bad luck! And now you don’t do sports? Well, no, but now we will have to do it, too, because of the friends so that they follow our steps because it is a very ... thing. My father-in-law for example runs three kilometers.’
With its flexible distribution tener que fills the gap in the deontic field. The relative unavailability of deber in an aspectual construction seems to broaden the use of tener que. In combination with an aspectual verb, the
244
Modal grounding and subjectification
speaker-oriented deontic readings of deber seem to be handed over to tener que. Rather than being merely an aspectual adjunct, the aspectual verb appears to be a pragmatic reinforcement of the modal in (23). It is a way of rendering the deontic force more speaker-oriented in the sense that the aspectual dimension becomes an expression of deontic compulsion or insistence. One could also try to explain the infrequent construction of deontic deber with ir a in terms of a different force structure. Deber most often conveys a strong intention of the speaker (i.e. the imposer) to move the addressee (or doer) to realize the event expressed in the infinitive, while tener que often expresses a more diffuse kind of force. However, a more plausible explanation is that deontic tener que is a more recent development and is thus less grammaticalized than deontic deber. This different distribution of deber and tener que shows that the deontic modals are undergoing a change towards less temporal grounding. 4. The shift from deontic/dynamic to epistemic modality Let us now deal with the link between epistemic/evidential commitment and deontic obligation. I will first discuss the metaphorical account of epistemic modality (4.1.). Next I will argue for an alternative account of modal grounding based on its reference-point function (4.2.) and on subjectification (4.3.). Finally, I will present the different epistemic readings in Spanish in terms of the subjectification of non-epistemic readings (4.4.). 4.1. Force dynamics and metaphor In the field of Cognitive Linguistics, epistemic modality and deontic (root) modality have broadly been analyzed in terms of force dynamics, i.e. (kinetic) forces and barriers (cf. Talmy 1988, 2002). Sweetser’s (1982, 1988, 1990) analysis of modals relies on the metaphorical extension of the force structure present in deontic (root) modality (24a) to epistemic modality (24b) in terms of barriers. (24) a. John may go. (deontic) b. John may be there. (epistemic) Yet, Pelyvás (2000: 235–240) convincingly argues that Sweetser’s account of may is problematic on the following grounds. First, the different scope of
From deontic/dynamic to epistemic modality
245
the negation with the deontic and the epistemic readings impedes an interpretation in terms of the metaphorical extension of barriers. In (25a), the deontic force relation is negated, while in (25b) not the epistemic relation but only the infinitival process is negated. (25) a. John may not go. b. John may not be there. Second, the structural relationship between the source domain – “the doer” (subject), i.e. John – and the target domain – the speaker and the premises (s)he relies on – is not preserved. If a real metaphorical shift took place, the elements of the source domain would have to correspond to those in the target domain. These two criticisms also hold for a metaphor-based analysis of the difference between Spanish deontic and epistemic modals: a more local approach to the shift toward epistemic readings seems to be required.206 In line with the basic principle of force dynamics, Pelyvás (2000) offers an alternative analysis of modal force in terms of the presence or absence of “counteracting forces of relative strengths”. Moreover he starts from Traugott’s (1989) hypothesis that epistemic may derives from the now extinct reading of ability. Langacker’s recent proposals also go in the direction of a more gradual view of the shift from deontic to epistemic modal readings. Langacker (1990: 28) seems to support the metaphorical mapping from potency in the social sphere to potency in the domain of reasoning proposed by Sweetser (1988), which is close to Langacker’s concept of realignment of deontic force. At the same time, however, Langacker (1990: 28) proposes an alternative account of the English deontic and epistemic modals in terms of the increasing “diffuseness” of the locus of potency, ending up with the epistemic modals expressing the “nature of evolving reality itself”. More recently, Langacker (1998, 1999, 2000: 298) somehow reconsiders the subjective realignment proposed in Langacker (1990). Now the potency relation in modals undergoes a process of attenuation of objectively construed elements and involves an increasingly diffuse source of potency from main verbs to epistemic modals (cf. Langacker 2000: 307– 309). This correlates with Langacker’s adoption of Verhagen’s (1995) argument that both dimensions are already “imminent in the objective conception”. Thus, rather than metaphorical mapping, there seems to be a continuum from concrete potency to extremely diffuse or nebulous potency.
246
Modal grounding and subjectification
Below, I follow Pelyvás’ (1996, 2003) adaptation of Langacker’s system which essentially considers the epistemic grounding by modals as a combination of reference-point function and subjectification. 4.2. Reference-point A reference-point serves “for purposes of establishing mental contact with another [entity]” [the target], as shown in Figure 6 (Langacker 1993: 1, 1999: 173).207 In a construction such as this girl, this serves as a referencepoint (R), which attracts the attention of the conceptualizer and has initial prominence. The reference-point R then shifts to the target, i.e. the instantiated girl (T), and the initial prominence of R fades away.
Figure 6. Reference point construction (Pelyvás 1996: 160)
Although reference-points are most developed in deixis (cf. Langacker 1993; Taylor 1996), epistemic modals also function as a reference-point by making the infinitival process the main focus of attention (cf. Pelyvás 2001). The dynamic nature of the reference-point is useful for the analysis of modals, since it enables the contextualization of a process: The dynamic aspect of the reference-point topic has a life history: it must first be established as such, often by a device that makes it salient as the explicit focus of attention. Once established, on the other hand, it may well remain implicit as subsequent expressions, themselves now in focus, are integrated into the context it provides. (Langacker 1993: 24)
The dynamic nature of the reference-point construction can account for the shift of prominence from the reference-point to the target, i.e. the infinitival process, which then becomes the grounded entity. While the modal has initial prominence, it ultimately profiles the target infinitival process, al-
From deontic/dynamic to epistemic modality
247
though the construction “occasionally retains some prominence of the reference-point” (Pelyvás 2001). Pelyvás’ (2001) proposal can help us elucidate the differences between the deontic and the epistemic modals in Spanish. The dynamic nature of the reference-point is crucial for a better understanding of the conflicting relation between aspectual markers and epistemic grounding (26a), on the one hand, and the felicitous combination of aspectual markers and deontic modals (26b), on the other. (26)
a. …* el pobre va a deber de haber estado sufriendo horrores. ‘* the poor guy will must have been suffering tremendously.’ b. ... tú vas a deber/tener que hablar un poquito más. ‘ you will have to speak a bit more.’
The fact that epistemic modals cannot combine with an aspectual verb, as in (26a), shows that they cannot be profiled themselves. Since the reference-point cannot be “left in profile” (Pelyvás 2001), the epistemic modal verb cannot be preceded by any other verb which blocks the dynamic nature of the epistemic grounding. This correlates with the reference-point function of modals that necessarily shift attention from the modal reference-point to the proposition as a whole. In (26b) the aspectual verb gives a full profile to the deontic potency relation, which keeps this profile. The result of the combination with an aspectual verb is that the infinitival process introduced by the deontic modal is less prominent than the deontic modal relation. Furthermore, the infelicity of the construction in which the epistemic/evidential modal stands alone in (27b) confirms that deber de has a reference-point function in that the verb only serves to modally ground the infinitival process, without remaining profiled itself.208 Evidential deber needs an infinitival process to which it can shift its own profile, as shown in (27c). The fact that in (27d) the proposition introduced by the epistemic modal can stand alone corroborates that the infinitival process is the modally grounded head. (27)
a.... el pobre debe de haber estado sufriendo horrores. (Habla Culta: Caracas: M5) ‘the poor guy must have been suffering tremendously.’ b. * el pobre debe (de)... ‘the poor guy must.’ c. ... debe (de) ser así. ‘it must be so.’
248
Modal grounding and subjectification
d. … ha estado sufriendo horrores, así es. ‘he has been suffering tremendously, so it is.’ By means of the examples in (28) I will now show that deontic deber cannot be considered a reference-point to the infinitival process.209 The independent modal in (28b) shows that the deontic relation in the deber construction in (28a) retains prominence. This does not mean, however, that the construction with ser así (28c) is not possible anymore. In (28d), it is shown that without the deontic modal relation the infinitival process does not refer to the whole proposition but to the deontic modal relation. This once again corroborates that deontic deber cannot be considered a mere reference-point to the infinitival process. (28)
a. ... tú deberías hablar un poquito más. (Habla Culta: San Juan: M16) ‘you should speak a bit more.’ b. … sí, deberías.... ‘yes, you should.’ c. … debería ser así. ‘It should be so.’ d. ... habla un poquito más, así es. ‘(s)he speaks a bit more, so it is.’
In sum, the reference-point function of the epistemic modals profiles the infinitival process as a grounded head and makes the latter the focus of attention. By contrast, the deontic modal relation is profiled and, since it retains prominence, it does not lend itself to a shift of attention to the infinitival process. This leads me to argue that, despite their tense inflection, the Spanish epistemic modals are similar to the modal grounding with the highest degree of subjectification. 4.3. Subjectification and interplay of forces Let us now look at the second dimension of modal grounding, viz. its advanced subjectification. Different degrees of subjectification point to the varying extent to which the locus of potency becomes increasingly diffuse. So far I have argued that the force relation impedes the deontic modals to be real modal grounding predications, like the Spanish epistemic modals. For the sake of clarity, I exemplify this difference by means of Pelyvás’ (1996) description of English may. The deontic permission reading of may
From deontic/dynamic to epistemic modality
249
usually involves a concrete and easily accessible force.210 The participants involved in the deontic force (F) present in may belong to the objective scene: the speaker is included in the scope of predication since s/he may be one of the participants in the interplay of forces, as shown in Figure 7 by means of the dotted line between the trajector (tr) and the landmark (lm).
Fig. 7. Deontic reading (narrow scope)
Fig. 8. Epistemic reading (wide scope)211
Figure 8 represents the epistemic modal reading of may. In this construction, there is no force relation driven by force F, as in Figure 4, but only viewing from the vantage point of the Speaker (S) or Ground (G). Pelyvás distinguishes between a ‘narrow scope’ (Figure 7) and a ‘wide scope’ (Figure 8) reading of the modal auxiliary may. The reprinted image schema of ‘narrow scope’ in Figure 8 illustrates that this kind of scope can include various types of interplay between deontic forces. With ‘narrow scope’, the speaker interferes more in the state of affairs than that (s)he profiles a special kind of elaborated reality with regard to the state of affairs. In the epistemic reading, by contrast, the force or potency is maximally diffuse, and thus broadens the scope of the predication. The speaker and the modal ground are part of the wide scope which leads to a subjective construal of the process. The narrow scope and wide scope correlate with the absence and. presence of a reference-point function. The profile of the deontic relation leads to a narrow scope view, while the non-profiling of an epistemic referencepoint enables wide scope on the state of affairs. Since the argumentation for English may is compatible with what I have argued for the Spanish modals so far, it can be taken to be valid for Spanish too.
250
Modal grounding and subjectification
4.4. Different epistemic and evidential readings As for the subjectification from deontic to epistemic modality, Pelyvás (1996: 167–168) proposes a ‘breakdown mechanism’ similar to attenuation which reduces the profile of the agentive structure step by step. There is a shift of the profile from the deontic relation with the doer (overall trajector), which has a prominent position in the interplay of forces, to the speaker and, ultimately, to the process itself. Hence, the subjectification of epistemic modals is based on the progressive attenuation of the interplay of forces to a scene in which no other force is involved. Since no participant other than the speaker actively interferes in the epistemic modals, the infinitival process becomes more salient than the potential interplay of forces: the only element in profile is the (infinitival) process itself. Still, the deontic or ability force patterns influence the different epistemic readings. In what follows, I will discuss how the epistemic readings of poder, deber and tener que can be understood as the result of subjectification of different force types and counterforces. All three verbs can be accounted for in terms of attentuation of the force structure. First, epistemic poder yields a potential reality reading including different possible realities, with the speaker not deciding on any of them. This is in line with the ability reading of poder, the subject of which bears the potential to perform some purposeful action and could–but does not necessarily–experience a strong counterforce. Epistemic poder does not express strong likelihood, since there is a potential resisting force against which the event is construed.212 The weak commitment of the speaker to the likelihood expressed in epistemic poder is also in line with the deontic reading of poder, where the speaker’s force is weaker than that of the addressee or the doer denoted by the grammatical subject (cf. Pelyvás 2000: 238–239). Thus the addressee may or may not perform the action allowed by the emitter of force. Second, the epistemic/evidential reading of deber stands for a combination of potential and projected reality (cf. Langacker’s dynamic evolutionary model). The deontic reading of this verb corresponds to the strong intention of the speaker, on the one hand, and the doer’s unspecified reluctance to perform the purposeful action, on the other (cf. Pelyvás 2000: 244). In line with this deontic reading, it is not surprising that the evidential readings of deber are most often based on inductive inferences, which attribute an important role to the speaker, but do not express the highest likelihood that the proposition is true. In epistemic deber, the subjectified counterforce involving unknown reality brings about a weaker speaker commitment than in an affirmative, unmodalized utterance.
Conclusions
251
Third, the epistemic/evidential modal reading of tener que ‘have to’ is somewhat different from that of deber, in that it expresses the highest likelihood that the proposition is true. With deontic modal tener que, the speaker usually invokes external arguments to commit the doer to the realization of the purposeful action. These arguments can be the volition of other unspecified participants or logical requirements associated with the action proposed. The subjectification of this pattern leads to a strongly deductive epistemic/evidential modal reading which correlates with the observation that the logical inferences are clearly related to rules and principles used in the world ‘out there’. As a matter of fact, in the modality expressed by tener que, the speaker him/herself is less directly involved in the assessment of reality, relying instead on the explanatory power of generally accepted criteria as implicit premises of the statement. 5. Conclusions On the basis of a slight adaptation of Langacker’s views of subjectivity and subjectification, I have shown that the epistemic/evidential readings of the Spanish modals can be considered modal grounding predications, while the deontic readings cannot. In Section (1.), I have presented a diachronic analysis of tener que and discussed the evolution of deber and poder. The conclusion of the first subsections was that the Spanish epistemic modals develop from deontic (or dynamic) modals. Next, I have discussed the status of grounding in the analysis of linguistic expressions in general and of modals in particular. There I have differentiated between temporal and modal grounding. In this way, I have nuanced Langacker’s (1990) claim that inflection prevents nonEnglish modal from being considered grounding predications (Type II Subjectification). I have reviewed the different types of subjectification proposed by Langacker (1990) and applied them to the Spanish system in Section (2.). The conclusion was that the temporal grounding of epistemic modals is more restricted than that of deontic modals. This has led me to argue that Type I and Type II subjectification are not only different in profile but also in conceptual content. The modal grounding predications are restricted to the epistemic modals. Although the epistemic readings of the Spanish modals formally differ from grounding predications, their conceptual content is very similar. The dynamic evolutionary model holds for all types of epistemic modals but cannot account for the deontic modals. Section (3.) has illustrated that Spanish epistemic modals distinguish
252
Modal grounding and subjectification
themselves from the deontic ones by the inability to be profiled themselves. Epistemic modals stand for the modal grounding that most completely profiles the infinitival process or the grounded entity. The impossibility of adding an aspectual auxiliary verb corroborates their dynamic referencepoint function. In Section (4.), the concepts of reference-point and subjectification have further underpinned the difference between epistemic modal grounding predications and deontic modals. Unlike the deontic modal, the epistemic modal is a reference-point which has only initial salience and profiles the infinitival process, which becomes more salient. Epistemic modals are also the result of subjectification in that they have undergone a shift from force directed to the doer (subject) in deontic modal readings to the process itself. In other words, the ‘narrow scope’ view of the prominent interplay of forces in the deontic reading shifts to a ‘wide scope’ view that includes the speaker and the epistemic modal grounding. Finally, I have discussed the relation between epistemic and other types of modality. In line with Pelyvás’ (1996, 2000, 2001, 2006) alternative to Sweetser (1990), I have shown that the attenuation of the force structure present in the dynamic and deontic readings leads to the different epistemic modal readings associated with poder, deber and tener que. Thus, the type of deontic force and counterforce appears to be related to the weak and the strong evidence used for potential or projected reality expressed in the epistemic/evidential readings of the Spanish modals.
Chapter 9 General conclusions
In this volume I have elaborated a cognitive-functional account of the relation between form and meaning of seven verbs with modal and/or evidential meanings. The approach, which considers both synchronic and diachronic data, is innovative in comparison with the syntax-oriented accounts of finite verb + infinitive constructions in Spanish in a number of ways which I will survey in this chapter. Three main conclusions emerge from the analyses presented. First, the evidential and modal (semi-)auxiliaries are characterized by a different frequency distribution in written and spoken Spanish, which points to a different relation between evidence and speaker. Second, the relation between evidentiality and epistemic modality is not a very straightforward one. Third, diachronic and synchronic evidence confirms that the verb + infinitive construction is a more appropriate form for a subjective statement than the que-clause construction. 1. Evidential and epistemic qualifications The central argument of the functional analysis is that parecer/resultar, amenazar/prometer and poder/deber/tener que can be analyzed in terms of evidentiality and that evidential qualifications are different from epistemic modal ones. The analysis of parecer and resultar in terms of evidentiality and (inter)subjectivity, which I elaborated in Chapter 2, is in two ways an improvement of the previous studies in the field. On the one hand, it refines the analysis of the broad group of non-factive predicates, to the extent that it shows that so-called “non-factive” parecer + infinitive and “near-factive” parece que constructions convey different evidential readings. As an alternative for the unjustified link between source-evidentiality and epistemic speaker commitment, I show that the notion of (inter)subjectivity can underpin the so-called “non-factive” and “near-factive” readings of parecer + infinitive and parece que: I account for “non-factive” parecer + infinitive in terms of a subjective statement based on subjective evidence, in contrast to “near-factive” parece que, which relies on intersubjective, i.e. shared, evidence to express a broader supported statement. The subjective state-
254 General conclusions ment conveyed by parecer + infinitive is perceived as expressing lower factivity than the intersubjective statement with parece que. As far as modal amenazar and prometer are concerned, the analysis presented in Chapter 4 is new in the field of Spanish linguistics, in that it has dealt in more detail with the modal qualification as distinct from the inferential qualification. I have shown that the different types of prediction originate from the different lexical semantics (namely the commitment of the subject), on the one hand, and the illocutionary force, on the other. The subject commitment and the presence of illocutionary force lead to a stronger speaker commitment in the modal readings of prometer which, hence, expresses a stronger prediction that the proposition will eventually come true. The absence of real speaker commitment and illocutionary force with amenazar, by contrast, results in a weaker speaker commitment in terms of a weaker prediction. The fact that, unlike prometer, the event predicted by amenazar is left unrealized in quite a number of cases in my corpus confirms this reading. The analysis of poder, deber and tener que in terms of evidentiality presented in Chapter 7 differentiates between the evidential readings of each verb which are, traditionally, called “epistemic”. It is generally observed that poder expresses weak commitment (possibility) and that deber conveys intermediate speaker commitment (strong possibility), while tener que conveys maximal commitment and near-certainty (cf. Gómez Torrego 1999, Fernández de Castro 1999: 189). My analysis gives an explanation for these different degrees of speaker commitment in terms of the modes of knowing involved. I have argued for a prototypical epistemic/evidential reading for each of the three verbs: poder usually does not rely on inference, while deber and tener que most often involve inductive inferences and deductive inferences, respectively. With modal verbs, the absence of inference appears to lead to weak speaker commitment to the proposition, the creative reasoning of inductive inferences results in intermediate speaker commitment, while the logical reasoning of deductive inferences results in strong speaker commitment. Interestingly, when deber and tener que involve deductive and inductive inferences, respectively, the speaker commitment is either strengthened or weakened in comparison with the most frequent readings of these verbs. Thus, the modal verbs are the only verbs examined that clearly link the evidential qualification with the epistemic modal qualification. In conclusion, I have argued that: (i) The concepts of epistemic modality and speaker commitment cannot readily be applied to evidential parecer and resultar, since these verbs do not convey a judgment of the likelihood that the event is true but only suggest that there is (shared or non-shared)
Linguistic representation
255
evidence to make a statement. (ii) Modal amenazar and prometer do not express the same speaker commitment to the proposition, i.e. they express a different judgment on the likelihood: while amenazar conveys weak prediction, prometer predicts strongly. It is not warranted to derive this difference from different evidential qualifications. (iii) The weak, intermediate and strong speaker commitment of epistemic poder and epistemic/evidential deber and tener que can be accounted for in terms of the different evidential modes of knowing involved in these three verbs. The overall conclusion is that only with the modal verbs, the type of inference determines the strength of the epistemic judgment or speaker commitment to the proposition. 2. Linguistic representation For all three verb types special attention has been paid to how the various evidential and epistemic readings are linguistically represented and to the cognitive process of change involved. Concepts such as transparency, grounding and subjectification led to a cognitive-linguistic account of the constructions both from a synchronic and a diachronic point of view. The different degrees of grammaticalization are underpinned by a corpus-based transparency analysis. The corpus data illustrate that, unlike resultar, parecer can be considered transparent in that it occurs more or less with the same verbs in the infinitive and in the que-clause. The analysis of the infinitive and the grammatical subject of modal amenazar and prometer has indicated that verbs are only partially transparent. The corpus data presented in Chapter 5 show that amenazar most often introduces change-of-state and action verbs and hardly ever combines with attributive verbs, while prometer does so in the majority of cases. I have argued that this has to do with the tendency to yield an internal or an external viewpoint on the state of affairs. Modal amenazar and prometer are not only less transparent than parecer in view of their specific viewpoint, they moreover do not combine with infinitives that activate the agentive patterns of lexical prometer and amenazar (deontic verbs, communication verbs etc.). The semantic classification of the subjects illustrates that amenazar and prometer are partially transparent. Although human and non-human subjects are possible with the two verbs, the latter are far more frequent than the former. With respect to poder, deber and tener que, Chapter 7 offered quantified corpus results on the infinitives and the grammatical subjects that combine with these verbs. These figures allowed for a differentiation between epis-
256 General conclusions temic poder, on the one hand, and evidential/epistemic deber and tener que, on the other. Epistemic poder combines with action verbs more frequently than deber and tener que, which most often introduce attributive infinitives. Moreover, I have shown that when poder combines with a subject denoting a thing, a state of affairs or a process, it receives an epistemic reading in half of the cases. Since evidential/epistemic deber and tener que do not manifest this pattern, one can conclude that poder is far more transparent than deber and tener que. Unlike poder, the latter verbs still face competition with the agentive patterns of the deontic reading. In addition to the transparency, the role of the complementation type of the seven verbs has been emphasized throughout the monograph. In line with the cognitive-linguistic tradition, I have argued that the ungrounded infinitives of the evidential and modal constructions lend themselves ideally to a more dynamic conceptualization of the event expressed. I have contended in Chapter 2 that the combination of the prominent anteposed subject and the ungrounded nature of the infinitival process enable a subjective conceptualization of parecer + infinitive. In such a construction, the speaker can compare different views of the state of affairs. The necessarily ungrounded complement of amenazar correlates with the conceptualization of the event from the subject’s perspective (cf. Chapter 4). The characteristics associated with the anteposed subject seem to play a role in this conceptualization. By contrast, the grounded que-clause with parece que gains in prominence and leads to a more objective conceptualization, which is also most often found with resultar. The modal verbs function as a reference-point for the ungrounded infinitival process in that they shift prominence to the process without remaining in profile themselves (cf. Chapter 8). With the modal verbs, the subject does not play a special role in the conceptualization which is directly driven by the speaker’s epistemic judgment. Finally, the limited diachronic analyses of the verbs have shown a steady increase of evidential constructions with infinitives, which testifies to their (involvement in a) subjectification process. Chapter 2 shows that the evidential use of parecer is progressively extended from constructions with nominal attributes to all types of infinitives. The same diversification of infinitive types is observed with modal amenazar and prometer (cf. Chapter 6). Verhagen’s hypothesis that a modal ‘threat’ shows up earlier than modal ‘promise’ in terms of the different infinitives that follow amenazar and prometer. The result of the diachronic evolution can also be approached synchronically. Langackerian (synchronic) subjectification in terms of the attenuation of the trajector relation or the absence of onstage conceptualizers can also account for the difference between evidential and
Further research
257
other readings. In the same way, I have explained evidential modal deber and tener que in terms of attenuation of the deontic force structure, which ends in complete absence of force in epistemic readings. Such a gradual shift refines the traditional account of deber and tener que based on (the metaphorization of) obligation. Working from the assumption that linguistic forms tell us something about their meaning, the combined results of the corpus analysis and the different dimensions of the cognitive-functional analysis support the idea that parecer vs. resultar, amenazar vs. prometer and poder vs. deber/tener que are three pairs of (semi-)auxiliaries. The corpus research shows that the first verb of each pair is more common in written discourse than in spoken discourse. With regard to the process of grammaticalization, parecer, amenazar and poder face fewer restrictions by non-evidential or nonepistemic readings than resultar, prometer and deber/tener que and can be considered more transparent. As for evidentiality and epistemic modality, the first group of verbs conveys lower likelihood that the speaker-oriented interpretation (parecer) is the correct one or that the prediction comes true (amenazar and poder). The other verbs involve a more general interpretation (resultar) or express stronger commitment. All the tendencies discussed above can be summarized and schematized in the following way:
Spoken discourse Written discourse Less grammaticalization More advanced grammaticalization Less transparent verbs More transparent verbs Subjective type of evidence or epis- Intersubjective type of evidence or evidential modality temic modality Strong statement Weak statement
3. Further research Several types of research could be undertaken to supplement the results obtained in the cognitive-functional analyses. It could be useful to extend the analysis of the grammaticalized evidential and modal verbs examined in this volume to other linguistic expressions that convey evidentiality and epistemic modality such as sentence adverbs, adjectives, particles and mental state predicates (cf. the function-to-form approach of Nuyts 2001b). Evidential and modal adjectives may lend themselves better to intersubjec-
258 General conclusions tive readings than constructions with the infinitive and mental-state predicates, and can moreover combine different qualifications (cf. parece posible ‘it seems possible). From the comparison of these different constructions, there may emerge an interesting view of how the speaker uses language and which constructions (s)he prefers for specific qualifications or combinations of qualifications. My research has emphasized the difference between evidentiality and epistemic modality with parecer/resultar and amenazar/prometer, and has analyzed how the modal verbs integrate evidentiality and epistemic modality. This investigation can further benefit from experimental empirical research along the lines proposed by Sanders and Spooren (1996) and Nuyts and Vonk (1999). Sanders and Spooren (1996) study epistemic modifiers (without excluding evidential verbs, however) and attempt to empirically underpin the categorization of certainty expressed by these modifiers. These authors want to measure the modal statement on a scale from ‘not certain at all’ to ‘very certain’, differentiate between knowledge-based and observational evidence and link the evidence with different kinds of certainty and subjectivity. In their pilot project, subjects were offered a list of epistemically modified sentences. The sentences are statements of facts, i.e. they do not refer to opinions, logical or natural laws or other generic expressions, nor intentional acts. The subjects were asked to perform two tasks: (i) they had to give their judgment on the certainty of the imaginary speaker about his/her utterance, by indicating this judgment on a seven-point scale, as shown in (1); (ii) they had to judge the appropriateness of the two evidential types in combination with the sentence by placing the labels (a) and (b) on a seven-point scale. (1)
It seems that Piet has gone shopping. not certain at all |---|---|---|---|---|---|---| very certain It seems that Piet has gone shopping, because ... a. his coat is gone. b. he always does so on Tuesdays. does not fit at all |---|---|---|---|---|---|---| fits very well (Sanders and Spooren 1996: 249)
A second experiment proposed by Sanders and Spooren (1996: 251) tests to what extent certain kinds of evidence determine which type of epistemic
Further research
259
modifier is appropriate. The authors set up a series of combinations of evidence and epistemic modifiers and ask the subjects to judge whether they are appropriate. The result of the investigation was a clear differentiation between evidential, epistemic modal and mental state verbs. Although the two evidential types (i.e. knowledge-based and observational evidence) proposed by Sanders and Spooren (1996) would have to be refined with the evidential types discussed in this volume, the application of such an analysis to a group of Spanish native speakers would certainly be a valuable complement to the evidentiality and (inter)subjectivity account presented in this book. Speaker intuitions can shed new light on the difference between evidentiality and modality, and the combination of the two qualifications (cf. modal auxiliaries). The analysis of evidentiality and epistemic modality could also be further extended in terms of their information structure. Nuyts and Vonk (1999) empirically analyze the (non-)focality of epistemic expressions by means of a method in which subjects are invited to produce epistemic expressions spontaneously on the basis of short cartoon strips depicting brief and simple stories. These cartoons leave sufficient room for alternative interpretations and are “therefore likely to cause some degree of uncertainty on the part of the speaker” (Nuyts 2001: 236). The authors differentiate between the “focality on the qualification” and “focality on the state of affairs” depicted in the cartoon strip. The condition of focus on the qualification is achieved by presenting subjects with an explanation why the story ends the way it does, and by asking their opinion about how good this explanation is. In this way their focus of attention is drawn to the likelihood that the explanation holds. The result of this analysis is that “speakers tend to avoid using epistemic expressions under conditions of focality” (Nuyts 2001b: 259). It would be interesting to do the same experiment with Spanish native speakers and investigate both the evidential and epistemic qualifications. In this way it might become possible to compare epistemic expressions with clearly evidential expressions. Research on focality may further corroborate the different nature of the two qualifications. For example, it might be the case that, unlike epistemic equalifications, evidential ones more easily allow for focus. The synchronic corpus research presented in this book has mainly focused on the frequency of the different constructions. The corpus research could also be further elaborated in terms of register and geographical variation. On the one hand, one could examine different types of newspapers (popular press, regional press, quality press), and on-line language in order to describe possibly diverging tendencies in the use of evidential and modal
260 General conclusions expressions (cf. Grondelaers, Speelman and Geeraerts 2002). On the other hand, a comparison between the language use of different countries would be required to find out whether the results obtained are generally valid for most of the Spanish-speaking countries or whether there are important regional differences (cf. the Leuven-based Research Unit “Quantitative lexicology and variational linguistics”). Since the diachronic corpus analysis conducted in this volume is limited, it only can be viewed as a basis for more fine-grained diachronic descriptions. In future research, the tendencies I have observed should be examined in greater depth and on the basis of a larger sample of examples, preferably drawn from a systematically compiled corpus based on scientific standards. The frequency of the constructions also needs to be taken into account in order to empirically underpin the tendencies commented upon. Furthermore, the functional analyses elaborated in this volume can help analyzing specific discourses in which the speaker carefully describes his/her own position without therefore giving the complete picture of reality. This strategy is systematically applied in various types of discourse, although political discourse may be an extraordinarily interesting market for such communicative strategies, as shown below. The invasion of Iraq (spring 2003) perfectly exemplifies how spin doctors use linguistic tools, at times to the detriment of the factual situation. Before, during and after the war, the discourse of the “coalition of the willing”, i.e. the United States and its allies, has shown how crucial it is to communicate to the public that one has evidence to make claims or statements (and to start a war) without making explicit the commitment to this information, as is the case with seem in (2). (2)
While stating (without any interest in proving it) that he has no weapons of mass destruction, Saddam seems to be threatening to use them against any invading military. (The electronic Warfare & Information Association)
An utterance such as the one in (2) can be analyzed in terms of the subjectivity of the seem construction and the ambiguity of the threat. The low likelihood associated with this construction allowed the spin doctors to put the alleged weapons of mass destruction on the agenda, without claiming explicitly that Saddam Hussein would use them in any case. Since evidential and epistemic qualifications are frequently used to guide, and sometimes deliberately misguide, the listener or reader to conclusions that serve the underlying message, the description of the different evidential and modal types can serve as a basis for analyzing political and other discourses in
Further research
261
Spanish and other languages. In other words, a descriptive and theoretical linguistic study such as the one offered here also ultimately stimulates the critical appreciation of day-to-day language as it is used to individually or collectively communicate about the world with its more than six billion inhabitants.
Notes Chapter 1 1. Besides the term “reading”, I follow Nuyts (2001b) in using the term “qualification” to refer to what evidential and epistemic modal modifiers express. 2. Besides the Spanish literature on modal auxiliaries, there are many other specialized publications which examine equivalents in other languages. See Sanders and Spooren (1996), Rooryck (2000) and Usoniene (2001) for the equivalents of parecer. For modal verbs see Halliday (1970), Palmer (1979, 1986, 2001), Lyons (1977, 1982), Perkins (1982, 1983), Coates (1983), Goossens (1982, 1985, 1987), Langacker (1985), Bybee et al. (1994), Bybee and Fleischman (1995), Pelyvás (1996), Mortelmans (1996, 2001), Achard (1998), van der Auwera and Plungian (1998), Nuyts (2001a,b), Verstraete (2001, 2002), Narrog (2005a,b), to name but a few. 3. Combé (1981), Gutiérrez Ordóñez (1986), Olbertz (1998) and Gómez Torrego (1999) discuss only the syntax of parecer. 4. Fernández de Castro (1999: 189) states that in the parecer + inf construction the speaker does not make an assertion but has information to think that the proposition is true. He gives the following interpretation: “From what I can see, the presupposition that I can formulate is...” or “What I can see is similar to...” [my translation]. 5. Marín-Arrese et al. (2004) differentiate between “evidential, epistemic and deontic modality” and list a series of evidential verbs in Spanish together with the calculation of their relative frequency. However, they do not account for the different constructions the evidential verbs allow for, nor do they detect an evidential reading in amenazar, prometer or deber and tener que. Hassler (2002) discusses reported speech in Romance and Hassler (2003) deals with evidential adjectives in Spanish, while Squartini (2001) addresses the evidentiality of the Romance tense system, but only briefly discusses the inferential use of the Spanish conditional and future tense. 6. I have omitted three verbs mentioned in Fernández de Castro’s (1999: 328) table. I will not discuss internal necessity expressed by haber de + inf ‘to have to/to be necessary/to be supposed to’, whose regional and register variation would require an extensive study in its own right. Neither will I deal with the expressions of approximativeness venir a ‘to be approximatively X’ and tender a ‘to tend to’ since they do not convey an evidential qualification. 7. Studies on the Dutch, English and French counterparts of parecer and resultar also refer to the label “epistemic”. Sanders and Spooren (1996: 242–247) consider the Dutch verbs schijnen ‘to seem, to look’, lijken ‘to seem’ and blijken ‘to turn out’ “nonsubjective epistemic modifiers” that express “uncertainty”, “semicertainty” and “certainty” respectively. Usoniene (2001: 186–188), for her part, discusses the seem + to be construction in terms of epistemic modal-
Notes
263
ity, which is the result of an extension of the meaning of seem from perception to modal evaluation. 8. González Vázquez (1999, 2000) gives probably the best Spanish written overview of the different types of modality, in general, and the discussion on epistemic modality and evidentiality, in particular. Yet, since the author discusses mainly English written studies which I refer to throughout the volume, I will not go into further details. 9. The Gramática Descriptiva de la Lengua española (1999) only mentions them on the side. Other reference grammars such as Butt and Benjamin (1996) and Matte Bon (2002) do not even mention them. 10. Hernanz (1999: 2234) also mentions semejar ‘resemble’. In this book I will not deal with the less frequent infinitival and clausal constructions of this verb. 11. Note that Vázquez-Laslop (2000, 2001) treats lexical prometer as a “deontic” verb. 12. The study of the sources of knowledge was first carried out by American structuralists who studied polysynthetic Native American languages that are rich in affixes, amongst them evidential ones (see the overviews in Chafe and Nichols 1986; Infantidou 2001). 13. Nuyts (2001b) adopts a function-to-form approach to epistemic modality, taking into account in addition to modal verbs, mental state verbs, e.g. I think, modal adjectives and adverbs, e.g. possible, probable, probably. The implementation of this approach to Spanish is a valuable topic for further research. 14. The existing corpus analyses do not describe the evidential and modal verbs either quantitatively and qualitatively. Although Olbertz (1998) describes the frequency distribution of the verbal periphrases, she only does so on the basis of a small corpus of peninsular Spanish. Fernández de Castro (1999) offers a qualitative description of the corpus data, while Marín-Arrese et al. (2004) limit the analysis to a quantitative description only. 15. The corpus can be consulted at http://www.corpusdelespanol.org. 16. One of the advantages of this corpus is its very fast and powerful search engine which allows a wider range of searches than almost any other large corpus in existence. Moreover, the frequency and the use of all morphological forms (conjugations, plurals, etc) are available for more than 20.000 words. The main problems in using this corpus for research purposes are the limited access to the context of the forms requested, the lack of direct geographical information and the time-consuming manual extraction of the data. 17. I used the complete 20th century corpus only for amenazar and prometer, because these verbs are less frequent than the other verbs examined in this volume. For parecer and resultar I have used the non-fiction prose section of the Corpus del español is (6.800.000 words) and a subpart of the oral section, namely the Corpus del habla culta (705.500 words), which contains Latin American and peninsular Spanish spoken by university students of several capitals. The examples of the modal verbs come from the same Corpus del habla culta and from a selection of opinion texts published in the Spanish
264 Notes newspaper El País (378 028 words). The restriction to the subcorpora for the other verbs was mainly motivated by practical reasons. Since the Corpus del español only gives access to linguistic forms that show up less than 2000 times and due to the fact that many constructions with parecer, poder and deber occur much more frequently, I have extracted data from an off-line subcorpus by means of Abundantia Verborum, which was developed by Dirk Speelman (1997), at the Leuven Department of Linguistics. 18. All participants are in their late twenties or thirties and have a university degree. The Spanish ones come from Castilia or Aragón. The Mexican informants live in central Mexico, namely Mexico DF and Querétaro. 19. Traditionally, philosophers of language have been involved in determining how the linguistic code conveying this information corresponds to logical representations in the mind. The modular view of language can be seen as coming from this tradition. 20. Note that I use the term “functional” in the Anglo-Saxon tradition, i.e. to refer to the communicative and cognitive function of language. In the field of Spanish linguistics, “functional” most often refers to the structuralist gramática funcional developed by Emilio Alarcos in the 1950s, which is widespread in Spain. 21. Grice’s William James Lectures (1967) paved the way for an inferential model of communication, where communication is viewed as intentional behavior, and understanding an utterance as recognizing the intentions behind it. Since Austin (1962) and Searle (1969) launched Speech Acts, Grice (1975) published his Conversational Maxims, and Sperber and Wilson (1986) presented their Relevance Theory, the pragmatic dimension of language has become increasingly important. 22. Ifantidou (2001) includes a very broad group of “evidentials” by discussing frankly, confidentially, seriously, unfortunately, sadly, happily, evidently, obviously, clearly, allegedly, reportedly, to believe, to admit, to suppose, to hear, to think, the Greek particle taha (equivalent of seem), to name just a few. 23. Ifantidou (2001) argues that evidential markers are higher level explicatures by which the speaker aims at optimal relevance. Sperber and Wilson (1986: 183): “On the analogy of “implicature”, we will call an explicitly communicated assumption an explicature. Any assumption communicated, but not explicitly so, is implicitly communicated: it is an implicature”. 24. Dendale (1994: 37) also claims that the hearer subjectively evaluates whether the information is reliable or not. More concretely, he states that the modal value of devoir, unlike the evidential qualification, is never stable, and hence suggests that the epistemic speaker commitment cannot generally be mixed with the source of information. 25. Some linguists make the link with truth-conditionality. Willett (1988: 55), for example, defines evidentiality as the domain which deals with “how the speaker obtained the information on which s/he bases an assertion”. Anderson (1986: 274), for his part, considers evidentiality a linguistic category that
Notes
26.
27. 28. 29. 30. 31.
32.
33.
265
“shows the kind of justification for a factual claim which is available to the person making the claim”. Finally, Rooryck (2001: 125) relates evidentials to the evaluation of “the truth value of a sentence both with respect to the source of the information contained in the sentence, and with respect to the degree to which this truth can be verified or justified”. In other words, this definition establishes a direct link between factivity and evidentiality. Besides the distinction between epistemic modality and evidentiality, there are more dimensions proposed in the literature. DeLancey (1997, 2001) argues for “mirativity”, as different from evidentiality, in that it refers to the unexpected status of information. When dealing with evidentiality, the French tradition refers to both evidentialité (Vet 1988; Dendale and Tasmowski 1994) and information médiatisée ‘mediated information’ (see Lazard 1956: 149; Guentchéva 1994: 8, 1996). Palmer’s position is heavily criticized in the literature amongst others by De Haan (1999) and Faller (2002). Here I refer to the evidential values that are not based on direct perception. The result of the speaker’s assessment is expressed in terms of speaker commitment to the proposition. Delbecque (2000: 156) uses the notion of conceptual distance to determine the conceptual constraints for the paradigmatic procedure of cliticization in Spanish. Interestingly, Nuyts (2001b: 11) points out that despite Langacker’s (1987) emphasis on the role of imagery in conceptualization, suggesting an imagistic view of conceptualization, Langacker (1987) also maintains that predicates – a clearly propositionalist notion – are the basic building blocks of so-called conceptual structure. Moreover, Nuyts (2001b: 11) shows that Langacker himself occasionally states that his pictorial diagrams are no more than “informal” renderings of meaning, and that his view “does not imply any necessary commitment to sensory imagery as an exclusive or essential facet of the meaning of linguistic expressions” (Langacker 1987: 111). Nuyts (2001b: 11) concludes that this is “somewhat surprising since in practice Langacker’s semantics is crucially determined by his schematic representations, and thus by a pictorialist or imagistic perspective”. In other words, the cognitive architecture is far more complex than sometimes assumed in Cognitive Grammar (a point which is also directly or indirectly advocated by Nuyts and Pederson 1997; Nuyts 2001b; Wilson and Sperber 2002; Seuren 2002, 2004). At this stage it is not clear, however, to what extent the conceptual level influences linguistic representation. Levinson (1997) argues that the conceptual representation and the linguistic representation must be separate and that cultural differences lead to different conceptual representations in languages.
266 Notes Chapter 2 34. The Chi-square significance test has been applied to this table. The ChiSquare probability is lower than 0.0001. Thus, it can be concluded that the relationships found in the data are statistically significant, and hence, not just due to coincidence. I am very grateful to Freek Van de Velde for his kind assistance with the statistical calculations. 35. Note that within the group of parecer constructions with a que-clause, there are 48 parece ser que ‘it seems to be that’ in the written corpus and 2 in the oral one. (respectively 2.07% and 0.24% of all parecer constructions). Besides ser ‘to be’ and estar ‘to be’, suceder ‘to happen’ can also be inserted in parece que ‘it seems that’, as in the following example: ... esos jóvenes habían iniciado los primeros contactos con la ciencia desde la curiosidad y hasta el entusiasmo. Sin embargo, de alguna manera parecía suceder que la propia enseñanza de las ciencias alejaba a una parte importante. (Encicl: Enseñanza de las ciencias) ‘... these youngsters had initiated the first contacts with science with curiosity and even with enthousiasm. Yet, in one way or another it seemed to happen that the teaching of the sciences itself drove a considerable portion away’ 36. The “other” group mainly corresponds to the comparison readings of parecer (parecerse, parece como si). 37. The me parece que construction constitutes 80 % of these constructions. 38. The Chi-square significance test is applied to this table. Since the Chi-Square probability is lower than 0.0001, it can be concluded that the relationships found in these data are statistically significant. 39. The “other” group includes mainly resultar de ‘to result from’ constructions. 40. Since the term “evidentials” refers to morphosyntactic evidential markers, I use the term “evidential types” or “evidential values” instead. 41. The overviews of Anderson (1986) and Plungian (2001) exceed the scope of the present volume. The indirect evidential values they comment on run parallel to those proposed above and the direct evidential values they extensively deal with, cannot be applied to parecer and resultar. See Squartini (2001) for a discussion of Anderson (1986) and Willett (1986). 42. Just like Faller (2002), I verticalize Willet’s (1988) taxonomy. 43. This is a strong reduction which reaffirms clear borders between grammar and lexicon. Grammaticalization and other studies have shown that the border is not neat. 44. See also a special issue of the Italian Journal of Linguistics (in press) on lexical evidentiality, edited by Mario Squartini. 45. The diversity of Willett’s evidential types is based on evidentials from morphosyntactically very different languages. 46. Given the fact that Nuyts (2001b) clearly differentiates between evidentiality and epistemic modality, it should be noted that this author considers the speaker commitment that corresponds to evidential qualification and the judgment of the likelihood of the event as two different things.
Notes
267
47. Verhagen (2005) distinguishes between personal and impersonal (inter)subjectivity and attributes a specific role to the conceptualizer. His model combines Cognitive Grammar subjectivist claims about conceptualization with attention to the speaker-hearer interaction. 48. Albeit without reference to (inter)subjectivity, Faller (2002: 8), in her dissertation on Quechua evidentials, also attempts to further subdivide between evidence used by the speaker and shared information in terms of two independent scales. On the one hand, the author wants to differentiate between direct and inferential knowledge on the basis of a scale (i), “that is ordered according to the amount of inference involved in arriving at a statement”. On the other hand, Faller proposes a second scale that is ordered according to the number of intervening speakers (ii). (i). visual > auditory > other sensory > inference from results > reasoning (ii). (direct) > secondhand > thirdhand > hearsay /folklore
49. 50. 51.
52. 53.
54. 55.
56.
Since Spanish is not rich in morphosyntactic evidentials, the ordering in scale (ii) will not be followed. Instead the concept of (inter)subjectivity will be used. My findings coincide with those of Bermúdez (2002: 26), who calls it deduction instead of inference: puede deducirse de la evidencia directa ‘can be deduced from direct evidence’. For the sake of the example, I only mention me ‘me’ and le ‘him/her’ although the whole dative paradigm (singular and plural) could be used. The dative with parecer could also be called experiencer. However, since the latter term is most often used in the context of psychological reaction, e.g. le molesta ‘it bothers him/her’, I prefer the term “conceptualizer” (cf. Langacker 1987). The conceptualizer refers to the entity that views the scene, which can be the speaker, the hearer or another entity present in the construction. Anderson (1986: 284) labels it seems to me “experiential inference”. That is, it is a kind of direct inference based on the conceptualizer’s own experience. Combé (1981: 190) describes constructions like parece ser ‘it seems to be’ as bearing “an emphatic, matter-of-fact or explanatory force”. These qualifications can be related to the different possible sources of information. The explanatory force is inferential in nature; the matter-of-fact reading has to do with the possible hearsay evidence. The left-dislocation is obviously not restricted to the subject but can also involve an object. This is in line with Chafe’s argument that inference, although produced on the basis of a concrete stimulus, takes place “without any indication of what the nature of the evidence was” (Chafe 1986: 266). The source of information used for an inference should necessarily remain “unspecified”. Unlike left-dislocation or the infinitival construction, the parenthetical use of parecer cannot be considered a construction type with a specific syntax. In a
268 Notes
57.
58. 59. 60. 61. 62. 63.
64.
65.
way, the parenthetical is a clause in a clause with a “relative syntactic autonomy with respect to the propositional content” (Bosque 1990: 32–33) [my translation]. In his study of (Wanka) Quechua repor(ta)tives, Floyd (1996: 931–932) words the relation between hearsay and ‘validation’ (commitment) this way: “The validational notions associated with the reportative are ambiguous and might be better thought of as a byproduct of the degree of confidence in the individual who served as the information source: if the speaker believes the origin’s account is trustworthy, he will also believe the story. In other words, they are implicatures from the speaker’s construal of the origin’s relationship to the designated conceptualization. This goes counter to the view in which reportative evidentials are viewed primarily as devices to relinquish personal reponsibility for the contents of an utterance”. The confidence in one’s own experience is in line with Anderson’s (1986) notion of “experiential inference” for the reading of it seems to me, i.e. the English counterpart of me parece que. See Ifantidou (2001: 119–160) for an overview. Unlike what Porroche (1990) says, there seems to be both a “near-factive” and a “non-factive” reading of parece que. For the sake of clarity, I will mainly deal with its “near-factive” reading. Rooryck (2001: 128) describes the parenthetical in terms of a “circumstantial inference”, which implies that the speaker’s statement can vary. Note that Subirats-Rüggeberg (1987: 122–123) only mentions ser ‘to be’, estar ‘to be’ and tener ‘to have’. Nicole Delbecque has drawn my attention to the combination with the existential haber ‘to be’. When a special pragmatic marker such as ahora ‘now’ is added, the resultar construction can yield an exclusive “hearsay” reading, by which the speaker tries to distance him/herself from the statement. I want to thank Álvaro Octavo de Toledo y Huerta for drawing my attention to the following example: Y ahora resulta que Bush es bueno, pero yo no me lo creo ‘And now it turns out that Bush is good, but I don’t believe it’ Note that, unlike the infinitive and que-clause constructions with resultar, a construction with resultar + adjectival attribute allows for an overt conceptualizer, as in (i). Thanks are due to Béatrice Lamiroy for reminding me of this type of construction. (i) Me resulta difícil describir este sentimiento. ‘It turns out to be difficult to describe this feeling.’ Since language is not static but always in evolution, extensions of readings of one of the described readings may take place. The combination of sentenceinitial resulta que and evidential modal deber ‘must’ in (i) indicates that resultar may be shifting to allow a more subjective statement. The inferential readings of deber are in line with the one rendered by resultar. Moreover, the modals appear to influence the reading of resultar by overruling the intersubjectivity.
Notes
269
(i) Resulta que debe ser hoy el día de las recriminaciones gratuitas y no ambivalentes. (http://www.zonalibre.org/blog/adrian/archives/003547.html) ‘It turns out to be that today must be the day of the free and unambivalent reproaches’ Another explanation of these constructions is that resultar is used by the speaker to make the evidential deber statement stronger. In this case, resultar backs up with intersubjective evidence the modal inferential statement expressed by deber. This points to a shift in the use of the resulta que construction from evidential marker to discourse particle. The speaker highlights that (s)he has generally valid evidence to support his/her statement. Chapter 3 66. Langacker’s paper called “Raising and Transparency” was first published in 1995 in Language 71.1. 7–62. The (2000) chapter is “adapted with substantial revisions”. 67. Other criteria for transparent verbs that Gaatone (1993) proposes, are: (1) they are not transitive, (2) they do not have valency, (3) they do not impede the second verb to be impersonal, (4) they do not affect the passive relation, (5) the syntactic relations are reduced to the agreement between the subject and the verb which precedes the infinitive. Lamiroy (1996) shows that, when one applies all these criteria, there are very few plainly transparent verbs. 68. The same criticism applies to Langacker’s (1995, 2000) list. 69. The Chi-square significance test is applied to this table. The Chi-Square probability is 0.0033, and is, hence, higher than 0.0001. The relationships found in the data cannot be considered statistically significant and may be due to a debatable sample. In other words, if we take another sample there is no guarantee that we would find a similar distribution. 70. Since the data of parecer + infinitive in the oral corpus are not statistically significant, no comparison with the written corpus was possible. 71. The Chi-square significance test has been applied to this table. Since the ChiSquare probability is 0.0034, the relationships found in the data cannot be considered statistically significant. 72. These data corroborate Hernanz’ (1999) observation that parecer + infinitive does not readily combine with inchoative verbs. 73. In the spoken corpus the attributive verbs (58%) are more frequent than in the written corpus. However, in view of the small number of infinitives that follows parecer in the spoken corpus, the distribution of the infinitives and the verbs in the que-clause cannot be analyzed as well as in the written language. 74. The Chi-square significance test has been applied to this table. Since the ChiSquare probability is 0.0780, the relationships found in the data cannot be considered statistically significant. 75. The Chi-square significance test has been applied to this table. Since the ChiSquare probability is 0.6240, the relationships found in the data cannot be considered statistically significant.
270 Notes 76. The fact that only attributive infinitives combine with resultar entails that all other verb types obligatorily appear in resultar’s que-clause. This can explain why action verbs show up in resultar’s que-clause more frequently than in parecer’s que-clause. 77. In Cognitive Grammar, focal participants are important elements of a grammatical relation. “Certain participants are singled out for special prominence irrespective of their semantic roles. Choosing a participant to be subject or object is very much akin to focusing a spotlight on it, by making these selections, the speaker directs attention to the focused participants (as well as to the interconnections which involve them directly), and thereby imposes a particular image on the scene (Langacker 1991: 301). As pointed out by Maldonado (2002: 6), focal participants show important “salience asymmetries which are claimed to correspond to figure/ground organization. The most salient element in a relation is called the trajector (tr) and is characterized as the primary figure. The second focal element the landmark (lm) is characterized as the secondary figure within the profiled relationship. [...] Given this hierarchical arrangement, the grammatical subject is equated with the trajector and the object with the landmark”. 78. See Rooryck (2000) for a comparative description of paraître and seem. Rooryck (2000: 21) states that “seem and appear can be analyzed straightforwardly as involving a configurationally expressed comparison of events or situations both in raising and nonraising contexts”. I argue that the semantics of comparison play a more crucial role in “raising” constructions than in “nonraising” ones. 79. In this chapter I do not deal with parece ser que and the parenthetical constructions. 80. Langacker (1995, 2000) does not address left-dislocation when he discusses focal prominence. I extend the argument presented so far to left-dislocation. 81. Note that in this analysis the 13th century is considered as the first stage of the Spanish language, although one could argue that the very first stage is much earlier. The analysis below is merely indicative and should be considered as a basis for a more refined, future investigation into the evolution of the evidential verbs. 82. Bolinger (1991: 37) only mentions that Keniston (1937: 505) only gives one example of this parecer + infinitive construction, namely one with the copular infinitive estar. 83. I have found only one example of 16th-century resultar + infinitive. The resulta ser construction is preceded by the pronoun se. Since no other examples have been found, it is impossible to draw conclusions from it. Mirárades ora que muy pocas eran las fuerças que contra una pobre donzella como lo yo soy avíades menester, donde se resulta ser poco el vencimiento mayormente para cavallero como lo vós presumís. (Anonymous. Platir. 1533)
Notes
271
‘Look now, very few were the forces that you needed against a poor lady as I am, where small turns out to be the victory, all the more so for the landlord you pretend to be.’ 84. For a discussion of the difference between pragmatic strengthening and subjective construal, see Brisard (2006). 85. However, the “onstage” or “offstage” conceptualizer is not the only question related to subjectification. In Chapter 6, I show that, despite the clear subjective import, prometer and amenazar are still bound to some of the characteristics present in the lexical reading. The subject’s role of these verbs will corroborate Langacker’s (2000: 358) proposal that full transparency and full nonparticipation are “the endpoint in a continuous spectrum of possibilities”: “If raising and control constructions are analyzed as forming a continuum, this has the advantage of not forcing an arbitrary dichotomization (involving substantially different grammatical structures) on examples that vary continuously or are distinguished only by subtle semantic nuances”. (Langacker 2000: 358) 86. I would like to thank Nicole Delbecque for these comments. 87. Importantly, there is another parecer construction that allows for the insertion of me with verbs other than ser ‘to be’: a. Me parece oír su voz. ‘I have the impression to hear his/her voice’ b. Me parece que he oído su voz. ‘I have the impression that I have heard his/her voice’ In the utterances (a) and (b), the “onstage” conceptualizer is not optional, but a constitutive part of this subjective construction, since there is co-reference between the conceptualizer and the infinitive. The pronoun me in the main clause refers to the speaker as the logical (implicit) subject of the infinitive. Chapter 4 88. Although Rooryck (2001a,b) has published two papers on evidentiality, he does not discuss promettre and menacer in terms of evidentiality. 89. These verbs also have an equivalent in the past, namely venir de ‘just finish V-ing’ for promettre and avoir failli ‘just not have + participle’ for menacer. 90. Rooryck (2000) seems to use the term “necessary” literally. However, Palmer (1979) has pointed out that “epistemic necessity is not to be paraphrased as ‘in the light of what is known it is necessarily the case that...’, but by something like ‘From what I know the only conclusion I can draw is ...” (see also Palmer 2001: 89). More recently van der Auwera and Plungian (1998) have argued that possibility and necessity are the endpoints on the modal scale. Future studies should also account for “intermediary values in some of the modality types (e.g. degrees of epistemic probability)” (cf. van der Auwera and Nuyts in print: 18).
272 Notes 91. Rooryck seems to suggest that the modal reading principally arises by virtue of the future perspective. However, since most of the evidential constructions with must have a past complement, this cannot be the only valid explanation (Nuyts 2001b: 200). 92. As shown before, Verhagen (1996, 2000) argues that the “descriptive” reading and the “epistemic” reading are construed in exactly the same line. He proposes the notion of “argumentational orientation” for both the lexical and the subjective reading of beloven/promise and dreigen/threaten (Verhagen 1996: 801). In contrast with Vliegen, however, this author does not link this up with different degrees of likelihood. 93. Against the background of the discussion on attenuation presented in the previous section, the term of “loss” is perhaps not the most appropriate one, since it has been shown that, depending on the framework used, the so-called “lost” elements are pragmatically strengthened or realigned in terms of another profiled relationship. 94. Still, Langacker’s analysis is very much in line with certain generative proposals concerning the link between raising and non-lexical readings (see, for example, Rooryck’s (2000: 44–47) account of raising with menacer and promettre). 95. Some authors question the unidirectionality of grammaticalization and argue for “degrammaticalization” (cf. Ramat 1992, Janda 2001, van der Auwera 2002 among others). 96. Vliegen (2006) mentions that he has found 805 occurrences of dreigen, but only 98 of beloven, in the corpora from the ‘Instituut voor Nederlandse Lexicologie’ in Leiden (about 38 million word forms). 97. The Woordenboek der Nederlandse Taal is the ‘Dictionary of the Dutch Language’, started in the second half of the nineteenth century (De Vries: 1864– 1998). 98. For other differences, see also Brisard (2006) and Lozowski (1999, 2002). 99. In line with the Gricean conversational maxims, it is expected that the promisor ideally tends to hold his/her promise. When one knows (s)he is not in a position to comply with the promise, it is socially preferred that no promise is uttered. Hence, when a promise is not held, the would-be promisor displays socially unacceptable behavior, while the promisee feels cheated. Starting from the assumption that communication is originally and intentionally positive for society, a promise is, thus, by default expected to be accomplished by a committed subject. Yet it should be clear that the explanation presented here describes general tendencies of the verbs and in no way pretends to offer a general guarantee. For example, the commitment of prometer’s subject does not necessarily mean that there is a 100 % guarantee that this realization ultimately takes place. The insincere promisor may have lied deliberately or may have changed his/her ideas about the promised action. In Searle’s (1969: 59) words, “the only way my audience can interpret my [promise] utterance is to assume that I believe that it is not obvious that I am going to do the thing
Notes
273
promised. A happily married man who promises his wife he will not desert her in the next week is likely to provide more anxiety than comfort”. 100. From a social-semiotic point of view, one can argue that it is expected that the threat will not last a long time. It is not linguistically encoded how long the threat expressed by amenazar remains present, nor whether it fades away or is realized, but one can imagine that the temporal range of a threat is kept as short as possible. Moreover, the addresser him/herself is often reluctant to undertake action and an unmaterialized threat is usually in the interest of the addressee. 101. Lo ‘it’ takes up the propositional content of the direct clause. 102. On the internet (a Blog page), I found one example of an emphatic assertion which was directly sent to the interlocutor: Hey, esa descripción no me hace justicia... yo no te pago, te amenazo!!! ‘Hey, this description does no justice to me.. I don’t pay you, I threaten !!’ (http://indiferencia.net/archivos/categorias/weblogs/_se_muerde_la_cola.p hp) 103. It may also be of interest to mention that these figures are higher than those available for Dutch and German. For Dutch dreigen, the context of 18.1 % of the examples points to non-realization, whereas with German drohen it does so in less than 14 % of the cases (cf. Vliegen 2006). 104. It is not possible to compare this with the Dutch and German corpus data, since Vliegen (2006) does not refer to the tense criterion. 105. Due to the number of cells with less than 5 occurrences, these data are not statistically significant. 106. The gerund examples in the past correspond to the use of a gerund construction in a past context, i.e. with preceding verbs in the past tense. 107. This is additional evidence for the argument that lower “factivity” (or here weaker epistemic value) correlates with the availability of features associated with the subject (cf. Chapter 3, Section 2.2.). 108. As far as Dutch beloven and German versprechen are concerned, 5.8 % and 2.9 % of the examples are accompanied by non-realization (cf. Vliegen 2006). 109. Palmer (2001: 89) mentions that “the word necessary is not used in an epistemic sense in ordinary language (as opposed to logical terminology). It would not be normal to say *It is necessary that John is in the office, although it is possible to say, in semi-logical language, It is necessarily the case that John is is in the office”. Previously, Palmer (1979) described “epistemic necessity” in terms of ‘From what I know the only conclusion I can draw is ...”. However, neither of both explanations applies to modal prometer, since this verb is not restricted to one exclusive conclusion. 110. Recall that Rooryck (2000: 43–44) judged the label “possible” too weak for “epistemic” promettre. 111. This is in line with Vázquez-Laslop’s (2001: 246–247) comment that epistemic prometer carries a “modal operator, the scope of which is the whole proposition”, that corresponds to “it is quite likely/it is expected”.
274 Notes 112. The term of mirativity refers to the grammatical marking of unexpected information (cf. DeLancey 1997). 113. This also points to a paradoxical side of prometer: although prometer can combine with both negatively and positively oriented complements, this double possibility does not increase the number of modal constructions with this verb. As shown before, modal amenazar is far more frequent. Chapter 5 114. In the lexical prometer and amenazar constructions the locus of the promise or the threat is the referent of the subject mentioned in the sentence itself, in the modal constructions it is the implicit conceptualizer who evaluates the likelihood. The subjects of the modal constructions do not play an agentive role in the construction. 115. Note that impersonal verbs are naturally excluded in the lexical reading of both verbs, while they readily show up in the modal readings. 116. Other semantic categories found in the infinitives that combine with parecer do not combine with modal amenazar and prometer. Communication verbs, modal verbs, and verbs expressing a mental state or a psychological reaction were not found in the corpus. 117. Note that when the present and past forms are counted together, the data in this table are statistically significant (Chi Square Probability: <.0001). 118. Table 2 also corroborates the general tendencies revealed by Vliegen’s (2006) analysis of the German and Dutch equivalents of amenazar and prometer. Vliegen (2006) shows that beloven/versprechen has a preference for copular verbs (94.2% and 93.2%, respectively), whereas dreigen/drohen most often combine with other verbs (80.8% and 67%, respectively). However, our data show that Spanish prometer is far more flexible in combining with other verbs than attributive ones (which take 50%), while amenazar does not combine as much with attributive verbs (only 1.62%) as Dutch dreigen and German drohen do. 119. This table has a Chi-Square probability of <0.0001, but 25% of the cells have less than 5 occurrences. Although this table gives strong indications of the use of modal amenazar and prometer, it may be not statistically significant. 120. Ruwet (1991: 80–81) bases the different points of view on syntactic properties. According to the author, an internal point of view reading, “where the content of the complement clause is seen from the point of view of the main subject”, is related to the absence of subject-to-subject raising. VázquezLaslop (2001: 259), for her part, distinguishes between an external point of view and an internal one on the basis of different types of modality: the former viewpoint corresponds to epistemic modality, whereas the latter viewpoint stands for “some sort of target-oriented modality” (in verbs like permitir ‘to allow’ and obligar ‘to oblige’). In this volume, I show that the different viewpoints of amenazar and prometer include epistemic modality and differ as to the status of the course of evolution (in the case of amenazar) and the
Notes
275
speaker’s own judgment (in the case of prometer). 121. Although this table gives strong indications about the use of modal amenazar, it is not statistically significant. 122. Other action verbs and verb phrases with modal amenazar are: arrancar el techo de la casilla ‘to tear loose the thatched roof of the box’, revelar misterios o intrigas ‘to reveal enigma’s and intrigues’, comerse los hospitales ‘to eat up the hospitals’, anular todo valor ‘to cancel all value’, destruir nuestra industria ‘to destroy our industry’, impedir su viaje ‘to impede his journey’, incendiar al país ‘to inflame the country’, deglutirla ‘to swallow it’, disparar ‘to shoot’, tragarse el patio ‘to eat up the patio’, llevarse la dentadura postiza ‘to take away the false teeth’, devorar al mundo ‘to devor the world’, quemarle la ropa ‘to burn the clothes to someone’, vomitarle encima ‘to vomit on him’, tragarse mi jardín de las delicias ‘to eat up my garden of delicacies’, obstruir el comercio ‘to hinder the commerce’, llevar a Clinton a pique ‘to lead Clinton to the ground’, confluir con el silencio ‘to converge with the silence’, disparar su carga ‘to drop his load’, llevarse la mujerada por delante ‘to take with him the women’, detener su guagua ‘to stop his car’, asentarse en mi mente ‘to fix in my mind’, apoderarse de todo el terreno nacional ‘to take possesion of the whole national territory, derribar la Tercera República ‘to bring down the Third Republic’, derribar el cobertizo de palma ‘to bring down the palm shelter’, destruir las frágiles carrozas ‘to destroy the fragile covered wagons’, descargar su ira cielo abajo ‘to relieve his rage down’, tomarse a Santa Fe ‘to occupy Santa Fe’, tirar la puerta ‘to kick down the door’, destruir barcos y vidas humanas ‘to destroy boats and human lives’, socavar la autoridad eclesiástica ‘to undermine the ecclesiastic hierarchy’, sobrepasarlas ‘to surpass’, llevarlos a un motín ‘to lead them to a mutiny’, arrastrar el resto ‘to drag the rest’, estallar mi cabeza ‘to burst my head’, devorarlo ‘to devour him/it’, llevarlo al exterminio ‘to lead him to the ground’, regalarme ‘to give away’, pasar a la arboleda y destruirlo todo ‘to go to wood and destroy it all’, asfixiarlo (1) ‘to asphyxiate him’, asfixiarle el corazón ‘to asphyxiate the heart’, llevar a la ruina al Bolshoi ‘to destroy Boshoi’, tirarla al suelo ‘to throw it/him to the ground’, quebrar la superficie de la tierra ‘to break the surface of the earth’, quebrar la celebración ‘to cancel the celebration’, sacar ‘to pull out’, hacer chirriar el engranaje de la programación ‘to make the chainwheel of the programmation crunch’, subir la tasa de interés ‘to raise the interest rate’, disolverle los huesos ‘to dissolve the bones’, soltarse ‘to break away’, hacerlo reventar ‘make him burst open’, borrar la huella de los hombres ‘to erase men’s track’, extender el fuego ‘to extend the fire’, invadirme ‘to descend upon me’, invadir la base fluvial ‘to enter in the fluvial base’, elevar el precio ‘to raise the price’, hacerla estallar ‘to make it burst’, romper el equilibrio constitucional ‘to break the constitutional balance’, estrangularme y acabar con la posibilidad de sobrevivir ‘to strangle and end the possibility of surviving’, meterse dentro de la habitación ‘to go inside the room’, aplastar (2) ‘to crush’. 123. Other change-of-state verbs and verb phrases with modal amenazar are: con-
276 Notes vertirse en un grave problema ‘to become a big problem’, convertirse en la ciudad de las puertas enrejadas ‘to become the city of fenced doors’, convertirse en la trampa más siniestra ‘to become the most sinister pitfall’, convertirse en un viento más ligero ‘to become a lighter wind’, convertirse en un conflicto internacional ‘to become an international conflict’, convertirse en un nuevo incidente diplomático con Italia ‘to become a new diplomatic incident with Italy’, convertirse en el inicio de un enfrentamiento regional ‘to be the beginning of a new regional conflict’, convertirse en delincuentes juveniles ‘to become young delinquents’, abrirse ‘to open’, desplazarse ‘to move on’, explotar ‘to explode’, subir ‘rise’, deshacerse (2) ‘to fall apart’, podrirnos a todos ‘to rot us all’, afectar los intereses de 10 hoteles ‘to affect the interests of 10 hotels’, subirse ‘to go up’, caer ‘to fall’, inundar la habitación entera ‘to drown the whole room’, venirse abajo (2) ‘to come down’, romper toda esperanza ‘to break all hope’, sumir al país en el caos ‘to lose the country in chaos’, terminar con su paciencia ‘to end his patience’, diluirse en una bronca insoportable ‘to end up in an unbearable tiff’, generalizarse ‘to generalize’, desplomarse ‘to crash’, apagarse ‘to die away’, extenderse ‘to extend’, evolucionar hacia una guerra civil ‘to evolve into a civil war’, derramarse y caerse ‘to flow away and fall down’, devorarme ‘to devour me’, llegar hasta donde se encontraban ‘to get where they were’, reventar las costuras de la prenda ‘to burst the sewing of the clothes’, derrumbarse (2) ‘to collapse’, hacerla desaparecer ‘to make her disappear’, perderse ‘to lose oneself’, sumir a Europa en una situación de anarquía ‘to lose Europe in a situation of anarchy’, desbordar toda medida ‘to surpass all measures’, quebrarse ‘to break’, desgranarse ‘to fall into pieces’, inundar los resquicios de su mente ‘to flood the meshes of his mind’, reventar ‘to burst’, radicalizarse ‘to radicalize’, tener serias repercusiones políticas ‘to have serious political repercussions’, afectar a zonas habitadas ‘to affect inhabited zones, acabar con la paciencia de los ciudadanos ‘to end the citizen’s patience’, romper la barrera del sonido ‘to break through the sound barrier’, caer sobre el pueblo silencioso ‘to fall over the silent village’, desbordarse ‘to overflow’, hacerse trizas ‘to be picked to pieces’. 124. The perspective given by amenazar is in line with a state of affairs in which the subject entity has already been affected. 125. Next to seguir + gerund ‘go on V-ing’, other aspectual verbs with amenazar found in the corpus are: no terminar ‘not to stop’, acabarse ‘to stop’ and acabar en guerra ‘to end up in war’. 126. The other attributive construction with modal amenazar is ser una rotunda esfera ‘to be a plain field’. 127. Although this table gives strong indications of the use of modal amenazar, it is not statistically signifcant. 128. I further deal with the role of the subject in Section (2.). 129. Other action verbs with modal prometer are: recuperar la armonía de ese barrio protagónico ‘to get back the harmony of this central quartier’, mostrar toda la riqueza ‘to show all wealth’, dilucidar misterios ‘to clarify enigmas’,
Notes
277
llevar todavía más lejos sus originales propuestas visuales ‘to improve his original visual proposals’, remover ‘remove’, llegar hacia ninguna parte medianamente iluminadora ‘to get to nothing enlightening’, sustituir a numerosos procesos fotográficos ‘to replace various photographic processes’. 130. Other attributive verbs with modal prometer are: ser la más agresiva contienda comercial de los próximos años ‘to be the most aggressive trade war of the coming years’, ser llovedor ‘to be rainy’, ser una de las satisfacciones de mi vida ‘to be one of the satisfactions of my life’, ser uno de los grandes atractivos ‘to be one of the big attractions’, ser duradero ‘to be lasting’, ser apasionante ‘to be fascinating’, ser más fría que de costumbre ‘to be colder than usual’, ser soleado y caluroso ‘to be sunny and warm’, ser grande, excitante ‘to be great, exciting’, ser inmisericorde ‘to be merciless’. 131. Note that in the corpus modal prometer also combines two times with the change-of-state verb convertirse ‘to become’: convertirse en una melena indómita ‘to become an untamed hairdo’ and convertirse en los mayores inversionistas publicitarios ‘to become the biggest advertising investors’. 132. In this context, Lamiroy (1998: 290) states that “la classe des verbes transparents sera par définition hétérogène, puisque tous les verbes ne se situent pas au même endroit de la chaîne [de grammaticalisation]”. [“The class of transparent verbs will be heterogeneous by definition, since not all verbs are situated at the same point in the grammaticalization chain”] 133. The action verbs that combine with modal prometer in the corpus do not involve patients or active participants, while they sometimes do with amenazar. 134. The label “Zero” refers to those subjects which remain implicit, i.e. are limited to the person marking expressed in the verbal morphology. 135. The data presented in this table are statistically significant (Chi Square Probability: <.0001). 136. Vliegen (2006) states that with Dutch dreigen and German drohen the typically weak commitment of the subject in the lexical reading explains why the majority of the subjects in modal readings are animate. My analysis does not confirm this observation for Spanish. Although the commitment and agentivity of the subjects of amenazar is certainly reduced, the animate subjects of amenazar bearing potential agentivity are not at all frequent. Chapter 6 137. Due to the limited number of examples, the periodization given in Traugott’s analysis should be taken with some caution. 138. Traugott’s (1996: 188) example of threaten, however, does not readily fit into the meaning of ‘prediction’. In this context it is worth mentioning that Traugott only mentions one example of threaten, whereas promise is represented in various examples. My own corpus research on amenazar shows that at the first stages the verb yields a “prediction” reading when it combines with
278 Notes a nominal complement. Further evidence could show a similar tendency for threaten. 139. Traugott and Dasher (2002: 28) prefer the notion of “conceptual structure” to that of “domain”, which is usually used by Sweetser (1990) and Langacker (1987, 1991) amongst others. The former authors argue that the borders of a domain are far from clear. 140. Sweetser’s account of the alternation between deontic and epistemic modals relies on the metaphorical extension of the force structure proposed by Talmy (1985), which is present in deontic (root) modality, to epistemic modality. Langacker’s (1991) proposed grounding predication function of modals is largely based on Sweetser’s (1990) metaphorical account, while Langacker (1999, 2000) proposes attenuation of subject control as a process parallel to subjectification. 141. Although, as shown in Chapter 4, the reading of prometer in constructions with strong illocutionary force can be one of threat. Thus, in some cases there is sociophysical force. 142. Note that Verhagen’s (2000) analysis situates this “personification” shift in the 16th and 17th century, which is a considerably earlier stage than the one Traugott (1997) proposes for English. It may be that in English, too, the “personification” reading was already available in the 16th and 17th century. 143. Neither Traugott (1997) nor Verhagen (2000: 206) give concrete examples of threaten with a nominal complement (although Verhagen refers to their existence). 144. As far as I know such a study has not been carried out for Spanish yet, nor does Corominas and Pascual’s (1984) etymological dictionary mention the modal readings of these verbs. 145. I am very grateful to Álvaro Octavio de Toledo y Huerta for his comments on my analysis of amenazar and prometer, which have helped me avoiding a series of pitfalls. All remaining problems are my own responsibility, of course. 146. In (9a), amenazar appears in the main clause, while the one in (9b) is situated in the que-clause. The que-clause position seems to be preferred by the early evidential expressions. 147. Note that in this amenazar construction, the complement introduced by the preposition de ‘from’ is an alternative construction to the nominal complement with an object function. In the following example, from 1559, the noun is not linked with amenazar through the preposition de: -… todas ellas [cosas criadas], cuanto de sí es, amenazan caída y, por decirlo que es, caminan cuanto es de suyo al menoscabo y al empeoramiento. (León, Luis de. De los nombres de Cristo 1559) ‘All these created things, insofar as they are his, threaten ruin and, to say like it is, insofar it is his, they go in the direction of discredit and decline’ 148. Note that the preposition de is used instead of con, the latter being the most common one in present-day Spanish.
Notes
279
149. Interestingly both constructions appear in a relative que-clause. Thus, from the early modal evidential readings on, a main clause position of the construction does not seem to be preferred. 150. This time the two verbs are linked by the preposition a ‘to’, instead of the preposition de ‘from’. The preposition con, which is nowadays most frequent, is a more recent evolution. 151. Note that several times modal amenazar shows up with the que-clause: No entiendo este misterio, podría ser que este fuese juicio y castigo dellos. Pero también podría ser que lo fuese de nosotros, y amenaza que las persecuciones de la fe no son acabadas. (Meneses, Felipe de. Luz del alma cristiana. 1556) ‘I do not understand this mystery, but it could be that it was a judgment and punishment of them. But it could also be that it was ours, and it threatens that the persecutions because of religion are not finished’ 152. This evolution is contrary to that of parecer, which first combines with the copular ser and then shifts to other verbs (see Chapter 3). 153. The corpus data presented in Chapter 5 (Section 1.3.) indicate that in presentday Spanish the attributive infinitives do not often combine with modal evidential amenazar. 154. The lack of diachronic evidence for the illocutionary force does not mean, however, that in the present-day use modal evidential prometer cannot be influenced by the lexical prometer reading with illocutionary force. It only questions that, historically speaking, illocutionary force motivates the early modal uses. 155. In this section, I use “objective” and “subjective” in a Langackerian way, i.e. referring to an objective scene or construing the normally objective path subjectively. 156. “Verbs which ‘lose’ their canonical thematic structure to function as raising verbs receive a meaning which combines temporal properties with epistemic modality” (Rooryck 2000: 45). 157. Other amenazar constructions in the 15th century are not completely exempt of threat. 158. In Chapter 3, I have argued that with parecer + infinitive the preposed subject co-determines the dynamic conceptualization of the infinitival process. Chapter 7 159. Nuyts (2001b: 207–208) argues that the Dutch modal auxiliary kunnen ‘can/may’ can yield an intersubjective reading in some cases. In my opinion, intersubjectivity does not readily apply to subjective modals. 160. Mortelmans (2001: 133) refers to the epistemic dürfte ‘may’ in German. 161. Gómez Torrego (1999: 3359–3362) mainly deals with poder in formal terms and makes no mention of a possible correlation between the “capability” or “obligation” readings and the “possibility” reading.
280 Notes 162. See Verstraete (2001) and Nuyts (2001b) for a discussion on objective and subjective modality. 163. Tasmowski and Dendale (1994) argue for a purely evidential reading of pouvoir. Their argument is based on logical operations that lead to several possible conclusions, in contrast to the unique conclusion with devoir. However, my corpus data presented in Chapter 7 will show that the inferential dimension of poder is only marginal. 164. Gómez Torrego (1999) respects the normative prescription that deber de is used for “probability” readings, while deber expresses “obligation”, although he recognizes that in present-day use both readings are associated with deber without de. My corpus data show that in Mexico deber de is frequently used to express an obligation. For a good description of the debate see Fernández de Castro (1999: 172–176). 165. Olbertz (1998) uses capitals for Proposition-oriented Epistemic Necessity. In line with the description of the other concepts, I prefer to put this and other notions between quotation marks. 166. Silva-Corvalán (1995: 90–91) concludes this from its absence in Moliner (1981). 167. The evidential dimension is not only to be based on visual information of another type of direct evidence but also involves the reasoning provoked by direct or indirect evidence: it is a process of “creation of information” (Dendale 1994). 168. In this utterance, devoir has either a deontic or an evidential reading. 169. This dynamic reasoning based on inference can be an alternative to the rather static belief that motivates the “certainty reading” of tener que (cf. Olbertz 1998: 413). 170. Abductive reasoning has the logical form of an inverse modus ponens and is “reasoning backwards” from consequent to antecedent. Therefore Peirce calls it also “Retroductive reasoning” (1965: 74). From a classic logical point of view, reasoning backwards is not a valid form of inference. It is conjectural, or presumptive thinking, aiming at matching pragmatic standards of plausibility, guided by the reasoner´s “guessing instinct” (1965 (7): 46). Peirce, however, claims that abduction is logical inference, because it can be represented in a “a perfect definite logical form” (1965 (5): 188). 171. See Goethals (2002) for a semiotic-linguistic study of the Spanish causal conjunctions porque ‘because’, ya que ‘since’ and como ‘since’. 172. Kronning (1994, 2001) and Squartini (2004) revise Dendale’s (1994, 1999, 2000) proposal for deductive inferentiality of devoir by reducing the ‘domain of application’ of inferential deduction and inferential evidentiality in general. 173. This hypothesis is not incompatible with Dendale’s argument that the evidential value is the basic value of devoir and that the modal value is derived from it. However, my position differs from Dendale’s claim that the evidential value is unique and stable (only inference), whereas its modal value is plural and variable (different degrees of certainty). Since reported
Notes
281
knowledge can also be invoked, the evidential value can hardly be called unique and stable. 174. Nuyts (2001b: 37–38, 207–208) relates the “objective modality” of Dutch kunnen ‘may’ with intersubjectivity, however, without discussing moeten ‘must’. It seems that, although must conveys an evidential qualification, its “objective” reading is not necessarily intersubjective. 175. Interestingly, Squartini (2004: 888) contrasts the counterfactual use of deber with the more factual use of tener que. Look at the deontic examples: a. El profesor tuvo que castigarme porque no estudié. ‘The professor had to punish me because I didn’t study.’ b. El profesor debió castigarme porque no estudié. ‘The professor should have punished me because I didn’t study.’ 176. Both verbs also have a deontic reading; sollen conveys a stronger obligation than müssen. 177. The result of the joined tables 1 and 2 has a Chi-Square probability lower than 0.0001 and is hence statistically significant. 178. In this context, Kronning (1996, 2001) argues for recognizing an “alethic necessity” reading as a bridge between deontic and epistemic readings. 179. Recall that Olbertz (1998) only considers deber to be an inferential modal. 180. Since these verb classes have been exemplified in the other chapters, there is no need to repeat them. The only specification required is the following: the ser/estar + past participle of the verb are both categorized as attributive verbs, notwithstanding the possible semantic differences in the passive infinitive. 181. The Chi-square significance test is applied to this table. The Chi-Square probability is lower than 0.0001 (without remarks on the low counts of certain cells). Thus, it can be concluded that the relationships found in the data are statistically significant. 182. The Chi-Square probability of this table is lower than 0.0001. However, since 25% of the cells have expected counts less than 5, the Chi-Square may not provide us with statistically significant information. 183. The Chi-Square probability of this table is lower than 0.0001. However, since 22% of the cells have expected counts less than 5, the Chi-Square may not be valid. 184. The Chi-Square probability of this table is lower than 0.0001. However, since 23% of the cells have expected counts less than 5, the Chi-Square may not be valid. 185. The Chi-Square probability of this table is lower than 0.0001. However, since 39% of the cells have expected counts lower than 5, the data are not statistically significant. 186. The Chi-Square probability of this table is lower than 0.0001. Since 41% of the cells have expected counts less than 5, this table is not statistically significant.
282 Notes Chapter 8 187. See also Goossens (1982, 1999, 2000) on the development of the English modals, Guo (1995) on Chinese modals, and Nuyts (2001b) on the Dutch modals. 188. Elvira (2002, 2003) outlines different stages of semantic change from deontic or dynamic modality to epistemic modality for deber and poder. 189. It stands to reason that the epistemic reading of poder (and perhaps the deontic reading) originates from dynamic poder, which expresses the subject’s capacity to do something. Thus, it does not seem to be justified to treat deontic deber and deontic poder on the same basis. Further research should shed a new light on this topic. 190. I base my analysis on the corpus data from the on-line Corpus del español, which offers lists per hundred examples. These overviews allow me to detect tendencies in the usage of the verbs under consideration. However, the examples remain to be converted into a relational database, for a more fine-grained quantitative description to be possible. This is left for further research. 191. Yllera (1980: 117) mentions only a few lines on the use of tener que. She observes that this construction is not frequent in the 15th century, and she neither discusses the possible link with tener de nor commenting on the rise of the evidential readings. 192. Interestingly, Olbertz (1998: 251) refers to the evolution of the Latin auxiliary habere and states that the Latin infinitive is used as a means of expressing non-factive, future oriented readings as an alternative to the Latin gerundivum form (dare aquam bibere vs dare aquam bibendam ‘give water to drink’): the infinitive gradually took over the function of the gerundivum. Importantly in Romance there are no gerundivum forms. 193. Interestingly, tener que also shows up as a kind of “mental state” verb. This use is regularly found until the beginning of the 17th century. Mas si alguno tiene que ha derecho en alguna cosa que otri touiere en iuro de paz demandegelo por el fuero. (Alfonso X. Fuero real. 13th century) ‘But if someone thinks he has the right to something that other people possessed, reclaim it before the tribunal’ 194. Grounding has inspired many cognitive linguists to analyze nominal and verbal constructions in terms of the objective perception and/or subjective elaboration of the linguistic expression (cf. Pelyvás 1996, Brisard 2002, Heyvaert 2003). 195. Some past forms of deontic modals can be performative. Debió haberlo hecho inmediatamente. ‘(S)he should have done it immediately’ 196. Pelyvás (1996, 2000, 2003) differs from Nuyts (2002) in that he subscribes to the basic assumptions of the Cognitive Grammar. Pelyvás (1996: 156) is not convinced by the “pragmatic strengthening” approach, which, according to
Notes
283
him, focuses too much on the alleged impact of Grice’s Maxims, which are “very often interpreted and reinterpreted in incompatible ways”. 197. Only (indicative/subjunctive) mood combines both the modal and the temporal type of grounding in Spanish. 198. Since Langacker (1990) does not give concrete examples, I reprint Radden’s (1999) examples when needed. 199. Langacker (1990: 28) refers to can and will as main verbs having a subject that functions as a trajector of both the verb and the associated process, although he “suspect[s] that they no longer profile the relationship of potency”. This “exception” shows that the formal criterion of absence of tense marking is not always on a par with the conceptual side of the construction. 200. The non-modal reading of deber ‘owe’ below does not have an onstage locus of potency either. The grammatical subject Su Alteza ‘Your highness’ has to comply with the reimbursement of the money. Tanbién te dise que yo he gastado para traher esa gente a Castilla mil y duzientos castellanos, los cuales me debe Su Alteza la mayor parte de ellos, y por esto le escriví que me mandase a tomar la cuenta. (Colón, Cristobal. Textos y documentos completos de Cristobal Colón. 15th century) ‘It also says that I have spent 1200 ducates to bring these people to Castilia, of which you owe me the major part, and therefore I wrote you to request you to make the bill for me.’ The absence of such examples in the 13th and 14th centuries of the Corpus del español leads me to hypothesize that this use shows up later than the modal readings. This deber construction is then an example of “degrammaticalization” (van der Auwera and Plungian 1998: 116–117). 201. Recall that deontic modality concerns the speaker’s (or another participant’s) intervention in the proposition, while epistemic modality refers to the belief state toward the proposition. 202. The restriction of grounding predications to epistemic modality contrasts with Mortelmans’ (2002) analysis, which considers speaker-oriented deontic readings of sollen ‘to have to’ and müssen ‘to have to’ as grounding predications, thereby excluding other deontic readings. 203. The dynamic evolutionary model is an elaboration of the basic epistemic model (cf. Langacker 1991: 240–9, 275–81). In the epistemic model, basic reality represents the level at which events are directly “observed, dreamed, imagined, ...” (Achard 2002: 208). The model is based on three theoretical notions: known reality, immediate reality, and irreality. These three ways of looking characterize the speaker’s basic conception of reality. 204. In English, the same opposition exists between must and should. The should construction cannot shed the deontic overtones. a. They must be dead by now. b. They should be dead by now. 205. Another construction I found on the Internet shows that deontic deber can be preceded by evidential parecer.
284 Notes No puede ser ninguno de los elementos, de modo que parece deber concluirse que tiene que ser algo así como la masa indiferenciada de los elementos previa a toda “formación” (José Ferrater Mora 1979: 2135– 2143) ‘It cannot be any of the elements, subsequently he seems to have to conclude that it has to do with something as the undifferentiated mass of elements previous to whatever “formation”.’ 206. In this context, Langacker’s (1999, 2000) proposal of “attenuation” or the progressive reduction of subject control in the finite verb forms facilitates the analysis of epistemic and deontic modal constructions in terms of a gradient shift from a strong interplay of forces to no force at all. 207. The term reference-point was first used by Rosch (1975) to refer to the relation of a prototype and other elements of a certain domain. Rosch (1975) proposes prototypes as a reference point in making judgements on the similarities and differences in other experiences and things in the world. Langacker gives another function to reference-point. The reference-point function derives from the cognitive procedure of mental scanning of what is linguistically encoded (Langacker 2003: 3). When the ground is implicit, an entity is usually grounded by means of a “Reference-point” (RP) which connects with the empirical world and delivers the referential value of the entity. 208. Gómez Torrego (1999: 3359–60) points out that not only deontic but also (root) epistemic poder can in certain contexts stand alone without an explicit infinitive, and provides the following example: A: ¿Puede estar tu padre en casa ahora? ‘Can your father be at home at the moment.’ B: Sí, puede. ‘Yes, he can/may.’ I am not sure whether this example points to a real epistemic use, since epistemic modifiers do not readily show up in interrogative sentences (cf Nuyts 2001b). The interpretation of the question in A is the following: “given the knowledge you have about the situation can your father be at home?” More prominence is given to the objective possibility that the father is at home. 209. Although evidential parecer and amenazar/prometer do not stand for a strong relation between participants, they cannot be considered to have the same reference-point function as the epistemic modals. For example, parecer can show up independently in constructions like por lo que parece ‘as it seems’ or lo parece ‘it seems so’, which illustrates that this expression can still be profiled itself. As far as amenazar and prometer are concerned, I have shown in Chapter 6 that the profile of these verbs does not completely fade away either, which is confirmed by the thematic relation these verbs maintain with the subject. Thus, these evidential verbs cannot be mere reference-points to the infinitival process.
Notes
285
210. Note that other deontic forces associated with more abstract entities are far more diffuse than that of permission. 211. These figures are taken from Pelyvás (1996: 168). 212. I am indebted to Ricardo Maldonado for drawing my attention to this specific point.
References Achard, Michel 1998 Representation of Cognitive Structures. Syntax and Semantics of French Sentential Complements. Berlin/New York: Mouton de Gruyter. 2002 Selección de modo en construcciones oracionales de complemento. In Estudios cognoscitivos del español. Revista Española de Lingüística Aplicada, Ricardo Maldonado (ed.), 111–130. Castellón/Querétaro: AELA/UAQ. Aijmer, Karin 1980 Evidence and the Declarative Sentence. Stockholm: Almquist/ Wiksell. Aikhenvald, Alexandra Y. 2004 Evidentiality. Oxford: Oxford University Press. Aikhenvald, Alexandra Y. and Robert M.W. Dixon (eds.) 2003 Studies in Evidentiality. Amsterdam/Philadelphia: Benjamins. Aksu-Koç, Ayhan A. and Dan I. Slobin 1986 A psychological account of the development and use of evidentials in Turkish. In Evidentiality: The Linguistic Coding of Epistemology, Wallace L. Chafe and Johanna Nichols (eds.), 159–167. Norwood NJ: Ablex. Anderson, Lloyd B. 1986 Evidentials, paths of change, and mental maps: typologically regular asymmetries. In Evidentiality: The Linguistic Coding of Epistemology, Wallace L. Chafe and Johanna Nichols (eds.), 273–312. Norwood NJ: Ablex. Austin, John L. 1962 How to Do Things with Words. Oxford: Clarendon Press. van der Auwera, Johan 2002 More thoughts on degrammaticalization. In New Reflections on Grammaticalization, Ilse Wischer and Gabriele Diewald (eds.), 19– 29. Amsterdam/Philadelphia: Benjamins. van der Auwera, Johan and Vladimir A. Plungian 1998 Modality's semantic map. Linguistic Typology 2: 79–124. Barcelona, Antonio (ed.) 2000 Metaphor and Metonymy at the Crossroads. A Cognitive Perspective. Berlin/New York: Mouton de Gruyter. Barnes, Janet 1984 Evidentials in the Tuyuca verb. International Journal of American Linguistics 50: 255–271. Benveniste, Emile 1958 Problèmes de Linguistique Générale. Paris: Gallimard.
References
287
Bermúdez, Fernando 2002 La estructura evidencial del castellano. Elevación de sujeto y gramaticalización. Romansk Forum 16: 19–29. 2003 Llegando a la conclusión: la escena del camino en los conectores consecutivos. Revue Romane 38 (2): 239–271. Blanche-Benveniste, Claire, José Deulofeu, Jean Stéfanini and Karel vanden Eynde 1987 Pronom et syntaxe: l'Approche Pronominale et son Application au Français. Paris: SELAF. Bolinger, Dwight L. 1946 Spanish parece que again. Language 22. 359–360. 1961 Syntactic blends and other matters. Language 37 (3): 366–381. 1980 Wanna and the gradience of auxiliaries. In Wege zur Universalien Forschung, Günter Brettschneider and Christian Lehmann (eds.), 292–299. Tübingen: Gunter Narr. 1991 Essays on Spanish: Words and Grammar. (Hispanic Monographs Essays). Newark, Delaware: Juan de la Cuesta. Bolkestein, Machtelt 1980 Problems in the Description of Modal Verbs. An Investigation of Latin. Assen: Van Gorcum. Bosque, Ignacio (ed.) 1990 Indicativo y Subjuntivo. Madrid: Taurus. Botne, Robert 1997 Evidentiality and Epistemic Modality in Lega. Studies in Language 21: 509–529. Brisard, Frank 2006 Logic, subjectivity, and the semantics/pragmatics distinction. In Subjectification: Various Paths to Subjectivity, Angeliki Athanasiadou, Costas Canakis and Bert Cornillie (eds.), 41–74. Berlin/New York: Mouton de Gruyter. Butt, John and Carmen Benjamin 1996 A New Reference Grammar of Modern Spanish. London: Edward Arnold. Bybee, Joan 1985 Morphology: A Study of the Relation between Meaning and Form. Amsterdam/Philadelphia: Benjamins. Bybee, Joan, Revere Perkins and William Pagliuca 1994 The Evolution of Grammar: Tense, Aspect and Modality in the Languages of the World. Chicago: University of Chicago Press. Bybee, Joan and Paul J. Hopper (eds.) 2001 Frequency and the Emergence of Linguistics Structure. Amsterdam/Philadelphia: Benjamins. Cabeza Pereiro, Carmen 1997 Las Completivas de Sujeto en Español. Santiago de Compostela: Universidade de Santiago de Compostela.
288
References
Chafe, Wallace L. 1986 Evidentiality in English conversation and academic writing. In Evidentiality: The Linguistic Coding of Epistemology, Wallace L. Chafe and Johanna Nichols (eds.), 261–272. Norwood NJ: Ablex. 1987 Cognitive constraints on information flow. In Coherence and Grounding in Discourse, Russel Tomlin (ed.), 21–51. Amsterdam/Philadelphia: Benjamins. 1994 Discourse, Consciousness and Time. Chicago: The University of Chicago Press. Chafe, Wallace L. and Johanna Nichols (eds.) 1986 Evidentiality: The linguistic coding of epistemology. Norwood NJ: Ablex. Claudi, Ulrike and Bernd Heine 1986 On the metaphorical basis of grammar. Studies in Language 10 (2): 297–335. Coates, Jennifer 1983 The Semantics of the Modal Auxiliaries. London: Croom Helm. Combé, Henk A. 1981 Some discrepancy Phenomena in Spanish. In Predication and Expression in Functional Grammar, A. Machteld Bolkestein, Henk A. Combé, Simon C. Dik et al., 185–203. London/New York: Academic Press. Company, Concepción 1995 Old forms for new concepts: the recategorization of possessive duplications in Mexican Spanish. In Historical linguistics 1993: Selected Papers from the 11th International Conference on Historical Linguistics, Henning Andersen (ed.), 77–92. Amsterdam/ Philadelphia: Benjamins. 1998 The interplay between form and meaning in language change: grammaticalization of cannibalistic datives in Spanish. Studies in Language 22 (3): 529–565. 2006a Zero in syntax, ten in pragmatics: Subjectification as syntactic cancellation. In Subjectification: Various Paths to Subjectivity, Angeliki Athanasiadou, Costas Canakis and Bert Cornillie (eds.), 375–397. Berlin/New York: Mouton de Gruyter. 2006b Subjectification of verbs into discourse markers. Semantic-pragmatic change only? In Topics in subjectification and modalization, Bert Cornillie and Nicole Delbecque (eds.), 97–121. Amsterdam/ Philadelphia: Benjamins. Copi, Irving and Carl Cohen 2003 Introduction to Logic. New York: Prentice Hall. Cornillie, Bert 2003 Subjetivización, predicaciones de anclaje y los modales españoles. Foro Hispánico 24: 21–34.
References 2004a
289
The shift from lexical to subjective readings in Spanish prometer ‘to promise’ and amenazar ‘to threaten’. A corpus-based account. Pragmatics. 14 (1): 1–30. 2004b Epistemic Modality and Evidentiality in Spanish (semi-)auxiliaries. A functional-pragmatic and cognitive-linguistic account. Ph. D. Diss., Department of Linguistics, K.U.Leuven. 2005a Reference point and subjectification in grounding predications: the case of the Spanish modals. Annual Review of Cognitive Linguistics 3: 56–77. 2005b Subjectivity in esperar-based constructions. In Reviewing Linguistic Thought: Perspectives into the 21st Century. Sophia Marmaridou, Kiki Nikiforidou and Eleni Antonopoulou (eds.), 77–96. Berlin/New York: Mouton de Gruyter. 2005c Agentivity and subjectivity in Spanish prometer and amenazar. A study of diatopical and constructional variation. Revista Internacional de Lingüística Iberoamericana. 3 (5): 171–196. 2005d A paradigmatic view of Spanish amenazar ‘to threaten’ and prometer ‘to promise’. Folia Linguistica 39 (3/4): 385–316. 2006 Conceptual and constructional considerations on the subjectivity of English and Spanish modals. In Subjectification. Various Paths to Subjectivity, Angeliki Athanasiadou, Costas Canakis and Bert Cornillie (eds.), 177–205. Berlin/New York: Mouton de Gruyter. Corominas, Joan 1998 Breve diccionario etimológico de la lengua castellana. Madrid: Gredos. Corominas, Joan and José Antonio Pascual. 1984 Diccionario crítico etimológico castellano e hispánico. Madrid: Gredos. Croft, William 1999 Some contributions of typology to Cognitive Linguistics, and vice versa. In Cognitive Linguistics: Foundations, Scope, and Methodology, Theo Janssen and Gisela Redeker (eds.), 61–93. Berlin/New York: Mouton de Gruyter. 2001 Radical Construction Grammar. Syntactic Theory in Typological Perspective. Oxford: Oxford University Press. Cuyckens, Hubert and Britta E. Zawada (eds.) 2001 Polysemy in Cognitive Linguistics: Selected papers from the International Cognitive Linguistics Conference, Amsterdam, 1997. Amsterdam/Philadelphia: Benjamins. Cuyckens, Hubert, Thomas Berg, René Dirven and Klaus-Uwe Panther (eds.) 2003 Motivation in Language: Studies in honor of Günter Radden. Amsterdam/Philadelphia: Benjamins.
290
References
Davidse, Kristin 1994 Fact projection. In Perspectives on English. Studies in Honour of Professor Emma Vorlat, Keith P. Carlon, Kristin Davidse and Brygida Rudzka-Ostyn (eds.), 259–286. Leuven/Paris: Peeters. Davies, Mark 2003 Corpus del español. http://www.corpusdelespanol.org/ De Haan, Ferdinand 1999 Evidentiality and Epistemic Modality: Setting Boundaries. Southwest Journal of Linguistics 18: 83–101. De Kock, Josse 1998 Norma, Libertad y Probabilidad. Ocho Soluciones del Español. Salamanca: Ediciones Universidad de Salamanca. De Kock, Josse (ed.) 2001 Lingüística con Corpus: Catorce Aplicaciones sobre el Español. Salamanca: Ediciones Universidad de Salamanca. Declerck, Renaat 1991 A Comprehensive Descriptive Grammar of English. Tokyo: Kaitakusha. DeLancey, Scott 1986 Evidentality and volitionality in Tibetan. Evidentiality: The Linguistic Coding of Epistemology, Wallace L. Chafe and Johanna Nichols (eds.), 203–213. Norwood NJ: Ablex. 2001 The mirative and evidentiality. Journal of Pragmatics 33 (3): 369– 382. Delbecque, Nicole 1996 Towards a cognitive account of the use of the prepositions por and para in Spanish. In Cognitive Linguistics in the Redwoods: the Expansion of a New Paradigm in Linguistics, Eugene H. Casad (ed.), 249–318. Berlin/New York: Mouton de Gruyter. 2000 Cognitive constraints on complement clause cliticization in Spanish. In Complementation. Cognitive and Functional Perspectives, Kaoru Horie (ed.), 149–196. Amsterdam/Philadelphia: Benjamins. 2003 La variable expresión del agente en las construcciones pasivas. Nueva Revista de Filología Hispánica 2: 373–416. Delbecque, Nicole and Béatrice Lamiroy 1999 La subordinación sustantiva. In Gramática descriptiva de la lengua española, Ignacio Bosque and Violeta Demonte (eds.), 1965–2083. Madrid: RAE/Espasa Calpe. Delport, Marie-France 2004 Deux Verbes Espagnols: haber et tener. Étude Lexico-Syntaxique. Perspective Historique et Comparative, Paris: Éditions Hispaniques. Dendale, Patrick 1994 Devoir: épistémique, marqueur modal ou évidentiel? Langue française 102: 24–40.
References 1999
291
‘Devoir’ au conditionnel: valeur évidentio-modale et origine du conditionnel. Cahiers Chronos 4. 7–28. 2000 Devoir épistémique à l’indicatif et au conditionnel: inférence ou prédiction? In Actes du XIIe Congrès International de Linguistique et de Philologie Romanes, Bruxelles, 23–29 juillet 1998, A. Englebert et al. (eds.), 159–169 (vol. 7.). Tübingen: Max Niemeyer. 2001 Le futur conjectural versus devoir épistémique: différences de valeur et restrictions d’emploi. Le Français Moderne 69 (1): 1–20. Dendale, Patrick and Walter De Mulder 1996 Déduction ou abduction: le cas de devoir inférentiel. In L’Énonciation Médiatisée, Zlatka Guentchéva (ed.), 305–318. Leuven/Paris: Peeters. Dendale Patrick and Johan van der Auwera (eds.) 2001 Les Verbes Modaux. Amsterdam: Rodopi (Cahiers Chronos 8). Dendale Patrick and Liliane Tasmowski. 2001 Introduction: evidentiality and related notions. Journal of pragmatics 33 (3): 339–348. Di Tullio, Ángela 1996 Verbos psicológicos en español. Signo y Seña. Revista del Instituto de Lingüística 5: 221–238. 1998 Complementos no flexivos de verbos de percepción física en español. Verba 25: 197–221. Diewald, Gabriele 2000 Scheinen als Faktizitätsmarker. In Wortschatz und Orthographie in Geschichte und Gegenwart. Festschrift für Horst Haider Munske zum 65. Geburtstag am 5. Mai 2000. Mechthild Habermann, Peter O. Müller and Bernd Naumann (eds.), 333–355. Tübingen: Niemeyer. 2001a A basic semantic template for lexical and grammaticalized uses of the German modals. Belgian Journal of Linguistics 14: 23–41. 2001b Scheinen–Probleme: Analogie, Konstruktionsmischung und die Sogwirkung aktiver Grammatikalisierungskanäle. Synchronie und Diachronie der Modalverben. Linguistische Berichte 9: 87–110. 2004 Faktizität und Evidentialität: Semantische Differenzierungen bei den Modal- und Modalitätsverben im Deutschen. In Tempus / Temporalität und Modus / Modalität im Deutschen - auch in konstrastiver Perspektive. Internationales Kolloquium. 8.-9. September 2000, Oddleif Leirbukt and Paul Valentin (eds.), 231–258. Bergen. Tübingen: Stauffenberg (Eurogermanistik). Dik, Simon C. 1978 Functional Grammar. Dodrecht: Foris. 1989 The Theory of Functional Grammar. Part 1: The Structure of the Clause. Dordrecht/Providence: Foris.
292
References
Dirven, René 1985 Metaphor as a basic means for extending the lexicon. In The Ubiquity of Metaphor, Wolf Paprotté and René Dirven (eds.), 85–119. Amsterdam/Philadelphia: Benjamins. 1989 A cognitive perspective on complementation. In Sentential Complementation and the Lexicon: Studies in Honour of Wim de Geest, Jaspers, Dany, Willem Klooster, Yvan Putseys and Pieter A.M. Seuren (eds.), 113–139. Dordrecht: Foris. Dirven, René and Marjolijn Verspoor (eds.) 1998 Cognitive Exploration of Language and Linguistics. Amsterdam/Philadelphia: Benjamins. Divjak, Dagmar 2004 Degrees of Verb integration. Conceptualizing and Categorizing Events in Russian. Doctoral dissertation. K.U. Leuven. Elvira, Javier 2002 Modalidad e inferencia pragmática. Paper presented at the conference on Les modalités et leur expression dans les documents espagnols du Moyen Age. Université de Paris X/Nanterre/Instituto Cervantes de París. 15-16 November 2002. 2003 Sobre el desarrollo del valor epistémico del verbo poder. Paper presented at the VI Congreso Internacional de Historia de la Lengua Española. Madrid. October 2003. Facchinetti, Roberta, Manfred, Krug and Frank Palmer 2003 Modality in Contemporary English. Berlin/New York: Mouton de Gruyter. Faller, Martina 2002 Semantics and Pragmatics of Evidentials in Cuzco Quechua. Ph. D. Diss., Department of Linguistics, Stanford University. Fernández de Castro, Félix 1999 Las perífrasis verbales en el español actual. Madrid: Gredos. Fernández Leborans, María Jesús 1999 La predicación. Las oraciones copulativas. In Gramática descriptiva de la lengua española, Ignacio Bosque and Violeta Demonte (eds.), 2357–2460. Madrid: RAE/Espasa Calpe. Fernández Leborans, María Jesús and María del Carmen Díaz Bautista 1990 Sobre la sintaxis del verbo español “parecer”. Boletín de la Real Academia Española. 353–420. Fisher, Cynthia, Henry Gleitman and Lila R. Gleitman 1991 On the semantic content of subcategorization frames. Cognitive Psychology 2. 231–254. Fitneva, Stanka 2001 Epistemic markers and reliability judgments. Evidence form Bulgarian. Journal of Pragmatics 33: 401–420.
References
293
Fleischman, Suzanne 1982 The Future in Thought and Language. Cambridge: Cambridge University Press. 1989 Temporal Distance: A basic linguistic metaphor. Studies in Language 13: 1–50. Floyd, Rick 1996a The radial structure of the Wanka reportative. In Cognitive Linguistics in the Redwoods: the Expansion of a New Paradigm in Linguistics, Eugene H. Casad (ed.), 895–941. Berlin/New York: Mouton de Gruyter. 1996b Experience, certainty and control, and the direct evidential in Wanka Quechua questions. Functions of Language 3: 69–93. Gaatone, David 1976 Il doit y avoir – *il faut y avoir. Revue Romane 9.2. 245–267. 1995 Syntaxe et sémantique: le cas des verbes transparents. Le Langage et le texte. Hommage à Alexandre Lorian. The Hebrew University of Jerusalem. 55–71. García, Erica C. 1967 Auxiliaries and the criterion of simplicity. Language 43 (4): 853– 870. Giacalone Ramat, Anna and Paul J. Hopper 1998 The limits of grammaticalization. Amsterdam/Philadelphia: Benjamins. Givon, Talmy 1971 Historical syntax and synchronic morphology: An archaeologist’s field trip. Chicago Linguistic Society 7. 394–415. 1979 On understanding grammar. New York/London: Academic Press. 1982 Evidentiality and Epistemic Space. Studies in Language 6. 23–49. 1993 English grammar: a function-based introduction. Amsterdam/ Philadelphia: Benjamins. (2 vols). Goethals, Patrick 2002 Las Conjunciones causales explicativas españolas como, ya que, pues y porque: un estudio semiotico-linguistico. Leuven/Paris: Peeters (Orbis Supplementa 21). González Fernández, María Jesús and Ricardo Maldonado. 2006 Syntactic Determinants of Pragmatic Markers of “Closure”. In Topics in subjectification and modalization, Bert Cornillie and Nicole Delbecque (eds.), 19–44. Amsterdam/Philadelphia: Benjamins. González Vázquez, Mercedes 1999 Evidencialidad y Modalidad epistémica. Análisis tipológico y semántico de su interacción. MA-Thesis. Universidade de Vigo 2000 De lo Posible a lo Necesario: Semántica de la Modalidad. Ph. D. Diss., Department of Linguistics. Universidade de Vigo. Gómez Torrego, Leonardo 1988 Perífrasis verbales. Madrid: Arco Libros.
294
References
1999
Los verbos auxiliares. Las perífrasis de infinitivo. In Gramática descriptiva de la lengua española, Ignacio Bosque and Violeta Demonte (eds.), 3323–3390. Madrid: RAE/Espasa Calpe. Goossens, Louis 1982 On the development of the modals and the epistemic function in English. In Papers from the Fifth International Conference on Historical Linguistics, Anders Ahlqvist (ed.), 74–84. Amsterdam/Philadelphia: Benjamins. 1985 Modality and the modals: a problem for Functional Grammar. In Predicates and terms in Functional Grammar, Machteld Bolkestein, Casper de Groot and J.L. Mackenzie (eds.), 203–217. Dordrecht: Foris. 1987 Modal shifts and predication types. In Ins and outs of predication, Johan van der auwera and Louis Goossens (eds.), 21–37. Dordrecht: Foris. 1999 Metonymic bridges in modal shifts. In Metonymy in Language and Thought, Klaus-Uwe Panther and Günter Radden (eds.), 193–210. Amsterdam/Philadelphia: Benjamins. 2000 Patterns of meaning extension, “parallel chaining”, subjectification, and modal shifts. In Metaphor and metonymy at the crossroads. A cognitive perspective, Antonio Barcelona (ed.), 149–169. Berlin/New York: Mouton de Gruyter. Grice, Paul H. 1975 Logic and Conversation. In Syntax and Semantics 3: Speech Acts, P. Cole and J. Morgan (eds.), 41–58. New York: Academic Press. 1989 Studies in the Way of Words. Cambridge MA: Harvard University Press. Grondelaers, Stefan, Dirk Speelman and Dirk Geeraerts 2002 Regressing on er. Statistical analysis of texts and language variation. In 6ièmes Journées internationales d'Analyse statistique des Données Textuelles – 6th International Conference on Textual Data Statistical Analysis, A. Morin and P. Sébillot (eds.), 335–346. Rennes: Institut National de Recherche en Informatique. Guentchéva, Zlatka 1994 Manifestations de la catégorie du médiatif dans les temps du français. Langue française 102: 8–23. Guentchéva, Zlatka (ed.) 1996 L’Énonciation Médiatisée. Leuven /Paris: Peeters. Guo, Jiansheng 1995 The Interactional Basis of the Mandarin Modal néng ‘can’. In Modality in Grammar and Discourse, Joan Bybee and Suzanne Fleischman (eds.), 205–238. Amsterdam/Philadelphia: Benjamins. Gutiérrez Ordóñez, Salvador 1986 Variaciones sobre la Atribución. León: Universidad de León. 1997 Temas, Remas, Focos, Tópicos y Comentarios. Madrid: Arco Libros.
References
295
2002a De Pragmática y Semántica. Madrid: Arco Libros. 2002b Forma y Sentido en Sintaxis. Madrid: Arco Libros. Halliday, Michael A. K. 1970 Language structure and language function. In New horizons in linguistics, John Lyons (ed.), 140–165. Aylesbury, U.K.: Penguin. 1994 An introduction to functional grammar. 2d ed. London/Baltimore: Edward Arnold. Harder, Peter 1996 Functional Semantics. A Theory of Meaning, Structure and Tense in English. Berlin/New York: Mouton de Gruyter. Hassler, Gerda 2002 Evidentiality and reported speech in Romance languages. In Reported Discourse. A meeting ground for different linguistic domains. Tom Güldemann and Manfred von Roncador (eds.), 143–172. Amsterdam/Philadelphia: Benjamins. 2003 Uso evidencial de adverbios modales. Paper presented at the 6th Congreso Internacional de Lingüística Hispánica at Leipzig. September 2003. Haverkate, Henk and Kees Hengeveld (eds.) 1993 Aproximaciones Pragmalingüísticas al Español. Amsterdam: Rodopi. Heine, Bernd 1993 Auxiliaries: Cognitive Forces and Grammaticalization. Oxford: Oxford University Press. 1997 Cognitive foundations of grammar. Oxford/New York: Oxford University Press. Heine, Bernd, Ulrike Claudi, and Friederike Hünnemeyer 1991 Grammaticalization: A conceptual Framework. Chicago: University of Chicago Press. Heine, Bernd and Tania Kuteva (eds.) 2002 World Lexicon of Grammaticalization. Cambrigde: Cambridge University Press. Heine, Bernd and Mechtild Reh 1984 Grammaticalization and Reanalysis in African languages. Hamburg: Helmut Buske Verlag Heine, Bernd and Hiroyuki Miyashita 2007 The structure of a functional category: German drohen. Language Sciences. Hengeveld, Kees 1988 Illocution, mood, and modality in a Functional Grammar of Spanish. Journal of Semantics 6.3/4. 227–269. Hernanz, María Lluïsa 1982 El infinitivo en español. Bellaterra: Ediciones de la Universidad Autónoma de Barcelona.
296
References
1999
El infinitivo. In Gramática descriptiva de la lengua española, Ignacio Bosque and Violeta Demonte (eds.), 2195–2356. Madrid: RAE/Espasa Calpe. Heyvaert, Liesbet 2003 A Cognitive-functional Approach to Nominalization in English. Berlin/New York: Mouton de Gruyter. Hopper, Paul J. 1991 On some principles of grammaticization. In Approaches to Grammaticalization Vol I, Elizabeth Traugott and Bernd Heine (eds.), 17– 35. Amsterdam/ Philadelphia: Benjamins. Hopper, Paul J. and Elizabeth Closs Traugott 1993 Grammaticalization. Cambridge: Cambridge University Press. Ifantidou, Elly 2001 Evidentials and relevance. Amsterdam/Philadelphia: Benjamins Keniston, Hayward 1937 Spanish Syntax List. New York: Henry Holt Company. Kiparsky, Paul and Carol Kiparsky 1970 Fact. In Progress in Linguistics, Manfred Bierwisch and K.E. Heidolph (eds.), 143–173. The Hague: Mouton. Kronning, Hans 1996 Modalité, Cognition et Polysémie: Sémantique du Verbe Modal devoir. Uppsala: Acta Universitatis Upsaliensis. (Studia Romanica Upsaliensia 54) 2001 Nécessité et hypothèse: devoir non déontique au conditionnel. In Le conditionnel en français, Patrick Dendale and Liliane Tasmowski (eds.), 251–276. Université de Metz. Kuteva, Tania 2004 Auxiliation: An Enquiry into the Nature of Grammaticalization. 2d ed. Oxford: Oxford University Press. Lamiroy, Béatrice 1983 Les Verbes de Mouvement en Français et en Espagnol: Étude Comparée de leurs Infinitifs (Linguisticae Investigationis supplementa 11). Amsterdam/Philadelphia: Benjamins. 1987 The complementation of aspectual verbs in French. Language 63: 278–298. 1991 Léxico y Gramática del Español; Estructuras Verbales de Espacio y Tiempo. Barcelona: Anthropos. 1994 Les syntagmes nominaux et la question de l’auxiliarité. Langages 115 : 64–76. 1995 La “transparence” des auxiliaires. In Tendances Récentes en Linguistiques Française et Générale, Hava Bat-Zeev Shyldkrot and Lucien Kupferman (eds.), 277–287. Amsterdam/Philadelphia: Benjamins. 1998 Prédication et auxiliaires. In Prédication, Assertion, Information, M. Forsgren, K. Jonasson and H. Kronning (eds.), 285-299. Uppsala: Acta Universitatis Uppsaliensis.
References 1999
297
Auxiliaires, langues romanes et grammaticalisation. Langages 135: 63–75. 2003 Les notions linguistiques de figement et de contrainte. Linguisticae Investigationis 26: 53–66. Langacker, Ronald W. 1985 Observations and speculations on subjectivity. In Iconicity in syntax, John Haiman (ed.), 109–150. Amsterdam/Philadelphia: Benjamins. 1987 Foundations of Cognitive Grammar. Volume 1: Theoretical prerequisites. Stanford: Stanford University Press. 1990 Subjectification. Cognitive Linguistics 1: 5–38. 1991 Foundations of Cognitive Grammar. Volume 2: Descriptive application. Stanford: Stanford University Press. 1991 Concept, Image and Symbol. Berlin /New York: Mouton de Gruyter. 1993 Reference-point constructions. Cognitive Linguistics 4. 1–38. 1995 Raising and transparency. Language 71: 1–62. 1997 Consciousness, construal, and subjectivity. In Language Structure, Discourse and the Access to Consciousness, Maxim I. Stamenov (ed.), 49–75. Amsterdam/Philadelphia: Benjamins. 1998 On subjectification and grammaticization. In Discourse and Cognition. Bridging the Gap, Jean-Pierre Koenig (ed.), 71–89. Stanford: CSLI Publications. 2000a Conceptual grouping and constituency. In Grammar and Conceptualization, Ronald W. Langacker, 147–170. Berlin/New York: Mouton de Gruyter. 2000b Subjectification and grammaticization. In Grammar and Conceptualization, Ronald W. Langacker, 297–315. Berlin/New York: Mouton de Gruyter. 2000c Raising and Transparency. In Grammar and Conceptualization, Ronald W. Langacker, 317–357. Berlin/New York: Mouton de Gruyter. 2002a Deixis and subjectivity. In Grounding: The Epistemic Footing of Deixis and Reference, Frank Brisard (ed.), 1–28. Berlin/New York: Mouton de Gruyter. 2002b Estructura de la cláusula en la gramática cognoscitiva. In Estudios Cognoscitivos del Español. Revista Española de Lingüística Aplicada, Ricardo Maldonado (ed.), 19–66. Castellón/Querétaro: AELA/UAQ. 2003 Extreme subjectification: English tense and modals. In Motivation in Language: Studies in honor of Günter Radden, Hubert Cuyckens, Thomas Berg, René Dirven and Klaus-Uwe Panther (eds.), 3–26. Amsterdam/Philadelphia: Benjamins. 2006 Subjectification, Grammaticization, and Conceptual Archetypes. In Subjectification: Various Paths to Subjectivity, Angeliki Athanasiadou, Costas Canakis and Bert Cornillie (eds.), 17–40. Berlin/New York: Mouton de Gruyter.
298
References
Lazard, Gilbert 1956 Caractères distinctifs de la langue tadjik. Bulletin de la société de linguistique de Paris. 52 (1): 117–186. 1999 Mirativity, evidentiality, mediativity, or other? Linguistic Typology 3: 91–109. 2001 On the grammaticalization of evidentiality. Journal of Pragmatics 33: 359–367. Lehmann, Christian 1982 Thoughts on Grammaticalization. A programmatic Sketch. Vol. I (Arbeiten des Kölner Universalien-Projekts 48). Köln: Universität zu Köln. Institut für Sprachwissenschaft. 2002 New reflections on grammaticalization and lexicalization. In New Reflections on Grammaticalization, Ilse Wischer and Gabriele Diewald (eds.), 1–18. Amsterdam/Philadelphia: Benjamins. Levinson, Stephen C. 1983 Pragmatics. Cambridge: Cambridge University Press. 1997 From Outer Space to Inner Space: Linguistic Categories and Nonlinguistic Thinking. In Language and Conceptualization, Jan Nuyts and Eric Pederson (eds.), 13–45. Cambridge: Cambridge University Press Lozowski, Przemyslaw 1999 Panchrony, or linguistics without synchrony. In Cognitive Perspectives on Language, Barbara Lewandowska-Tomaszczyk (ed.), 23–35. Frankfurt am Main: Peter Lang. 2002 Deconstructing the world-view: from actuality to virtuality in cognitive linguistics. In Cognitive Linguistics Today, Barbara Lewandowska-Tomaszczyk and K. Turewicz (eds.), 65–79. Frankfurt am Main: Peter Lang. Lunn, Patricia 1989 Spanish mood and the prototype of assertability. Linguistics 27. 687– 702. 1995 The Evaluative Function of the Spanish Subjunctive. In Modality in Grammar and Discourse; Joan Bybee and Suzanne Fleischman (eds.), 419–449. Amsterdam/Philadelphia: Benjamins. Lyons, John 1977 Semantics. Cambridge: Cambridge University Press. 1981 Language and Linguistics: An Introduction. Cambridge: Cambridge University. 1982 Deixis and subjectivity: Loquor, ergo sum? In Speech, Place, and Action: Studies in Deixis and Related Topics, Robert J. Jarvella and Wolfgang Klein (eds.), 101–124. New York: Wiley. 1994 Subjecthood and subjectivity. In Subjecthood and Subjectivity: The Status of the Subject in Linguistic Theory, Marina Yaguello (ed.), 9– 17. Paris: Ophrys.
References 1995
299
Linguistic Semantics: An Introduction. Cambridge: Cambridge University Press. Maldonado, Ricardo 1988 Energetic Reflexives in Spanish. Berkeley Linguistics Society 14: 153–165. 1999 A Media Voz: Problemas Conceptuales del Clítico se en Español . México: Instituto de Investigaciones Filológicas UNAM. 2002 Objective and subjective datives. Cognitive Linguistics 13 (1): 1–65. Marín-Arrese, Juana, Laura Hidalgo Downing and Silvia Molina Plaza 2002 Evidentiality and modality in English and Spanish: connections and interactions. In English Modality in Perspective: Genre Analysis and Contrastive Studies, Roberta Facchinetti (ed.), 135–146. Verona: Università di Verona. Mori, Olga 1974 Sobre las construcciones españolas ‘tener que + infinitivo’ y ‘haber que + infinitivo’ y sus equivalentes en inglés. Filología moderna 15: 155–181. Mortelmans, Tanja 2000 Konjunktiv II and epistemic modals in German: a division of labour. In Constructions in Cognitive Linguistics, Ad Foolen and Frederike Van der Leek, 191–215. Amsterdam/Philadelphia: Benjamins. 2001 On the 'Evidential' Nature of the 'Epistemic' Use of the German Modals müssen and sollen. Belgian Journal of Linguistics. 131–149. 2002 “Wieso sollte ich dich küssen, du hasslicher Mensch!” A study of the German modals sollen and müssen as `grounding predications’ in interrogatives. In The Epistemic Footing of Deixis and Reference, Frank Brisard (ed.), 391–432. Berlin/New York: Mouton de Gruyter. Mushin, Ilana 2001 Evidentiality and Epistemological Stance: Narrative Retelling. Amsterdam/Philadelphia: Benjamins. Narrog, Heiko 2005a Modality, mood and change of modal meanings: a new perspective. Cognitive Linguistics 16 (4): 677–731. 2005b On defining modality again. Language sciences 27 (2): 165–192. Noël, Dirk 2002 Believe-type matrix verbs and their complements. Corpus-based investigations of their function in discourse. A collection of articles. Ph. D. Diss., Department of Linguistics. Universiteit Gent. Nuyts, Jan 1992 Aspects of a Cognitive-pragmatic Theory of Language: On Cognition, Functionalism, and Grammar. Amsterdam/Philadelphia: Benjamins. 2000 Tensions between discourse structure and conceptual semantics: The syntax of epistemic modal expressions. Studies in Language 24: 103–135.
300
References
2001a
Subjectivity as an evidential dimension in epistemic modal expressions. Journal of Pragmatics 33: 383–400. 2001b Epistemic Modality, Language, and Conceptualization: A Cognitivepragmatic perspective. Amsterdam/Philadelphia: Benjamins. 2002 Grounding and the system of epistemic expressions in Dutch: a cognitive-functional view. In The Epistemic Footing of Deixis and Reference, Frank Brisard (ed.), 433–466. Berlin/New York: Mouton de Gruyter. 2004 The modal confusion. On terminology and the concepts behind it. In Modality: Studies in Form and Function, Alex Klinge and Hendruk Müller (eds.), 5–38. London: Equinox. 2005 Modality: Overview and linguistic issues. In The Expression of Modality. William Frawley ed., 1–26. Berlin/New York: Mouton de Gruyter. Nuyts, Jan and Eric Pederson (eds.) 1997 Language and Conceptualization. Cambridge: Cambridge University Press. Nuyts, Jan and Wietse Vonk 1999 Epistemic modality and focus in Dutch. Linguistics 37: 699–737. Olbertz, Hella 1998 Verbal Periphrases in a Functional Grammar of Spanish, Berlin/New York: Mouton de Gruyter. Palmer, Frank R. 1979 Why auxiliaries are not main verbs. Lingua 47. 1–25. 1990 Modality and the English modals. Second edition. London: Longman. 2001 Mood and modality. 2d ed. Cambridge: Cambridge University Press. Perkins, Michael R. 1982 The Core Meanings of the English Modals. Journal of Linguistics 18: 245–273. 1983. Modal expressions in English. London: Pinter. Pelyvás, Péter 1996 Subjectivity in English: Generative Grammar versus the Cognitive Theory of Epistemic Grounding. Frankfurt am Main: Lang. 2000 Metaphorical extension of may and must into the epistemic domain. In Metaphor and metonymy at the crossroads. A cognitive perspective, Antonio Barcelona (ed.), 233–250. Berlin/New York: Mouton de Gruyter. 2001a On the development of the category modal: a cognitive view. How changes in image-schematic structure led to the emergence of the grounding predication. In Wort und (Kon)text, Piroska Kocsány and Molnár Anna (eds.), 103–130. Frankfurt am Main: Peter Lang. 2001b The development of the grounding predication. Epistemic modals and cognitive predicates. In Pragmatics and the Flexibility of Word Meaning, E. Németh T. and K. Bibok (eds.), 151–174. Amsterdam: Elsevier.
References 2006
301
Subjectification in (expressions of) epistemic modality and the development of the grounding predication. In Subjectification: Various Paths to Subjectivity, Angeliki Athanasiadou, Costas Canakis and Bert Cornillie (eds.), 121–150. Berlin/New York: Mouton de Gruyter. Plungian, Vladimir A. 2001 The place of evidentiality within the universal grammatical space. Journal of Pragmatics 33: 349–357. Porroche Ballesteros, Margarita 1990 Aspectos de atribución en español. Zaragoza: Pórtico. Radden, Günter 1999 Modalverben in der Kognitiven Linguistik. In Grammatik und Mentale Prozesse. Angelika Redder and Jochen Rehbein (eds.), 261-294, Tübingen: Stauffenbrug Verlag. Ramat, Paolo 1992 Thoughts on degrammaticalization. Linguistics 30:549–560. Reis, Marga 2004 Modals, So-Called Semi-Modals, and Grammaticalization in German. Paper delivered at Berkeley Germanic Linguistics Roundtable. 2-3 April 2004. Riddle, E. 1975 Some pragmatic conditions on complementizer choice. Chicago Linguistic Society 11. 467–474. Rivero, María-Luisa 1971 Mood and presupposition in Spanish. Foundations of Language 7. 305–336. 1980 On left dislocation and topicalization in Spanish. Linguistic Inquiry 11. 363–395. Rodríguez Espiñeira, María José 2000 Percepción directa e indirecta en español. Diferencias semánticas y formales. Verba 27. 33–85. 2001 Alternancias de esquema sintáctico con predicados de valoración intelectual. In Léxico y Gramática. Aleixandre Veiga, M. González and M. Souto (eds.), 313–326. Lugo: Tris Tram. Rojo, Guillermo 1974 Perífrasis verbales en el gallego actual. Verba. Annex 2. Santiago de Compostela: Universidade de Santiago de Compostela. Rooryck, Johan 1997 On the interaction between Raising and Focus in sentential complementation. Studia Linguistica 50. 1–49. 2000 Configurations of Sentential Complementation: Perspectives from Romance languages. Routledge: London. 2001a Evidentiality (Part I). Glot International. 5.4. 2001b Evidentiality (Part II). Glot International. 5.5.
302
References
Rosch, Eleanor 1973 Natural categories. Cognitive psychology 4: 328–350. 1975a Cognitive representations of semantic categories. Journal of Experimental Psychology 104: 192–233. 1975b Cognitive Reference Points. Cognitive Psychology 7: 532–547. Rudzka-Ostyn, Brygida (ed.) 1988 Topics in Cognitive Linguistics. Amsterdam/Philadelphia: Benjamins. Ruwet, Nicolas 1972 Théorie syntaxique et syntaxe du français. Paris: Éditions du Seuil. 1983 Montée et contrôle. Une question à revoir. Revue Romane 24: 17–37. 1991 Syntax and Human Experience. Chicago: University of Chicago Press. Sanders, José and Wilbert Spooren 1996 Subjectivity and certainty in epistemic modality: A study of Dutch epistemic modifiers. Cognitive Linguistics 7: 241–64. Schiffrin, Deborah 1990 The principle of intersubjectivity in communication and conversation. Semiotica 80: 121–151. Schwenter, Scott A. 1994 ‘Hot News’ and the Grammaticalization of Perfects. Linguistics 32: 995–1028. 1999 Evidentiality in Spanish morphosyntax: A reanalysis of (de)queísmo. In Estudios de Variación Sintáctica, María José Serrano (ed.), 85– 87. 1999. Madrid: Iberoamericana. Searle, John. R. 1969 Speech Acts. An Essay in the Philosophy of Language. Cambridge: Cambridge University Press. 1983 Intentionality. An Essay in the Philosophy of Mind. Cambridge: Cambridge University Press. Seuren, Pieter A.M. 1969 Operators and Nucleus, a Contribution to the Theory of Grammar. Cambridge: Cambridge University Press. 2002a A View of Language. Oxford: Oxford University Press. 2002b How the cognitive revolution passed linguistics by. In Language and evolution. Frank Brisard and Tanja Mortelmans (eds. Antwerp Papers in Linguistics 101. 2003 Verb clusters and branching directionality in German and Dutch. In Verb Constructions in German and Dutch, Pieter A.M. Seuren and Gerard Kempen (eds.), 247–296. Amsterdam/Philadelphia: Benjamins. Silva-Corvalán, Carmen 1995 Contextual conditions for the interpretation of ‘poder’ and ‘deber’ in Spanish. In Modality in Grammar and Discourse, John Bybee and
References
303
Suzanne Fleishmann (eds), 67–105. Amsterdam/Philadelphia: Benjamins. Speelman, Dirk 1997 Abundantia Verborum. A computer tool for carrying out corpusbased linguistic studies. Doctoral dissertation. K.U.Leuven. Sperber, Dan and Deirdre Wilson 1986 Relevance: Communication and Cognition. Oxford: Blackwell. Squartini, Mario 1998 Verbal Periphrases in Romance. Aspect, Actionality and Grammaticalization. Berlin/New York: Mouton de Gruyter. 2001 The internal structure of evidentiality in Romance. Studies in Language 25 (2): 297–334. 2004 Disentangling evidentiality and epistemic modality in romance. Lingua 114 (7): 873–889. Lexical vs. grammatical evidentiality in French and Italian. Linguis2006 tics. Stein, Dieter and Susan Wright (eds.) 1995 Subjectivity and Subjectivisation in Language. Cambridge: Cambridge University Press. Subirats-Rüggeberg, Carlos 1987 Sentential complementation in Spanish: a lexico-grammatical study of three classes of verbs. Amsterdam/Philadelphia: Benjamins. Sweetser, Eve E. 1982 Root and epistemic modals: causality in two worlds. Berkeley Linguistics Society 8: 484–507. Sweetser, Eve E. 1990 From Etymology to Pragmatics. Metaphoric and Cultural Aspects of Semantic Structure. Cambridge: Cambridge University Press. Talmy, Leonard 1985 Lexicalization patterns: semantic structure in lexical forms. In Language Typology and Syntactic Description: Grammatical Categories and the Lexicon, Timothy Shopen (ed.), 57–149. Volume 3. Cambridge: Cambridge University Press. 1988 The Relation of Grammar to Cognition. In Topics in Cognitive Linguistics, Brygida Rudzka-Ostyn (ed.), 165–205. Amsterdam/Philadelphia: Benjamins. 2000 Toward a Cognitive Semantics. Cambrigde MA: MIT Press. (2 vols). Tasmowski, Liliane 1989 A look at sembler from different angles. In Sentential Complementation and the Lexicon: Studies in Honour of Wim de Geest, Dany Jaspers et al. (eds.), 403–423. Dordrecht: Foris. Tasmowski, Liliane and Patrick Dendale 1994 Pouvoir: un Marqueur d'évidentialité. Langue Française 102 : 41– 55.
304
References
1998
Must/Will and doit/futur simple as epistemic modal markers. Semantic value and restrictions of use. In English as a Human Language. To honour Louis Goossens, Johan van der Auwera, Frank Durieux and Ludo Lejeune (eds.), 325–336. München: Lincom Europa.
Taylor, John 1989 Linguistic Categorization: Prototypes in Linguistic Theory. Oxford: Oxford University Press. 1996 Possessives in English: an Exploration in Cognitive Grammar. Oxford: Clarendon Press. Traugott, Elizabeth Closs 1982 From propositional to textual and expressive meanings: Some semantic-pragmatic aspects of grammaticalization. In Perspectives on Historical Linguistics, Winfred P. Lehmann and Yakov Malkiel (eds.), 245–271. Amsterdam/Philadelphia: Benjamins. 1988 Pragmatic strengthening and grammaticalization. Berkeley Linguistics Society 14: 406–415. 1989 On the rise of epistemic meanings in English: an example of subjectification in semantic change. Language 65: 31–55. 1993 The conflict promises to erupt into war. Berkeley Linguistics Society 17. 348–358. 1995 Subjectification in grammaticalisation. In Subjectivity and Subjectivisation in Language, Dieter Stein and Susan Wright (eds.), 31–54. Cambridge: Cambridge University Press. 1997 Subjectification and the development of epistemic Meaning: The Case of Promise and Threaten. In Modality in Germanic Languages, Toril Swan and Olaf Jansen Westvik (eds.), 185–210. Berlin – New York: Mouton de Gruyter. 1999 The rhetoric of counter-expectation in semantic change. In Historical semantics and cognition, Andreas Blank and Peter Koch (eds.), 177– 196. Berlin/New York: Mouton de Gruyter. 2003 From subjectification to intersubjectification. In Motives for Language Change, Raymond Hickey (ed.), 124–139. Cambridge: Cambridge University Press. Traugott, Elizabeth and Bernd Heine (eds.) 1991 Approaches to Grammaticalization, Vol I&II. Amsterdam/ Philadelphia: Benjamins. Traugott, Elizabeth and Ekkehard Koenig 1991 The semantics-pragmatics of grammaticalization revisited. In Approaches to Grammaticalization, Vol 1, Elizabeth Traugott and Bernd Heine (eds.), 189–218. Amsterdam/Philadelphia: Benjamins. Traugott, Elizabeth Clossand Richard Dasher 2002 Regularity in Semantic Change. Cambridge: Cambridge University Press.
References
305
Usoniene, Aurelia 1999 Perception verbs revisited. Lund: Department of Linguistics. Working Papers. 211–225. 2001 On the Modality of the English Verbs of Seeming. Belgian Journal of Linguistics 14: 183–204. Van Gelderen, V. 1999 Direct and indirect discourse: an alternative view on the role of complementizers. Project Description Paper. Leiden University. Vázquez Laslop, María Eugenia 2001a La Arquitectura Lingüística del Compromiso: las Oraciones de deber ser. México: Colegio de México. 2001b Epistemic prometer and full deontic modal verbs. Belgian Journal of Linguistics. 207–241. Vendler, Zeno 1967 Linguistics in Philosophy. Ithaca/New York: Cornell University Press. Verhagen, Arie 1995 Subjectification, syntax, and communication. In Subjectivity and Subjectivisation in Language, Dieter Stein and Susan Wright (eds.), 103–128. Cambridge: Cambridge University Press. 1996 Sequential conceptualization and linear order. In Cognitive Linguistics in the Redwoods: the Expansion of a New Paradigm in Linguistics, Eugene H. Casad (ed.), 793–817. Berlin/New York: Mouton de Gruyter. 2000 'The girl that promised to become something': An exploration into diachronic subjectification in Dutch. In The Berkeley Conference on Dutch Linguistics 1997: the Dutch Language at the Millennium, Thomas F. Shannon and Johan P. Snapper (eds.), 197–208. Lanham MD: University Press of America. 2005 Constructions of Intersubjectivity. Oxford: Oxford University Press. Verstraete, Jean-Christophe 2001 Subjective and objective modality: interpersonal and ideational functions in the English modal auxiliary system. Journal of Pragmatics 33. 1505–1528. 2002 Interpersonal Grammar and Clause Combining in English. Ph. D. Diss., Department of Linguistics. K.U. Leuven. Veyrat Rigat, Montserrat 1993 Aspecto, Perífrasis y Auxiliación: un Enfoque Perceptivo. LynX (a monographic series in Linguistics and World Perception). Annex 6. Universidad de Valencia. Vliegen, Maurice 2006 The relation between lexical and epistemic readings: the equivalents of promise and threaten in Dutch and German. In Topics in subjectification and modalization. Bert Cornillie and Nicole Delbecque (eds.), 73–95. Amsterdam/ Philadelphia: Benjamins
306
References
Wierzbicka, Anna 1988 The semantics of grammar. Amsterdam/Philadelphia: Benjamins. 1994 Semantics and epistemology: The meaning of ‘evidentials’ in a cross-linguistic perspective. Language Sciences 16.1. 81–137. Wilson, Deirdre 1975 Pressupositions and Non-Truth Conditional Semantics. New York: Academic Press. Wilson, Deirdre and Dan Sperber 2002 Pragmatics, Modularity and Mind reading. Mind and Language 17. 3–23. 2004 Relevance theory. In Handbook of pragmatics. L. Horn and E. Ward (eds.), 607–632. Oxford: Blackwell. Willett, Thomas 1988 A cross-linguistic survey of the grammaticalization of evidentiality. Studies in Language 12: 51–97. Yllera, Alicia 1980 Sintaxis histórica del verbo español: las perífrasis medievales. Zaragoza: Departamento de Filología Francesa, Universidad de Zaragoza.
Author index Achard, Michel 63, 232, 262, 283 Aikhenvald, Alexandra 7, 18, 19 Anderson, Lloyd 7, 223, 264, 266, 267, 268 van der Auwera, Johan 11, 115 Barcelona, Antonio 51, 157, 204 Benjamin, Carmen 263 Benveniste, Émile 23 Bermúdez, Fernando 4, 24, 42, 44, 45, 46, 267 Boas, Hans 10 Bolinger, Dwight 70, 72, 73, 74, 78, 270 Bolkestein, Machteld 225 Brisard, Frank 271, 272, 282 Butt, John 263 Bybee, Joan 6, 187, 225, 262 Cabeza Pereiro, Carmen 63, 83 Chafe, Wallace 6, 7, 10, 18, 19, 20, 21, 263, 267 Coates, Jennifer 87, 199, 262 Combé, Henk 48, 262, 267 Cornillie, Bert 3, 6, 104, 122, 125, 224 Dasher, Richard 12, 70, 157, 158, 278 De Haan, Ferdinand 5, 10, 195, 196, 265 De Mulder, Walter 192 De Vries, Gijs 272 DeLancey, Scott 265, 274 Delbecque, Nicole 136, 213, 265, 268, 271 Delport, Marie-France 227 Demonte, Violeta 6 Dendale, Patrick 5, 10, 191, 192, 193, 194, 195, 264, 265, 280 Díaz Bautista, Carmen 3
Diewald, Gabriele 90, 106 Dixon, Robert 7 Elvira, Javier 282 Faller, Martina 1, 10, 265, 266, 267 Fernández de Castro, 4, 5, 6, 181, 184, 185, 188, 189, 206, 222, 254, 262, 263, 280 Fernández Leborans, María Jesús 3, 15, 37 Fischer, Cynthia 12 Fitneva, Stanka 20 Fleischman, Suzanne 197, 262 Floyd, Rick 20, 37, 268 Gaatone, David 59, 58, 123, 136, 269 Geeraerts, Dirk 260 Givón, Talmy 12 Goethals, Patrick 280 Gómez Torrego, Leonardo 5, 181, 187, 188, 190, 206, 254, 262, 279, 280, 284 González Vázquez, Mercedes 5, 263 Goossens, Louis 187, 199, 262, 282 Grice, Paul 264, 283 Grondelaers, Stefan 260 Guentchéva, Zlatka 265 Guo, Jiansheng 282 Gutiérrez Ordóñez, Salvador 63, 262 Halliday, Michael 262 Hassler, Gerda 262 Heine, Bernd 106 Hernanz, María Lluïsa 3, 6, 17, 42, 44, 48, 55, 86, 263, 269 Heyvaert, Liesbet 282 Hopper, Paul 165 Ifantidou, Elly 9, 264, 268
308
Author index
Kronning, Hans 191, 193, 197, 204, 280, 281 Lamiroy, Béatrice 59, 58, 59, 92, 122, 123, 130, 268, 269, 277 Langacker, Ronald 12, 13, 59, 58, 59, 63, 64, 69, 79, 80, 81, 92, 93, 95, 122, 132, 136, 149, 150, 175, 176, 179, 180, 224, 230–239, 240, 245, 246, 250, 251, 262, 265, 267–272, 278, 283, 284 Lazard, Gilbert 265 Levinson, Stephen 265 Lozowski, Przemyslaw 272 Lunn, Patricia 232 Lyons, John 184, 195, 262 Maldonado, Ricardo 29, 102, 127, 270, 285 Marín-Arrese, Juana 262, 263 Matte Bon, Francisco 263 Miyashita, Hiroyuki 106 Mori, Olga 226 Mortelmans, Tanja 184, 196, 198, 239, 262, 279, 283 Narrog, Heiko 12, 262 Nichols, Joanna 7, 10, 263 Nuyts, Jan 1, 5, 10, 12, 13, 20, 24, 28, 42, 102, 106, 223, 224, 230, 231, 232, 233, 240, 257, 258, 259, 262, 263, 265, 266, 271, 272, 279, 280, 281, 282, 284 Olbertz, Hella 5, 188, 189, 190, 197, 207, 208, 210, 225, 226, 262, 263, 280, 281, 282 Palmer, Frank 7, 10, 11, 20, 37, 41, 182, 184, 262, 265, 271, 273 Pederson, Eric 12, 265 Peirce, Charles 192, 280 Pelyvás, Peter 154, 224, 230, 232, 233, 240, 242, 244, 245, 246, 247, 248, 249, 250, 252, 262, 282, 285
Perkins, John 262 Plungian, Vladimir 11, 20, 115, 116, 121, 181, 262, 266, 271, 283 Porroche, Margareta 3, 15, 37, 42, 55, 63, 83, 268 Radden, Günter 231, 234, 235, 283 Reis, Marga 90, 106 Rooryck, Johan 6, 87, 90, 91, 92, 104, 114, 115, 120, 175, 262, 265, 268, 270, 271, 272, 273, 279 Rosch, Eleonor 284 Ruwet, Nicolas 93, 274 Sanders, Ted 258, 259, 262 Sapir, Edward 10 Schiffrin, Deborah 24 Searle, John 99, 103, 154, 193, 264, 272 Seuren, Pieter 265 Silva-Corvalán, Carmen 5, 184, 185, 188, 190, 207, 213, 215, 216, 243, 280 Speelman, Dirk 260, 264 Sperber, Dan 264, 265 Spooren, Wilbert 258, 259, 262 Squartini, Mario 19, 193, 195, 196, 197, 262, 266, 280, 281 Sweetser, Eve 154, 179, 244, 245, 252, 278 Talmy, Leonard 244, 278 Tasmowski, Liliane 10, 265, 280 Taylor, John 246 Traugott, Elizabeth 6, 12, 23, 69, 70, 76, 78, 80, 87, 88, 93, 94, 95, 150, 151, 152, 153, 155, 156, 157, 158, 167, 172, 174, 175, 179, 196, 198, 233, 245, 277, 278 Usoniene, Aurelia 262 Van de Velde, Freek 266 Vázquez-Laslop, Eugenia 6, 86, 91, 263, 273, 274
Author index Verhagen, Arie 6, 14, 23, 87, 88, 89, 90, 93, 94, 95, 100, 116, 118, 150, 152, 155, 156, 167, 170, 179, 245, 256, 267, 272, 278 Verstraete, Jean-Christophe 231, 262, 280 Vliegen, Maurice 92, 104, 120, 146, 157, 272, 273, 274, 277
309
Willett, Thomas 7, 10, 18, 19, 264, 266 Wilson, Deirdre 264, 265 Yllera, Alicia 225, 226, 227, 282
Subject index agentive, 14, 59, 60, 62, 122–149, 164, 170, 175, 178, 185, 218, 221, 250–256, 274 alternation, 13, 15, 61, 83, 154, 155, 278 amenazar, 2, 6, 8, 9, 11, 14, 85–284 animate, 137, 146–148, 151, 277 assessment, 10, 11, 24, 77, 87, 90, 175, 182–187, 202, 211, 233, 241, 251, 265 attenuation, 14, 23, 52, 92, 95, 119, 122, 132, 136, 149, 174, 175, 176, 177, 178, 180, 245, 250, 252, 256, 272, 278, 284 auditory evidence, 18, 19, 29, 30, 31, 36, 267 auxiliary, 1, 2, 70, 249, 252, 279, 282 belief, 6, 18, 20, 27, 28, 36, 37, 44, 45, 55, 69, 76, 89, 113, 152, 157, 167, 186, 189, 201, 203, 207, 208, 210, 211, 233, 280, 283 beloven, 88–94, 117, 146, 150, 155, 156, 170–274 categories, 5, 17, 59, 60, 62, 115, 123, 136, 274 chi-Square, 266, 269, 274, 281 cognition, 11, 157 Cognitive Grammar, 13, 59, 84, 176, 230, 232, 233, 265, 267, 270, 282 cognitive-functional, 3, 13, 86, 92, 183, 224, 232, 253, 257 commitment, 4, 5, 9, 10, 11, 13, 14, 15, 20, 21, 22, 23, 37, 38, 39, 40, 41, 42, 48–56, 85, 92–125, 136, 140–144, 150–153, 157, 161, 168, 181, 185–190, 194–212, 222–231, 244, 250– 277 complementation, 8, 13, 84, 256
conceptualization, 11, 12, 13, 17, 23, 64, 65, 66, 67, 68, 83, 84, 174, 178, 256, 265, 267, 268, 279 conceptualizer, 16, 21, 27, 28, 31, 38–45, 53, 60, 62, 69, 80–84, 125, 129, 130, 176, 181, 187, 200, 211, 234, 246, 267, 268, 271, 274 corpus, 7, 8, 70–75, 158, 169, 226– 229, 263, 282, 283 deber, 2, 3, 4, 5, 8, 9, 11, 14, 21, 22, 181–269, 280–284 deduction, 14, 18, 185, 191, 192, 193, 202, 203, 208, 267, 280 deductive, 4, 5, 7, 14, 181, 187, 191– 212, 222, 251, 254, 280 definition, 6, 7, 10, 19, 24, 80, 239, 265, 277 deontic, 14, 86, 130, 132, 133, 135, 149, 154, 182–257, 262, 263, 278, 280–285 devoir, 5, 58, 191, 192, 193, 194, 196, 197, 204, 264, 280 diachronic, 3, 8, 13, 14, 59, 69, 70, 77, 83, 93, 94, 95, 125, 150, 155, 158, 167, 174, 175, 179, 180, 224, 225, 229, 230, 251, 253, 255, 256, 260, 279 discourse, 1, 7, 8, 9, 16, 17, 34, 62, 65, 95, 106, 172, 183, 195, 257, 260, 269 distribution, 7, 15, 16, 17, 36, 48, 49, 56, 59, 60, 86, 96–98, 106, 107, 111, 112, 122, 123, 136, 143, 167, 181–183, 190, 213–215, 219, 222–224, 238, 243, 244, 253, 263, 269 dreigen, 88–94, 117, 120, 150, 156, 157, 167, 170, 272–277 Dutch, 6, 24, 88, 89, 92, 94, 146, 167, 195, 196, 262, 272–282
Subject index dynamic, 13, 14, 59, 63–67, 83, 125– 128, 133, 164, 182, 184, 213–216, 221–230–256, 279–283 English, 6, 7, 58, 88, 89, 94, 103, 115, 150–155, 167, 174, 179, 187, 195, 196, 198, 199, 224, 231, 232, 234, 235, 236, 239, 240, 245, 248, 249, 251, 262, 263, 268, 278, 282, 283 epistemic modality, 1, 3, 5, 7, 8, 9, 10, 11, 83, 87, 92, 115, 181–198, 210, 212, 222, 224, 239–259, 263–266, 274, 278, 279, 282, 283 evidentiality, 1, 3, 5, 6, 7, 8, 9, 10, 11, 13, 15, 18, 19, 20, 24, 28, 36, 37, 39, 48, 92, 115, 116, 121, 181, 185, 187, 188, 190, 195, 196, 198, 210, 212, 222, 253, 254, 257, 258, 259, 262, 263, 264, 265, 266, 271, 280 focal prominence, 63–69, 83, 174, 178, 270 force, 14, 92, 96, 100–103, 120, 133, 141, 154–157, 167, 170, 189, 212, 238, 239, 240, 241, 244–249, 250–257, 267, 278, 279, 284 French, 6, 23, 58, 87, 90, 150, 191, 195, 204, 262, 265 general knowledge, 28, 31, 189, 194, 200 German, 24, 92, 114, 141, 146, 177, 194, 195, 198, 237, 239, 240, 273, 274, 277, 279 grammaticalization, 12, 93, 153, 196, 226, 227, 233, 239, 255, 257, 272, 277 grounding, 13, 14, 65, 66, 224, 225, 230–255, 278, 283 hearsay, 6, 9, 10, 18–55, 187, 193, 198, 205, 212, 267, 268 human agent, 134, 140, 172
311
illocutionary force, 1, 14, 86, 92, 96, 102, 103, 112, 117, 120, 156, 160, 168–171, 179, 254, 278, 279 impersonal, 58, 65, 130, 131, 132, 135, 227, 267, 269, 274 induction, 14, 18, 19, 191, 193, 200 inductive, 6, 14, 181, 187, 191, 194, 195, 198, 199, 200, 201, 202, 203, 206, 208, 209, 210, 212, 222, 250, 254 inferential, 4, 5, 10, 11, 14, 18, 19, 20–29, 36–50, 59, 87, 88, 89, 90, 111, 115, 116, 121, 179, 181, 185–212, 222, 226, 241, 254, 262, 264, 267–269, 280, 281 interlocutor, 1, 21, 26, 30, 46, 47, 93, 102, 103, 109, 122, 134, 150–157, 177–179, 194, 211, 215, 238, 273 interplay of forces, 132, 135, 155, 233, 248, 249, 250, 252, 284 intersubjectivity, 13, 16, 23, 24, 45, 46, 53, 56, 82, 268, 279, 281 Latin American Spanish, 7, 9, 263 left-dislocation, 15, 32, 33, 34, 37, 40, 42, 46, 47, 56, 68, 69, 74, 78, 267, 270 lexical reading, 92, 117, 128, 133– 135, 141–148, 154, 157, 161, 164, 173, 271, 274, 277 likelihood, 1, 3, 11, 38, 85, 87, 88, 89, 92, 109, 113, 115, 120, 152, 175, 181–208, 233, 237, 240, 250, 251, 254, 257–260, 266, 272, 274 main clause, 8, 23, 59, 101, 142, 271, 278, 279 may, 2, 6, 11, 22, 23, 32, 37, 38, 39, 40, 41, 43, 47, 52, 58, 63, 65, 68, 75, 78, 83, 87, 88, 89, 91, 92, 94, 100, 108–112, 119, 123, 126, 134, 144, 152–157, 169, 175–179, 182, 184, 185, 186–198, 204, 206, 213, 224, 231–240, 244–249, 250, 257–260, 268–284
312 Subject index mental operation, 23 metaphor, 150, 152, 153, 154, 155, 157, 179, 245 metonymy, 69, 76, 80, 150, 153, 156, 157, 161, 179 middle voice, 130 modal reading, 6, 14, 97, 117, 118, 126, 128, 131, 132, 133, 134, 139, 142, 143, 144, 145, 146, 147, 148, 149, 152, 154, 155, 156, 162, 163, 164, 168, 170, 171, 185, 236, 249, 251, 272, 283 modes of knowing, 1, 11, 14, 15, 18, 19, 181, 202, 207, 254, 255 must, 2, 3, 5, 6, 7, 9, 11, 18, 21, 22, 28, 30, 58, 93, 99, 115, 151, 157, 187, 188–212, 224, 228, 229, 231, 235, 239–247, 265, 268–272, 281, 283 narrow scope, 233, 249, 252 necessity, 4, 11, 46, 90, 91, 92, 114– 116, 121, 189–196, 203, 239, 262, 271, 273, 281 obligation, 4, 14, 22, 99, 182, 187, 188, 198, 203, 210, 212, 215, 225, 228, 244, 257, 279, 280, 281 offstage, 79, 80, 81, 82, 175, 234, 271 onstage, 79, 80, 81, 82, 175, 234, 235, 236, 238, 256, 271, 283 parecer, 1–86, 123, 181, 213, 253– 274, 279, 283, 284 parenthetical, 15, 34, 35, 36, 40, 47, 48, 53, 267, 268, 270 participant structure, 1, 8 past tense, 106, 107, 110, 111, 113, 120, 200, 231, 238, 273 peninsular Spanish, 7, 263 perceptual evidence, 4, 5, 153, 193 poder, 2–14, 30, 54, 133, 181–190, 204, 210–218, 221–232, 236,
238–243, 250–257, 264, 279, 280–284 possibility, 4, 11, 46, 53, 64, 65, 68, 90, 91, 92, 104, 106, 114, 115, 149, 152, 157–160, 167, 174, 184–190, 195, 210, 221, 222, 254, 271, 274, 275, 279, 284 potency, 175–180, 234–236, 241– 249, 283 prediction, 6, 14, 85, 86, 87, 88, 92, 100–132, 157, 159, 172, 178, 179, 254, 255, 257, 277 present tense, 102, 106, 107, 109, 111, 112, 113, 120, 126, 231, 232, 234, 237 profile, 132, 133, 175, 176, 180, 224, 232–235, 239–251, 256, 283, 284 prometer, 2, 6, 8, 9, 11, 14, 85–181, 213, 253–263, 271–279, 284 prominence, 13, 59, 63, 64, 66, 68, 95, 103, 118, 138, 144, 147, 179, 206, 218, 230, 233, 242, 247, 250, 252, 256 qualification, 5, 10, 11, 12, 20, 21, 22, 23, 24, 29, 32, 37, 41, 44, 55, 88, 166, 167, 181, 185, 188, 201, 212, 254, 259, 262, 264, 266, 281 raising, 6, 17, 63, 86, 92, 93, 157, 175, 270, 271–274, 279 reasoning, 11, 18, 19, 20, 25, 28, 29, 36, 87, 88, 109, 155, 186– 200, 206–211, 245, 254, 267, 280 Relevance Theory, 9, 264 reliability, 10, 15, 20, 21, 22, 23, 37– 52, 55, 194 reported information, 18, 19, 22, 35, 36, 205 reportive, 14, 187, 195, 196, 197, 198, 199, 202, 204, 205, 206 resultar, 1–20, 48–83, 181, 253–270 semantic classification, 59, 60, 61, 62, 137
Subject index sentence position, 31 source of information, 5, 9, 10, 18, 19, 20, 21, 25, 28, 30, 31, 33, 35, 37, 39, 43, 50, 88, 110, 120, 190, 198, 264, 267 Spanish, 3, 4, 5, 6, 7, 9, 15–21, 37, 53, 64–68, 70–81, 86, 91, 94, 146, 150, 152, 158–280, 283 speech participants, 23, 43, 45, 65, 66 spoken language, 10, 16, 17, 97, 181–183, 201, 204, 253, 257, 263, 269 stance, 1, 23, 49 subjectification, 9, 12, 13, 14, 56, 59, 69, 70, 76–87, 92– 95, 121, 149, 150, 153, 157, 174–177, 179, 180, 224–230, 234, 235, 236, 244–252, 255, 256, 271, 278 subjectifier, 20, 29, 36, 38, 40 subjectivity, 13, 15, 18, 23, 24, 25, 42–56, 64–69, 83, 84, 93, 95, 103, 156, 230, 235, 251, 253, 258–260, 267 subordinate clause, 8, 59, 142, 173, 191 synchronic, 3, 8, 12, 13, 14, 59, 69, 79, 83, 93–96, 150, 174, 224, 230, 253, 255, 256, 259
313
tener que, 2, 3, 4, 5, 8, 9, 11, 14, 132, 133, 181, 183, 185, 187, 189, 190, 206–213, 219–229, 236, 240, 241–244, 247, 250, 251–255, 257, 262, 280, 281, 282 trajector, 176, 177, 234, 235, 249, 250, 256, 270, 283 transparency, 12, 13, 14, 59, 58, 59, 60, 69, 78, 83, 92–95, 121–125, 130, 149, 175, 181, 213, 255, 256, 271 transparent, 13, 58, 59, 60, 61, 62, 63, 76, 92, 122, 125, 135, 136, 148, 149, 175, 213, 222, 255, 256, 257, 269, 277 Tuyuca, 10 verb type, 59, 61, 62, 83, 123, 255, 270 verbal periphrases, 4 viewpoint, 123, 125, 126, 128, 129, 130, 148, 237, 255, 274 visual, 11, 18, 19, 29, 30, 31, 36, 145, 200, 267, 277, 280 wide scope, 233, 249, 252 written language, 16, 17, 34, 59–62, 181–183, 253, 257, 263, 266, 269