Cognitive Approaches to Tense, Aspect, and Epistemic Modality (Human Cognitive Processing)

Cognitive Approaches to Tense, Aspect, and Epistemic Modality Human Cognitive Processing (HCP) Cognitive Foundations...

Author: Adeline Patard | Frank Brisard (editors)

92 downloads 1340 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

DOWNLOAD PDF

Cognitive Approaches to Tense, Aspect, and Epistemic Modality

Human Cognitive Processing (HCP)

Cognitive Foundations of Language Structure and Use This book series is a forum for interdisciplinary research on the grammatical structure, semantic organization, and communicative function of language(s), and their anchoring in human cognitive faculties. For an overview of all books published in this series, please see http://benjamins.com/catalog/hcp

Editors Klaus-Uwe Panther University of Hamburg

Linda L. Thornburg

Editorial Board Bogusław Bierwiaczonek

Higher School of Economics and Humanities, Poland

Mario Brdar

Josip Juraj Strossmayer University, Croatia

Barbara Dancygier

University of British Columbia

N.J. Enfield

Max Planck Institute for Psycholinguistics, Nijmegen & Radboud University Nijmegen

Elisabeth Engberg-Pedersen University of Copenhagen

Ad Foolen

Radboud University Nijmegen

Raymond W. Gibbs, Jr.

University of California at Santa Cruz

Rachel Giora

Tel Aviv University

Elżbieta Górska

University of Warsaw

Martin Hilpert

Freiburg Institute for Advanced Studies

Zoltán Kövecses

Eötvös Loránd University, Hungary

Teenie Matlock

University of California at Merced

Carita Paradis

Lund University

Günter Radden

University of Hamburg

Francisco José Ruiz de Mendoza Ibáñez University of La Rioja

Doris Schönefeld

University of Leipzig

Debra Ziegeler

Paul Valéry University, France

Volume 29 Cognitive Approaches to Tense, Aspect, and Epistemic Modality Edited by Adeline Patard and Frank Brisard

Cognitive Approaches to Tense, Aspect, and Epistemic Modality Edited by

Adeline Patard Frank Brisard University of Antwerp

John Benjamins Publishing Company Amsterdamâ•›/â•›Philadelphia

8

TM

The paper used in this publication meets the minimum requirements of American National Standard for Information Sciences – Permanence of Paper for Printed Library Materials, ansi z39.48-1984.

Library of Congress Cataloging-in-Publication Data Cognitive approaches to tense, aspect, and epistemic modality / edited by Adeline Patard, Frank Brisard. p. cm. (Human Cognitive Processing, issn 1387-6724 ; v. 29) Includes bibliographical references and index. 1. Cognitive grammar. 2. Grammar, Comparative and general--Tense. 3. Grammar, Comparative and general--Modularity. 4. Psycholinguistics. I. Patard, Adeline. II. Brisard, Frank. P165.C636â•…â•… 2011 415’.6--dc22 isbn 978 90 272 2383 8 (Hb ; alk. paper) isbn 978 90 272 8521 8 (Eb)

2011014902

© 2011 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Co. · P.O. Box 36224 · 1020 me Amsterdam · The Netherlands John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa

Table of contents List of contributors Acknowledgments Introduction: Cognitive approaches to tense, aspect, and epistemic modality Frank Brisard and Adeline Patard

vii ix 1

part i.â•‡ Theoretical foundations The definition of modality Renaat Declerck

21

The English present: Temporal coincidence vs. epistemic immediacy Ronald W. Langacker

45

The organization of the German clausal grounding system Elena Smirnova

87

Grounding in terms of anchoring relations: Epistemic associations of ‘present continuous’ marking in Turkish Ceyhan Temürcü

109

part ii.â•‡ Descriptive application: Cognitive Grammar Some remarks on the role of the reference point in the construal configuration of “more” and “less” grounding predications Elena Smirnova and Tanja Mortelmans New current relevance in Croatian: Epistemic immediacy and the aorist Mateusz-Milan Stanojević and Renata Geld Aspect as a scanning device in natural language processing: The case of Arabic Lazhar Zanned

137

159

181



Cognitive Approaches to Tense, Aspect, and Epistemic Modality

part iii.â•‡ Descriptive application: Other cognitive approaches Imperfective aspect and epistemic modality Ronny Boogaart and Radoslava Trnavac

217

Communicating about the past through modality in English and Thai Katarzyna M. Jaszczolt and Jiranthara Srioutai

249

The epistemic uses of the English simple past and the French imparfait: When temporality conveys modality Adeline Patard

279

Name index Subject index

311 315

List of contributors Ronny Boogaart University of Leiden [email protected] Renaat Declerck K.U.Leuven Campus Kortrijk [email protected] Renata Geld University of Zagreb [email protected] Katarzyna M. Jaszczolt University of Cambridge [email protected] Ronald W. Langacker University of California at San Diego [email protected] Tanja Mortelmans University of Antwerp [email protected] Adeline Patard University of Antwerp [email protected]

Elena Smirnova Leibniz University Hanover [email protected] Jiranthara Srioutai Chulalongkorn University, Bangkok [email protected] Mateusz-Milan Stanojević University of Zagreb [email protected] Ceyhan Temürcü Middle East Technical University, Ankara [email protected] Radoslava Trnavac Simon Fraser University [email protected] Lazhar Zanned University of Manouba, Tunis [email protected]

Acknowledgments Some of the contributions included in the present volume were presented during the 7th Chronos conference, held at the University of Antwerp, Belgium, from September 18–20, 2006. Others (Jaszczolt & Srioutai, Patard, Smirnova, Smirnova & Mortelmans, Stanojević & Geld) have been especially commissioned. The theme uniting them, the interaction between tense, aspect, and modality, was also the special topic of this Chronos meeting. We would like to thank the authors involved for the extreme amiability and patience that they exhibited throughout this somewhat laborious project. We are much obliged to all of them, and excited that we are now able to present their work collectively as an important attempt to further the study of time and modality within the domain of grammar in a (more or less) cognitive vein. This publication also owes a great deal to the (once again very professional) services offered by various collaborators at John Benjamins. We are more than grateful to the acquisition editors for the Human Cognitive Processing: Cognitive Foundations of Language Structure and Use book series, Bertie Kaal and particularly Hanneke Bruintjes, who showed enough confidence to continue her unwavering support and encouragements. In the last stages, the assistance provided by the series editors, Klaus-Uwe Panther and Linda L. Thornburg, proved invaluable as well. Finally, we wish to thank an anonymous referee for useful suggestions and challenging, though sympathetic, observations. Frank Brisard Adeline Patard Antwerp, February 2011

Introduction Cognitive approaches to tense, aspect, and epistemic modality Frank Brisard and Adeline Patard University of Antwerp

1. Background Various issues of grammatical time, including tense and aspect, are inextricably linked, to varying extents, with concerns of mood and modality, in particular of an epistemic nature. This is not an especially striking or new observation, given the frequent treatment of markers of tense, aspect, and mood/modality in terms of one integrated super-category, usually referred to as ‘TAM’ or ‘TMA’ and covering the three main semantic domains marked on the verb (this holds especially for studies of a typological orientation, where TAM categories are seen as pervasive in the world’s languages and therefore generally considered to be universals of human language; see, e.g., Bybee et al. 1994, Chung & Timberlake 1985, Dahl 1985, Givón 2001). This treatment, however, is more often than not motivated on a purely formal basis, reflecting, if not the regular co-occurrence of such markers, then at least their intricate ways of interacting. At the same time, the three categories, in standard generative work, are not even supposed to be represented at the same syntactic level, thereby failing (a priori) to address the notional parallelisms that unite the meanings expressed by them.1 Notably, the exclusive characterization of tenses as 1. In formal treatments of tense and aspect, reference is sometimes made to a modal analysis, e.g., in Dowty’s (1977) work on the progressive, which is analyzed as a combined temporal/ modal operator (of necessity; see also Portner 1998). Also, Hornstein (1990) deals with tenses, modals, and the perfect as having the same temporal organization. However, the conception of modality in such types of truth-conditional semantics, as a function of possible worlds, is rather far removed the more traditional way in which this category is presented in the present volume, viz., as the grammatical expression of various degrees of (un)certainty. One of the few more philosophically inspired attempts so far to see temporality in terms of epistemic modality seems to be the recent monograph by Jaszczolt (2009).



Frank Brisard and Adeline Patard

indicating (relative or absolute) location in time, i.e., as essentially temporal markers, has inspired many scholars to deal with non-temporal usage types of tense markers, which are abundant cross-linguistically, as either derived pragmatically from a more basic or ‘pure’ temporal semantics (i.e., to treat them as secondary meanings), or to deny the link with their temporal uses altogether and propose an account in terms of homonymy, according to which the two types of usage are semantically unconnected. The same holds, a fortiori, for markers of aspect, it seems.2 Thanks to this tactic, it has been possible to preserve the age-old assumption, backed up by numerous philosophical models dealing with the relation between language and reality, that tense is not a psychological category dependent on the speaker’s subjective choices, but rather an objective property of the world (see Ludlow in preparation for a recent overview of this discussion). At the same time, students of tense have thus largely succeeded in banning concerns of non-temporal uses from the discussion of grammatical time, arguing, quite simply, that calling a tense or aspect modal must boil down to a case of miscategorization (except perhaps in the paradigm of the future, where the link is less obviously denied). Cognitive theories of grammar have suggested an alternative to this mainstream approach to the grammatical marking of time.3 In Cognitive Grammar (Langacker 1991), tenses, like moods, are called ‘grounding predications’, because they relate a proposition to the ground or situation of speech. This relation is said to be an epistemic one, since the speaker/conceptualizer, when talking about concrete states of affairs, is primarily interested in the knowledge status that is to be assigned to them: how certain can she be of the state of affairs conjured up in a clause, and how accessible or immediate are the means to verify that status? When referring to tenses in the indicative, this status is necessarily one of certainty, meaning that the speaker presents a state of affairs as belonging to her conception of reality at the time of speaking. The idea, then, is to draw up a typology of various kinds of reality between which speakers of a given language care to distinguish. Thinking about different types of reality might appear something of a contradiction at first, but it is not different from, and basically analogous to, distinguishing different types of non-reality (possibility, probability, counterfactuality, etc.). For instance, the use of a simple present in English to refer to an unqualified or 2. This is so despite various hints in the past regarding the systematic nature of the link between certain aspectual and modal categories (Fleischman 1995, on the connotations of non-reality associated with past imperfectives, is a classic example in this respect; cf. also Barceló and Bres 2006, James 1982, Martin 1991, Mellet 1990, and Patard 2007; for general past-time reference and modality, see Dahl 1997, Fleischman 1989, Hogeweg 2009, Iatridou 2000, Larreya 2003). 3. Not only cognitive approaches present such an alternative. For instance, the French enunciativist tradition treats tense as a linguistic category that touches upon the speaker’s subjectivity by definition.

Introduction

immediate reality (either a constant state or a dynamic process construed generically or habitually) represents another type of reality than with that of a present progressive, which bestows more of an incidental or contingent character upon an (equally real, but qualified) situation. This interaction between tense and aspect may be said to affect the whole of the (English) tense system, and part of the work cut out for those less formally inspired accounts included in the present volume consists precisely in uncovering such differences, which more often than not go beyond the strictly temporal characteristics ascribed to tense and aspect categories. Not much of that has been done yet, it should be noted, though in the wake of Cognitive Grammar a definite start appears to have been taken in the more recent past (see the contributions on TAM in Brisard 2002a and Boogaart & Janssen 2007 for an overview). Relevant studies undertaken in both mental-space theory (e.g., Cutrer 1994, Doiz-Bienzobas 1995) and Construction Grammar (e.g., Michaelis 2004, Croft in preparation) follow up on some of the leads proposed by the discussion of grounding in Cognitive Grammar, or come up with their own functional explanations. More generally, linguists (e.g., in the French tradition) are beginning to grasp the complex interconnections between the domains of tense, aspect, and modality, both at the level of grammar and at that of discourse (see Abraham & Leiss 2008, Barceló et al. 2006, Condoravdi 2002, Declerck 2005, Gosselin 2005, Stowell 2004, and Trnavac 2006). Not all contributions in this volume are strictly to be situated within the theory of Cognitive Grammar, or even within the wider paradigm of cognitive linguistics. However, all of them do subscribe to the view that the choice of grammatical markers of time is motivated by semantic factors. Moreover, these studies also, to varying degrees, present an explicit attempt to unify the different usage types characterizing a particular (set of) construction(s), which is an important point of methodology in the analysis of tense and aspect semantics. In passing, most of the present contributions comment on issues of subjectivity linked with the description of these markers, which ties in neatly with the more general usage-based perspective suggesting that elements of grammar develop out of, and should therefore be seen as, subjectified or speaker-oriented uses of linguistic constructions. 2. Overview of the contributions A first set of chapters, presented in Part I, deals with some fundamental issues in thinking about modality, and how it relates to (the grammatical marking of) temporality. First, Declerck proposes a model that accounts for modality as a notional category and its expression in English, among other means, through the use of modal auxiliaries. In the second chapter, Langacker addresses the contrasting





Frank Brisard and Adeline Patard

modal values of ‘epistemic immediacy’ and ‘epistemic contingency’, suggested by Brisard (2001, 2002c, 2005) to describe the meanings of the two English present tenses (simple and progressive). In the third chapter, Smirnova presents a systematic discussion of temporal and modal grounding in German, which turns out to be quite different from the English system. And in the final chapter of Part I, Temürcü takes up the connection between epistemic immediacy vs. contingency and grammatical present-time reference again, and confronts it with the meaning of the ‘present continuous’ marker in Turkish. He describes and analyzes this meaning within the framework of his own model of ‘anchoring relations’, which exhibits an explicitly cognitive orientation while at the same time challenging the assumption, entertained by students in Cognitive Grammar, that all grounding predications have an epistemic and thus modal, rather than temporal, core meaning. To begin with, Renaat Declerck makes a distinction in his contribution, “The definition of modality”, between different types of world: the ‘factual word’ comprises all the situations that are actualizing at S (the speech time) or have actualized before S, a ‘counterfactual world’ is a world distinct from the factual world, and a ‘nonfactual world’ or ‘theoretical world’ is a possible world that is not interpreted as being the factual world. He suggests a comprehensive and concrete definition of modality according to which a situation ‘is represented as actualizing in a nonfactual world’. Four modalizing devices are observed in English. The first two mechanisms rely on the marked use of a verbal tense instead of the unmarked form so as to produce a tentative or counterfactual interpretation: ‘modal backshifting’ consists in using one or more backshifting form(s) (past and/or perfect tense), and modal conditionalization concerns the use of a conditional rather than a present tense in a non-conditional context. Declerck accounts for the category of ‘nonneutral theoretical’ conditionals, i.e., non-factual conditionals that denote a specific link between the suppositional world expressed and the factual world. This link can be of four different types: closed, open, tentative, or counterfactual, ranging from a full coincidence with the factual world to an asserted divergence from it. Finally, the notion of ‘not-yet-factual at t’ is introduced, characterizing a world that is non-factual/modal at the time in which it is anchored. Declerck further defines the notions of epistemic modality and root modality. Epistemic modality is claimed to establish an ‘epistemic relation’, i.e., a degree of compatibility between the modal world and the factual world. This epistemic relation can be left indeterminate, e.g., in imaginary worlds, or it can be specified and assigned a factuality value. To account for the meaning of English modal auxiliaries (and conditionals), Declerck outlines an epistemic scale of factuality values ranging from ‘factual’ to ‘counterfactual’. Root modality, on the other hand, is concerned with ‘factors that determine the actualization of a [...] situation in a nonfactual world’. Contrary to epistemic modality, root modality does not imply any epistemic

Introduction

relation, but a ‘modal position’ assumed by the speaker as regards the conditions of a situation’s actualization. By way of conclusion, Declerck looks at ‘world-evoking’ lexical verbs and identifies three types of them: ‘factive verbs’ (e.g., know), which automatically refer to a factual world, ‘intensional verbs’ (e.g., believe), which create a specified epistemic modal world, and ‘attitudinal verbs’ (e.g., intend), evoking a root modal world. The chapter by Ronald W. Langacker, “The English present: Temporal coincidence vs. epistemic immediacy”, tackles the question of whether the English (simple) ‘present tense’ is best described as marking tense or modality. Langacker confronts his account of the English present tense as indicating temporal coincidence with the time of speaking (Langacker 1991, 2001) with Brisard’s analysis (2001, 2002c, 2005) in terms of ‘epistemic immediacy’. He first tackles the puzzle of present perfectives in English: in the present, perfective processes are compatible with the progressive construction but not with the simple present (at least if they are to indicate actual coincidence with the present), whereas, in the domain of the past, they can occur with both. The problem is viewed as conceptual (i.e., semantic), with both a durational and an epistemic component: canonically, dynamic events (typically short changes of state, like kicking a ball) cannot both be observed and reported at exactly the same time, and thus do not fall under Langacker’s definition of the English present tense as indicating ‘full and exact coincidence with the time of speaking’. This is so because the time it takes to linguistically report a token event, like kicking a ball, is rarely coextensive with the event itself (durational) and/or the starting point of the event rarely coincides with the moment of its recognition (epistemic) as a type of event, like the ‘kicking a ball’ type – unless, of course, the event in question is in some sense predictable or stereotypical (as with habituals and play-by-play sporting reports, for instance) or is by definition coincident with the speech event (as with performatives). Langacker also examines some non-present uses of the present tense: the ‘historical present’, the present of ‘scheduled future’ and in singular generics. To explain them, he suggests a special viewing arrangement in which the event coincident with the time of speaking is not the actual ‘represented event’, but a virtual (or mental) ‘representing event’ through which the actual process is apprehended. Next, Langacker considers an epistemic account of the present tense in the line of Brisard’s proposal, reinterpreting the semantic contrast in English between the present tense and the past tense as expressing the inclusion of the designated process in the immediate or non-immediate reality of a speaker/conceptualizer. This explicitly modal conception allows him to explain the present perfective puzzle and the non-present uses of the present tense, as well as the combination of the past tense with modals (might, would etc.) to refer to present situations, typically in a more mitigated way. These past modals are shown to indicate that the reality at issue, as the basis of a





Frank Brisard and Adeline Patard

proposition’s modal assessment, is ‘non-immediate’, i.e., distinct from that of the current speaker: it might, as with past-tense conditionals for instance, involve an imagined or virtual conception of an alternative immediate reality, differing only slightly from the speaker’s actual one, viz., in the premises contained by the protasis. Given that the so-called tense marking (past or present) in such cases does not apply to the proposition itself but to the status of the modal assessment, where temporal location is relatively irrelevant, Langacker concludes that ‘epistemic immediacy’ more adequately describes the schematic meaning of the English present tense, including on modals, and that ‘temporal coincidence’ corresponds rather to its prototypical value. In “The organization of the German clausal grounding system”, Elena Smirnova presents a systematic discussion of temporal and modal grounding in German, which for one turns out quite different from the English system. Smirnova starts by discussing the notion of grounding, developed in Cognitive Grammar, as a specific case of deixis or indexicality. She underlines the necessarily relational nature of deictic signs in general, and of grounding predications in particular. Three basic components are suggested in the characterization of grounding predications: (i) the nature of the grounded entity (nominal/clausal) that is related to the ground, as an immanent reference point, (ii) the nature of the grounding relation itself, yielding different concrete grounding values or meanings, and (iii) the different conceptualizations of, or viewing arrangements on, the ground. Smirnova proposes to focus on the third aspect and argues for the possibility of having two alternative deictic construals of the ground. In the default construal, the speaker and the hearer are identified as a common deictic center. The epistemic region that does not coincide with this ground and is thus non-immediate or distal corresponds to a third party. In the second configuration, the ground is construed as a speakerexclusive deictic center, excluding the hearer. In this case, the distal epistemic region also comprises the position of the hearer. The contrast is then applied to the German grounding predications, starting with temporal expressions. According to Smirnova, this system is marked in German by the opposition between the present tense, which situates the profiled event is in the temporal region of the ground, and the past tense, locating the profiled event outside of this region. As for the ground, its construal configuration is characterized by the adoption of a temporal vantage point shared by both the speaker and the hearer (common deictic center). Next, Smirnova deals with the modal grounding system of German, which is founded on the opposition between the indicative and the subjunctive mood. She posits that all indicative markers situate the profiled event within the realm of reality, whereas subjunctive marking locates the profiled event outside of reality. It is further assumed that the present and the preterit subjunctive entail different construal configurations of the ground. The preterit subjunctive is argued to relate

Introduction

the profiled event to the common deictic center, thus locating it outside of the reality shared by the speaker and hearer. In comparison, the present subjunctive relates the profiled event to the speaker-exclusive deictic center and consequently situates the event in a region outside of the speaker’s reality, but which may include the hearer’s reality. Finally, epistemic uses of the German modals are claimed to constitute an altogether different sub-system from that of German mood: they situate the profiled event in various spheres of (the speaker’s) knowledge of reality, and allow for only one configuration of the ground construed as a speaker-exclusive deictic center. The last chapter, by Ceyhan Temürcü, on “Grounding in terms of anchoring relations: Epistemic values of ‘present continuous’ marking in Turkish” explores the association between present-time reference and the evidential category of ‘new information’ (also known as ‘immediate meaning’ (Nichols 1986) or ‘mirativity’ (DeLancey 1997, 2001)). In particular, Temürcü examines the epistemic interpretations of the ‘continuous aspect’ marker -Iyor in Turkish (when used as the only TAM marker on the verb). For the sake of the analysis, he has developed his own theoretical framework of ‘anchoring relations’, inspired both by Cognitive Grammar’s conceptual account of ‘grounding relations’ and by semantic versions of Functional Grammar that deal with ‘underlying clause structure’. He expands on the concept of ‘grounding’ (see, among others, Langacker 1991, 1997, 2002; Brisard 2002b, c) and proposes a set of anchoring relations that are purely semantic in nature and that account for the use of tense, aspect and mood markers in utterances of natural languages. They involve epistemic but also aspecto-temporal categories and, moreover, each of the meanings of so-called grounding predications is decomposed using temporal, epistemic, and volitional ‘building blocks’. Temürcü then gives a comprehensive description of the temporal (present continuous or restricted habitual) and epistemic (‘certainty’ or ‘new information’) uses of Turkish -Iyor. He notices that this marker typically serves to convey epistemic contingency, i.e., “to express information incidental to a certain knowledge state (that of the speaker at the time of utterance or a displaced epistemic center)”. Temürcü departs from Brisard’s (2005) analysis of the English present progressive, semantically a close counterpart of the -Iyor form. Indeed, he adopts a ‘family resemblance’ conception of the polysemy of -Iyor, and, accordingly, does not derive its temporal uses from a more modal ‘schematic meaning’ in terms of epistemic contingency. Instead, he posits a pragmatic relation of metonymy, whereby epistemic contingency most frequently (but not necessarily) arises within the same conditions of use as present-time reference. By way of conclusion, epistemic contingency is identified as a sufficient condition for the use of -Iyor in many temporal contexts: it is what remains after a process of subjectification turns a marker of present-progressive aspect into one with persistent perfect, future, and habitual-gnomic values.





Frank Brisard and Adeline Patard

Part II contains applications of the grounding model proposed in Cognitive Grammar to deal with the meanings of particular tense and mood/modal markers in various languages. The study by Smirnova & Mortelmans argues for a modal characterization of verb forms in German, suggesting specific ‘construal configurations’ triggered by the context. Stanojević & Geld investigate the ‘aorist’ in Croatian, applying the notion of ‘epistemic immediacy’ to the domain of past-time reference. And in his chapter, Zanned analyzes aspectual categories in Arabic (the inflectional indication of perfective/past vs. imperfective/present, but also derivational, syntactic, and lexical marking) against the background of cognitive psychology. The chapter by Elena Smirnova & Tanja Mortelmans, “Some remarks on the role of the reference point in the construal configuration of ‘more’ and ‘less’ grounding predications”, investigates the different degrees of grounding and subjectification exhibited by the German construction würde + infinitive in its different uses. In Cognitive Grammar, a ‘grounding predication’ relates a profiled/designated entity to the ground or situation of speech, thus locating it in a specific epistemic region (e.g. ‘immediate reality’; cf. above). Following Mortelmans (2002, 2006), a gradual view on grounding and on subjectification is proposed. Smirnova & Mortelmans suggest the grammatical relevance of positing intermediate stages between objective construals, which imply no relation to the ground or where that relation is effected in an explicit manner, and subjective construals. They assume that such intermediate configurations can be described by adding a relevant reference point that cannot automatically be equated with the ground or the clausal subject, and that this reference point can be inferred via information present in the surrounding context. This criterion, indicative of an intermediate degree of grounding and subjectification, is tested against the German construction würde + infinitive, which is compared to the preterit subjunctive and the construction werden + infinitive. The preterit subjunctive is argued by Smirnova & Mortelmans to be a grounding predication that locates the profiled situation outside of reality. For them, it involves a ‘true’ grounding relation since it construes the ground as the only relevant reference point to which the profiled situation is linked. However, according to Smirnova & Mortelmans werden + infinitive has only acquired an intermediate degree of grounding and subjectification. Indeed, they demonstrate that, for werden + infinitive to be interpreted in an unambiguous way (as either future or epistemic), it always requires an additional reference point that cannot coincide with the ground. As for würde + infinitive, this construction is analyzed, in line with Smirnova (2006), as a polysemous construction that may function either as an analytical alternative to the synthetic preterit subjunctive or as the preterit counterpart of the indicative construction werden + infinitive. Once again, these two variants show different degrees of grounding and subjectification.

Introduction

The subjunctive würde + infinitive and the preterit subjunctive are both viewed as true grounding predications which construe the ground as the sole relevant reference point, whereas the indicative würde + infinitive entails, just like werden + infinitive, a construal configuration calling for an additional relevant reference point that is responsible for its temporal or epistemic reading. The temporal/ epistemic use of würde + infinitive is consequently not considered an instantiation of a true grounding predication. In “New current relevance in Croatian: Epistemic immediacy and the aorist”, Mateusz-Milan Stanojević & Renata Geld suggest a new and unified characterization of the Croatian ‘aorist’, classically analyzed as an absolute past tense (in narrative sequences), in the light of more recently developed usage types. They first briefly survey the different uses of the aorist traditionally described in grammars: the narrative aorist, the aorist of recent actions, the aorist of proverbs, and the aorist of future actions. These uses and the observation that the aorist predominantly occurs with the first person singular suggest that an epistemic definition of the aorist is more adequate than a temporal one. Stanojević & Geld investigate present-day uses of the aorist by means of a corpus study and a questionnaire. The corpus study consisted in extracting, from the Croatian National Corpus, all the tokens of six highly frequent verbs in the first person singular. Here, it emerges that the aorist in contemporary Croatian tends to behave like a perfect tense insofar as it is typically used without absolute past-time reference (as in contexts of current relevance). As for the questionnaire, its aim was to harvest grammaticality judgments from young native speakers about their uses of the aorist and of other tenses like the Croatian ‘perfekt’. The results show that the aorist is preferred in recent-past and immediate-future contexts, especially in spoken language, and that it is not typical of narrative counter-sequential contexts, which prefer the ‘perfekt’ to refer to events preceding previously mentioned events. To account for these patterns within the framework of Cognitive Grammar, a schematic definition of the aorist is proposed as marking ‘epistemic immediacy [...] and precedence of the action with relation to the moment of speech’. More specifically, Stanojević & Geld posit a specific viewing arrangement according to which the process is construed with extreme subjectivity (Langacker 2003) as immediately preceding the ground on a virtual plane distinct from the actual plane. This description allows them to explain both the recent-past and the immediate-future uses. Finally, a distinction is maintained between two virtual planes V1 and V2, which reflect the usage differences between the aorist and the perfekt but are also suggested to be relevant to the analysis of the Croatian tense system as a whole: V1 is closer to the actual plane, including the speaker within its immediate scope, and is therefore dedicated to epistemically immediate processes marked by the aorist, whereas V2 is further away from the actual plane, characterizing epistemically distant actions signaled





Frank Brisard and Adeline Patard

by the perfekt. Stanojević & Geld thus conclude that the semantic contrast between the aorist and the perfect is one of subjectification, and not necessarily of temporal reference. In the chapter by Lazhar Zanned, “Aspect as a scanning device in natural language processing: The case of Arabic”, the claim is defended that aspect is a device for constructing mental images of processes (states and events), schematically indicating how the process unfolds. In Cognitive Grammar, this is described in terms of an attention-driven scanning operation in processing time, whereby a standard of comparison and a target are related by a conceptualizer as she tracks a stative or dynamic situation, with the corresponding shifts in focal area and often leading to chains of scanning events. Scanning may occur either summarily or sequentially, depending on whether the conception builds up gradually to yield a composite structure with all components simultaneously made available, or whether an evolving scene is followed through its successive stages, resulting in a continuous series of transformations. Zanned puts forward the hypothesis that summary scanning involves no change and is thus said to characterize imperfective processes, while sequential scanning is appropriate for representing the change that is supposed to be implied in perfectives. In addition, for the latter type he introduces a distinction between different types of ‘slicing’ or cutting up a dynamic process: monosequential scanning, profiling a single sequence of the process, vs. polysequential scanning, where more than one sequence and possibly the entire process is being profiled, through chaining. Zanned concludes by pointing out the links between scanning as a general cognitive skill and other mechanisms of attention management, such as scope of awareness and the ability of zooming in. Finally, in Part III, three chapters are devoted to the relationship between the past tense and/or imperfective aspect and modality, mainly the expression of (epistemic) distance. Two of these studies, Boogaart & Trnavac and Patard, derive the value of epistemic modality from the temporal and/or aspectual contributions of tense morphemes in English, Russian and French. In contrast, Jaszczolt & Srioutai argue, for English and Thai, that a temporal interpretation may be arrived at by way of ‘cognitive default’. In their chapter, “Imperfective aspect and epistemic modality”, Ronny Boogaart & Radoslava Trnavac aim at unveiling a general connection that may exist crosslinguistically between imperfectivity and epistemicity or, more broadly, subjectivity. As a starting point, Boogaart & Trnavac present a set of contexts, mainly illustrated by data from English, in which the link between imperfective aspect and epistemic modality is clearly manifested: modal verbs, counterfactual conditionals, and counterfactual wishes. They compare these contexts with ‘perspectivized discourses’ as found in English (e.g., in (free) indirect speech), where imperfectives

Introduction

imply a perspective or point of reference from which the designated situation is viewed (and that is not the actual speaker/writer’s). Both types of context share an element of ‘subjectivity’, in that ‘information is not presented, objectively, as actually being the case in the world, but rather situated in a different ‘space’ (or ‘world’)’. Next Boogaart & Trnavac elaborate the traditional conception of aspect by adopting an anaphoric analysis of imperfectivity in the line of Kamp & Rohrer (1983) and Kamp & Reyle (1993). They assume that ‘simultaneity [of the situation E] with an independently provided reference time (E, R)’ constitutes the basic meaning of imperfective aspect. As for the perfective past, it is claimed to express the strict precedence of a situation E in relation to the point of speech. According to Boogaart & Trnavac, it is the nature of the antecedent reference point R, required by the anaphoric property of imperfectives, that determines the subjective (and epistemic) readings in the mentioned contexts. In perspectivized discourse, the reference time corresponds to a past perspective and conveys ‘character subjectivity’ (Verhagen 2005). On analogy, with epistemic modals and counterfactual wishes the reference point is an epistemic evaluation point provided by the speech time or its ‘equivalent in a counterfactual world’, bringing about ‘speaker-hearer subjectivity’ (Verhagen 2005). To conclude, Boogaart & Trnavac also examine the generality of the link between aspect and modality. In this respect, Russian is said to present some apparent counterexamples, like the modal moč (‘can/could’), the perfective present, conditional imperatives, and dative-infinitive constructions. To explain the lack of a systematic connection, in Russian narrative, between imperfective aspect and the various subjective readings previously discussed for English, it is argued that the semantics of these Russian imperfectives differs from that of their English counterparts, since the former require no anaphoric linking with a contextually provided antecedent time. Lastly, Boogaart & Trnavac underline two structural differences between both languages which further explain the contrast between Russian and English as regards the aspect-modality connection: the absence of a perfect form combining with an imperfective auxiliary, as in Germanic and Romance languages, with only a simple perfective past form in Russian to cover both perfective and perfect (subjective, epistemic) meanings, as well as the absence of sequence of tenses in Russian, with Russian retaining a (relative) present tense in contexts where English and French would have an anaphoric imperfective past form to express simultaneity in the past. Both factors effectively weaken the assumed link between imperfectivity and epistemic modality, though this does not undo the value of the authors’ unified explanation of those imperfective constructions that do exhibit an anaphoric element in their meaning configuration. In “Communicating about the past through modality in English and Thai”, Katarzyna M. Jaszczolt & Jiranthara Srioutai argue that the semantic representation of temporal expressions is best conceived of as involving a type of modality.





Frank Brisard and Adeline Patard

They support this position by offering an analysis, developed in the framework of Default Semantics, of forms expressing past-time reference in English and Thai. The discussion starts with some general considerations on the structure and properties of time as reflected in McTaggart’s (1908) distinction between the A-series and the B-series. Jaszczolt & Srioutai adopt the latter conception, assuming that the linguistic categories of time (tense, aspect) are supervenient on more basic, non-temporal categories. Hence, they argue that the indication of time in utterances boils down to expressing ‘a degree of detachment from the content of the expression uttered’, and so to a type of epistemic modality. The model of Default Semantics is founded on the idea that the meaning of an utterance mainly depends on the pragmatic enrichment of an underdetermined semantic content. Correspondingly, Default Semantics distinguishes four semantic and pragmatic sources of information that interact (‘merge’) so as to produce the meaning of an utterance: (i) word meaning and sentence structure (WS), (ii) conscious pragmatic inference (CPI), (iii) cognitive defaults (CD), and (iv) social/cultural defaults (SCD). Crucially, the third source of information, cognitive defaults, concerns a salient referential interpretation which arises as a shortcut in the inferential process and which is motivated by the cognitive property of (strong) speaker intentionality (or ‘aboutness’). This type of default meaning is considered decisive in the interpretation of temporal expressions. Jaszczolt & Srioutai apply this model to the analysis of two ways (a standard and a nonstandard one) of referring to the past in English: the narrative present (called past of narration) and the simple past. The past of narration, which is viewed as a departure from the default interpretation of the present tense, is said to be derived from conscious pragmatic inference (CPI 1, in the first stage of processing truth-conditional content, or what is said). By contrast, the past use of the simple past is seen as a default interpretation conveyed by cognitive default (CD). Finally, Jaszczolt & Srioutai apply the model of Default Semantics to the Thai modal auxiliary d1ay1II, which can receive different interpretations depending on its position (verb-initial or verb-final) and the context: circumstantial ability/opportunity or dynamic possibility, past or present reference, strong referential intentionality or not. To deal with this polysemy, previous studies treated d1ay1II as a past-tense marker in verb-initial position and as a modal marker in verb-final position. Instead, Jaszczolt & Srioutai propose a unified account of d1ay1II as a marker of dynamic modality. By default, d1ay1II triggers a referential, past interpretation (CD), but it may also give rise to a present-time reading if a pragmatic inference (CPI 1) blocks the past temporal default. Jaszczolt & Srioutai conclude that Default Semantics is a pragmatic framework that may accommodate languages like Thai, in which temporal interpretations do not rely on overt grammatical indicators.

Introduction

The following chapter by Adeline Patard, “The epistemic uses of the English simple past and the French imparfait: When temporality conveys modality”, seeks to account for elements of epistemic modality as observed in some uses of the English simple past and the French imparfait. Patard first argues for a temporal definition of both tenses, against an ‘epistemic’ (Brisard 2010) or ‘inactual’ one (Iatridou 2000). In her definition, she uses the notions of ‘reference point’, ‘topic time’, and ‘viewpoint’ (cf. Reichenbach 1947, Klein 1994, and Gosselin 1996). She describes the basic meaning of the simple past and the imparfait by means of a combination of temporal and aspectual semantic features. The traditional opposition between ‘temporal’ and ‘modal’ uses is then reinterpreted in terms of the speaker’s referential or subjective intentionality. Patard further distinguishes, for the modal usage types, between the ‘epistemic uses’, which express the speaker’s assessment of the probability of the designated situation, and the ‘illocutionary uses’, which convey the illocutionary stance of the speaker, i.e., the degree of commitment to the speech act at issue. Patard’s analysis is essentially rooted in the tradition of ‘dialogism’ (Bakhtin 1977, 1984) and its application to the description of tense uses. Dialogism is defined as an enunciative phenomenon whereby a given utterance exhibits two or more different enunciative sources, with (i) a primary enunciator E1 equated with the speaker and (ii) one or more secondary enunciator(s) whose utterance(s) also contribute(s) to the speaker’s actual utterance (cf. Bres 2001, 2003). Patard demonstrates that the simple past and the imparfait may have dialogic uses in which they evoke the presence of a secondary enunciator e1 in the past. In these dialogic uses, a past reference point, allegedly needed for the description of imperfective past tenses, comes to be identified with e1. Finally, Patard tackles two epistemic uses of the simple past and the imparfait, the counterfactual conditional use and the optative use, paying particular attention to the following points: the sense of lower probability generally associated with these uses, their non-past interpretations, the fact that imperfective morphology is ‘fake’ in these contexts (Iatridou 2000), and the paradigmatic and syntagmatic relationships with the other tenses in English and French, respectively. Taking into account contextual elements that trigger the counterfactual reading of the simple past and imparfait (conjunctions, the conditional structure [p, q], and the presence of a conditional tense in the apodosis), Patard comes up with a dialogic analysis of the conditional and optative uses. She assumes that these tense forms are used to refer to a past enunciator e1 responsible for the utterance of a proposition and, consequently, to convey an epistemic judgment about the lower probability of the situation designated by that proposition (to which the speaker, in other words, does not fully commit). Patard concludes that, in these epistemic uses of the simple past and the imparfait, modality originates in temporality (the presence of a past enunciator), and that the connection between pastness and counterfactuality is therefore enunciative in nature.





Frank Brisard and Adeline Patard

3. Conclusion The study of markers of tense and aspect in relation to notions of an epistemic nature looks like it is developing into an exciting, if not entirely new, field of research. With this volume, we wish to provide an impetus and thematic unity to this research, and suggest a number of key questions that lie at the heart of it: unification, subjectivity, and, not least, the importance of a clear and precise definition of modality. Most of all, we do not wish to convert any non-believers, neither to Cognitive Grammar nor even to the view (not shared by some of the contributors involved) that epistemic modality is the basic meaning of (all) tense-aspect constructions. However, if it contributes to establishing an awareness that modal meaning elements or usage types are directly relevant to the analysis of the grammar of time, we believe that the volume may have served its main purpose. References Abraham, Werner & Elisabeth Leiss (eds.). 2008. Modality-Aspect Interfaces: Implications and Typological Solutions. Amsterdam/Philadelphia: John Benjamins. Bakhtin, Mikhaïl. 1977. Le Marxisme et la philosophie du langage: Essai d’application de la méthode sociologique en linguistique. Paris: Minuit. Bakhtin, Mikhaïl. 1984. Les genres du discours. In Esthétique de la création verbale, 265–272. Paris: Gallimard. Barceló, Gérard Joan & Jacques Bres. 2006. Les temps de l’indicatif en français. Paris: Ophrys. Barceló, Gérard Joan, Jacques Bres, & Adeline Patard (eds.). 2006. Aspectualité, temporalité et modalité [Cahiers de Praxématique 47]. Montpellier: Université Paul-Valéry. Boogaart, Ronny & Janssen, Theo 2007. Tense and aspect. In The Oxford Handbook of Cognitive Linguistics, D. Geeraerts & H. Cuyckens (eds.), 803–828. Oxford: Oxford University Press. Bres, Jacques. 2001. Dialogisme. In Termes et concepts pour l’analyse du discours, C. Détrie, P. Siblot & B. Vérine (eds.), 83–86. Paris: Champion. Bres, Jacques. 2003. Mais oui, il était un joli temps du passé comme les autres, le petit imparfait hypocoristique. Langue française 138: 111–124. Brisard, Frank. 2001. Be going to: An exercise in grounding. Journal of Linguistics 37: 251–285. Brisard, Frank. 2002a. Grounding: The Epistemic Footing of Deixis and Grammar. Berlin: Mouton de Gruyter. Brisard, Frank. 2002b. Introduction: The epistemic basis of deixis and reference. In Grounding: The Epistemic Footing of Deixis and Reference, F. Brisard (ed.), xi–xxxiv. Berlin: Mouton de Gruyter. Brisard, Frank. 2002c. The English present. In Grounding: The Epistemic Footing of Deixis and Reference, F. Brisard (ed.), 251–297. Berlin: Mouton de Gruyter. Brisard, Frank. 2005. Epistemic interactions of tense and aspect in the English verb: The paradigm of the present. In Cognitive Linguistics: A User-Friendly Approach, K. Turewicz (ed.), 65–82. Szczecin: University of Szczecin Press.

Introduction

Brisard, Frank. 2010. Aspects of virtuality in the meaning of the French imparfait. Linguistics 48: 487–524. Bybee, Joan, Revere Perkins & William Pagliuca. 1994. The Evolution of Grammar: Tense, Aspect and Modality in the Languages of the World. Chicago/London: University of Chicago Press. Chung, Sandra & Alan Timberlake. 1985. Tense, aspect and mood. In Language Typology and Syntactic Description, vol. 3: Grammatical Categories and the Lexicon, T. Shopen (ed.), 202– 258. Cambridge: Cambridge University Press. Condoravdi, Cleo. 2002. Temporal interpretation of modals: Modals for the present and for the past. In The Construction of Meaning, D. Beaver, S. Kaufmann, B. Clark & L. Casillas (eds.), 59–88. Stanford: CSLI Publications. Croft, William. In preparation. Verbs: Aspect and Argument Structure. Oxford: Oxford University Press. Cutrer, Michelle L. 1994. Time and Tense in Narrative and Everyday Language. PhD Dissertation, University of California at San Diego, United States. Dahl, Östen. 1985. Tense and Aspect Systems. Oxford: Blackwell. Dahl, Östen. 1997. The relation between past time reference and counterfactuality: A new look. In On Conditionals Again, A. Athanasiadou & R. Dirven (eds.), 97–114. Amsterdam: John Benjamins. Declerck, Renaat. 2005. The relation between temporal and modal uses of indicative verb forms. In Crosslinguistic Views on Tense, Aspect and Modality. [Cahiers Chronos 13], B. Hollebrandse, A. van Hout & C. Vet (eds.), 215–227.Amsterdam: Rodopi. DeLancey, Scott. 1997. Mirativity: The grammatical marking of unexpected information. Linguistic Typology 1: 33–52. DeLancey, Scott. 2001. The mirative and evidentiality. Journal of Pragmatics 33: 369–382. Doiz-Bienzobas, Aintzane. 1995. The Preterit and the Imperfect in Spanish: Past Situation vs. Past Viewpoint. PhD Dissertation, University of California at San Diego, United States. Dowty, David. 1977. Towards a semantic analysis of verb aspect and the English ‘imperfective progressive’. Linguistics and Philosophy 9: 37–61. Fleischman, Suzanne. 1989. Temporal distance: A basic linguistic metaphor. Studies in Language 13: 1–50. Fleischman, Suzanne. 1995. Imperfective and irrealis. In Modality in Grammar and Discourse, J. Bybee & S. Fleischman (eds.), 519–551. Amsterdam: John Benjamins. Givón, Talmy. 2001. Syntax: An Introduction, vol. 1. Amsterdam: John Benjamins. Gosselin, Laurent. 1996. Sémantique de la temporalité en français: Un modèle calculatoire et cognitif du temps et de l’aspect. Louvain-la-neuve: Duculot. Gosselin, Laurent. 2005. Temporalité et modalité. Brussels: Duculot. Hogeweg, Lotte. 2009. What’s So Unreal about the Past? In Tsangalidis, Anastasios and Roberta Facchinetti (eds.), Studies on English Modality – in honour of Frank R. Palmer, 181–208. Bern: Peter Lang. Hornstein, Norbert. 1990. As Time Goes by: Tense and Universal Grammar. Cambridge, MA: MIT Press. Iatridou, Sabine. 2000. The grammatical ingredients of counterfactuality. Linguistic Inquiry 31: 231–270. James, Deborah. 1982. Past tense and the hypothetical: A cross-linguistic study. Studies in Language 6: 375–403.





Frank Brisard and Adeline Patard Jaszczolt, Katarzyna M. 2009. Representing Time: An Essay on Temporality as Modality. Oxford: Oxford University Press. Kamp, Hans & Uwe Reyle. 1993. From Discourse to Logic. Dordrecht: Kluwer. Kamp, Hans & Christian Rohrer. 1983. Tense in texts. In Meaning, Use and Interpretation, R. Bäuerle, C. Schwarze & A. von Stechow (eds.), 250–269. Berlin: Mouton de Gruyter Klein, Wolfgang. 1994. Time in Language. London: Routledge. Langacker, Ronald W. 1991. Foundations of Cognitive Grammar, vol. 2: Descriptive Application. Berlin: Mouton de Gruyter. Langacker, Ronald W. 1997. Consciousness, construal, and subjectivity. In Language Structure, Discourse and the Access to Consciousness, M. I. Stamenov (ed.), 49–75. Amsterdam: John Benjamins. Langacker, Ronald W. 2001. The English present tense. English Language and Linguistics 5: 251–271. Langacker, Ronald W. 2002. Remarks on the English grounding systems. In Grounding: The Epistemic Footing of Deixis and Reference, F. Brisard (ed.), 29–38. Berlin: Mouton de Gruyter. Langacker, Ronald W. 2003. Extreme subjectification: English tense and modals. In Motivation in Language, H. Cuyckens, T. Berg, R. Dirven & K.-U. Panther (eds.), 3–24. Amsterdam: John Benjamins. Larreya, Paul. 2003. Irrealis, past time reference and modality. In Modality in Contemporary English, R. Facchinetti, M. G. Krug & F. R. Palmer (eds.), 21–45. Berlin: Mouton de Gruyter. Ludlow, Peter. In preparation. Tensism and Presentism. Oxford: Oxford University Press. Martin, Robert. 1991. Types de procès et systèmes hypothétiques: De l’aspect ‘de re’ à l’aspect ‘de dicto’. Travaux de linguistique et de philologie 22: 87–95. McTaggart, John E. 1908. The unreality of time. Mind 17. Reprinted in Philosophical Studies, J. E. McTaggart (1934), 110–131. London: Edward Arnold. Mellet, Sylvie. 1990. Temps et mode en latin: À propos de l’imparfait. Bulletin de la Société de Linguistique de Paris XXXV: 161–171. Michaelis, Laura. A. 2004. Type shifting in Construction Grammar: An integrated approach to aspectual coercion. Cognitive Linguistics 15: 1–67. Mortelmans, Tanja. 2002. Wieso sollte ich dich küssen, du hässlicher Mensch! A study of the German modals sollen and müssen as “grounding predications” in interrogatives. In Grounding: The Epistemic Footing of Deixis and Reference, F. Brisard (ed.), 391–432. Berlin: Mouton de Gruyter. Mortelmans, Tanja. 2006. Langacker’s subjectification and grounding: A more gradual view. In Subjectification: Various Paths to Subjectivity, A. Athanasiadou, C. Canakis & B. Cornillie (eds.), 151–176. Berlin: Mouton de Gruyter. Nichols, Johanna. 1986. The bottom line: Chinese Pidgin Russian. In Evidentiality: The Linguistic Encoding of Epistemology, W. Chafe & J. Nichols (eds.), 239–257. Norwood, NJ: Ablex Publishing. Patard, Adeline. 2007. L’un et le multiple. L’imparfait de l’indicatif en français: Valeur en langue et usages en discours. PhD Dissertation, Paul-Valéry University – Montpellier 3: Montpellier, France. Portner, Paul. 1998. The progressive in modal semantics. Language 74: 760–787. Reichenbach, Hans. 1947. Elements of Symbolic Logic. New York: Free Press.

Introduction  Smirnova, Elena. 2006. Die Entwicklung der Konstruktion würde + Infinitiv im Deutschen. Berlin: Mouton de Gruyter. Stowell, Tim. 2004. Tense and modals. In The Syntax of Time, J. Guéron & J. Lecarme (eds.), 621–636. Cambridge, MA: MIT Press. Trnavac, Radoslava. 2006. Aspect and Subjectivity in Modal Constructions. PhD Dissertation, Leiden University, The Netherlands. Verhagen, Arie. 2005. Constructions of Intersubjectivity: Discourse, Syntax and Cognition. Oxford: Oxford University Press.

part i

Theoretical foundations

The definition of modality Renaat Declerck

University of Leuven This chapter offers a definition of modality that is as concrete and complete as possible. Modality means that there is reference to actualization of a situation in a world that is not represented as being the factual world. All types of modality are pigeonholed, regardless of whether the ‘modalizer’ is an auxiliary, lexical verb, adverb, conditional clause, a morphological operation like ‘modal backshifting’ or ‘conditionalization’, etc. Clear definitions are given of root and epistemic modality, and it is shown that not all epistemic values are modal values. On the basis of a rich modal world typology and a number of possible relations between a modal world and the factual world, different subtypes of modality are distinguished. Keywords: modality, epistemic, root modality, condition

1. Introduction1 The entry for ‘modality’ in Leech’s (2006: 64) Glossary of English Grammar reads: “see modal (auxiliary verb)”. In the entry ‘modal (auxiliary) (verb)’ we read: “a member of a small class of verbs that have meanings relating to modality, that is to such concepts as possibility or permission (can, may), obligation, necessity or likelihood (must, should), prediction, intention or hypothesis (will, would)”. This definition of modality reduces modality to the use of modal auxiliaries and gives us little tangible to go by. This is not surprising, because modality is a most elusive concept, which in pedagogical grammars is usually illustrated (rather than defined) by a list of possible meanings that ‘modal auxiliaries’ can have. Of course, more detailed linguistic treatments of the phenomenon of modality often do offer

1. I wish to thank Ilse Depraetere, Susan Reed, Bert Cappelle and An Verhulst for the fruitful discussions we have had in connection with modality. Special thanks are due to Susan Reed and Caroline Gevaert for their comments on a previous draft of this manuscript.



Renaat Declerck

a kind of definition, but there does not seem to be a standard definition that is widely accepted. The issue is also clouded by the fact that the class of ‘modal auxiliaries’ is usually treated as a waste-paper basket in which all the verbs are put which have the formal and syntactic characteristics of auxiliaries but are clearly not auxiliaries used to build tense forms, aspectual forms or passive verb forms. At least two of the auxiliaries that are often (even in major recent grammars) included in the set of modal auxiliaries, viz. used to and dare, arguably do not belong to that class.2 The meaning expressed by used to is aspectual, not modal: My late uncle used to go to bed early predicates of my uncle the past characteristic (habit) of going to bed early.3 I daren’t do that has exactly the same nonmodal meaning as I don’t dare to do that. Since to dare to is a lexical verb without a modal meaning, the auxiliary dare in I daren’t do that should be treated as what we could call a ‘lexical auxiliary’ – something which to my knowledge nobody has ever suggested doing. A lexical auxiliary combines the morpho-syntactic characteristics of an auxiliary with the lexical meaning of a full verb (rather than with the meaning of an auxiliary expressing tense, voice, modality or aspect). In this article, in which we aim to give a satisfactory definition of ‘modality’, we have to find (among other things) what all the (true) modal auxiliaries have in common, apart from their specific meanings, which mostly have to do with some kind of possibility or necessity. It is this common core that constitutes the essence of modality and thus offers a basis for a definition of modality. As we will see, that common element of meaning is that a situation is represented as actualizing in a nonfactual world. 2. Terminological preliminaries a. We use ‘situation’ as a cover term for the various possible types of contents of clauses, i.e. as a cover term for anything that can be expressed in a clause. According to Lyons (1977), a situation is either a state, an action, a process (=change, development) or an event (=a nonagentive dynamic situation, e.g. a fall). As a cover-term for the various verbs that are typically associated with one of these kinds of situation we use the term actualize. The sentence The 2. Quirk et al. (1985: 137) treat used to and dare as “marginal modals”. Biber et al. (1999: 484) call them “marginal auxiliary verbs”, adding that they “can behave like modals in taking auxiliary negation and yes-no question inversion”. 3. Huddleston & Pullum (2002) recognize that the meaning of used to is “aspectual, not modal” (p. 115) but they still treat the auxiliary dare as a modal auxiliary (p. 110).

The definition of modality 

situation is actualizing can thus be said of a state that is holding, an action that is being performed, a change that is taking place or an event that is happening. In all four cases the sentence refers to the actualization of a situation. b. A ‘possible world’ – we are using the term in its linguistic meaning rather than in the way it is used in logic – is always a t-world, i.e. a world which is anchored to a given time t. This means that a tensed proposition can be true at one time but false at another, in other words, that it may be true of one t-world but false of another. Thus, J.F. Kennedy is the President of the U.S. is true of any world holding at (=anchored to) some time in the course of 1961 but is false of the objective S-world, i.e. the actual world holding at S (=speech time) – see section (d) below. By contrast, omnitemporal situations (referred to by generic or universal sentences like A horse is an animal) actualize in every objective world holding at any possible time. The unmarked form of t-anchoring is S-anchoring (where S means ‘speech time’). If the world referred to is an S-world, i.e. a world that is anchored to S, and the sentence referring to it is in the present tense, no anchor time needs to be specified in the sentence, nor in its context. This explains why, unlike The weather was nice (which begs the question “When?”), The weather is nice is fully interpretable in isolation: the hearer assumes that the speaker represents the situation as actualizing at S. It also explains why omnitemporal situations are as a rule referred to in the present tense: this follows naturally from the fact that S is one of the times at which the situation is factual and is the unmarked anchor time in the linguistic mind of the speaker. c. There is no essential difference between saying that a proposition is “true of ” (Lyons 1977: 687) a particular t-world and saying that the situation denoted by the proposition ‘actualizes in’ that t-world. Thus, if the tensed proposition John is walking home is true of (=true with reference to) the objective S-world (=the actual world holding at S), the situation referred to is represented as actualizing in that world. If the untensed proposition ‘John be walking home’ is not true of the objective S-world, as in [I wish] John was walking home, the situation does not actualize in the objective S-world, i.e. it does not belong to the actualizations making up the objective S-world. In this particular case John is walking home is true of a counterfactual S-world (i.e. a world which is just the reverse of the factual world), which means that the situation actualizes in that counterfactual S-world. d. We can distinguish between ‘objective’ and ‘subjective’ t-worlds. An objective t-world is the unique real world that holds at a given time and which is judged real by an (imaginary) ideal outside observer viewing the world as it is at that given time. A subjective t-world is an alternative world which is not judged real



Renaat Declerck

by such an ideal outside observer but which is conceived of as real by some consciousness at a certain time. Such a t-world consists of the tensed (=anchored in time by their finite verb form) propositions that are deemed true by the world-creating consciousness at the given time. Thus, the situation referred to by Amsterdam lies in Belgium (which is counterfactual in the objective S-world) actualizes in the counterfactual S-world existing in the mind of a speaker who is convinced that this assertion is true at S. For simplicity, we disregard subjective t-worlds in the discussion below. e. Another distinction we can make is that between ‘narrow t-worlds’ and ‘extended t-worlds’. A narrow t-world is a t-world comprising all the situations that are actualizing at a given time t. Tensed propositions can only be true of such a world if the tense represents the (actualization of the) situation referred to as simultaneous with t. An extended t-world is a world comprising all the situations that are actualizing at t or have actualized before t. Various tenses can be used to represent the actualization of a situation as factual in an extended t-world. Thus, Caesar was Emperor of Rome, There have been many battles between the French and the English and Paris is the capital of France all refer to actualizations that are factual in the speaker’s extended S-world. Since an extended ti-world includes a narrow ti-world, any actualization that is factual in an objective narrow t-world is also factual in the corresponding objective extended t-world. Thus if John is in London now, the tensed proposition John is in London is true not only of the narrow S-world but also of the extended S-world. For the sake of simplicity, we ignore the distinction between narrow and extended t-worlds whenever it is not crucial. As regards S-worlds, we use ‘factual world’ in the sense of ‘extended factual S-world’. f. Whether a sentence is positive or negative is irrelevant to the question whether the actualization of the situation is factual or not in a given t-world, or to the question whether the sentence (=tensed proposition) is true or not of that t-world. Thus both London lies on the Thames and London does not lie on the Rhine represent the (positive or negative) situation referred to as actualizing in the factual world, and both sentences are true of the factual world. It follows that we must distinguish between ‘negative sentences’ and ‘counterfactual sentences’, as well as between ‘negative situations’ and ‘counterfactual situations’. London does not lie on the Rhine is a negative sentence, referring to the actualization of a negative situation (i.e. a situation referred to in a negative sentence). That negative situation is represented as actualizing in the factual world. By contrast, If I were you... refers to a situation that actualizes in a counterfactual

The definition of modality 

S-world. This clause refers to a counterfactual situation (=a situation that does not actualize in the factual world because it is incompatible with this world). g. There are many different kinds of t-world, such as worlds that are factual, ‘theoretical’ (see below) or counterfactual at t. Situations that are located in such worlds are represented as actualizing in them. Thus, the situations referred to by John is ill, If only John was ill! and John is perhaps ill are each time represented as actualizing in a given S-world, but that S-world is, respectively, the factual world, a counterfactual S-world (=an S-world which is conceived of by the speaker as being incompatible with the factual world) and a ‘theoretical’ S-world which is represented as possibly coinciding with the factual world. If there exist several ti-worlds (=worlds anchored to the same time ti), the situation referred to can only actualize in one of these ti-worlds. Thus, if it is indeed the case that John is ill at S, the sentence John is ill represents the situation of John being ill as actualizing in the factual world. In If only John {was/were} ill!, the situation denoted by ‘John be ill’ is represented as actualizing in the counterfactual S-world evoked by If only (=I wish...). And in John is perhaps ill, the same situation is represented as actualizing in the theoretical S-world evoked by the modal adverb perhaps (=It is possible that...). h. A situation represented as actualizing in a nonfactual world is automatically nonfactual in the factual world. Thus, the situation referred to by If only John was ill! is represented as actualizing in the counterfactual world created by if only, and hence as having the factuality value [+counterfactual] (rather than [+factual]) in the factual world. Similarly, John is perhaps ill refers to a theoretical S-world (created by perhaps) in which the situation of John being ill actualizes. The factuality value of this actualization in the factual world is ‘possibly factual’ (which, as we will see in Section 6.2, is an instance of a ‘relative’ factuality value.) ‘Possibly factual’ is a kind of nonfactuality. i. All nonfactual worlds (i.e. possible worlds that are not represented and/or interpreted as being the factual world) are created by some kind of ‘nonfactualworld creating device’. Such a device can be, for example, an intensional verb like believe, an attitudinal verb like intend, a modal auxiliary, a conditional clause, a modal adverb like perhaps, etc. – see also Section 4(a). Since a nonfactual world is by definition a modal world (and vice versa – see Section 3), we call the device in question the modalizer. A proposition underlying a clause containing a modalizer is a modalized proposition. Thus, in Suppose John were here, the modalizer is suppose and the modalized proposition is ‘John be here’. In Bill must leave now, the modalizer is the modal auxiliary must and the modalized proposition is ‘Bill leave now’.



Renaat Declerck

Huddleston (1984: 167–8) suggests the label ‘residue’ for what is left of a clause with a modal auxiliary when the auxiliary is abstracted from it, viz. the subject plus the (nontensed and nonmodal) proposition that has the form of an infinitival clause. Thus, in Tom must swim, ‘Tom swim’ is the residue proposition or simply ‘residue’. We adopt this terminology. The situation denoted by the residue proposition we call the residue situation. In John may be ill, the residue situation is the state denoted by the proposition ‘John be ill’. The speaker says that that state is actualizing in a nonfactual world that possibly coincides with the factual world, which comes down to saying that the state in question is possibly actualizing in the factual world. The term ‘residue’ corresponds to what we have called the ‘modalized proposition’, but it is a handier term because it is easier to speak of the ‘residue situation’ than of ‘the situation denoted by the modalized proposition’. We therefore adopt the terms ‘residue’, ‘residue situation’ and ‘residue proposition’. Moreover, we extend the use of these terms to those cases in which the modalizer is not a modal auxiliary. In I wish John was here, the verb wish is the modalizer and the proposition ‘John be here’ is the residue proposition. j. The referent of a modalizer is a modal state, i.e. the state of a situation being necessary, possible, permitted, wished for, etc., or the state of it being possible, likely, necessary, etc. that a proposition is true in a given world.

(1) That rumor may be true. (The modal auxiliary asserts the existence of the possibility that the rumor is true. The existence of something is a state. The modalizer “may” thus refers to a modal state.) (2) You must leave now. (The modal auxiliary expresses that it is obligatory for the addressee to leave now. The existence of an obligation is a modal state.) (3) [Don’t go through the woods.] You could get lost. (The modal auxiliary “could” expresses the state of it being possible for the addressee to get lost.)

Like any state, a modal state is a situation actualizing in a particular world. In the above examples, the world in question is the factual world, i.e. the extended S-world – see Section 2(e). When the modal state is thus represented as actualizing in the factual world, we can say that the modal world is grounded in the factual world. This is usually the case, although there are also instances in which one modal world is grounded in another. Thus, in I might have been killed if I had not given in the modal suppositional world (created by if) is grounded in the modal epistemic world (created by might) comprising theoretically possible actualizations: the sentence means ‘It is possible that I would have been killed if I had not given in’. Note that this example shows that grounding is a different phenomenon from anchoring, which is a question of the temporal relation between two worlds

The definition of modality 

– see Section 2(b) above – although the time to which a grounded t-world is anchored is usually also the time to which the grounding t-world is anchored. In the above example, ‘It is possible that’ is anchored to S, while ‘I would have been killed if I had not given in’ is anchored to a t earlier than S. Both t-worlds, however, are anchored to the ‘extended S-world’ – see Section 2(e) – which we have agreed to refer to as the ‘factual world’. k. It should be stressed that by ‘nonfactual world’ we do not mean a world that is necessarily different from the factual world but rather a possible world that is not represented and/or interpreted as being the factual world. This is clear from John may be here, which refers to a nonfactual world in which the state of John being here actualizes and which may or may not coincide with the factual world. A nonfactual world could not be represented as possibly coinciding with the factual world if ‘nonfactual world’ were defined as ‘world that is necessarily different from the factual world’. 3. The definition of modality a. Modality can be defined as the phenomenon that a situation is located in a nonfactual world (as defined above). This means that the residue situation is represented as actualizing in a nonfactual world, in other words that the proposition underlying the sentence is ‘true of ’ a modal world. The nonfactual world in question can be related to (grounded in) the factual world in one of several ways. For example:

(4) [For all we know,] John may be dead. (“May” is a modalizer that refers to a present state: “It is possible...”. It creates a nonfactual world in which the state of John being dead is actualizing. This nonfactual world is related to the factual world in terms of the ‘factuality value’ (see Section 6.1.) ‘possibly factual’: the world in which John is dead possibly coincides with the factual world. Hence the interpretation ‘It is possible that John is dead’.) (5) Val must leave town as soon as possible. (“Must” is a modalizer that refers to a present state. It creates a nonfactual world in which the situation of Val leaving town actualizes. This nonfactual world is interpreted as a future world because of “as soon as possible”. This modal world is not related to the factual world in terms of a ‘factuality value’, because “must” here expresses root (nonepistemic) modality – see Section 5(a). But there is a relation between the two worlds because the sentence is interpreted as ‘Actualization of Val leaving town in a future nonfactual world is obligatory’. The modal state thus concerns the obligation of the actualization of



Renaat Declerck

the residue situation in a nonfactual (more specifically, not-yet-factual) modal world. In other words, the relation between the two worlds is that in the factual world the existence of the nonfactual world in which Val leaves town is deemed necessary.) b. What has been given in the previous subsection is the definition of modality in a nutshell. In the following sections further elements are introduced, which gradually render the definition more detailed and concrete. 4. Kinds of modalizers a. As we have seen, any linguistic device creating a modal world is a modalizer. In English, we can distinguish the following types of modalizer: a modal auxiliary (e.g. can, must, may...), a modal adverb (e.g. perhaps, possibly, duly, obligatorily), an intensional verb like believe, suppose, imagine, an attitudinal verb like intend, want, hope, wish, the subjunctive mood, the imperative mood, a conditional clause creating a ‘theoretical world’ (see Section 4(d)), a tense auxiliary creating a future world (e.g. will, be going to, be about to) or expressing posteriority, an inserted comment clause with an intensional verb (e.g. I think), ‘modal backshifting’ (see Section 4(b) below) or ‘modal conditionalization’ (see Section 4(c) below) or a combination of the latter two. Some of these modalizing devices are well-known, others deserve some comment. The latter are treated in the next subsections. b. ‘Modal backshifting’ (or ‘formal distancing’) is a basic mechanism producing either a tentative or a counterfactual interpretation. It is important to see that this is not the same type of ‘backshifting’ (Jespersen 1931: 151) as occurs in indirect speech, for example when I am ill is reported as She said she was ill. In modal backshifting the conditional tense can be replaced by the conditional perfect, e.g. I would be happy if she came → I would have been happy if she had come. This kind of ‘backshift’ is never possible in indirect speech, where would and would have remain unchanged: I would help him if I could → She said she would help him if she could. Modal backshifting can produce various results. In some cases only the verb is a backshifted form. For example, It is time we leave can be modally backshifted to It is time we left to stress the not-yet-factuality of our leaving. In other cases only the infinitival clause representing the residue proposition is a backshifted form, as in He was to have been in his office. The result of the modal backshifting here is that the actualization of the residue situation is interpreted as counterfactual. Henry was to have been in his office says that there was a past obligation (resulting from

The definition of modality 

some kind of arrangement or rule) for Henry to be in his office at a given time, but that he was not there at that time. The existence of the obligation itself is not counterfactual. Another example is the following:

(6) The decision should have been made tomorrow [but now we understand that there is now another week’s grace]. (Google UK)

In this example should be made is modally backshifted to should have been made. This does not affect the existence of the obligation but renders the representation of the actualization of the residue situation counterfactual. In some cases the result of modal backshifting is that both the infinitive and the preceding verb are backshifted forms. Thus, He may be absent, which expresses present epistemic possibility, can be modally backshifted to He might have been absent, which represents the (past, present or future) actualization of the residue situation as counterfactual. Similarly, in I {could/might} have been in Iraq now, both the modal auxiliary and the infinitive are backshifted to create a counterfactual interpretation. The sentence is interpreted something like ‘It would have been possible for me to be in Iraq now, (but I am not)’. In this case, both the modal state (root possibility) and the residue situation are represented as counterfactual. The following are further examples in which both the verb and the verb of the residue are backshifted forms. (It should be noted, however, that not all speakers of British English find such instances acceptable.)

(7) I hoped to sit beside her. → I had hoped to sit beside her. → I had hoped to have sat beside her. (8) I would like to see her tonight. → I would have liked to see her tonight. → I would have liked to have seen her tonight.

There is also the mechanism of double modal backshifting. This means that the same verb form undergoes modal backshifting twice. Examples of this are largely restricted to a highly informal register and may be typical of a particular regional variety of English. For instance:

(9) If he had a car, he would collect us. → If he’d had a car, he’d have collected us. → If he had’ve had a car, he’d have collected us.

c. Another modalizing device is modal conditionalization. This means that the present tense form of a verb is replaced by the corresponding conditional tense form without there being an overt or implicit condition: (10) [What a nice view!] We might be in Switzerland! (This is interpreted as ‘It would be possible for us to be in Switzerland’, hence ‘It would seem as if we were in Switzerland’. This is the conditionalized version of We may be in



Renaat Declerck

(11) (12) (13) (14)

Switzerland (on the interpretation ‘It is possible for us to be in Switzerland’, i.e. ‘The possibility exists that we are in Switzerland’).) [“John is very happy with his new girl-friend.”] – “It {seems/would seem} so”. (The addition of would renders the reply more tentative or doubtful.) I {wish/should wish} to voice a different opinion. (id.) As a first step to the new life you desperately need, I {suggest/would suggest} assertion training. (id.) I {think/should have thought} he could have left something for us, at least. (Should have thought is an example of double modal conditionalization. It stresses the counterfactuality of the actualization of the residue situation and in doing so changes the sentence from a neutral statement into an expression of criticism.)

d. Another modalizing device is the use of a conditional clause evoking a ‘nonneutral theoretical world’. In Declerck & Reed (2001), conditionals with such a clause are called ‘nonneutral theoretical-P conditionals’. This is because ‘conditional clause’ and ‘head clause’ are abbreviated to ‘P-clause’ and ‘Q-clause’, respectively, given that the standard logical representation of a conditional sentence is ‘if P, (then) Q’. Declerck and Reed (2001) propose the ‘possible-world typology of conditionals’ that is represented in Figure 1. In this typology, P-propositions are either factual or theoretical. The theoretical-P conditionals are subclassified into ‘neutral theoretical-P’ and ‘nonneutral theoretical-P’ conditionals. In the former, the P-proposition represents a mere supposition, while in the latter the supposition is accompanied by a presupposition (assumption) concerning the relation between the theoretical P-world and the factual world. The assumed relation may be that of identity (=closed condition), possible identity (=open condition), likely nonidentity (=tentative condition) or definite nonidentity (=counterfactual condition).

Possible P-worlds Factual

Theoretical (nonfactual) Neutral theoretical Closed

Figure 1.

Nonneutral theoretical Open Tentative

Counterfactual

The definition of modality

In factual-P conditionals, the world referred to in the P-clause is the factual world. This means that factual P-clauses are not modalizers. The following exemplify the type: (15) I (always) avoided her if I could. (16) [Yet according to the government the number of homeless people in the UK has now reached 300,000, over 50% of these are under 25 and this figure is continuing to grow at an alarming rate.] If you add to these those who are at risk or who have been considered as ‘no hopers’, the number of people who need our help is truly of crisis proportions. (Cobuild corpus) A conditional whose P-clause refers to a theoretical P-world – a theoretical-P conditional – may refer to that theoretical world without specifying the relation between the suppositional world and the factual world. In this case we speak of a neutral theoretical conditional clause. A speaker who uses a neutral-P conditional does not convey a specific assumption (presupposition) about the likelihood that the theoretical-P world coincides with the factual world. The following are some examples: (17) If a woman has a history of cancer in her family, she should have herself checked at least once a year. (Because a woman has nonspecific reference, the P-clause refers to a set of people in general. Though the speaker naturally assumes that there are people belonging to this set, he does not assume (nor denies, for that matter) that the hearer (or any other specific person) forms part of the set. There is simply no presupposition that the application of P to a specific person is true, false, unlikely to be true, or a real possibility.) (18) Emergency lighting should be fitted in the hall, if there is no glass in the front door to let in light. (Cobuild corpus) (19) If a shop’s good enough, it’s worth making a little effort to get to. (Cobuild corpus) A nonneutral theoretical P-clause evokes a suppositional world which is epistemic because it stands in a given factuality relation to the factual world. The factuality values can be of four different kinds, which leads to a distinction between closed-P, open-P, tentative-P and counterfactual-P conditionals. In these four types, different factuality relations are assumed to hold between the theoretical P-world and the factual world. To begin with, a P-world is ‘closed’ if it is assumed (or ostensibly assumed) to coincide with the factual world: (20) [“The picture you are now looking at is a Van Gogh.”] – “Well, if this is a Van Gogh, I’m rather disappointed by it.”





Renaat Declerck

(21) If I’m going to catch the 5.12 train, I’d better be on my way now. (The P-clause presupposes that I want to catch that train.) (22) If (as I am told now) he will come here himself tomorrow, I no longer need to try and get into contact with him today. In an ‘open-P conditional’, the theoretical world called up by the P-clause is assumed to be such that it may or may not turn out to correspond with the factual world or the expected future extension of the factual world. It follows that the Pclause often refers to the future: since the future is unknown, the P-situation and the Q-situation are easily viewed as an ‘open possibility’ (i.e. as neither certain, probable nor unlikely). (23) If the train is late, we will miss our connection in London. (24) [I’ve been told that] it will be worth a lot more if we can get his signature on it. (Cobuild corpus) (25) [I don’t know if Liverpool won their match yesterday.] If they did, they must be top of the League now. The basic meaning of a ‘tentative-P conditional’ is that the theoretical P-world is assumed to be such that it might correspond with the factual world, but that this possibility is considered as rather unlikely (but not impossible) to be (or to have been or to come) true. (26) If he {did/were to do} that, he would be in real trouble. (tentative-P: the Pactualization is contrary to expectation) (27) I would be surprised if that man over there were the man we’re looking for. Finally, in a ‘counterfactual-P conditional’, the theoretical P-world is assumed to be contrary to fact, i.e. to be definitely different from the factual world (or, if the reference is to the future, from what is going to become the factual world). Thus, in I would have been happy if she had come, the P-clause is presupposed to be true of a counterfactual theoretical P-world but false of the factual world. In sum, a P-clause is a modalizer creating an epistemic nonfactual world provided the condition is represented and interpreted as closed, open, tentative or counterfactual. e. Some nonfactual t-worlds can be characterized in terms of a modal concept that has been neglected in the literature on modality, viz. the idea that the tworld in question is envisaged by the speaker but not yet factual at the time t to which the world in question is anchored. This kind of modal world, which we call ‘not-yet-factual at t’, is evoked by any expression that has posteriority as part of its meaning. The clearest cases are those in which the reference is exclusively to a future world. Thus, John will trim the hedge evokes a t-world

The definition of modality 

which is nonfactual in the specific sense that it is not-yet-factual at S but is predicted (at S) to become factual at some future time t. The proposition ‘John trim the hedge’ is true of that not-yet-factual t-world (in which actualization is envisaged by the speaker), but the not-yet-factual t-world itself is neither factual nor counterfactual at S: it is an S-world which is partly counterfactual (because there is no actualization at S) and partly theoretical (because actualization at a time later than S is envisaged at S) – see Declerck (2009) for a more detailed analysis.4,5 Another type of example in which a situation is located in a not-yet-factual world is I saw Sam before she had seen me. Here the situation of the before-clause is represented as not-yet-factual at the time of the head clause situation. The sentence is therefore interpreted as ‘I saw Sam at a time when she had not yet seen me’, or ‘When I saw Sam, it was not yet a fact that she had seen me’ (which leaves it vague whether she ever saw me or not). In It’s high time they left and I’d rather they left ‘modal backshifting’ (i.e. the use of left rather than leave – see Section 4(b)) locates the actualization of the guests’ leaving in a world that is not-yet-factual at S: though the situation has not actualized yet, its actualization is envisaged for the future. In I intended to call Bill up when he was at home, both the situation of my calling up Bill and the when-clause situation are interpreted as not-yet-factual at the time when my intention was a past fact. 5. Epistemic modality a. We speak of epistemic modality when the degree of compatibility (or overlap) between the modal world and the factual world is at stake. Epistemic utterances express the speaker’s evaluation of the relation between the modal world in which the residue situation actualizes and the factual world. Thus, Bill may be stuck in a traffic jam expresses that the situation of Bill being stuck in a traffic jam is possibly actualizing in the factual world. 4. Envisage actually refers to something weaker than prediction. It just means that the speaker reckons with the possibility that a situation will actualize. 5. One might object that Rita will do it tomorrow does not rule out that Rita is already doing it at S, as witnessed by the fact that we can say Rita will still be doing it tomorrow. However, the sentences Rita will do it tomorrow, Rita is doing it and Rita will still be doing it tomorrow all refer to different situations. Doing something tomorrow is not the same thing as doing it now or as still be doing it tomorrow. In Rita will do it tomorrow, the situation referred to is located, not in the factual S-world, but in a world that is expected to exist tomorrow. In other words, the situation is counterfactual in the factual S-world, but at the same time envisaged as possible in a future world.



Renaat Declerck

By contrast, utterances that are interpreted in terms of root modality (or nonepistemic modality) are concerned with factors that determine the actualization of the residue situation, as in the following examples: (28) You may leave tomorrow. (The speaker is granting permission.) (29) You {can/cannot} buy tickets in the office over there. (The addressee’s ability or inability to buy tickets at a certain place depends on a circumstantial factor.) (30) The fugitives must leave the country because their visas expired last week. (The expiration of the visas makes it necessary for the fugitives to leave the country, because it is illegal to remain in the country without a valid visa.) In sum, we can say that an epistemic modal world stands in a particular relation to the factual world or to a future extension of it: there is an epistemic relation between the two of them. This relation concerns the chances that the epistemic world coincides with, or will coincide with, the factual world or its future extension. By contrast, as a rule – see, however, Section 7(b) – a root modal world is not epistemically related to the factual world or to its future extension. It is an ‘epistemically dangling’ world. Thus, The car can be repaired says nothing about the chances that the car will be repaired. b. Within epistemic modality, two major possibilities must be distinguished. In most cases there is specified epistemic modality. This means that the epistemic relation between the modal world and the factual world is specified. For example, the modal world created by may in Dave may be lying is related to the factual world in terms of epistemic possibility: the residue proposition has the factuality value ‘possibly factual’. This means that the situation of Dave being lying is possibly actualizing in the factual world, in other words that the modal world (in which the situation actualizes) possibly coincides with the factual world. When there is specified epistemic modality, one of various ‘factuality values’ on the ‘epistemic scale’ – see Section 6 – is applicable. Specified epistemic modality can be indicated by epistemic modal auxiliaries, epistemic modal adverbs, intensional verbs (believe, imagine...), comment clauses with an intensional verb (e.g. I think), expressions of posteriority (e.g. before in He died before I had seen him), modal backshifting, modal conditionalization and ‘nonneutral theoretical’ conditional clauses. The second type of epistemic modality is indeterminate epistemic modality. In this case the question of the epistemic status of the nonfactual world (i.e. its factuality value vis-à-vis the factual world) is not specified (though it is not irrelevant). Indeterminate epistemic modality is found in conditionals with a ‘neutral

The definition of modality 

theoretical’ conditional clause – see Declerck & Reed (2001: 72–80). The following are some illustrations: (31) If a woman has a history of cancer in her family, she should have herself checked at least once a year. (subject NP with nonspecific reference) (32) If water boils, it changes into steam. (general truth) (33) If she has a problem, she goes to her grandmother. (permanent habit) (34) Children are orphans if their parents are dead. (set-identifying conditional clause) (35) If he noticed Brand, he didn’t comment. (Cobuild corpus) (‘anchoring conditional clause’, i.e. the function of the conditional clause is to anchor the head clause into the ongoing discourse) In none of these examples is there an assumption or presupposition on the part of the speaker that the supposition is treated as true, untrue, unlikely to be true, or as a real possibility. There is no specified epistemic relation (=relation in terms of a factuality value) between the suppositional modal world and the factual world. The same is true of imaginary worlds, which form another instance of neutral theoretical suppositional worlds. In the following examples, nothing is specified, implicated or presupposed in connection with the (in)compatibility of the suppositional world with the factual world. For example: (36) In your place I wouldn’t react if he wrote me a threatening letter. (This sentence neither affirms nor denies the speaker’s not reacting to a threatening letter. In other words, the head clause situation is presented neither as factual nor as counterfactual. The utterance is a pure supposition, i.e. a mere thought experiment.)6 (37) [I didn’t go to the party, so I don’t know if I would’ve become as drunk as you all appear to have been. In fact, I’m afraid] I would’ve been drunk too, unless I’d had my wife with me. (38) [It seemed unlikely that such a frail woman would have been capable of strangling her husband. However] if it had been her, the clues that would have been left would have been exactly the same as those we did find. [So we could not cancel her from our list of suspects.] (39) [Suppose the door was actually locked.] If the door had been locked, the burglar would first have tried to open a window, [and that would explain why we found those footprints in the flower-beds] (40) [If you’re naughty, a wild dog might come in and] it would eat you. 6. Unlike Harder (1996: 456), who applies the term “thought experiment” to tentative-P and counterfactual-P conditionals, we speak of ‘thought experiment’ in connection with imaginaryP conditionals only.



Renaat Declerck

As is clear from these examples, imaginary P-conditionals are not formally distinguishable from counterfactual conditionals. This explains why imaginary P-conditionals have often been mistaken for counterfactuals. 6. The epistemic scale of specified factuality values 1. ‘Epistemic’ has to do with knowledge. We can say that there is ‘specified epistemic’ meaning when either knowledge or a degree of (un)certainty is expressed about whether there {is/was/has been/will be} actualization of a situation in the factual world or in the expected future extension of it. In other words, a clause has specified epistemic meaning if the speaker expresses his opinion on the chances that the nonfactual world in which the situation actualizes coincides with the factual world or its expected future extension. These chances represent different values on an epistemic scale that ranges from ‘factual’ to ‘counterfactual’. This scale consists of a number of factuality values. The exact number of these values is irrelevant: we can split the continuum into as many categories as we wish. However, to account for all the meanings of the modal auxiliaries in English we need to distinguish between (at least) the following values:7 a. factuality: there is no doubt about the actualization of the (positive or negative) situation in the speaker’s factual world because that actualization is affirmed (asserted as being a fact in that world): John is ill./John is not ill.8 b. strong necessity (strong assumption of truth): [John is not here.] He must be ill. c. weak necessity (weak assumption of truth): [The money is not in the till, so] it should already be in the safe. d. probability: John may well be ill. e. possibility: John may be ill./It might be true. f. improbability: It should not be difficult to find his address. g. impossibility: It cannot be true./You cannot be serious! h. not-yet-factuality: John will come tomorrow. [I read the letter] before he had read it. i. counterfactuality: [You shouldn’t have gone through the woods.] You might have got lost. Because a counterfactual world is the opposite of the factual world, 7. The ordering of the values on the scale may be subject to debate, but that need not concern us here. 8. It may be worth stressing that, contrary to what is sometimes claimed, negation is not a modalizer. A negative clause like John is not ill does not express counterfactuality (but rather what we might call ‘the factuality of nonactualization’.) This means that both John is ill and John is not ill represent the actualization of a (positive or negative) situation as ‘factual’ – see also Section 2(f). Both sentences should therefore be treated as nonmodal sentences.

The definition of modality 

there is no doubt about the actualization of a (positive or negative) situation in a counterfactual world. (Note that, unlike the actualization of a negative situation, the actualization of a counterfactual situation is not located in the factual world. It is located in a counterfactual world, which is defined in relation to the factual world.) 2. There are some further remarks to be made. Firstly, some of these factuality values can also be expressed by conditional clauses – see also Section 4(d): a. Factuality is expressed by ‘factual’ conditional clauses, as in If I had a problem, I always went to my grandmother. b. Strong assumption of truth is expressed if the conditional clause is ‘closed’, i.e. when the condition is assumed by the speaker to be fulfilled, as in If the work will be done anyhow, I might as well have a lie down or If, as you say, you were late this morning, I would like you to do some overtime this afternoon. e. Possibility is expressed by ‘open’ conditional clauses, as in I’ll be relieved if she phones me. f. Improbability is expressed by ‘tentative’ conditional clauses, as in I’d be relieved if she phoned me, Call the police if someone should try to break in or I’d be very angry if she were to disturb me tomorrow. i. Counterfactuality is expressed by ‘counterfactual’ conditional clauses like If I were you... or If I had been there at the time... Secondly, the factuality values in between the extremes ‘factual’ and ‘counterfactual’ can be referred to as relative factuality values. (In the case of conditionals, they are the nonneutral theoretical values ‘closed’, ‘open’, and ‘tentative’ – see Section 4(d).) Unlike the extremes, they all imply some degree of uncertainty on the part of the speaker: the modal world is represented neither as coinciding with the factual world nor as being incompatible with it. Thirdly, of the values on the specified epistemic scale, only the relative factuality values and the counterfactuality value are modal values. Factuality is an epistemic value, but a world that is represented as factual is not a modal (=nonfactual) world. Obviously, a sentence with the epistemic value ‘factual’ cannot refer to a nonfactual world. In Section 2(i), ‘nonfactual world’ has been defined as a world that is not represented as factual. Fourthly, neutral theoretical conditional clauses, such as those expressing an ‘imaginary’ condition – see Section 5(d) –, do not express a factuality value on the specified epistemic scale, because the relation between the modal world and the factual world remains unspecified (indeterminate). Fifthly, counterfactuality differs from the other values on the specified epistemic scale in that it is not a relative factuality value: it is the only nonfactual epistemic modal value that does not imply a degree of uncertainty about the actualization



Renaat Declerck

Specified epistemic scale Nonmodal = factual

Modal = nonfactual Purely theoretical

Not-yet-factual

Counterfactual

= absolute factuality = relative factuality = relative factuality = absolute factuality value (a) values (b–g) value (h) value (i)

Figure 2.â•‡ Epistemic Values Nonmodal

Modal Specified

Indeterminate Factual

Neutral theoretical

Purely theoretical

Notyet-factual

Counterfactual

Figure 3.â•‡

of the residue situation in the relevant modal world. Because counterfactuality is defined in relation to the factual world – a situation is counterfactual if it is located in a ti-world which is the reverse of the factual ti-world – there is no doubt about the absolute factuality status of counterfactual situations. Thus, in [Kim should not have gone through the woods.] She might have got lost, there is a clear understanding that Kim did not get lost when she went through the woods. Unlike She may get lost, the clause She might have got lost does not leave any room for uncertainty. Figure 2 summarizes the conclusions arrived at in this section in connection with the specified epistemic scale. (The indications (a), (b-g), (h) and (i) refer to the examples in Section 6.1.) This chart can be expanded by adding ‘indeterminate epistemic modality’ – see Section 5(b). The result is Figure 3. 7. Root (nonepistemic) modality a. As noted in Section 5(a), sentences that are interpreted in terms of root modality (or nonepistemic modality) are concerned with factors that determine the actualization of the residue situation in a nonfactual world, as in the following examples:

The definition of modality 

(41) You may leave tomorrow. (The speaker is granting permission.) (42) You {can/cannot} buy tickets in the office over there. (The addressee’s ability or inability to buy tickets at a certain place depends on a circumstantial factor.) (43) The fugitives must leave the country because their visas expired last week. (The expiration of the visas makes it necessary for the fugitives to leave the country.) In such sentences no idea is evoked of an epistemic world that stands in a particular factuality-value relation to the factual world or to a future extension of the factual world. There is no ‘epistemic relation’ between two worlds, i.e. no relation that concerns the chances that a given epistemic world coincides with, or will coincide with, the factual world or its future extension. In the above examples, the nonfactual world in which the residue situation actualizes is an ‘epistemically dangling’ world – see also Section 5(b). Thus, The car can be repaired says nothing about the chances that the car will be repaired. It follows that we may wonder what is modal about ‘root modality’. Can we not say that My cat can swim is a simple statement of fact, without any reference to a nonfactual world? The answer is no. In My cat can swim, the residue proposition ‘My cat swim’ is true of a world that is not represented as the factual world, i.e. of a nonfactual world. The sentence says nothing about whether the actualization of the residue situation (which actualizes in that nonfactual world) is factual, doubtful, not-yet-factual, counterfactual, etc. in the factual world. The sentence My cat can swim may well be true even if my cat has never swum in its life and is never going to swim. This is because no epistemic relation is expressed between the nonfactual world in which my cat swims and the factual world. It does not follow, however, that there is no relation whatever between the two worlds. In the case of root modality, the relation is expressed by the lexical meaning of the modalizer. In The fugitives must leave the country, it is must that expresses the relation between the factual world and the nonfactual world in which the fugitives leave the country: the link between the two worlds is the modal position assumed by the speaker. In this example, the speaker expresses the modal position that it is necessary for the nonfactual world to coincide with the factual world. Similarly, in This problem can be solved, the speaker expresses the modal position that it is possible for the nonfactual world in which the problem is solved to coincide with the factual world. It follows that there is only a minimal difference between root modality and epistemic modality, whereas there is a lot that the two have in common. Let us start by looking at what they have in common, which is the following:



Renaat Declerck

a. The sentence contains a modalizer, e.g. must. (Compare the nonepistemic John must leave with the epistemic John must be leaving.) b. The auxiliary functioning as modalizer refers to a modal state (e.g. ‘X is necessary’) which is located in the factual world. c. The modalizer creates a world which is not represented as factual, i.e. a nonfactual world. d. The residue situation (denoted by e.g. ‘John leave’ or ‘John be leaving’) is conceptualized as actualizing in the nonfactual modal world. e. The lexical semantics of the modal auxiliary expresses the relation (e.g. necessity, possibility) between the modal world and the factual world. The difference between epistemic modality and root modality is determined by the interpretation-in-context of the auxiliary. It is only when the auxiliary is interpreted as an epistemic modal auxiliary that the modal world is an epistemic world and that the relation between that world and the factual world is an epistemic relation (e.g. ‘necessarily factual’). In the case of root modality, the relation between the nonfactual world and the factual one is not an epistemic relation. For example, John must leave tells us nothing about a factuality relation between the nonfactual world and the factual world, i.e. about the chances that the two worlds coincide. The relation that there is is the modal position (e.g. volition, obligation, possibility) assumed by the speaker. The number of possible modal positions is restricted. The set minimally comprises root necessity (including obligation) and root possibility (including ability and permission), probably also volition (willingness) and possibly some others. The question how many root modal meanings there exist is a moot question, which we will not go into here. b. Root modality and epistemic modality are not necessarily mutually exclusive. Sometimes an epistemic modalizer can be added to a clause expressing root modality. In such a case the root modal world is no longer an ‘epistemically dangling’ world, because the root modality is overlaid by epistemic modality. For example: (44) The soldiers must guard the ammunition bunker. (root modality, viz. obligation) (45) The soldiers must probably guard the ammunition bunker. (Probably expresses a relative factuality value, so that the root modality is overlaid by epistemic modality.) Similarly: (46) According to the arrangement, the decision should be made tomorrow. (root necessity)

The definition of modality 

(47) According to the arrangement, the decision should have been made tomorrow. (root necessity overlaid by epistemic modality, viz. a sense of counterfactuality resulting from modal backshifting) c. Root modality can be created not only by root modal auxiliaries but also by root modal adverbs (duly, obligatorily...) and by ‘attitudinal verbs’ like wish, want, etc. (see the next section). 8. World-evoking lexical verbs There are three types of world-evoking lexical verbs: factive verbs, intensional verbs and attitudinal verbs. The latter two are modalizers. a. ‘Factive verbs’ (see Kiparsky & Kiparsky 1970) are verbs like realize, know, acknowledge, etc. which evoke a world (expressed by their complement clause) which is automatically interpreted as being the factual world: (48) I {realize/know/acknowledge} that John is a better player than myself. The world expressed by the that-clause is treated as factual, and hence as nonmodal. Factive verbs are therefore not modalizers. b. ‘Intensional verbs’ like believe, suppose, think, dream, etc. create an ‘intensional’ world which may or may not coincide with the factual world. These verbs thus create a ‘specified epistemic modal world’ – see Section 5(b) – which is related to the factual world in terms of a factuality value. The value in question is mostly one of the purely theoretical values (e.g. what John believes may or may not be true, i.e. has the factuality value ‘possibly factual’). However, those verbs that have an implication of posterior actualization of the residue situation (e.g. I {foresee/predict} that they will soon surrender./I forgot to talk to her.) express the value ‘not-yet-factual’, whereas at least one of them, viz. imagine can imply counterfactuality, as in He imagined he was Napoleon. In sum, verbs like think, believe, etc. can create a world that is not represented as being the factual world. The residue situation is conceived of as actualizing in that nonfactual world. The stronger the intensional meaning of the verb, the less likely it is for the intensional world to coincide with the factual world. In the case of the strongest intensional verb (imagine) the intensional world is even normally interpreted as counterfactual. The semantics of the verb thus has to do with the likelihood of compatibility with the factual world. This means that such an intensional verb expresses a kind of factuality value, so that the intensional world can be called an epistemic modal world. Intensional verbs are therefore epistemic modalizers.



Renaat Declerck

c. ‘Attitudinal verbs’ are verbs like intend, wish, hope, expect, etc. that create a world which is not represented as being the factual world but which has the characteristics of a root modal world: there is little difference of meaning between I want him to do it and He must do it. Attitudinal verbs are thus modalizers creating a nonfactual world that is related to the factual world in terms of root modality. As in the case of root modality expressed by a modal auxiliary, the root modal interpretation of a world created by an attitudinal verb can be supplemented with an epistemic modal interpretation. Most of these nonfactual worlds have the epistemic value ‘not-yet-factual’. Moreover, an epistemic modalizer can often be inserted into the sentence, as in the following: (49) Perhaps John wants to leave. (The epistemic modalizer is perhaps.) (50) I had intended to have left this place earlier. (A counterfactual reading is produced by modally backshifting intended to had intended and by modally backshifting leave to have left.) (51) I wish Jane were here. (The use of the past subjunctive form were creates a counterfactual interpretation of the residue.) 9. Overview of modal worlds A nonfactual world has been defined as a possible world that is not represented as being the factual world. A nonfactual world is by definition a modal world (and vice versa). The world that is evoked by a ‘factive verb’ (realize, acknowledge, etc.) or by a ‘factual conditional clause’ – see Section 4(d) – is the factual world, not a modal world. The world that is evoked is a modal world in the following cases: a. The evoked world is a modal world if the sentence expresses ‘indeterminate epistemic modality’ – see Section 5(b) –, as is the case in ‘imaginary’ and other ‘neutral theoretical’ conditionals. Indeterminate epistemic modality means that the question of the epistemic status of the nonfactual world (i.e. its factuality value vis-à-vis the factual world) is relevant but is not answered because there is no reference to a specific factuality value on the epistemic modal scale. b. The evoked world is also a modal world if the sentence expresses ‘specified epistemic modality’, which means that the question of the epistemic status of the nonfactual world is relevant and is also answered: we know the factuality value (on the epistemic scale) that is applicable. Specified epistemic modality can be indicated by epistemic modal auxiliaries, epistemic modal adverbs,

The definition of modality 

intensional verbs (believe, imagine...), comment clauses with an intensional verb (e.g. I think), expressions of posteriority (e.g. before in He died before I had seen him), modally backshifted indicative forms (e.g. [I wish] it was true), modal conditionalization (e.g. I would suggest something different), the subjunctive (e.g. So be it, then./[I wish] it were true) and conditional sentences with a ‘nonneutral theoretical’ conditional clause, i.e. a conditional clause expressing a condition that is ‘closed’, ‘open’, ‘tentative’ or ‘counterfactual’. c. The evoked world is also a modal world if root modality is expressed. The relevant relation between the modal world and the factual world can be specified by a root modal auxiliary, by a root modal adverb (e.g. duly, obligatorily), by an attitudinal verb like want, intend, etc. or by the imperative mood. 10. Conclusion In this article we have attempted to give a definition of modality that is as concrete and as complete as possible. We have argued that there is modality whenever there is reference to actualization of a situation in a world that is not represented as being the factual world. All types of modality have been pigeonholed, regardless of whether the modalizer is a modal auxiliary, a lexical verb, an adverb, a conditional clause, a morphological operation like modal backshifting or conditionalization, etc. Clear definitions have been given of root modality and epistemic modality, and it has been shown that not all epistemic values are modal values: ‘factuality’ is one of the two extremes of the epistemic scale but is not a modal value. On the basis of a rich modal world typology and a relatively great number of possible relations between a modal world and the factual world, many different subtypes of modality have been distinguished, such as ‘indeterminate epistemic modality’ versus ‘specified epistemic modality’, the latter of which can be ‘purely theoretical’, ‘not-yet-factual’ or ‘counterfactual’. ‘Purely theoretical epistemic modality’ has been shown to comprise a variety of relative factuality values, which in the case of conditionals give rise to ‘closed’, ‘open’ and ‘tentative’ P-clauses. It has been shown that root modality and epistemic modality have much in common, in fact everything except the kind of relation existing between the modal world and the factual world. World-evoking verbs, which are usually disregarded in treatments of modality, have been drawn into the picture too. Whereas factive verbs are not modalizers, intensional verbs like believe or imagine can be used as epistemic modalizers, and attitudinal verbs like want or wish as root modalizers.



Renaat Declerck

References Biber, Douglas, Stig Johansson, Geoffrey Leech, Susan Conrad, & Edward Finnegan. 1999. Longman Grammar of Spoken and Written English. Harlow: Pearson Education Ltd. Declerck, Renaat & Susan Reed. 2001. Conditionals. Berlin & New York: Mouton de Gruyter. Declerck, Renaat. 2009. ‘Nonfactual at t’: A neglected modal concept. In Modality in English, R. Salkie, P. Busuttil, & J. van der Auwera, (eds.), 31–54. Berlin: de Gruyter. Harder, Peter. 1996. Functional Semantics: A Theory of Meaning, Structure and Tense in English. Berlin & New York: Mouton de Gruyter. Huddleston, Rodney. 1984. English Grammar: An Outline. Cambridge: Cambridge University Press. Huddleston, Rodney, & Geoffrey K. Pullum. 2002. The Cambridge Grammar of the English Language. Cambridge: Cambridge University Press. Jespersen, Otto. 1931. A Modern English Grammar on Historical Principles. Part 4: Syntax (3rd volume): Time and Tense. London: Allen & Unwin/Copenhagen: Ejnar Munksgaard. Kiparsky, Paul & Carol Kiparsky. 1970. Fact. In Progress in Linguistics, M. Bierwisch & K. E. Heidolph (eds.), 143–173. The Hague: Mouton. Leech, Geoffrey. 2006. A Glossary of English Grammar. Edinburgh: Edinburgh University Press. Lyons, John. 1977. Semantics. Cambridge: Cambridge University Press. Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech, & Jan Svartvik. 1985, A Comprehensive Grammar of the English Language, London: Longman.

The English present Temporal coincidence vs. epistemic immediacy Ronald W. Langacker

University of California at San Diego Two accounts of the English present are outlined and compared: one that treats it as tense, indicating coincidence with the time of speaking; and one that treats it as modality, indicating epistemic immediacy. Well-known problems with a time-based account are overcome in a conceptual semantic description that makes explicit the mental constructions invoked in non-present uses. With a modal account, the problem is to make the notion of epistemic immediacy sufficiently clear and precise for explicit analysis and comparison to be possible. There is no doubt that the modal account is more general, but for a wide range of central uses the two are effectively equivalent, temporal coincidence providing the basis for epistemic immediacy. The former can be seen as the category prototype, the latter as a schematic (fully general) characterization. Keywords: present tense, Cognitive Grammar, English

1. The issue In line with the theme of this volume, the basic question I address is whether the English “present tense” is best described as marking tense or modality. While this has long been a matter of concern from different perspectives, I will address it specifically in the context of cognitive linguistics and conceptual semantics. Linguists generally agree that the English present does not mark present time, at least not consistently or exclusively. After all, as seen in (1), the simple present cannot be used for present-time events, while it is used for events that are past, future, or time-independent. Nevertheless, I have argued (Langacker 1991: 2001) that the English present can in fact be analyzed quite successfully as marking present time. I have suggested that its “naïve” characterization, as indicating coincidence with the time of speaking, is basically valid. This account has been criticized by Brisard (1999, 2001, 2002), who has argued in general against time-line



Ronald W. Langacker

descriptions of tense, proposing in particular that the English present is best characterized as having a kind of modal import, which I will refer to as epistemic immediacy.

(1) a. *I read a book right now. [present] b. I get home last night and find a note on my door. [past (“historical present”)] c. They leave next week for Darfur. [future] d. Sugar dissolves in water. [time-independent]

I think Brisard would agree that it is not really a matter of choosing between these two accounts. In large measure they are interchangeable, representing alternate perspectives on the same basic vision.1 But while the temporal coincidence account works quite well within its intended scope, i.e. for clauses grounded by “tense” alone, sentences with modals require an epistemic characterization. Thus, if the temporal account is considered prototypical, a schematic (fully general) characterization has to be epistemic. The basic challenge is then to explicate the notion “epistemic immediacy” and how it applies in a broad range of cases. Neither account can be understood without the supporting context of conceptual semantics as developed in cognitive linguistics during the last quarter century. I must therefore start by introducing some basic notions of cognitive semantics and Cognitive Grammar (Langacker 1987a, 1990, 1991, 1999a). 2. Prerequisites 2.1

Conceptual semantics

Cognitive semantics starts from the supposition that meaning resides in conceptualization (in a broad sense of the term). Semantic structures are the conceptualizations evoked and symbolized by linguistic expressions. Since the mind is in the loop, linguistic meanings are not determined solely by objective properties of the situations described – what counts is how those situations are apprehended and viewed for linguistic purposes. Crucially, we have the ability to conceive and portray the same situation in alternate ways. This is known as construal. Every expression incorporates a particular way of construing the conceptual content it evokes.

1. “... I have not sufficiently emphasized epistemic immediacy, which I see as the flip side of coincidence with the time of speaking ...” (Langacker 2001: 271 [fn. 6]). My recognition of the need to develop an epistemic account has been reinforced by numerous discussions over the years with Brisard and with Mariko Higuchi.

The English present 

Construal is an essential aspect of linguistic meaning, part of the conventional semantic value of lexical and grammatical elements. Among the dimensions of construal are specificity, perspective, scope, and prominence. By specificity I mean the degree of precision and detail at which a situation is characterized. In relative terms, for instance, the noun dove is semantically specific, describing its referent in fine-grained detail, while bird is more schematic, offering only a coarse-grained characterization applicable to a wider range of entities. Perspective includes such factors as vantage point, orientation, and whether a scene is viewed in static or dynamic terms. The expressions in (2) illustrate the contrast between a static and a dynamic construal. The other two factors are exemplified by the phrases next week and last week. As shown in Figure 1, each evokes as its content the conception of an endless series of weeks (w) extending through time (t). They further presuppose a temporal vantage point (VP), located in one of these weeks, as well as an orientation, looking either forward or backward in time, represented by arrows. Given this content and perspective, we can simply describe the expressions as designating the week adjacent to the vantage point (in the direction of orientation). (2) a. We’re very close to Christmas. b. We’re rapidly coming up on Christmas.

[static] [dynamic]

Scope is reasonably viewed as an aspect of perspective, but here it merits separate attention. By scope I mean the extent of the conceptual content an expression invokes as the basis for its meaning – its “coverage” in relevant domains of knowledge and experience. One domain is space, which is central to the conception of physical objects: they exist in space, occupy spatial locations, have a certain spatial extension (size), and manifest a particular spatial configuration (shape). Consider, then, the notion table. We can hardly conceptualize a table without invoking the domain of space. However, it is not necessary to bring the entire universe within our scope of conception; we only need a spatial expanse sufficient for a table to be manifested in it. As shown in Figure 2(a), this vaguely delimited region – invoked to support the concept table – constitutes its spatial scope.

(a) next week W t

Figure 1.â•‡

W

(b) last week VP

W

W

W

W

W t

W

W

VP W

W

W



Ronald W. Langacker (a) table space spatial scope

(b) next week MS

IS VP

W

W

W

t

Figure 2.â•‡

From the mental standpoint, considered as a subjective representational space, scope is always bounded – we can only be aware of so much at any one moment. It is like a window on our mental world, with a certain fixed size and bounded by the frame. There is however enormous flexibility in regard to how much of the world appears inside this frame. In particular, we can apprehend most any facet of the world at different scales, and more appears within the frame as scale increases. For example, the following expressions pertain to successively larger time scales, so as their temporal scope (i.e. as the basis for characterization) they invoke successively longer time spans: tomorrow < next week < next month < next year < the next century ... Within a domain, there are often grounds for distinguishing between an expression’s maximal scope (MS), i.e. its overall content, and that portion which is directly relevant for a particular purpose. This I call the immediate scope (IS). Metaphorically, I describe this as the onstage region, the general locus of viewing attention. In the case of next week, the immediate temporal scope is roughly as shown in Figure 2(b), a time span sufficient to support the conception of the designated entity, as apprehended from an offstage vantage point.2 It is however still recognized that the time span being attended to is part of a series of weeks that extends indefinitely. Providing a clearer example are successive terms in a whole-part hierarchy, like the following: body > arm > hand > finger > knuckle. Each expression designates an entity which functions as the immediate scope for purposes of characterizing the next (e.g. a hand is specifically part of an arm, and a finger part of a hand). Of course, the body as a whole still figures in their conception, providing the maximal scope for the part terms. A final aspect of construal is prominence, of which there are many sorts. For instance, onstage elements (those within the immediate scope) are more prominent than offstage elements. For present purposes, the most important kind of prominence is profiling, indicated diagrammatically with heavy lines. We thus 2. I regard the vantage point as being offstage, hence outside the immediate scope, because it remains implicit and is not itself being viewed.

The English present 

observe in Figure 2(b) that the phrase next week, for example, profiles a certain week, and not, say, the vantage point or the domain of time overall. An expression’s profile is the entity it designates, or refers to – its conceptual referent (as opposed to its referent in “the world”, if indeed it has one). As such it is the specific focus of attention within the expression’s immediate scope (the general locus of attention). If the immediate scope is like a stage, the profile is like the actor currently being watched. The role of the mind in constructing linguistic meaning is by no means limited to imposing alternate construals on conceptual content. In large measure, that content – and more generally the world we live in and talk about (what counts as “the world” for linguistic purposes) – is mentally constructed. Crucial here are various mental phenomena, reasonably described as imaginative in nature, which have been extensively explored in cognitive linguistics. Among these are mental spaces, fictivity, metaphor, and blending. These give rise to elaborate mental constructions that are often essential for the semantic and grammatical coherence of expressions. Mental spaces (Fauconnier 1985, 1997; Fauconnier & Sweetser 1996) are separate “working areas”, each hosting a conceptual structure representing some facet of an overall conception. Though each has a measure of autonomy, a crucial factor is their configuration: the nature of the spaces, their relations to one another, and correspondences between their elements. Consider the sentence Joe wants to meet an actress. To describe its meaning, we have to distinguish between reality, the space invoked by default, and a space representing the content of Joe’s desire. The difference between the specific and non-specific interpretations is a matter of whether the actress figuring in his desire corresponds to one existing in reality, or is merely imagined, existing only in the desire space. On the non-specific interpretation, the actress referred to is said to be a fictive (or virtual) entity, rather than an actual individual. Of the entities we refer to linguistically, surprisingly many are fictive, even in discussions of actual situations (Langacker 1999b, 2005; Matsumoto 1996a, 1996b; Sweetser 1997; Talmy 1996). In Each boy threw a stone, for example, all the entities mentioned are virtual rather than actual. Each boy is not any actual boy, the stone mentioned stands for the actual stones involved but cannot be identified with any of them, and likewise for the profiled event of throwing that these fictive entities participate in. Still, the sentence is used to describe an actual complex occurrence. Cognitive linguists have amply documented the pervasive importance of metaphor in thought and language (e.g. Lakoff & Johnson 1980; Lakoff & Núñez 2000; Lakoff & Turner 1989; Turner 1987). A metaphor consists in correspondences established between two domains of experience, a source domain and a target domain, the former providing a means of apprehending the latter. In (2b), the source



Ronald W. Langacker

domain of movement through space is used to apprehend the target domain of time and sequences of events. Metaphor proves to be a special case of blending (Fauconnier & Turner 2002). In blending, selected elements of two input spaces are projected and integrated to form another space, called the blend. The concept robot, for instance, represents a blend of the notions person and machine. A blend has its own, emergent properties and is often purely fictive, something which could not exist in actuality. This happens in metaphor, a special case of blending, where a blend results from using the source domain to apprehend the target domain. Despite its fictive nature, the resulting blend provides the situation described by metaphorical expressions. Thus, just as its form suggests, (2b) describes the virtual process of moving through time as if it were a spatial landscape. This mental construction, involving metaphor and blending, is responsible for the expression’s semantic and grammatical coherence. 2.2

Symbolic grammar

A properly formulated conceptual semantics, strongly justified in its own terms, makes possible an account of grammar which views it as inherently meaningful. More specifically, I claim that lexicon and grammar form a continuum consisting solely in assemblies of symbolic structures. A symbolic structure is the pairing of a semantic structure with a phonological structure (its semantic and phonological poles). The aspect of this approach that most concerns us is doubtless the most controversial, even in cognitive linguistics. It is the claim that basic grammatical categories, such as noun and verb, can be characterized semantically – not just at the prototype level, but schematically, for the classes in general. Standard arguments against their semantic definability are erroneous because they ignore the vital role of conceptualization (considering only objective properties of the entities described), and also because the definitions considered (and rightfully rejected) – notions like ‘object’, ‘person’, ‘place’, and ‘event’ – are too specific to be valid for all instances. Conceptual characterizations have to based on how situations are apprehended and construed, and since the same content is subject to alternate construals, it can perfectly well support expressions representing different categories. Furthermore, characterizations schematic enough to cover all instances cannot be tied to specific conceptual content. They must instead be based on mental operations (or cognitive abilities). A key point is that an expression’s meaning depends on the nature of its profile, not its overall conceptual content. Consider the verb throw, sketched in Figure 3(a). It profiles an event in which an agent exerts force (double arrow) on another participant, thereby causing it to move along a spatial path (single arrow). The same conceptual content figures in thrower, represented in Figure 3(b), which is not however a verb but rather a noun. It is a noun because, with respect to this

The English present

content, it does not profile the event – a complex relationship that unfolds through time – but rather the agent, a kind of thing, in an abstract sense of the term. This difference in profiling is not a matter of the objective situation (which may be the same), or the conceptual content representing it, but rather one of construal, specifically the directing and focusing of attention. The same content also gives rise to the noun throw, a count noun designating one instance of the event (e.g. He hurried his throw). While this noun does refer to the same event as the verb, it is not semantically equivalent, because it construes it very differently: it takes the entire, temporally extended relationship and views it holistically as a single, object-like entity, represented by the ellipse in diagram (c). Throw is a noun because it profiles this abstract, mentally constructed thing. There is no reason to doubt that we have this capacity for conceptual reification. It is no more remarkable than the construal of time through spatial metaphor, as in (2b), or its personification in (3a). I believe it reflects the same mental operations that produce the referents of nouns like the ones in (3b). While it clearly consists of multiple entities, the profile of a noun such as stack is construed as a single, bounded entity at a higher level of conceptual organization. For this reason it functions grammatically as a singular count noun. (3) a. Time creeps up on us and robs us of our youth. b. stack, team, set, herd, flock, orchestra, alphabet, orchard, collection, archipelago ... Thus I define a thing abstractly, in terms of mental operations (Langacker 1987b, 1991: ch. 1). Here I merely suggest that these are basically contractive in nature, producing a unitary entity from constitutive elements. Conversely, I define a relationship abstractly in terms of basically expansive mental operations, which take an entity as the basis for further assessment, usually involving other entities (though it may also be internal). Granted these abstract notions, schematic characterizations can be offered for basic grammatical categories. A nominal expression – a noun or a noun phrase – profiles a thing. Verbs, adjectives, and prepositions profile different kinds of relationships. In particular, a verbal expression – a verb or a finite clause – profiles a process, by which I mean a relationship followed sequentially in its evolution through time. (a) throw (V)

⇒ Figure 3.â•‡

(b) thrower (N)

→

⇒

(c) throw (N)

→

⇒

→





Ronald W. Langacker

Clearly, the event described by throw evolves through time. There is first the agentpatient interaction, involving the transmission of force, and then the patient’s rapid movement along a spatial path. The verb throw represents the default way of apprehending this event, by simply tracking this relationship as it unfolds through time. As a noun, throw invokes the same event, which develops through time in just the same manner, but represents a non-default way of apprehending it. Instead of tracking it through time, instant by instant, it takes a summary view, construing it as a unitary entity at a higher level of conceptual organization. This is not unlike the summation producing the higher-order things profiled by the nouns in (3b), except that the constitutive elements, instead of being things themselves, are relationships (the successive “time-slices” of the event). Thus an expression is not a verb just because its content involves a relationship with temporal extension – it must further be apprehended and profiled as such. The essential role of construal is further evident in the three-way contrast in (4) among the verb resemble, the preposition like, and the adjective alike. All three expressions profile a relationship consisting in an assessment of similarity, indicated in Figure 4 with a dashed, double-headed arrow. Objectively, this relationship endures indefinitely, without essential change. Linguistically, the three expressions all have a different take on it. Being a verb, resemble specifically views it as continuing through time; this tracking through time is indicated by the bar along the time arrow. But since the relation is constant through time, the assessment can also be made with respect to any single moment during this indefinite time span. The preposition and the adjective construe it in this manner. The relationship they profile is equivalent to any time-slice of the verbal process. They simply present the essential content of the relation, in abstraction from its temporal duration. They do not themselves follow it through time.3 Hence they are not processual in nature, and thus not verbs. The basic difference between a preposition and an adjective is that the former has two salient participants, the latter just one. Within the class of adjectives, alike is atypical because the entity it puts in focus as its one salient participant is a higher-order thing consisting of distinct individuals. It thus takes either a plural or a coordinate subject (e.g. Bush and Putin). This is then a case where the profiled relationship is internal to what is construed (albeit at a higher level of organization) as a single participant. (4) a. Bush resembles Putin. b. Bush is like Putin. c. Bush and Putin are alike. 3. They can however combine with be, as they do in (4). This schematic verb gives them temporal extension at a higher level of conceptual organization.

The English present  (a) resemble (V)

(b) like (P)

(c) alike (ADJ)

t

Figure 4.â•‡

Especially important here is a basic conceptual distinction observable within the noun and verb categories. In the case of nouns, it is manifested in the grammatical opposition count vs. mass. For verbs, the analogous opposition is “active” vs. “stative”, or in my terminology, perfective vs. imperfective. The verb throw, for instance, profiles a perfective process, resemble an imperfective one. Once allowance is made for the inherent difference between nouns and verbs, the count/mass and perfective/imperfective oppositions turn out to be exactly the same (Langacker 1987b). For ease of presentation, I limit my discussion of nouns to the prototypes: physical objects for count nouns, physical substances for mass nouns. We think of these as existing primarily in space – I will say that space is their domain of instantiation. The account, though, is fully general, accommodating nouns with other domains of instantiation (e.g. time for a noun like week). With verbs the domain of instantiation is always time. The parallelism of count nouns and perfective processes is spelled out in (5). With respect to each property, they are polar opposites of substances and imperfective processes, as described in (6). Let us consider each property in turn, starting with objects and substances. (5) An object or a perfective process is: a. conceived as being bounded (in the immediate scope in its domain of instantiation); b. not conceived as being internally homogeneous; c. not contractible (an arbitrary subpart is not itself an instance); d. replicable (combining two instances results in multiple instances).



Ronald W. Langacker

(6) A substance or an imperfective process is: a. not conceived as being bounded (in the immediate scope in its domain of instantiation); b. conceived as being internally homogeneous; c. contractible (an arbitrary subpart is itself an instance); d. not replicable (combining two instances results in a single, larger instance). A physical object, such as a shoe, is conceived as being bounded in space (its domain of instantiation). The boundary and the shape it defines are inherent in the conception of an instance, so they fall within the immediate scope, i.e. the spatial expanse required to support this conception. This is shown in Figure 5(a), where shading represents the material substance constituting the object. Objects in general are not conceived as being internally homogeneous – they have distinguishable parts, often consisting of different substances. From these two properties it follows that an object is usually not contractible, i.e. a subpart is not itself an instance of the same type (part of a shoe is not itself a shoe). Finally, due to bounding an object is replicable: putting two instances together results in multiple instances (not a single, larger instance). Substances, for instance leather, exhibit the opposite properties. Though bounding is not precluded, it is not part of the characterization of a substance. Instead, a substance is characterized in qualitative terms (rather than by shape or observation of a spatial boundary). As such it is conceived as being internally homogeneous (the same qualitative properties hold throughout its spatial extension). From these two properties it follows that any portion of a substance is itself an instance of the same type, i.e. a substance is contractible. Hence the configuration in diagram (b): starting from a large expanse of substance, if we “zoom in” and direct attention to just a subpart – which is thus the immediate scope of conception – that portion is itself identifiable (on qualitative grounds) as an instance of the substance in question.4 Finally, a substance is not replicable, in the sense that combining two instances does not result in multiple instances, but rather a single, larger instance. For example, we talk about the leather in those two shoes – although the leather and the shoes are spatially coextensive, in terms of their linguistic construal there are two instances of shoe but only one instance of leather comprising them both. Turning now to processes, let me first introduce the notational convention of using an elongated box with square corners to represent a perfective process, as in Figure 5(c), and one with rounded corners for an imperfective process, as in 5(d). For either sort of process, the domain of instantiation is time: just as a physical 4. Note that only this portion is profiled, since by definition an expression’s profile is the focus of attention within its immediate scope.

The English present  (a) Object

(b) Substance

MS IS

space

(c) Perfective process MS

MS

MS IS

IS

IS space

(d) Imperfective process

t

t

Figure 5.â•‡

object or substance has spatial extension, a process has temporal extension. To carry the analogy one step further, let us ask what corresponds to the material substance that constitutes a physical thing. With a process, what extends through time (in the same way that substance extends through space) is nothing other than the profiled relationship, as shown in Figure 4(a). And like the material substance of an object, the component states of a process – the relations obtaining at each instant (its “time-slices”) – may or may not be the same throughout. That is, a process may consist in either a change through time or the continuation through time of a stable situation. A perfective process, like throw, is conceived as being bounded in time. And since the bounding is part of the characterization of such a process, something required to identify an instance of the type, it falls within the immediate scope in this domain, as seen in Figure 5(c). A perfective process is not conceived as being internally homogeneous. For instance, throwing has different phases (swinging of the arm, motion of the projectile), and since even a single phase involves a change through time, the constitutive relations (time-slices) differ from instant to instant. Thus a perfective process is not contractive, e.g. swinging the arm is not itself an instance of throwing. And it is of course replicable: successive acts of throwing constitute two instances of this process type. By contrast, an imperfective process like resemble has indefinite temporal extension, in the sense that bounding is inessential to its characterization. This goes along with its being internally homogeneous: the same relationship holds throughout, thus providing a qualitative means of identification. As a consequence, any portion of the overall process is itself an instance of the same type. If the statement Bush resembles Putin is valid throughout their official careers, it is also valid for any part of that time span, e.g. during 2005. So if we limit our attention to that immediate scope, in the manner of diagram (d), the same statement applies to the profiled segment thus delimited. An imperfective process is also non-replicable, i.e. combining two instances yields a single, longer instance. If Bush resembled Putin all during 2004, and all during 2005 as well, we would not say that Bush resembled Putin twice, but rather that he resembled him over the entire period.



Ronald W. Langacker

Obviously, I have provided only the bare minimum in regard to theoretical and descriptive prerequisites. With this background, we can now turn to the main order of business: presenting and assessing alternative accounts of the English present tense. 3. Temporal coincidence What is surprising about the account based on temporal coincidence is that it works so well, given the undeniable facts in (1): normally the present cannot be used for present-time events, but can be used for non-present occurrences (past, future, or time-independent). Nonetheless, I have proposed and defended the following characterization:

(7) The English present tense specifies that an instance of the profiled process occurs and precisely coincides with the time of speaking.

Let us first deal with the problem posed by examples like (1a), *I read a book right now, returning later to “non-present” uses of the present. 3.1

Present perfectives

Linguists recognize the need to posit two basic classes of English verbs – perfective (or “active”) vs. imperfective (or “stative”) – on the basis of grammatical behavior. In particular, perfectives resist occurring in the simple present tense but do take the progressive with be...-ing, while imperfectives do the opposite, as seen in (8). In (8ai) I include the adverb right now to force a “true present” interpretation, as opposed to one of the non-present interpretations (e.g. as habitual). Some linguists might define the two classes on the basis of these and other grammatical behaviors. In Cognitive Grammar, however, I take such behaviors as being merely symptomatic of a conceptual contrast, characterized in (5)–(6). Thus it is not a rigid lexical contrast. Although most verbs have a primary categorization as either perfective or imperfective, numerous factors are capable of inducing an alternate construal resulting in the opposite behavior. I do not attempt to survey these here. (8) a. Perfective: b. Imperfective:

(i) *He mows the lawn (right now). (ii) He is mowing the lawn. (i) He knows Italian. (ii) *He is knowing Italian.

With respect to the classic aspectual categories described by Vendler (1967), perfective verbs subsume three of the four – achievement, accomplishment, and

The English present 

activity verbs – while imperfectives correspond to stative verbs. Activity verbs, such as sleep, sit, and swim, behave like other perfectives in regard to the present tense and the progressive. While the processes they designate are easily construed as being internally homogeneous, they are nonetheless conceived as occurring in bounded episodes, and bounding is the critical property for the perfective/imperfective contrast. Note that verbs like these – bounded yet internally homogeneous – are directly comparable to count nouns such as lake, board, or lawn (bounded expanses of water, wood, and grass). Their conceptual characterizations make it possible to explain why perfectives and imperfectives behave grammatically as they do. Why do only perfectives take the progressive? The reason is that the progressive construction serves an imperfectivizing function, so its application to imperfectives would be superfluous. Its effect is sketched in Figure 6. On the conception of a perfective process, it imposes an immediate temporal scope internal to its boundaries, thus confining the profiled relationship to the onstage portion so delimited. Since bounding is no longer evident within the immediate scope, the derived process is imperfective. It is further construed as being internally homogeneous (the constitutive relationships are effectively equivalent, being identified only as time-slices of the original perfective process). Turning now to tense, we observe that the English past tense is compatible with both perfectives and imperfectives (He mowed the lawn; He knew Italian), and that imperfectives occur in the present (He knows Italian). It is only presenttense perfectives that are problematic (*He mows the lawn). To see why we must first describe the tense morphemes. Like the progressive, they impose an immediate temporal scope, i.e. they direct attention to a particular span of time for purposes of viewing the profiled clausal process. Unlike the progressive, which is aspectual, the present and past are deictic elements that identify this time span relative to the speech event. In diagrams, I represent the speech event by means of a box with squiggly lines. Obviously, the past morpheme imposes an immediate scope that is prior to the time of speaking, as shown in Figure 7(a). The profiled Progressive (be Ving) MS IS

t

Figure 6.â•‡



Ronald W. Langacker (a) Past IS

t

(b) Present MS

IS

t

(c) Performative MS

IS

MS

t

Figure 7.â•‡

process must appear in this frame. In the case of the present, shown in diagram (b), the immediate temporal scope and the speech event are coextensive. The hypothesis at issue is that the designated process precisely coincides with this time span. Past perfectives are unproblematic because there is no inherent limit on the size of the immediate temporal scope – it need only precede the speech event. Hence it can always be made large enough to encompass a bounded process, as required. Past imperfectives are also unproblematic, but for a different reason. Since they may extend indefinitely, there is no guarantee that a limited immediate scope always contains them in their entirety. But full inclusion is unnecessary, due to contractibility. Any subpart of an imperfective process is itself a valid instance of its type. Suppose, then, that an imperfective process overflows the limits of the immediate scope, as seen in diagram (a). The portion subtended by it, and profiled by virtue of being onstage, itself constitutes an instance of the process type, hence the requirements are met. In just the same way, imperfectives work in the present tense because, as seen in diagram (b), the portion of an imperfective process which coincides with the time of speaking itself counts as an instance of its type. Perfective and imperfective processes are naturally associated with different modes of apprehension. Since bounding is inherent in its characterization, a perfective process has to be observed in its entirety to be identified. But with imperfectives this is not an issue, as there may not be any “entirety”. For a stable situation that extends indefinitely, the usual mode of apprehension is one of sampling some limited portion. A limited sample is sufficient for qualitative identification and may be all that is possible. Why, it may now be asked, are present perfectives problematic? There is nothing conceptually incoherent about a bounded process being coincident with the speech event. Rather than their being semantically anomalous, I have suggested that present perfectives pose problems of use from the pragmatic standpoint. One problem is durational: most bounded events that we might have occasion to talk about happen not to be the same in length as the time required to utter a clause describing them. It takes at most a second to produce the sentence He mows the

The English present 

lawn, but the actual mowing is likely to take far longer. There is also an epistemic problem: by the time we observe an event to find out what is happening, it is already too late to initiate a description that precisely coincides with it. Striking confirmation for this explanation comes from a major exception to the non-occurrence of present-tense perfectives, namely performatives (Austin 1962): (9) a. I order you to destroy those files. b. I hereby sentence you to 30 days in the county jail. c. I promise that I will be more careful. Speech act verbs are perfective (note that they take the progressive when used descriptively, e.g. She is ordering him to destroy the files). But when used to actually perform the speech act named, they necessarily occur in the present.5 Performatives represent a special case of language use where the durational and epistemic problems simply do not arise. The durational problem does not arise because, in a performative, the profiled process and the speech event are one and the same, hence necessarily the same in duration, as shown in Figure 7(c). The epistemic problem does not arise because the speaker has prior intention to perform the act in question – she does not have to observe the event in order to identify it. The infelicity of present tense perfectives is therefore not due to inherent semantic anomaly, but rather to the durational and epistemic problems that arise in a particular kind of circumstance. This circumstance does however have defaultcase status. In the default viewing arrangement, the interlocutors are together in a fixed location from which they observe and report on actual occurrences in the world around them. It is on this basis that a sentence like He mows the lawn is judged ungrammatical and given a star, even though it is perfectly acceptable assuming a different arrangement (e.g. to describe a habitual practice). The incompatibility stems from the interaction between the default viewing arrangement (involving observation and description) and the meaning of the present tense. We can therefore predict that present tense perfectives will also be acceptable in other viewing circumstances where the durational and epistemic problems happen not to arise. This will be so when the event in question has the right duration to be made coincident with the time of speaking, and where the speaker is the actor, acting intentionally and with full awareness. We can imagine, for example, a friendly game of chess in which a player describes his move as he makes it. The first clause in (10a) would be uttered while moving the rook, and the second clause 5. I do not consider cases with the progressive, e.g. I am ordering you to destroy those files, to be true performatives. Rather, they represent the politeness strategy of indirectness: the speaker avoids the direct assertion of authority, achieving the same effect by merely describing what he is actually doing. Alternatively, the sentence could be used to reinforce an order already made or merely implied, the explicit description leaving no room for ambiguity.

 Ronald W. Langacker

while removing the opponent’s knight. Or, we can imagine a situation where, for some reason, the speaker carries out a series of bodily movements and describes each one as it occurs, as in (10b). Although the situation may be contrived, use of the present with perfectives is perfectly natural in this context. (10) a. I move my rook to QB3, and capture your knight. b. I raise my hand. I lower my hand. I turn to the left. I turn back to the right. A very common use of present-tense perfectives is for the play-by-play description of sporting events, as in (11a). This would seem to be a clear instance of the default viewing arrangement. Why, then, are present-tense perfectives not just acceptable, but the norm? For one thing, the events described do in fact have just about the right duration. A longer event, like the manager walking from the dugout to the pitcher’s mound, would normally not be reported in the simple present. Also, the events in question are highly stereotyped, to the point that they can be identified very quickly and even anticipated. Observe that events that depart from the usual script, like the fight in (11c), are not reported in the present. There is also a certain amount of tolerance in normal language use. If I say, for instance, that All politicians are dishonest, you will not consider me a liar if we both recognize that there are a few exceptions. So given the reality of the situation, we are no doubt prepared to cut the announcer a bit of slack in regard to temporal coincidence. We might also say that the play-by-play style of announcing is based on a fictive viewing arrangement, where the fiction of simultaneous reporting is adopted for sake of convenience and “vividness”. (11) a. He hits it into the hole. Jeter makes a nice stop. He fires to first, and gets him by a step. b. The manager {is walking/*walks} slowly toward the mound. c. A fight {has just broken out/*breaks out} in the stands! 3.2

Non-present uses

Play-by-play reporting represents a transitional case between present and nonpresent uses of the present. To the extent that it relies on scripting or a fictive viewing arrangement, it resembles the latter. Non-present uses of the present are all based on departures from the default viewing arrangement. When these are properly recognized, an account in terms of temporal coincidence can be rescued. As another transitional case, imagine a group of detectives watching a surveillance tape which has caught the commission of a crime. One detective, who has previously viewed the tape, narrates what is happening as it happens, in the manner

The English present 

of (12a), describing each event as it appears on the monitor. In another arrangement, the clerk might recall the same events using the “historical present”, as in (12b). In both cases present tense perfectives are perfectly felicitous. (12) a. The suspect enters the store. Now he approaches the counter. He hands the clerk a note. Now he pulls back his coat, and shows her the gun. b. I’m working late last night, just getting ready to close up, when this guy walks in. He comes to the counter and gives me a note. Then he pulls back his coat and I see a gun. Why the present, when the events were in the past? What is crucial here is that both situations involve a special viewing arrangement more elaborate than the default one. It is not just a matter of observing and describing actual occurrences. In (12a), the interlocutors are not observing the events themselves, directly, but are watching a video replay. In (12b), the events once observed directly are being accessed through recall. The historical present is not unreasonably described as a “mental replay”, where previous events are relived by running the tapes of memory. In both cases, a distinction can be drawn between the events themselves and representations of the events. A surveillance video represents events. Memory records experience for later access. We must therefore ask which level the expressions in (12) apply to. It seems to me quite evident that they are directly describing the representations rather than the events themselves. When the detective says He hands the clerk a note, he is directly describing something that happens in the context of the surveillance video. Of course, since the events on the video are representations of actual events, the expressions also describe the latter – but only indirectly, via the description of the video events representing them. Similarly, the historical present consists in describing events that occur as part of a mental replay; only indirectly, via this replay, do the expressions pertain to the original, actual events. I propose, then, that the present perfectives in (12) are based on special viewing arrangements more elaborate than the default arrangement. As shown in Figure 8, these special viewing arrangements involve the distinction between a represented event, which may be actual, and a representing event, which – as a representation – is necessarily virtual. In these arrangements, what is being “viewed”, and directly described linguistically, is the representing event. However, the entire configuration, including the representing event and the relation it bears to an actual one, is part of an expression’s overall meaning. Every expression presupposes a multifaceted conceptual substrate that supports the content overtly expressed and renders it coherent. The viewing arrangement is one facet of this substrate. On this account the present perfectives in (12) do in fact describe events coincident with the time of speaking. In (12a), the detective describes each virtual event as it occurs on the monitor. They are all roughly of the proper length. For



Ronald W. Langacker (a) Default viewing arrangement Expression Direct description Actual event

(b) Special viewing arrangement Expression Direct description Representing event (Virtual) Representation Represented event (actual)

Figure 8.â•‡

longer events, the detective would probably resort to the progressive: Now he’s walking around the store to be sure there’s no security guard. Likewise, a speaker using the historical present describes events coincident with their successive recall. Or to put it another way, the speaker relives the events by recounting them. And since they are only being recalled (not actually occurring), duration is not a problem: by adjusting the speed of the replay, the representing events can always be made to coincide with the time of speaking. I suggest that Figure 8(b) represents a general scheme for (so-called) nonpresent uses of the present. They are all based on mental constructions involving representations of occurrences. It is these virtual occurrences – rather than the actual ones they represent – that are directly described linguistically and coincide with the time of speaking. The constructions involved are numerous and varied in nature. I consider just a few more (representative?) cases. First is the future use of the present, as in (13a). I call this the “scheduled future”, since it evokes some kind of plan or schedule governing the occurrence of future events. Although it may be physically manifested, in printed form or on a monitor, what matters is the existence of a schedule as a mentally and socially constructed entity – an accepted plan expected to govern the timing and occurrence of future events. Observe that we do not use the scheduled future for events that cannot be planned or controlled, e.g. (13b). Note further that we have to posit this mental construction and acknowledge its linguistic relevance quite independently of the present tense. In (13c), for instance, the second clause still describes a scheduled future occurrence, even though the verb is marked for past tense rather than present. In other words, the scheduled future is not just a meaning of the present tense, but arises from the interaction of the present tense with an independent conceptual phenomenon.

The English present 

(13) a. We have to hurry. The plane leaves in ten minutes. b. *A nuclear accident destroys Seattle next week. c. She was rushing through the airport. The plane left in ten minutes. The scheduled future use of the present is sketched in Figure 9. A schedule comprises a series of virtual events, each the mental representation of an anticipated actual event. While the represented events lie in the future, the virtual representing events are presently available whenever the schedule is known and in effect. In expressions like (13a), the speaker is not referring directly to the planned future event. What she is doing, instead, is consulting the schedule and “reading off ” an entry. Reading off an entry consists in activating or mentally reconstituting the profiled representing event. This virtual occurrence, which is just a matter of the speaker apprehending what she is saying, coincides with the time of speaking.6 A script is something like a schedule, one difference being that the represented events can occur at any time – indeed, any number of times – provided that they all occur in the proper (scripted) sequence. In “reading” a script, the virtual representing events are mentally reconstructed from their descriptions, and interpreted as models to be emulated in producing the actual events constituting a performance.7 In the case of scripts, some situations are ambivalent as to whether the events being described are the actual ones of a performance or the virtual ones the performance instantiates. Imagine a cooking show on TV, where the chef describes each step roughly coincident with performing it, as in (14a). Possibly the chef is directly describing the actions themselves, in the manner of (10). However, the chef is also following a script, so despite the first-person pronoun (which I take as reflecting the actual performance), he might also be describing the virtual MS

IS schedule (virtual events) t

t

(anticipated actual events)

Figure 9.â•‡ 6. A diagram for the historical present would be comparable, except that the represented events are prior to the time of speaking, accessible by recall rather than scheduling. 7. Of course, these “actual” events may themselves be fictive representations of something that may or may not have happened in reality.



Ronald W. Langacker

representing events which constitute it, doing so in sync with the actual events instantiating them. In another style of narration, the chef describes each action before performing it, as in (14b). If the descriptions pertain directly to the actual events, we have to say that this speech genre allows a generous tolerance in regard to simultaneity (more generous than in the play-by-play mode). What I incline to say, instead, is that the expressions directly describe the virtual events of the script. The chef evokes each successive representing event by way of introducing the actual event that will follow. (14) a. First I take an egg. I crack it and empty it into a bowl. Now I take a cup of flour, and put it in the bowl with the egg. I mix them together ... b. First I take six eggs ... I crack them and empty them into a mixing bowl ... Now I measure out two cups of flour ... I put them into the bowl with the eggs ... Next I beat the mixture until it is well blended ... As a final example, consider a singular generic, as in (15a). Generics invoke a cultural model in which certain kinds of events are taken as reflecting the world’s essential nature, which provides a stable structure within which other, incidental events occur (Goldsmith & Woisetschlaeger 1982; Langacker 1997). Accordingly, our knowledge of the world includes a set of structural generalizations, representing how the world is thought to work. These generalizations are not limited to laws of nature or physical matters; part of what constitutes the structure of our “world” is established social practice. Sentence (15a) describes a structural generalization of this sort. As a generalization, it is necessarily a virtual entity, representing what is common to any number of actual occurrences. The process profiled by this clause is thus a virtual event corresponding to an open-ended set of actual ones, as sketched in Figure 10. The man and woman referred to by the subject and object nominals are virtual instances of their types rather than actual individuals. In producing a sentence like (15a), the speaker is “reading off ” an entry in this representation of the world’s essential structure. So while the represented events have no particular temporal location, this virtual occurrence of the profiled representing event coincides with the time of speaking. (15) a. A man proposes to a woman. [That’s how it’s done.] b. In those days, a man proposed to a woman. [Now anything goes.] Once more, this mental construction – the notion of structural generalizations pertaining to the world’s essential nature – is independent of the present tense. It is also invoked in past-tense sentences like (15b), describing what the world’s structure used to be like (before the world went to hell). Thus genericity is not, in and of itself, a meaning of the present tense. Rather, the mental construction provides a particular context in which the present can be used in its normal value.

The English present  MS

IS structual generalizations

…

m

w

m

m

w

w

m

w

m

w

…

t

Figure 10.â•‡

4. Epistemic immediacy The temporal coincidence account of the English present enjoys a considerable degree of descriptive success. To compare it with an account based on epistemic immediacy, the latter needs to be formulated with comparable detail and precision. While not the same in all specifics, the following attempt is in the same spirit as the proposals made in Brisard (2002). 4.1

General considerations

In presenting the temporal coincidence account, I spoke of the “virtual occurrence” of events. For the profiled event and the speech event to temporally coincide they have to both occur, and when the former is only virtual, its occurrence must be virtual as well. What it means to say that a virtual event “occurs” depends on its nature and provenance. A taped event occurs in the sense of appearing on a monitor, where it is subject to observation in much the same manner as the original, actual event. A recalled event consists in a mental simulation, a partial reactivation of the original experience. In the case of scripts, mental schedules, or generalizations about the world’s inherent structure, a virtual event’s occurrence may simply be a matter of its apprehension by the speaker as the essential content of a finite clause.8 While I take this notion as being both coherent and defensible, it has the drawback that words like occur and occurrence suggest the actual realization of events. 8. This too is a kind of mental simulation, granted the view – now prevalent in cognitive semantics – that simulation has a central role in conceptual meaning (e.g. Johnson 1987; Barsalou 1999; Matlock 2004; Bergen 2005).



Ronald W. Langacker

An epistemic approach has the advantage of allowing us to avoid these terms. Instead of saying that an event’s occurrence coincides with the time of speaking, we can simply say that the present indicates epistemic immediacy of the profiled process. This is what Brisard (2002: 263) is getting at with expressions like “immediate givenness”, “present givenness”, “epistemic certainty”, and “immediate certainty”. Of course, I am less concerned with the term than with its import. What does this “immediacy” or “certainty” consist of? And if the profiled clausal process is specified as being immediate, who is it immediate to? And when? Our concern is limited to English tense and the basic modals. These constitute what is referred to in Cognitive Grammar as clausal grounding (Brisard [ed.] 2002; Langacker 2002a, 2002b): grammaticized elements that specify the epistemic status of the profiled process vis-à-vis the ground (the speech event and its participants). To avoid extraneous issues, I will make the standard simplifying assumption that the proposition expressed by a finite clause represents the actual view of the current speaker.9 Assuming this default, the present indicates epistemic immediacy to the speaker at the time of speaking. So as soon as one tries to be explicit about immediacy, the notion of temporal coincidence re-enters the picture. This should come as no surprise, as temporal and epistemic concerns are closely bound up with one another. It has often been observed that present, past, and future are not just locations along a line, but differ in essential respects (Brisard 1999; Evans 2004: 186–188). This is sometimes phrased in terms of mental experience and the kind of access we have to occurrences: we access the present through direct experience, the past through memory, and the future through anticipation. Alternatively, it is sometimes phrased more objectively, with no explicit reference to cognition: the past is fixed and the future undetermined, while the present mediates between them as the ongoing transition from potential to realized. But in either case these are fundamentally epistemic notions. I accept the basic validity of these qualitative distinctions. Moreover, I see in them a motivation for certain aspects of linguistic structure. In and of themselves, however, they cannot serve as an actual characterization of tense. Obviously, for example, the English present cannot be characterized semantically as indicating that the speaker directly perceives or experiences the profiled occurrence – if only because so much of what we talk about is abstract, mentally and socially constructed, or learned through communication or instruction. We cannot simply describe the past tense as meaning that the profiled occurrence is accessed through memory. After all, much of what we describe in 9. The extent to which this oversimplifies matters, and how to approach the full range of cases (not just this special, albeit privileged case), are spelled out in Langacker (2004a).

The English present 

present-tense sentences is also retrieved through memory (e.g. Lincoln is the capital of Nebraska). Suppose we base semantic characterizations of tense on the idea that the past is fixed, the present inchoate, and the future undetermined. It would then seem problematic that the present is used for “timeless” statements of general validity, e.g. A triangle is a three-sided polygon. And if the relevant notion is “epistemic certainty”, why does this not point to the past rather than the present? Something that is fixed would seem to be epistemically more certain that something merely inchoate. Simple characterizations of this sort – e.g. defining the present tense in terms of direct experience, and the past tense in terms of memory – might be workable if the use of language were limited to personally witnessed physical occurrences (statements like I see an apple; The apple may fall; The apple fell; The apple is on the ground). But general characterizations have to be more abstract and broadly applicable. An epistemic account has to accommodate knowledge irrespective of its nature and source. In particular, a general description of the English present requires a notion of epistemic immediacy that is elaborate and less than obvious, being based on cognitive models and mental constructions. 4.2

An epistemic model

My account (Langacker 1991: ch. 6; 2004a; 2008) starts from the premise that living creatures are continually striving to gain or maintain control of their circumstances. This striving for control occurs on many levels: physical, perceptual, mental, and social. It includes such varied activities as eating, focusing attention, acquiring possessions, and establishing social relationships. And crucially, at the mental level, it includes our ongoing effort to build up a coherent conception of “the world” (in the broadest sense of that term). The result of this activity, for a given conceptualizer at a given time, constitutes what I refer to as a reality conception.10 As life continues, a conceptualizer is constantly engaged in adjusting and augmenting his conception of reality on the basis of new experience. So reality, for a particular conceptualizer (C), is what C currently accepts as valid or established. At least for the time being, it is a settled matter, as opposed to being merely suspected or simply being considered. It is what C “knows” (rightly or wrongly). Considering an element P for inclusion in reality (R) is to some extent a force-dynamic activity (Talmy 1988), a manifestation of C’s striving for epistemic control. But once the issue has been resolved, P’s inclusion as part of R 10. The word “reality” should not be taken too strictly, since “the world” – in the broadest sense – includes innumerable entities that are not real in any narrow sense, e.g. mental constructions like a schedule or a structural generalization about the world’s essential nature.



Ronald W. Langacker (a) Epistemic striving

⇒

C R

(b) Epistemic control

P

C

P R

Figure 11.â•‡

represents a locally stable situation, a state of relaxation (rather than mental effort). Note that this abstract characterization makes no reference to the source of knowledge or its means of incorporation in R. It makes no difference whether P’s acceptance in R is based on perception, hearsay, inference, instruction, intuition, etc. This much is sketched in Figure 11, where a double dashed arrow indicates the striving for epistemic control. The English tense-modal system (clausal grounding) is based on this distinction between epistemic striving and epistemic control. The system consists of two binary oppositions: the presence vs. the absence of a modal; and “present” vs. “past”, which – as a general characterization – is better described as “proximal” vs. “distal”, or immediate vs. non-immediate. The presence of a modal indicates epistemic striving, in which case the choice of the modal (may, can, will, shall, must, and their non-immediate counterparts) specifies the degree of force involved (Sweetser 1982; Talmy 1988; Langacker 2008). The absence of a modal indicates epistemic control: the profiled process is part of C’s conception of reality. We will return to the modals. For now, let us concentrate on cases where a modal is absent. Part of C’s reality conception is C herself, her existence, and her current experience. These are taken for granted as the foundation for cognition. Not only that, but – in a very real sense – C and C’s location function as the center of C’s mental universe. For linguistic purposes, they define the deictic center: the I-here-now from which the world is apprehended, and relative to which other entities are situated. At a given moment, C’s experience (E) comprises what C is consciously aware of – internally, through perception, or by thinking about it. Perception (e.g. seeing something) and internal awareness (e.g. feeling pain) are forms of direct experience, which provides the epistemic foundation for our mental world. But obviously, the world we inhabit and talk about goes far beyond what we ourselves directly experience. It includes what others have perceived and credibly reported. Much of it is abstract, being mentally, socially, and culturally constructed, but is nonetheless as real and significant to us as the core layers. I accept it as valid, for instance, that Lincoln is the capital of Nebraska, even though this is not something that can be

The English present 

directly perceived. Part of our world, and accepted as part of reality, are mental constructions like schedules, scripts, and structural generalizations about the world’s essential nature. Also part of our world are entities like linguistic theories, and while their content may not be accepted as real, their existence certainly is. As a first approximation, then, our epistemic landscape has the configuration shown in Figure 12(a). Direct experience accounts for only part of what we accept as real, and what we accept as real is only part of our mental world (W). For a second approximation, we have to introduce time. We do not experience the world as a series of discrete time-slices, but as something that is continuously developing. One thing we experience is the passage of time itself, and one thing we know is that the world, and our experience of it, are constantly changing through time. We must therefore elaborate our basic epistemic model to incorporate the temporal dimension, as shown in diagram (b). Part of what we know is that our current experience – or immediate experience (IE), as I call it – is just the latest phase in the ongoing process of experiencing. Likewise, the world itself, as well as that portion of the world we know about (R), have a developmental history, of which the current (momentary) phase is only the latest. Those aspects of our knowledge that pertain to the world as presently constituted will be called immediate reality (IR). Immediate reality is the locus of immediate experience, i.e. one thing we accept (or take for granted) as valid about the current moment is that we are having our current experience.11 (a)

(b)

W

W C

C E R

E

IE

R

IR

Figure 12.â•‡

11. I should point out a subtle difference in how IR and IE are defined. Whereas IE is the current phase of experience, IR is not the current phase of knowledge – it is rather (what is accepted as) knowledge of the present. At a given moment, we know something about the past as well as the present, but we currently experience only what we currently experience. The past figures in IE only to the extent that we recall or apprehend it. The current phase of knowledge includes all of R, of which IR is only a portion.



Ronald W. Langacker

At a given moment, most of what C accepts as real is only latent, hence not an aspect of immediate experience. I know, for example, that Lincoln is the capital of Nebraska, but this figures in my experience only on those rare occasions when I happen to think about it. The point, of course, is crucial. The distinctive property of the human mental capacity is precisely the fact that immediate experience is not confined to what we currently feel and perceive. We are capable of thinking about anything at all, mentally peering into any corner of our immense conceptual universe, of which reality itself is only a limited portion. When we think about something, we thereby bring it within the scope of awareness, so that – in a sense – it is part of immediate experience. It is not that the object of thought, per se, moves through the epistemic landscape and enters IE; merely thinking about something does not make it real or perceptible. Rather, it is part of IE in the sense that its apprehension is an aspect of our conscious awareness, experienced internally. Thus, in and of itself, thinking about something does not make it real or otherwise change its epistemic status. Its apprehension is however real and directly experienced. In Figure 13(a), the object of thought, X, is shown at an arbitrary location. Its apprehension, X′, is internal to C, an aspect of C’s conscious awareness. To keep the diagrams simple, I abbreviate this as shown in diagram (b), where X′ is left implicit. Instead, a dashed arrow stands mnemonically for the process we conceptualize metaphorically as C “reaching out” and “apprehending” X. As noted, X can be anywhere in the epistemic landscape. It may lie outside reality, e.g. in entertaining future hopes. In thinking about the past, X is real but non-immediate. It may be in immediate reality but outside immediate experience, for instance invoking the knowledge that Lincoln is the capital of Nebraska. In the case of external perception, like seeing a cat on a mat, X – the situation perceived – is part of IE but outside C, X′ being the perceptual experience. And in the special case where X is internal to C, e.g. the experience of pain, X and X′ effectively (a)

(b) MS

X′ C IE IR R

Figure 13.â•‡

X

IS

MS

C IE IR R

X

IS

The English present 

collapse – there is no essential difference between apprehending an internal experience and simply having that experience.12 I will understand “thinking about X” to mean that we are specifically attending to X, as the focus of our conscious awareness. This does not of course imply that we are aware of nothing else. Even if I am focusing on the fact of Lincoln being the capital of Nebraska, I am still aware to some extent of my immediate physical surroundings, of being slightly hungry, etc. In particular, I have some awareness of the overall epistemic landscape and the position of X within it. If I focus attention on a cat being on a mat, I am generally aware as well of the epistemic status of this situation: whether it is purely imaginary, something that happened previously, something I actually see right now, or something I know of by other means (e.g. someone tells me). There are, in short, levels and degrees of awareness. Accordingly, in Figure 13 the box containing X is labeled as the immediate scope of conception (IS), i.e. the general locus of attention (the “onstage region”), and the outer box as the maximal scope (MS), the full extent of awareness.13 For present purposes, C can be characterized as the subjective center of consciousness. We can think of the arrow from C to X as an epistemic path, the path C follows in “reaching out” and “apprehending” X. The path is “longer” or “shorter” depending on X’s position in the epistemic landscape. In the case of internal experience, where X and X′ are the same, the path has a length of zero. Presumably the path is progressively longer as it passes beyond the limits of C, IE, IR, and R. Measured in this way, the path gives a rough estimate of “epistemic distance”. These boundaries are all natural places where a language might choose to make a coding distinction. For example, “evidential” markers might indicate whether or not an occurrence is (or was at some stage) part of immediate experience. English makes coding distinctions at IR and R. This is an epistemic model pertaining to facets of cognitive organization. Inclusion in R does not depend on something being true in any objective sense (i.e. true in the world), but rather on C accepting it as being valid with respect to C’s mentally constructed world. One difference is that C is truly the center of this mentally constructed world, while probably not enjoying this privileged position objectively (in the “real” world out there). Another difference is that past occurrences accepted as part of R can be modified on the basis of new considerations – our conception of previous history is subject to adjustment. We could only wish that the objective course of events were so flexible. 12. Of course, we can also think about the nature of this experience and the fact that we are having it, in which case it also functions as an object of thought, itself being apprehended. I take this as representing another level of conceptual organization. 13. The latter’s apprehension is also an aspect of internal experience. I have not attempted to show this in the diagrams.



Ronald W. Langacker

C1

C0

C2

R2

X

R1 R0

Figure 14.â•‡

Figure 12(b) shows E as the core of a cylinder, suggesting (not unreasonably) that personal experience has a privileged status. But obviously – and obviously essential for language (though not emphasized here) – we accept as real the existence of other conceptualizers each of whom engages in conceptualization and builds up their own conception of reality. We thus have configurations like Figure 14, where C0 accepts as real the proposition that C1 accepts as real the proposition that X is accepted as real by C2.14 In this case X does figure in C0’s reality conception (R0), but is only indirectly accessible within it. To “reach” X, C0 has to trace along a multi-step epistemic path leading through other conceptualizers and their own conceptions of reality. Reality, then, is a complex affair involving often elaborate mental space configurations, with any number of hierarchical levels. Among the requisite spaces are those corresponding to mental constructions like schedules, scripts, and structural generalizations. These are virtual entities comprising virtual (representing) events. Their existence as such is nonetheless part of reality, as defined. Being based on accepted validity, real vs. unreal is not the same as actual vs. virtual, since even virtual entities can be accepted as valid (cf. Langacker 1999b). If R is conceived as a cylinder, as in Figure 12(b), we can view it metaphorically as “growing” along the temporal axis as time passes and there is more to know about. On this view, future occurrences per se are not (or not yet) part of R. One could argue this on philosophical or general epistemic grounds – since the future has not yet been determined, it cannot yet be real or known. Actually, though, my basis for this position is linguistic. I am not engaged here in a philosophical exercise. Rather, I am trying to explicate the tacit cognitive model underlying English clausal grounding, i.e. its grammaticized tense-modal system. In the basic English system, future is marked by the modal will. A fundamental feature of the model I am proposing is precisely that the presence of a modal indicates 14. An example would be the sentence Cheney knows that Rice is convinced that Bush is a genius.

The English present 

epistemic striving, as opposed to epistemic control, and R is simply defined as what is under epistemic control. This is not the only possible way to deal with future occurrences – other languages may handle them differently – but it makes sense and appears to be how English is doing it.15 4.3

Non-modal clauses

That brings us back to tense. Let us start with the simplest case: non-modal clauses describing actual occurrences. The absence of a modal indicates that the profiled occurrence (P) is accepted as real. What, then, is the import of “present” vs. “past”, which I have characterized more generally as immediate vs. non-immediate? Since P is confined to R (due to the absence of a modal), the obvious proposal is that immediate vs. non-immediate is a matter of whether P is located in immediate reality (IR) or its complement, “non-immediate reality”. That is indeed what I propose. This is fundamentally an epistemic characterization, for R is defined as what C (by default, the speaker) accepts as valid (or “knows”). Clearly, though, the distinction is equivalent to a temporal one, since immediate reality consists of those facets of R pertaining to the world at the present time (the locus of current experience). This is the basis for my previous description (Langacker 1991) of present vs. past time as being the prototypical values of the more general immediate vs. non-immediate opposition. The English present and past (for non-modal expressions) are sketched in Figure 15. In both cases, the time of speaking delimits what counts as immediate experience and immediate reality for this purpose. For the present tense, the time of speaking also defines the immediate temporal scope for apprehending the profiled clausal process, P.16 The difference between present and past tense is thus a matter of whether P is found in immediate reality or non-immediate reality. The past tense involves a slightly greater epistemic distance because the epistemic path crosses an additional boundary (IR). These configurations are basically equivalent to those in Figure 7 and yield the same explanation for the distribution of present and past with perfective and imperfective verbs. A non-immediate perfective is always acceptable because the immediate temporal scope can always be long enough to encompass P (e.g. She drove 15. This discussion pertains to the grammaticized tense-modal system. English does have other ways of dealing with the future, e.g. with gonna, is to, or has to, not to mention the scheduled future use of the present. I consider these to represent different ways in which constraints on future occurrences are incorporated as part of present reality. 16. Of course, the time of speaking is actually a temporal interval, not just a point, as the diagrams might suggest. If only time is considered, the box enclosing P can be eliminated in diagram (a), since IS = IR in this domain. Other domains do however figure in P’s conception.



Ronald W. Langacker (a) Immediate (“Present”)

(b) Non-immediate (“Past”)

MS P

MS

IS P

IS C

C R

IR

R

IR

Figure 15.â•‡

to Michigan). An immediate perfective is usually problematic (*She drives to Michigan) owing to the durational and epistemic problems. Imperfectives can either be immediate (She is in Michigan) or non-immediate (She was in Michigan) because whatever portion of the overall process is put onstage and profiled itself constitutes a valid instance of the process type. And performatives still work, despite being immediate perfectives, because P is the speech event itself, hence an aspect of immediate experience. With some justice, one might argue that I have merely rephrased the temporal coincidence account in terms of whether P is or is not in IR, where IR itself is defined in terms of temporal coincidence with the time of speaking. More charitably, one could say that I have sketched an epistemic approach from which the temporal coincidence account can be derived. Either way, more can be said from an epistemic standpoint. As noted earlier, the temporal characterization of perfective vs. imperfective processes implies a difference in their typical mode of apprehension. Though flexible in size and scale, subjectively the immediate temporal scope is always bounded (like the visual field) – there is a limit to the temporal expanse we can invoke at one time as the locus for focused viewing. Being temporally bounded itself, a perfective process will always fit within IS just by making this large enough. It is therefore possible for a perfective occurrence to be apprehended in its entirety. And since bounding is inherent in the characterization of a perfective process type, full apprehension – including the endpoints – is required to identify an instance of the type. By contrast, an imperfective process with indefinite temporal extension cannot be apprehended in its entirety, for there is no entirety. All we can do is observe that portion of the overall process which happens to fall within whatever temporal scope we adopt for viewing purposes. In other words, the natural mode of apprehension for imperfectives is one of sampling. But since an imperfective process is conceived as exhibiting qualitative uniformity throughout, a sample is enough to identify it and counts as an instance of the process type.

The English present 

These contrasting modes of apprehension have different epistemic consequences for what lies outside the immediate temporal scope. Being apprehended in its entirety, within IS, a perfective occurrence tells us nothing about what might occur beyond its limits. On the other hand, since the profiled portion of an imperfective is merely a sample, with no evident bounding, the presumption is that the overall process extends beyond IS in both directions. So at least in local terms, imperfectives correlate with stability both within and beyond the immediate scope of observation. The presumed stability they thereby offer does not necessarily rise to the level suggested by Brisard’s term “epistemic certainty”. It does however amount to a kind of “givenness” (also his term), a local framework with respect to which more transient occurrences can be assessed. On this basis Brisard (2002) describes all imperfective clauses as being “structural” statements. Even contingent situations, stable only when viewed on a very small time scale, qualify as structural knowledge by this definition (e.g. I see it now; The cat is on the mat; He’s straightening his tie). I myself use the term more narrowly, notably for generics, but I acknowledge that we are dealing with a gradation rather than discrete categories (cf. Langacker 1997). Regardless of terminology, we will see that this notion of “givenness”, or “basis for assessment”, is relevant to the meaning of the English present. Let us now extend this epistemic account of the present to so-called “nonpresent” uses. I have described these as involving the distinction between a representing event and a represented event. In these expressions, I propose, what is directly coded linguistically is the virtual representing event, as opposed to the actual event(s) it represents. These uses of the present all conform to the characterization proposed, i.e. the profiled occurrence (P) is found in immediate reality (IR). What differs is the nature of the mental construction in which it figures. I briefly examine three cases: the scheduled future, singular generics, and the historical present. The first point to make about the scheduled future, one already noted, is that the mental construction involved is independent of the present tense. Scheduled future expressions are possible for the past, the present, or even the future, as seen in (16). In the last sentence, the intended interpretation – an entirely possible one – is that will indicates futurity of the schedule (not the actual event itself), i.e. at 3 it is appropriate to say Our plane leaves at 6. (16) First our plane left at noon. Then they rescheduled it, so now it leaves at 3. I’m sure they will reschedule it again, so that when 3 rolls around it will leave at 6. Use of the present in this construction therefore does have to do with the time of speaking, in the sense that it indicates that the schedule being consulted is in effect at the present moment. While it is in effect, the entire schedule can be consulted at any time. A particular expression, e.g. Our plane leaves at 6, is then a matter of “reading



Ronald W. Langacker

off ” (or accessing) one schedule entry, whose apprehension is coded linguistically. This is sketched in Figure 16(a). The schedule is represented as a box labeled V, to indicate its status as a virtual entity (which is not to deny that some schedules have physical or at least visual instantiation). The boxes inside it stand for schedule entries, i.e. virtual events representing anticipated actual events. Among these virtual events is the one profiled by a particular sentence, like Our plane leaves at 6. This situation conforms to the epistemic characterization of the present: that the profiled occurrence (P) is located in immediate reality (IR). This despite its virtual nature – recall that the speaker’s conception of reality includes many virtual entities, not just actual ones. Here the speaker accepts it as valid that the schedule is currently in effect.17 The case of a past-tense scheduled future, e.g. First our plane left at noon, is shown in diagram (b). It is simply a matter of referring to a mental schedule that was in effect at an earlier time. Granted this conceptual configuration, the expression conforms to the epistemic characterization of past tense: P is located in nonimmediate reality.18 What about a future scheduled future, like the last clause in (16)? It simply involves the schedule (hence P) being projected as in effect at a future time, in the manner specified by will or another modal. Singular generics work analogously.19 The mental construction involved is like a virtual plan or schedule, except that its entries – instead of representing specific (a)

(b) MS

MS V

P

IS

V

P

IS C

C R

IR

R

IR

Figure 16.â•‡ 17. Does P precisely coincide with the time of speaking, as in the temporal account? In this case (as opposed to the situation with actual events), I think the point is moot. P is part of a schedule that is in effect throughout the time of speaking, and its apprehension is taken as coinciding with the time of speaking. However, since I am not (in the epistemic approach) talking about a virtual occurrence of P, the time of such an occurrence is not at issue. 18. Observe that this does not specify whether the scheduled event was anticipated to occur before or after the time of speaking. It merely has to be subsequent to the time when the schedule was consulted (given the very nature of a schedule). This consequence “falls out” from the characterization proposed. 19. Singular and plural generics represent different strategies of generalization (Langacker 1997). To keep things simple, I confine attention to the former.

The English present 

actual events – are virtual occurrences taken as representing facets of the world’s essential nature. Each of these structural generalizations projects to any number of actual occurrences for as long as it remains valid. We see in (17) that even these “structural statements” may have a limited shelf life, so that we have to specify the span of time when they are in effect. Actually, the diagrams in Figure 16 work perfectly well for singular generics in the present and past tense. We simply have to interpret the box labeled V as describing the world’s essential structure (i.e. its entries map onto actual events in a different way than with a schedule). (17) In the good old days, a woman obeyed her husband. Now things are all messed up: a man obeys his wife. But in the future, when we put all this craziness behind us, a woman will obey her husband again. The historical present is sketched in Figure 17. It too conforms to the epistemic definition, as it profiles a virtual event located in immediate reality. Here, though, this event is an aspect of C’s internal experience, consisting in the recall (or mental replay) of a previous occurrence, given as P′. Observe that P′ is shown as part of E, i.e. it is something C previously experienced, though generally not internally. On the other hand, the recall is necessarily internal.20 I conclude that the epistemic characterization works for both actual and virtual occurrences, given a proper description of the mental constructions in which the latter figure. In all cases, the English present situates P in immediate reality, the past in non-immediate reality. But we have not yet considered modals, which MS

V IS P

P′

C E

IE

R

IR

Figure 17.â•‡

20. The case of watching a video replay on a monitor is analogous, except that P is external to C within IE (being apprehended via perception), and P′ need not be part of E (i.e. C may not have witnessed the original event).



Ronald W. Langacker

specify that P has not yet been accepted as real. It is still a target of epistemic striving, the different modals indicating the degree of epistemic force involved. Consideration of modal expressions leads to a more schematic characterization of the immediate/non-immediate opposition. 4.4

Modals

I am only concerned with the core of the grammaticized modal system: the elements may, can, will, shall, and must, and their non-immediate counterparts might, could, would, and should.21 Except as necessary, I will not describe these forms individually, the focus here being their common properties. I am not claiming that the non-immediate forms are fully analyzable or derived in a wholly regular fashion from their basic counterparts; each modal has its semantic peculiarities and its own ecological niche in the epistemic landscape. Still, there are general points to be made. Just a word about the distinction between root and epistemic senses of the modals, exemplified in (18a,b). Involving notions like obligation and permission, root modals exhibit more clearly the force-dynamic character of modals (Sweetser 1982; Talmy 1988; Langacker 1991). Ranging in degree from the absence of a barrier (may) to compulsion (must), the force is generally manifested socially in the case of root modals, mentally with epistemic modals. Both involve what we might call “the evolution of reality”, as well as a striving for control, as sketched in Figure 11(a). The difference pertains to the level at which this occurs (Langacker 2008). Root modals are aimed at effective control – determining what happens in the world itself. And obviously, epistemic modals are aimed at epistemic control – evolution in our knowledge of the world. They register the force subjectively experienced as we mentally simulate reality (R) “growing” in such a way that it comes to encompass the profiled event (i.e. we accept the profiled occurrence as valid). Here we are only concerned with the epistemic modals. (18) a. Root modals: You {may/should/must} report the theft. b. Epistemic modals: It {may/should/must} be hot in Chicago. A key observation is that the morphological distinction immediate vs. non-immediate is manifested on different elements depending on whether or not the clause contains a modal. In the absence of a modal, it is marked on the verb, as present vs. past tense (which I have argued to be a reasonable characterization even in the epistemic approach). With a modal, however, immediate vs. non-immediate is not 21. In Langacker (1991: §6.2.2.3), I give an argument that must simply lacks a non-immediate form.

The English present 

marked on the verb, but rather by the form of the modal itself: may vs. might, will vs. would, etc. Semantically as well, so-called “tense” does not apply directly to the verbal process (P), but only indirectly. It is rather the modal, in either its immediate or non-immediate form, which applies to P and specifies its epistemic status. This is one respect in which the epistemic approach is advantageous relative to the temporal coincidence approach. With modals, the “temporal” component of grounding does not even apply to P, so it can hardly be said to specify P’s coincidence or non-coincidence with the time of speaking. And while some modal uses are present-oriented, as in (18b), very often P is future relative to the speech event (e.g. It {may/might} rain tomorrow), in which case immediate vs. non-immediate certainly does not indicate temporal coincidence vs. precedence vis-à-vis the time of speaking. This is not to say that the non-immediate modals are unanalyzable or that the non-immediacy they convey has no connection with past time. They do correlate with past time in the so-called “sequence of tenses” phenomenon, exemplified in (19), where use of an immediate form is subsequently reported with a non-immediate form. Now, English does not actually have a sequence-of-tenses rule (Langacker 1991: §6.2.2.2). It is however important to understand why the nonimmediate form appears in these reports. In the case of (19a), without a modal, it is simply a matter of the situation described being in the past relative to the time of speaking; the speaker describes the reported situation from her own temporal vantage point. But with a modal, in (19b,c), the temporal location is indeterminate – from the perspective of the current speaker, the projected time of helping could be in the past or the future. Why, then, is the non-immediate form of the modal employed?22 (19) a. Tom says “I am tired”. b. Tom says “I may help”. c. Tom says “I will help”.

Report: Tom said he was tired. Report: Tom said he might help. Report: Tom said he would help.

We must first consider immediate forms of the epistemic modals, e.g. the original use of may and will in (19b,c). This is represented in Figure 18(a). I have omitted the arrow for time, since with modals time is clearly secondary. When marked by a modal, P can be located in the present, as in (18b); in the future, as when Tom says I will help; or even in the past, as in Tom said he would help yesterday. The essential factor is that the modal situates P outside reality. Hence the single dashed 22. In all these cases the immediate form can also be used, showing the non-existence of a sequence-of-tenses rule. The epistemic status thus indicated is always appropriate from the speaker’s perspective. For instance, Tom said he may help – where may indicates future potentiality – can only be used if P is still future relative to the current time of speaking.



Ronald W. Langacker (a)

MS C

R

IR

P

(b)

IS

MS

C′ R′

P

IR′

R

IS

C IR

Figure 18.â•‡

arrow in diagram (a), representing the epistemic path from C to P, crosses the boundary of R. The double arrow indicates the nature or degree of the epistemic force, distinguishing one modal from another. Thus may specifies that the evolution of reality – as mentally simulated by C – has the potential of it “reaching” and encompassing P (there is no barrier to this happening), whereas will specifies that R is projected (or predicted) to reach it. Let us now consider non-immediate forms in reported speech, e.g. might and would in (19b,c). The key factor, clearly, is that the current speaker (C) does not herself assume responsibility for the epistemic assessment, but is rather reporting on the assessment previously made by another conceptualizer, C′. As shown in Figure 18(b), C accepts as real that C′ has previously made this assessment, using his own conception of reality (R′) as its basis. That assessment placed P outside R′, its precise status being indicated by the base form of the modal (may or will). Thus P is not per se part of C’s conception of reality (R). It figures in R only derivatively: C accepts as real the fact that C′ had an epistemic stance in regard to P, namely the stance of not accepting it as real. In short, as shown by the arrows, C reaches P only via a two-step epistemic path leading through the mind of another conceptualizer (C′). We need to be precise about the modal and how it relates to the situation shown. First, it has to be recognized that P is not itself part of R, even though it appears inside R in the diagram, in contrast to Figure 16, where P is understood as being real (albeit virtual). The difference is that, in Figure 16, P is an entry on an accepted virtual schedule which C herself accesses directly, while in Figure 18(b) C only reaches P by apprehending someone else’s epistemic judgment. Given the limitations of a two-dimensional representation, to interpret these diagrams properly we have to consider the epistemic path (not just spatial inclusion). In Figure 16, the epistemic path leads directly from C to P (though it may cross boundaries), whereas in 18(b) the path from C only leads to another conceptualizer, C′, from which a second epistemic path leads to P. We can go even further and isolate the contributions of each path in determining the modal form. Specifically, the first path – anchored by C – is responsible for

The English present 

the choice of the non-immediate forms, might and would, instead of may and will. The choice of may or will, as the modal base, reflects the second path, anchored by C′. That is, the modal base (may or will) indicates the epistemic assessment made by C′, on the basis of his own conception of immediate reality (IR′). The speaker expresses this assessment with the non-immediate might or would because the speaker’s own conception of immediate reality (IR) does not support the assessment coded by the modal base. In short, the basis for the modal assessment (may vs. will) is IR′, and IR′ is non-immediate to C. It is non-immediate to C in two respects: first, it is earlier in time; and second, it is the reality conception of another individual. This apportionment is the key to understanding other uses of the non-immediate modals, as in (20). I suggest that in each case the non-immediacy coded by might indicates that IR – the speaker’s conception of immediate reality – is not the basis invoked for the assessment associated with the modal base, may. In each case, the potentiality expressed by may is calculated relative to a conception of immediate reality, IR′, distinct from the speaker’s actual one. Using might (in lieu of may) indicates the non-immediacy of this presumed basis of modal assessment. As shown in Figure 19, the speaker (C) invokes a virtual situation, involving an imagined version of immediate reality (IR′), with respect to which the modal assessment may would be appropriate. Just as in Figure 18(b), might signals the nonimmediacy from IR of the assumed basis for modal assessment (IR′). (20) a. If it weren’t so cloudy it {*may/might} get hot. b. When the clouds disperse it {may/might} get hot. c. Even though it’s cloudy it {may/might} get hot. This is most evident in (20a), a counterfactual conditional. The counterfactual construction invokes a mental space specifically portrayed as being distinct from reality. The import of (20a) is that the modal assessment may would be appropriate given an imagined conception of immediate reality (IR′), namely immediate reality as it would appear to C in the hypothetical situation. There is thus a two-step MS V C

C IR′

IR

Figure 19.â•‡

P

IS



Ronald W. Langacker

epistemic path: from IR to IR′, and from IR′ to P. It is further presumed that P is only accessible via this path, i.e. it would not occur except in the counterfactual situation.23 For this reason may is not a permissible alternative. In (20b), the when-clause sets up a mental space associated with the specified time (the time when the clouds disperse). Because the situation has not yet been realized, it is still (broadly speaking) a virtual one. Its realization is however anticipated – in contrast to (20a), where the path of evolving reality excludes both the situation and its consequences. Hence there are two different ways of making the modal assessment. On the one hand, it can proceed as shown in Figure 19, resulting in might. As the basis for modal assessment, the speaker invokes the imagined conception of immediate reality (IR′) expected to obtain at the time specified by the when-clause (a reality in which the clouds have dispersed). On the other hand, because that situation is projected as occurring, the potentiality expressed by may is equally appropriate whether IR′ or IR itself is taken as the basis for assessment. With may, in other words, P (it getting hot) is directly accessed from IR via a one-step epistemic path, just as in a single-clause expression (It may get hot). In this case the function of the when-clause is simply to specify P’s temporal location. Finally, in (20c) the adverbial clause does not itself invoke a virtual situation, but merely describes an aspect of IR. The adverbial has no direct bearing on the choice of may vs. might, since either would be appropriate even without it. With may, IR itself is the basis for modal assessment, as in Figure 18(a). What about might? Does it still imply a virtual basis for assessment, as in Figure 19, despite the absence of a clause that explicitly introduces it? I suggest that it does. Choosing the non-immediate form (might instead of may) itself signals that the potentiality normally conveyed by may is not an appropriate assessment from IR as presently constituted. No other mental space being indicated, the basis for assessment may simply be an alternative to IR that differs from it in relatively minor respects. In (20c), for instance, the alternative IR′ could be just like IR except for there being fewer clouds. The assessment It may get hot would then be supported, but since that situation does not in fact obtain, might is used instead to register its non-immediacy. This is essentially equivalent to saying that might expresses a more tenuous potentiality than does may. In (21), I summarize the proposed epistemic description of the immediate/nonimmediate opposition in English. In the absence of a modal, where P is confined to reality (R), immediacy implies that P is in IR. Obviously, then, non-immediacy 23. This feature of the counterfactual construction can however be overridden: If it weren’t so cloudy, it might get hot – but it may get hot anyway. Note that but and anyway indicate the suspension of this feature.

The English present 

indicates that P is not in IR. We can take this as being the default situation (the prototype), in which immediate vs. non-immediate is largely equivalent to present vs. past in time. In combination with a modal, where P is excluded from R, immediacy implies that IR is the direct basis for the modal assessment of P. Here non-immediacy indicates that some other basis (IR′) must be found. These two characterizations clearly have a lot in common. From them we can abstract the schematic description in (21c), which covers both modal and non-modal clauses. In both cases, immediacy implies that IR is the direct basis for P’s epistemic assessment – either containing P or providing direct access to it (via a one-step epistemic path). (21) a. In the absence of a modal, where P is confined to R, (non-)immediacy implies that P is (not) in IR. [prototype] b. In combination with a modal, where P is excluded from R, (non-)immediacy implies that IR is (not) the direct basis for the modal assessment of P. c. In either case, (non-)immediacy implies that IR is (not) the direct basis for the epistemic assessment of P. [schema] 5. Conclusion This has been a preliminary attempt at pinning down the specifics of an epistemic account of the English present. While fairly comprehensive, it is certainly not exhaustive. I have not, for instance, considered the use of the present in adverbial clauses, e.g. the initial clause in (20b).24 Two possible advantages over the temporal coincidence account have come to light. First, it avoids the need to talk about the “virtual occurrence” of events, a notion that may be defensible but might be considered dubious. Second, it is readily extended to clauses with modals, whereas temporal coincidence cannot be. That the epistemic approach would prove advantageous is hardly a surprise. My time-based account of the English present (Langacker 2001) was limited to nonmodal clauses, where the label “tense” is not altogether inappropriate. Here immediate vs. non-immediate is essentially equivalent to present vs. past. With modals, however, immediacy vs. non-immediacy requires a non-temporal characterization. Thus, in my overall description of English clausal grounding (Langacker 1991: ch. 6), I characterized it schematically as a “proximal”/“distal” contrast, with present and past tense representing prototypical values.

24. An analysis of this use that conforms to (21) is proposed in Langacker (2007).



Ronald W. Langacker

I conclude that the English present is reasonably described as having “present time” (coincidence with the time of speaking) for its prototypical value, and “epistemic immediacy” as its schematic characterization (valid for all instances). It thus exemplifies a general finding of Cognitive Grammar concerning the meanings of grammatical elements: many such elements have both prototypical and schematic semantic values, the latter consisting in some mental experience inherent in the former. For instance, the schematic import of possessives – that of invoking one element as a “reference point” to mentally access another – is immanent in the possessive prototypes of ownership, kinship, and whole-part relations (Langacker 1993, 2004b). More controversially, I have claimed that nouns are characterized schematically in terms of contractive mental operations inherent in the conception of physical objects, the category prototype (Langacker 1987b). The English present can be seen as another instance of this general pattern: epistemic immediacy is an aspect of the mental experience inherent in the apprehension of present-time occurrences. References Austin, John L. 1962. How to Do Things with Words. Cambridge, MA: Harvard University Press. Barsalou, Lawrence W. 1999. Perceptual Symbol Systems. Behavioral and Brain Sciences 22: 577–660. Bergen, Benjamin. 2005. Mental Simulation in Literal and Figurative Language Understanding. In The Literal and Nonliteral in Language and Thought [Studies in Language 11], S. Coulson & B. Lewandowska-Tomaszczyk (eds.), 255–278. Frankfurt am Main: Peter Lang. Brisard, Frank. 1999. A Critique of Localism in and about Tense Theory. PhD Dissertation, University of Antwerp, Belgium. Brisard, Frank. 2001. Be going to: An Exercise in Grounding. Journal of Linguistics 37: 251–285. Brisard, Frank. 2002. The English Present. In Grounding: The Epistemic Footing of Deixis and Reference [Cognitive Linguistics Research 21], F. Brisard (ed.), 251–297. Berlin & New York: Mouton de Gruyter. Evans, Vyvyan (ed.). 2004. The Structure of Time: Language, Meaning and Temporal Cognition [Human Cognitive Processing 12]. Amsterdam & Philadelphia: John Benjamins. Fauconnier, Gilles. 1985. Mental Spaces: Aspects of Meaning Construction in Natural Language. Cambridge, MA & London: MIT Press/Bradford. Fauconnier, Gilles. 1997. Mappings in Thought and Language. Cambridge: Cambridge University Press. Fauconnier, Gilles & Eve Sweetser (eds.) 1996. Spaces, Worlds, and Grammar. Chicago & London: University of Chicago Press. Fauconnier, Gilles & Mark Turner. 2002. The Way We Think: Conceptual Blending and the Mind’s Hidden Complexities. New York: Basic Books.

The English present  Goldsmith, John & Erich Woisetschlaeger. 1982. The Logic of the English Progressive. Linguistic Inquiry 13: 79–89. Johnson, Mark. 1987. The Body in the Mind: The Bodily Basis of Meaning, Imagination, and Reason. Chicago & London: University of Chicago Press. Lakoff, George & Mark Johnson. 1980. Metaphors We Live By. Chicago & London: University of Chicago Press. Lakoff, George & Rafael E. Núñez. 2000. Where Mathematics Comes From: How the Embodied Mind Brings Mathematics into Being. New York: Basic Books. Lakoff, George & Mark Turner. 1989. More than Cool Reason: A Field Guide to Poetic Metaphor. Chicago & London: University of Chicago Press. Langacker, Ronald W. 1987a. Foundations of Cognitive Grammar, vol. 1, Theoretical Prerequisites. Stanford: Stanford University Press. Langacker, Ronald W. 1987b. Nouns and Verbs. Language 63: 53–94. Langacker, Ronald W. 1990. Concept, Image, and Symbol: The Cognitive Basis of Grammar [Cognitive Linguistics Research 1]. Berlin & New York: Mouton de Gruyter. Langacker, Ronald W. 1991. Foundations of Cognitive Grammar, vol. 2, Descriptive Application. Stanford: Stanford University Press. Langacker, Ronald W. 1993. Reference-Point Constructions. Cognitive Linguistics 4: 1–38. Langacker, Ronald W. 1997. Generics and Habituals. In On Conditionals Again [Current Issues in Linguistic Theory 143], A. Athanasiadou & R. Dirven (eds.), 191–222. Amsterdam & Philadelphia: John Benjamins.. Langacker, Ronald W. 1999a. Grammar and Conceptualization [Cognitive Linguistics Research 14]. Berlin & New York: Mouton de Gruyter. Langacker, Ronald W. 1999b. Virtual Reality. Studies in the Linguistic Sciences 29 (2): 77–103. Langacker, Ronald W. 2001. The English Present Tense. English Language and Linguistics 5: 251–271. Langacker, Ronald W. 2002a. Deixis and Subjectivity. In Grounding: The Epistemic Footing of Deixis and Reference [Cognitive Linguistics Research 21], F. Brisard (ed.), 1–28. Berlin & New York: Mouton de Gruyter. Langacker, Ronald W. 2002b. Remarks on the English Grounding Systems. In Grounding: The Epistemic Footing of Deixis and Reference [Cognitive Linguistics Research 21], F. Brisard (ed.), 29–38. Berlin & New York: Mouton de Gruyter. Langacker, Ronald W. 2004a. Aspects of the Grammar of Finite Clauses. In Language, Culture and Mind, M. Achard & S. Kemmer (eds.), 535–577. Stanford: CSLI Publications. Langacker, Ronald W. 2004b. Possession, Location, and Existence. In Linguagem, Cultura e Cognição: Estudios de Linguística Cognitiva, A. Soares da Silva, A. Torres, & M. Gonçalves (eds.), vol. I, 85–120. Coimbra: Almedina. Langacker, Ronald W. 2005. Dynamicity, Fictivity, and Scanning: The Imaginative Basis of Logic and Linguistic Meaning. In Grounding Cognition: The Role of Perception and Action in Memory, Language and Thinking, D. Pecher & R. A. Zwaan (eds.), 164–197. Cambridge: Cambridge University Press. Langacker, Ronald W. 2007. The Present Tense in English Adverbial Clauses. In Cognition in Language: Volume in Honour of Professor Elzbieta Tabakowska, W. Chłopicki, A. Pawelec, & A. Pokojska (eds.), 179–209. Cracow: Tertium. Langacker, Ronald W. 2008. Enunciating the Parallelism of Nominal and Clausal Grounding. In Du fait grammatical au fait cognitif [From Gram to Mind: Grammar as Cognition], vol. 1,



Ronald W. Langacker J.-R. Lapaire, G. Desagulier, & J.-B. Guignard (eds.), 17–65. Pessac: Presses Universitaires de Bordeaux. Matlock, Teenie. 2004. Fictive Motion as Cognitive Simulation. Memory and Cognition 32: 1389–1400. Matsumoto, Yo. 1996a. How Abstract is Subjective Motion? A Comparison of Coverage Path Expressions and Access Path Expressions. In Conceptual Structure, Discourse and Language, A. E. Goldberg (ed.), 359–373. Stanford: CSLI Publications. Matsumoto, Yo. 1996b. Subjective-Change Expressions in Japanese and Their Cognitive and Linguistic Bases. In G. Fauconnier & E. Sweetser (eds.), 124–156. Sweetser, Eve E. 1982. Root and Epistemic Modals: Causality in Two Worlds. Proceedings of the Annual Meeting of the Berkeley Linguistics Society 8: 484–507. Sweetser, Eve E. 1997. Role and Individual Interpretations of Change Predicates. In Language and Conceptualization [Language, Culture and Cognition 1], J. Nuyts & E. Pederson (eds.), 116–136. Cambridge: Cambridge University Press. Talmy, Leonard. 1988. Force Dynamics in Language and Cognition. Cognitive Science 12: 49–100. Talmy, Leonard. 1996. Fictive Motion in Language and “Ception”. In Language and Space, P. Bloom, et al. (eds.), 211–276. Cambridge, MA & London: MIT Press/Bradford. Turner, Mark. 1987. Death is the Mother of Beauty: Mind, Metaphor, Criticism. Chicago: University of Chicago Press. Vendler, Zeno. 1967. Linguistics in Philosophy. Ithaca: Cornell University Press.

The organization of the German clausal grounding system Elena Smirnova

Leibniz University Hanover This chapter examines the verbal categories of German in terms of grounding predications. Using the broad notion of deixis as a starting point for the examination, it first proposes a unified description of grounding predications as a particular class of deictic expressions. Second, it suggests that three general classification criteria – the nature of the grounded entity, the nature of the grounding relation, and the construal configuration of the ground – constitute a model which may be applied on grounding predications in different languages in order to discover their similarities and differences. The proposed model is applied to the German clausal grounding system, concentrating on German temporal and modal grounding predications. Keywords: grounding, deixis, construal, verbal categories, German

1. Introduction In this chapter, I try to shed some light on the German clausal grounding system. More specifically, I attempt to establish a systematic view of German temporal and modal grounding which I suggest are different in many ways from the English grounding system described by Cognitive Grammar. The chapter starts with some theoretical explorations. First, I discuss the notion of grounding in relation to the broader conception of deixis and deictic signs. I offer general classification criteria that may be applied to every grounding predication insofar as it represents a particular class of deictic expressions. Furthermore, I deal with the notion of construal. I put forward the hypothesis that there exist alternative ways of deictically construing the ground, i.e. the immanent reference point every grounding predication refers to. Albeit remaining unprofiled and implicit, the ground, being differently construed, plays a crucial role in the interpretation of the status of grounding predications.



Elena Smirnova

Applying the theoretical considerations made in the first part of the chapter to German grounding predications, I explore how the German clausal grounding system is organized. In doing so, I focus first on temporal grounding, investigating relevant characteristics of the temporal deictic center and of different temporal values of German tenses. Second, I concentrate on the modal grounding system, arguing that, although being closely connected, the verbal mood markers and the epistemically used modal verbs should be best described as different grounding (sub-)systems in German. 2. Deictic characteristics of grounding predications In Cognitive Grammar, a grounding predication is defined as “a predication that locates a thing or process with respect to the ground in fundamental ‘epistemic’ domains pertaining to reality and speaker/hearer knowledge” (Langacker 1991a: 548). Grounding predications are thus considered a special class of deictic expressions since they anchor a conceptualized entity to the ground, i.e. the speech event, its participants, and the immediate circumstances of the utterance. Grounding predications differ from the overall class of deictic expressions by virtue of some distinct features (cf. Langacker 2002a: 8 ff.). They are indispensable to the formation of a nominal or finite clause from nouns and verbs. They are grammatical, as opposed to lexical, elements. Therefore, their semantics tend to be abstract and schematic in nature. They neither mention the (aspect of the) ground nor the grounding relation explicitly: “rather than being profiled, the ground remains implicit and serves as a reference point” (Langacker 2002a: 10). However, in spite of these differences, both linguistic elements – (the more lexical) deictic expressions like I, here, now and (the more grammatical) grounding predications like tense, mood markers, and demonstratives – share some symptomatic features, since they belong to a common broader class of deictic signs. The notion of deixis comes from Bühler (1934) and rests on the assumption that there is a Zeigfeld, i.e. a deictic space or field, which is made up of the I-herenow orientation of the utterance, including both speaker and addressee(s) who interpret the linguistic symbols (constituting the Symbolfeld) with reference to this field. Therefore, every act of linguistic communication proceeds and is anchored in the Zeigfeld. The Zeigfeld is seen as the actual situation frame, as well as the current mental space of vision, centered on the speaker. The speaker’s vantage point is the origo, or the I-here-now starting point in the centre of the deictic coordinate system. From this ‘point of viewing’, the speaker usually makes his statements, and the hearer succeeds in understanding them, relating what is said to that same zero-point.

The organization of the German clausal grounding system 

Consequently, deictic signs are generally characterized by their dependence on the communication context, or the current speech situation (cf. e.g. Bühler 1934, Fillmore 1997, Langacker 1991, Hanks 1990). They anchor linguistic contents to the deictic zero-point, which is normally (a certain aspect of) the speech situation. In doing so, they also encode a semantic relation between linguistic contents and the (aspect of the) context. The former function is often considered the relational, or reflexive, aspect of a deictic sign; the latter describes the referential, or denotative, aspect. Hence, the most important feature of a deictic sign is that it is relational in nature. As far as such a relation falls within the scope of investigation, (at least) three aspects are in focus: (i) the entity being related to the origo, i.e. the referent of a deictic sign, (ii) the relation itself, i.e. its semantic value, defining the respective deictic dimension, and (iii) the point of reference, i.e. the origo. For instance, when describing personal pronouns, one needs to define (i) the referent of this sign, which is a person, (ii) the relational content of the sign, which is the specification of the communicative role (of this person), and (iii) the reference point the sign refers to, which is the speaker’s role in the speech event. Likewise, when specifying the spatial deictic here (or there), one has to diagnose that it is about (i) a location (ii) in space relative to (iii) the current location of the speech event. Similarly, in Cognitive Grammar, every grounding predication is claimed to establish a link between a grounded entity and the ground. In practice, it very often coincides with a link to the speaker, who functions as a central point of reference within the ground. The link to the ground, i.e. the grounding relationship, however, is not of the same nature for every grounding predication. Langacker (2003) claims that speaker’s and hearer’s specific conceptualizations, i.e. ‘viewing’, of the profiled entity constitute grounding relationships. The nature of grounding relationship distinguishes one grounding predication from another: [The ground] is implicit and subjectively construed, serving almost literally as the locus of ‘viewing’, from which the speaker and hearer direct their attention to the profiled entity onstage. Although the distinctive semantic value of a grounding element resides in the profile’s status and location with respect to the ground, this grounding relationship is not per se the focused object of conception. Rather, it is inherent in the ‘viewing arrangement’ in terms of which the speaker and hearer conceptualize the nominal or clausal referent. (Langacker 2003: 7)

Thus, linking this to the general aspects of a relational deictic sign, I propose that the aforementioned threefold schema applies to grounding predications: (i) the referent of a deictic sign is the profiled entity, (ii) the deictic relational content and the respective deictic dimension are defined by the grounding relation, and (iii) the deictic center is the (relevant aspect of the) ground. I try to show that, while two of the three proposed levels of classification have been accounted for in the

 Elena Smirnova

Cognitive Grammar view of grounding predications, little attention has been paid to the third one, namely the ground itself. It has been assumed that the ground, defined as the situation of speech and very often equivalent with the speaker, builds a homogeneous point of reference. Without being fixed in space or time, the ground is terminologically conceived of as a set of some elements and dimensions. What has been seen to be crucial for the structuring and understanding of a grounding predication are (i) the (semantic) nature of the referent and (ii) the nature of the grounding relation, cf.: The aim of grounding a predication is to establish mental contact with, or direct someone’s attention to, a referent which discourse participants are presumably able to determine, given 1) the semantic content of the “phrase” to which the grounding predication attaches, and 2) the nature of the grounding relation proper. (Brisard 2002: xv)

I suggest, however, that all three aspects of a grounding predication should be integrated. Specifications of these three basic aspects underlie different status of grounding predications: 1. The distinction between clausal and nominal grounding predications resides in the nature of the grounded entity. The basic semantic opposition consists of two linguistic entities: it is either a verb, designating a process, or a noun, designating a thing; 2. Distinctions in the nature of the grounding relationship underlie conceptual distinctions between grounding predications. As I noted above, the grounding relationship defines the deictic dimension in which grounded entities are situated. Moreover, semantic values of single grounding predications specify certain locations in the respective deictic space. Generally, the nominal grounding relationship concerns the accessibility of things, whereas clausal grounding revolves around the existence of processes. For nominal grounding, definite and quantificational grounding predications are distinguished. For English, the demonstratives this, that, these, and those, the articles the, a and some, and certain quantifiers like some, most, all, no, each, every, any are identified as nominal grounding elements (Langacker 2002b). For clausal grounding, modal and temporal grounding predications are distinguished.1 As for temporal grounding, a present tense signals that the designated event is immediate to the ground, whereas a past tense morpheme is associated with non-immediacy, i.e. with some kind of (epistemic or temporal) distance between the 1. As far as I know, the status of evidential markers as grounding predications has not been a matter of discussion so far. I would suggest, however, that evidentials serve as grounding predications of the third type, in addition to temporal and modal grounding predications.

The organization of the German clausal grounding system 

ground and the profiled event (see Langacker 1991a, b, 2002b, Brisard 2002). As for modal grounding, the absence of a modal indicates that the profiled event is part of reality, whereas the presence of a modal verb locates the profiled event in the realm of non-reality (Langacker 2002b). 3. The third element, which is prominent in every description of a grounding predication, is the ground itself. Albeit the link to the ground is implicit and a grounding predication does not actually profile the ground, different conceptualizations of the ground, as I suggest, may determine different status of grounding predications. This can be seen parallel to the role of the grounding relationship which, despite being not explicitly profiled by a grounding predication, underlies its conceptual status. Similarly, a certain construal of the speech situation, especially regarding conceptualizations of the speaker’s role in it, may influence distinctions between grounding predications. Such distinctions may be found either in specific contextual interpretations, which a grounding predication allows for, or they may be included in language-specific grammatical paradigms. In the following section, I deal with this in more detail. Moreover, I suggest that the construal phenomenon which Langacker (2003: 23) calls “extreme subjectification”, i.e. eliminating “any direct concern with the occurrence of actual events”, may be described according to the threefold schema proposed here. The shift from actuality to virtuality (=“conceptualizer’s mental activity”) in the temporal domain, as well as from reality to the knowledge about reality (=“reality as a mental construct”) in the modal domain may be also understood as implying different ways of the construal configuration of the ground. Since the three components of every grounding predication – the grounded entity, the grounding relationship, and the ground – are distinguished only analytically here, insofar as they form a sign unit, a significant change in the meaning of the sign suggests a shift involving all components of the sign. For the English present tense, there occurs a shift from the actual speech event, which is shared by speaker and hearer, to the speaker’s conceptualization of the speech event whose coincidence with the grounded event is asserted. Consequently, the grounding relationship represents not the domain of actual reality, but rather the domain of virtual reality. As for the grounded entity, it shifts from actual events to virtual ones. And as for modal grounding, the vantage point is extrapolated from actual reality, which is assumed to be common for speaker and hearer at the time of speaking, to the (conceptualizer’s) knowledge about reality, where no identity between speaker’s and hearer’s knowledge about reality can be claimed. In short, “extreme subjectification” touches all components of the sign: it pertains to the construal (i) of the grounded entity (from the actual process to its representation or conceptualization), (ii) of the grounding relationship (from reality to the knowledge about



Elena Smirnova

reality, from actuality to virtuality), and (iii) of the ground itself (from the speech event to the conceptualization event, where the speaker’s role in the actual communication process is interpreted as the conceptualizer’s role in the actual conceptualization process). 3. Two basic kinds of construal configurations of the ground As I have argued in the previous section, there are three basic criteria for the description and classification of grounding predications: they correspond to the three components of every deictic sign and thus consist of (i) the nature of the profiled entity, (ii) the nature of the grounding relationship, and (iii) the nature of the ground’s conceptualization itself. In the following, I focus on the third criterion in more detail. Notice that I am not concerned with the idea of the ground in terms of the context of every actual speech situation, which is essentially dynamic and created through linguistic and non-linguistic actions. In that view, the ground is basically sociocentric, not egocentric; it is something that is jointly created by the participants and that is constantly shifting in interaction, as it is not only maintained but also created and modified by the participants. This point has been made by, inter alia, Goffman (1981), Hanks (1990), and Laury (2002). Without denying the fact that the ground, i.e. the context of speech, is permanently shifting in processing communication, I nonetheless agree with Langacker (2002a: 20) that “the speaker and hearer need to establish a coherent view of a complex situation...”, which is organized around a certain deictic center. Thus, I am not concerned here with the problem of how the ground is established in every communication process. Instead, assuming that there is a stable point of reference that is implicitly referred to by using grounding predications, I attempt to provide a more differentiated view on the structuring of the deictic center. In so doing, I take up the idea of construal and apply it to the perspectival structuring of the speech situation itself. The concept of the (current) speech event plays the essential role in the linguistic phenomenon of deixis (and grounding). I argue that there are two basic alternative ways of deictic construal being inherent in every deictic and grounding process. At this point, I explicitly refer to the definition of the ground, claiming that it is the speech event together with its participants, i.e., both speaker and hearer. Therefore, the speaker, who without any doubt plays the pivotal role in every speech situation, is notwithstanding (merely) a part of the ground. Relating linguistic expressions to the speaker as the central entity of the speech event, grounding predications can and do distinguish between different conceptualizations of the speaker’s role in the current communication process.

The organization of the German clausal grounding system 

I suggest that a closer look at the conceptualization of the ground itself, as it makes difference between two basic ways of construing the speaker’s role in it, allows for better understanding of the phenomenon of grounding. Two alternative ways of the fundamental cognitive structuring of the ground may be described in terms of deictic construal: On the one hand, the construal of the ground is based on the immanent coincidence of the mutual communicative roles of speaker and hearer. Indeed, the speech event necessarily consists of two participants (otherwise it would not be a speech event). This is the default structural configuration of the speech situation: speaker and hearer are identified as a communicative whole, which is accompanied by immediate circumstances of the speech situation. The configuration that implies coincidence of the discourse roles, or “coordinating cognition” (Verhagen 2007), in a speech situation serves as the default viewing point. Speaker and hearer share among other things a common point of reference, which is the coincidence of their communicative roles. This common point of viewing may be internalized by the speaker when using deictic expressions. Therefore, using deictic elements, among them grounding predications, the speaker identifies this viewing point as the point of reference. In the following, I call this kind of the ground construal common deictic center (other terms are “basic construal configuration” (Verhagen 2007), “first-person-inclusive” (Janssen 2002)), since the hearer’s role is part of the common vantage point in this viewing arrangement. On the other hand, the deictic construal of the ground may be dominated by only the speaker’s role. As far as the speaker concerns himself as individual conceptualizer, the hearer is separated from his internal cognitive world. Thus, the hearer belongs to the external world, albeit part of the current speech event. The ground is then coincident with the speaker’s role and excludes the communicative role of the hearer. In this case, only the speaker serves as the point of viewing and thus defines the deictic center. Using deictic expressions including grounding predications, the speaker may relate what is being said to this point. I use the term speaker-exclusive deictic center to name this kind of the ground construal, since the role of the hearer is excluded from the vantage point of this viewing arrangement (other terms are “first-person-deixis” (Verhagen 2007) and “first-person-exclusive” (Janssen 2002)). I prefer this term because it refers to the communication situation in general, whereas other terms pertain rather to the category of person, which is a deictic category of its own and encompasses only the distribution of the discourse roles of the participants in communication. Note that the so called “second-person-deixis” (Verhagen 2007), where the situation is presented from the hearer’s point of view, i.e. where the reference point of the addressee is explicitly adopted by the speaker and thus serves as deictic center, cannot be considered the third kind of the ground construal, in my view.



Elena Smirnova

Instead, in such cases the speaker-exclusive construal configuration is shifted from the speaker to the hearer. The positions of the speaker and hearer are reversed: adopting the hearer’s point of view, the speaker gives up his own central position and is thus excluded, at least as linguistically coded, from the deictic center. To summarize, while the first type of the construal configuration of the ground – the common deictic center – favors the coincidence of the mutual communication roles of speaker and hearer, the second one – the speaker-exclusive deictic center – is based on the distinction between those roles and foregrounds the role of the speaker only. As I show below, German tenses and the preterit subjunctive mood employ the first construal configuration of the deictic center in their prototypical uses. On the other hand, grammatical expressions of such epistemic notions as knowledge and experience, which normally differ from one person to another, are predominantly based on the second construal of the viewing arrangement. For German, e.g., the epistemically, or deictically (cf. Diewald 1999), used modal verbs fall into this category of grounding predications. The following general rule is valid, irrespective of which type of the construal configuration of the ground is involved. A particular deictic space (or epistemic domain), in which the grounded entities (processes or things) are situated, is “divided” into regions in respect to the ground. There are at least two discrete regions of the deictic space: one including the ground and the other excluding it. The first one is defined positively in respect to the ground; every entity located in this (epistemic) region is situated within the same deictic space where the deictic center is located (in other words: this region is proximal to the ground). The second deictic region is defined negatively in respect to the ground; every entity located in this region is situated outside the region that includes the ground (in other words: this region is distal). The distribution of proximal (including the ground, coincident with the ground, overlapping with the ground, etc.) and distal (i.e. non-proximal, noncoincident, non-inclusive, non-overlapping, etc.) deictic values, however, may be based on the two alternative deictic construals of the ground respectively. Consequently, the distal deictic values around the shared vantage point, i.e. the common deictic center, lie outside the shared viewing point of speaker and hearer and refer typically to third party(-ies). On the other hand, the organization of the deictic space around the speaker-exclusive deictic center implies that the position of the hearer is included in the distal deictic region with respect to that vantage point. Therefore, combining the general rule of the value distribution (positive vs. negative with respect to the ground) with the distinct construal configurations of the ground, we get different formations of the deictic space for each configuration. Covering to a great extent identical realms of a deictic space, these formations

The organization of the German clausal grounding system 

differ in regard to whether the hearer is situated in the same region in which the speaker finds himself or not, i.e. whether the hearer’s position belongs to the ‘inside’ or to the ‘outside’ region of the deictic space. As I show below, due to the overlapping of the deictic spheres organized around different deictic centers, there are correspondences between particular values of grounding predications. Because of this, they are interchangeable to some degree, as is the case for the German verbal mood markers. To briefly illustrate the binary distinction between two alternative ways of construal configuration of the ground made above, I would like to allude to the idea of the twofold perspectival structure of the cognitive framework of the speech situation presented in Janssen (2002). Janssen (2002: 158 ff.) summarizes the twofold conceptualization of the discourse roles in the speech situation in the following manner. The figure shows that “the framework of the speech situation is divided into an inner dimension forming the 1/2-perspective, and an outer dimension forming the 1/3-perspective” (Janssen 2002: 158). The first-person discourse role may be filled by the speaker (possibly with one or more associates differing from addressee), which corresponds to the exclusive ‘we’. The first-person discourse role may be filled by a party consisting of both the speaker and the hearer, which corresponds to the inclusive ‘we’: “Whoever the speaker’s possible associate(s) may be, the discourse role of the speaker is primordial or central in both systems” (Janssen 2002: 159). Applying the proposed framework of the speech situation to the English demonstratives, Janssen convincingly shows that this and that allow for two interpretations which are closely related to the twofold perspectival system presented in Figure 1 (examples (1) to (3) are taken from Janssen 2002: 170). II – 1/3-perspective II – 1/2perspective

1 protagonist party

2 antagonist party

[party initiating the current conversational turn]

[party participating in the current conversation]

1 protagonist party (incl. antagonist party) [party constituting the current conversation] 3 third party [party not constituing the current conversation]

Figure 1.â•‡ Double interactive perspective of the roles in the speech situation (Janssen 2002: 160)



Elena Smirnova

(1) (2) (3)

I can see everything from this window. And you? Do you have a good view from that window? Do you also find it so awfully hot in this room? Did you also find it so awfully hot in that room?

In (1), by using this window, the speaker refers to the window he is looking through. On the contrary, using that window, the speaker refers to the window the addressee is looking through. In this interpretation, the use of this and that is determined by the 1/2-perspective. Thus, the opposition this-that in (1) differentiates between speaker and hearer, defining the speaker only as a reference point. In (2) and (3), on the other hand, the use of this and that is based on the 1/3-perspective. Here, the addressee is conceptualized as part of the (current) speech situation, his role being equated with the role of the speaker. Their shared current position serves as the common point of reference, to which the entity room is related. The twofold cognitive frame of reference of the speech situation proposed by Janssen (2002) is in line with my distinction between two alternative ways of the ground construal. The viewing arrangement, which Janssen centers on the ‘1/2perspective’ or ‘the first-person-exclusive’, corresponds with the deictic construal of the speech situation, whereby the individual role of the speaker is set up as the point of reference. The other possibility, in Janssen’s words, the ‘1/3-perspective’ or ‘the first-person-inclusive’ vantage point, can be consider equivalent to the deictic construal of the speech situation, whereby the coincidence of the mutual speaker and hearer roles is set up as the reference point. To sum up, I propose that there are two basic alternative ways in construing the ground, therefore resulting in two different settings of the deictic centre. I suggest that these two possibilities of the deictic construal configuration of the vantage point often cause ambiguous interpretations of grounding predications. In processing communication, the addressee mostly receives sufficient hints from the context that enable him to establish the deictic centre of the current utterance correctly. In the next section, I argue that certain grounding predication types are generally associated with the certain construal configuration of the ground, whereas others allow for both interpretations in different contexts. 4. German grounding predications German grammatical devices have not been extensively investigated from the Cognitive Grammar point of view. Sporadically, modal grounding predications have been the focus of attention. In the chapter on subjectification, Langacker (1991b: ch. 12.6) claims that the German modals, in contrast to their English

The organization of the German clausal grounding system 

counterparts, are main verbs rather than grounding predications. As a reaction to this, a lot of research has been done on this topic by Tanja Mortelmans (1999, 2001, 2002, and 2004). She has convincingly shown that the German modals in their epistemic use have acquired a high degree of subjectification and thus function as grounding predications. However, the investigation of the overall German grounding system including tense, mood and other grammatical categories is still to be done. This chapter presents the first attempt to describe the system of the German clausal grounding. It focuses particularly on the temporal (4.1.) and modal (4.2.) grounding, while also considering the relations between verbal mood (4.2.1.) and epistemic modality (4.2.2.). 4.1

Temporal grounding

For English, temporal grounding is said to be prototypically manifested as “present” versus “past tense” (Langacker 1991a, 2002a,b, 2003 etc.). The contrast between present and past tense marking is identified with the deictic opposition proximal vs. distal, i.e. by virtue of temporal (non-)coincidence or (non-)overlapping with the moment of speech. Additionally, this contrast may be understood in terms of epistemic (non-)immediacy of the event in respect to the ground (Brisard 1999, 2002). What tenses do (in the absence of a modal) is “locate the profiled process within the realm of reality, where the proximal and distal morphemes are susceptible to a temporal construal” (Langacker 1991a: 250). In my view, a similar distribution of temporal grounding values can be assumed for German. The basic opposition within the temporal grammatical domain is represented by the overtly morphologically marked opposition between present (Präsens) and past tense (Präteritum).2 Using the unmarked, neutral present tense form, the speaker locates the profiled process in the same temporal region of actuality in which he finds herself. Thus, the present tense indicates the (temporal) overlapping of the actual event with the ground. On the contrary, the 2. I do not include the German periphrastic tense forms in my discussion for several reasons. First, the forms Perfekt and Plusquamperfekt are formed with the auxiliaries haben and sein, manifesting aspectual rather than temporal values. The increasing tendency of the Perfekt to be used synonymously with the Präteritum, especially in the South German regions, however, does not influences, in my view, the genuine conceptual temporal opposition I propose here. Instead, the Perfekt is gradually overtaking the place of the Präteritum in the temporal grounding system of German by losing its aspectual characteristics and favoring merely its temporal value. Second, the German Futur II form (werden + infinitive perfect) is no longer considered temporal but modal. Third, it is still a matter of intense debate whether the verbal periphrasis werden + infinitive (Futur I) serves as future tense form in German.



Elena Smirnova

overtly marked past tense (Präteritum) locates the profiled actual event outside the temporal region of the speaker. However, there is a crucial difference between temporal grounding in English and in German. The difference, as I suggest, lies in a ‘more objective’ status of the temporal grounding relationship in German (which also means that the stage of “extreme subjectivity” of temporal grounding predications has not (yet) been reached). Moreover, this difference supposedly originates from the overall structuring of the clausal grounding system of German. While English does not exhibit verbal mood distinctions (anymore), German still possesses the highly grammaticalized mood markers indicative and (present and preterit) subjunctive. As a consequence, German temporal grounding is not necessarily restricted to the epistemic dimension of reality (which by definition “extends no farther than the present” (Langacker 2003: 13)). The German prototypical temporal values do not essentially adhere to the domain of reality, i.e. they do not imply that the grounded entity necessarily belongs to reality. Instead, temporal grounding predications in German only situate the profiled event in the actuality that extends along the time line. They instantiate the actual (=not conceptual) occurrence of a process stating that the process pertains to the external developments ‘out there in the world’. Assuming that (dynamic) processes occur running along the horizontal line, which evolves from past to future, and that the conceptualizer’s direction of gaze following the occurrences of such processes coincides with the direction of time evolving, the binary distinction between ‘behind now’ and ‘not behind now’ mirrors the German temporal opposition between past and non-past. Following the general rule of the distribution of deictic values, the non-past region of the temporal deictic dimension is defined positively with respect to the ground, i.e. the events located in this region are situated in the same temporal sphere in which the speaker finds himself. The processes grounded by the present tense carry the neutral temporal value that can be termed as coincident, overlapping with, proximal to, or including the ground. In contrast, the ‘past’ region of the temporal deictic space is defined negatively with respect to the ground, i.e. it is the ‘outside’ region, valued as non-coincident with, non-overlapping with, non-immediate with, distal to, or exclusive of the ground. The temporal deictic dimension is thus identified with these two regions, and the temporal grounding relationship is identified with the location of the profiled event within one of the two described regions.3 3. Further specifications in the temporal domain are made by aspectual marked forms Perfect and Plusquamperfect. The perfect tense (Perfekt), as opposite to the simple past (Präteritum), specifies a past event as bounded. The Plusquamperfect indicates that a bounded past event is located relative to the ground which is shifted further to the past along the time line (often called Vorvergangenheit).

The organization of the German clausal grounding system 

The view of German temporal grounding proposed here allows e.g. for the explanation of an important difference between the German and English present tense: the German present tense, unlike English, is usually employed for referring to future events. In English, the use of the present tense for future events is “infelicitous unless there is some indication of when the event is expected to occur. It is also problematic for events not amenable to scheduling” (Langacker 2003: 22). On the contrary, the reference to past events is not problematic (historical present). For German, however, both options are felicitous, cf.: (4) (5)

Meine Mutter fliegt (morgen/in zwei Wochen) nach Australien. ‘??My mother flies (tomorrow/in two weeks) to Australia’ Vor zwei Tagen, meine Mutter kommt zu mir. ‘Two days ago, my mother comes to my place’

Such uses of the German present tense are in line with the description given above. The non-prototypical, i.e. future and past, uses of the (unmarked) present tense may now be explained by suggesting a shift in temporal vantage point along the temporal line. In German, unlike English, the vantage point allows movements in both directions. As the shift of the temporal vantage point (deictic center) does not imply a shift with respect to reality, there are no restrictions in degree or direction of temporal shifts. Moreover, the reality status of the profiled events remains untouched, thus these events are usually considered actual events occurring either in the past or in the future. Importantly, the ground, or the way that it is construed in the case of temporal grounding, allows only for one construal configuration. In order to suffice in communication of temporally grounded clauses, the speaker and the hearer always need to share a common deictic center. This common temporal point of reference is usually provided by the position of the speaker. In other words, the hearer always has to adopt the vantage point of the speaker to understand the temporal location of the profiled process properly. I suggest that such a restricted configuration of the deictic center in the case of temporal grounding originates in the relatively ‘objective’ character of our conceptualization of time. We are accustomed to the (shared) view of time as a line which is evolving from past to future. This line is built of points that are strung together. Furthermore, processes usually take time to proceed, and they proceed by arriving at and departing from chronologically situated points on the time line. Moreover, if one experiences temporally evolving processes, he usually presupposes that other persons who are present at the same time and place would experience the occurrence and duration of these processes in the same manner. The exact location of the occurrence of the events in time is traditionally anchored in our knowledge about the world, supported by means of numerous mechanisms for calculating time (e.g. calendar, watch).

 Elena Smirnova

Thus, this “objectively” conceptualized picture of time influences the linguistic coding of temporality. Even if the speaker is able to imagine his own time, mentally scanning through the constitutive phases of a process, a common reference point is needed to communicate the temporality of events to the addressee. To summarize, the German temporal grounding system is prototypically represented by the unmarked present tense vs. the marked past tense, which manifests the opposition between the grounding values ‘non-past’ vs. ‘past’. The temporal grounding relationship in German concerns only the actual occurrence of the evolving processes or their phases along the time line. Thus, the temporal grounding relation may be interpreted in terms of reality of the described events, but does not inherently pertain to the epistemic domain of reality. In the next section, I deal in more detail with modal grounding, which is exclusively oriented upon the epistemic domain of reality. Furthermore, the German temporal grounding system uses only one construal configuration of the ground, i.e., it always establishes the common deictic center. The adoption of the shared temporal vantage point by both the speaker and hearer in the particular speech event is to a larger extent anchored in our ‘objective’ and conventionalized conceptualization of temporality: it is not possible to establish such vantage points as “in respect to my time” and “in respect to your time”, which are independent from each other and do not coincide in a particular speech situation. As I show below, modal grounding predications allow for both construal configurations of the ground. This, in my view, is the crucial difference between temporal and modal clausal grounding, at least in languages like German and English. 4.2

Modal grounding

4.2.1 Verbal mood markers The German modal grounding system is represented by the overtly marked opposition between the indicative and the (present and preterit) subjunctive mood (see also Mortelmans 2001). The status of the imperative mood is controversial. Following Donhauser (1987), who convincingly demonstrated that the imperative is a semi-finite category which lacks relevant semantic as well as formal features to be incorporated into the system of German verbal mood, I exclude the imperative from the present discussion. The highly grammaticalized German verbal mood markers, i.e. the indicative (Indikativ) and subjunctive (Konjunktiv), are grounding predications. They modify whole propositions, i.e. they locate the profiled events with respect to the ground. The modal grounding relationship comprises the reality status of the profiled event, thus the modal grounding operates in the epistemic domain of reality.

The organization of the German clausal grounding system 

The unmarked neutral indicative mood situates the profiled event in the realm of reality, whereas the subjunctive locates the profiled process outside of reality, cf.: (6) (7)

Sie lobt deine Arbeit. ‘She praises your work’ Sie würde deine Arbeit loben (, wenn sie sie sähe). ‘She would praise your work (, if she saw it)’

The German preterit subjunctive is generally associated with irrealis, potentialis, non-factual, and counterfactual functions. In this respect, it is uncontroversial that the profiled event does not belong to reality. It is however problematic to integrate Konjunktiv I (the present subjunctive) into the reality-based conception of modal grounding in German. The default meaning of the present subjunctive is said to be “less concerned with epistemic than with personal matters: it does not indicate a shift as far as the epistemic region (reality > irreality) is concerned, but signals a change with respect to the personal perspective (speaker > third person) on the SoA [state of affairs]” (Mortelmans 2001: 7). The function of the present subjunctive consists of signaling indirect speech, cf.: (8) (Er sagt,) er habe sich getäuscht. ‘(He says,) he has-KONJI erred’ In this connection, Mortelmans (2001) claims: [T]he indicative generally marks a positive attitude of the speaker, whereas the past subjunctive – apart from its use as a marker of indirect speech – generally marks some kind of negative attitude of the speaker. The present subjunctive, then, could be regarded as marking a neutral speaker attitude: the speaker presents a SoA from the viewpoint of a third person, she neither commits herself to it nor distances herself explicitly from it. (Mortelmans 2001: 9)

In her view, then, German verbal mood is not necessarily concerned with the location of the profiled event within the realm of (non-)reality, but is rather characterized as signaling the speaker’s attitude. In my view, however, the German mood markers are concerned with the reality status of the proposition. They set up the link between the profiled event and the ground, whereby the event is positioned with respect to a modal deictic center. Thus, the grounding relationship shared by the German mood markers describes the reality status of the designated event. Albeit being unprofiled, the grounding relationship specifies the epistemic dimension in which the profiled event is situated – the domain of reality. The vantage point of this dimension is centered on the speaker – he evaluates the reality-status of the profiled event. But, as I suggest, it is

 Elena Smirnova

not primarily the speaker’s positive, negative, and neutral attitude that is indicated by the German mood markers (cf. Mortelmans 2001), though such an interpretation is possible in most cases. Rather, it is the status of the profiled event that is marked by the mood markers indicative and subjunctive. The basic modal opposition consists in the contrast between whether the described event is located in the region in which the ground is situated, i.e. within reality, or outside this region, i.e. outside of reality. While the indicative modal value encompasses the realm of reality, the subjunctive extends through the realm of non-reality. In this connection, the unmarked, neutral indicative mood is opposite to the marked subjunctive mood in terms of (non-)reality of the event presented in the clause. The distinction between the values of present and preterit subjunctive lies in the different construal configurations of the ground, which trigger prototypical interpretations of these markers. The preterit subjunctive relates the profiled event to the common deictic center, identified with (the default conception of) reality shared by speaker and hearer in the speech situation. Using the preterit subjunctive, the speaker claims that the profiled event is not real, i.e. irreal, potential, counterfactual etc., without specifying the exact location of the event within the realm of nonreality. In other words, the speaker situates the profiled event outside the reality of the actual speech situation (shared by the speaker and hearer). That the preterit subjunctive is often used to express wishes corresponds with its prototypical function, as the event is considered not real with respect to the speech situation. The present subjunctive, on the other hand, relates the profiled event to the speaker-exclusive deictic center, identified with speaker’s (conception of) reality. Thus, the addressee with his (conception of) reality is seen as part of the region outside the speaker’s reality. Therefore, using the present subjunctive prototypically as a marker of indirect speech, the speaker locates the profiled event outside the region of reality in which he finds himself. However, the region outside speaker’s reality may include the hearer’s reality. Thus, claiming that the designated event does not belong to reality as internalized by the speaker, the present subjunctive allows for interpretations where the neutral attitude of the speaker towards the (universal) reality status of the event can be assumed. Foregrounding the non-real status of the proposition with respect to the ground construed as “my reality”, the speaker does not claim the general non-real status of the proposition. He only states that the profiled event does not belong to his own (conception of) reality. It has been suggested that the present subjunctive shifts the deictic center from the speaker to another (cited or reported) speaker (see e.g. Diewald 1999) without locating the profiled event within the realm of non-reality. In this view, the present subjunctive locates the reference point, not the profiled entity. Here I argue for the analysis of the present subjunctive presented above, as it provides a more integrated view on subjunctive vs. indicative mood markers.

The organization of the German clausal grounding system 

Moreover, the main function of modal grounding, which is to locate the profiled event with respect to the (modally defined) ground, remains the same for all mood markers. Furthermore, the following interrelation between the preterit and present subjunctive values could be explained on the basis of the proposed analysis. The preterit subjunctive also appears in indirect speech, being able to replace the present subjunctive without considerable change in utterance meaning. It has been suggested that the speaker marks greater distance with regard to the truth of the utterance using the preterit subjunctive instead of the present subjunctive. However, as is correctly pointed out by Fabricius-Hansen (1997), Diewald (1999) and Mortelmans (2001), this cannot be assumed for every use of the preterit subjunctive in indirect speech. Rather, the preterit subjunctive should be seen as a second (and in many cases unambiguous) means to indicate indirect speech. In the view of the analysis proposed here, the interrelation between the values of the preterit and present subjunctives could be explained in terms of great overlapping of the regions in which the profiled events are located by these mood markers. Despite the construal configuration of the ground, which both subjunctive mood markers are based on, the epistemic domain in which they locate the profiled events is the same. Moreover, this epistemic domain of (non-)reality is divided into the same distinct regions by both present and preterit subjunctives – the region defined positively with respect to the ground, i.e. including, overlapping with, coincident with, proximal to the ground, and the region defined negatively with respect to the ground. Consequently, for both the present and preterit subjunctive markers, the epistemic regions that exclude the ground, albeit depending on which construal configuration of the ground is chosen, are equivalent to a great extent. For the past subjunctive, however, this region is defined in less specific terms than for present subjunctive (“reality” vs. “my [speaker’s] reality”). The less specifically a value is defined, the more extensive the region which accommodates this value. Hence, the epistemic region represented by the present subjunctive is included in the epistemic region referred to by the preterit subjunctive. This overlapping, then, explains why the preterit subjunctive form can replace the present subjunctive in most cases (note that the reverse situation does not hold for German). 4.2.2 Epistemic modals In the following, I do not address the question of the grounding status of the German epistemic modals in detail (for this see Mortelmans 1999, 2001, and 2004). Instead, starting from the hypothesis that the highly grammaticalized German epistemic modals do function as grounding predications, I attempt to determine the place of the German epistemic modals in the modal grounding system of German. Mortelmans (2001) proposes that German mood and epistemic modals can be integrated in one consistent model that provides an application of Langacker’s

 Elena Smirnova

dynamic evolutionary model (1991a: 277). The German modal können ‘can’ is said to locate the profiled process within potential reality, whereas müssen ‘must’ locates the profiled process within projected reality. Dürfte (which is the preterit subjunctive form of dürfen ‘may’) “expresses a degree of likelihood somewhere between mere possibility (können) and near certainty (müssen)” (Mortelmans 2001: 13). The modals sollen and wollen (like the present subjunctive mood) are said to typically mark a shift of the origo, since they are used to introduce reported speech. (9) Er kann/muss/dürfte/will/soll deinen Geburtstag vergessen haben. ‘He may/must/has probably/claims to/is said to have forgotten your birthday’ I agree with Mortelmans that the epistemically used German modal verbs, similar to the German subjunctive mood markers, affect the location of the profiled event outside reality, i.e. outside the epistemic domain in which the speaker finds himself. Situating the process outside reality, they also specify the region of non-reality in which the process is situated (potential reality, projected reality). I suggest, however, that there is a crucial difference between the German verbal mood markers and the epistemically used modal verbs. The difference lies in different nature of the grounding relationship. Whereas the verbal mood markers refer to reality, the epistemic modals imply a cognitive transfer into the knowledge of reality, the difference pointed out for the English epistemic modals by Langacker (1991b): This usage reflects a shift in the domain where the notion of “momentum” is manifested: this conception is transferred from the evolution of REALITY ITSELF to the evolution of our KNOWLEDGE OF (PRESENT) REALITY. [...] reality is largely external, while knowledge of reality lies wholly within the province of the conceptualizer, the notion of evolutionary momentum is construed more subjectively when applied to the latter. (Langacker 1991b: 336f.; original emphasis)

I suggest that the epistemic domain ‘reality itself ’ applies to the German verbal mood markers, while the domain ‘knowledge of reality’ is occupied by the epistemic modals. In contrast to English, where, due to the absence of the verbal mood markers, these two options are simultaneously used by the modals, there is a clear distinction between the verbal mood markers and the epistemic modals in German. There is another important aspect of the German epistemic modals that distinguishes them from the mood markers. As I mentioned above, the semantic mood distinctions are built upon the basic deictic binary opposition ‘real’ (the epistemic region positively defined in respect to the ground) vs. ‘non-real’ (negatively defined region). For German epistemic modals, however, the distribution of their grounding values does not follow such binary distinctions. Resting on

The organization of the German clausal grounding system 

the basic modal distinction ‘real’ vs. ‘non-real’ (in respect to the knowledge of reality), they characterize only the ‘non-real’ region. They represent the respective region as a scale where less or more (non-)real spheres of knowledge are distinguished, in which the profiled events are situated. Thus, the scale of the epistemic values of müssen (müsste) – dürfte – können (könnte) represent a scale from the near certainty (=less non-real) to mere possibility (=more non-real). Sollen and wollen, on the other hand, express a neutral non-real status of the profiled event in respect to the speaker’s (present) knowledge of reality.4 As for the alternative construal configuration of the ground, I propose that the epistemically used modal verbs allow only for one configuration, namely that they are oriented around the speaker-exclusive deictic center. Due to the highly subjective status of the modal grounding represented by epistemic modals, they refer to the internal mental world (=knowledge) of the conceptualizer. In this connection, it is quite unlikely that the speaker and hearer share the same knowledge about reality, even if they share the same speech situation. Therefore, epistemic modals are traditionally understood in terms of a speaker’s evaluation (of the truth) of a proposition, of a speaker’s personal attitude towards a proposition, or of a speaker’s degree of commitment to the truth of a proposition. Using epistemic modals, the speaker unambiguously signals his own view of the profiled event. This interpretation enables the addressee’s knowledge to be separated from the viewing arrangement and thus to be represented explicitly as knowledge which the speaker does not (need to) share. 5. Conclusion In this chapter, I have examined the verbal categories of German in terms of grounding predications. I have taken the broader notion of deixis as a starting point for an examination of German clausal grounding. I first proposed a unified explanation and classification of grounding predications being a particular class of deictic expressions. I suggested that three main aspects are relevant in distinguishing grounding predications: (i) the nature of the grounded entity, defining the (nominal vs. clausal) status of grounding; (ii) the nature of the grounding relation, defining the epistemic domain including different grounding values; and (iii) the construal configuration of the ground, defining the viewing arrangement (on the epistemic domain).

4. I am not further concerned with the values of wollen and sollen here, since their epistemic modal status is somewhat problematic and currently a matter of considerable debate.

 Elena Smirnova

Then, employing the prominent idea of construal, I showed that there are two different ways in which the ground may be construed. The ground is either represented as the common deictic center, which is normally shared by the speaker and hearer in a current speech situation, or as the speaker-exclusive deictic center that favors only the speaker’s role in communication. Grounding predications differ from each other in the way in which construal configuration of the ground is prominent. For German, the common deictic center is prominent in the case of temporal grounding, anchoring the grounded processes to the time of speaking, which has to be shared by speaker and hearer. As for German modal grounding, different construal configurations underlie the distinction between the past and the present subjunctive, both being opposite to the unmarked verbal mood indicative. The common deictic center, defined as (the conception of present) reality of the speech event, serves as the vantage point of the preterit subjunctive, whereas the speaker-exclusive deictic center, identified with speaker’s (conception of present) reality, functions as the vantage point of the present subjunctive. German epistemic modals, however, relate the grounded processes in respect to the ground construed as the speaker-exclusive deictic center, anchoring the profiled event to the speaker’s internal knowledge of reality. References Brisard, Frank. 1999. A critique of localism in and about tense theory. PhD Dissertation, University of Antwerp, Belgium. Brisard, Frank (ed.). 2002. Grounding: The epistemic footing of deixis and reference. Berlin/New York: Mouton de Gruyter. Bühler, Karl. [1934] 1999. Sprachtheorie. Jena: Fischer. Diewald, Gabriele. 1999. Die Modalverben im Deutschen: Grammatikalisierung und Polyfunktionalität. Tübingen: Niemeyer. Donhauser, Karin. 1987. Mood and Morphology: An alternative Approach to the Syntax and Semantics of German Moods. Lingua 73: 53–77. Fabricius-Hansen, Cathrine. 1997. Der Konjunktiv als Problem des Deutschen als Fremdsprache. In Aspekte der Modalität im Deutschen – auch in kontrastiver Sicht, Debus Friedhelm & Oddleif Leirbukt (eds.), 13–37. Hildesheim/Zürich/New York: Georg Olms. Fillmore, Charles. 1997. Lectures on Deixis. Stanford: CSLI. Goffman, Erving. 1981. Forms of Talk. Philadelphia: University of Pennsylvania Press. Hanks, William F. 1990. Referential Space: Language and Lived Space among the Maya. Chicago: University of Chicago Press. Janssen, Theo A.J.M. 2002. Deictic principles of pronominals, demonstratives and tenses. In F. Brisard (ed.), 151–193. Langacker, Ronald W. 1991a. Foundations of Cognitive Grammar 2. Descriptive application. Stanford: Stanford University Press.

The organization of the German clausal grounding system  Langacker, Ronald W. 1991b. Concept, image, and symbol. The cognitive basis of grammar. Berlin/ New York: Mouton de Gruyter. Langacker, Ronald W. 2002a. Deixis and subjectivity. In F. Brisard (ed.), 1–28. Langacker, Ronald W. 2002b. Remarks on the English grounding systems. In Brisard (ed.), 29–38. Langacker, Ronald W. 2003. Extreme subjectification: English tense and modals. In Motivation in Language: Studies in honor of Günter Radden, Hubert Cuyckens, Thomas Berg, René Dirven, & Klaus-Uwe Panther (eds.), 3–26. Amsterdam/Philadelphia: John Benjamins. Laury, Ritva. 2002. Interaction, grounding and third-person referential forms. In Brisard (ed.), 83–111. Mortelmans, Tanja. 1999. Die Modalverben sollen und müssen im heutigen Deutsch unter besonderer Berücksichtigung ihres Status als subjektivierter ‘grounding predications’. PhD dissertation, University of Antwerp, Belgium. Mortelmans, Tanja. 2001. An introduction to Langacker’s grounding predications: Mood and modal verbs in German. In Modalität und mehr/Modality and more, Heinz Vater & Ole Letnes (eds.), 3–26. Trier: Wissenschaftlicher Verlag Trier. Mortelmans, Tanja. 2002. ‘Wieso sollte ich dich küssen, du hässlicher Mensch!’: A study of the German modals sollen and müssen as ‘grounding predications’ in interrogatives. In Brisard (ed.), 391–432. Mortelmans, Tanja. 2004. The status of the German auxiliary werden as a ‘grounding redication’. In Modalität und Übersetzung, Heinz Vater & Ole Letnes (eds.), 33–56. Trier: Wissenschaftlicher Verlag Trier. Verhagen, Arie. 2007. Construal and perspectivisation. In Handbook of Cognitive Linguistics, Dirk Geeraerts and Hubert Cuyckens (eds.), 48–81. Oxford: Oxford University Press.

Grounding in terms of anchoring relations Epistemic associations of ‘present continuous’ marking in Turkish Ceyhan Temürcü

Middle East Technical University This chapter aims to contribute to the understanding of the connection between ‘present time reference’ and ‘epistemic contingency’ (including the evidential category ‘new information’) by investigating the epistemic associations of the Turkish ‘continuous aspect’ marker -Iyor (when used as the only TAM marker on a verbal predicate), with analyses based on the framework of anchoring relations developed in Temürcü (2007). Similar to Brisard’s (2002b, 2005) analysis of the English present progressive construction, -Iyor is shown to be associated with ‘epistemic contingency’. The treatment is then compared to the epistemic account of tenses and aspects in Cognitive Grammar. It is argued that temporal and epistemic categories should be taken as distinct but interacting components of sentence meaning and that the connection between present time reference and epistemic contingency is best explained in terms of metonymy. Keywords: epistemic contingency, epistemic immediacy, mirativity, continuous aspect, present time reference, grounding, anchoring, metonymy

1. Introduction Cross linguistically, certain types of tense and aspect markers are known to be connected with certain epistemic values.1 For instance, markers of ‘past tense’ are often associated with ‘epistemic distance’ (e.g., ‘counterfactuality’ or ‘unreality’) (Fleischman 1989, Dahl 1997, Iatridou 2000, Palmer 2001: 203–221), markers of ‘perfect aspect’ with ‘inference’ (Comrie 1976: 108–110, Anderson 1982, Bybee, Perkins & Pagliuca 1994: 95–97, Johanson 2000a: 121–123), and markers of ‘future’ 1. The term ‘epistemic’ is used in this chapter as a cover term for both specifications of probability (‘epistemic modality’) and evidential dependency (‘evidentiality’).

 Ceyhan Temürcü

with ‘uncertainty’ (e.g., ‘irrealis’, ‘epistemic modality’, or ‘indefiniteness’) (Lyons 1977: 677, Yavaş 1982b, Comrie 1985: 45, Givón 1994, Mithun 1995: 385, Palmer 2001: 104–105). What is less commonly investigated is the relation between the present time reference (as expressed by markers of ‘present imperfective’, ‘present progressive’ or ‘present continuous’) with the evidential category most often called ‘new information’ (also called ‘immediate information’, ‘immediate evidence’ and sometimes subsumed under ‘mirative’).2 Nichols (1986: 254–255) asserts a universal covariance between ‘perfective’ and ‘inference’ on the one hand, and between ‘imperfective’ and ‘immediate’ on the other, on the basis of data from Chinese Pidgin Russian, Turkish (Aksu-Koç & Slobin 1986), Tibetan (DeLancey 1986) and Sherpa (Woodbury 1986). DeLancey (2001) supports Nichols’s generalization by illustrating the association between ‘mirative’ and ‘imperfective’ and that between ‘inferential’ and ‘perfective’ with data from Hare, Sunwar, Korean (citing DeLancey 1997) and Tadzhik (citing Lazard 1996). Brisard (2002b, 2005) presents a comparable analysis for the English ‘present progressive’ construction. Using the account of epistemic grounding in Cognitive Grammar, he identifies the schematic value of this construction as ‘epistemic contingency’. This characterization includes ‘new information’, as it covers the expression of unexpected and surprising events. For instance, the ‘imperfectivization’ of the verb in the example below indicates that the subject’s behavior “is not what one would expect, given what is known about the subject, but that it constitutes something of a surprise” (Brisard 2002b: 270):

(1) You are being silly.

(Brisard 2002b: 270)

This chapter investigates epistemic associations of the Turkish marker -Iyor as used on its own on a verb (henceforth designated as -Iyor#, # showing that no further TAM marker is attached), which then typically expresses ‘continuous aspect’ in present tense.3 Section 1 outlines the theoretical background, which is based on 2. ‘Present time reference’ is distinguished here from ‘present tense’, which is a grammatical category. Present time reference involves designating an event or state as ongoing or obtaining at the time of utterance. Markers labeled as ‘present tense’ in many languages, including English, do not consistently refer to the present time. 3. Turkish markers are referred to in the text using abstract morphemic representations, which are underspecified as to phonetic alternations conditioned by morphoâ•‚phonological rules, such as vowel harmony and consonant assimilation. The letters used for abstract phonemes and their phonetic realizations (in the Turkish orthography) are listed below. A: I: X: D: G:

a, e ı, i, u, ü a, e, ı, i, u, or ü d, t ğ, k

Grounding in terms of anchoring relations 

the framework of anchoring relations developed in Temürcü (2007). Section 2 presents an analysis of temporal and epistemic values of -Iyor# and concludes that it is strongly associated with ‘epistemic contingency’, which is taken as covering both ‘new information’ and ‘personal certainty’. Section 3 compares the results of this analysis with Brisard’s (2002b, 2005) epistemic characterizations of English present tense and present progressive construction. It then expands on how the present account differs from the epistemic model of grounding in Cognitive Grammar in approaching the connection between temporality and epistemicity. 2. Temporal and epistemic categories within the framework of anchoring relations The framework of anchoring relations (Temürcü 2007) is a cross-linguistically applicable, semantic-pragmatic theory of tense, aspect, and mood (TAM). It motivates and defines purely semantic categories that are meant to comprise the building blocks of aspecto-temporal and modal values. This section depicts the semantic space of anchoring relations and expands on the temporal and epistemic categories that will be needed in the analysis of -Iyor in Section 3. 2.1

Anchoring relations in three semantic domains

The framework of anchoring relations imports insights and tools from various frameworks within the cognitive linguistics tradition. The very idea of ‘anchoring’ is inspired by the account of ‘grounding’ in Cognitive Grammar (Langacker 1991: 315–342, 1997, 2002a, b, 2004; Brisard 2002a, b). Anchoring relations are conceived as purely semantic components of grounding predications in three semantic domains.4 These are called the temporal, epistemic and volitional domains and coincide, to a certain extent, with what Sweetser (1990) called the ‘content’, ‘epistemic’, and In morpheme representations, parentheses indicate buffer phonemes, and a hyphen (â•‚) marks a bound morpheme. The hash sign (#) at the end of a marker is used to indicate that no further TAM marker is attached to it in the given context. 4. ‘Grounding’ in Cognitive Grammar accounts for the deictic nature of tenses and grammatical modality markers in finite clauses, and of demonstratives, articles, and certain quantifiers in fully specified nominal expressions. Grounding predications “indicate the relationship of a designated entity to the ground or situation of speech, including the speech event itself, its participants, and their respective spheres of knowledge” (Brisard 2002a: xi). While grounding predications are grammatical elements, anchoring relations are defined in purely semantic terms. As such, they can be expressed not only by grammatical strategies, but also by lexical items, independent particles, sentence types, intonational distinctions, a combination of these, or with zeroâ•‚marking. As a consequence, while grounding predications always express offstage (nonâ•‚focal, backgrounded) and general (schematic) meanings, anchoring

 Ceyhan Temürcü

‘speech act’ domains. The temporal domain includes meanings typically expressed by viewpoint (or, relational) aspects and tenses, the epistemic domain those expressed by markers of epistemic modality and evidentiality, and the volitional domain those expressed by volitional (deontic) and illocutionary moods. All the three semantic domains of anchoring are conceived from the perspective of a center (i.e., viewed from a certain vantage point). The temporal center stands for the ‘reference time’, the epistemic center for the ‘reference knowledge state’, and the volitional center for the ‘reference state of will’ at the given state of the discourse. By default, these three centers are tied, respectively, to the speaker’s conception of ‘now’, to her judgment, and to her will as manifested at the communicative situation. These centers can also be displaced to be associated with other times, knowledge states and states of will. Such displacements, which are analogous to the building of ‘child spaces’ in the mental space theory (Fauconnier 1985, 1998; Fauconnier & Turner 1994, 1998; Sweetser & Fauconnier 1996), create new temporal, epistemic or volitional frames around shifted centers. Anchoring relations are defined in terms of qualifications of higher order entities relative to the centers in each domain. These entities are temporal locations (in the temporal domain), thoughts (in the epistemic domain) and projections (in the volitional domain). A higher order entity can be located in one of the four schematically distinguished regions within a domain. A temporal location, a thought or a projection can be actual (metaphorically, ‘preceding’ the center), immediate (metaphorically, ‘at’ the center), potential (metaphorically, ‘following’ the center), or generic (having no specific relation to the center): actuality current state (center)

immediacy

genericity

potentiality current frame

Figure 1.â•‡ A schematic delimitation of the possible qualifications of entities in each domain of anchoring

An utterance can express a higher order entity (or a plurality of higher-order entities) that resides in any of these four regions. As a fifth option, an utterance can relations can also appear onstage (e.g., when expressed by mental state predicates) and can be very specific (e.g., when specified by adverbial elements).

posterior

prospective

atemporal

hypothetical

conjectured

Current epistemic frame

epistemic center

new information

probable

general fact

envisioned

wanted

Current volitional frame

volitional center

immediate contribution

acceptable

asserted

accepted

volitional domain

Figure 2.â•‡ The basic anchoring categories defined in relation to the centers (reference states) in the three domains of anchoring

Current temporal frame

temporal center

simultaneous

recurrent

inferred

certain

anterior

perfect

epistemic domain

temporal domain

general statement

Grounding in terms of anchoring relations 

 Ceyhan Temürcü

quantify over a range of higher-order entities around the center. Figure 2, in which small circles represent higher-order entities, shows the labels for the basic anchoring categories delimited on the basis of this perspectival configuration. In addition to five category types (categories of actuality, immediacy, potentiality, genericity and quantification), two compound categories (categories of relevance) are depicted in each domain. The containment properties of the three types of higher-order entities yields a hierarchically organized structure for sentence meaning, somewhat similar to the ‘underlying clause structure’ (UCS) in Functional Grammar (Dik 1989, Hengeveld 1987, 1990).5 The main clause of every utterance is analyzed as designating a ‘stateof-affairs’ (SoA) (in CG terms, as ‘profiling a process’) and signaling one anchoring category from each domain of anchoring: (2) Volitional category â•… (Epistemic category â•…â•… (Temporal category(SoA))) Brief characterizations for the basic temporal and epistemic categories are given in Table 1. These categories are identified on purely semantic grounds, regardless of whether or not they receive explicit (grammatical or lexical) marking. (Volitional anchoring categories are not included here because they are not relevant for the specific aims of this chapter.) In addition to these basic anchoring categories, two categories of relevance are defined in each domain of anchoring. These are compound categories that link a category of immediacy with either a category of actuality or one of potentiality. As such, they express causal, logical, or intentional relevance to the current reference states. The temporal and the epistemic categories of relevance are characterized in Table 2. While the basic anchoring categories are schematically delimited, utterances often express more specific, fine-grained TAM meanings. When necessary, such distinctions can be represented as specifications to the basic anchoring categories, using descriptive labels between slash signs (/). For instance, the English sentence in (3) expresses the temporal category anterior with further specification of the time of occurrence (anterior/at five/). Similarly, both (4) and (5) match the schematic definition of probable, but they specify different degrees of likelihood, which can be distinguished as probable/high/and probable/low/, respectively. 5. Unlike the semantic structure presented here, the UCS in the standard versions of Functional Grammar incorporates formalâ•‚structural features (see Nuyts 2001: 256, 305–306). The current account shows most affinity with Anstey’s (2002) purely semantic version of Functional Grammar. See Temürcü (2007: 18–19, 30) for motivations for the containment hierarchy ‘projection > thought > temporal location > SoA’.

Grounding in terms of anchoring relations 

Table 1.â•‡ Characterizations of the basic anchoring categories in the temporal and epistemic domains6

Immediacy

Actuality

Potentiality

Quantification

Generality

Temporal domain

Epistemic domain

sımultaneous presents a SoA (an event or a state) as ongoing at the temporal center. In other words, it expresses an immediate condition. anterıor presents a SoA as having occurred before the temporal center. posterıor presents a SoA as potentially occurring after the temporal center.

new information presents a proposition as new and not-wellassimilated to the current knowledge state. In other words, it expresses a piece of immediate evidence. certaın presents a proposition as well-assimilated to the current, personal knowledge state. hypothetıcal presents a proposition as merely unknown relative to the current knowledge state, without specifying any specific degree of likelihood. probable expresses an epistemic evaluation for the truth of a proposition. It can be coupled with a degree of likelihood that may range from ‘very unlikely’ to ‘virtually certain’.*

recurrent expresses repeated occurrences of a SoA in a temporal range around the temporal center. It can be coupled with a frequency of occurrence that may range from ‘rarely’ to ‘always’. atemporal presents a SoA as free from temporal restrictions.

general fact presents a proposition as one for which the truth is taken to be generally valid, rather than as being incidental to a specific knowledge state.

(3) John came at five. (4) John must be sleeping. (5) John might be sleeping.

* The idea of characterizing probability (in general, epistemic modality) in terms of ‘quantification over a range of thoughts’ is basically similar to the idea of ‘quantification over possible worlds’ in modal logic and formal semantics, where the modal value of a proposition is expressed in terms of the worlds in which it is true and those in which it is false. But the current approach differs from ‘possible words’ accounts in that it characterizes linguistic meaning with reference to speakers’ construal of reality, rather than objectively defined models of reality.

 Ceyhan Temürcü

Table 2.â•‡ Characterizations of the temporal and epistemic categories of relevance6 Temporal categories of relevance

Epistemic categories of relevance

perfect establishes a causal relation between an anterıor and a simultaneous temporal location. In its grammatical expression, it explicitly conveys a past SoA and implies immediate conditions, conceived as ‘results’ of this past occurrence.

ınferred establishes a logical relation between a thought qualified as certaın and one qualified as new information. In its grammatical expression, it explicitly conveys a definitive conclusion and implies immediate evidence, conceived as ‘proofs’ for this conclusion. conjectured establishes a logical relation between a thought qualified as hypothetical and one qualified as new information. In its grammatical expression, it explicitly conveys a non-definitive conclusion and implies immediate evidence conceived as the ‘indications’ for this conclusion.

prospective establishes a causal relation between a posterior and a simultaneous temporal location. In its grammatical expression, it explicitly conveys a potential event in the future and implies immediate conditions conceived as the ‘causes’ of this potential occurrence.

2.2

More on epistemic anchoring categories

In this section, epistemic anchoring categories are further specified with reference to (i) the presence or absence of evidentiality, (ii) the presence of absence of full certainty, and (iii) the distinction between personal vs. general knowledge. Each category is illustrated with a grammatical strategy. Since grammatical TAM markers are typically polysemous, a minimal specification of the utterance context is given as necessary, in order to remove any artificial ambiguities.

6. A ‘logical relation’ is distinguished from a ‘causal relation’ in that it involves an epistemic inference, rather than (just) a causeâ•‚effect relation. This difference can be identified in Sweetser’s (1990) distinction between ‘enablement’ in the ‘contentâ•‚domain’ vs. in the ‘epistemicâ•‚domain’. Sweetser’s (1990: 77) following examples of two uses of ‘because’ illustrate this distinction. In (a), the two clauses are connected by “real world causality”, whereas in (b) the connection is only epistemic (logical); the speaker expresses her conclusion on the basis of a premise:

(a) John came back because he loved her. (Sweetser 1990: 77) (b) John loved her, because he came back. (Sweetser 1990: 77)

In the framework of anchoring relations, the distinction between ‘causal’ vs. ‘logical’ relevance is used in contrasting the focal meanings of perfect aspect vs. inferential evidentiality markers. While a perfect implies present effects (causal consequences) of an overtly expressed past event, an inferential marker implies immediate evidence for (logical premises of) an overtly expressed conclusion.

Grounding in terms of anchoring relations 

new information, inferred and conjecture are evidential anchoring categories. While new information explicitly indexes immediate evidence (i.e. a newly apprehended piece of information), both inferred and conjecture conventionally imply the existence and/or type of such evidence.7 New ınformation is covered in DeLancey’s (1997: 35–36, 2001: 369, 379) definition of ‘mirativity’ as “the marking of a proposition as representing information which is new to the speaker” and most closely matches what Nichols (1986) calls the ‘immediate meaning’. Nichols identifies this category in the Chinese Pidgin Russian marker est, which can be used “in a more or less spontaneous reaction to a new, salient, often surprising event just as it happens” (Nichols 1986: 248).8 For instance, the sentence in (6) is uttered by Dersu Uzala, who “standing at a riverbank as a storm approaches, notices high water and remarks on it” (Nichols 1986: 250). But the corresponding sentence without est (7) does not convey any such immediate reaction; “Dersu has been aware for some time that a flood might take place” (Nichols 1986: 251): (6) Voda pribavljaj est. water rise est ‘The water’s rising’

(Arsen’ev 1960: 42; cited in Nichols 1986: 250)

(7) Kamni smotrju: voda pribavljaj. rocks look water rise ‘I’m looking at the rocks: the water’s rising’ (Arsen’ev 1960: 42; cited in Nichols 1986: 250) Inferred overlaps, to a large extent, with the traditional category ‘inference’. It expresses a definitive (certain) conclusion deduced from a piece of immediate evidence. It is illustrated below with Aikhenvald’s (2003: 135) Tariana example. This sentence can be uttered, e.g., in a context where the speaker “saw Cecília with a stick in her hand, and the scarred dog running away”: (8)

Ceci t∫inuâ•‚nuku duâ•‚kwisaâ•‚nihka. Cecília dogâ•‚top.non.a/s 3sgfâ•‚scoldâ•‚spec.infr.rec.p ‘Cecília scolded the dog’ (I infer it on the basis of obvious evidence.) (Aikhenvald 2003: 135)

7. ‘New information’ is often subsumed under ‘mirative’ or ‘admirative’, and considered to be related to, but distinct from, evidentiality (e.g., DeLancey 1997, 2001, Plungian 2001, Aikhenvald 2004: 196–215). But in the framework of anchoring relations new information is taken as the most central evidential category, since it directly expresses a piece of immediate evidence (with or without an indication of surprise). 8. Nichols (1986) also mentions that est, when used with punctual verbs, can also convey the ‘inferential’ meaning (in past contexts), and the ‘predictive’ meaning (in future contexts).

 Ceyhan Temürcü

The third evidential anchoring category, conjectured, also implies an immediate piece of evidence, but differs from inferred in expressing a non-definitive (hypothetical) conclusion. For example, in Mao Naga, while -ahi is used for definitive conclusions, the combination of le with -oTi conveys a weak inference in contexts where the evidence is not ‘infallible’ or ‘reliable’ (Giridhar 1994: 337–338; Bhat 1999: 69–70): (9) pfano idu rüâ•‚oTi le. he yesterday writeâ•‚irrelevant ırrealıs (Glosses from Bhat 1999) ‘He must have written yesterday’(Giridhar 1994: 337; cited in Bhat 1999: 69) The other epistemic categories, namely, probable, hypothetical, certain and general fact, are non-evidential, as they explicitly refer to nor conventionally imply any immediate evidence. Probable corresponds to the traditional category ‘epistemic modality’; it expresses a – vaguely or precisely specified – degree of likelihood. Some grammatical markers can be used for any degree of likelihood, like the Mao Naga suffix -amolo (labeled as ‘Dubitative’ in Giridhar 1994: 308–309) illustrated in (10) below. A language can also recruit different markers to grammatically distinguish different degrees of likelihood, as is the case in English (11): (10) pfo ta-amoloe. he go-may/might ‘He may/might go’

(Giridhar 1994: 309)

(11) John might/may/must have failed the exam. Hypothetical is similar to probable, as it too involves uncertainty and is nonevidential. But unlike probable, it does not positively suggest any degree of likelihood: It merely identifies a proposition as one of which the truth is ‘unknown’. It is illustrated below by the Mao Naga construction v-e v (where v denotes a verb and -e is a concessive mood marker) (Giridhar 1994: 309–310): (12) siboâ•‚no hayi shoâ•‚e shoâ•‚tie (peâ•‚loshüe) Sibo ricebeer drinkâ•‚e drinkâ•‚that can’t say ‘It is possible that Sibo drank ricebeer (can’t say)’

(Giridhar 1994: 310)

Both certain and general fact express categorical certainty, i.e., the speaker’s full epistemic commitment. But while general fact qualifies a proposition as a generally valid, well-established fact, certain presents a piece of information as an item of personal knowledge. Typical manifestations of general fact include expressions of encyclopedic knowledge, scientific information and culturallysanctioned information. Certain is typically involved in the expression of knowledge acquired via the speaker’s direct perception or sensation, or information

Grounding in terms of anchoring relations 

personally received from other sources.9 The difference between certain and general fact can be illustrated with Woodbury’s (1986) examples given below, which reveal the opposition between the affixes -nok and -wi in Sherpa (TibetoBurman). Their epistemic values, identified in Woodbury (1986: 191–192) as + experiential (13) and -experiential (14), corresponds to that between certain and general fact in the current framework: (13) 󰁘aa saaâ•‚p mi ti yemburâ•‚laa deâ•‚kiâ•‚nok. rice eatâ•‚nomnlzr man be Katmanduâ•‚dat stayâ•‚he ‘The man who is eating rice lives in Katmandu’ (I see, have seen...) (Woodbury 1986: 191) (14) 󰁘aa saaâ•‚p mi ti yemburâ•‚laa deâ•‚kiâ•‚wi. rice eatâ•‚nomnlzr man be Katmanduâ•‚dat stayâ•‚gn ‘The man who is eating rice lives in Katmandu’ (It is known...) (Woodbury 1986: 191) 2.3

Combinations of temporal, epistemic and volitional categories in the sentence meaning

As reflected in the hierarchical structure for sentence meaning given in (2), the overall semantic value of an utterance is analyzed as conveying a volitional, an epistemic and a temporal anchoring category at the same time, taking scope over each other in this respective order. Since the semantic space is exhaustively (though schematically) partitioned by seven basic anchoring categories in each domain, any given utterance will match one and only one category in each domain. The examples below illustrate various combinations of temporal, epistemic and volitional categories in some English sentences. (Note that these examples are just expository; they are not meant as complete semantic analyses): (15)

John took the train at five. Temporal category: anterıor Epistemic category: certaın Volitional category: asserted

(16)

John seems to have taken the train. Temporal category: perfect Epistemic category: conjectured Volitional category: asserted

9. certain overlaps what Faller (2002) calls ‘personal information’ and general fact, ‘encyclopedic information’ (Faller 2002: 18, 133). This contrast is also reflected in Goldsmith & Woisetschlaeger’s (1982) distinction between ‘phenomenal’ and ‘structural’ knowledge.

 Ceyhan Temürcü

(17)

John may take the train at five. Temporal category: posterior Epistemic category: probable Volitional category: asserted

(18)

John is busy now. Temporal category: simultaneous Epistemic category: certain Volitional category: asserted

(19)

You are smoking! Temporal category: simultaneous Epistemic category: new information Volitional category: asserted

(20)

Bring the book back to me tomorrow. Temporal category: posterıor Epistemic category: hypothetıcal Volitional category: wanted

3. Epistemic values of the Turkish ‘present continuous’ marker â•‚Iyor# 3.1

The data

Certain Turkish markers with primarily aspecto-temporal values are known to be associated with epistemic dimensions. For instance, the ‘resultative-perfect’ marker â•‚mIş is associated with ‘inference’ (e.g., Lewis 1967: 101, 122; Yavaş 1980: 41–65; Slobin & Aksu 1982; Aksu-Koç 1988; Johanson 2000b, 2003) the ‘habitualâ•‚generic’ marker â•‚Xr with ‘uncertainty’ (e.g., Yavaş 1980: 104–110, 1982a, b; Tura-Sansa 1986, Johanson 1994: 254–255), the ‘perfective’ marker â•‚DI with ‘direct experience’ and ‘definiteness’ (e.g., Aksuâ•‚Koç 1998, Kornfilt 1997: 337), and the ‘past’ marker â•‚(y)DI with ‘doubt’ and ‘counterfactuality’ (e.g., Yavaş 1980: 19–38, Kornfilt 1997: 368). However, the epistemic associations of the ‘continuous aspect’ marker â•‚Iyor have not been sufficiently investigated.10 â•‚Iyor# typically signals that the occurrence of a SoA coincides with the deictic temporal center (time of utterance), thus conveying a ‘present continuous’ 10. Descriptive labels used for â•‚Iyor include ‘progressive aspect’ (Underhill 1976/1987: 111–114; Yavaş 1980: 125–133), ‘focal intraterminal view’ (Johanson 1994: 254–255; Csató & Johanson 1998: 214), and ‘imperfective aspect’ (Aksuâ•‚Koç 1988: 18, 1994: 332, 1998: 268). However, as Kornfilt (1997: 357–358) notes, â•‚Iyor is best identified with Comrie’s (1976: 25–26) ‘continuous aspect’ (rather than ‘imperfective’ or ‘progressive’), since it is not the only imperfective marker (others are â•‚Xr and â•‚mAktA) and since it can be used with both dynamic and stative verbs.

Grounding in terms of anchoring relations 

meaning. This is illustrated below with an activity-type (21) and a state-type SoA (22). Both of these examples straightforwardly match the temporal anchoring category simultaneous. â•‚Iyor# can also be used for relatively extended periods around the temporal center, as shown in (23) and (24) below. Such utterances are also analyzed as expressing simultaneous, since the designated SoAs are meant to be true at the temporal center: (21) Ahmet (şimdi/ şu anâ•‚da) uyuâ•‚yor.11 Ahmet now this momentâ•‚loc sleepâ•‚cont ‘Ahmet is sleeping (now/at the moment)’ (22) (Şimdi/ şu anâ•‚da) dinlenâ•‚mek istâ•‚iyorâ•‚um. now this momentâ•‚loc restâ•‚inf wantâ•‚contâ•‚a1sg ‘I want to rest (now/at the moment)’ (23) Bu hafta çalışâ•‚mâ•‚ıyorâ•‚um. this week workâ•‚negâ•‚contâ•‚a1sg ‘I am not working this week’ (24) Ekolojik denge hızla bozulâ•‚uyor. Ecological balance rapidly deteriorateâ•‚cont ‘The ecological balance is being destroyed rapidly’ â•‚Iyor# can also express what is sometimes called a ‘restricted habitual’, i.e., repeated occurrences around the temporal center (the anchoring category recurrent):12 (25) Hasan arada bir uğrâ•‚uyor. Hasan occasionally drop.byâ•‚cont ‘Hasan drops by occasionally’ 11. The following table shows the gloss entries used in the Turkish examples. Agreement with the 3rd person singular, which is formally unmarked, is not indicated in the glosses. a1sg first person singular agreement loc locative abl ablative nec necessity acc accusative neg negative an action nominalizer p1sg first person singular pronoun cont continuous p2sg second person singular pronoun epaste evidential past enclitic pass passive fn factive nominalizer past past fut future plu plural habg habitualâ•‚generic pos possibility inf infinitive poss possessive jdge judgment enclitic prog progressive 12. The main habitualâ•‚gnomic (atemporal) marker in Turkish is â•‚Xr rather than â•‚Iyor (see examples (31)–(32) and the next note).

 Ceyhan Temürcü

Such sentences with â•‚Iyor#, in addition to their temporal values (simultaneous or recurrent), also signal the speaker’s personal certainty about the truth of the expressed proposition; a dimension captured in the definition of the epistemic anchoring category certain. As seen in the examples below, â•‚Iyor# is also used when a presently ongoing SoA is newly apprehended, hence not fully assimilated to the speaker’s knowledge state, i.e. when the epistemic value of the sentence is new information. For instance, (26) can be uttered by a speaker who is surprised to witness her addressee smoking. In (27) too, the sentence expresses a newly apprehended piece of information, even though no surprise is signaled: (26) Sen sigara içâ•‚iyorâ•‚sun! you cigarette inhaleâ•‚contâ•‚a2sg ‘You are smoking!’ (27) İşte Ahmet de gelâ•‚iyor. here Ahmet too comeâ•‚cont ‘Here, Ahmet is coming too’ This association of â•‚Iyor# with personal certainty (certain) and with epistemic immediacy (new ınformation) is better observed if the above examples are compared with those which express ‘present continuous’ (simultaneous relative to the deictic temporal center) with different epistemic values. Different degrees of epistemic modality (probable) require the appendage of â•‚DIr, olabilir or olmalı to â•‚Iyor (28). The inference (inferred) of an ongoing event or state requires the combination â•‚Iyorâ•‚(y)mIş (29). Finally, â•‚Iyor leaves its place to â•‚mAktAâ•‚DIr in expressions of strong, wellâ•‚established factuality (general fact). For instance, in scientific discourse, the propositional content of (24) above will be expressed as (30): (28) Ali uyuâ•‚yor‑dur /uyuâ•‚yor ol‑abilir /uyuâ•‚yor ol‑malı. Ali sleepâ•‚contâ•‚jdge /sleepâ•‚cont beâ•‚pos /sleepâ•‚cont beâ•‚nec ‘Ahmet may/might/must be sleeping’ (29) Meğer Ali uyuâ•‚yor‑muş. it.turns.out.that Ali sleepâ•‚contâ•‚epaste ‘It turns out that (to my surprise) Ali is sleeping’ (30) Ekolojik denge hızla bozulâ•‚makta‑dır. Ecological balance rapidly deteriorateâ•‚progâ•‚jdge ‘The ecological balance is being destroyed rapidly’ The epistemic association of â•‚Iyor# with certain and new information is stronger than one might think on the basis of these examples. This becomes more evident

Grounding in terms of anchoring relations 

when â•‚Iyor is contrasted with the ‘habitualâ•‚generic’ aspect marker â•‚Xr.13 Gnomic and characterizing utterances are most typically expressed by â•‚Xr# in Turkish, as shown in (31) and (32) below. Since these sentences lack any specific temporal reference, they match the temporal anchoring category atemporal. They also convey a sense of general, wellâ•‚established validity, which matches the epistemic anchoring category general fact. But â•‚Iyor# replaces â•‚Xr# when the speaker presents an atemporal predication as an item of personal knowledge (certain) or as a newly discovered fact that is not yet well-assimilated to her knowledge state (new ınformatıon). For instance, a scientist who has discovered the ability of lizards to see in the dark would use â•‚Iyor#, with or without an indication of surprise (33). Similarly, a speaker who has come to know Mustafa’s love for animals by personal experience or by immediate observation would utter (34) rather than (32). (31) Kertenkeleâ•‚ler karanlıkâ•‚ta görâ•‚ür. lizardâ•‚plu darknessâ•‚loc seeâ•‚habg ‘Lizards see in the dark’ (32) Mustafa hayvanâ•‚larâ•‚ı sevâ•‚er. Mustafa animalâ•‚pluâ•‚acc loveâ•‚habg ‘Mustafa loves animals’ (33) Kertenkeleâ•‚ler karanlıkâ•‚ta görâ•‚üyor(!) lizardâ•‚plu darknessâ•‚loc seeâ•‚cont ‘Lizards see in the dark(!)’ (34) Mustafa hayvanâ•‚larâ•‚ı sevâ•‚iyor(!) Mustafa animalâ•‚pluâ•‚acc loveâ•‚cont ‘Mustafa loves animals(!)’ This epistemic difference between â•‚Xr# and â•‚Iyor# in habitualâ•‚gnomic contexts is further illustrated below. Only â•‚Iyor# is congruent in the existence of adverbial constituents which indicate immediate epistemic apprehension (new information), such as anlaşılan ‘evidently’ or galiba ‘apparently’: (35) Anlaşılan/ Galiba evidently apparently ... Kertenkeleâ•‚ler karanlıkâ•‚ta ??görâ•‚ür/ görâ•‚üyor. lizardâ•‚plu darknessâ•‚loc seeâ•‚habg seeâ•‚cont ‘Evidently/apparently, lizards see in the dark’ 13. â•‚Xr is traditionally identified as the ‘aorist’ in Turkish, but its aspectoâ•‚temporal value is better reflected by labels such as ‘habitualâ•‚generic aspect’ (Yavaş 1980: 96–100, Aksuâ•‚Koç 1995: 276, 1998: 272–273), ‘timeless tense’ (Menges 1968: 128), ‘extended tense’ (Reichenbach 1947: 290–291) and ‘nonâ•‚focal intraterminal viewpoint’ (Johanson 1994: 254–255).

 Ceyhan Temürcü

Similarly, both â•‚Xr# and â•‚Iyor# can convey a perfect of persisting situation (perfect/persisting/) relative to the time of utterance. A minimal pair is given in (36) below. Here too, â•‚Xr# and â•‚Iyor# differ only in their epistemic values. While â•‚Xr# presents the proposition as a wellâ•‚established, objective fact (general fact), â•‚Iyor# reflects either the speaker’s certainty on the basis of her personal knowledge (certain) or a newly discovered piece of information (new information). Thus, â•‚Iyor# is better suited than â•‚Xr# in (37), where bildiğim kadarıyla ‘as far as I know’ indicates personal certainty rather than general factuality. Similarly, in (38) and (39), where demek ki ‘as I understand’ and aa! ‘oh!’ indicate new information (the latter with surprise, e.g., when the speaker has just discovered this upon examining employment records), â•‚Iyor# is used rather than â•‚Xr#: (36) Ali yılâ•‚larâ•‚dır bu şirketâ•‚te çalışâ•‚ır/ çalışâ•‚ıyor. Ali yearâ•‚pluâ•‚since this company workâ•‚habg/ workâ•‚cont ‘Ali has been working in this company for years’ (37)

Bildiğim kadarıyla adam on yılâ•‚dır buraâ•‚da As.far.as.I.know man ten yearâ•‚since hereâ•‚loc ?çalışâ•‚ır/ çalışâ•‚ıyor. workâ•‚habg/ workâ•‚cont ‘As far as I know, the man has been working here for ten years’

(38)

Demek ki bu köyâ•‚de uzun zamanâ•‚dır kimse as.I.understand this villageâ•‚loc long timeâ•‚since no.one ??yaşaâ•‚maz/ yaşaâ•‚mâ•‚ıyor. liveâ•‚neg.habg/ liveâ•‚negâ•‚cont ‘As I understand, no one has been living in this village for years’

(39) Aa! Bu adam on yılâ•‚dır buraâ•‚da ??çalışâ•‚ır!/ çalışâ•‚ıyor! oh this man ten yearâ•‚since hereâ•‚loc workâ•‚habg/ workâ•‚cont ‘Oh! This man has been working here for ten years!’ Alternative markings for future sentences in Turkish show that â•‚Iyor# is actually more closely associated with epistemic immediacy (new information) than with personal certainty (certain). Categorical futures are marked by the ‘definite future’ marker â•‚(y)AcAG (40), predictive futures are marked with â•‚Xr (41), and ‘scheduled futures’ are marked with â•‚Iyor (42). All these examples include simple future reference (posterior) with identical specifications (‘at five’). Differences among them are neatly captured by epistemic distinctions in the semantic space of anchoring: â•‚(y)AcAG# associates the future occurrence with wellâ•‚assimilated, personal certainty (certaın); â•‚Xr# with an estimation of likelihood (probable); and â•‚Iyor#, with immediate epistemic significance (new information). This immediate significance may obtain when a present schedule, plan, or program exists for a future

Grounding in terms of anchoring relations 

event (Yavaş 1982b), as would normally be the case in (42). But such premeditation is not strictly necessary; the sentence in (43), for instance, can be uttered on the basis of weather forecast information: (40) Tren saat üçâ•‚te kalkâ•‚acak. train hour threeâ•‚loc take.offâ•‚fut ‘The train will (certainly) take off at three o’clock’ (41) Tren saat üçâ•‚te kalkâ•‚ar. train hour threeâ•‚loc take.offâ•‚habg ‘The train will (probably) take off at three o’clock’ (42) Tren saat üçâ•‚te kalkâ•‚ıyor. Train hour threeâ•‚loc take.offâ•‚cont ‘The train takes off at three o’clock’ (43) Pazar günâ•‚ü hava açâ•‚ıyor. Sunday dayâ•‚poss weather openâ•‚cont ‘On Sunday clouds will dissipate’ In â•‚Iyor futures, immediate epistemic significance (new information) may stem from the ‘newness’ of the information relative to the knowledge state(s) of the addressee(s) (a ‘displaced’ epistemic center), as could be the case in the two examples above. It can also be ‘new’ to the speaker herself, in case it has just been discovered and not yet been fully assimilated to her current knowledge state. For instance, a speaker who has just checked the train schedule to find out that the departure hour is earlier than expected, would prefer â•‚Iyor# to â•‚(y)AcAG# or â•‚Xr# (44). But information conveyed in â•‚Iyor futures does not have to be literally ‘new’ to a reference knowledge state. It can also be used when the speaker wants to ‘foreground’ the immediate epistemic significance of already presupposed information, as can be the case in (45): (44) Olamaz! Tren saat üçâ•‚te kalkâ•‚ıyor! beâ•‚neg.habg train hour threeâ•‚loc take.offâ•‚cont ‘Impossible! The train leaves at three o’clock!’ (45)

Yarın gidâ•‚iyorâ•‚um ama tomorrow goâ•‚contâ•‚a1sg but senâ•‚in umrâ•‚unâ•‚da bile değil. p2sgâ•‚gen concernâ•‚poss.a2sgâ•‚loc even neg ‘I’m leaving tomorrow but you don’t even care’

 Ceyhan Temürcü

3.2

â•‚Iyor# and epistemic contingency

The analysis presented above reveals that â•‚Iyor#, in addition to its temporal values, is associated with personal certainty (certain) and, more strongly, with immediate epistemic significance (new information).14 These two categories can be subsumed under the label epistemic contingency, since they both express information incidental to a certain knowledge state (that of the speaker at the time of utterance or a displaced epistemic center). Epistemic contingency contrasts with epistemic generality, which is captured in the definition of the anchoring category general fact and shown to be typically expressed by â•‚Xr# in Turkish. This epistemic opposition between the ‘present continuous’ marker â•‚Iyor# and the ‘habitualâ•‚generic’ marker â•‚Xr# is strikingly parallel with Brisard’s (2002b, 2005) characterization of the difference between the English ‘present progressive construction’ and ‘present tense’. Brisard identifies English present tense as expressing ‘epistemic immediacy’, by which he means ‘structural knowledge’ (or, knowledge which is constitutive of the ‘ground’, which is conceived as a mental model of reality). This closely matches what is called ‘epistemic generality’ and associated with â•‚Xr# here.15 And what Brisard attributes to the English present progressive, ‘epistemic contingency’, is argued to be associated with â•‚Iyor# in this study. However, while Brisard derives ‘epistemic contingency’ from the ‘internal temporal perspective’ imposed by the English progressive, this label is used more descriptively here, to cover information incidental to a certain knowledge state (either wellâ•‚assimilated or not), and is applied only to the epistemic values of â•‚Iyor#. These differences actually reflect a more general theoretical difference between the two treatments in accounting for the nature of the relationship between aspectoâ•‚temporality and epistemicity, which is explained below. Brisard attributes general (schematic) epistemic meanings to tenses and aspects, identifies their prototypical meanings, and explains their peripheral uses in terms of their interactions with the discursive context. This ‘prototype’ approach to polysemy, which is akin to a monosemist one, is rooted in the basic tenet of Cognitive Grammar that grounding is ‘epistemic’ in nature, in the sense that it specifies the epistemic status of a profiled entity with regard to the coordinates of the immediate communicative context (Brisard 2002a: xiii–xv; 14. As shown by Tura-Sansa (1986), the semantic counterpart of â•‚Iyor in nonâ•‚verbal predication is zeroâ•‚marking (â•‚Ø). The epistemic connotations of â•‚Iyor can be shown to hold for â•‚Ø too. See Temürcü (2007) for semantic maps that show the overall ranges of polysemy of â•‚Iyor and â•‚Ø on the one hand, and those of â•‚Xr and â•‚DIr on the other. 15. By an unfortunate clash of terminology, what is labeled as ‘epistemic generality’ in the current framework corresponds to Brisard’s ‘epistemic immediacy’, while ‘epistemic immediacy’ is taken here as a subcategory of epistemic contingency rather than epistemic generality.

Grounding in terms of anchoring relations 

Langacker 2002b: 29, 2004). At the core of Brisard’s analysis lies the idea that different kinds of time reference have different epistemic statuses due to their distinctive modes of cognitive apprehension. Brisard (2002b) maintains that the English present tense inherits its value of ‘epistemic immediacy’ from the epistemic character of present time reference, namely, ‘immediate givenness’. In the same manner, Brisard (2005: 72–81) derives the schematic meaning of the English present progressive construction, ‘epistemic contingency’, from the type of construal involved in its prototypical aspectual use, which involves the imposition of an ‘internal perspective’ to the designated process. Although Langacker (this volume) advises caution for not directly attributing types of mental access to tenses, he goes along with Brisard (2002b) in assigning the English present tense the schematic meaning ‘epistemic immediacy’, which he takes to be “an aspect of the mental experience inherent in the apprehension of presentâ•‚time occurrences”. The framework of anchoring relations differs from the epistemic model for grounding predications in Cognitive Grammar in two related respects. First, it does not attempt to assign general (schematic) meanings to grammatical TAM markers. It rather adopts a ‘family resemblance’ approach to polysemy, where each conventional use of a marker is â•‚synchronically and/or diachronicallyâ•‚ related to at least one other use, without necessarily being tied to a core or prototypical meaning. Second, it takes temporal reference and epistemic reference as distinct but interacting components of sentence meaning, rather than deriving aspectoâ•‚temporal values from epistemic features (or viceâ•‚versa). While grounding defines one ‘epistemic’ relation between the ground and a profiled content, the framework of anchoring relations recognizes three vantage points (centers) in three distinct domains, and accordingly motivates three layers of anchoring relations. As such, it allows one to identify various combinations of temporal, epistemic and volitional categories in the compositional meaning of an utterance. In consequence of these theoretical differences, the analysis in this chapter does not identify ‘epistemic contingency’ as a general (or schematic) characterization of the overall range of polysemy of â•‚Iyor. It mainly shows that â•‚Iyor#, in its various uses, is more consistently associated with epistemic contingency (certain and new information) than with present time reference (simultanoeus). Epistemic contingency, on its own, does not explain why â•‚Iyor can also be used to express recurrent (repeated occurrences around the reference time) in addition to simultaneous, or why it cannot express perfects other than persisting perfect. Such explanations should also take into account the ways different TAM markers compete for regions on the semantic space, possibly necessitating diachronic evidence for semantic extensions, reductions and shifts which have lead to their

 Ceyhan Temürcü

current ranges of polysemy. Such analyses may indeed reveal that the ‘internal temporal perspective’ involved in continuous or progressive aspect provides the chief psychological motivation for their association with epistemic contingency (more specifically, with epistemic immediacy, as conceived here). Such an explanation would also comply with the perspectiveâ•‚based delimitation of anchoring categories, shown in Figure 2: both simultaneous and new information are categories of immediacy, expressing contents which hold ‘at’ the current temporal and epistemic states, respectively. Still, such semanticâ•‚structural similarities will not be enough to establish a necessary link between epistemic contingency and present time reference, as the latter can also occur with other epistemic values. For instance, in the examples given in Section 3.1 and replicated below, present time reference (simultaneous) combines with epistemic modality (probable), inference (ınferred), and general validity (general fact), respectively: (28) Ali uyuâ•‚yorâ•‚dur. Ali sleepâ•‚contâ•‚jdge ‘Ali may be sleeping now’

(sımultaneous + probable)

(29) Meğer Ali uyuâ•‚yorâ•‚muş. (sımultaneous + ınferred) it.turns.out.that Ali sleepâ•‚contâ•‚epaste ‘It turns out that (to my surprise) Ali is sleeping’ (30) Ekolojik denge hızla bozulâ•‚maktaâ•‚dır. (sımultaneous + general fact) Ecological balance rapidly deteriorateâ•‚progâ•‚jdge ‘The ecological balance is being destroyed rapidly’ In the same vein, epistemic contingency, as conceived here, can be expressed along with temporal categories other than simultaneous. Turkish examples of such combinations, where - Iyor# is used for a combination of atemporal, perfect or posterior with certain or new information, were already given in (33)–(39) and (42)–(45). Similarly, the Turkish ‘direct past’ marker â•‚DI can express a past (anterior or perfect) occurrence either as an item of direct personal experience (certain), as in (46) and (47), or as an immediate discovery (new information), as in (48): (46) Ayşe gece buraâ•‚da kalâ•‚dı. Ayşe night hereâ•‚loc stayâ•‚past ‘Ayşe stayed here yesterday night’ (47) Ben bu filmâ•‚i görâ•‚düâ•‚m. I this movieâ•‚acc seeâ•‚pastâ•‚a1sg ‘I have seen this movie’

(anterıor + certaın)

(perfect+ certaın)

Grounding in terms of anchoring relations 

(48) Kara görünâ•‚dü! (perfect + new ınformatıon) land appearâ•‚past ‘Land is in sight!’ (Lit: ‘the land has appeared’) These examples also show that the association of epistemic contingency (critically including epistemic immediacy, or mirativity) with present time reference, even if crossâ•‚linguistically significant, is not absolute. What I want to argue is that their relation is not a matter of conceptual necessity but one of pragmatic metonymy: the conditions of use of certaın and of new information frequently overlap with the conditions of use of simultaneous. In other words, epistemic contingency is often, but not always, triggered by the perceptual access to a SoA that is ongoing at the time of utterance. This contextual overlap is reflected in the sentence meaning as the coâ•‚occurrence of epistemic contingency with present time reference. In languages that lack separate grammatical marking for epistemic contingency, this combination will be most likely to be cumulatively expressed by a ‘present imperfective/continuous/progressive’ strategy, as is the case in Turkish and English.16 Furthermore, diachronic processes may lead to the generalization of the epistemic values of such ‘present time’ markers. What I am going to propose for â•‚Iyor is not very different from Langacker’s (1990, 1997, 1998, 2003) account of ‘subjectification’:17 â•‚Iyor, which started its TAM career as a marker of progressive aspect (e.g., Erdal 2004: 252), must have resumed wider semantic applicability through a process of loosening (partial bleaching) of its more objective (temporal), and strengthening of its more subjective (epistemic) associations. Eventually, epistemic contingency became a sufficient condition for the use of â•‚Iyor even when the temporal component is persisting perfect, future or habitualâ•‚gnomic.

16. In some languages, the combination of ‘present time reference’ with ‘new information’ requires the use of an evidential marker in conjunction with ‘present imperfective’ marking. See, e.g., DeLancey’s (1997: 39, 2001: 376) Hare data and Aikhenvald’s (2004: 199) Tsakhur example. Although ‘new information’ has mostly been attributed to more or less dedicated markers, a closer look at the epistemic behavior of the ‘present imperfective/continuous/progressive’ markers in languages other than Turkish and English may reveal further insights about the connection between present time reference with epistemic immediacy. 17. Langacker (1990: 17, 1998: 73) describes subjectification as a kind of semantic change that involves “the realignment of some relationship from the objective axis to the subjective axis”. This process involves the reinforcement of subjective (offstage) meanings and the erosion of less subjective (onstage) ones.

 Ceyhan Temürcü

4. Conclusion I agree that grounding is an ‘epistemic’ (or, ‘epistemological’) operation in the sense that it specifies the relation of a designated content to the coordinates of the ‘ground’, which is represented in the speech participants’ mental models of reality. What this chapter suggests is that, in a more restricted sense of the term ‘epistemic’, epistemic categories should be distinguished from aspectoâ•‚temporal ones. The framework of anchoring relations achieves this by decomposing the meanings expressed by grounding predications into their temporal, epistemic and volitional building blocks. This makes it possible to explain the connection between present time reference and epistemic contingency with reference to their frequent coâ•‚occurrence in sentence meaning. References Aikhenvald, Alexandra Y. 2003. Evidentiality in Tariana. In Studies in Evidentiality. [Typological Studies in Language 54], A. Y. Aikhenvald & R. M. W. Dixon (eds.), 131–164. Amsterdam & Philadelphia: John Benjamins. Aikhenvald, Alexandra Y. 2004. Evidentiality. Oxford: Oxford University Press. Aksuâ•‚Koç, Ayhan. 1988. The Acquisition of Aspect and Modality: The Case of Past Reference in Turkish. Cambridge: Cambridge University Press. Aksuâ•‚Koç, Ayhan. 1994. Development of linguistic forms: Turkish. In Relating Events in Narrative, A Crossâ•‚linguistic Developmental Study, S. Strömqvist & L. Verhoeven (eds.), 393–457. Hillsdale, NJ: Lawrence Erlbaum. Aksuâ•‚Koç, Ayhan. 1995. Some connections between aspect and modality in Turkish. In Temporal Reference, Aspect and Actionality, vol. 2: Typological Perspectives, P. M. Bertinetto, V. Bianchi, Ö. Dahl & M. Squartini (eds.), 271–289. Torino: Rosenberg and Sellier. Aksuâ•‚Koç, Ayhan. 1998. The role of input vs. universal predispositions in the emergence of tenseâ•‚aspect morphology: evidence from Turkish. First Language 18: 255–280 Aksuâ•‚Koç, Ayhan & Dan I. Slobin. 1986. A psychological account of the development and use of evidentials in Turkish. In Evidentiality: The Linguistic Coding of Epistemology, W. L. Chafe & J. Nichols (eds.), 159–167. Norwood, New Jersey: Ablex Publishing. Anderson, Lloyd B. 1982. The ‘perfect’ as a universal and as a languageâ•‚particular category. In Tenseâ•‚Aspect: Between Semantics and Pragmatics, P. J. Hopper (ed.), 227–264. Amsterdam: John Benjamins. Anstey, Matthew P. 2002. Layers and Operators Revisited [Working Papers in Functional Grammar 77]. Amsterdam: University of Amsterdam. Arsen’ev, Vladimir K. 1960. Po Ussurijskomu kraju. Moscow: Gos. izdâ•‚vo geografičeskoj literatury. Bhat, Darbhe Narayana Shankara 1999. The Prominence of Tense, Aspect and Mood. Amsterdam & Philadelphia: John Benjamins.

Grounding in terms of anchoring relations 

Brisard, Frank. 2002a. Introduction: The epistemic basis of deixis and reference. In Grounding: The Epistemic Footing of Deixis and Reference, F. Brisard (ed.), xi–xxxiv. Berlin & New York: Mouton de Gruyter. Brisard, Frank. 2002b. The English present. In Grounding: The Epistemic Footing of Deixis and Reference, F. Brisard (ed.), 251–297. Berlin & New York: Mouton de Gruyter. Brisard, Frank. 2005. Epistemic interactions of tense and aspect in the English verb: the paradigm of the present. In Cognitive Linguistics: A Userâ•‚friendly Approach, K. Turewicz (ed.), 65–82. Szczecin: University of Szczecin Press. Bybee, Joan, Revere Perkins & William Pagliuca. 1994. The Evolution of Grammar. Tense, Aspect and Modality in the Languages of the World. Chicago & London: University of Chicago Press. Comrie, Bernard. 1976. Aspect. Cambridge: Cambridge University Press. Comrie, Bernard. 1985. Tense. Cambridge: Cambridge University Press. Csató, Éva Á. & Lars Johanson, 1998. Turkish. In The Turkic Languages, Lars Johanson & Éva Á. Csató (eds.), 203–235. London & New York: Routledge. Dahl, Östen. 1997. The relation between past time reference and counterfactuality: a new look. In On conditionals again, A. Athanasiadou & R. Dirven (eds.), 97–114. Amsterdam & Philadelphia: John Benjamins. DeLancey, Scott. 1986. Evidentiality and volitionality in Tibetan. In Evidentiality: The Linguistic Coding of Epistemology [Advances in Discourse Processes 20], W. L. Chafe & J. Nichols (eds.), 203–13. Norwood, New Jersey: Ablex Publishing. DeLancey, Scott. 1997. Mirativity: The grammatical marking of unexpected information. Linguistic Typology 1(1): 33–52. DeLancey, Scott. 2001. The mirative and evidentiality. Journal of Pragmatics 33(3): 369–382. Dik, Simon C. 1989. The Theory of Functional Grammar. Part I: The Structure of the Clause. Dordrecht: Foris. Second, revised edition by Kees Hengeveld. 1997. Berlin & New York: Mouton de Gruyter. Erdal, Marcel. 2004. A Grammar of Old Turkic [Handbook of Oriental Studies, Section 8: Uralic and Central Asian Studies]. Leiden & Boston: Brill. Faller, Martina. 2002. Semantics and Pragmatics of Evidentials in Cuzco Quechua. PhD Dissertation, Stanford University United States. Fauconnier, Gilles. 1985. Mental Spaces: Aspects of Meaning Construction in Natural Language. Cambridge, MA: MIT Press. Reprinted. 1994. Cambridge: Cambridge University Press. Fauconnier, Gilles. 1998. Mental spaces, language modalities, and conceptual integration. In The New Psychology of Language: Cognitive and Functional Approaches to Language Structure. M. Tomasello (ed.), 251–279. Mahwah, NJ: Lawrence Erlbaum. Fauconnier, Gilles & Mark Turner. 1994. Conceptual projection and middle spaces. Cognitive Science Technical Report 9401. San Diego: University of California, Department of Cognitive Science Technical Reports. Available at: http://www.cogsci.ucsd.edu/research/files/ technical/9401.pdf Fauconnier, Gilles & Mark Turner. 1998. Principles of conceptual integration. In Discourse and Cognition, J.-P. Koenig (ed.), 269–283. Stanford: CSLI Publications. Fleischman, Suzanne. 1989. Temporal distance: a basic linguistic metaphor. Studies in Language 13(1): 1–50. Giridhar, Puttushetra Puttuswamy. 1994. Mao Naga Grammar [CIIL Silver Jubilee Publication Series]. Mysore: Central Institute of Indian Languages. Givón, Talmy. 1994. Irrealis and the subjunctive. Studies in Language 18: 265–227.

 Ceyhan Temürcü Goldsmith, John & Erich Woisetschlaeger. 1982. The logic of the English progressive. Linguistic Inquiry 13(1): 79–90. Hengeveld, Kees. 1987. Clause structure and modality in Functional Grammar. In Ins and Outs of the Predication [Functional Grammar Series 6], J. van der Auwera & L. Goossens (eds.), 53–66. Dordrecht: Foris. Hengeveld, Kees. 1990. The hierarchical structure of utterances. In Layers and Levels of Representation in Language Theory: A Functional View [Pragmatics and Beyond New Series 13], J. Nuyts, A. M. Bolkestein, & C. Vet (eds.), 1–23. Amsterdam: John Benjamins. Iatridou, Sabine. 2000. The grammatical ingredients of counterfactuality. Linguistic Inquiry 31(2): 231–270. Johanson, Lars. 1994. Türkeitürkische Aspektotempora. In Tense Systems in European Languages, R. Thieroff & J. Ballweg (eds.), 247–266. Tübingen: Max Niemeyer Verlag. Johanson, Lars. 2000a. Viewpoint operators in European languages. In Tense and Aspect in the Languages of Europe, Ö. Dahl (ed.), 27–187. Berlin & New York: Mouton de Gruyter. Johanson, Lars. 2000b. Turkic indirectives. In Evidentials: Turkic, Iranian and Neighbouring Languages, L. Johanson & B. Utas (eds.), 61–87. Berlin & New York: Mouton de Gruyter. Johanson, Lars. 2003. Evidentiality in Turkic. In Studies in Evidentiality [Typological Studies in Language 54], A. Y. Aikhenvald & R. M. W. Dixon (eds.), 273–290. Amsterdam & Philadelphia: John Benjamins. Kornfilt, Jaklin. 1997. Turkish. London: Routledge. Langacker, Ronald W. 1990. Subjectification. Cognitive Linguistics 1: 5–38. Langacker, Ronald W. 1991. Concept, Image and Symbol: The Cognitive Basis of Grammar. Berlin & New York: Mouton de Gruyter. Langacker, Ronald W. 1997. Consciousness, construal, and subjectivity. In Language Structure, Discourse and the Access to Consciousness, M. I. Stamenov (ed.), 49–75. Amsterdam & Philadelphia: John Benjamins. Langacker, Ronald W. 1998. On subjectification and grammaticization. In Discourse and Cognition, J.-P. Koenig (ed.), 71–89. Stanford: CSLI Publications. Langacker, Ronald W. 2002a. Deixis and subjectivity. In Grounding: The Epistemic Footing of Deixis and Reference, F. Brisard (ed.), 1–28. Berlin & New York: Mouton de Gruyter. Langacker, Ronald W. 2002b. Remarks on the English grounding systems. In Grounding: The Epistemic Footing of Deixis and Reference, F. Brisard (ed.), 29–38. Berlin & New York: Mouton de Gruyter. Langacker, Ronald W. 2003. Extreme subjectification: English tense and modals. In Motivation in Language [Current Issues in Linguistic Theory 243], H. Cuyckens, T. Berg, R. Dirven, & Klausâ•‚Uwe Panther (eds.), 3–26. Amsterdam & Philadelphia: John Benjamins. Langacker, Ronald W. 2004. Aspects of the grammar of finite clauses. In Language, Culture and Mind, M. Achard & S. Kemmer (eds.), 535–577. Stanford: CSLI Publications. Lazard, Gilbert. 1996. Le médiatif en persan. In L’Enonciation Médiatisée, Z. Guentchéva (ed.), 21–30. Leuven & Paris: Peeters. Lewis, Geoffrey. L. [1967] 1978. Turkish Grammar. 3rd edition. Oxford: Clarendon Press. Lyons, John. 1977. Semantics. Cambridge: Cambridge University Press. Menges, Karl Heinrich. 1968. The Turkic Languages and Peoples: An Introduction to Turkic Studies. Wiesbaden: Otto Harrasowitz. Mithun, Marianne. 1995. On the relativity of irreality. In Modality in Grammar and Discourse, J. Bybee & S. Fleischman (eds.), 367–388. Amsterdam & Philadelphia: John Benjamins.

Grounding in terms of anchoring relations 

Nichols, Johanna. 1986. The bottom line: Chinese Pidgin Russian. In Evidentiality: The Linguistic Coding of Epistemology, W. L. Chafe & J. Nichols (eds.), 239–257. Norwood, New Jersey: Ablex Publishing. Nuyts, Jan 2001. Epistemic Modality, Language, and Conceptualization: A Cognitiveâ•‚Pragmatic Perspective. Amsterdam: John Benjamins. Palmer, Frank R. 2001. Mood and Modality, 2nd Edition. Cambridge: Cambridge University Press. Plungian, Vladimir. A. 2001. The place of evidentiality within the universal grammatical space. Journal of Pragmatics 33(3): 349–357. Reichenbach, Hans. 1947. Elements of Symbolic Logic. New York: McMillan. Slobin, Dan & Ayhan Aksu. 1982. Tense, aspect, and modality in the use of Turkish evidential. In Tenseâ•‚Aspect: Between Semantics and Pragmatics, P. J. Hopper (ed.), 185–201. Amsterdam: John Benjamins. Sweetser, Eve. 1990. From Etymology to Pragmatics: Metaphorical and Cultural Aspects of Semantic Structure. Cambridge: Cambridge University Press. Sweetser, Eve & Gilles Fauconnier. 1996. Cognitive links and domains: basic aspects of mental space theory. In Spaces, Worlds, and Grammar, G. Fauconnier & E. Sweetser (eds.), 1–28, Chicago & London: University of Chicago Press. Temürcü, Ceyhan. 2007. A Semantic Framework for Analyzing Tense, Aspect and Mood: An Application to the Ranges of Polysemy of â•‚Xr, â•‚DIr, â•‚Iyor and â•‚Ø in Turkish. PhD Dissertation. University of Antwerp, Belgium. Tura-Sansa, Sabahat. 1986. DIR in modern Turkish. In Proceedings of the Turkish Linguistics Conference, A. Aksuâ•‚Koç & E. Erguvanlıâ•‚Taylan (eds.). 145–159. İstanbul: Boğaziçi University Publications. Underhill, Robert. 1976. Turkish Grammar. Cambridge, Massachusetts: MIT Press. (Fifth edition printed in 1987.) Woodbury, Anthony. 1986. Interactions of tense and evidentiality: A study of Sherpa and English. In Evidentiality: The Linguistic Coding of Epistemology, W. L. Chafe & J. Nichols (eds.), 188–202. Norwood, New Jersey: Ablex Publishing. Yavaş, Feryal. 1980. On the Meaning of the Tense and Aspect Markers in Turkish. Unpublished PhD Dissertation. University of Kansas, United States. Yavaş, Feryal. 1982a. The Turkish Aorist. Glossa 16(1): 40–43. Yavaş, Feryal. 1982b. Future reference in Turkish. Linguistics 20: 411–429.

part ii

Descriptive application Cognitive Grammar

Some remarks on the role of the reference point in the construal configuration of “more” and “less” grounding predications Elena Smirnova and Tanja Mortelmans

Leibniz University Hanover and University of Antwerp In Cognitive Grammar, grounding predications are by definition highly grammaticalized elements. As grammaticalization is a gradient phenomenon, this chapter argues for a more gradual view on grounding. It is proposed that a particular linguistic unit may show a weaker or a stronger degree of grounding. The degree of grounding is closely connected with the degree of grammaticalization. Moreover, different degrees in grounding and grammaticalization match different degrees in subjectification. The model suggested here highlights the role of the relevant reference point in the construal configuration of a linguistic unit. The application of the model is illustrated using the example of the German preterit subjunctive, and the constructions werden/würde + infinitive. Keywords: grounding predication, grammaticalization, subjectification, reference point

1. Introduction In Cognitive Grammar, it is uncontroversial that every grounding predication is a highly grammaticalized linguistic element (e.g. Langacker 1991b: 321). Therefore, as far as a linguistic element is considered to serve as a grounding predication, its grammatical (as opposed to lexical) status may be taken for granted. However, it has turned out that the question whether an item is to be considered as a grounding predication, cannot always be answered with a clear yes or no. As has been noted by Mortelmans (2002, 2006), a more gradual view on grounding and grounding predications should be taken, whereby a particular linguistic unit may show a stronger or a weaker degree of grounding. Formally, the degree of grounding is closely connected with varying degrees of grammaticalization. As far as conceptual

 Elena Smirnova and Tanja Mortelmans

content is concerned, it can be assumed that different degrees in grounding match different degrees in subjectification. What this actually means and in what manner different degrees of grounding and subjectification may be discovered, are the questions this chapter attempts to answer. The chapter is mainly concerned with the polyfunctional German construction würde + infinitive whose grammatical status in the German tense-mood-system is still highly controversial. Using this construction as an example, we propose a way in which different degrees of grounding of the same linguistic unit may be discovered. Our approach here is primarily synchronic rather than diachronic, as we do not deal with the historical development of the construction. Rather, we investigate the construction würde + infinitive in the different readings it has in Present Day German. Uncovering the nature of the construal configuration in terms of subjectivity and implicitness, we will argue that two main variants of this construction can be distinguished, which are linked up with two different construals of the reference point. However, our findings also have diachronic relevance, in that we assume that a split has occurred in the grammaticalization process of the construction that has led to the establishment of two functional variants of the same form. Furthermore, as the degree of grounding is assumed to correspond to the degree of formal grammaticalization, it is different for the two variants of the construction würde + infinitive. The chapter is structured as follows. We start by giving a brief introduction to grounding predications, stressing the close connection between grounding and (subjective) construal. Next, we argue for a view on grounding as a gradient phenomenon bringing together the notions of grounding, subjectification and grammaticalization. In Sections 3 and 4, we examine the grounding status of the German preterit subjunctive and of the German construction werden + infinitive, respectively. Section 5 presents the results of our analysis of the controversial construction würde + infinitive, claiming that two different variants of this construction have to be distinguished which differ in their degree of grounding. In the conclusion, we suggest that the notion of (construal of the) reference point provides a helpful tool for testing the grounding status of various linguistic elements. 2. Grounding predications: Conceptual structure The function of a grounding predication consists in the localization of a profiled entity in a particular epistemic domain. This localization always takes place relative to the ground, which includes the speech event, its participants speaker and hearer, and their spheres of knowledge.

“More” and “less” grounding predications 

Every finite clause is grounded, by definition. In English, the requisite grounding is effected by the tense-modal complex [...]. Because an epistemic predication profiles the grounded entity (not the subjectively construed grounding relationship), the tense-modal complex designates a schematic process, which it locates vis-à-vis the ground in an “epistemic” domain. (Langacker 1991a: 245, our emphasis)

A crucial feature in the characterization of a grounding predication is the maximally subjective construal of the ground. Despite its pivotal role as a relevant reference point, the ground remains offstage and unprofiled. Grounding predications, by definition, only profile the grounded entity, whereas the ground itself as well as the grounding relationship are left unprofiled. A grounding element does not specifically mention the ground, despite invoking it as a reference point. [...] a grounding element does not explicitly refer to any facet of the ground (e.g. the speaker, the addressee, or the time of speaking) and cannot be made to do so (Langacker 2002: 13); [...] a grounding element construes the ground itself with maximal subjectivity. (Langacker 2002: 18; original emphasis)

The maximally subjective, i.e. implicit, character of the ground is therefore decisive for the conceptual structure of every “true” grounding predication. As we have seen, every grounding predication implies a strong link to the ground. However, neither the ground itself nor the grounding relationship can be brought onstage, i.e. can be explicitly coded by a grounding predication. The aspect of implicitness (or maximal subjectivity) of the ground and of the grounding relationship will play a crucial role in our analysis. Applying this property to the German (synthetic) preterit subjunctive, to the German construction werden + infinitive and, in a final step, to the construction würde + infinitive, we will show that these linguistic forms exhibit different degrees of grounding. 3. A more gradual view on grounding and its consequences As has already been noted above, we argue for a more gradient view on the phenomenon of grounding. Originally, it was more or less suggested that grounding was a binary phenomenon: linguistic units that do not fulfill all grounding properties (a list of which is provided in Langacker (2002), for instance), have been mostly excluded from the class of grounding predications. [...] one of the problems conjured up [...] is the question to what extent one could entertain the possibility of also including non- or semi-grammatical predications in the discussion of grounding. In other words, the extent to which grounding is a binary or a gradient notion is still in dispute. (Brisard 2002: xxi)

 Elena Smirnova and Tanja Mortelmans

One of the first explicit attempts towards treating grounding as a gradient phenomenon has been made by Mortelmans (2002), based on her observations of the German modal verbs: The question whether a modal verb functions as a grounding predication, is in need of a more qualified answer that the one suggested by Langacker. I have argued that a number of parameters, functional as well as formal, determine the grounding status of a modal. This is particularly the case for the German modals, which as a group have not acquired the same degree of formal grammaticalization as their English counterparts, a reason for Langacker to deny them the status of grounding predication. If one investigates the individual members of the German modal verb category, though, finer distinctions can be made. (Mortelmans 2002: 423–424)

In view of the close connection between grounding status and the degree of formal grammaticalization of a linguistic element, whereby grammaticalization is conceived of as a gradient phenomenon extending through long periods of time and involving successive steps leading from a “less” to a “more” grammaticalized status of a linguistic unit, it becomes clear that grounding should also be seen as a gradient phenomenon. Moreover, conceptual changes involved in the (diachronic) process of developing grounding predications, such as subjectification and attenuation, also provide evidence for treating grounding as a gradient phenomenon. If subjectification and grammaticalization are to a large extent interrelated – as they undoubtedly are – and if grounding predications evolve through the processes of subjectification and grammaticalization, it is obvious that grounding must be gradient as well. As has been mentioned above, we are primarily concerned with functional (rather than formal) parameters of grounding predications. One of the crucial aspects here is the precise nature of their construal configuration. In the context of subjectification and subjectivity, two distinct construal configurations of an entity are distinguished: the objective and the subjective one. An entity is construed objectively to the extent that it is distinct from the conceptualizer and is put onstage as a salient object of conception. Being the focal point within the onstage region, an expression’s profile has a high degree of objectivity. An entity receives a subjective construal to the extent that it functions as the subject of conception but not as the object. The highest degree of subjectivity thus attaches to the speaker and hearer, specifically in regard to those expressions that do not in any way include them within their scope. (Langacker 2002: 17)

These two construal possibilities are exemplified in Figure 1. Subjectification, then, may be described as a process leading from the left to the right. Its starting point is the canonical viewing arrangement, in which an

“More” and “less” grounding predications 

tr

lm

tr

lm

XY

Y’ X’

G

G

G = ground tr = trajector lm = landmark XY = profiled relation Y’ = the rest of the original profile X’ = subjectively construed relation

Figure 1.â•‡ Objective and (maximal) subjective construal (Langacker 1991a: 216)

external relationship (XY) is construed objectively, i.e. without being linked to the ground (which is outside the scope of predication). Subjectification occurs when some relationship is realigned from the objective to the subjective axis. Such realignment results in the configuration shown on the right: instead of X there is an analogous relationship X’ that holds between the objective situation and some aspect of the ground. These two possibilities, as Langacker himself states, describe two extreme poles on the subjectification (or subjectivity) scale: Thus I do not envisage attenuation and eventual full subjectification as occurring in a single step. It is more likely a gradual evolutionary process involving small steps along a number of possible parameters. (Langacker 1999: 155)

This view implies that there are intermediate stages in a subjectification process. Such intermediate construal configurations are mostly described by implementing an additional relevant reference point (R) which can neither be equated with the clausal trajector (tr) nor with the ground itself. This is the case, for instance, in the popular ‘across’-example Vanessa is sitting across the table from Veronica, where Veronica serves as a relevant reference point (Langacker 1991b: 326–327). This is the case also for the evolution of go-futures, where a temporal reference point serves as a starting point for the conceptualizer’s mental scanning, without being

 Elena Smirnova and Tanja Mortelmans

automatically equaled with either the trajector or with the ground: both French aller and English be going to are themselves inflected for tense and therefore allow for reference points other than the ground which depend on the tense marking of the respective go-form. It is important to note here that the notion of the reference point R does not mean the same as the temporal reference point in the sense of Reichenbach. The reference point R may be any entity that is put onstage and objectively construed. In the ‘across’-example it is the person Veronica who serves as R, for the go-futures it is a point in time. Usually, the subjectification process reaches its endpoint when the reference point R is necessarily and implicitly equated with (some aspect of) the ground, so that R=G. At this point, “true” grounding predications arise, which conform to a maximally subjective construal configuration (Langacker 1991b: 332). We would like to put forward a general description of the intermediate stage(s) in subjectification by establishing a distinct feature for linguistic elements that show an intermediate degree of subjectification. This distinct property of linguistic signs can serve as a crucial parameter that allows us to distinguish between different degrees in subjectification and grounding. As a consequence, we can distinguish between weaker and stronger grounding elements. If a linguistic expression relates the profiled entity to (some aspect of) the ground, it can be assumed to gain in degree of subjectivity. However, this expression cannot be considered a “true” grounding predication, as long as it relates the profiled entity to a reference point R which cannot be fully equated with the ground G or whose equation with the ground must be effected in an ‘objective’ (i.e. explicit) manner. Let us exemplify this by means of the familiar case of the go-future: It should be noted that French aller and English gonna are not grounding elements, even with their future value – they are themselves main verbs that require grounding to form a finite clause. Observe that the temporal reference point (R) need not coincide with the time of speaking [...] The reference point [of an actual future-tense grounding element; ES/TM] is specifically identified with the ground [...] R becomes more subjective by being equated with G, and as a consequence, the temporal relationship it anchors winds up offstage and hence unprofiled. (Langacker 2002: 23) [emphasis added; ES/TM]

In other words: the presence of a relevant reference point R, to which the profiled entity is related by a linguistic expression, and which is not automatically identified as the ground, prevents us from treating this linguistic expression as a “true” grounding predication. Nevertheless, it does not preclude the expression from having already reached an intermediate degree of subjectification and hence from serving at least partially as a grounding predication.

“More” and “less” grounding predications 

4. The German preterit subjunctive as a grounding predication The German verbal mood forms indicative, present subjunctive (Konjunktiv Präsens) and preterit subjunctive (Konjunktiv Präteritum) are straightforward grounding predications, since they are highly grammaticalized items that locate the designated event in a particular epistemic region relative to the ground: The meaning of the moods Indikativ and Konjunktiv must indeed be characterized relative to the ground: using the unmarked neutral mood ‘Indikativ’, the speaker locates the state of affairs in reality, i.e. in the same epistemic region in which she finds herself. With a Konjunktiv-II, on the other hand, a SoA [state of affairs] is situated in some part of irreality, i.e. outside the epistemic dominion of the speaker, as the German past subjunctive is generally associated with irrealis and potentialis-like functions. (Mortelmans 2001: 6–7; original emphases)

We are mainly concerned here with the preterit subjunctive. It locates the profiled event outside reality, i.e. outside the speaker’s own epistemic region of reality. As mood operators select the reality status of the ground as its relevant aspect (rather than e.g. its temporal orientation) the modal ground can be described as the “speaker’s reality conception” or, in short, as “speaker’s reality”. For the German preterit subjunctive, it has been claimed (e.g. Kasper 1987, Diewald 1999, and Smirnova 2006) that the localization of the profiled event in the realm of non-reality primarily depends on the presence of so-called “Bedingungen der sinnvollen Behauptbarkeit” [conditions of reasonable assertion; translation ES/TM]. So, the preterit subjunctive does not simply locate a state of affairs in non-reality, but points to (a) particular condition(s), the non-fulfillment of which entails the non-fulfillment of the designated state of affairs. This analysis is confirmed by the (proto)typical use of the German preterit subjunctive in conditionals. (Note that the German preterit subjunctive occurs both in the protasis and the apodosis of a conditional sentence.) Since there are factors – conditions – which are recognized by the speaker as being not fulfilled at the time of speaking, the profiled situation that depends on the fulfillment of this condition cannot belong to the epistemic realm of reality, cf.: (1) (2)

Sie würde deine Arbeit loben, wenn sie sie sähe. ‘She would praise your work, if she saw it’ Wäre ich an seiner Stelle gewesen, hätte ich gehandelt. ‘If I had been in his place, I would have taken action’

This analysis of the German preterit subjunctive might lead to the conclusion that these unfulfilled (or non-given) conditions serve as a relevant (onstage) reference point R in the preterit subjunctive’s construal configuration and are thus crucial

 Elena Smirnova and Tanja Mortelmans

for the particular location of the profiled event. Consequently, since these nonfulfilled conditions cannot be equated with the ground (as they clearly fall outside “speaker’s reality”), it might be suggested that the German preterit subjunctive does not function as a “true” grounding predication after all. However, a more careful view should be taken on this phenomenon. We propose that the non-given conditions do not serve as the relevant reference point in the construal configuration of the German preterit subjunctive. Instead, we argue for treating the non-given conditions as possible reference points that may be added to the predication for further specification of the non-real status of the profiled event. So, they do not make a (particular aspect of the) ground explicit, but rather specify the (non-real status of the) profiled event. As the (non-real status of the) profiled event is onstage, its specification through explicit mentioning of non-given conditions does not objectify any aspect of the ground: the ground remains implicit and is construed with maximal subjectivity. The status of the profiled event is anchored only with respect to the ground, i.e. with respect to the reality conception of the speaker. It is not determined, but only specified by a number of unfulfilled facts. To demonstrate this, let us first present an analogous analysis of the German past tense. The German past tense (Präteritum) is a “true” grounding predication: it locates the profiled event in the past relative to the ground, i.e. relative to the time of speaking. There is no further reference point(s) that may be assumed to be relevant in the construal configuration of the German past tense. However, the explicit mentioning of a (past) temporal reference point is not excluded by the use of the past tense, cf.: (3) (4) (5) (6)

Er war krank. ‘He was ill’ Er war gestern krank. ‘He was ill yesterday’ Er war die ganze Woche vor Weihnachten krank. ‘He was ill the whole week before Christmas’ Als ich ankam, war er krank. ‘When I arrived he was ill’

In (3), the profiled event is ‘simply’ located in the past by means of the grounding predication of past tense. In (4), the past in which the profiled situation is located, is specified by the temporal expression gestern ‘yesterday’, which is closely linked to the time of speaking by its deictic nature. In (5), the temporal specification of the past is carried out by the expression die ganze Woche vor Weihnachten ‘the whole week before Christmas’, which is largely independent of the speech situation and gives the exact location of the event in calendar time. Finally, in (6), the

“More” and “less” grounding predications 

temporal specification is effected by means of a subclause which itself gets a past tense marking. What all four sentences have in common is that they describe past situations from the point of view of the current speech situation. Whether an event occurred at an exactly defined time, as in (5), or somewhere in the past, as in (3), makes no difference for the past tense form used. Moreover, additional temporal specifications, whatever they may be, do not influence the temporal value of the grounding element Präteritum. Hence, for the German past tense grounding predication Präteritum, we cannot treat the explicit temporal expressions in (4), (5) and (6) as relevant reference points that are crucial for the location of the profiled event. In all cases, the relevant reference point remains the time of speaking, i.e. the ground, from the point of view of which the profiled event is (implicitly) located. Moreover, temporal specifications that are not in line with the past time orientation of Präteritum are blocked. One cannot normally say, for example, *Er war morgen/ nächste Woche krank ‘He was ill tomorrow/next week’ (unless for stylistic effects in specific textual configurations). Therefore, temporal expressions, which may always be added to a clause grounded by the past tense form, cannot be considered relevant for the temporal grounding predication. Rather, they are possible specifications of the (temporal grounding status of the) profiled event, which are themselves (implicitly) anchored in the ground. They do not affect the grounding value of the past tense form. Now, let us turn again to the German preterit subjunctive, as the same analysis may be applied to it, cf.: (7) (8) (9)

Sie wäre in ihrem Zimmer, (wenn...) ‘She would be in her room, (if...)’ Wenn sie schon zuhause wäre, wäre sie in ihrem Zimmer. ‘If she were at home already, she would be in her room’ Wenn ihre Mutter sie dahin geschickt hätte, wäre sie in ihrem Zimmer. ‘If her mother had sent her there, she would be in her room’

The list of examples can be continued by adding more possible conditions, the realization of which is considered non-given at the time of speaking. In (7), the profiled event is located outside reality: the state of affairs ‘she is in her room’ is construed as being non-real. Examples (8) and (9) have the same modal value, i.e. the same situation of the subject she being in her room is considered non-real. The difference between (8) and (9) is that in (8) the condition wenn sie schon zuhause wäre ‘if she were at home already’ is not fulfilled, whereas in (9) it is the condition wenn ihre Mutter sie dahin geschickt hätte ‘if her mother had sent her there’. The fact is, however, that these conditions do not influence the location of the profiled event within non-reality. The preterit subjunctive alone, without any

 Elena Smirnova and Tanja Mortelmans

further means, sufficiently effects a particular localization of the profiled event. The profiled situation is located outside reality: the reasons for this particular localization may be further specified through explicit mention of non-given conditions. Similarly to the additional temporal expressions in the case of past tense grounding, non-fulfilled conditions are themselves (implicitly) anchored in the ground, i.e. they are themselves evaluated by the speaker relative to the ground. This might be a reason why in German, the conditional clause always gets the preterit subjunctive form – ‘wenn sie schon zuhause wäre’, ‘wenn ihre Mutter sie dahin geschickt hätte’. Consequently, contexts like *Wenn sie schon zuhause ist (indicative), wäre (preterit subjunctive) sie in ihrem Zimmer are not acceptable. We hope to have shown that, despite their important role, non-given conditions cannot be considered the relevant reference point R for the preterit subjunctive construal configuration. By way of conclusion, we claim that the German preterit subjunctive serves as a “true” grounding predication, construing the ground as the only relevant and implicit reference point from the perspective of which the profiled event is localized. 5. The status of werden + infinitive as a grounding predication Now, let us turn to the German construction werden + infinitive. The functional status of this construction is still a matter of considerable debate in German linguistics, as it can be conceived as a future tense form, on the one hand, or as a modal construction, indicating the speaker’s epistemic assessment of the likelihood of the described situation, on the other. The former account is mainly represented by the so-called temporalists (to name but a few: Matzel and Ulvestad 1982, Fabricius-Hansen 1986, Thieroff 1992 et al.). The proponents of the latter view are the so-called modalists (e.g. Vater 1975, Itayama 1993); an accurate overview of the debate between temporalists and modalists and of the general problems associated with the semantics of the German werden + infinitive construction is presented in Diewald (2005). In the context of grounding and subjectification, Mortelmans (2004) has claimed that werden + infinitive in both functions has acquired an intermediate degree of subjectification. Starting from this assumption, we will analyze the grounding status of the construction werden + infinitive using the parameter of the relevant reference point that we have established above. Let us start with the following example: (10) Anna wird ein Buch lesen. ‘Anna will read a book/will be reading a book’

“More” and “less” grounding predications 

In isolation, the German sentence has at least two readings. On the one hand, the described situation may be interpreted as being located in the future relative to the time of speaking, i.e. werden + infinitive is a future tense form. On the other hand, the described situation may be understood as being epistemically assessed by the speaker to the extent that the speaker assumes that the profiled situation is likely to occur (either in the present or the future). The question is then: why do such different interpretations of one and the same form arise? In our view, this can be explained if the role of the relevant reference point in the construal configuration of this construction is taken into account. Various formal and functional analyses have shown (e.g. Kotin 2003, Diewald & Habermann 2005, Smirnova 2006 et al.) that the construction werden + infinitive is relatively highly grammaticalized in both functions. As for the degree of subjectification, Mortelmans (2004) has demonstrated that werden + infinitive has reached an intermediate degree of subjectification, since every use of this construction implies a strong link to the speaker and thus to the ground. Therefore, we may suggest that the construction is linked to the ground in the sense that part of the conceptual relationship expressed by werden is realigned from the objective to the subjective axis. This means that the process of subjectification has already performed its initial steps. Despite its relatively high degree of formal grammaticalization, the construction cannot be considered to be a “true” grounding predication, though. We propose that, for each of the two readings of the construction, a distinct relevant reference point must be assumed, which to a large extent determines a particular interpretation of the construction. To demonstrate this, let us perform some modifications of the sentence in (10): (11) Anna wird heute Abend/morgen etc. ein Buch lesen. ‘Anna will read a book this evening/tomorrow etc.’ The explicit mention of the future oriented temporal points (heute Abend ‘tonight’, morgen ‘tomorrow’) considerably reduces the interpretational ambiguity of the sentence in (10): an epistemic interpretation of the sentence seems to be less likely now. Instead, a temporal (future) reading of werden + infinitive is more likely. By means of an additional contextually present temporal expression, the profiled event is located in the future relative to the time of speaking, which serves here as the relevant reference point R. It should be mentioned here that werden + infinitive considerably differs from the future tenses in other languages like e.g. English or French in some relevant aspects. Most importantly, werden + infinitive cannot be assigned the future tense function as its default function. It does not show any signs of obligatory use as a future tense form; instead, the present tense is mostly used to refer to future events.

 Elena Smirnova and Tanja Mortelmans

Therefore, the modal value of the construction cannot be derived from a basic temporal value, as this does not qualify as the core function of werden + infinitive. Thus, an analysis which assumes that the modal meaning of the construction arises as the reinterpretation of its core future meaning, whereby it is not the event itself but the confirmation or verification of the event that is located in the future (cf. e.g. Barceló & Bres 2006 for French & Ziegeler 2006 for English), cannot be applied here. Now, consider the following modification of the sentence in (10): (12) Anna wird jetzt/heute Abend/morgen etc. ein Buch lesen, wie ich sie kenne. ‘Anna will be reading a book (now, tonight, tomorrow), as far as I know her’ The textual presence of the expression wie ich sie kenne ‘as far as I know her’ in (12) seems to trigger an epistemic reading of the sentence: it is now most likely interpreted as the speaker’s epistemic assessment of the described situation on the basis of his personal information about the person involved. In this case, the temporal location of the event, either in the present or in the future (jetzt ‘now’ vs. morgen ‘tomorrow’) appears to be more or less irrelevant, as the epistemic evaluation is the main meaning of werden + infinitive. The pieces of information, known and presented by the speaker, seem to be most relevant for the epistemic interpretation of the sentence with werden + infinitive. This speaker-oriented information serves here as the relevant reference point R. The profiled situation is thus located in a particular epistemic region that pertains to the realm of (knowledge about) reality rather than to the realm of temporality. Comparing (11) and (12), we can state that the inherent functional ambiguity of the construction werden + infinitive exemplified in (10) is radically reduced if particular functions are explicitly mentioned in the text (or are somehow contextually present). The temporal interpretation of the construction is supported (or triggered) by the presence of temporal expressions, whereas the modal (epistemic) interpretation is supported by the presence of speaker-oriented information. We can conclude, then, that the (con-)textual presence of temporal or other facts to which the speaker has access is crucial for the interpretation of werden + infinitive. These elements may be considered to serve as relevant reference points in the construal configuration of the construction werden + infinitive. This supports our starting assumption that werden + infinitive has acquired an intermediate degree of subjectification. Werden + infinitive typically requires an additional relevant reference point to be interpreted in an unambiguous way. This reference point is crucial for the interpretation of werden + infinitive: it serves as a device pointing towards a particular aspect of the ground or another piece of information the speaker has mental access to that is crucial for a particular interpretation of werden + infinitive.

“More” and “less” grounding predications 

In sum, werden + infinitive cannot be considered a “true” grounding predication, since it always implies a relevant reference point (which may, but need not be equal with the ground; R≠G) in its construal configuration. Viewing this from another perspective, however, it can be stated that werden + infinitive has already acquired a relative high degree of subjectification, since the presence of a relevant reference point in the construal configuration testifies that some conceptual relationship has been realigned from the objective to the subjective axis. 6. The status of würde + infinitive as a grounding predication The German periphrastic würde + infinitive construction is a polyfunctional construction. The fact that it serves as an analytical alternative to the synthetic preterit subjunctive is uncontroversial, cf.: Mit dem Grundtempus des Konjunktivs II (Konj. Prät.) bezieht der Sprecher sich auf Gegenwärtiges oder Zukünftiges (d.h. Nichtvergangenes). [...] Die würdeForm wird in diesem Funktionsbereich des Konjunktivs II weitgehend gleichbedeutend mit der Grundform gebraucht. ‘With the basic tense of the Konjunktiv II the speaker refers to the present or the future (i.e. non-past). [...] In this functional domain of the Konjunktiv II the würde form is largely used in a synonymous way’ (Duden 2005: 523) (see also Helbig & Buscha 2001, Flämig 1991, Eisenberg 2004, Weinrich 2003 et al.)

Würde + infinitive can replace the preterit subjunctive forms of most verbs in German, particularly in conditional environments. Thus, in (13) würde kommen is functionally equivalent to käme, and in (14) würde sprechen is equivalent to spräche: (13) Ich würde kommen, (wenn ich Zeit hätte.) [=käme] ‘I would come (if I had time)’ (14) “Das war, als hätte man einen Teddybär dabei”, erzählt die Managerin, “oder als würde man mit einem Menschen sprechen.” [=spräche] ‘“It was like having a Teddy bear with you”, the manageress said, “or like talking to a real person”’ However, it has often been noted that the “replacement” of the synthetic preterit subjunctive is not the only function of this construction. Würde + infinitive exhibits variable readings in different contexts which are different from the typical uses of the preterit subjunctive. In Smirnova (2006), it was suggested that the construction würde + infinitive has two main functions in Present Day German: it may either function as an analytical variant of the synthetic preterit subjunctive (see (13) and (14)) or it may

 Elena Smirnova and Tanja Mortelmans

function as the preterit counterpart of the indicative construction werden + infinitive. The second reading is exemplified in (15), whereby würde wählen ‘would vote’ cannot be substituted by the preterit subjunctive form wählte: (15) Martin Walser wusste am Vortag der Wahl noch nicht, wen er wählen würde. [≠ wählte] ‘Martin Walser did not know on the day before elections who he was going to vote for’ In what follows, we will show that the two assumed variants of the construction würde + infinitive can also be reconstructed if the relevant reference point parameter is applied. Let us first give an example in order to show the inherent functional ambiguity of würde + infinitive (the example together with the added modifications is taken from Fritz 2000: 197f.): (16) Wir wussten alle, dass das Frühjahr turbulent werden würde. ‘We all knew that spring-time would become turbulent’ (17) Alle verschleppten Entscheidungen laufen zusammen. Das Frühjahr wird turbulent werden. ‘All postponed decisions come together. Spring-time will be turbulent’ (18) Wenn alle verschleppten Entscheidungen zusammenliefen, würde das Frühjahr turbulent werden. ‘If all postponed decisions came together, spring-time would be turbulent’ In (16), the construction würde + infinitive (‘würde werden’), in connection with the embedding expression wir wussten ‘we knew’ (which is adopted here only to make the past temporal perspective of the described event clear), has an ambiguous interpretation. Morphologically, it may be reconstructed in two ways: it is either the past tense form of the present indicative form wird werden ‘will become’ > würde werden ‘would become’ or the analytical preterit subjunctive form of the verb werden ‘become’ > würde werden ‘would become’. This functional ambiguity can be overruled in two ways. This is shown in (17) and (18), which represent the reformulation of (16) into the present temporal perspective. The explicit mention of some known facts (alle verschleppten Entscheidungen laufen zusammen ‘all postponed decisions run together’) that are considered real (they are formulated in the indicative) leads to the conclusion das Frühjahr wird turbulent werden ‘spring-time will be turbulent’. So, the original form würde werden ‘would become’ in (16) appears as the (present) indicative construction wird werden in (17). Thus, in case there are some known facts which are interpreted as pointing towards a particular conclusion, würde + infinitive can be paraphrased only by the indicative construction werden + infinitive (which we already described

“More” and “less” grounding predications 

in Section 4). In this respect, würde + infinitive form can best be described as the mood-indifferent preterit form of the construction werden + infinitive. As the German language does not have the indicative preterit form of the verb werden – *wurde – as a morphologically available option for the construction with infinitive anymore, würde + infinitive may be seen either as a marked preterit subjunctive form of werden + infinitive or as a neutral preterit form of the same construction. In (18), on the contrary, the same explicitly mentioned facts are presented as conditions that are considered unfulfilled at the time of speaking (wenn alle verschleppten Entscheidungen zusammenliefen ‘if all postponed decisions ran together’), as they are expressed in the preterit subjunctive form (for the status of the unfulfilled conditions see Section 3 of this chapter). Such an explicit representation of “unreal” facts makes clear that the form würde werden in (16) must be considered the analytic preterit subjunctive form of the infinitive verb werden, which has the same form in the past as well as in the present temporal perspective. Hence, the construction würde + infinitive allows for two distinct readings, one of which is the analytical variant of the preterit subjunctive and the other one is the preterit (subjunctive) form of the construction werden + infinitive. The following examples exemplify the ambiguity of würde + infinitive shown above (examples are taken from the DWDS-Corpus; www.dwds.de): (19) Ich denke mal, ich würde mich bestechen lassen, falls jemand genug bieten würde. (Die ZEIT 49/2000) ‘I think, I would accept a bribe, if I was offered enough money’ (20) Bei dem heutigen Angebot würde das Unterhaltungsbedürfnis des Publikums sicher nicht unerfüllt bleiben, wenn sich ARD und ZDF zurückziehen würden. (Die ZEIT 44/2003) ‘With today’s offer the need for entertainment would certainly not remain unrealized, if ARD and ZDF backed out’ The examples above show that the context sometimes triggers ambiguous interpretations of würde + infinitive (that is another argument for treating this construction as polyfunctional, rather than assuming two homonyms, see below). So, the expression ich denke mal ‘I think’ in (19) suggests that würde + infinitive serves as the preterit (subjunctive) form of werden + infinitive (werde bestechen lassen), while the conditional expression falls jemand genug bieten würde ‘if I was offered enough money’ suggests that würde + infinitive is the analytical preterit subjunctive form. In (20), the modal expression sicher ‘certainly’ together with the explicitly specified known facts bei dem heutigen Angebot ‘with today’s offer’ trigger the epistemic reading of würde + infinitive, while the if-clause wenn sich ARD und ZDF zurückziehen würden ‘if ARD and ZDF backed out’ seems to make the subjunctive reading prominent.

 Elena Smirnova and Tanja Mortelmans

However, we cannot assume that there are two different constructions würde + infinitive which are homonymous in Present Day German. Diachronic investigations (e.g. Kotin 2003, Smirnova 2006) have clearly shown that würde + infinitive has evolved through one more or less consistent grammaticalization process, whereby both its present-day readings have arisen out of one lexical source meaning. We are particularly interested in the grounding status of the construction würde + infinitive. Our starting assumption is that the two described variants of this construction exhibit different degrees of grounding. It is quite obvious that, formally, these two variants of würde + infinitive differ in their degrees of grammaticalization. The subjunctive würde + infinitive shows a very high degree of grammaticalization, since würde can be claimed to function as a preterit subjunctive auxiliary in Present Day German. In this function, the auxiliary occurs only in one morphological form (there is no other tense or mood form available for this function), the construction is obligatory in most subjunctive (e.g. irrealis, potentialis, and conditional) contexts, and every German verb can participate in this construction as an infinitival complement. Moreover, the subjunctive construction würde + infinitive is a constitutive member of the German verbal mood paradigm, together with the synthetic preterit subjunctive, the (synthetic) present subjunctive and the periphrastic pluperfect subjunctive. The temporal/epistemic würde + infinitive, on the other hand, shows a lesser degree of grammaticalization as compared to the subjunctive construction. In this reading, würde + infinitive is an inflectional – preterit subjunctive or past tense – form of the indicative construction werden + infinitive, i.e. it is part of the inflectional paradigm of the construction [werden + infinitive] and cannot be attributed a particular grammatical function of its own. Moreover, the temporal/epistemic würde is typically restricted to specific contexts, e.g. to past tense contexts or to past tense contexts with cognitive matrix verbs. There is very little evidence that this construction is obligatory in certain contexts (the only obligatory context for würde + infinitive in this reading would be the so-called free indirect speech). Finally, there is no straightforward higher-level category (as compared to the category of verbal mood in case of subjunctive würde) which würde or werden + infinitive would enter in Present Day German (yet). Therefore, we are looking for differences in subjectivity status, i.e. for differences in the construal configurations of the two main variants of würde + infinitive, which would support our assumption regarding the direct correspondence between the degree of grammaticalization and the degree of grounding. Our attempt is rather trivial. We will show that würde + infinitive in its subjunctive reading is functionally equivalent to the preterit subjunctive. In doing so, we will argue that their construal configurations are equal. In the subjunctive reading, therefore, würde + infinitive serves as a “true” grounding predication. As for

“More” and “less” grounding predications 

the second reading of würde + infinitive, we will show that würde + infinitive shares the relevant properties with the construction werden + infinitive, most importantly concerning its relevant reference points in the construal configuration. We will try to show that the contexts in which this reading arises typically contain pieces of textual advice that have the status of the relevant reference point in the construal configuration. Moreover, we will show that the conceptual status of the textual advice is identical with the conceptual nature of the reference points we have discovered for werden + infinitive in Section 4. As for the analytical alternative to the synthetic preterit subjunctive (as in (18)), we propose to treat würde + infinitive as a “true” grounding predication. We claim that, in the construal configuration of this reading of würde + infinitive, the relevant reference point is by default equated with the ground. Consider (13), (14), and the following examples: Wenn ich eine stilvolle Villa buchen würde, gäbe es keine Reklamationen. (Die ZEIT 44/2003) [= buchte] ‘If I booked a stylish villa, there would not be any complaints’ Anders würde es dagegen aussehen, wenn Schwitters bei der Komposition der Ursonate seinerseits Elemente von Staren- und anderem Vogelgesang verwendet hätte... (Die ZEIT 26/2001) [= aussähe] ‘However, it would look differently if Schwitters, while composing the Ursonata, had used elements of starling and other birdsongs...’

(21) (22)

In these examples würde + infinitive locates the profiled event outside reality, which is the main function of the preterit subjunctive in German. Furthermore, the würde-form can always be substituted by a synthetic subjunctive form of the infinitive verb. Moreover, in most instances, some unfulfilled conditions are implied by the use of this form or explicitly presented in the text (see e.g. (22) and (13)). This observation is supported particularly by example (21) in which würde + infinitive itself is used to formulate the unfulfilled condition. Taking into account our preceding observations on the status of the unfulfilled conditions, we suggest that würde + infinitive does not have any entity as the relevant reference point but the ground itself. Moreover, the ground is implicit and unprofiled, i.e. it is construed with a maximal degree of subjectivity. Hence, we can treat the construction würde + infinitive in its subjunctive reading as a “true” grounding predication, i.e. as a grammatical construction invoking a maximal subjective construal configuration. As for the temporal-epistemic reading of würde + infinitive (as in (17)), we claim that it has not acquired the same degree of subjectification and grounding as its subjunctive counterpart. In this respect, the temporal-epistemic reading of

 Elena Smirnova and Tanja Mortelmans

würde + infinitive may be assigned to the class of so-called “less” grounding elements showing an intermediate degree of grounding and subjectification. Consider the following examples (example (15) is repeated here as (25)): (23) Mein Vater war so erschüttert, dass ich Angst hatte, er würde es für geschmacklos halten, jetzt wieder von Geld anzufangen.(Böll, 202) [≠ hielte] ‘My father was so shocked that I was afraid (that) he would take it as being tasteless to speak about money now’ (24) Ich glaube, so etwas würde den Zuschauern auf den Keks gehen. (Spiegel 1994/42) [≠ ginge] ‘I believe, such a thing would bore the audience’ (25) Martin Walser wusste am Vortag der Wahl noch nicht, wen er wählen würde. (Spiegel 1994/42) [≠ wählte] ‘Martin Walser did not know on the day before elections who he was going to vote for’ (26) Wir waren überzeugt, dass es zu einem Zusammenstoß kommen würde. (Die ZEIT 49/2002) [≠ käme] ‘We were sure that there was going to be a clash’ In these examples würde + infinitive cannot be paraphrased by the synthetic preterit subjunctive form of the infinitive verb, as the substitution would change the meaning of the whole sentence. The only exception is the sentence in (24) where a substitution of würde gehen by ginge is possible if so etwas is interpreted in terms of a non-given condition, e.g. ‘if such a thing happened’. This interpretation, however, heavily depends on the expression so etwas presented in the text and represents only one possible interpretation of the sentence. The alternative interpretation, where würde gehen cannot be substituted by ginge, seems more appropriate in this context. An important characteristic of the contexts cited above is that würde + infinitive is mostly embedded in a cognitive matrix predicate (glauben ‘believe’, Angst haben ‘be afraid’, wissen ‘know’, überzeugt sein ‘be sure’ etc.). In such contexts, würde + infinitive serves as an inflectional form (morphologically the preterit subjunctive form) of the construction werden + infinitive. The affinity between würde + infinitive and werden + infinitive becomes evident when these contexts are put into the present temporal perspective (such modifications can be easily performed also for (23), (25), and (26)): (27) (28)

Ich glaubte, er würde ohnehin im Krieg fallen. ‘I believed he would fall in the war anyway’ Ich glaube, er wird ohnehin im Krieg fallen. ‘I believe he will fall in the war anyway’

“More” and “less” grounding predications 

On the basis of this replacement possibility Thieroff (1992) proposes that würde + infinitive is the indicative future in the past (“FuturPräteritum im Indikativ”). We suggest, however, that würde + infinitive fulfils the same (temporal or modal) function in past contexts as werden + infinitive fulfils in present and future ones. As for the status of the relevant reference point, the cognitive matrix predicates indicate a strong link to (the knowledge sphere of) the speaker. In (23), (24) and (26) this is the first person subject (singular and plural), whose epistemic stance is expressed by means of würde + infinitive. Moreover, the temporal perspective seems to be irrelevant since the speaker’s personal involvement is concerned: it can either be past in (23), (25), and (26) or present in (24). Furthermore, the text mostly contains some explicitly mentioned pieces of information which are interpreted by the speaker as pointing towards the conclusion that is formulated by würde + infinitive: mein Vater war erschüttert ‘my father was so shocked’ in (23), so etwas ‘such a thing’ in (24), dass in der Gegend Guerilleros und Paramilitärs sind ‘that there are guerilleros and paramilitaries in this area’ in (26). Starting from these information pieces, the speaker comes to the conclusion (introduced by a cognitive predicate) about the likelihood of the occurrence of the described event. We can therefore treat the explicitly mentioned information and the speaker-oriented matrix predicates as pieces of textual advice that serve as relevant reference points for the interpretation of the construction würde + infinitive. The nature of such reference points is identical with the relevant reference point we have assumed for the epistemic reading of the construction werden + infinitive (see Section 4). For (25), there is no link to the actual speaker: wusste ‘knew’ refers to the grammatical subject of the sentence, Martin Walser, and not to the speaker. However, there is a crucial textual advice in this sentence which triggers the interpretation of würde + infinitive: the temporal expression am Vortag der Wahl ‘the day before the elections’ builds a temporal frame in which the sequence Vortag der Wahl – Tag der Wahl ‘the day before the elections – the day of the elections’ is established. This temporal expression suggests that the situation formulated with würde + infinitive takes place in the future relative to the time of the matrix clause. Hence, the relevant reference point for the interpretation of würde + infinitive is a temporal expression pointing toward future time. In this respect this relevant reference point is equal to the reference point of the construction werden + infinitive elaborated in Section 4. In sum, for würde + infinitive in its non-subjective readings, we propose a construal configuration with two relevant reference points, which are identical with the reference points established for the indicative construction werden + infinitive. In this connection, we may claim that würde + infinitive is an inflectional form of werden + infinitive that itself is grounded by the preterit subjunctive form.

 Elena Smirnova and Tanja Mortelmans

To conclude, we can state that würde + infinitive indeed has two distinct variants in Present Day German. These two variants exhibit different construal configurations. The analytical alternative to the synthetic preterit subjunctive is a “true” grounding predication with only one relevant reference point fully equated with the ground, i.e. this variant exhibits a maximally subjective construal configuration of the ground. On the other hand, würde + infinitive serves as a variant of the indicative construction werden + infinitive which is grounded by the mood marker. In this reading, würde + infinitive exhibits a construal configuration which is not maximally subjective: two different relevant reference points may be assumed in its construal configuration, whereby each of them triggers another functional interpretation of this construction (either temporal or epistemic). The functional interpretations of würde + infinitive directly correspond to the functional interpretations of werden + infinitive. 7. Conclusion In this chapter, we have argued that grounding is a gradient phenomenon, similar to the closely connected phenomena of grammaticalization and subjectification (subjectivity). On the basis of the gradual nature of grounding, we suggest that there are linguistic elements that may exhibit a stronger or a weaker degree of grounding, i.e. they are grounding predications to a higher or lesser extent. In order to elaborate the degree of grounding of a particular linguistic element, we have highlighted the role of the relevant reference point in the construal configuration of a linguistic unit. We have discussed the status of contextually present pieces of information that serve as relevant reference points and the extent to which they are able to influence the functional status of linguistic elements. In doing so, we suggested a useful tool for distinguishing various degrees of grounding: if a relevant reference point (R) in the construal configuration of a linguistic element can be established, which is associated with (some aspect of) the ground but cannot be automatically or fully equated with it (R≠G), the linguistic element under consideration cannot be considered a “true” grounding predication. For this reason, the construal configuration does not show a maximally subjective construal of the ground. Using the example of the synthetic preterit subjunctive, and the analytical constructions werden + infinitive and würde + infinitive, we have shown how the proposed instrument can be applied in the particular analysis of linguistic elements. Moreover, we hope that the proposed tool may also be used for testing and examining other linguistic elements in other languages whose status as grounding predication is problematic.

“More” and “less” grounding predications 

References Barceló, Gérard J. & Jacques Bres. 2006. Les temps de l’indicatif en français. Paris: Ophrys. Brisard, Frank. 2002. Introduction: The epistemic basis of deixis and reference. In Grounding: The epistemic footing of deixis and reference [Cognitive Linguistics Research 21], F. Brisard (ed.), xi–xxxiii. Berlin & New York: Mouton de Gruyter. Diewald, Gabriele. 1999. Die Modalverben im Deutschen: Grammatikalisierung und Polyfunktionalität [Germanistische Linguistik 208]. Tübingen: Niemeyer. Diewald, Gabriele. 2005. Werden und Infinitiv: Versuch einer Zwischenbilanz nebst Ausblick. Deutsch als Fremdsprache 42: 23–32. Diewald, Gabriele & Mechthild Habermann. 2005. Die Entwicklung von werden & Infinitiv als Futurgrammem: Ein Beispiel für das Zusammenwirken von Grammatikalisierung, Sprachkontakt und soziokulturellen Faktoren. In Grammatikalisierung im Deutschen [Linguistik: Impulse und Tendenzen 9], T. Leuschner, T. Mortelmans, & S. De Groodt (eds.), 229–250. Berlin: de Gruyter. Duden. 2005. Die Grammatik. Unentbehrlich für richtiges Deutsch. 7., völlig neu erarbeitete und erweiterte Auflage. Hg. von der Dudenredaktion. Mannheim: Dudenverlag. Eisenberg, Peter. 2004. Grundriß der deutschen Grammatik. Vol.2: Der Satz. 2., überarbeitete und aktualisierte Auflage. Stuttgart & Weimar: Metzler. Fabricius-Hansen, Cathrine. 1986. Tempus fugit. Über die Interpretation temporaler Strukturen im Deutschen. Düsseldorf: Schwann. Flämig, Walter. 1991. Grammatik des Deutschen: Einführung in Struktur- und Wirkungszusammenhänge. Erarbeitet auf der theoretischen Grundlage der “Grundzüge einer deutschen Grammatik”. Berlin: Akademie. Fritz, Thomas. 2000. Wahr-Sagen: Futur, Modalität und Sprecherbezug im Deutschen [Beiträge zur germanistischen Sprachwissenschaft 18]. Hamburg: Helmut Buske. Helbig, Gerhard & Joachim Buscha. 2001. Deutsche Grammatik: Ein Handbuch für den Ausländerunterricht. Berlin: Langenscheidt. Itayama, Mayumi. 1993. Werden – modaler als die Modalverben. Deutsch als Fremdsprache 30: 233–237. Kasper, Walter. 1987. Semantik des Konjunktivs II in Deklarativsätzen des Deutschen [Reihe Germanistische Linguistik 71]. Tübingen: Niemeyer. Kotin, Michail L. 2003. Die werden–Perspektive und die werden–Periphrasen im Deutschen. Frankfurt etc. : Peter Lang. Langacker, Ronald W. 1991a. Foundations of Cognitive Grammar. Volume II: Descriptive Application. Stanford: Stanford University Press. Langacker, Ronald W. 1991b. Concept, Image and Symbol. The cognitive Basis of Grammar. Berlin & New York: Mouton de Gruyter. Langacker, Ronald W. 1999. Losing Control: Grammaticization, Subjectification, and Transparency. In Historical Semantics and Cognition, A. Blank & P. Koch (eds.), 147–175. Berlin & New York: Mouton de Gruyter. Langacker, Ronald W. 2002. Deixis and subjectivity. In Grounding: The epistemic footing of deixis and reference [Cognitive Linguistics Research 21], F. Brisard (ed.), 1–28. Berlin & New York: Mouton de Gruyter. Matzel, Klaus & Bjarne Ulvestad. 1982. Futur I und futurisches Präsens. Sprachwissenschaft 7: 282–328.

 Elena Smirnova and Tanja Mortelmans Mortelmans, Tanja. 2001. An introduction to Langacker’s grounding predications: Mood and modal verbs in German. In Modalität und mehr/Modality and more, H. Vater & O. Letnes (eds.), 3–26. Trier: Wissenschaftlicher Verlag Trier. Mortelmans, Tanja. 2002. Wieso sollte ich dich küssen, du hässlicher Mensch! A study of the German modals sollen and müssen as “grounding predications” in interrogatives. In Grounding: The epistemic footing of deixis and reference [Cognitive Linguistics Research 21], F. Brisard (ed.), 391–432. Berlin & New York: Mouton de Gruyter. Mortelmans, Tanja. 2004. Grammatikalisierung und Subjektivierung: Traugott und Langacker revisited. Zeitschrift für germanistische Linguistik 32: 188–209. Mortelmans, Tanja. 2006. Langacker’s subjectification and grounding: A more gradual view. In Subjectification: various paths to subjectivity [Cognitive Linguistics Research 31], A. Athanasiadou, C. Canakis, & B. Cornillie (eds.), 151–176. Berlin & New York, Mouton de Gruyter. Smirnova, Elena. 2006. Die Entwicklung der Konstruktion würde + Infinitiv im Deutschen. Berlin: de Gruyter. Thieroff, Rolf. 1992. Das finite Verb im Deutschen: Tempus – Modus – Distanz [Studien zur deutschen Grammatik 40]. Tübingen: Narr. Vater, Heinz. 1975. werden als Modalverb. In Aspekte der Modalität [Studien zur deutschen Grammatik 1], J. P. Calbert & H. Vater (eds.), 71–148. Tübingen: Narr. Weinrich, Harald. 2003. Textgrammatik der deutschen Sprache. Unter Mitarbeit von M. Thurmair, E. Brendl und E.-M. Willkop. Zweite revidierte Auflage. Hildesheim: Olms. Ziegeler, Debra. 2006. Omnitemporal will. Language Sciences 28: 76–119.

New current relevance in Croatian Epistemic immediacy and the aorist Mateusz-Milan Stanojević and Renata Geld Zagreb University

Croatian grammarians make conflicting claims about the Croatian aorist. Some consider it a general past tense used with perfective verbs denoting completed past while others believe that it denotes a recent past action. Both accounts focus on one of aorist uses, failing to explain some others, notably future reference. Based on a corpus study and speaker judgments, we claim that the Croatian aorist has a dual role as an aspectual and tense phenomenon expressing epistemic immediacy. We propose an analysis that contrasts an epistemically immediate virtual plane (which includes the aorist) and an epistemically distant virtual plane (which includes the Croatian perfect tense), offering a unified explanation of all aorist uses and fitting in with other Croatian and cross-Slavic data. Keywords: epistemic immediacy, current relevance, aorist, Croatian

1. Introduction1 There are two ways of analyzing the aorist in Croatian grammars. Traditional grammars (e.g. Katičić 1991: 57–59; Barić et al. 1995: 413) describe the aorist as a general past tense used with perfective verbs denoting completed past. More contemporary-minded grammarians analyze the aorist as expressing recent past action (e.g. Raguž 1997: 185; Težak & Babić 1994: 265; Silić & Pranjković 2005: 192). For instance, a sentence such as: (1) Stigoh iz daleka. arrive.perf.aor.1stsg from far ‘I have arrived from far away’ 1. The authors would like to thank Ricardo Maldonado Soto, Günter Radden, and the editors of the volume for their valuable suggestions and discussions. All remaining inadequacies are our own.

 Mateusz-Milan Stanojević and Renata Geld

would be analyzed differently by the proponents of the two views. Whereas the former would claim that the aorist refers to absolute past time (with no connection to the present), the latter would analyze it as a tense expressing recent past action, completed just before the moment of speaking (this will be the view taken in this chapter; hence we have translated (1) using the English present perfect). Things are additionally complicated by views on the Croatian perfect tense,2 that is claimed to symbolically express current relevance (at least in some contexts) by both schools (cf. e.g. the “traditionalist” Katičić (1991, 1992) and the more recent views by Silić & Pranjković (2005: 192–193)). Thus, if the Croatian perfect tense was to be used in this sentence: (2)

Stigao sam arrive.perf.participle.sg.masc be.imperf.pres.1stsg iz daleka. from far ‘I arrived from far away’

the very fact that the verb form stigao sam is in the perfect tense is claimed to connect it with the present. However, in spite of these claims and its name, the Croatian perfect is a general past tense, because it does not symbolically express perfect meaning (in the sense of current relevance, cf. e.g. Comrie 1976: 52), as shown by Stanojević & Geld (2005). Indeed, the Croatian perfect tense does not correspond to Dahl’s criteria for a perfect – the answers on the perfect questionnaire (Dahl 2000: 800–809) show that it is a general narrative past tense rather than a perfect (Lindstedt 2000: 366), as reflected in our translation of (2) where we used the English past simple.3 Thus, the proposed neat distinction between a currently relevant perfect and a general past aorist does not stand. Traditional grammars also claim that the aorist is stylistically marked because it is primarily used in narrative fiction, and not in non-fictional texts or everyday speech. Thus, its use is claimed to evoke “vividness, [...] succinctness and [...] intimacy” (Katičić 1991: 58), whereas the use of the much more common perfect tense is considered stylistically neutral (Katičić 1991: 54). However, more recently, the aorist seems to be gaining momentum among young speakers, both in everyday spoken language and in various types of electronic communication – short text 2. The Croatian perfect tense is formed using the imperfective present tense of the verb biti (‘to be’) and the l-participle of the (perfective or imperfective) content verb, which is inflected for gender and number. 3. Indeed, this is acknowledged by Croatian grammarians who claim that the Croatian perfect tense expresses current relevance only in some contexts, whereas it is “neutralized” in others, which means that it expresses absolute past (Katičić 1991: 53; Katičić 1992: 177–178; Barić et al. 1995: 407).

New current relevance in Croatian 

messages (Žic-Fuchs 2002–2003), emails and web forums (Tošović 2006). This is also noted by a more recent Croatian grammar (Silić & Pranjković 2005: 192). The question that arises is: what does the aorist express and how is it integrated in the grammatical system of Croatian. Building on the points concerning the relationship between the aorist and the perfect in Croatian put forward by Stanojević & Geld (2005) we claim that the aorist expresses epistemic immediacy, i.e. the immediacy of the process expressed in relation to the speaker and her knowledge about the event. We propose an analysis of the epistemic immediacy of the aorist using an epistemically immediate virtual plane, V1, which is in contrast to an epistemically distant virtual plane V2. Such an analysis of the aorist sides with its dual role as an aspectual and tense phenomenon (put forward by Lindstedt 1985), and is in accordance with cross-Slavic data (specifically findings about the aorist in Serbian, Bulgarian and Macedonian). In order to substantiate our claims, we will be examining examples of usage of the aorist listed in Croatian grammars, appearing in the Croatian National Corpus (www.hnk2.ffzg.hr) and on the Internet. Based on our findings, we developed a questionnaire to obtain judgments on grammaticality and usage tendencies of the aorist and the perfect, which we used with 159 native speakers of Croatian. The chapter is organized as follows. In Section 2 we identify usage patterns of the Croatian aorist. This will serve as the basis for the third section, in which we examine aorist usage tendencies and report on the results of the questionnaire. Section 4 discusses the results, presents a cognitive analysis of the symbolic properties of the aorist on the epistemically immediate plane. The chapter ends with a conclusion and gives possible avenues for future research. 2. Usages of the Croatian aorist As we mentioned above, the core meaning of the aorist in traditional grammars of Croatian is described as denoting actions that happened in the past and have no relevance for the current state of affairs. These features correspond to Givón’s features of the past tense rather than the perfect tense (Givón 2001: 283–297): it should refer to past action, be used with absolute reference, be terminated, have no current relevance and be sequential. As a result, it should also be primarily used in narrative contexts. Such a view of the aorist focuses on only one of its uses ignoring the three others. Thus, the aorist may be used for recent actions, proverbs, and actions following the speech event. Interestingly, these uses appear both in traditional fiction (which is the basis of grammatical analysis in most traditional grammars of Croatian) and in current everyday spoken and written communication. Moreover, the aorist is used more frequently in the first person singular than in any

 Mateusz-Milan Stanojević and Renata Geld

other form. In this section we will look at these usages of the aorist: the narrative aorist, the aorist of recent actions, the aorist of proverbs, the aorist of future actions, and we will briefly examine the predominance of the first person singular. 2.1

Narrative aorist

Here is a typical example of the aorist as used in a narrative context (from a Croatian novel):4 (3)

Onda me zaokupio then i.gen overcome.perf.participle.masc.sg stid: pobjegoh u šumu, shame.nom run away.perf.aor.1stsg in forest.loc tamo se isplakah i there refl. cry.perf.aor.1stsg and zaspah. fall asleep.perf.aor.1stsg ‘Then I was overcome by shame: I ran away to the forest, where I cried my eyes out, and fell asleep’

The aorist is used to describe actions occurring sequentially one after another. Significantly, in this story, the string of three aorists seems to move the story forward, possibly indicating the culmination (and coda) of a story (Lindstedt 1994: 38–39). Although more seldom, such usages of the aorist also appear in everyday communication on Internet forums. Here is a quote from a blog (http://crome. mojblog.hr/1/jedan-neobican-i-divan-dan/68122.aspx):5 (4)

Odgovorih, i u roku par reply.perf.aor.1stsg and in term.loc couple minuta dobih odgovor. minute.gen.pl get.perf.aor.1stsg answer.nom. Pa ja opet odgovorih. Pa then i.nom again reply.perf.aor.1stsg then

4. Unless stated otherwise, all examples cited from this point on have been taken from the Croatian National Corpus (CNC) available at: www.hnk2.ffzg.hr. The CNC contains only written language taken from traditional and more recent fictions and newspapers. See Section 3.1 for a short discussion of the CNC. 5. We used the Webcorp (www.webcorp.org.uk) and Google (www.google.hr) to search for examples of the aorist on the Internet. Although no guarantee can be made as to the mother tongue of the authors of the blogs and posts, we narrowed our search only to those appearing on the .hr domain.

New current relevance in Croatian 

dobih odgovor. get.perf.aor.1stsg answer.nom ‘I replied, and I got an answer in a matter of minutes. So I replied again. And I got another answer’ Thus, in examples (3) and (4) several perfective verbs in the aorist appear in sequence (thus iconically matching the order of events in the world), in this sense being similar to a past (rather than a perfect) tense. They refer to bounded events, seen in their totality, and following one another, providing a way to move the story forward. 2.2

The aorist of recent actions

The aorist may also refer to an event that is recent, i.e. closely precedes the speech event (see e.g. Raguž 1997: 185; Težak & Babić 1994: 265). Here is an example of such usage from a nineteenth century novel: (5)

Upravo danas dođoh sa just today come.perf.aor.1stsg with zaručnicom amo da se fiancée.instr here to refl. poskrbimo za vjenčanje. take care.perf.pres.1stpl for wedding ‘I have just arrived with my fiancée to make arrangements for the wedding’

In this example upravo ‘just’ shows that the action denoted by the aorist is recent, and may, in fact, be considered currently relevant by implicature (cf. Stanojević & Geld 2005; Radden & Dirven 2007: 215). Let us consider a more recent example from an online forum, in which the person is saying that s/he has just come back from the city and logged on (http://kikyyy.bloger.hr/default.aspx?tag=hrana): (6) Eto mene, dodjoh iz grada.6 here i.gen come.perf.aor.1stsg from city ‘Here I am, I have returned from the city’ The sequence eto mene ‘here I am’, just like upravo, is a clear sign showing that the action of returning from the city has happened recently.7 6. No changes have been made to the original spelling of the examples from the internet, where changes or omissions of Croatian graphemes (such as č, dž, š, ž) are frequent. Thus, in (6) the digraph dj is used for đ. 7. For more on the meaning of evo, eto and eno see Žic-Fuchs (1996: 57–59), and works cited therein.

 Mateusz-Milan Stanojević and Renata Geld

2.3

The aorist of proverbs

Thirdly, the aorist is also used in proverbs, and in this case it is said to express a general, timeless truth (cf. Katičić 1991: 59; Težak & Babić 1994: 265). For instance, the proverb: (7) Dva loša ubiše Miloša. two bad.pl kill.perf.aor.2ndpl Miloš.gen ‘Two bad persons have killed Miloš’ is used when (two or more) people (i.e. dva loša ‘two bad persons’) conspire against somebody (i.e. Miloš), which is harmful to this person (denoted by ubiše ‘kill’). In examples such as (7) the aorist, rather than being timeless (in other words generic, i.e. a result of the generic is specific mapping common in proverbs), seems to be warranted by its characteristic of referring to an action that is relevant for the situation at hand. 2.4

The aorist of future actions

In addition to this, the aorist may be used to express future events (Težak & Babić 1994: 265; Katičić 1991: 59). More specifically, the action seems to be inevitable or planned, as in (example from Težak & Babić 1994: 265): (8) Ljudi, pogiboh! people.voc die.perf.aor.1stsg ‘People, I’m dying!’ In (8) the aorist means that one’s dying is inevitable, inviting the inference that the person needs immediate help – it would in fact be interpreted as a cry for help. Indeed, as reported in one of the grammars, the speaker is so certain that the action will take place “that he sees it as a past action” (Silić & Pranjković 2005: 192). The action may also be planned by the speaker to take place in the future (cf. http://www.reiki-hara.hr/forum2/viewtopic.php?p=36154): (9) 2.5

odoh i ja sutra, ali go.perf.aor.1stsg and I tomorrow but se vraćam kroz tri dana. refl. return.imperf.pres.1stsg through three days ‘I am also leaving tomorrow, but I will be back in three days’

The predominance of the first person singular

Data from a frequency dictionary of Croatian (Moguš, Bratanić & Tadić 1999) indicate that the first person singular is the most frequently used form of the aorist.

New current relevance in Croatian 

Thus, for instance, the data for the most frequent Croatian perfective verb reći ‘say’ shows that the frequency of the first person singular in sentences with a verb in the aorist is three times higher than that of the next most frequent pronoun, the third person plural. The same pattern of much higher frequencies for the first person singular is observable in all six of the Croatian perfective verbs we chose for this study: doći ‘come’, otići ‘leave’, reći ‘say’, vidjeti ‘see’, dobiti ‘get’ and kupiti ‘buy’. This may be due to the importance people attach to their own direct experience of an event (as noticed by traditional grammarians, Katičić 1991: 58; Težak & Babić 1994: 265). In summary, the Croatian aorist is not only used to refer to the time of completed past events but also to many more temporal situations. The aorist is used for currently relevant actions (cf. examples (5)–(7)), both generically and specifically, and for imminent and planned future actions (examples (8)–(9)). Moreover, it is more frequently used in the first person singular than in any other form. Thus, the issue we are faced with is the following: which of the usages (the completed past, the recent past, the future) is to be considered the core motivating other usages? Our claim in this chapter is that that the aorist primarily describes events in the epistemic region closest to the speaker’s vantage point (Langacker 1991), and that all other usages are based on this one. In order to prove this claim, we will look at the tendencies of usage of the aorist in everyday language. 3. Current usage tendencies of the aorist In order to investigate the patterns of usage of the aorist we listed the most frequent Croatian perfective verbs based on a frequency dictionary of Croatian (Moguš, Bratanić & Tadić 1999). We chose six verbs according their frequency, which resulted in the following list: doći ‘come’, otići ‘leave’, reći ‘say’, vidjeti ‘see’, dobiti ‘get’ and kupiti ‘buy’.8 Secondly, we manually searched the Croatian National Corpus for the aorist forms of the verbs.9 Then, we examined the usage of the 8. Although the Croatian National Corpus is tagged, its tagging is very unreliable for forms that are as rarely used as the aorist. For instance, 169 of the 170 forms of vidjeh see-PERF. AOR.1st.sg in the corpus were tagged as nouns (and the remaining form was tagged as an adjective). Therefore, we had to search for particular verbs rather than a random sample of the aorist. Although the choice of verbs may have influenced our results, taking the most frequent verbs was the only practicable way of performing a corpus study. Note that the verbs were different in their semantic characteristics: there were two verbs of movement, a perception verb, a verb of speaking and two benefactive verbs. 9. The aorist second and third person singular are homographs with the third person perfective present, and because of a great number of them (and problems with tagging) they were excluded from the analysis.

 Mateusz-Milan Stanojević and Renata Geld

aorist in current Croatian on the basis of two procedures: investigation of usage in the Croatian National Corpus and on the Internet, and using speaker judgments concerning grammaticality and usage tendencies. In the following two subsections we will look at each in turn. 3.1

Current usage tendencies: The corpus

We extracted all of the tokens of the aorist from the Croatian National Corpus as described above (i.e. excluding the second and third person singular which are homographs with the third person imperfective present tense). We examined the entire corpus, i.e. both its fictional and non-fictional segment. It is important to note that this corpus has not yet reached its planned size and that it is not yet representative of current Croatian usage, because it largely consists of various fictional and newspaper texts, with no other functional styles or spoken language. Despite these deficiencies, it is still the best available source of authentic language material for Croatian. Our search resulted in 273 examples of the aorist of doći (21 of which from recent non-fictional texts), 168 of otići (37 from recent texts), 885 of reći (261 in recent non-fictional texts), 265 of vidjeti (15 from recent non-fictional texts), 49 of dobiti (11 of which were from recent texts) and 13 tokens of kupiti (none from non-fictional texts). We checked how time reference was constructed, looking for time adverbials and other ways of referring to the time of the action. We expected the aorist to be mainly used without precise time identification (such as jučer ‘yesterday’, u srijedu ‘on Wednesday’, etc.) or with time identification that enables the construction of the event from a present point of view (including items such as upravo ‘just’, još ‘yet’, etc.). This would go to show that the time of the event is not in focus, although the reported event happened in the past. The results show that the majority of tokens of the aorist of the six selected verbs are indeed used without precise time identification, which was the case in 67.5% of the cases. Here is a typical example: (10)

Besmislenijih rečenica i pointless.comp.gen sentence.pl.gen and kadrova dugo ne vidjeh, čak scen.pl.gen long not see.perf.aor.1stsg even ni na hrvatskoj dalekovidnici. nor on Croatian.loc television.loc ‘I have not seen more pointless sentences or scenes for a long time, even on Croatian television’

In example (10) no reference to the time of the action is given, i.e., there is no precise time reference in the example. The only indication of time is dugo ‘for a long

New current relevance in Croatian 

time’, which is a way of including the present moment into the time frame. Although the process of “not seeing” refers to the past, it ends at the time when the present moment starts, i.e. it refers to the recent past. Recent past usage is clearly visible in five of the six verbs in the sample: vidjeti, doći, reći, dobiti and kupiti, from markers of recent past action such as upravo ‘just’ (cf. example (5)) eto and evo (cf. example (6)), još ‘yet’ and markers expressing consequences of the action (e.g. usage of the present tense in the following sentence, and visible results of the action).10 For instance, rekoh is frequently used as a discourse device reminding the collocutor of what the speaker has said during the present speaking situation. This is most frequently done by saying kako rekoh or kao što rekoh ‘as I have said’: (11)

Nije ni sigurno, kao što not.be.pres.3rdsg nor certain, like that rekoh, da je say.perf.aor.1stsg that be.imperf.pres.3rdsg prisega protuustavna. oath.nom.sg unconstitutional.nom ‘It is not certain, as I have said, that the oath is unconstitutional’

Thus, in five of the verbs (dođoh, rekoh, vidjeh, dobih and kupih) epistemic immediacy is exhibited through referring to recent past action. Aorist may also express inevitable future action (see Section 2.4.). This was the case with odoh, which was sometimes used in the corpus with future markers such as sutra ‘tomorrow’, prekosutra ‘the day after tomorrow’, referring to the immediate future (as exemplified in (9)). Judging by the number of examples from the corpus and grammars, it would seem that in current Croatian only some verbs invite planned/inevitable future interpretation (such as: otići ‘leave’, poginuti ‘get killed’, nestati ‘disappear’, which are frequently mentioned in grammars; cf. e.g. Težak & Babić 1994: 265; Silić & Pranjković 2005: 192). However, older grammatical treatises show that this was quite a regular use when the aorist was generally more frequently used (cf. Stojićević 1956: 34 and references cited therein). Therefore, the immediate future use of otići is not a vagary of grammar, but a systematic fact – a result of the epistemic immediacy of the aorist. The remaining 32.5% of examples are primarily used with deictic expressions (yesterday, two days ago), and never include exact dates, years or months. This is in accordance with the nature of the aorist, which does not focus on the time of the 10. We also extracted a random sample of 743 tokens of the first person masculine of the perfect tense for all of the six verbs in order to compare it with the aorist forms. There were 462 cases that were used with precise time reference (62%), and the remaining 38% were not. Moreover, dates, months and years are not uncommon with the perfect tense. This is indicative of the perfect being a basic general past tense that does not symbolically express current relevance but epistemic distance (Stanojević & Geld 2005).

 Mateusz-Milan Stanojević and Renata Geld

event, but its immediacy from the point of view of the speaker. Deictic expressions are crucially determined by the position of the deictic center (normally I-herenow), and require the accessibility of the deictic center in order to be understood, no matter whether they are precise in pointing to a time (as in yesterday) or not (as in just). This is precisely why they can be used with the aorist, which requires the event to be immediately accessible from the present time (which usually coincides with the deictic center). In other words, although deictic expressions of time do not need to express recent past, they always allow an immediacy interpretation by virtue of their mandatory connection with the deictic center. It is this trait that coincides with the nature of the aorist. Current usage tendencies and corpus results suggest that the aorist exhibits the characteristics of both aspect and tense, as suggested by Lindstedt (1985). It takes only (morphologically) perfective verbs and refers to an event as a bounded whole which is available to the speaker in its entirety. Being perfective, the action referred to by the aorist cannot indicate the present moment. Thus, it is removed to the immediate past or (rarely) to the immediate future, which is a clearly aspectual trait. In this sense, the aorist in Croatian is parallel to the usage of the present tense of perfective verbs in all Slavic languages, which can never refer to an ongoing situation (cf. Dalewska-Greń 1997: 348). Note, however, that the default point of removal for the aorist tends to be the immediate past (and may be the immediate future only in particular cases). This is in opposition to the perfective present tense in Croatian, where the default point of removal is (immediate) virtuality (Geld & Zovko Dinković 2007). It is because of this default removal to the past that aorist exhibits characteristics of tense. Importantly, the data suggest that the removal is to an immediately available point. In the next section we turn to speakers’ judgments concerning the immediacy of the aorist. 3.2

Current usage tendencies: Speakers’ judgments

In order to test the epistemic immediacy of the action in the aorist, we constructed a questionnaire to obtain grammaticality judgments and usage tendencies from speakers of Croatian. Based on the corpus results, we hypothesized that the speakers’ reports would reflect a trend for the aorist to be distinguished from the perfect and the present tense by being predominantly used for epistemically immediate actions (recent past and immediate future). In the following subsection we will describe the sample and the questionnaire and then report on the results. 3.2.1 Participants, questionnaire and results In line with the view that the aorist has been gaining momentum among younger speakers of Croatian (Žic Fuchs 2002–2003) we decided to test primarily younger

New current relevance in Croatian 

speakers of Croatian. The sample comprised 159 native speakers of Croatian, all university students (124 female and 32 male) aged 18–33 (median age: 19).11 The questionnaire consisted of five tasks, in which participants were asked to report on their usage and metalinguistic judgments concerning the aorist, the perfect and the present tense. Methodologically speaking, a four-point Lickert scale was used (forced choice; I would never say this; I could say this; I often say this and I always say this), as well as Y/N questions and a question requiring participants to rate the order of likelihood of usage on a scale from 1 to 5. In three of the Y/N questions space was left for additional comments. The questionnaire was designed to have the participants report on the immediate future and recent past usage of the aorist (in opposition to the perfect of a perfective verb and the present tense of an imperfective verb) and for the possibility of counter-sequentiality in narrative usages. 3.2.2.1 The aorist and the immediate future In the first task, we tested the immediate future usage of the aorist. Participants were given a situation (“you are leaving the house and you are letting others know that you are leaving”), and were asked to report on their usage of the immediate future meaning of odoh (cf. example (9)) and the perfective perfect otišao sam or otišla sam. Participants reported that the immediate future meaning of the aorist odoh is a strong possibility in their usage. In the first task, a total of 52% of participants said they could use the aorist, and 20% report that they use it with some frequency. The remaining 28% of participants said they would never use the aorist. As opposed to this, participants’ reports indicate that the perfect tense is not normally used with immediate future meaning – 65% of participants say they would never use the perfect tense in this construction, 26% say they could use it, and the remaining 9% say that they use it frequently or always. Thus, the perfect and the aorist are clearly distinct with regard to the immediate future meaning. 3.2.2.2 The aorist and the recent past Recent past usage of the aorist (as compared with the perfect) was tested in three final tasks. One of the tasks was purely diagnostic, and its aim was to see whether participants would declare a sentence in the perfect more distant in time than a sentence in the aorist. Participants were asked to say whether they considered one of two sentences (one in the aorist: Dobih na lotu ‘I have won the lottery’, and the other in the perfect tense: Dobio sam na lotu ‘I won the lottery’) more distant in 11. Participants spent most of their life in different regions of Croatia, which does seems to have an influence on their usage of the aorist (e.g. speakers from the coastal region report less aorist usage), but we will not be focusing on this aspect in the paper.

 Mateusz-Milan Stanojević and Renata Geld

time than the other. Indeed, 48% of participants consider the perfect usage more distant in time, 19% consider the aorist to be more distant, and 33% consider the two sentences equally distant. Conversely, speakers judge the aorist to be more immediate in time than the perfect. The next task tested for the recent past usage of the verb reći. Participants were given a situation (“your collocutor has not heard what you have just said, and you are repeating your words”). They were asked to report on the recent past usage of the aorist rekoh and perfect rekao sam/rekla sam. The results of this task reinforce the results for immediate future, clearly showing that the aorist is perceived as referring to the recent past. Most participants rate the aorist form as something they could say (43%), or as a construction that they use with some frequency (29%). Only 28% of participants say they would never use the aorist in this situation. As opposed to this, the perfect construction is vague as to the time of the action and resultativeness (Stanojević & Geld 2005), so we expected it to be reported as being used with some frequency in this type of construction. Indeed, only 11% of participants say they would never use the perfect, whereas the rest report to use it with some frequency (51%) or indeed always (18%). These results confirm vagueness of the perfect and its status as a “neutral” option in various past contexts. High scores for the aorist show that it is not on the decline and that the recent past usage is indeed one of the contexts where it is used. The aim of the fourth task was to test speakers’ judgments concerning the time reference of the aorist and the perfect. Participants were given two sentences: Kupih cipele (aorist: I have bought a pair of shoes) and Kupio sam/Kupila sam cipele (perfect tense: I bought a pair of shoes). They were also given a set of four time adverbials: upravo ‘just’, danas ‘today’, jučer ‘yesterday’ and prošle godine ‘last year’, and the option of using no adverbial of time. Participants’ task was to rank the five options for each sentence, 1 being their first choice for this sentence and 5 being their least likely choice. Results of this task show that speakers consider upravo ‘just’ to be clearly connected with the aorist, as opposed to the perfect tense. It was ranked as the first choice with the aorist sentence in 42% of the cases, and as the second choice by additional 25% participants. As opposed to this, the perfect was ranked as the first or second choice in only 27% of the cases. Figure 1 shows these trends. The remaining data for the aorist show that it is preferred in more recent past contexts. Thus, the more distant the time point, the lower the participants ranked it on the scale. The median ranking for danas ‘today’ was 3 (with most participants (66%) ranking it at 2 and 3), the median ranking for jučer ‘yesterday’ was also 3, but most participants (70%) ranked it at 3 and 4. Finally, the median ranking for prošle godine ‘last year’ was 4 (with 26% of participants ranking it at 4, and as many as 47% of participants ranking it at 5).

New current relevance in Croatian  45 40 35 Percent

30 25

Aorist Perfect

20 15 10 5 0 1

2

3

4

5

Ranking

Figure 1.â•‡ Ranking of upravo in the perfect and the aorist

As opposed to this, the results of the perfect do not present such a clear-cut picture, which is to be expected because of its vagueness. The perfect is not used with upravo ‘just’, as seen above. The remaining rankings are somewhat different than for the aorist (median ranking for danas ‘today’ is 2, median ranking for jučer ‘yesterday’ is 3, and median ranking for prošle godine ‘last year’ was 4), but a similar trend of the relation of distance of time with lower ranking is evident. The results for upravo seem to suggest that recent past is not usually connected with the perfect. Danas and jučer may easily be construed both as being within the same time frame as the reference point and as not being within the same time frame. This could be the reason for their relation with the perfect, which is vague in this respect. Finally, the type of action described (buying shoes) may also influence the ranking of the adverbials – it is an action that would easily be relevant only if it happened relatively recently (therefore prošle godine ‘last year’ is ranked as low down on the scale). Although such reasoning suggests that some additional research would be necessary, our results may be taken as an indicator of existing trends. 3.2.2.3 The aorist and counter-sequentiality Task 2 tested the narrative usage of the perfect, the pluperfect and the aorist. Participants were given the following three sentences: Ušao sam u sobu. Pogledao sam oko sebe. Već sam ranije na stolu vidio svoj rođendanski poklon (‘I entered the room. I looked around. I had already seen my birthday present on the table’). In the first item all three sentences were in the perfect tense (which, intuitively, corresponds to neutral spoken language usage). In the second item all of the verbs were in the aorist. In the last item, the first two verbs were in the perfect tense, and the last

 Mateusz-Milan Stanojević and Renata Geld

verb was in the pluperfect tense (which is considered to be the prescriptively correct). We asked the speakers to say whether they would use a particular sequence (by choosing yes or no), and asked them to comment when they would do it (space was left for comments). Corpus results and other research have shown that, in current language, narrative usage is not in the domain of the aorist, but rather in that of the perfect tense. Moreover, counter-sequentiality is not characteristic of the aorist: strings of aorists iconically refer to actions following one another (Lindstedt 1994: 37–39). As predicted, 74% of the participants judged that they could use the perfect in this type of construction, and 55% said that they could use the pluperfect, and some judged both as possible. Only 27% of participants said that they could use the aorist. This result for aorist is expected because the sequence was narrative and counter-sequential, as opposed to relating to spoken language and being sequential. All those who said they could use the aorist felt the need to comment on its usage. They predominantly mention various stylistic features of the aorist as the reason behind their choice (most comments refer to sounding poetic, archaic, creative, etc.), and some mention written language, emails and short text messages as appropriate for aorist usage. The results of the questionnaire show that speakers see the aorist as a tense that they primarily use in recent past and immediate future contexts. The immediate future and recent past spoken language usage (which includes the speech event within the same time frame) distinguish the aorist from the perfect tense. Although some speakers report that the aorist may be used narratively in countersequential contexts, the pluperfect and particularly the perfect are the more usual tenses in these contexts. 4. Discussion and cognitive analysis Let us briefly sum up what has been said so far. In addition to appearing in narrative contexts, where it is normally used to move the action forward, the aorist exhibits a tendency not to be used with precise time reference and to refer to recent past. The aorist can also be used in inevitable or planned future actions, and is seldom reported as an option in counter-sequential contexts. Finally, it is used with perfective verbs. A schematic characterization of the Croatian aorist must include all of the facts about its usage, including the recentness of the action, its general past and future usage. The Croatian aorist exhibits the marks of both an aspect and a tense: it is perfective – it refers to actions in their totality, hence necessarily removing them from the present time. However, it is also a tense in the sense that the default point of removal for most verbs is the recent past, unlike, for instance, for

New current relevance in Croatian 

the present tense of perfective verbs (see Section 3.1.). All of these components can be subsumed as epistemic immediacy of the action expressed by the aorist. 4.1

The aorist and epistemic immediacy

The aspectual nature of the aorist – its perfectiveness – warrants its removal from the speech event. The aorist usage tendencies and speakers’ judgments suggest that the aorist describes events in the epistemic region closest to the speaker’s vantage point. These events represent conceptual occurrences that belong to virtual reality based on the speaker’s knowledge about the world and its structure (Langacker 1987, 1991, 2001a, 2001b, 2003). Although the aorist does indicate actions completed before the speech event, it also involves departures from canonical viewing arrangements (Langacker 1990, 2003). In the canonical viewing arrangement, the grounded process is construed with a high degree of objectivity and the ground (G) itself is construed subjectively. The speaker (S) and the hearer (H) are viewers or conceptualizers viewing the profiled process from offstage. In such canonical viewing arrangements, the speaker simply reports on what happens in the world (Figure 2). However, in the case of the aorist, conceptualizers are not simple observers of the actual world events. They describe processes whose occurrence is conceptual rather than actual and the temporal occurrence of such processes is construed with extreme subjectivity (Langacker 2003: 16). The process designated by the aorist is construed in such a way that the speaker/hearer and the ground are included in the immediate scope (Figure 3). This element of the aorist schema explains the intuition of the grammarians who explain the aorist as a vivid and directly experienced action (e.g. Stojićević 1951: 25; Rončević 1956: 15–16; Katičić 1991: 58). Extreme subjectivity of the speaker facilitates seeing the action as recent if it refers to the past and inevitable or immediate if it refers to the future. Moreover, this analysis is supported by the fact that the aorist is predominantly used in the first person singular.

G

S H

PROCESS IS

Figure 2.â•‡ Grounding predication (adapted from Langacker 2003: 8)

 Mateusz-Milan Stanojević and Renata Geld

S G

H

PROCESS

IS

Figure 3.â•‡ Epistemic immediacy in the Croatian aorist

4.2

The aorist and the moment of speech

Whereas the aspectual nature of the aorist requires its removal to an epistemically immediate region, its tense (i.e. its contrast with the perfective present tense) requires this immediate region to be defaulted to the past. The second element of the aorist schema refers to the precedence of the action with relation to the moment of speech. Given that past usages predominate in the aorist, we believe that they are to be taken as the basis for schema formation. However, the schema also needs to be general enough to include future usages. Therefore, we propose that the aorist be analyzed as temporal precedence on the virtual plane, and that the type of correspondence between the ground on the actual plane and the ground on the virtual plane (i.e. whether the two grounds coincide or not) determine the time of the action (past or future). Figure 4 shows the time of speech as a box with a squiggly line, and the event preceding it as a box with a bold curved line: Let us reconsider example (6), repeated below as (12): (12) Eto mene, dodjoh iz grada. here i.gen come.perf.aor.1stsg from city ‘Here I am, I have returned from the city’

virtual plane

Figure 4.â•‡ Precedence of the action in the virtual plane

New current relevance in Croatian 

Virtual plane

Actual plane

Figure 5.â•‡ Recent past usage of the aorist

The action of the aorist precedes the time of speech, which means that the ground on the virtual plane coincides with the ground on the actual plane (cf. Figure 5).12 Let us now consider examples of general past usage in which several actions follow one another, such as (4) (here repeated as (13)): (13)

Odgovorih, i u roku par reply.perf.aor.1stsg and in term.loc couple minuta dobih odgovor. minute.gen.pl get.perf.aor.1stsg answer.nom Pa ja opet odgovorih. Pa then i.nom again reply.perf.aor.1stsg then dobih odgovor. get.perf.aor.1stsg answer.nom ‘I replied, and I got an answer in a matter of minutes. So I replied again. And I got another answer’

The analysis of such examples remains the same as the previous one, only taking into consideration multiple actions in the past. Figure (6) is a diagram of example (13). Finally, let us reconsider example (9), repeated below as (14): (14)

odoh i ja sutra, go.perf.aor.1stsg and I tomorrow ali se vraćam kroz tri dana. but refl. return-imperf.pres.1stsg through three days ‘I am also leaving tomorrow, but I will be back in three days’

12. The (relatively rare) usage of the aorist in proverbs is an offshoot from the recent past usage. The context and our knowledge (of the fact that we are dealing with the proverb) connected with the recentness/immediacy of the action basically facilitate a generic is specific reading.

 Mateusz-Milan Stanojević and Renata Geld Virtual plane

Actual plane

Figure 6.â•‡ General past usage of the aorist: Several actions in sequence Virtual plane

Actual plane

Figure 7.â•‡ Inevitable/planned future usage of the aorist

According to the schema we have proposed for the aorist, the action of the aorist (Figure 4) is presented in the virtual plane, and its virtual occurrence takes place immediately before the speech event in the same plane. The ground in the virtual plane does not coincide with the one of the speaker the time of speaking, thus the viewpoint is shifted and the event is conceptually viewed from future. The speech event in the virtual plane corresponds to the speech event in the actual plane. The action is completed before the speech event in the virtual plane. The expected occurrence of the actual event is indicated by the time expression sutra ‘tomorrow’. The event is placed after the speech event in the actual plane and it is linked to its virtual occurrence by the dotted correspondence line. The fact that the action has happened in the virtual plane, corresponds to its inevitability or certainty in the actual plane (cf. Figure 7). 4.3

Epistemic immediacy revisited: Two virtual planes

Note that our drawings so far do not include any sign of the first element of the aorist schema, i.e. of its immediacy with relation to the speaker. We believe that

New current relevance in Croatian 

the contrast of the aorist and the perfect warrants the introduction of two virtual planes, V1 and V2. Distinguishing between these two planes will enable us to explain the difference between aorist forms and l-participle forms and all their usages in Croatian. The first of these planes, V1, is closer to the actual plane, and enables the inclusion of the speaker within the immediate scope. All actions presented in this plane are epistemically immediate, and the aorist is a case in point. As noted above, recentness of the past actions, their directness and inevitability of future actions are all signs of this. Possible current relevance of the recent past usage of aorist is a matter of pragmatic inference in context, greatly facilitated by the perfectivity of the verb and the epistemic immediacy of the action. In other words, since the process is in the immediate scope together with the speaker, she can easily infer it as being currently relevant. In contrast, V2 is further away from the actual plane, and is used for epistemically distant actions (signaled, e.g., by the l-participle). Like the aorist, the l-participle also shows that the action has been completed before a reference point. Its placement on V2 means that it is does not enable a subjective construal. The lack of inclusion of the speaker in the immediate scope has another consequence – the speaker has no direct experience of the action.13 In line with this, the l-participle is used in referring to distant actions – such as the perfect tense (which is used with clear past meaning, and easily allows distant past contexts; cf. footnotes 8 and 9 and Stanojević & Geld 2005), the past perfect, the optative (expressing a wish rather than inevitable future action).14 Thus, the difference between the aorist and the perfect is one of subjectification. Whereas the aorist “pulls” the action of the virtual plane closer to the actual plane, including grounding in the immediate scope of view, the perfect “pushes” the virtual plane away from the ground, and the scope of view is no longer large enough to include the ground; thus the action described by the l-participle is distant. Such an analysis of epistemic immediacy of the aorist and the possible epistemic distance of the perfect has some specific consequences for the analysis of the two tenses and some more general consequences for the Croatian tenses system and the tense systems of other languages. In the description of the Croatian aorist and perfect, the V1 – V2 distinction would mean that the pragmatic inference of current 13. In this respect this interpretation is akin to a possible dominion analysis, in which the processes on the V1 would be within the speaker’s dominion (and thus controllable and directly experienced), whereas processes on V2 would be outside the speaker’s dominion (not directly experienced and outside of the speaker’s control). 14. Evidence from Bulgarian seems to support this analysis, because the l-participle is also used in expressing imperceptivity, a case in point of epistemic distance (cf. Koseska-Toszewa & Korytkowska 1993).

 Mateusz-Milan Stanojević and Renata Geld

relevance for the aorist and the perfect has two different sources: perfectivity of the verb and contextual factors for the perfect (in line with Stanojević & Geld 2005) and (in addition to the previous two) the immediate scope for the aorist. More generally, the V1 – V2 distinction invites further investigation of the combination of the two, for instance in the Croatian conditional (which is formed using the aorist form of the auxiliary and the l-participle of the main verb). Finally, the viability of this notion and its possible relation to dominion phenomena should be checked cross-linguistically and diachronically. 5. Conclusion In this chapter we have attempted to prove that the aorist in Croatian expresses epistemic immediacy. The examination of the aorist usage patterns has shown that the aorist may be used sequentially with absolute past reference, that it may be used for recent actions and for imminent and planned future actions. Aorist forms are predominantly used in the first person singular. In actual written usage, aorist predominantly refers to actions that have happened recently and that are currently relevant. Speakers report the aorist to be primarily used in recent past and immediate future contexts. Based on usage patterns, actual usage tendencies and speakers’ judgments, we have offered a schema of the aorist according to which the aorist expresses a virtual occurrence of an epistemically immediate action preceding the virtual ground. This schema subsumes all the usages of the aorist. Finally, we have proposed a distinction between two virtual planes, V1, which enables including the speaker in the immediate scope, and V2, which does not, and have offered some preliminary thoughts on how these notions may function in the Croatian tense system. References Barić, Eugenija. et al. 1995. Hrvatska gramatika. Zagreb: Školska knjiga. Comrie, Bernard. 1976. Aspect: An introduction to the study of verbal aspect and related problems. Cambridge & New York: Cambridge University Press. Dahl, Östen (ed.). 2000. Tense and Aspect in the Languages of Europe. Berlin & New York: Mouton de Gruyter. Dalewska-Greń, Hanna. 1997. Języki słowiańskie. Warsaw: Wydawnictwo Naukowe PWN. Geld, Renata, & Irena Zovko Dinković. 2007. Perfectives, imperfectives and the Croatian present tense. In Cognitive Paths into the Slavic Domain, D. Divjak & A. Kochańska (eds.), 111–148. Berlin & New York: Mouton de Gruyter. Givón, Talmy. 2001. Syntax: An Introduction. Vol. 1. Amsterdam & Philadelphia: John Benjamins.

New current relevance in Croatian  Katičić, Radoslav. 1991. Sintaksa hrvatskoga književnog jezika. Zagreb: Hrvatska akademija znanosti i umjetnosti. Katičić, Radoslav. 1992. Novi jezikoslovni ogledi. II, dopunjeno izdanje. Zagreb: Školska knjiga. Koseska-Toszewa, Violetta & Małgorzata Korytkowska. 1993. Z problematyki modalności imperceptywnej. Studia gramatyczne bułgarsko – polskie, Vols. 5 and 6: 177–193. Langacker, Ronald W. 1987. Foundations of Cognitive Grammar. Vol. 1: Theoretical Prerequisites. Stanford: Stanford University Press. Langacker, Ronald W. 1990. Subjectification. Cognitive Linguistics 1: 5–38. Langacker, Ronald W. 1991. Foundations of Cognitive Grammar. Vol. 2: Descriptive Application. Stanford: Stanford University Press. Langacker, Ronald W. 2001a. Cognitive linguistics, language pedagogy, and the English present tense. In Applied Cognitive Linguistics I: Theory and Language Acquisition, M. Pütz, S. Niemeier & R. Dirven (eds.), 3–37. Berlin & New York: Mouton de Gruyter. Langacker, Ronald W. 2001b. The English present tense. English Language and Linguistics 5.2: 251–272. Langacker, Ronald W. 2003 Extreme Subjectification: English Tense and Modals. In Motivation in Language, H. Cuyckens, T. Berg, R. Dirven, & K.-U. Panther (eds.), 3–24. Amsterdam & Philadelphia: John Benjamins. Lindstedt, Jouko. 1985. Bulgarian aorist and imperfect. X Nordiska Slavistmötet, Föredrag 102: 81–88. Lindstedt, Jouko. 1994. On the development of the South Slavonic Perfect. Three papers on the perfect. EUROTYP Working Papers. European Science Foundation. Lindstedt, Jouko. 2000. The perfect – aspectual, temporal and evidential. In Tense and Aspect in the Languages of Europe, Ö. Dahl (ed.), 365–383. Berlin & New York: Mouton de Gruyter. Moguš, Milan, Maja Bratanić, & Marko Tadić. 1999. Hrvatski čestotni rječnik. Zagreb: Školska knjiga. Radden, Günter & René Dirven. 2007. Cognitive English Grammar. Amsterdam & Philadelphia: John Benjamins. Raguž, Dragutin. 1997. Praktična hrvatska gramatika. Zagreb: Medicinska naklada. Rončević, Nikola. 1956. Aorist u hrvatskom jeziku. Zagreb: štampano kao rukopis. Silić, Josip & Ivo Pranjković. 2005. Gramatika hrvatskoga jezika za gimnazije i visoka učilišta. Zagreb: Školska knjiga. Stanojević, Mateusz-Milan & Renata Geld. 2005. Current relevance: A cognitive account. Glossos 6. http://www.seelrc.org/glossos/issues/6/stanojevic.pdf Stojićević, Aleksandar. 1951. Značenje aorista i imperfekta u srpskohrvatskom jeziku. Ljubljana: Slovenska akademija znanosti in umetnosti. Težak, Stjepko & Babić, Stjepan. 1994. Gramatika hrvatskog jezika: priručnik za osnovno jezično obrazovanje. 9. popravljeno izdanje. Zagreb: Školska knjiga. Tošović, Branko. 2006. Aoristno emajliranje i čatiranje. In Proceedings of the conference Jezik i mediji – jedan jezik: više svjetova. J. Granić (ed.), 703–710. Zagreb & Split: Croatian Applied Linguistics Society. Žic Fuchs, Milena. 1996. ‘Here’ and ‘there’ in Croatian: A case study of an urban standard variety. In The Construal of Space in Language and Thought, M. Pütz & R. Dirven (eds.), 49–62. Berlin & New York: Mouton de Gruyter. Žic Fuchs, Milena. 2002–2003. Communication technologies and their influence on language: an example from Croatian, Studia Romanica et Anglica Zagrebiensia XLVII–XLVIII: 597–608.

Aspect as a scanning device in natural language processing The case of Arabic* Lazhar Zanned

University of Manouba This chapter investigates the claim that aspect is a verbal device for constructing mental images of processes and these aspectual values correspond to the way(s) a process is profiled in different scanning modes: summary vs. sequential scanning and monosequential vs. polysequential scanning. This occurs through other cognitive abilities, such as chaining, scope of awareness, and zooming ((in) and (out)). The analysis is based on data from Standard Arabic. The study proceeds by presenting an overview of the aspectual system in Arabic and investigating the different facets involved: aspect as an image-building tool, aspect as a scanning device, scanning modes in Arabic (summary scanning, monosequential scanning, polysequential scanning) and, finally, scope of awareness and the process of zooming in aspectual representation. Keywords: summary vs. sequential scanning, monosequential vs. polysequential scanning, chaining, scope of awareness, zooming, Cognitive Grammar

1. Introduction Notions such as imagery, scanning, summation, and sequentiality are well established in Mental Imagery and Cognitive Grammar. In both theories, mental imagery (Kosslyn 1980, 1994, 1995, 1999; Tversky 1999) or mental simulation (Langacker 1987, 2008, 2009, 2010) involves basic functions in cognition in general and in verbal conceptualization in particular. Scanning is one of the basic modes of cognitive processing in making mental images by recalling the events we experienced in the past or by describing a * I would like to thank Adeline Patard, Frank Brisard and Mark Turner for their helpful comments.

 Lazhar Zanned

situation using symbolic resources. In Mental Imagery, scanning is characterized in terms of tracking distances and configurations of the recalled images. In Cognitive Grammar, summary and sequential scanning are seen as playing a role in characterizing the semantics of verbs (Langacker 1987, 2008), but this is still a matter of debate (Broccias & Hollmann 2007). In this chapter, we assess the claim that aspect is a verbal device for constructing mental images of processes (states and events) and that all aspectual values may be handled in terms of scanning. This occurs through other cognitive abilities, such as chaining, attention management based on scope of awareness, and zooming ((in) and (out)). The analysis is based on data from Standard Arabic. It is divided into three sections. Section (2) presents an overview of the aspectual system in Arabic derivational markers (reduplication, vowel lengthening, and affixation) and lexical markers expressing aspectual values such as nearness, inchoation, duration, habituality, and termination. Section (3) is devoted to aspect in imagery and presents an attempt to demonstrate that aspect is as an image-building tool (3.1), and that it is a scanning device in verbal communication (3.2). Section (4) deals with modes of aspectual scanning in Arabic. Three modes are investigated: aspectual summary scanning (4.1.1), aspectual monosequential scanning (4.1.2), and aspectual polysequential scanning (4.1.3). Finally, the roles of aspectual representation in attentional management are examined: framing the scope of awareness (4.2) and zooming (4.3).

2. The aspectual system in Arabic: An overview Historically, studies of aspect in Arabic focused on external elements, i.e. the verbal conjugations and the context that helps to indicate tense. Aspect was considered to be a matter of simple perfective-imperfective opposition. Studies that considered the interaction of time and aspect improved upon this simplicity (Comrie 1976, Fleisch 1979, Cohen 1989). The literature does indicate some basic aspectual values, such as those of inflectional, derivational, and lexical markers, but the internal constituency of the process denoted by the verbal stem has not been given sufficient attention yet. Arabic morphology is nonconcatenative. The formation of the lexical item is multi-tiered: each morpheme bears its own meaning and associates with the CVskeleton (template) following the rules of association (McCarthy 1979, 1981). The verb takassar, for instance, meaning that something broke in the past by itself into pieces, is derived as shown in (1). The simple opposition perfective vs. imperfective is inflectional. The inflected verbal forms are of two types:

Aspect as a scanning device 

(1) a

Tense, voice, aspect (perfective).

CV-skeleton

CV CVCCVC

Reflexive, durative, cumulative…

Special morpheme(s)

t

Reflexive.

Vocalic melody

Consonantal root

k s r

Lexical meaning (break).

→ takassar (‘break’, reflexive, repetitive, past tense, perfective)

a. stem + inflection: katab + ta (write + 2M.SG): past perfective (‘you wrote’), b. inflection + stem: ta + ktub (2M.SG + write): present, imperfective (‘you write/ you are writing’). Considered at this level, the aspectual system is based on the binary opposition: perfective vs. imperfective (Comrie 1976: 78, Cohen 1989: 183) and derivational, syntactic, and lexical markers optimize it. Derivational markers Derivational markers are morphemes incorporated in the CV-skeleton to create the augmented verbs, which comprise more consonants or vowels than the simple triliteral verbs. The triliteral verbs always keep the lexical aspect (Aktionsart) related to the lexical meaning expressed by the consonantal root, but the augmented verbs express other aspectual values which may overlap with the original lexical aspect. Derivational markers of aspect are of three types: (a) reduplicated consonantal roots, (b) lengthened vowels, and (c) special morphemes affixed to the verbal stem. Some verbs may undergo reduplication and vowel lengthening. These markers bear their aspectual value(s) regardless of the conjugation of the verb. Reduplication (gemination) Reduplication indicates a variety of aspectual values including intensive, iterative, durative, distributive, and gradual meanings. It comes in two kinds: partial and total. Partial reduplication occurs on the second consonant in the root and total reduplication bears on the first and second consonants in the root. The lexical root √ksr, for instance, is related to the punctual process of breaking. The simple form kasara means that the act of breaking occurred once and it

 Lazhar Zanned

expresses punctuality as in (2a). However, the augmented form kassara means that the act occurred many times and expresses iteration: the act of breaking is repeated until the object (cup) is reduced into pieces as in (2b); or, until the whole set of the object is exhausted as in (2c): (2) √ksr (‘to break’) a. kasar-a al-walad-u al-ka�s-a. break.punctual.pst-3m def-boy-nom def-glass.sg-acc ‘The boy broke the glass’ b. kassar-a al-walad-u al-ka�s-a. break.iterative.pst-3m def-boy.sg-nom def-glass.sg-acc ‘The boy broke the glass into pieces’ c. kassar-a al-walad-u al-ku�ūs-a. break.iterative.pst-3m def-boy.sg-nom def-cup.pl-acc ‘The boy broke all the cups’ Partial reduplication also expresses duration and continuity of the process. The reduplicated verb related to the root √drb (‘to exercise, to be skillful’) means that the process of training and exercising covers an extensive amount of time: (3) √drb (‘to exercise, to be skillful’) a. darib-a al-walad-u alā al-sibāħat-i. be skilled.pst-3m def-boy-nom on def-swimming-gen ‘The boy is skilled at swimming’ (The boy is skillful at swimming) b. darrab-a al-mudarrib-u al-walad-a. train.dur-pst-3m def-coach-nom def-boy-acc ‘The coach trained the boy (for a long time)’ c. ta-darrab-a al-walad-u alā al-sibāħat-i. refl-train.dur.pst-3 def-boy-nom on def-swimming-gen ‘The boy trained for swimming on his own for a long time’ Furthermore, partial reduplication expresses the gradualness of the process. It occurs in successive or continuous steps: the process √xrŠ (‘to go out’), for instance, is conceived of happening in a gradual fashion when reduplicated verbal forms are used: (4)

√xrŠ (‘to leave, to go out’) ta-xarraŠ-a al-tālib-u min jāmiat-i Manouba. refl-go out.stepped fashion.pst-3m def-student-nom from University of Manouba ‘The student graduated from the University of Manouba (after some years)’

Aspect as a scanning device 

Some processes, like √mwt ‘die’ and √qtl ‘kill’, usually conceived as a single whole that cannot be partitioned, are multiplied by partial reduplication. Therefore, reduplicated verbs derived from those roots are used in situations where the agents (experiencers) and/or the patients (victims) are multiple, to express iteration and cumulation. In order to be iterated, the process √mwt ‘die’ requires only multiple experiencers, as exemplified in (5a), but √qtl ‘kill’ requires a single agent and a multiple victim as in (5b), or both a multiple agent and a multiple victim as in (5c): (5) a. mawwat-at al-�ibil-u. die.cumulative.pst-3f def-camels-nom ‘All the camels died in the same period’ b. qattal-a Zayd-un al-�ibil-a. kill.cumulative.pst-3m zayd-nom def-camels-acc ‘Zayd killed all the camels’ (Zayd slaughtered the camels) c. qattal-at al-šurtat-u al-mutazāhir-īna. kill.cumulative.pst-3f Def-police-nom def-demonstrators-acc ‘The police slaughterd the demonstrators’ In addition, the total reduplication of the consonantal root, usually biliteral, expresses duration or iteration of the process, as in (6b): (6) √zl (‘to slip’) a. √zll (‘to slip (once)’) zall-a Zayd-un. slip.punctual.pst-3m Zayd-nom ‘Zayd slipped’ b.

√zlzl (‘to quake, to shake’) zal~zal-at al-�arđ-u. slip~slip.iterative.pst-3f def-earth-nom ‘The earth shook (for a long time)’

Vowel lengthening Along with reduplication, mostly partial, vowel lengthening is used in Arabic morphology to express duration in time or in space, thus changing a punctual process into a durative one as exemplified in (7). The notion related to √qtr (‘to drip’), for instance, is conceived as a punctual action when the simple verb form is used ((7a)), and as a continuous action that lasts for a long time when the second vowel of the template is lengthened ((7b)):

 Lazhar Zanned

(7) √qtr (‘to drip (+liquid)’) a. qatar-a al-mā�-u. drip.pst-3m def-water-nom ‘Water dripped’ b. ta-qātar-a al-mā�-u. refl-drip.durative.pst-3m def-water-nom ‘Water dripped for a long time’ Reduplication and vowel lengthening may occur in the same template to express intensity, mostly in the field of colors. The notion of ‘being red,’ or ‘getting red’ related to √ħmr, for instance, is conceived as a normal or neutral hue in (8a) and as an intensive one in (8b): (8) √ħmr (‘to (be, get) red’) a. �iħmarr-at ayn-u Zayd-in. get red.pst-3f eye-nom Zayd-gen ‘Zayd’s eye reddened’ b. �iħmārr-at ayn-u Zayd-in. get red.intensive.pst-3f eye-nom Zayd-gen ‘Zayd’s eye got very red’ Affixation Other special morphemes are affixed to the template to convey a variety of aspectual values: inchoation, intensity, duration, etc. For instance, the prefix (�a) means inchoation, generally in processes of entering a space, as in (9a), or being at the beginning of a period of time, as in (9b): (9) √bħr (‘sea’) a. �a-bħar-a Zayd-un. get.into-sea.pst-3m Zayd-nom ‘Zayd got into the sea’ (Zayd sailed) √sbħ (‘morning’) b. �a-sbaħ-a Zayd-un. get.into-morning.pst-3m Zayd-nom ‘Zayd got in the morning’ (Zayd started his morning) The infixed morpheme (w) expresses duration and intensity of the process, generally related to states (shapes), feelings and colors as exemplified in (10):

Aspect as a scanning device 

(10) √ħdb (‘to curve, to be curved’) a. �iħdawdab-a zahr-u Zayd-in. be.curved.intensive.pst-3m back-nom Zayd-gen ‘Zayd’s back got very curved’ √ħlw (‘to be sweet, sweetness’) b. �iħlawl-ā ayš-u Zayd-in. be.sweet.pst.intensive-3m life-nom Zayd-gen ‘Zayd’s life got very sweet for a long time (intensely sweet)’ √xdr (‘to (be, get) green’) c. �ixdawdar-a al-ušb-u. be.green.intensive.pst-3m def-grass-nom ‘The grass got very green’ (The grass was intensely green) Lexical markers In addition to the derivational markers of aspect (reduplication, affixation and vowel lengthening), Arabic has lexical markers. They are a closed set of grammaticalized verbs, usually used as inflected ‘auxiliaries’ and followed by an imperfective verb expressing the main process in the sentence. Therefore, the marking of aspect bears on the main process conceived as a whole or sliced into successive stages. Depending on the way the process is conceived, the array of aspectual values they convey varies: inchoation, restarting, duration or continuation, habituality, and completion or termination, as summarized in the following paragraphs. a. Nearness Verbs derived from the roots √kwd (‘to near’) and √wšk (‘to hurry, to near’) are used in Arabic to express aspectual nearness, i.e. the state of being near or on the point of implementing the process. Nearness does not necessarily require intention or awareness of the agent, as illustrated in (11a). The state of nearness may be captured at any stage of the unfolding process; for instance, (11b) conveys the ultimate phase before termination: (11) √kwd (‘to near’) a. kād-a Zayd-un ya-ksir-u al-ka�s-a. near.pst-3m Zayd-nom 3m-break.prs-ind def-cup-acc ‘Zayd almost broke the cup’ b. kād-a Zayd-un �an yu-nhiy-a al-qasīd-a. near.pst-3m Zayd-nom that 3m-finish.prs-sbjv def-poem-acc ‘Zayd almost finished (writing) the poem’

 Lazhar Zanned

b. Inchoation Inchoation is expressed in Arabic by verbs related to many roots with different meanings: √nš� (‘to create’), √bd� (‘to start, to begin’), √�xð (‘to take, to grasp’), √šr (‘to start, to open’), √tfq (‘to start’), √rwħ (‘to go away’) and √df (‘to push’). As regards the characterization of inchoation, all these verbs only indicate inchoation, except the last one √df (‘to push’), which indicates the rate at which the agent started the process, as exemplified in (12b): (12) √�xð (‘to take, to grasp’) a. �axað-a Zayd-un ya-ktub-u. take.pst-3m Zayd-nom 3m-write.pres-ind ‘Zayd started writing’ √df (‘to push’) b. �in-dafa-a Zayd-un ya-ktub-u. ref-push-pst-3m Zayd-nom 3m-write.pres-ind ‘Zayd rapidly started writing’ c. Duration Verbs indicating immobility are always used to express duration and continuation. The absence of motion is conceived in terms of maintaining the same posture, which causes the same state to last for a long time; the result is duration. Consequently, verbs related to the roots √byt (‘to stay during the night’), √zll (‘to stay, to stand still’) and √qd (‘to sit, to rest’) etc. are always used in the affirmative mode (i.e. with positive ‘polarity’). Some other verbs express a duration bounded in time: the night in (13b), for instance, is the location in time for the process of writing. However, durative aspect is conveyed by other verbs related to roots meaning separation and distantiation, such as √fkk (‘to separate, to dissociate’), √zwl (‘to vanish, to get away’), √ft� (‘to stop’) and √brħ (‘to leave (a place)’). All these verbs are negated (by the particles mā (‘not’) or lam (‘not’)), to express duration in the process as in (13c): (13) √zll (‘to stay, to stand still’) a. zall-a Zayd-un ya-ktub-u šir-an. stay.pst-3m Zayd-nom 3m-write.pres-ind poetry-acc ‘Zayd was writing poetry for a long time’ √byt (‘to stay during the night’) b. bāt-a Zayd-un ya-ktub-u šir-an. stay.night.pst-3m Zayd-nom 3m-write.pres-ind poetry-acc ‘Zayd spent the night writing poetry’

Aspect as a scanning device 

√zwl (to vanish, to get away) c. mā zāl-a Zayd-un ya-ktub-u šir-an. neg get.away.pst-3m Zayd-nom 3m-write.pres-ind poetry-acc ‘Zayd did not stop writing poetry’ d. Habituality Habituality is expressed by a verb derived from √kwn (‘to be, to exist’). It is always in the past form (kāna) followed by the main verb of the sentence in the imperfective; therefore it marks durative (progressive) aspect in the past (14a). However, some adverbs make the aspect of habituality more explicit by indicating the cyclical or continuous way the process occurred in time, as illustrated in (14b) and (14c), respectively: (14) √kwn (‘to be, to exist’) a. kān-a Zayd-un ya-Šrī. be.pst-3m Zayd-nom 3m-jog.pres-ind ‘Zayd was jogging’ b. kān-a Zayd-un ya-Šrī kull-a yawm-in. be.pst-3m Zayd-nom 3m-jog.pres-ind every-loc day-gen ‘Zayd used to jog every day’ c. kān-a Zayd-un ya-Šrī bākir-an. be.pst-3m Zayd-nom 3m-jog.pres-ind early-loc ‘Zayd used to jog early (in the morning)’ e. Termination Terminative aspect is expressed in Arabic by the use of full verbs derived from either the root √nhy (‘to end, to reach’), or √frγ (‘to empty’), where the process is conveyed by a nominal form in the completive position, as exemplified in (15): (15)

√frγ (‘to empty’) faraγ-a Zayd-un min kitāba-ti al qasīd-i. empty.pst-3m Zayd-nom from writing-gen def-poem-gen ‘Zayd finished writing the poem’

Furthermore, Arabic uses the same verb related to the root √wd (‘to come back’) to mark some related aspectual values such as restarting, termination, and habituality. The affirmative use of the verb expresses restarting aspect, which is a kind of (new) inchoation since it presupposes that the process was interrupted and then resumed, as illustrated in (16a). When the verb is negated, it indicates the termination of the process regardless of the frequency at which it occurs: it may be an activity that stopped suddenly or a habit that lasted for long, as in (16b). However,

 Lazhar Zanned

the use of the nominal form ādat (‘habit’) as an adverbial expresses habituality in expressions such as (16c): (16) √wd (‘to come back’) a. ād-a Zayd-un ya-ktub-u šir-an. come.back.pst-3m Zayd-nom 3m-write.prs-ind poetry-acc ‘Zayd is writing poetry again’ b. lam ya-ud Zayd-un ya-ktub-u šir-an. neg 3m-come.back.prs Zayd-nom 3m-write.prs-ind poetry-acc ‘Zayd no more writes poetry’ c. ya-ktub-u Zayd-un al-šir-a ādat-an. 3m-write.prs-ind Zayd-nom def-poetry-acc habit-loc ‘Zayd usually writes poetry (or habitually, regularly)’ The preceding presentation of the aspectual system of Arabic is offered as a basic overview, skipping many details. I now turn to analyzing this system with respect to imagery and scanning as treated in cognitive linguistics. 3. Aspect in imagery This section deals with the extent to which aspect might be considered as a verbal device to construct mental images related to states, events, and processes. The analysis will be based on attested evidence concerning mental imagery as presented in the literature on cognitive psychology. In turning to cognitive psychology, we follow the Cognitive Commitment principle (Lakoff 1990), which stipulates that “principles of linguistic structure should reflect what is known about human cognition from other disciplines, particularly the other cognitive sciences (philosophy, psychology, artificial intelligence and neuroscience).” (Evans & Green 2006: 40–41). 3.1

Aspect as an image-building tool

Imagery is the cognitive faculty that elaborates perceptual representations (Langacker 1987: 110, Wraga & Kosslyn 2003) into mental images. Mental images are the internal representations that produce the experience of perception while the appropriate sensory input is absent. Therefore, images (visual and mental) play a central role in human cognition. Normal language use involves the description of actual, virtual, or fictive entities (Langacker 1999). Events, processes, and states have a particular status in the

Aspect as a scanning device 

perception, the organization, and the memorization of the actual or fictive world, since they stand for the individual actions performed by the agents and the interactions between the agents: in brief, they profile the changing (dynamic) part of the world. Hence, change, by itself, is a constituent of the perceived or represented scene and may be dissociated from the other entities present in it. We can consider the mental representations of a dynamic scene to include aspectual representations that focus on the changing part(s) in it, i.e. the participants and their roles in the course of their interaction. This might explain why the verb is the most important component of the sentence and bears the markings of time, aspect, and modality. Language offers the symbolic resources to express actions conceived from a variety of vantage points. The mental images of actions are conveyed (or triggered) by the lexical items (“run”, “jump”, “sleep”...) and aspectual forms express the ways in which they occur (duration, phases, intensity, etc.) Therefore, aspect as the verbal coding of mental images related to processes and aspectual forms is one of many grammatical tools constituting the mental representations of states, events, and processes. The functions of mental imagery are multiple. They are used in predicting, creating, retrieving, navigating, tracking, and reaching objects or movements (Kosslyn 1995; Wraga & Kosslyn 2003). Imagery is the ability to predict the outcome of an action like planning something ahead. The notion of ‘making a chair’, for instance, is based on that ability. From a cognitive view, the maker is following a general plan (prototype) made up of many partial plans. Making the different parts of the chair, following precise steps, is guided by this plan at two levels: the partial plan elaborates the individual part; the general plan outlines its place and function in the whole representation of the chair. In aspectual terms, it implies telic aspect, because the verb “make” indicates the result of the process, which presupposes that the agent achieves all the stages mentioned earlier. Mental imagery is the ability to create a mental model in a variety of situations, such as problem solving, learning new tasks, or improving skills by repeating an action (Kosslyn 1995). In aspectual terms, this ability is exemplified by the verbal description of actual scenes (events, processes, states) or the elaboration of fictive ones planned ahead to give directions to perform an act or operation with reference to the internal organization structuring the whole. Furthermore, the analogy between visual and verbal representations is widely acknowledged (Denis, Gonçalves & Memmi 1995), and language may be considered a source of constructing mental images. Wraga & Kosslyn (2003) view mental imagery as relying more on visual perception to construct representations than verbal descriptions, since mental (visual) images include more details than verbal images do. The visual mental image of a ‘dog under the tree’, for instance, comprises the

 Lazhar Zanned

shape, size, orientation, and proportions of the observed objects, etc., while the sentence “the dog is under the tree” only specifies these two entities’ physical relation. Mental images based on verbal descriptions convey a minimal amount of information. This might be true for images of static situations like the one mentioned by Wraga & Kosslyn, but when it comes to dynamic situations in general and to processes in particular, verbal mental images are richer in specifications, since their aspectual representation indicates how the process unfolds in much more detail. The act of writing, for instance, is lexically represented, through the verb “write”, as hand movement(s) making graphic symbols on a material surface. However, other relevant details are conveyed by other linguistic elements expressing aspectual values, as illustrated in (17b–e): (17)

a. b. c. d. e.

I wrote the book. I am writing the book. I wrote the book line by line. I started writing the book. I finished writing the book.

As for depiction in mental imagery, Denis & Cocude (1997) argue that visual images incorporate the metric structure of represented objects and their configurations. Depiction is considered as the correspondence between the representation and the represented object: each portion of the representation is a representation of a portion of the object, such that the distances among portions of the representation correspond to the distances among the corresponding portions of the object. Depictive representations always stand for parts of the object, even when they are divided in an arbitrary way, just like the fragments of a picture cut out randomly still represent the corresponding portions of the represented object or scene (Kosslyn 1994; Kosslyn, Thompson & Ganis 2002). Moreover, not only do images based on verbal descriptions have the same structural properties as images derived from perception (Denis & Kosslyn 1999), the sequencing of a verbal description is also reflected in the internal structure of images (Denis, Gonçalves & Memmi 1995). Representations of scenes constructed from verbal descriptions tend to preserve the spatial relationships occurring in the scene (Tversky & Lee 1998). However, aspectual representation is depictive, too: it indicates the schematic temporal architecture of the process. The basic spatial relationships are related to the timing of the action as it unfolds: the notion of ‘jump’, for instance, is always indicated schematically, without indicating any precise details about the length or height of the jump, unless some other grammatical means are used to give more details about them (as in “jump high”, “jump higher and higher”, “jump backward” or “forward”). In some contexts, more information about portions of the action is available, in a schematic fashion, such as information about the start, the middle, or

Aspect as a scanning device 

the end sequences, as in “start jumping”, “be jumping”, and “stop jumping,” etc. Aspectual representations reflect the basic distances between the different phases that constitute the process as it unfolds in conceived or real time. They are conveyed by aspectual forms bearing notions such as repetition, frequency, habituality, etc. I argued in this section that aspect is a verbal tool for building mental images. It is an inherent part of the symbolic resources available in grammar to depict the dynamic part of the perceived (or represented) situations and to construct mental images based on verbal descriptions. Aspectual mental imagery – if we may call it that – is schematic and depictive and may be considered as the verbal facet of mental imagery. 3.2

Aspect as a scanning device

In the previous section, I argued that aspect might be considered as a verbal image-building tool. In this section, I try to show that building aspectual mental images of actions is based on scanning as a general cognitive mental ability, a step that leads to considering aspect as a verbal scanning device. In the end, I argue that all aspectual values related to events, processes, and states commonly discussed in the linguistic literature can be handled in terms of scanning, be it summary or sequential. In cognitive psychology, mental scanning is considered to be the basic mechanism of mental imagery. It involves the systematic shifting of attention over visualized objects. Kosslyn (1994) affirms that two mechanisms are at the base of mental scanning: shifting attention and transformation (rotation). Shifting attention occurs while moving from one portion of the mental image to another. A portion is selected by focusing attention on it. Rotating the visual mental image of an object preserves the motor properties of skills involved during the actual perception of this object. It relies on motor programs that ordinarily move the eyes, the head, and different parts of the body. Scanning or rotating a mental image occurs in the same way as scanning or rotating an actual object, since the cognitive system simulates actual behavior (Berthoz 1997). In Cognitive Grammar, scanning is a type of cognitive processing that occurs ubiquitously and automatically (Langacker 1987). It does not relate to attention, neither in its general sense nor in its specialized sense as involving shifts in focal area. Scanning is the operation relating a standard of comparison to a target and registering any discrepancies between them. Scanning and comparison are inherently interrelated cognitive processes, since they occur in all domains and at all levels of cognitive functioning (Langacker 1987: 108). Scanning is an event on its own, distinguishable from the representation of standard of comparison and the target. It is a complex directional event: complex

 Lazhar Zanned

because it consists of a chain of scanning operations, and directional because the chain defines a path through the field of representation through which the perceiver scans in a constant direction until a contrast is registered (Langacker 1987: 109). Three sets of scanning events are listed in Langacker (1987): (a) scanning events recognizing the standard as separate entity, (b) scanning events to recognize the target as a distinct entity, and (c) a comparison of the two entities thus recognized with their relative positions in the mental field. In the case of a falling object, set (a) would establish the top-most structure of the field as a standard, set (b) would present the lower structure as a target, and set (c) would compare the two positions of the same object in multiple phases. The perception and the conceptualization of any situation (event, process, state) are based on scanning to detect change or stativity through time. Mental representation follows the same process. The concept of ‘walking’, for instance, is characterized by changes in the location of the mover. The latter starts at a given location in space, which coincides with a certain point in time. This location is the standard, recognized as a distinct entity by a first set of scanning events. Then, as (s)he moves and changes location, every new location is considered as a target of comparison. The comparison stage then back and forth between two contingent locations, turning the perception and the conception of the movement to one complex image. Processing goes on until the mover reaches some (arbitrary) point and stops. Mental scanning is thus made up of countless continued acts of comparison (Langacker 1987: 108), and this fact makes for the complexity of the scanning event. In aspectual terms, scanning is based a comparison between different successive phases of the action as it unfolds, such that each phase is at the same time a standard and a target: a target for the previous and a standard for the following act of comparison. The starting point is a target for a standard that corresponds to the initial state (the absence of movement); the end is a target for the pre-final phase and a standard for the following phase (again, the absence of movement). The different successive phases making up the entire (uniform) movement make scanning chains. Given that an action unfolds through chained successive phases and that it is perceived and conceptualized through chained scanning operations, we may consider that aspectual values can be handled in terms of standards and targets. Nearness, or the state of preparing to start an action, for instance, is a target for a precedent state and a standard for the following one, which is the first phase of the process itself. Furthermore, nearness may coincide with any phase in the unfolding process, under one condition, which is the presence of a landmark somewhere in or outside the action: in (18a), nearness is captured just before the breaking of

Aspect as a scanning device 

the cup would have occurred. It is captured just before the terminative phase of writing the poem in (18b): (18) a. kād-a Zayd-un ya-ksir-u al-ka�s-a. near.pst-3m Zayd-nom 3m-break.prs-ind def-cup-acc ‘Zayd almost broke the cup’ b. kād-a Zayd-un yu-nhī al-qasīd-a. near.pst-3m Zayd-nom 3m-finish.prs.ind def-poem-acc ‘Zayd almost finished (writing) the poem’ Inchoative aspect is related to the initial phase recognized as a separate state different from the other state(s) occurring or imagined in the background. The initial state may be conceived as a single point in time as in (19a), where the starting of the running is an intermediate short state between resting or walking (the standard) and speeded movement (the target). It may be conceived as an intermittent state or intermittent states varying in length, as in (19b), where the agent is starting the process of writing poetry. This process does not necessarily take place in a single moment: (19c) presents writing a poem as an ongoing process. The process is renewable, manifested through trials and drafts, as is common for poets. In general, scanning in inchoatives is expansive, since it involves only the subsequent phases of the process:1 (19) a. bada�-a Zayd-un ya-Šrī. start-pst-3m Zayd-nom 3m-run.prs.ind ‘Zayd started running’ b. bada�-a Zayd-un ya-ktub-u šir-an. start-pst-3m Zayd-nom 3m-write.prs.ind poetry.indf-acc ‘Zayd started writing poetry’ c. bada�-a Zayd-un ya-ktub-u qasīd-an. start-pst-3m Zayd-nom 3m-write.prs.ind poem.indf-acc ‘Zayd started writing a poem’ Terminative aspect relates to the final phase recognized as a terminative state of the process. It is the target for the immediate pre-final phase on the one hand and for the whole process on the other. Finishing a poem, for instance, is a terminative phase of the whole process of writing, which may take a long time, and its final step, when the writer puts away his or her pen, as illustrated in (20). Scanning in

1. Langacker (1987: 225) uses “expansive” and “contractive” as two kinds of scanning operations in the case of ‘above’ and ‘below’. This applies in aspectual scanning too.

 Lazhar Zanned

terminative aspect is contractive when the target is the pre-final phase. However, it is expansive when the target includes the whole precedent process: (20) faraγ-a Zayd-un min kitābat-i al-qasīd-i. empty-pst-3m Zayd-nom from writing.f-gen def-poem.gen ‘Zayd finished writing the poem’ Durative aspect involves comparison between the phases of the process in a way that is necessarily bidirectional. To detect continuation, scanning goes back and forth, switching standards and targets and adding new ones as the process unfolds. Even though landmarks are moving and comparison operations are numerous, discrepancies are neglected in favor of similarity between different successive phases. This similarity makes us conceive the state as lasting and the process as enduring. Duration may be bounded and the process may be bounded, too, as in (21a), where the two boundaries (standard/target) are fixed and continuation occurs within those boundaries: writing poetry is bounded by the night between evening and dawn. Duration may also be unbounded and the process may be unbounded, too, as illustrated in (21b, c, d): (21) a. bāt-a Zayd-un ya-ktub-u ši r-an. stay.night-pst-3m Zayd-nom 3m-write.prs.ind poetry.indf-acc ‘Zayd spent the night writing poetry’ b. kān-a Zayd-un ya-ktub-u šir-an. be-pst-3m Zayd-nom 3m-write.prs.ind poetry.indf-acc ‘Zayd was writing poetry’ c. zall-a Zayd-un ya-ktub-u šir-an. stand.still-pst-3m Zayd-nom 3m-write.prs-ind poetry.indf-acc ‘Zayd kept writing poetry’ d. mā zāl-a Zayd-un ya-ktub-u šir-an. neg get.away-pst-3m Zayd-nom 3m-write.prs-ind poetry.indf-acc ‘Zayd did not stop writing poetry’ Restarting, as an aspectual value, relates to durative aspect and inchoation. It is a new inchoation in a durative process that stopped at a given time. It also relates to habituals in this respect, since they refer to an intermittent durative process. Writing poems, for instance, is a habit that occurs every day for a long time in (22a), and that is restarted in (22b). The comparison is built on the precedent durative process of writing as a standard and the new episode of the process or its renewal as a target. Thus, comparison in aspectual scanning can be based on episodes that last a long time:

Aspect as a scanning device 

(22) a.

kān-a Zayd-un ya-ktub-u šir-an kull-a be-pst-3m Zayd-nom 3m-write.prs.ind poetry.indf-acc every-loc yawm-in. day-gen ‘Zayd used to write poetry every day’

b.

ād-a Zayd-un ya-ktub-u šir-an come.back.pst-3m Zayd-nom 3m-write.prs-ind poetry-acc kull-a yawm-in. every-loc day-gen ‘Zayd is writing poetry every day again’

I have argued so far that aspectual representations might be dealt with in terms of scanning as a cognitive operation based on comparisons between successive phases of the action. In the following section, I try to show that aspectual representations include spatial and temporal dimensions related to actions and processes. In imagery, the mental scanning effect resides in a linear relationship between scanned distances and scanning times: the visual images incorporate the metric structure of represented objects or configurations (Denis & Cocude 1997). Kosslyn (1995: 283) argues that visual mental images contain references to space. As distance is an intrinsic part of space, it is part of the representation, and, thus, it affects the processing time: scanning times increase linearly with increasing distances scanned across. Borst, Kosslyn, and Denis (2006: 476) take the view that mental images incorporate the metric information present in the original stimulus. Aspectual representations are also based on distances in space and/or time. Distances in aspectual terms are schematic representations of real or virtual dimensions. Aspectual scanning depicts them in two ways: as a whole or as partitioned, as illustrated in (23), (24) and (25). In Arabic, verbal morphological forms with an inner long vowel convey continuation or duration in space and/or time, as a basic dimension, (see vowel lengthening (Section 2)). Precisely, distances in space and in time are expressed by the morphological template ((CV)C1aaC2aC3), where C1C2C3 corresponds to the consonantal root bearing the lexical meaning (the process) and the long vowel (ā) marks the extent or duration of the action in space and in time. In (23a), the event is rendered as a whole and the act of receiving the award is summarized, although it incorporates many different actions related to the ceremony and to the actual act of stretching out the hand to grasp the awarded object. In this case, distances are not considered relevant, so they are ignored. In (23b) and (23c), distance in space is a basic component of the representation: in (b) Syrine hands Salmā the cup by taking it in her hand and stretching her arm to reach Salmā, covering all of the distance separating them. We may imagine that

 Lazhar Zanned

the distance is longer when, for example, one of them is on the ground and the other is in a tree. In (c), Syrine does the same thing as is done in (b), until she reaches for the cup: (23) √nwl (‘to acquire, to get’) a. nāl-at Syrine jā�izat-an. get.pst-3f Syrine prize.indf-acc ‘Syrine got a prize’ b. nāwal-at Syrine Salmā al-ka�s-a. hand.pst-3f Syrine Salmā def-cup-acc ‘Syrine handed the cup to Salmā’ c. ta-nāwal-at Syrine al-ka�s-a. refl-grasp.pst-3f Syrine def-cup-acc ‘Syrine grasped the cup – by stretching her arm to reach it’ Likewise, aspectual scanning may depict distance in time. The process √qtr (‘to drip (+liquid)’), for instance, is scanned in two different ways in (24) and has two aspectual construals: in (24a), it is represented as a punctual event that does not last because it is rendered as a completed whole. However, it is construed in (24b) as a continuous action in time. The meaning of both expressions is referential, but the difference bears on the mental scanning translated into aspectual marking: (24) √qtr (‘to drip (+liquid)’) a. qatar-a al-mā�-u. drip.pst-3m def-water-nom ‘Water dripped’ b. ta-qātar-a al-mā�-u. refl-drip.durative.pst-3m def-water-nom ‘Water dripped for a long time’ The duration of a process may be conceived as successive cycles, as illustrated in (25). The transition of possession, as in inheritance for instance, is expressed in two ways: a single cycle or transition as in (25a), where the agent inherits the object from his immediate ancestor(s), or a chain of successive transitions between generations, where each generation is at the same time an ancestor and a descendant, as in (25b): (25) √wrθ (‘to inherit’) a. wariθ-a Zayd-un al-qasr-a. inherit.pst-3m Zayd-nom def-palace-acc ‘Zayd inherited the palace’

Aspect as a scanning device 

b. ta-wāraθa-t �usrat-u Zayd-in al-qasr-a. refl-inherit.durative.pst-3m family-nom Zayd-gen def-palace-acc ‘Zayd’s family inherited the palace, generation after generation’ Furthermore, aspectual scanning implies directionality in the represented process. According to Langacker (1987: 224), scanning involves transition, which makes chains. The chaining may be linear as illustrated in (23) and (25), where linearity is spatial and temporal, respectively. However, it may also be conceived as a multitude of divergent lines starting at a real or virtual center. The event is then a multilinear process occurring in all directions simultaneously, where the trajector (agent) is multiple (plural, generic etc.), as shown in (26): (26) √tyr (to fly) a. tāra-t al-farāšat-u. fly.pst-3f def- butterfly.f.sg-nom ‘The butterfly flew away’ b. ta-tāyar-a al-farāš-u. refl-fly.pst-3m def-butterfly.pl-nom ‘The butterflies flew away in all directions, simultaneously and sporadically’ In (26a), the trajector is singular and the flying is linear. Scanned in a summary fashion, the flying is reported as a whole event and there is no indication of the way in which it happened. In contrast, the trajector is multiple in (26b), and the template (ta-C1aaC2aC3) expresses the way the process is performed: a multitude of butterflies are flying simultaneously, sporadically and in all directions, starting from a central point where they were gathered. The scanning involves many acts of comparison between different butterflies at two levels: the first is the level of the individual whose movement is scanned, making an individual linear chain following the trajector’s different locations in space. The second is the level of the mass, where many acts of comparison between the individuals occur to register the simultaneity and the variety of flying directions and to elaborate the scene as a homogenous coherent percept. Linearity in space also involves static configurations. In (27), the same event of spreading is construed in two ways: in (27a), the scene is dynamic because it involves moving objects (‘grains’) thrown by an agent (Zayd) and distances are not in focus. The scene in (27b) is static, since there is no movement. Comparison is based on a fictive movement starting at the center of a circle and reaching the periphery in a kind of network established through scanning. Two crossing views indicate distances. The first one involves fictive motion by starting from the center and going in a radial fashion until the individual location of each grain is covered. The other is

 Lazhar Zanned

tracking the location of each grain circularly. Both views hold simultaneously and need to be mutually superimposed in order to constitute a coherent scene: (27) √nθr (‘to spread’) a. naθar-a Zayd-un al-ħabb-a. spread.pst-3m Zayd-nom def-grain-acc ‘Zayd spread the grains’ b. ta-nāθar-a al-ħabb-u. refl-spread.pst-3m def-grain-acc ‘The grains spread all the way around’ I have argued in this section that aspect may be considered as a verbal scanning device: aspectual scanning is the basic mechanism in building linguistic representations of processes. It is based on comparisons between successive phases of the action as it unfolds. Inchoative aspect focuses on the initial phase(s); terminative aspect relates to the final phase(s); durative aspect involves a bidirectional comparison between all the phases of the process. Aspectual scanning includes spatial and temporal configurations and thus linearity and directionality in the represented (dynamic or static) situation. In section (4), I now deal with modes of aspectual scanning, based on data from Arabic. 4. Aspectual scanning: The case of Arabic Comrie (1976: 13) proposes a difference between states, events, and processes. States are static, they continue until changed. Events and processes are dynamic and require continuous input of energy. They differ in the way they are viewed: while events are viewed as a complete whole (perfective aspect), processes are viewed in progress, from within (imperfective aspect). In Cognitive Grammar (Langacker 1987: 244), the conceptualization of a process follows its temporal evolution. This evolution is made up of a series of successive states standing for the different phases of the process and is conceived as occupying a continuous series of points in conceived time, where each state consists of a relation between the trajector and the landmark. Langacker takes the view that the mental flexibility that humans have to dissociate processing time from the conception of objective time should be the basis for analyzing linguistic structure in general and aspect in particular. In the light of this dichotomy, the verb profiles a process, which is a relationship, viewed with respect to its evolution through conceived time, while adjectives, adverbs, and prepositions profile a nonprocessual relationship; they are atemporal since evolution through time is not in focus. Two kinds of processes of verbs are thus established: perfectives and

Aspect as a scanning device 

imperfectives. The perfectives involve a change through time; imperfectives do not (Langacker 1987: 254). A perfective verb profiles a process bounded within the immediate temporal scope, while an imperfective verb profiles a constant situation persisting indefinitely through time (Langacker 2001: 257). Imperfective processes are conceived as a mass made up of parts of the same event, so any portion may be an instance of the process type, and any portion that coincides with the speech event will qualify as a representative instance. Imperfective processes are nonreplicable while perfectives are replicable (Langacker 1987: 260). However, this is not a rigid dichotomy partitioning the class of verbs into disjoint subsets. Some imperfective verbs (as in (28a)) can receive a perfective, dynamic construal in specific contexts, as in (28b) (Langacker 1987: 257): (28) a. This road winds through the mountains. (Imperfective) b. This road is winding through the mountains. (Perfective) The scope of predication determines the aspect of the verb by either encompassing the entire scene simultaneously or only successively through time. 4.1

Scanning modes

Aspect is one of many grammatical tools for epistemic grounding since it indicates the way a profiled process is related to time, construed, and scanned. In CG, scanning has two modes: summary and sequential. A complex scene may be processed in either mode; the scene of “getting in a room through the door,” for instance, may be profiled in a summary or sequential fashion (Langacker 1987: 248). When the component states are profiled individually as relations, then sequential scanning is employed and the verb ‘enter’ is used. When the components are profiled collectively as one thing, then summary scanning is employed and the noun ‘entrance’ is used. The difference between the verb (“enter”) and the noun (“entrance”) thus reflects the difference between two ways a process is scanned. In this view, the basic opposition between summary and sequential scanning is related to the opposition of the verb (infinitive, conjugated) and the noun (action) that share the same lexical root. In summary scanning, the situation is conceived as stable, but in sequential scanning, it is conceived as changing, with one single state captured at a time (Langacker 1987: 264). On the one hand, summary scanning is additive in that all aspects are coexistent and simultaneously available (Langacker 1987: 144–145, 248). On the other hand, sequential scanning involves disparity related to the recognition of change and implies the transformation of a configuration into another or a continuous series of transformations. Sequential scanning profiles a coherent unfolding scene divided into phases where each configuration serves as the standard for an act of comparison and where each configuration is successively

 Lazhar Zanned

recognized as different from the other. Summary scanning is the ability we display when examining a still photograph and sequential scanning is what we do in watching a motion picture (Langacker 1987: 248). In CG, the notion of change is pivotal for cognitive processing. It is the source of the difference between summary scanning (zero change) and sequential scanning (presence of change). In aspectual terms, it is the source of the difference between perfectives (presence of change) and imperfectives (zero change). We may establish a correspondence between the pairs in a one-to-one fashion: summary scanning/imperfectives, on the one hand, and sequential scanning/perfectives, on the other. This fact may suggest that imperfectives should be defined as processes scanned in a summary fashion and perfectives as processes scanned sequentially. In processes scanned in a summary fashion, the trajector (agent, subject, mover...) is occupying all the positions in its trajectory simultaneously (imperfective) and the view is global and fixed. However, in those processes scanned sequentially, the trajector is at a determined position (location in space and/or in time etc.) and the view is partial and cut into slices. Slicing occurs in two ways. The first applies when a single sequence of the process is profiled. I will call this the monosequential scanning. The second applies when all the sequences or some of them are profiled. I will call this polysequential scanning. We have the ability to capture (one of) the internal sequences of a process. This fact may be unnoticed in everyday use, but we may imagine the following scene. A detective and his colleague are screening a tape of a scene recorded in a shopping center: someone is stealing money from a drawer. He has many ways to comment it: a. Using summary scanning and focusing on the sequence of taking money would give something like “Now, he takes the money.” b. Using sequential scanning, he has the ability to follow the successive phases of the operation as if he were shooting each one of them, but instead of using a camera, he uses verbal tools. This would give “Now, he opens the drawer with his right hand, and now, he introduces the left one, grasps the money, takes it out, closes the drawer, ...”. Moreover, he can screen every sequence as it starts, unfolds, and ends. Opening the drawer, for instance, would be described in the following terms: “Now, he reaches out the right hand, grasps the edge of the drawer, starts pulling it carefully, pulls it, pulls it... now, the drawer opens...”. The last step he may reach is the ultimate component of the different processes involved in the sequence: the atomic elements of the process are suitable for monosequential scanning. In this respect, Chang, Gildea, & Narayanan (1998) established some schematic sequences of the processes, based on ordered phases: ready-start-ongoing-iteratefinish-done. Some other properties are related to other parameters, including the

Aspect as a scanning device 

rate (slow-fast), continuity (stop-restart-suspend-cancel), etc. This may help us assess the link between scanning and aspectual representations in general and the link between some standard aspectual values and monosequential or polysequential scanning in particular. Tversky (1999) established an analogy between drawings and language since they both include a small number of segments (words) that combine to form an infinite number of drawings or sentences. We may posit that aspectual constructions are made of a limited number of schematic forms (aspectual schemas) that combine to form an infinite number of representations of processes: these forms are the grammatical elements expressing all the aspectual values. 4.1.1 Aspectual summary scanning Considering that a process is bounded, we have the faculty to scan it in a summary fashion and construe it as a whole including its beginning, evolving, and ending. This corresponds to perfectives, in Langacker’s conception. Summary scanning applies to all types of processes regardless their semantic properties (punctual, duration etc.): (29) Action State Experience

kataba He wrote �alima He suffered marida He was sick nāma He slept

yaktubu writes ya�lamu suffers yamradu is sick yanāmu sleeps

sa-yaktubu will write sa-ya�lamu will suffer sa-yamradu will be sick sa-yanāmu will sleep

4.1.2 Aspectual monosequential scanning Processes are of two types: punctual and non-punctual. Punctual processes do not take a long time to occur. They are conceived, depicted, and scanned as a single event that cannot be divided into parts. They are non-countable, unless repeated: √qtr (‘drip (liquid)’), √ksr (‘break’), √qt (‘cut’), √qsm (‘divide’) etc. Other processes behave like punctuals, but they may have some inherent duration, like √mwt (‘die’), √qtl (‘kill’), √sqt (‘fall’) etc. Punctual processes are the best candidates for monosequential scanning due to their nature. However, we have the ability to scan any process in a single mood to capture just one of its sequences, as illustrated in (30): (30) a. kād-a Zayd-un ya-squt-u. near.pst-3m Zayd-nom 3m-fall.prs-ind ‘Zayd almost fell’

 Lazhar Zanned

b. �aħsad-a al-zar-u. harvest.ready.pst-3m def-crop-nom ‘The crops are ready for harvest’ c. �axað-a Zayd-un ya-ktub-u. take.pst-3m Zayd-nom 3m-write.prs-ind ‘Zayd started writing’ d. �abħar-a Zayd-un. get.into-sea.pst-3m Zayd-nom ‘Zayd got into the sea’ (Zayd sailed) e. �atfal-at Zaynab-u. child.get.pst-3f Zaynab-nom ‘Zaynab gave birth to a child’ 4.1.3 Aspectual polysequential scanning We may scan a process in a polysequential fashion to represent some of its fragments or its repetition or continuation as a whole. An ongoing process is scanned sequentially and the chaining of the sequences makes it polysequential. Progressive forms are the best representatives of this mode of scanning, as illustrated in (31): (31) a.

kān-a Zayd-un ya-rfa-u al-alam-a lammā be.pst-3m Zayd-nom 3-hoist.prst-ind def-flag-acc when daxal-tu. enter.pst-1 ‘Zayd was hoisting the flag when I entered’

b. �ar-ā Zayd-an ya-rfa-u al-alam-a. 1-see.prst Zayd-acc 3-hoist.prst-ind def-flag-acc ‘I see Zayd hoisting the flag’ Arabic shows a frequent use of the name of the agent to profile a continuous process related to the time of speaking or to a constant state. Being a noun derived from the verb, the name of the agent oscillates between the verb and the adjective. Thus, it may have the valence of the verb when it expresses a continuous process; otherwise, it is an adjective. Some names of the agent are proper nouns. The name of the agent kātib (‘writer’), for instance, profiles a perfective process, which is already finished as in (32a), or an imperfective process, which occurs or will occur as in (32b), or it may designate a political position with no act of writing and no aspectual value as in (32c): (32) a. Zayd-un kātib-u al-risālat-i. Zayd-nom write.agentive-nom def-letter-gen ‘Zayd is the one who wrote the letter (the writer of the letter)’

Aspect as a scanning device 

b. Zayd-un kātib-u-n al-risālat-a. Zayd-nom write.agentive-nom-indf def-letter-acc ‘Zayd (is) will(ing to) write the poem’ c. Zayd-un kātib-u al-dawlat-i. Zayd-nom write.agentive-nom def-state-gen ‘Zayd is Secretary of State’ The progressive verb and the name of the agent may profile a continuous process. For instance, the process is scanned in a polysequential mode by the verb yarfau in (31) and by the name of the agent rāfi in (33): (33) a.

kān-a Zayd-un rāfi-an al-alam-a lammā be.pst-3m Zayd-nom hoist.agentive-acc def-flag-acc when daxal-tu. enter.pst-1 ‘Zayd was hoisting the flag when I entered’

b. �a-rā Zayd-an rāfi-an al-alam-a. 1-see.prst Zayd-acc hoist.agentive-acc def-flag-acc ‘I see Zayd hoisting the flag’ Polysequential scanning profiles multiple phases in the process itself and/or multiple participants. Multiple phases correspond to some aspectual values, such as iteration, repetition, duration, cumulation, habituals, and partitioning. They constitute a variety of polysequential scanning, making a continuum of interconnected grades on a scale. Multiple participants correspond to either the plurality of the agent or the plurality of the patient. Adverbs can also express the multiplicity of the process or participants. Iteration relates to the repetition or frequency of the process without focusing on the engagement of the agent to repeat it. The second consonantal root reduplication indicates iteration, as mentioned in Section 2: (34) a. qattal-a Zayd-un al-γuzāt-a. kill.iteration.pst-3m Zayd-nom def-invader.pl-acc ‘Zayd killed all the invaders, one by one’ b. nawwam-a al-�atfāl-u bākir-an. sleep.iteration.pst-3m def-boy.pl-nom early-loc ‘All the boys went to bed early, one after the other’ Duration is expressed by the long vowel in the verbal stem. The process is conceived as lasting for a while without necessarily being continuous, as shown in (35):

 Lazhar Zanned

(35) a. zāwal-a Zayd-un dirāsat-a-hu fī Manouba. follow.durative.pst-3m Zayd-nom study-acc-his in Manouba ‘Zayd studied at Manouba for a long time’ b. ta-qātar-a al-mā�-u. refl-drip.durative.pst-3m def-water-nom ‘Water was dripping for a long time’ Reciprocal processes based on alternation between participants are profiled as continuous in polysequential scanning. The processes in (36), for instance, occur necessarily by alternation between two people and last for a certain time. Nevertheless, they are profiled in polysequential scanning as occurring in a continuous way: (36) a. qātal-a Zayd-un al-γuzāt-a. fight.durative.pst-3m Zayd-nom def-invaders-acc ‘Zayd fought the invaders for a long time’ b. ta-nāqaš-at Syrine wa Salmā. recp-argue.durative.pst-3f Syrine and Salmā ‘Syrine and Salmā argued for a long time’ Cumulation is a staged operation, in which scanning follows consecutive sequences until it reaches the ultimate stage, that of saturation. Verbal stems expressing cumulation are attested in the description of colors as in (37a, b), and in a few other domains, as in (37c, d). The most intense hue of a color, for instance, is reached after a series of layering operations where the following one adds to the preceding one, making the color more and more intense: (37) a. �iħmārr-a wajh-u Zayd-in. get.red.cumulative.pst-3m face-nom Zayd-gen ‘Zayd’s face got intensely red’ b. �ixdawdar-a al-ušb-u. get.green.cumulative.pst-3m def-grass-nom ‘The grass got intensely green’ c. �iħdawdab-a Zayd-un. get.curved.cumulative.pst-3m Zayd-nom ‘Zayd got his back very curved’ d. �išawšab-a al-ħaql-u. get.grass.cumulative.pst-3m def-field-nom ‘The field got a lot of grass’ Grammaticalized verbs indicating the length or the iteration of the process conveyed by the main verb express habituals and continuation:

Aspect as a scanning device 

(38) a.

kān-a Zayd-un ya-ktub-u al-šir-a fī be.pst-3m Zayd-nom 3m-write.prst-ind def-poetry-acc in sibā-hu. youth-he ‘Zayd used to write poetry in his youth’

b.

tāl-a mā katab-a Zayd-un al-šir-a fī get.long.pst-3m that write.pst-3m Zayd-nom def-poetry-acc in sibā-hu. youth-he ‘Zayd used to write poetry for long time in his youth’

c.

mā fati�-a Zayd-un ya-ktub-u al-šir-a neg stop.pst-3m Zayd-nom 3m-write.prst-ind def-poetry-acc fī sibā-hu. in youth-he ‘Zayd did not stop writing poetry in his youth’

Adverbs are one of the most common tools of aspectual representation in languages. Some adverbs in Arabic indicate polysequential scanning in general and the partitioning of the process in particular. The process of writing, for instance, is scanned in a summary fashion in (39a), but its sequences are profiled in (39b), where they coincide with the items of writing or with those of reading such as kalimat (‘word’) in (39b) or str (‘line’) in (39c): (39) a. katab-a Zayd-un al-kitāb-a. write.pst-3m Zayd-nom def-book-acc ‘Zayd wrote the book’ b.

katab-a Zayd-un al-kitāb-a kalimat-an write.pst-3m Zayd-nom def-book-acc word.indf-acc kalimat-an. word.indf-acc ‘Zayd wrote the book word by word’

c. katab-a Zayd-un al-kitāb-a satr-an satr-an. write.pst-3m Zayd-nom def-book-acc line.indf-acc line.indf-acc ‘Zayd wrote the book line by line’ To conclude, two kinds of aspectual scanning were established: summary scanning and sequential scanning. The former corresponds to imperfectives and the latter corresponds to perfectives. In aspectual summary scanning, the process is depicted in a summary fashion where the view is global and involves no change. However, when the process is scanned sequentially, the view is partial and cut into

 Lazhar Zanned

slices. Slicing occurs in two ways. The first one is monosequential scanning, in which a single sequence of the process is profiled. This corresponds to aspectual values such as nearness, inchoation, termination, and stop. The second one is polysequential scanning, in which multiple sequences of the process itself are rendered and/or multiple participants are described. This corresponds to the aspectual values usually related to progressive forms, such as iteration, habituals, continuation, and cumulation. 4.2

Scope of awareness in aspectual representation

Aspectual scanning involves the representation of scope. In CG, an expression’s maximal scope is the full array of conceptual content it evokes as the basis of its meaning. Its immediate scope is the limited portion directly relevant for a particular purpose, and its profile is the specific focus of attention within the immediate scope. In aspectual terms, the maximal scope is the whole scene of the process. Summary scanning, where the focus of attention coincides with the entire process including all its parts, captures this, as shown in (26), where the immediate scope and the maximal scope coincide. The immediate scope is the portion of the process made salient by the aspectual marker in monosequential and polysequential scanning. As salience relates to size, centrality, or importance (Tversky 1999: 97), the inchoative form, for instance, profiles the initial phase(s) of the process and its starting point in time. As it is captured, it constitutes the focus of attention, and all the other phases of the process are in the background. The aspectual values sketched in (26) render different portions of the process, where each one is focalized regardless of its position in the sequentiality of the whole process. Polysequential scanning is of an intermediate step between summary scanning and monosequential scanning. The aspectual values dealt with in 4.1.3, namely ongoing, iteration, duration, cumulation, habituals, continuation, and partitioning, represent a scope of awareness maximally encompassing the single process as a maximal scope. Iteration, for instance, is made up of the replication of any replicable process. However, the replicates are considered as identical, thus making up a single entity. The maximal scope will shift from the single process to a more general level, which includes the replicates of that process. The enlargement of the scope, creating a maximal scope where the immediate scope is the entire process, accompanies this shift: the verb qattala in (34a) expresses the iteration of killing. We cannot conceive of the process as occurring instantly for all the victims, so we have to conceive of it as repeated continuously or occurring in bounded episodes (Langacker 2001: 258) as long as there remain some living creatures

Aspect as a scanning device 

among the γuzāt (‘invaders’). Verbs profiling processes construed as occurring in bounded episodes are the best candidates for polysequential scanning. In polysequential scanning, the aspectual value itself is the immediate scope, since it is the basic value profiled by the verbal form, and the maximal scope is the larger scene or historical context in the common background of speakers and hearers. 4.3

Aspect as a zooming device

Zooming, in mental imagery, is the process of focusing attention on an imaged object to see finer details or particular features (Kosslyn 1980, 1995; Wraga & Kosslyn 2003). Langacker (2001: 259) considers the progressive as a kind of ‘zooming in’ and taking an internal view of a bounded event. Zooming has two directions in and out. The first relates to the registration of details, increasing image accuracy (Tversky & Lee 1998), whereas the second relates to the general view, in which details are ignored. Between the two extremes, there is some graded scale. In aspectual terms, summary scanning is a standard for zooming since the process is captured as a whole. It is the base for two kinds of zooming operation: a zooming in which focuses on a limited portion, and a zooming out which includes more than one portion. Monosequential scanning is a zooming in device, whereby we capture, select and single out a single phase of a process. The phase may be a portion or an interval in the process: nearness, starting, restarting, ending or any other phase in between. Polysequential scanning is a zooming out device. The aspectual values related to it treat the process as a unit (bounded or unbounded) and make it multiple to express duration, replication, cumulation, and partitioning. Moreover, inherent semantic properties of the process determine the mode of aspectual scanning for which the verb is a candidate and the type of zooming. In theory, we may scan all types of processes in a summary fashion. However, some processes are more suitable for monosequential than for polysequential scanning and vice-versa. In some cases, the conflict between the semantic properties of the verb and the aspectual scanning mode makes the utterance ungrammatical, as shown in (40): (40) √nwm (to sleep) a. nāma Zaydun. ‘Zayd slept’ b. *bada�a Zaydun yanāmu.’ ‘*Zayd started sleeping’ c. bada�a Zaydun yanāmu bākiran. ‘Zayd started to sleep early’

 Lazhar Zanned

The process √nwm (‘sleep’) cannot be partitioned; it is an obligatory candidate for summary scanning, as in (40a). The element of inchoation bada�a (‘started’) in (40b) makes the sentence ungrammatical because of the conflict between the process of sleeping, which cannot be partitioned, and the aspectual value of inchoation. However, the adverb bākiran (‘early’) in (40c) solves the conflict by introducing another temporal dimension without altering the process itself: the starting relates to the new habit Zayd is developing, namely to sleep early. The agent does not start to sleep. He starts a new habit related to sleeping. The conflict may be solved also by other grammatical means of polysequential scanning. Punctual processes cannot be conceived as lasting unless replicated, as shown in (41): (41) √qfz (‘to jump’) a. qafaza Zaydun alā al-ħā�iti. ‘Zayd jumped over the wall’ b. qafaza Zaydun āliyan. ‘Zayd jumped high’ c. qafaza Zaydun baīdan. ‘Zayd jumped away’ d. qafaza Zaydun qafzatan. ‘Zayd jumped once’ e. qafaza Zaydun qafzatayni. ‘Zayd jumped twice’ f. qafaza Zaydun kaθīran. ‘Zayd jumped a lot’ g. qafaza Zaydun tawīlan. ‘Zayd jumped for a long time’ h. bada�a Zaydun yaqfizu. ‘Zayd started jumping’ i. rāħa Zaydun yaqfizu. ‘Zayd continued to jump’ The process √qfz (‘jump’) involves a vertical movement starting from the ground. The trajector leaves the ground, rises and reaches a certain height, and then falls back to the ground, not necessarily at the same location he left. We have the cognitive ability to grasp any portion of the process as we do in slow-motion mode. In aspectual terms, we cannot consider the sequence of ‘leaving the ground’ as inchoation, or ‘the landing’ as termination, and so forth. As a punctual process, √qfz (‘jump’) is a candidate for summary scanning, as shown in (41a, b, c, d). The jumping process is not suitable for monosequential scanning, so it is not compatible with inchoation, termination, or duration. When the process is

Aspect as a scanning device 

replicated, which is the case in current language use; it becomes suitable for both monosequential and polysequential scanning. This is possible thanks to the use of verbs of inchoation (41h) and continuation (41i) or to the use of adverbs indicating plurality (41f, e) or duration (41g), as long as the process is repeated. Furthermore, the use of numerals as an aspectual marker makes a clear cut between two modes of scanning: summary and polysequential. In this case, polysequential scanning is indicated in Arabic by numerals (marratan (‘once’), marratayni (‘twice’), marrātin (‘many times’), xamsa marrātin (‘five times’)... n times) or by the noun of the action (qafzatan (‘one jump’), qafzatayni (‘two jumps’), qafazātin (‘many jumps’)). When the number indicated is one, then the process is scanned in a summary fashion. Otherwise, the scanning is polysequential. 5. Conclusion As “drawings reveal people’s conceptions of things, not their perceptions of things” (Tversky 1999: 94), aspectual forms reveal people’s conceptions of processes, not their perceptions of processes. I have argued that aspect in language should be considered as a verbal scanning device used in profiling processes. Using data from Arabic, I have shown that aspect profiles processes either as a whole (bounded or unbounded) in summary scanning, or as sliced into (bounded) sequences in sequential scanning. Sequentiality bears on a single phase (part or slice) of the process in monosequential scanning to convey nearness, inchoation, termination, etc. It also bears on multiple successive phases in polysequential scanning to convey duration, iteration, habituality, cumulation, etc. In both cases, aspect functions as a verbal zooming device to determine the scope of attention in predication, framed in between a large scope of the process in summary scanning and a limited portion of this process in sequential scanning. This chapter is a preliminary attempt to establish a link between insights from cognitive psychology (Mental Imagery) and cognitive linguistics (Cognitive Grammar) regarding the use of aspect in language. A very important link is the cognitive ability of scanning. This chapter contributes to enlarging the scope of study in cognitive framework(s) to less-investigated languages such as Arabic. References Berthoz, Alain. 1997. Le sens du mouvement. Paris: Odile Jacob. Broccias, Cristiano & Willem B. Hollmann. 2007. Do we need summary and sequential scanning in (Cognitive) Grammar? Cognitive Linguistics 18 (4): 487–522.

 Lazhar Zanned Borst, Grégoire, Stephen M. Kosslyn, & Michel Denis. 2006. Different cognitive processes in two image-scanning paradigms. Memory and Cognition 34 (3): 475–490. Chang, Nancy, Daniel Gildea, & Srini Narayanan. 1998. A dynamic model of aspectual composition. In Proceedings of the Twentieth Annual Meeting of the Cognitive Science Society COGSCI-98. Madison. (http://www.icsi.berkeley.edu/NTL/papers/asp.pdf). Cohen, David. 1989. L’aspect verbal. Paris: Presses Universitaires de France. Comrie, Bernard. 1976: Aspect, An introduction to the study of verbal aspect and related problems. Cambridge: Cambridge University Press. Denis, Michel & Marguerite Cocude. 1997. On the metrical properties of visual images generated from verbal descriptions: Evidence for the robustness of the mental scanning effect. European Journal of Cognitive Psychology 9: 353–379. Denis, Michel & Stephen M. Kosslyn. 1999. Scanning visual mental images: A window on the mind. Current Psychology of Cognition 18: 409–465. Denis, Michel, Marie-Rose Gonçalves & Daniel Memmi. 1995. Mental scanning of visual images generated from verbal descriptions: Towards a model of image accuracy. Neuropsychologia 33: 1511–1530. Evans, Vyvyan & Melanie Green. 2006. Cognitive Linguistics: An Introduction. Edinburgh: Edinburgh University Press. Fleisch, Henri. 1979. Traité de philologie arabe, vol.2, Beyrouth: Dar El Machriq Editeurs. Kosslyn, Stephen M. 1980. Image and Mind. Cambridge, MA: MIT Press. Kosslyn, Stephen M. 1994. Image and Brain. Cambridge, MA: MIT Press. Kosslyn, Stephen M. 1995. Mental Imagery. In Visual Cognition. An Invitation To Cognitive Science, vol.2, S. M. Kosslyn & D. N. Osherson (eds.), 267–296. Cambridge, MA: MIT Press. Kosslyn, Stephen M. 1999. Visual Mental Images as re-presentations of the world: A cognitive neuroscience approach. In Visual and Spatial Reasoning in Design, J. S. Gero & B. Tversky (eds.), 83–92. Sydney: Key Centre of Design Computing and Cognition, University of Sydney. Kosslyn, Stephen M. 2002. Einstein’s Mental Images: The Role of Visual, Spatial, and Motoric Representations. In The Languages of the Brain, A. M. Galaburda, S. M. Kosslyn, & Y. Christen (eds.), 271–287. Cambridge, MA: Harvard University Press. Kosslyn, Stephen M., William L. Thompson, & Giorgio Ganis. 2002. Mental imagery doesn’t work like that. Behavioral and Brain Sciences 20: 198–200. Lakoff, George. 1990. The invariance hypothesis: Is abstract reason based on image schemas? Cognitive Linguistics 1 (1): 39–74. Langacker, Ronald W. 1987. Foundations of Cognitive Grammar, Volume 1: Theoretical Prerequisites. Stanford, CA: Stanford University Press. Langacker, Ronald W. 1999. Virtual reality. Studies in the Linguistic Sciences 29 (2): 77–103. Langacker, Ronald W. 2001. The English present tense, English Language and Linguistics, 5 (2): 251–272. Langacker, Ronald W. 2008. Sequential and summary scanning: A reply. Cognitive Linguistics 19 (4): 571–584. Langacker, Ronald W. 2009. The English present: Temporal coincidence vs. epistemic immediacy, Investigations in Cognitive Grammar, 185–218. Berlin & New York: Mouton de Gruyter. Langacker, Ronald W. 2010. Reflections on the functional characterization of spatial prepositions, (http://corela.edel.univ-poitiers.fr/index.php?id=999#descriptionauteurs) McCarthy, John. 1979. Formal Problems in Semitic Phonology and Morphology. PhD Dissertation, MIT, United States.

Aspect as a scanning device  McCarthy, John. 1981. A Prosodic Theory of Nonconcatenative Morphology. Linguistic Inquiry 12: 373–418. Tversky, Barbara. 1999. What does drawing reveal about thinking? In Visual and spatial reasoning in design, J. S. Gero & B. Tversky (eds.), 93–101. Sydney: Key Centre of Design Computing and Cognition. Tversky, Barbara & Paul U. Lee. 1998. How space structures language. In Spatial Cognition: An interdisciplinary approach to representing and processing of spatial knowledge, C. Frecsa, C. Habel, & K. F. Wender (eds.), 157–175. Berlin: Springer-Verlag. Wraga, Mary & Stephen M. Kosslyn. 2003. Imagery. Encyclopedia of Cognitive Science, vol. 2, 466–470. London: Nature Publishing Group.

part iii

Descriptive application Other cognitive approaches

Imperfective aspect and epistemic modality Ronny Boogaart and Radoslava Trnavac

University of Leiden and Simon Fraser University As is well known, the epistemic reading of modal verbs typically arises with imperfective complements. It is argued that this is related to a more general connection between imperfective aspect and subjectivity: imperfective forms express simultaneity of a situation with an independently provided point of reference. This may be a point of perspective, an epistemic evaluation time, or the point of speech itself. Data from Russian, however, suggest that this particular link between imperfective aspect and epistemic modality is restricted to imperfective aspect of the Germanic, and Romance, kind. Keywords: aspect, perspective, subjectivity, epistemic modality, English, Russian

1. Introduction A well known observation on the connection between imperfective aspect and epistemic modality is illustrated in (1) (cf. Coates 1983, Heine 1995, Barbiers 1996, Abraham 2001 and the contributions to Abraham & Leiss 2008).1 (1) a. He must read that book. b. He must be reading that book.

deontic (?epistemic) epistemic (deontic)

The complements of the modal verb must in (1) differ from one another aspectually: in (1a), the complement presents a telic situation, in (1b) it contains a progressive construction, which is a way to express imperfective aspect in English.2 The aspectual distinction between (1a) and (1b) has definite consequences for the kind of modal readings that are compatible with these sentences: (1a) resists an epistemic reading, whereas for (1b), this reading is in fact the most easily available 1. The authors wish to thank the editors of this volume and an anonymous reviewer for their inspiring comments on an earlier version of this chapter. 2. Our use of the terms ‘perfective’ and ‘imperfective’ aspect for English will be explained in Section 2.1. We are well aware of the differences with Slavic aspect, see Section 4.

 Ronny Boogaart and Radoslava Trnavac

one. In this chapter, we will take a closer look at the imperfective-epistemic link. Two main questions guiding the discussion are the following: 1. Why is imperfective aspect required for an epistemic reading of the modal verb in sentences such as (1)? 2. How general is the connection between imperfective aspect and epistemic modality cross-linguistically? In the first part of the chapter, focusing on data from English, we will broaden the scope of our investigation beyond the deontic/epistemic interpretation of modal verbs. We will suggest that the phenomenon illustrated in (1) is part of a more general connection between imperfective aspect and subjective information that can be postulated at least for Romance and Germanic languages (Section 2). The semantic property that unifies all the different readings of imperfectives in these languages, including epistemic-modal readings, is constituted by the anaphoric temporal ordering relation of simultaneity with a point of reference (Section 3). In the second part of our chapter we confront this analysis with counterexamples from Russian (Section 4). Some of these data, such as the modal readings of the present perfective, do not constitute real counterexamples (Section 4.1), but other data show that there is no clear link between imperfective and epistemic in Russian (Section 4.2). In the final section, we argue that the Russian data do not invalidate our initial explanation for the imperfective-epistemic link in Germanic and Romance. In fact, the semantic property that we hold responsible for the epistemic reading of imperfectives in these languages is clearly not part of the semantics of the imperfective in Russian (Section 5). 2. Imperfective aspect and subjectivity In this section, we show that the close link between imperfective aspect and epistemic modality that was illustrated in (1) can be observed in other contexts than the complements of modal verbs as well. More specifically, the phenomenon in (1) is arguably related to the predominant use of imperfective rather than perfective forms in counterfactual contexts that has been noted for various languages and that we will demonstrate to exist in English as well (Section 2.2). Moreover, there is an interesting parallel with the use of imperfective aspect in ‘perspectivized’ discourse, such as, most notably, in (free) indirect speech (Section 2.3). For the time being, we use the term ‘subjective’ to indicate what is shared by epistemic modals, counterfactuals, and represented speech and thought: in all these types of contexts, information is not presented, objectively, as actually being the case in the world, but rather situated in a different ‘space’ (or ‘world’), which may be either a

Imperfective aspect and epistemic modality 

particular ‘viewpoint space’ or a ‘counterfactual space’.3 After our discussion of the subjective contexts in which languages typically prefer the use of imperfective aspect, we will define more precisely the semantic property that we feel is responsible for the subjective, including epistemic, potential of imperfectives, namely anaphoric temporal reference (Section 3). 2.1

Modal verbs

From the start of our discussion, we want to extend the observations on the interaction of aspect and modality in English to include the lexically stative complement in (1c) and the perfect complement in (1d).

(1) a. He must [read]perf that book.

deontic (?epistemic)

b. He must [be reading] imperf that book.

epistemic (deontic)

c. He must [be] imperf someone who likes to read.

epistemic (deontic)

d. He must [have] imperf read that book.

epistemic (deontic)

As far as the modal interpretation of must is concerned, (1c) and (1d) clearly pattern with (1b) rather than (1a). A deontic reading of the modal verb, i.e. an obligation reading, is not incompatible with the sentences in (1b–d), but without further contextual support (for instance, a temporal phrase such as by next week in (1d) or a purpose clause such as in order to get that job in (1c)), an epistemic reading is more readily available for these sentences. The reverse is true for (1a), for which an epistemic reading is in fact hard to come up with at all. At this point, it will be necessary to explain in what way the data in (1) confirm the connection between imperfective aspect and epistemic modality, since it may not be obvious that the complements in (1b–d) are in fact imperfective. After all, English verb forms, with the exception of the progressive in (1b), are not overtly marked for perfective or imperfective aspect. However, the aspectual interpretation of unmarked verb forms can generally be inferred on the basis of, on the one hand, lexical content (events versus states), and, on the other hand, the grammaticalization of aspectually marked forms such as the progressive. For English, such an analysis boils down to the following: lexical statives are standardly imperfective, whereas event-clauses are perfective unless they are marked by a progressive verb form (Smith 1991, Boogaart 1999, 2004).4 3. Fleischman (1995) refers to the latter two as ‘irrealis’, but we feel that is too strong a label for the perspectivized use of imperfectives in (free) indirect speech, as well as for the epistemic reading of modal verbs that she does not discuss. 4. More precisely, the English progressive construction covers the imperfective domain for events and stage-level statives (Boogaart 1999: 173–174).

 Ronny Boogaart and Radoslava Trnavac

As for the semantic definition of the perfective/imperfective distinction, we do not principally depart from the tradition established at least since Comrie (1976): being imperfective, in our analysis, means presenting a situation, ‘from the inside’, as ‘ongoing’ at a particular moment in time, without reference to the boundaries of the situation; a perfective verb form instead presents a situation, ‘from the outside’, as a completed whole, thus including both its starting point and endpoint. (We will return to the semantics of aspect in Section 3.1.) Thus, in (1b) the complement of must is imperfective because it contains a progressive verb form; in (1c), the complement is imperfective because the English predicative construction with to be is inherently imperfective, and does not need (or even allow) to be marked explicitly as such (*He must be being someone who likes to read). As for the perfect complement in (1d), we assume that the situation presented by means of the auxiliary have is an imperfective state as well, just like the situation presented by be in (1c) (or in (1b), for that matter). The sense of anteriority expressed by the perfect construction should be attributed to the meaning of the participle: the time of have in (1d) is simultaneous with the time of must, the participle read presents a situation that precedes this time. Thus, we assume that in (1d) the epistemic reading is triggered for the same reason as in (1b) and (1c) (or, for instance, in he must have that book). Such a compositional analysis of the perfect construction, assuming both the (imperfective) auxiliary and the (perfective) participle to present a situation in their own right (see e.g. Janssen 1993 and Wada 2001), shows that, in addition to (1b) and (1c), (1d) may be regarded as illustrating the imperfective-epistemic link as well. 2.2

Counterfactual

In a seminal paper, Fleischman (1995: 519) has shown convincingly that “in many languages of the world we encounter a more than chance connection between the aspectual category imperfective and irrealis modality”. Fleischman’s notion of ‘irrealis’ covers many phenomena that we would rather not term ‘irrealis’, nor ‘modality’, but she does treat two sets of data that we feel are highly relevant for our present discussion of the epistemic-imperfective link. One concerns the use of aspect in ‘speech and thought quotation’ (see next section), the other one concerns the use of imperfective aspect for ‘unreal conditionals’ (1995: 523–24). Of the latter, Fleischman provides examples from Romance languages, such as French (2a) (‘counterfactual to the present’) and (2b) (‘counterfactual to the past’). (2) a. Si j’ avais le temps, je t’ écrirais. if I haveimperf, past the time I you writeconditional

‘If I had time, I would write to you’

Imperfective aspect and epistemic modality 

b. Si j’avais eu le temps, je t’ aurais écrit. if I havepast, perfect the time I you writepast, conditional

‘If I had had time, I would have written to you’

Both the protasis (subclause) and the apodosis (main clause) of these conditional constructions contain forms that are formally imperfective past. The crucial point of such examples is not that a past tense is used to express something other than temporal precedence, but rather that an imperfective past, and not a perfective past such as the passé simple, is used here. Iatridou (2000), in fact, considers imperfective aspectual morphology in counterfactuals to be ‘fake’: it fails to receive its usual aspectual interpretation. This is most evident in examples such as Modern Greek (3) (Iatridou 2000: 237). (3)

An echtizes to spiti se ena mina if builtimperf the house in one month tha prolaveness na to pulisis prin to kalokeri. fut have-time-enoughimperf to it sell before the summer

‘If you built this house in a month, you would be able to sell it before the summer’ Even though the verb form in the protasis displays imperfective morphology, the event of building the house is actually interpreted perfectively (because of the ‘completive’ adverbial). From the list of languages that, according to Fleischman, use imperfective aspect in ‘unreal conditionals’, English is absent. This is correct since English does not systematically use states and progressives (if he was/were coming...) rather than eventives (if he came...) in if-clauses of the type illustrated in (2) and (3), as is also evidenced by the English translation of (3).5 However, the English data in (4), presenting counterfactual wishes, are compatible with Fleischman’s claims about a link between imperfective aspect and ‘irrealis modality’.

(4) a. ?I wish I [climbed] perf Mount Everest. b. I wish I [was/were climbing] imperf Mount Everest. c. I wish I [was/were] imperf a mountaineer. d. I wish I [had] imperf climbed Mount Everest.

5. Since it is unproblematic to use an eventive (‘perfective’) predicate in English if-clauses (If I won the lottery, I would buy a house), the connection between aspect and modality is not entirely clear in this context. Schouten (2000), however, shows that such eventive clauses are typically interpreted as presenting real possibilities, whereas stative if-clauses (if I was/were sick) prefer counterfactual readings. See Section 4.1.3 for a highly similar observation from Russian.

 Ronny Boogaart and Radoslava Trnavac

The sense of counterfactuality implied by wish is incompatible with a perfective complement clause, as in (4a), whereas it can be felicitously combined with a progressive (and thus imperfective) complement, as in (4b), as well as with a lexically stative complement or a perfect complement, as in (4c) and (4d). Assuming that lexical states as well as (the finite, stative part of) progressive and perfect constructions express imperfective aspect (see 2.1), these sentences show that in English as well there is a connection between imperfective aspect and counterfactuality. To put it the other way around, there is apparently something ‘factive’ about eventive (and thus perfective) simple past clauses such as I climbed Mount Everest, which makes them unsuited for use in a counterfactual wish context. Moreover, the parallel between (4) and (1) is striking: the same kind of complements that, in (1), allow for an epistemic reading of the modal verb must, in (4) allow for a counterfactual reading. This suggests that in addition to the links between ‘imperfective and irrealis’ and between ‘imperfective and epistemic’ there is also a connection between ‘irrealis and epistemic’. As mentioned earlier, the common denominator of these two can be characterized in terms of subjectivity. In order to clarify this point, and to connect it to the aspectual semantics of imperfective aspect, we will discuss one more context that typically prefers the use of imperfective forms, namely that of reported speech and thought. 2.3

Perspectivized discourse

Given the semantics of perfective and imperfective aspect, presenting ‘completed’ and ‘ongoing’ situations respectively, it is understandable that their main function in narrative discourse is quite different: whereas perfective forms are typically used to present a sequence of, iconically ordered, events that constitutes the ‘backbone’ or ‘foreground’ of a story, imperfective forms are used to provide information on situations that are going on in the ‘background’ (Hopper 1979). In temporal terms, perfective forms move narrative time forward by introducing a new point of reference, whereas imperfective forms present situations holding at the time of the preceding perfective (Kamp & Reyle 1993). A standard example of the combination of a perfective and an imperfective in English narrative discourse is given in (5a). (5) a. John entered. It was pitch dark in the room. b. John switched off the lights. It was pitch dark in the room. On closer scrutiny, however, it is a more accurate description of the interpretation of (5a) to say that the state of darkness is holding at a point in time ‘just after’ the entering of the room, which can be identified with the time at which John noticed or realized the room to be dark; such an analysis is needed for (5b) anyway, since

Imperfective aspect and epistemic modality 

the room most likely was not dark yet before (or at) the moment at which John switched off the lights (Hinrichs 1986). Now, it is true more generally that situations presented by means of imperfective forms – thus including, in our analysis, lexical states as well as progressives and perfects in English – are often interpreted as holding not at the time of an event presented explicitly in the surrounding discourse, but rather at a point of perspective that has to be inferred (Caenepeel 1989; Sandström 1993).6 For the progressive, such examples are discussed by Ehrlich (1987); see (6) from Virginia Woolf ’s novel To the lighthouse.

(6) Looking at this hand he thought that if he had been alone dinner would have been almost over now; he would have been free to work. Yes, he thought; it is a terrible waste of time. The children were dropping in still. ‘I wish one of you would run up to Roger’s room’, Mrs Ramsay was saying.

Even though there is a progressive verb form, expressing imperfective aspect, in the last sentence of this fragment, the clause does move narrative time forward. The progressive is there only to maintain the ‘perspectivized’ mode of presentation that was introduced earlier in the fragment. Thus, the situation of Mrs Ramsay talking is presented from the perspective of the story character referred to by ‘he’ in the preceding sentences. An example involving the past perfect is given in (7) (Boogaart 1999: 242).

(7) At that moment, he dropped the vase. He had broken Mary’s most precious belonging!

The standard, temporal function of the past perfect in narrative is to present a situation that precedes a reference point in the past, typically provided by the situation of the preceding sentence (E < R < S, in Reichenbach’s 1947 representation). In (7), however, the situation of breaking clearly does not precede the dropping of the vase. Rather, the second sentence describes what ‘he’, or some other observer of the scene, thinks right after the time of the preceding sentence: the imperfective state referred to by had holds at a point of perspective just after the dropping of the vase (cf. (5b)); the event of breaking itself precedes this time but does not have to precede the event of the preceding clause. Such uses of the past perfect, described in some detail by Oversteegen & Bekker (2002), may in fact move narrative time forward, like simple past eventive (and thus perfective) clauses are supposed to do. 6. This analysis converges with one in terms of ‘dialogism’ or ‘polyphony’, which can be found in the French literature on tenses (see, for example, Bres 2009). According to this approach, the imperfect tenses can in certain contexts serve to indicate the presence of a ‘secondary enunciator’ who is reporting the situation described as it is unfolding.

 Ronny Boogaart and Radoslava Trnavac

The perspectival potential of imperfective forms is, however, not restricted to cases of narrative discourse, let alone to sophisticated literary fiction. It may be observed also in complement clauses of indirect speech, as in (8).

(8) a. He told us that he [climbed] perf Mount Everest.

b. He told us that he [was climbing] imperf Mount Everest. c. He told us that he [was] imperf a mountaineer. d. He told us that he [had] imperf climbed Mount Everest. In (5), (6) and (7), a time of seeing, hearing and thinking is inferred. In (8), the matrix clause provides an explicit point of perspective for the situation of the complement clause to be anchored to. The exact analysis of tense in indirect speech is an issue of quite some debate, but for present purposes it suffices to note that the embedded situation in (8b–d) is most easily interpreted as holding at the time of the matrix clause: the time of was in (8b) and (8c) and the time of had in (8d) is simultaneous with the time of told in the dominating clause. In these instances, the situation is presented at least partly from the perspective of the reported speaker (the ‘quotee’, referred to as ‘he’). Since we are dealing with indirect rather than direct speech, the point of view of the reporting speaker, the one who utters the entire sentence (‘the quoter’), is also involved to some extent, but for imperfective complements as in (8b–d) it is clear that the time of the matrix may provide the time at which the situation is holding. For the perfective past complement in (8a), it may be argued that its time is not linked to the time of the matrix at all: the embedded past tense in (8a) only expresses ‘past’ with respect to the time of the reporting speaker (Declerck 1991, Salkie & Reed 1997, Boogaart 1999; cf. Stowell 1993 and Abusch 1997 for discussion). Such an ‘absolute’, as opposed to ‘relative’, construal is not incompatible with the syntactic context of indirect speech, as should be clear from (9).7

(9) He told us that he is a mountaineer.

Just like the embedded present tense in (9) is necessarily interpreted with respect to the deictic center of the reporting speaker, the embedded past tense in (8a) may be interpreted, independently of the matrix time, as past with respect to the point of speech (paraphrase: ‘he climbed Mount Everest – that’s what he told us’, or ‘according to him, he climbed Mount Everest’). An important argument in favor of such an analysis of (8a) is that it provides us with an explanation for the difference between (8a) and (8d). In both sentences, the resulting temporal interpretation is 7. Cf. Verhagen (2005: 78–154) for the more general claim that finite complement clauses should very often not be interpreted as in any way ‘subordinated’, either semantically or syntactically, to the main clause.

Imperfective aspect and epistemic modality 

that the event of climbing the mountain precedes him telling us about it.8 The difference, then, is not temporal but perspectival: whereas in (8a) the perspective is maintained at the here-and-now of the reporting speaker (as in (9)), in (8d), the situation is explicitly (also) presented from the perspective of the reported speaker (as in (8b–c)). Thus, in perspectivized discourse, we see, both in indirect speech and in ‘free indirect speech’ (which includes reported thoughts and perception as in (5)–(7)), a preference for imperfective aspect. In French, in fact, the perfective past passé simple hardly ever occurs in the complement clause of indirect speech, which prefers the imperfective past imparfait. If the passé simple does occur, as in (10), then the complement clause receives an independent reading, i.e. the claim made in the complement clause is presented as a fact by the reporting speaker himself at the time of speaking, just as in our analysis of (8a) and (9) (Landeweerd & Vet 1996).9 (10) Marie savait qu’Einstein fut un grand savant. Marie knew that Einstein wasperf a great scholar ‘Marie knew that Einstein was a great scholar’10 In the following section, we connect these observations on the ‘perspectivized’ use of imperfectives to the counterfactual and epistemic readings of these forms as well as to the overall semantics of imperfective aspect. 3. Imperfective aspect and anaphoric temporal reference 3.1

The semantics of tense and aspect

In addition to strictly aspectual/temporal definitions of perfective and imperfective aspect, as referring to completed and incomplete situations, traditional 8. In our ‘absolute’ analysis of the embedded past tense in (8a), this precedence reading cannot be attributed to the past tense as such, which, in our view, expresses past with respect to the point of speech rather than with respect to the time of the matrix. The precedence reading in such cases follows from (a) perfective aspect ruling out a simultaneity reading, and (b) pragmatics ruling out a so-called ‘forward shifted’ reading, where the climbing of the mountain would be in the future relative to the time of the matrix; on the latter, see Cutrer’s (1994) factprediction principle. 9. The passé simple does give rise to a relative reading following a future tense in the matrix clause, but this is true for past-under-future generally, also in English (he will say that he climbed Mount Everest). 10. In the English translation of (10) this is less evident than it is in the French example (or in (8a) and (9)), since both Einstein était un grand savant (imperfective past) and Einstein fut un grand savant (perfective past) will be rendered as Einstein was a great scholar in English.

 Ronny Boogaart and Radoslava Trnavac

definitions of aspect often include some notion of perspective. Thus, as mentioned already in Section 2.1, perfective and imperfective forms are said to present situations ‘from the outside’ and ‘from the inside’, respectively (Comrie 1976). Even though such characterizations of aspect already suggest a possible way of unifying all the ‘subjective’ uses of imperfective aspect reviewed above, they are clearly not very precise. To remedy this, we will adopt, with some modifications, the anaphoric analysis of imperfective aspect that has become popular for the French imparfait since Kamp & Rohrer (1983). For a felicitous use of this form, a reference time has to be independently provided by the surrounding discourse; the situation is interpreted as simultaneous with (holding at) this reference time.11 If we use R to represent the ‘antecedent time’, the semantics of imperfective past may thus be represented in Reichenbachian terms as in (11). (11) imperfective past: E, R < S As demonstrated for the sequences in (5), Kamp & Reyle (1993) extend this claim to include lexical statives (and progressives) in English past tense narrative discourse as well. And indeed, in narrative as well as in non-narrative discourse, English sentences such as he was sick and he was reading a book cannot be used in ‘out of the blue’ contexts either: these sentences claim that the situation of him being sick or reading a book held at a definite moment in the past that is somehow already under discussion.12 Crucially, such imperfective forms do not claim that the situation entirely precedes the point of speech (he was sick when I last saw him and, for all I know, he is still sick now) and, therefore, a characterization such as ‘E < S’ would be insufficient. In fact, when using an imperfective, the speaker does not make any claims about what happens either before or after the reference point. We take the anaphoric analysis of imperfective aspect one step further: we claim that ‘simultaneity with an independently provided reference time (E, R)’, constitutes the basic meaning of imperfective aspect as such.13 More specifically, it is independent of tense: it applies to imperfective present as well as to imperfective past. In the case of an imperfective present, as in he is sick or he is reading a book, the reference time is simply constituted by the point of speech, which is an 11. In the most typical use of imperfectives, R is ‘properly included’ in E, but this is ultimately dependent on lexical content rather than part of the semantics of aspect (Boogaart 1999: 170–172). 12. A potentially problematic class of examples for the anaphoric analysis is constituted by individual level predicates as in Einstein was a great scholar (see example (10) and footnote 10), where the antecedent time seems to be constituted by the whole lifetime of Einstein (or at least the relevant subpart thereof). We may have to allow for the possibility that, being unmarked for aspect, the English simple past is compatible with a perfective reading even in stative clauses. 13. At least for Romance and Germanic languages; cf. Section 4 on Slavic-style aspect.

Imperfective aspect and epistemic modality 

indispensable part of any speech event and is thus per definition ‘independently given’, see (12). (12) imperfective present: E, R, S Moreover, the anaphoric analysis of imperfective aspect also captures an important characteristic of the interpretation of perfect forms. As was already mentioned at several points in the preceding sections, we take the finite verb form of these complex constructions to present a situation in its own right. Thus, the perfect presents two situations: an imperfective state by means of the auxiliary, and a perfective event by means of the past participle. The semantics of the perfect, then, includes the semantics of imperfective aspect in (11)–(12) as part of its meaning: the past perfect presents an imperfective state holding at a reference point in the past (cf. (11)), whereas the present perfect presents an imperfective state holding at the point of speech (cf. (10)). In either case, this does not suffice as a semantic description since the contribution of the past participle must be taken into account: the time of the past participle precedes the time of the auxiliary. Combining the semantics of the auxiliary and the participle results in (13) and (14) as our representation for past perfect and present perfect.14 (E1 refers to the situation presented by means of the past participle, whereas E2 represents the state referred to by the perfect auxiliary, cf. e.g. Wada 2001.) (13) past perfect: (14) present perfect:

E1 < E2, R < S E1 < E2, R, S

But now how to represent the meaning of ‘perfective past’ forms as in he read a book or he climbed Mount Everest? Since Reichenbach (1947), it has become customary to represent the meaning of the simple past in English as E, R < S. However, using the notion of R to represent the antecedent time required for the interpretation of imperfectives, we already reserved this very formula for the category of imperfective past in (12). This representation captures the intuition that imperfective past forms cannot be used in ‘out of the blue’-contexts and that they do not express that the entire situation E precedes the point of speech S; imperfective forms only assert what is valid at R. Precisely these considerations are often used to motivate the postulation of a point of reference R, independently of aspect, in the semantics of the simple past per se (cf. also the motivation given by Klein 1994 for his Topic Time), but they clearly do not apply if the clause is eventive/perfective rather than stative/imperfective. Thus, in eventive clauses, the English simple 14. These depart from Reichenbach’s (1947) representation only by including a second E to represent the independent contribution of the finite verb form, presenting a stative – and thus imperfective – situation.

 Ronny Boogaart and Radoslava Trnavac

past can be used in ‘indefinite’, out of the blue contexts (I hear that John found a diamond in his garden! – Stowell 1993) and such sentences do claim that the entire situation E precedes the point of speech S (?He read the book and he may still be reading it now). Therefore, the representation in (15) is in fact sufficient to represent the semantics of perfective past (cf. Löbner 1988: 175).15 (15) perfective past:

E<S

As for perfective present, combining the semantics of present and perfective would boil down to something like ‘the entire situation E occurs at the point of speech S’ (E, S), which shows why perfective aspect and present tense are basically incompatible (?he reads a book), unless the combination is given a habitual reading (He reads a book per week). In the latter case, the situation is coerced into a special kind of stative/imperfective situation, with the time of speech serving as reference time at which it is holding, as in (12). We will return to the topic of perfective present in our discussion of Russian in 4.2. 3.2

The subjective potential of imperfective aspect

The semantic characterization of imperfective aspect as presenting a situation holding at a contextually determined reference time (E, R) is general enough to capture all uses of imperfective forms, including the subjective and epistemicmodal uses presented in Section 2. We assume there to be a cline going from temporal (‘objective’) to modal (‘subjective’) uses of this form, depending on the nature and source of the reference point. In this section, we apply the anaphoric analysis of imperfective aspect to the three types of subjective contexts from Section 2: perspectivized discourse (3.2.1), counterfactual wishes (3.2.2) and epistemic modals (3.2.3.). 3.2.1 Perspectivized discourse While in the case of an imperfective present, the reference time is simply given by the point of speech (see (12)), imperfective past forms present us with many more possibilities. The reference time may be provided, for instance, by a temporal adverbial or a when-clause, as in (16a) and (16b), or by the time of a syntactically dominating clause, as in (16c), or an immediately preceding clause, as in (16d). (16) a. At 8 o’clock he was still writing his paper on aspect. b. When I came in, he was writing a paper on aspect. 15. The ‘definite’ reading of eventive clauses such as in Partee’s (1973) famous example I didn’t turn of the stove, which does not refer to just any moment in the past at which the speaker could have turned off the stove, we consider to be a matter of pragmatics (Wilson & Sperber 1993).

Imperfective aspect and epistemic modality 

c. I met a guy who was writing a paper on aspect. d. I came into the room. He was writing a paper on aspect. Such examples may thus be regarded as illustrating the strictly temporal use of imperfective aspect since the reference time in these cases is a temporal reference point. However, as was already noted for the sentences in (5), the reference time may also be a point of perspective that is not given explicitly in the linguistic context but merely implied. This, in fact, seems to be the most plausible interpretation of the sentences in (16) as well, which suggests that even if a temporal reference point is given explicitly in the context, hearers/readers are still inclined to assign a perspectivized reading to an imperfective form in order to arrive at a maximally coherent interpretation.16 It will be clear by now why, as illustrated in 2.3, imperfective past forms dominate discourse representing the speech or thought (or perception) of a story character, that may in fact be the speaker himself at an earlier moment in life (as in (16b) and (16d)). Taking a point of perspective in the past to function as reference point is just one way of fulfilling the ‘anaphoric requirement’ (E, R) imposed by an imperfective past form. If such a viewpoint is explicitly provided by a matrix clause, i.e. in indirect speech such as in (8), repeated below, then it is difficult not to regard it as the antecedent time for the embedded imperfective, just like the point of speech is the default reference time for an imperfective present.17

(8) a. He told us that he [climbed] perf Mount Everest.

b. He told us that he [was climbing] imperf Mount Everest. c. He told us that he [was] imperf a mountaineer. d. He told us that he [had] imperf climbed Mount Everest. The embedded perfective past in (8a) simply presents the event of climbing Mount Everest as preceding S (see (12)) and it does not even allow for a simultaneity reading – for exactly the same reason that He climbs Mount Everest cannot present an event actually going on at the point of speech. (However, just like for the imperfective present, habitual readings are allowed: he told us that he climbed Mount Everest every year.) 16. There does not have to be a prior linguistic context for the use of an imperfective to be felicitous. Thus, one may open a conversation by saying ‘He was sick!’ as long as the relevant reference time is identifiable for the hearer (like the reference of he in the same utterance). 17. For an embedded imperfective past it is, of course, not impossible to override the matrix time: the reference time may also be provided by the preceding discourse or by a temporal adverbial in the complement clause (he told us that he was climbing Mount Everest this time last year; he told us that he had already climbed Mount Everest before he got married to her).

 Ronny Boogaart and Radoslava Trnavac

The ‘objective’ or ‘factive’ feel of sentences such as he climbed Mount Everest in our analysis follows from the fact that a perfective past form does not require the hearer to look for any additional reference point or perspective point other than that provided by the actual point of speech itself. Imperfective forms, on the other hand, allow for a particular kind of subjective reading, called ‘character subjectivity’ by Verhagen (2000), that consists of taking a point of perspective in the past as the reference point needed for an adequate interpretation of the form (cf. Brisard 2010, who claims that it is the very meaning of the French imparfait to construe a viewpoint distinct from that of the speaker at the time of speaking). 3.2.2 Counterfactual The sentences in (4), repeated below, illustrated that the use of a perfective past in counterfactual wishes is infelicitous whereas the use of an imperfective past – be it a lexical stative, a progressive or a perfect – is fine.

(4) a ?I wish I [climbed] perf Mount Everest. b. I wish I [was/were climbing] imperf Mount Everest. c. I wish I [was/were] imperf a mountaineer. d. I wish I [had] imperf climbed Mount Everest.

Even though wish in the matrix clause of (4a–d) looks like a present tense and the speaker wishes something to be true at the moment of speaking, such sentences require the use of a past tense in the complement, as is evidenced by the ungrammaticality of (17). (17) *I wish I am climbing Mount Everest. This is not unexpected since a present tense form presents a situation as actually holding at the point of speech, and it is the very point of these sentences that these situations are not true (hence the term ‘counterfactual wish’). However, (4a) fulfills this requirement – a perfective past is a past tense – and yet most people prefer the use of a past perfect in this context, as in (4d). In our view, this is because the perfective past in (4a) actually shares an important feature with the present tense in (17). In the preceding section, we claimed that a perfective past form does not require the hearer to look for any additional reference point or perspective point other than that provided by the actual point of speech itself, which is basically a paraphrase of our non-anaphoric analysis of perfective past (as E < S). Now, a present tense is like a perfective past in this respect: since for the present tense the point of speech functions as reference time no alternative perspective has to be construed either. Thus, both a perfective past and a present tense present situations as true from the speaker’s perspective, which in (4a)

Imperfective aspect and epistemic modality 

and (17) obviously clashes with the semantics of wish. Whereas after verbs of saying (as in (8)) or knowing (as in (10)) the use of the ‘factive’ present tense or perfective past, independently linking the situation to S, is still allowed, in wish-contexts as in (4) the situation wished for simply does not exist independently of the wishing. Now, the reference point introduced by the matrix clause in (4) is not merely a temporal reference point, as in (16), or a point of perspective, as in (8), as it is the ‘time’ of a counterfactual world; it is, as it were, the counterfactual counterpart of the point of speech. The perfective past is incapable of expressing simultaneity with this time, just like a perfective present cannot express a situation that is actually going on at S (I climb Mount Everest) – unless given a habitual reading (I climb Mount Everest every year). And, indeed, (18) is a definite improvement over (4a). (18) I wish I climbed Mount Everest every year. An imperfective past, as in (4b–d) and (18), per definition expresses simultaneity with an independently provided reference time and for an adequate use of the form it does not matter whether this reference time is an actual temporal reference point or merely a virtual one (cf. Brisard 2010 on the imparfait). 3.2.3 Modal verbs Turning finally to the puzzle presented in (1), it is important to make a distinction between (i) the aspectual value of the modal verb itself, and (ii) the aspectual interpretation of its complement. (1) a. He must [read]perf that book.

deontic (?epistemic)

b. He must [be reading] imperf that book.

epistemic (deontic)

c. He must [be] imperf someone who likes to read.

epistemic (deontic)

d. He must [have] imperf read that book.

epistemic (deontic)

As for the first issue, it is argued by Boogaart (2007) that the ‘epistemic evaluation time’ required for the interpretation of an epistemic modal verb is a specific instantiation of the temporal reference point that is needed for a coherent reading of any imperfective. Thus, must in both non-epistemic (1a) and epistemic (1b–d) presents an imperfective state holding at a point of reference which, on the most plausible reading of these sentences, is constituted by the point of speech. The difference between epistemic modals and other imperfectives (including nonepistemic modals) is that for an epistemic reading a highly specific kind of reference point is needed, namely a point in time that can function as epistemic evaluation time. Now, the most likely candidate for this is obviously constituted by the point of speech. Epistemic modality has, in fact, been defined as ‘speaker oriented’ (as opposed to ‘agent oriented’) modality: the truth of a proposition is

 Ronny Boogaart and Radoslava Trnavac

judged by the speaker to be either ‘possible’ or ‘necessary’ (to different degrees). Thus, by uttering (1b–d) the speaker expresses that, at the moment of speaking, he feels it is very likely that the embedded proposition is true. Following Verhagen (2000), we are dealing here, then, with a case of ‘speaker-hearer subjectivity’, as opposed to the ‘character subjectivity’ discussed in Section 3.2.1 above.18 The specific semantics of epistemic modality, together with our anaphoric analysis of imperfective aspect, explains why the perfect of a modal typically resists epistemic readings, as is exemplified for Dutch in (19b). (See Eide 2011 for a recent, cross-linguistic overview and a discussion of counterexamples.) (19) a. Hij moet dat boek gelezen hebben. he must that book readpst, part haveinf

‘He must have read that book’

b. Hij heeft dat boek moeten lezen. he has that book mustinf readinf

‘He has had to read that book’

(19a) is the Dutch equivalent of English (1d). In these sentences, the modal verb presents an imperfective state holding at a point of reference in the present (E, R, S), i.e. the point of speech can be interpreted as epistemic evaluation time and an epistemic reading of the modal is therefore compatible with the linguistic information provided by the utterance. Because of the perfect form of the complement, the epistemic judgment is about a past event: the time of hebben (‘have’) coincides with the time of moet (‘must’), and thus with the point of speech, but the participle expresses precedence. In (19b), however, the modal verb does not allow for an epistemic reading. In this sentence, it is not the complement of the modal but the modal verb itself that appears in perfect form. For English must, the perfect form does not exist (anymore), but the difference between Dutch (19a) and (19b) may be compared to that between English (20a) and (20b), containing the ‘semi-modal’ have to (Eide 2011: 3). (20) a. He must have done it. b. He has had to do it. 18. This does not mean that epistemic modal verbs can never occur in the past tense. In such rare cases, that seem to be restricted to narrative discourse, the epistemic evaluation time is a time preceding the point of speech at which a story character (that may be the narrator himself, as in (16b) and (16d)), is considering the likelihood of a proposition being true: While going through the files once again she suddenly realized the solution to this crime. The butler must have done it (she thought). Such cases are not really different from the other cases of ‘character subjectivity’ discussed in 3.2.1. See Boogaart (2007) for a detailed discussion of the past tense of epistemic modals; cf. Fagan (2001) for German examples.

Imperfective aspect and epistemic modality 

Whereas in (20a), like in Dutch (19a), an epistemic reading is available, the modal in (20b), like in (19b), can only get a non-epistemic, obligation reading. This follows from our semantic representation for the present perfect in (14), repeated here. (14) present perfect:

E1 < E2, R, S

For an epistemic reading of a modal verb to be possible, a reference time (typically the point of speech) is required to function as an epistemic evaluation time. However, the participial part of a perfect construction, corresponding to E1 in (14), does not make available such a reference time and therefore an epistemic reading is ruled out if the modal verb itself is presented in the perfect form. The perfect construction coerces, as it were, the imperfective state of ‘being possible/necessary’ expressed by modal verbs in the present tense, as in (1b–d), (19a) and (20a), into a perfective event type and thereby makes epistemic-modal, or any other kind of ‘perspectivized’ reading, for the modal verb impossible. The non-epistemic reading of modal verbs, such as the obligation reading of moeten (‘must’) and have to in (19b) and (20b), is different from the epistemic one in the sense that it is not per definition dependent on a reference point and it does not necessarily have any special relationship to the speaker and the point of speech either. Thus, non-epistemic modals behave more like ordinary lexical statives: they impose less restrictions on the nature and source of the reference point and they are compatible with perfective aspect and the non-anaphoric construal that goes with it, as in (19b) and (20b). Our claim that at the heart of the epistemic-imperfective link is the temporal ordering relation of simultaneity (E, R) is further confirmed by the kind of complements that are shown in (1) to typically give rise to epistemic readings. That the modal verb itself presents an imperfective state may be a necessary condition for an epistemic reading, but it is not a sufficient condition. As an additional requirement, the situation of the complement in a modal verb construction needs to be linked to the time of the modal verb. In accordance with our discussion thus far, we are treating epistemic evaluation as a specific kind of subjectivity, or ‘perspectivization’, and we may thus again draw an analogy with the linking of imperfective forms that can be seen in (free) indirect speech contexts (3.2.1) and, especially, in counterfactual wish-contexts (3.2.2). Relevant examples are repeated in (21) and (22). (21) a. He told us that he [climbed] perf Mount Everest. b. He told us that he [was climbing] imperf Mount Everest. c. He told us that he [was] imperf a mountaineer. d. He told us that he [had] imperf climbed Mount Everest. (22) a. ?I wish I [climbed] perf Mount Everest.

b. I wish I [was/were climbing] imperf Mount Everest.

 Ronny Boogaart and Radoslava Trnavac

c. I wish I [was/were] imperf a mountaineer. d. I wish I [had] imperf climbed Mount Everest. For the embedded perfective past in (21a) it was argued that it links the situation of the complement clause to the point of speech independently of the time of the matrix, not unlike a present tense would (he told us that he is climbing Mount Everest this year) (cf. also French (10)). Only the imperfectives in (21b–d) are dependent for their interpretation on a previously established (or inferable) point of reference which, in the absence of further context, is most likely given by the time of the matrix clause. The sentences in (22) differ from those in (21) mainly because the complement of wish cannot be interpreted independently of the time of the matrix at all since the situation that is wished for does not exist outside the counterfactual world that is triggered by wish. Therefore, the choice for a non-anaphoric past, as in (22a), is not expected after a verb such as wishing. Only using an anaphoric imperfective past, as in (22b–d), results in the dependent construal that is required by the semantics of wish. The non-finite complement of a modal verb in an epistemic construction behaves just like the finite complements in (21) and (22): for a dependent reading to be possible, the complement needs to be imperfective. The epistemic constructions are actually more like (22) than like (21) in the sense that an independent construal is not even allowed. On an epistemic reading of the modal it is, after all, the proposition of the complement that is being evaluated at a definite moment in time, typically by the speaker at the point of speech. Thus, the situation of the complement E needs to be linked to the time of epistemic evaluation R (=S), resulting in a construal that we have argued, on independent grounds (see 3.1), to constitute the very meaning of imperfective aspect (E, R).19 On the deontic reading of a modal verb construction, the modal and its complement are less dependent. In (1a), for instance, the obligation expressed by must exists prior to, and independent of, the execution of the reading event. Obligation readings, then, typically imply non-simultaneity of the time of the modal and the time of the complement (i.e. ‘bi-phasicness’ in the terminology of Abraham 2001) and, therefore, prefer the use of perfective forms. Thus, we have shown that the close link between imperfective aspect and epistemic modality in modal verb constructions is part of a more general semantic phenomenon, that includes at least the use of imperfective forms in (free) indirect 19. However, just like in indirect speech and in counterfactual wishes, an alternative reference time may be provided by a temporal adverbial, as in he must be swimming tomorrow at 8, which allows for an epistemic reading, as opposed to he must swim tomorrow at 8. Crucially, linking to a punctual reference time is required for an epistemic reading and this point of evaluation is typically but not necessarily constituted by the point of speech.

Imperfective aspect and epistemic modality 

speech, as in (21), and in counterfactual wishes, as in (22): what all of these have in common is the temporal ordering relation of simultaneity, either with the point of speech or with an additional reference time. The anaphoric property of imperfective forms explains its potential for subjective readings, which may be either instances of ‘speaker-hearer subjectivity’ (if the point of speech is the reference point) or ‘character subjectivity’ (if a perspective point in the past is construed as reference point). Our analysis thus provides a unified explanation for a number of seemingly divergent phenomena, which further include the lack of an epistemic reading for (23) (see this section) and the well-known observation that the simple present in English cannot express situations that are actually ongoing at the point of speech, as in (24) (see Section 3.1). (23) He has had to do it. (24) He climbs Mount Everest. The constraints on the interpretation of (23) and (24), like those on the interpretation of (1a), follow from the incompatibility of perfective aspect and anaphoric temporal reference, or, phrased the other way around, from the fact that imperfective aspect is required for simultaneity (either with the point of speech or with another punctual reference time that can function as point of evaluation). In the final section of the chapter, we will review some counterexamples from Russian and discuss the implications thereof for our anaphoric analysis of imperfective aspect. 4. Aspect and modality in Russian 4.1

Counterexamples to the imperfective-epistemic link

Discussing the imperfective-epistemic link in the first part of our chapter, we focused on data from English, which is a ‘tense prominent’ language (Bhat 1999) that does not systematically encode the perfective/imperfective distinction. When talking about aspect in English one is basically talking about the lexical semantics of predicates, as eventive versus stative, in combination with the grammatical semantics of aspectually marked constructions such as the perfect and the progressive. It will, therefore, be interesting to see to what extent the imperfective-epistemic link can also be observed in an ‘aspect prominent language’ such as Russian. In this language, aspectual choice is salient morphologically. Most verbs are not single lexical entities but occur in morphologically linked pairs in which ‘perfective’ is opposed to ‘imperfective’. Aspect in Russian is less dependent on tense and time reference than in many other languages. For instance, Russian aspect is expressed not only in inflected past tense forms, as is typically the case in Romance languages,

 Ronny Boogaart and Radoslava Trnavac

but also in the present and the future tense, as well as in non-finite verb forms such as imperatives, infinitives, and gerunds; in addition, it is expressed obligatorily in the subjunctive as well. In the following subsections, we will show that there is, in fact, no systematic connection between imperfective aspect and epistemic modality in Russian (cf. Divjak 2009, also for an overview of the literature). We will start out with a brief remark on modal verbs in Russian (4.1.1) and then go on to discuss the use of aspect in other modal constructions, such as the perfective present (4.1.2), the conditional imperative (4.1.3) and the dative-infinitive (4.1.4). (The interaction of aspect and modality in these constructions is described in more detail by Trnavac 2006.) 4.1.1 Modal verbs Russian does not have a system of modal auxiliary verbs that is as developed as it is in English and the other Germanic languages. Instead of these, it mostly uses adverbs and finite modal predicates to express modality, as well as the imperative and infinitive constructions that will be discussed in the following subsections. However, the frequently used moč (‘can’) is very much like English can/could, since it takes an infinitival complement and it allows for both epistemic and nonepistemic readings. These different readings are illustrated in (25) (with an imperfective infinitive) and (26) (with a perfective infinitive). (25)

Mama u Peti očen’ liberal’nyj čelovek, i Pet’a svobodno mother in Petja very liberal person and Petja freely možet gul’at’ na ulice každyj den’ do pozdna. can walkINF, IMP on street every day till late

‘Petja’s mother is a very liberal person, and he can go out every day till late in the evening’ (26)

Est’ li osnovanija verit’ Pet’ke? Pet’ka mog basis trustinf, ipf Petjka Petjka could be prt i sovrat’. and lieinf, pf

‘Is there a basis to trust Petjka? Petjka could have lied’ (Bulygina and Šmelev 1997: 210) Example (25) illustrates the deontic use of moč, expressing permission. Thus, the choice for an imperfective complement does not automatically result in an epistemic reading and in (25) is motivated by ‘aspectual’ rather than ‘modal’ considerations, more specifically by the repetition as expressed by každyj den’ (‘every day’). Given another context, the combination of moč with an imperfective complement is compatible with an epistemic reading as well (‘he could be going out

Imperfective aspect and epistemic modality 

every day till late, but I am not sure of it’). Similarly, the use of a perfective complement with moč does not by itself indicate what kind of modality is involved. In (26), the context triggers an epistemic reading, but in other contexts non-epistemic readings are possible as well (‘he can lie very well when it is needed’). The crucial example in view of our discussion is, of course, the one in (26) since it shows the compatibility of perfective aspect and epistemic modality in Russian and, therefore, constitutes a direct counterexample to our hypothesis in Section 3.20 4.1.2 Perfective present Instead of modal verbs, Russian may use the perfective present to express a range of modal notions. Such uses of the perfective present obviously constitute a most convincing argument against any theory that wishes to postulate an exclusive connection between imperfective aspect and modality. For the specific hypothesis developed in the first part of our chapter, however, the case of the perfective present in Russian does not pose a problem since the modal readings that are compatible with this form are not epistemic readings. In Section 3.1, it was noted that the combination of present tense and perfective aspect in English gives rise to special interpretations: since a situation cannot at the same time be ‘completed’ (perfective aspect) and ‘ongoing at the point of speech’ (present tense), such forms can only be given a habitual/iterative reading (I climb Mount Everest every year). Not all languages, however, solve the semantic clash between perfective aspect and present tense in the same way as English does. In Russian, the form is typically interpreted as referring to a future situation, which may already be considered a ‘modal’ reading, and, in addition, allows for modal readings of the kind illustrated in (27)–(29). (27) Ja nikak ne pojmu. I no way not understandpres, pf ‘I just can’t understand’ (28) On silen, čto ugodno podnimet... he strong whatever liftpres, pf ‘He is strong, he can lift everything you want...’ (29) On vsegda skažet dobroe slovo. he always tellpres, pf good word ‘He will always put in a good word’ These examples confirm that Russian may use the perfective present in contexts where English prefers the use of modal verbs (see the translations of (27)–(29)). 20. Note, however, that the perfective infinitive in Russian (26) is rendered in English by means of a perfect complement, including the ‘imperfective’ auxiliary have. See discussion in 4.2.

 Ronny Boogaart and Radoslava Trnavac

There are two reasons why these cases do not constitute a problem for our hypothesis. First, it is well known, and documented for many languages, that forms that are used to present future situations can also serve modal functions.21 Our claim on the imperfective-epistemic link does, of course, not imply that imperfective aspect is the only means to express modality and future forms are definitely one of the other candidates that are used as such in many languages (That will be the postman!). Second, the modal notions associated with the perfective present include such things as ability and volition, i.e. ‘participant-internal modality’ in the sense of van der Auwera & Plungian (1998), but, crucially, not epistemic-modal readings in the sense that we have been using this term. If anything, therefore, the data on the perfective present confirm rather than contradict our hypothesis. 4.1.3 Conditional imperatives Since, following Fleischman (1995) and Iatridou (2000) (see Section 2.2), we included conditional clauses in our discussion of the imperfective-epistemic link, and in our general notion of ‘speaker-hearer subjectivity’, it makes sense to look at the contribution of aspect in the Russian conditional imperative construction as well. The conditional use of imperatives has been noted to exist in many languages, including Dutch and English (miss that train and you’ll be late for the concert) (Fortuin & Boogaart 2009), but the Russian construction is more general. In particular, its use is not restricted to 2nd person (implied) subjects, as it is in English and Dutch (Boogaart & Trnavac 2004). Another specific property of the Russian construction is that it is mostly used in counterfactual rather than hypothetical contexts.22 Examples (30), (31) and (32) illustrate the conditional (hypothetical and counterfactual) readings of the Russian imperative. (30)

Mne kažet’sja, što vyskažis’ my i vse Idat seems that sayimper, pf we and everything pojdet po-staromu. gopres, pf as before

‘It seems to me that if we speak out, everything will become as before’

21. According to Barentsen (1998), the modal readings of the perfective present in cases such as (27)–(29) are not based on its future meaning, but rather on the perfective form requiring a ‘sequential link’, see 4.2. 22. A rare example of a counterfactual conditional imperative in English is Turn a different corner and we never would have met from the George Michael song ‘Different corner’. In Dutch, the counterfactual use of the imperative requires the use of a past perfect imperative (as in had het gedaan, lit. had it done). A detailed discussion of Dutch and Russian conditional imperatives is offered by Fortuin & Boogaart (2009).

Imperfective aspect and epistemic modality 

(31) Svari ja etu kartošku vo vremja, my by ne opozdali. boilimper, pf I this potatoes in time we irr not late ‘If I had boiled the potatoes in time, we would not have been late’ (32) Zanimajsja ona vo vremja, sejčas by ne bojalas’ ekzamena. studyimper, ipf she in time now irr not be afraid exam ‘If she had studied in time, now she wouldn’t have been afraid of the exam’ The occasional hypothetical reading of the conditional imperative can be distinguished from the counterfactual reading on formal grounds, since, on the hypothetical reading, the apodosis usually contains perfective present, as in (30), or imperfective future, while on the counterfactual reading, the apodosis usually contains the conditional mood (subjunctive), as in (31) and (32). As for the interaction between aspect and modality, Trnavac (2006) argues that the distinction between hypothetical and counterfactual use is also reflected in the apodosis of the construction: on the hypothetical reading, the imperative form prefers perfective aspect (see (30)), while on the counterfactual reading, it can have either perfective or imperfective aspect (see (31) and (32), respectively).23 For present purposes it suffices to note that the imperative in this (typically counterfactual) conditional construction is standardly perfective, which is not in line with our observations on conditional clauses and counterfactual wishes in Section 2.2. If aspect is used in this construction to distinguish between hypothetical and counterfactual readings, then we are dealing with a split within the domain of subjective, epistemic meaning rather than with a split between epistemic and non-epistemic meaning. 4.1.4 Dative-infinitive constructions A final modal construction of Russian that we will briefly discuss is the dative-infinitive construction. As is the case with the conditional imperative, discussed in the previous section, aspect seems to play a subtle role in distinguishing between different kinds of modal readings for this construction. Again, however, the modal distinction indicated by aspect is not one between epistemic and deontic modality. While the dative-infinitive construction does allow for both deontic and 23. The specific interaction between perfective/imperfective aspect and hypothetical/counterfactual interpretation in Russian, and possibly in other languages as well, needs more investigation. Interestingly, a similar connection has been noted for French conditionals: Martin (1991) and Gosselin (1999) show that, with a ‘perfective reading’ of the imparfait in the protasis (due to a telic situation), the meaning is preferably hypothetical, whereas with an imperfective reading (due to an atelic situation) the reading is more counterfactual. Schouten (2000) makes strikingly similar observations for English and Dutch (cf. fn.5).

 Ronny Boogaart and Radoslava Trnavac

epistemic readings, both of these readings, in fact, seem to prefer the use of imperfective forms. An example of the deontic use is given in (33), which expresses obligation. (33) Ložis’, tebe že zavtra idti v sadik. go to bed youdat prt tomorrow goinf, ipf in kindergarten ‘Go to bed, you have to go to kindergarten tomorrow’ (Bricyn 1990: 208) Marginally, the deontic cases can be perfective, especially when used with negation, as in (34). (34) Emu ved’ ne ujti iz goroda. hedat prt not goinf, pf from town ‘After all, he can’t leave the town’ The uses of the dative-infinitive that are sometimes labeled ‘epistemic’ (Rappaport 1985) are those that express inevitability or predestined future (cf. the modal use of the to-infinitive in English sentences such as He is to become the next king). An example would be (35). (35) Gorodu- horošet’. citydat get prettierinf, ipf ‘The city will become nicer (with time)’

(Bricyn 1990: 219)

The main motivation for treating such utterances as epistemic is that they are used to express certainty on behalf of the speaker that some situation will become true. Now, just like the deontic case in (34), the epistemic-like use of the dative infinitive is usually imperfective; the use of perfective aspect is possible, however, with scalar/temporal modifiers, as in (36), containing the adverbial skoro ‘soon’ (Fortuin 2000: 370). (36) Im skoro stat’ soldatami. theydat soon becomeinf, pf soldiers ‘They are to become soldiers soon’ In any case, the distinction between deontic and epistemic-like uses of the dativeinfinitive construction in Russian does not correlate with an aspectual distinction between perfective and imperfective. There is one other type of modal use of the construction that does seem to prefer the perfective form of the infinitive (Trnavac 2006) but that use is an instance of ‘participant-internal’ rather than deontic modality (in the terminology of van der Auwera & Plungian 1998), as in (37).

Imperfective aspect and epistemic modality 

(37)

Vy začem prišli? you why come Mne pogovorit’ s učitelem. Idat talkinf, pf with teacher

‘Why did you come?’ – ‘I need/have to talk to a teacher’(Fortuin 2000: 348) The term ‘participant-internal modality’ refers to a kind of possibility or necessity internal to a participant engaged in the state of affairs: (39) expresses a need inherent to the dative participant. Sentences like these can be paraphrased with nužno (‘need’) or hotet’ (‘want’). Thus, if aspect is used to distinghuish between different modal readings of the dative-infinitive in Russian at all, it is perhaps used to make a distinction between participant-internal modality, on the one hand, and other types of modality (including both deontic and epistemic modality), on the other hand. 4.2

Discussion: Imperfective and anaphoric in Russian

In Section 2, we provided data from Romance and Germanic languages that are used in the literature to illustrate a link between imperfective aspect and epistemic or, more generally, perspectivized information. In Section 3, we suggested that all uses of imperfective aspect, including both ‘objective’ temporal uses and subjective uses, are compatible with an anaphoric analysis of the meaning of imperfective aspect. Situations presented by means of imperfective forms are always interpreted as holding at a contextually determined time, which may be a temporal reference point or, more specifically, a point of perspective, an epistemic evaluation time or the point of speech itself. In the preceding subsections, however, we showed that in Russian there is no systematic connection between imperfective aspect and the various kinds of subjective readings that we discussed in the first part of the chapter (cf. Divjak 2009). Now, does this need to have consequences for our analysis of the (mainly) English data in Section 2? We believe it does not. Rather, we claim that in a crucial respect the semantics of imperfective aspect in Russian is different from the semantics of imperfective aspect in a language such as English. More specifically, the semantic property that we hold responsible for the subjective potential of imperfective forms in Germanic and Romance languages – i.e. anaphoric temporal reference – is clearly not part of the inherent semantics of the imperfective in Russian. The fact that the Russian imperfective does not require any such linking is evidenced most clearly by the so-called ‘general-factual’ use of the form, illustrated in (38) and (39). (38) Kto stroil etot zavod? who buildpret, ipf this factory ‘Who built this factory?’

(Padučeva 1996: 33)

 Ronny Boogaart and Radoslava Trnavac

(39) Ja otkryval okno. I open pret, ipf window ‘I opened the window’

(Padučeva 1996: 36)

In Section 3.1, we argued for an anaphoric analysis of imperfective aspect, defined as simultaneity of a situation E with a reference point R, provided by (or inferable from) the surrounding discourse. On the basis of English examples we showed that imperfective past forms do not claim that the entire situation E precedes the point of speech S (E < S), but merely that the reference point R at which E is holding precedes S (E, R < S). The general-factual interpretation of the imperfective past in Russian, however, as in (38) and (39), can be captured by ‘E < S’: the action is understood to be fully completed before the point of speech and the interpretation does not require any additional linking to some alternative reference point or perspective point in the past. In fact, it has been argued for Russian by Barentsen (1998) that the perfective rather than the imperfective has an anaphoric-like property, which he calls sekventnaja svjaz’ (‘sequential connection’). Any situation presented by means of a perfective verb in Russian has to be linked to an external situation, located either before or after the event itself. A well-known manifestation of the property of ‘sequential connection’ is shown in the narrative use of perfective forms, where consecutive perfectives present a chain of sequential situation: every situation is linked to the result state of the previous situation. See example (40) from Barentsen (1998: 52). (40)

On vstal, pošel k oknu i he get up pret, pf go-pret, pf to window and mahnul rukoj. waved pret, pf hand

‘He got up, went to the window and waved’ This type of linking in narrative discourse is of course well known, also from studies on Germanic and Romance languages. In fact, Kamp & Reyle (1993) use the term ‘anaphoric’ to refer to such interclausal temporal relations in discourse, including both simultaneity readings (as in our restricted use of the term ‘anaphoric’) and sequence readings. According to Barentsen, however, the property of ‘sequential connection’ is an inherent part of the semantics of perfective aspect in Russian and it may also be used to explain many other uses of perfectives in addition to the narrative use in (40). Specifically, the modal use of the perfective present that we discussed in 4.1.2 and is illustrated by Barentsen with the example in (41), would equally follow from this property. (41) Horošij grim vse nedostatki skroet. good makeup all defects hidepres, pf ‘Good make-up can hide all defects’

Imperfective aspect and epistemic modality 

In this case, the event of hiding defects is linked to a preceding situation that is not concrete or explicit as in (40). Rather, in (41), this ‘starting situation’ consists of the presence of some defects: good make-up has the potential to respond successfully to this situation since its application will hide the defects. This way, the (nonepistemic) modal use of perfective present can be related to the abstract overall meaning of perfective aspect, much like we have been trying to relate the (epistemic) modal use of imperfective forms to the semantics of imperfective aspect. The difference between our notion of ‘anaphoric’ as applied to imperfective aspect and Barentsen’s notion of ‘sekventnaja svajz’ corresponds to the distinction between the temporal ordering relations of simultaneity (imperfective) and sequence (perfective). The discussion by Barentsen and the general-factual examples in (38) and (39) suggest that in Russian the ‘sequential link’ required by the perfective form is felt much more strongly than the ‘simultaneity link’ that we associate with imperfective aspect. In a language such as English it is clearly the other way around, as is also evidenced in the English translations of Russian (38) and (39): as argued in Section 3.1, eventive clauses in English may be used to present ‘all new’ information that does not have to be linked to any other, external situation. Since we explained the subjective and epistemic uses of imperfective forms on the basis of their anaphoric nature, it is not surprising – and, in fact, in line with our theory – that non-anaphoric imperfective forms, as in Russian, do not show a similar range of modal interpretations. To conclude, we want to mention two further general differences between the tense-aspect systems of English and Russian that make clear why we should not expect the link between imperfective and epistemic to be as strong in Russian as it is in English. The first one is that Russian lacks a separate perfect form like the English past and present perfect. The relevance of this difference for the present issue should be clear from the English example in (42) and its Russian translation in (43). (42) He told us that he had climbed Mount Everest. (43) On skazal nam, što podnjalsja na Mont Èverest. he told us that climbpret, pf on Mount Everest ‘He told us that he had climbed Mount Everest’ The perfect form in English has many ‘perspectivized’ uses, like the one in (42), and in the complement of a modal verb it triggers an epistemic reading of the modal (he must have done it). From the start of our chapter (see 2.1), we assumed that the data on the English perfect support our hypothesis on the link between imperfective and epistemic. We could do so only because the perfect construction in a language such as English is a compound verb construction that, in addition to a ‘perfective’ participle, includes an imperfective auxiliary. Thus, in (42) the

 Ronny Boogaart and Radoslava Trnavac

imperfective state denoted by had is simultaneous with the reference point provided by the matrix clause (told). However, Russian uses a perfective past form to cover both ‘perfective’ and ‘perfect’ meaning. As a result, English (42) is rendered in Russian as (43), containing a perfective past in the complement clause. Thus, we cannot attribute the ‘relative’ meaning of (43) to anything ‘imperfective’ as we did for English (42). The same problem, in fact, arises with all other subjective and epistemic uses of perfect forms that we discussed in the first part of our chapter (see also the crucial Russian example in (26) and its English translation). This means that in Russian the perfective form will cover the subjective and epistemic semantic domain that in English is covered by the (at least partly imperfective/ anaphoric) perfect form, which, of course, has quite dramatic consequences for the link between imperfective and epistemic in Russian. A final, related, point concerns the fact that Russian does not have sequence of tenses. Contrary to English and French, that use imperfective past in the complement clause of indirect speech to express simultaneity in the past, Russian uses the present tense in such contexts, cf. (44) and (45). (44) He told us that he was climbing Mount Everest. (45) On skazal nam, što podnimaetsja na Mont Èverest. he told us that climbpres, ipf on Mount Everest ‘He told us that he was climbing Mount Everest’ In other words, languages such as Russian use relative tense: the time of the matrix clause is treated as the deictic center for the use and interpretation of tense in the complement clause. This basically boils down to indirect speech using the same tense as would be used in the corresponding direct speech. (Therefore, the embedded perfective past in the earlier example (43) is interpreted as past-of-past.) In Section 3, we used the ‘perspectivized’ use of imperfective past forms, that is most clearly manifested in indirect speech as in (44), as a kind of intermediate step between the ‘objective’ (temporal/aspectual) use of imperfectives and its use in subjective (counterfactual, conditional, epistemic) contexts. Languages that lack sequence of tenses do not use (imperfective) aspect but rather (present) tense to express the relative ordering of situations in indirect speech, as in (45). Thus, imperfective aspect in such languages does not have to serve this particular ‘anaphoric’ function. This, then, might constitute a ‘missing link’ between the strictly aspectual use and the purely modal use of imperfectives. We believe it is at least an interesting hypothesis, to be tested on more languages in the future, that there would be such a connection between sequence of tenses (or the lack thereof) and the modal use of imperfective aspect (or the lack thereof).

Imperfective aspect and epistemic modality 

5. To conclude In this chapter, we did not provide new data but rather tried to give a unified explanation for a range of observations from the linguistic literature concerning the connection between aspect and modality. In particular, we connected the epistemic reading of imperfective complements in modal verb constructions to the more general perspectivized use of imperfective forms. What these, as well as all other, uses of imperfectives have in common is the temporal ordering relation of simultaneity with a contextually provided, or inferable, reference point. For an epistemic reading to be possible, a specific kind of reference time, typically constituted by the point of speech, is needed to function as epistemic evaluation time. This answers the first general question from Section 1, i.e. why is imperfective aspect required for an epistemic reading? As for the second question – how general is the connection between aspect and modality cross-linguistically? – we discussed some counterexamples from Russian. We argued that the link between epistemic modality and imperfective aspect in Russian is not as strong as in our data from Romance and Germanic languages since the semantics of imperfective aspect itself is different. Crucially, the Russian imperfective does not require a contextually provided antecedent time for its interpretation and, following our theory, we should therefore not expect to find a special affinity between this form and subjective-epistemic contexts. To further substantiate our analysis and to provide a more comprehensive answer to our second question, we should obviously look at more languages and more data. While doing so, it will be particularly interesting to take into account the interaction with sequence of tenses noted at the end of Section 4.

References Abraham, Werner. 2001. Modals: toward explaining the epistemic non-finiteness-gap. In Modalität und Modalverben im Deutschen, Reimar Müller & Marga Reis (eds.), 7–36. Hamburg: Buske. Abraham, Werner & Elisabeth Leiss (eds.). 2008. Modality-Aspect Interfaces. Implications and Typological Solutions. Amsterdam & Philadelphia: John Benjamins. Abusch, Dorit. 1997. Sequence of tense and temporal de re. Linguistics and Philosophy 20: 1–50. Barbiers, Sjef. 1996. The Syntax of Interpretation. The Hague: Holland Academic Graphics. Barentsen, Adrie A. 1998. Priznak “sekventnaja svaz’’’ i vidovoe protivopostavlenie v russkom jazyke. In Tipologija vida: problemy, poiski, rešenija, M. Ju. Čertkova (ed.), 43–58. Moskva: Škola “Jazyki russkoj kul’tury”. Bhat, D.N. Shankara. 1999. The Prominence of Tense, Aspect, and Mood. Amsterdam & Philadelphia: John Benjamins.

 Ronny Boogaart and Radoslava Trnavac Boogaart, Ronny. 1999. Aspect and Temporal Ordering: A Contrastive Analysis of Dutch and English. The Hague: Holland Academic Graphics. Boogaart, Ronny. 2004. Aspect and Aktionsart. In Morphology: An International Handbook on Inflection and Word-formation, Vol. 2, G. Booij, C. Lehmann, J. Mugdan (eds.), in collaboration with W. Kesselheim & S. Skopeteas, 1165–1180. Berlin & New York: Mouton de Gruyter. Boogaart, Ronny. 2007. The past and perfect of epistemic modals. In Recent Advances in the Syntax and Semantics of Tense, Mood and Aspect, L. de Saussure, J. Moeschler, & G. Puskas (eds.), 47–70. Berlin: Mouton de Gruyter. Boogaart, Ronny & Radoslava Trnavac. 2004. Conditional imperatives in Dutch and Russian. In Linguistics in the Netherlands, Leonie Cornips and Jenny Doetjes (eds.), 25–35. Amsterdam & Philadelphia: John Benjamins. Bres, Jacques. 2009. Dialogisme et temps verbaux de l’indicatif. Langue Française 163: 21–39. Brisard, Frank. 2010. Aspects of virtuality in the meaning of the French imparfait. Linguistics 48: 487–524. Bricyn, V.M. 1990. Sintaksis i semantika infinitiva v sovremennom russkom jazyke. Kiev: Naukova Dumka. Bulygina, Tatiana.V. & Andrej. D. Šmelev. 1997. Jazykovaja konceptualizacija mira: Pragmatika. Semantika. Leksikografija. Vid. Vremja, Lico, Modal’nost’. Moskva: Škola “Jazyki russkoj kul’tury”. Caenepeel, Mimo. 1989. Aspect, Temporal Ordering and Perspective in Narrative Fiction. PhD Dissertation, University of Edinburgh, UK. Coates, Jennifer. 1983. The Semantics of Modal Auxiliaries. London: Croom Helm. Comrie, Bernard. 1976. Aspect. Cambridge: Cambridge University Press. Cutrer, L. Michelle. 1994. Time and Tense in Narrative and in Everyday Language. PhD Dissertation, University of California, United States. Declerck, Renaat. 1991. Tense in English: Its Structure and Use in Discourse. London: Routledge. Divjak, Dagmar. 2009. Mapping between domains. The aspect-modality interaction in Russian. Russian Linguistics 33: 249–269. Eide, Kristin M. 2011. Modals and the present perfect. In In the mood for mood, T. Mortelmans, J. Mortelmans, & W. de Mulder (eds.), 1–20. Amsterdam & New York: Rodopi. Ehrlich, Susan. 1987. Point of View: A Linguistic analysis of Literary Style. London: Routledge. Fagan, Sarah M.B. 2001. Epistemic modality and tense in German. Journal of Germanic Linguistics 13: 197–230. Fleischman, Suzanne. 1995. Imperfective and irrealis. In Modality in grammar and discourse, J. Bybee & S. Fleischman (eds.), 519–551. Amsterdam: John Benjamins. Fortuin, Egbert. 2000. Polysemy or Monosemy: Interpretation of the Imperative and the Dativeinfinitive Construction in Russian. Amsterdam: ILLC. Fortuin, Egbert & Ronny Boogaart. 2009. Imperative as conditional: from constructional to compositional semantics. Cognitive Linguistics 20: 641–673. Gosselin, Laurent. 1999. Les valeurs de l’imparfait et du conditionnel dans les systèmes hypothétiques. Cahiers chronos 4: 29–51. Heine, Bernd. 1995. Agent-oriented vs. epistemic modality: Some observations on German modals. In Modality in Grammar and Discourse, J. Bybee & S. Fleischman (eds.), 19–53. Amsterdam & Philadelphia: John Benjamins.

Imperfective aspect and epistemic modality  Hinrichs, Erhard. 1986. Temporal anaphora in discourses of English. Linguistics and Philosophy 9: 63–82. Hopper, Paul J. 1979. Some observations on the typology of focus and aspect in narrative language. Studies in Language 3: 37–64. Iatridou, Sabine. 2000. The grammatical ingredients of counterfactuality. Linguistic Inquiry 31: 231–270. Janssen, Theo A.J.M. 1993. Tenses and demonstratives: Conspecific categories. In Conceptualization and Mental Processing in Language, R. A. Geiger & B. Rudzka-Ostyn (eds.), 741– 783. Berlin: Mouton de Gruyter. Kamp, Hans & Christian Rohrer. 1983. Tense in texts. In Meaning, use and interpretation, R. Bäuerle, C. Schwarze, & A. von Stechow (eds.), 250–269. Berlin: Mouton de Gruyter. Kamp, Hans & Uwe Reyle. 1993. From Discourse to Logic. Dordrecht: Kluwer. Klein, Wolfgang. 1994. Time in Language. London: Routledge. Landeweerd, Rita & Co Vet. 1996. Tense in (free) indirect discourse in French. In Reported Speech: Forms and Functions of the Verb, T. A.J.M. Janssen & W. van der Wurf (eds.), 141–164. Amsterdam & Philadelphia: John Benjamins. Löbner, Sebastian. 1988. Ansätze zu einer integralen semantischen Theorie von Tempus, Aspekt und Aktionsarten. In Temporalsemantik, V. Ehrich & H. Vater (eds.), 163–191. Tübingen: Niemeyer. Martin, Robert. 1991. Types de procès et systèmes hypothétiques. De l’aspect de re à l’aspect de dicto. In Les typologies de procès, C. Fuchs (ed.), 87–95. Paris: Klincksieck. Oversteegen, Eleonore & Birgit Bekker. 2002. Computing perspective: the pluperfect in Dutch. Linguistics 40: 111–161. Padučeva, Elena V. 1996. Semantičeskie Issledovanija. Moskva: Škola “Jazyki russkoj kul’tury”. Partee, Barbara H. 1973. Some structural analogies between tenses and pronouns in English. The Journal of Philosophy 70: 601–609. Rappaport, Gilbert. 1985. Aspect and modality in the context of negation. In The Scope of Slavic Aspect, M. S. Flier & A. Timberlake (eds.), 194–223. Columbus: Slavica. Reichenbach, Hans. 1947. Elements of Symbolic Logic. New York: the Free Press; London: CollierMacmillan. Salkie, Raphael & Susan Reed. 1997. Time reference in reported speech. English Language and Literature 1: 319–348. Sandström, Görel. 1993. When-clauses and the Temporal Interpretation of Narrative Discourse. PhD Dissertation, University of Umeå, Sweden. Schouten, Edith. 2000. The role of the Native Language in the Non-Native Acquisition of Hypothetical Conditional Structures. The Hague: Holland Academic Graphics. Smith, Carlota S. 1991. The Parameter of Aspect. Dordrecht: Kluwer. Stowell, Timothy A. 1993. Syntax of Tense. Manuscript, University of California at Los Angeles. Trnavac, Radoslava. 2006. Aspect and Subjectivity in Modal Constructions. The Hague: Holland Academic Graphics. van der Auwera, Johan & Vladimir A. Plungian. 1998. Modality’s semantic map. Linguistic Typology 2: 79–124.

 Ronny Boogaart and Radoslava Trnavac Verhagen, Arie. 2000. The girl that promised to become something: An exploration into diachronic subjectification in Dutch. The Berkeley Conference on Dutch Linguistics 1997: The Dutch Language at the Millennium, T. F. Shannon & J. P. Snapper (eds.), 197–208. Lanham, MD: University Press of America. Verhagen, Arie. 2005. Constructions of Intersubjectivity: Discourse, Syntax and Cognition. Oxford: Oxford University Press. Wada, Naoaki. 2001. Interpreting English Tenses. A Compositional Approach. Tokyo: Kaitakusha Co. Wilson, Deirdre & Dan Sperber. 1993. Pragmatique et Temps. Langages 112: 8–25.

Communicating about the past through modality in English and Thai Katarzyna M. Jaszczolt and Jiranthara Srioutai

University of Cambridge and University of Chulalongkorn The chapter concerns semantic representation of past-time reference. It begins with the proposal that temporality be subsumed under the category of epistemic modality, understood as a degree of detachment from the content of the expressed proposition. We present arguments in support of this stance and refer to other arguments already aired in the linguistic and philosophical literature. We briefly summarise the contextualist theory of Default Semantics and apply it to the analysis of past-time expressions in Thai and English. We point out that the semantic representation of temporal expressions requires a perspective that combines the insights of cognitive approaches to meaning with a formal account of ‘pragmatic compositionality’. We focus on the Simple Past and Narrative Present in English and on the modal auxiliary d1ay1II in Thai that frequently assumes past-time reference. Keywords: past-time reference, epistemic modality, Thai, Default Semantics

1. The rationale and objectives This chapter concerns semantic representation of past-time reference. We begin with presenting our assumption that temporality is best thought of as a species of modality, understood as a degree of detachment from the content of the expressed proposition. We present some arguments in support of this stance and refer to other arguments already aired in the linguistic and philosophical literature. Next, we introduce the theory of Default Semantics in which the analysis of past-time/modal expressions will be conducted. We point out that the semantic representation of temporal expressions requires a perspective that combines the insights of cognitive approaches to meaning with a formal account of ‘pragmatic compositionality’. Finally, we apply this theory to the analysis of selected expressions with past-time reference in English and in Thai. In particular, we focus on the Simple Past and Narrative Present in English (called Past of Narration in that

 Katarzyna M. Jaszczolt and Jiranthara Srioutai

‘past’ refers here to temporality rather than tense) and on the modal auxiliary d1ay1II in Thai that frequently assumes past-time reference. The reason for the choice of these forms of expressing past-time reference in these two languages is as follows. Thai is a language in which the temporality of the situation presented in the utterance is frequently left for pragmatic inference. In other words, grammatical and lexical markers of tense and aspect are optional. Moreover, there are auxiliaries in Thai which, although they are frequently regarded as tense markers, are in effect markers of modality in that while they can assume different temporal meanings, they can be easily analyzed as markers of the speaker’s detachment from the situation presented in the sentence.1 We analyze the meaning of d1ay1II2 and demonstrate how the past-time reference can be pragmatically inferred. We describe the dependence of the meaning of d1ay1II on word order. Next, we introduce the problem of classifying d1ay1II as a past tense or a modal marker. The main argument of Section 5 of the chapter consists of the presentation of a unified account of d1ay1II as a modal marker whereby pastness is carried by its default interpretation. We conclude that a semantic framework in which pragmatic inference can be easily integrated is required for the analysis of (i) English sentences with past-time reference without overt markers of tense or aspect, as well as, and even more so, of (ii) Thai sentences with past-time reference in which marking of temporality is optional and often missing. Default Semantics proves to be an adequate theory to represent the meaning of such expressions in that various sources of information about meaning are distinguished and are treated with parity. Like Discourse Representation Theory (DRT, e.g. Kamp & Reyle 1993) from which it is derived, it combines the cognitive perspective on representing meaning with an aim to offer a formal account of how meaning is composed. Default Semantics proposes that compositionality is to be regarded as a pragmatic phenomenon and sought not on the level of linguistic semantics but rather on the level of the merger of information coming from diverse sources – an idea that is compatible with Jackendoff ’s conceptualist semantics (Jackendoff 2002; Culicover & Jackendoff 2005). But it goes beyond DRT in giving no priority to the syntactic structure. For the sake of comparison and contrast, before we move to the analysis of d1ay1II, we first discuss two types of expressions that can assume past-time reference 1. D1ay1II is regarded as temporal in Supanvanich (1973), as modal in Noss (1964), Muansuwan (2002), and Iwasaki & Ingkaphirom (2005), and as both temporal and modal in Kanchanawan (1978). 2. We use here the transliteration system from Diller (1996). In summary, vowel phonemes are transliterated as (high) i, u’, u, (mid) e, oe, o, and (low) ae, a, o’; three diphthongs are ia, u’a, and ua; long vowels are transliterated with a colon; the tone class of each syllable-initial consonant is specified by subscripts 1, 2, and 3; and tone markers are indicated by superscripts I and II.

Communicating about the past 

in English: the Simple Past, where the reference to a past interval or moment is fully grammaticalized in the form of the tense, and the Past of Narration, where the form that normally refers to the present time assumes past-time reference. We demonstrate how pragmatic inference allows the reference to the past to arise. We also briefly discuss how strictly modal expressions in English fit in the typology of past-time-referring constructions. 2. Time as modality The best place to begin discussing the structure and properties of time is McTaggart’s (1908) distinction between the A-and B-series. McTaggart (1908: 127) argues as follows: Why do we believe that events are to be distinguished as past, present, and future? I conceive that the belief arises from distinctions in our own experience. At any moment I have certain other perceptions, I have also the memory of certain other perceptions, and the anticipation of others again. The direct perception itself is a mental state qualitatively different from the memory of the anticipation of perceptions. (McTaggart 1908: 127)

McTaggart points out here the fact that the present moment has a special status in temporality: it allows us to go back into our memories or project forward with anticipation. Reality of the past and future does not necessarily have to be the case. Instead, we have two options: to think of time as (a) real, moving from the past through the present to the future (or alternatively events as moving from the future into the present and then into the past), or as (b) a way in which human mind organizes timeless reality: I shall speak of the series of positions running from the far past through the near past to the present, and then from the present to the near future and the far future, as the A series. The series of positions which runs from earlier to later I shall call the B-series. (McTaggart 1908: 111)

In short, the A series is tensed, time belongs to events, and there is no real change.3 One way of conceptualizing it would be as a stationary observer watching events that move in front of his/her eyes, as in Figure 1. The B series is untensed, events are related by means of the earlier-than/laterthan relations, there is no real change, and time is psychological: it is not real but rather belongs to the observer. For example, for Mellor (1998: 123), the experience 3. The main representatives of this view are Arthur Prior (e.g. 1957, 1967, 1968, 2003), Peter Ludlow (1999), Joshua Parsons (2002, 2003), and Quentin Smith (1993).

 Katarzyna M. Jaszczolt and Jiranthara Srioutai

Figure 1.â•‡ A possible conceptualization of the A series

Figure 2.â•‡ A possible conceptualization of the B series

of time consists of, and can be explained as, the accumulation of memories.4 One way of conceptualizing the B series is as an observer who moves between the timeless, real and stationary events as in Figure 2. The following analysis will be founded on the B-theory principles. In particular, we shall employ the following assumption for representing time in semantics: The categories of tense, aspect, and whole propositions are founded on mental representations of events, organized on non-temporal principles.

The idea that time is not the most basic epistemological category but rather is founded on some other categories is not new. Moens & Steedman (1988) and Steedman (1997) argue that temporality is supervenient on the perspective and contingency. They claim that tense and aspect systems in languages are founded on the same conceptual primitives as evidentiality. In a similar vein, van Lambalgen & Hamm (2005) say that goals, planning, and causation link the past with the present and the future. In general, “the linguistic coding of time is (...) driven by the future-oriented nature of our cognitive makeup” (van Lambalgen & Hamm 2005: 13). Next, in truth-conditional semantics, Parsons (2002, 2003) puts forward a counterfactual theory of tense: he develops a semantics in which truth-at-a-time

4. The main representatives of this position are Hans Reichenbach (1948), D. H. Mellor (1993, 1998), and also recently Thomas Sattig (2006) who argues that although humans think in terms of the A series, the A series is supervenient on the B series. In a wider intellectual domain, Albert Einstein and Bertrand Russell also belong to B-theorists.

Communicating about the past 

is replaced with truth-at-a-world. Expression (1) is then analyzed as a sum of (2) and (3), where times are bound to counterfactual situations:

(1) M was wholly future and will be wholly past. (2) There is some past time such that, were it that time, M would be wholly future. (3) There is some future time such that, were it that time, M would be wholly past. (adapted from Parsons 2002: 10)

For the purpose of this chapter we shall not engage with detailed discussions concerning the ontology of time but instead take the extant arguments as sufficient for postulating that time is not a primitive category but instead supervenes on something more basic. We shall assume that time is a degree of detachment from what, in timeless reality, would be affirmed with certainty. For example, to come back to the first quotation from McTaggart at the beginning of this section, time is a detachment from the possibility of present perception. For the purpose of semantic analysis, we shall conceive of time as a degree of detachment from the content of the expression uttered. That content can in principle be construed as a proposition, a sentence, a situation, or an event. In Section 3 we shall argue that event is the most adequate construct for depicting this content as an object of this detachment. We can now move to making this notion of detachment more precise. The best candidate for the semantic/pragmatic detachment as conveyed by means of the utterance itself is the semantic notion of epistemic modality, understood rather broadly, as for example in van der Auwera & Plungian (1998: 81–82), according to whom epistemic modality “refers to a judgment of the speaker: a proposition is judged to be uncertain or probable relative to some judgment(s).” The authors also point out that “...epistemic modality concerns (has scope over) the whole proposition” (1998: 82). This is an important characteristic of epistemic modality and it will allow us to represent it in our semantics as an operator that is akin to sentential operators in modal logic. We shall also follow van der Auwera & Plungian and assume that modality is a category compatible with evidentiality in that inferential evidentiality overlaps with epistemic necessity, as for example, the behavior of the English must + base verb form or must have + past participle demonstrate. The speaker looks at the watch and says:

(4) Tom must be in London by now. (5) Mary must have finished her piano practice by now.

A fortiori, we shall not be adopting more restricted, grammar-driven definitions of evidentiality such as Aikhenvald (2004) according to which evidentials are restricted to “the grammatical means of expressing information source” (2004: xi) and “[t] o be considered as an evidential, a morpheme has to have “source of information”

 Katarzyna M. Jaszczolt and Jiranthara Srioutai

as its core meaning; that is, the unmarked, or default interpretation” (2004: 3). For our purposes, evidentiality, when it is used to show detachment from the situation conveyed in the expressed proposition, is a sub-species of modality. In order to argue that time is supervenient on the category of detachment or modality one has to consider the future, the present, and the past. We discussed the first at length elsewhere, both for English (Jaszczolt 2003, 2005, 2006) and for Thai (Srioutai 2004a, b, 2006). We used the theory of Default Semantics (Jaszczolt 2005) in order to show that the English will and the Thai c1a are best analyzed as markers of modality. We are not alone with this claim. Enç (1996) convincingly demonstrated that will behaves like a modal in that the sequence of tenses does not apply and hence will does not behave analogously to the past-tense markers but rather seems to side with the modals (see Enç 1996: 350). Fleischman (1982), in her work on Romance languages, showed that the future is closely related to irrealis or nonfactive modality and deontic modality in that in the historical perspective we observe a bidirectional semantic shift: from clearly modal meaning of a form, to tense, and to modal coloring again. Similarly, van der Auwera & Plungian (1998) derive the future from the participant-external necessity and argue that the future form in turn acquires the meaning of epistemic necessity: the process they call a remodalization cycle. There are also ample arguments in the literature to the effect that futurity as an epistemological category. Ludlow (1999), for example, derives futurity from predictability and potentiality, or a disposition of the world. But while the future can be more readily conceived of as modal in that the future is to a large extent unknown and speakers normally talk about it expressing a greater or lesser degree of detachment, the claim that the present or the past are modal yields much less readily to common intuitions. While there are many possible ‘futures’, there is only one past. And yet there does not necessarily seem to be a qualitative difference between them. Just as the future, the past is not ‘here and now’, in front of our eyes. Evidence we have for it and memories we collect are permeated with a degree of detachment from what really was the case. That is why, in English, we can use a phrase with the past tense form ‘went to London’, as well as a weaker, modal equivalent ‘may have gone to London’, depending on the degree of certainty. The way to conceive of the past as modality has to proceed according to phenomenological principles, that is taking the past to mean memories of experiences, and hence more, or less, accurate and more, or less, epistemically reliable recollections. Another way to argue for pastness as modality is to look at counterfactuals, or similarity across possible worlds. Thomason (2002) develops a formal-semantic account of the past where pastness is modality in the sense of historical necessity: if two possible worlds are similar at a particular moment in time, then they share the same past up to and including that moment in time. He gives a formal account of

Communicating about the past 

the fact that what was a possibility in the future, becomes necessity in the past: historical possibilities diminish monotonically with the passage of time. We will share this view of the past as modality and in what follows will provide an analysis of some expressions with past-time reference in English and Thai where expressing time does not necessarily rely on grammatical and lexical markers of temporality, or, in other words, on tense and temporal adverbials. While the observation that expressing temporality does not require tense, aspect or temporal adverbials is not in itself new or controversial, we will follow it through and show that the fact that information about the time of the eventuality can be pragmatically inferred is not to be regarded as an anomaly or exception in semantic theory and can be easily accounted for in a semantic representation of utterances, as long as semantics is understood to be a ‘pragmatics-rich’ semantics, where information from inference, conventions, and other non-linguistic sources contributes to the representation of meaning. This semantics becomes a truth-conditional theory of utterance meaning. In the next section we introduce such a suitably ‘interactive’ view on the sources of information about meaning called Default Semantics. 3. Merger representations It has been widely accepted that the interpretation of the speaker’s utterance by the addressee has to rely both on the process of decoding and on inference. In addition to composing the meaning of the sentence out of the meanings of the lexical items and the structure in which these lexical items are placed, the hearer has to engage in inferential processes in order to enrich and supplement this meaning and arrive at the intended message. The exact interaction of this pragmatic inference with the output of syntactic processing varies from one account to another, and so do the assumptions as to the extent of application and the conscious/subconscious nature of these processes (see e.g. Recanati 2004). Nevertheless, the interaction of these two components is now common currency in post-Gricean pragmatics, founded on the ideas of the underdeterminacy of semantics, the pragmatic enrichment of the output of syntactic processing, and the truth-conditional relevance of the representation so enriched. Since the 1980s, truth-conditional semantics has been understood rather broadly and freely by Gricean pragmaticists, in that the truth conditions that are of interest in the analysis of utterance interpretation are not the truth conditions of the sentence, that is the unit that is physically uttered and processed according to the rules of syntax, but rather the truth conditions of the utterance, or speech act, that is the representation that is enriched as a result of pragmatic inference and better approximates the meaning as intended by the speaker (see e.g. Sperber and Wilson, e.g. 1986; Carston, e.g. 1988, 2002; Recanati,

 Katarzyna M. Jaszczolt and Jiranthara Srioutai

e.g. 1989, 2004). One of the best developed variants of this view is at present truthconditional pragmatics (Recanati 2003, 2004).5 According to this view, the truth value is predicated of an utterance or what is said by the speaker. Recanati argues that what is said is the only level at which meaning is consciously accessible to the interlocutors: going below this level is always subdoxastic. For example, in (6) and (7), the meanings that are available to the communicators in a particular context are those in (6a) and (7a) respectively. They result from the pragmatic enrichment of the output of syntactic processing.

(6) Everybody went to London. (7) I haven’t eaten.

(6) a. Everybody in this house went to London. (7) a. I haven’t eaten breakfast yet.

Truth-conditional pragmatics subscribes to contextualism, a standpoint according to which this pragmatic contribution is always present. In other words, utterances are always processed in context and this context affects their interpretation.6 In contextualism, “there is no level of meaning which is both (i) propositional (truth-evaluable) and (ii) minimalist, that is, unaffected by top-down factors” (Recanati 2004: 90). To compare, for quasi-contextualists, there is such a level of a minimal proposition but it does not play a role in utterance interpretation (see Recanati 2004: 86).7 However, it is by no means certain that such top-down factors always play a part. Be that as it may, what matters for our current purpose is the assumption that this level of pragmatically enriched meaning is salient in communication and we have to have an adequate theory to show how the interaction of the semantic and the pragmatic aspects of meaning takes place. Or, at least, one has to be able to propose a model of this interaction. Remaining in the overall orientation of semantic underdeterminacy, we now introduce the main theses of Default Semantics and discuss the process of construction of utterance meaning proposed in this theory. We distinguish the sources of information about utterance meaning and introduce the notion of a merger representation.

5. For developments in truth-conditional pragmatics that took place after this research was completed see e. g. Recanati (2010). 6. See Recanati (1994, 2004). 7. Recanati (2004: 83ff) also discusses the so-called literalist positions according to which the truth-conditional content can be ascribed to sentences rather than utterances or speech acts. We will not contribute to this debate in this chapter but will take contextualism, the view opposed to literalism, as our starting point.

Communicating about the past 

Default Semantics (henceforth DS)8 distinguishes four sources of information about utterance meaning: (1) word meaning and sentence structure (WS); (2) conscious pragmatic inference (CPI); (3) cognitive defaults (CD); and (4) social/cultural defaults (SCD). Information from these sources combines to produce the what is said as it is understood in DS, that is the utterance meaning intended by the speaker and recognized as so intended by a model hearer. On the DS account, what is said need not constitute a development of the logical form of the uttered sentence but instead is free from any syntactic constraints imposed by it. In order to avoid confusion, the label ‘what is said’ is not used in the theory in the technical sense but instead the discussion proceeds in terms of the representation of the main meaning that is normally intended by the speaker and recovered by the addressee. The representation of the output of this interaction is called merger representation. The model in Figure 3 represents the process of utterance interpretation. Stage I: Processing of the truth-conditional content combination of word meaning and sentence structure (WS)

compositional

conscious pragmatic inference1 (CPI1)

social/cultural defaults1(SCD1)

cognitive defaults (CD)

Stage II: Processing of implicatures – social/cultural defaults2 (SCD 2) – conscious pragmatic inference2 (CPI 2)

Figure 3.â•‡ Utterance interpretation in DS (adapted from Jaszczolt 2005: 73)

8. This article is based on the 2005 version of Default Semantics as it predates the revisions introduced to the theory. For the revised account of the sources of information see Jaszczolt (2009, 2010).

 Katarzyna M. Jaszczolt and Jiranthara Srioutai

Stage I is what will interest us in the current investigation as it represents the process of the recovery of the main meaning. The output of the syntactic processing of the sentence is represented as WS. This output can interact with the result of pragmatic inference (CPI 1) as in (8), producing the meaning in (8a). (8) Anne missed the train and was late for work. a. Anne missed the train and as a result was late for work. The juxtaposition of the simple sentences about the two events: Anne’s missing the train and Anne’s being late for work produces the effect of the strengthening of the ordinary, truth-functional conjunction and to the relation of consequence: and as a result.9 In addition to WS and CPI 1, there are two more sources that contribute to the merger representation. They are shortcuts through pragmatic inference that occur either due to the properties of mental operations or due to the shared cultural and social knowledge of the interlocutors. The first one is called a cognitive default (CD), and the latter a social/cultural default (SCD 1). Cognitive defaults explain the reason behind the salience of, say, referential interpretations of definite descriptions such as ‘the author of The Man without Qualities’ in (9). It is intuitively plausible that the addressee B is likely to react to A’s statement by posing a question that assumes that A knows who the author of the novel was. (9) a. The author of The Man without Qualities was a very good writer. b. Really? Who is it? We can also think of a situation in which speaker A talks about whoever happened to write this novel, without knowing, or remembering, the identity of the person. In yet another scenario, speaker A is referentially mistaken and thinks of, say, Marcel Proust (instead of the correct Robert Musil) as the author of the novel. Now, the preferred, most typical use of definite descriptions is the one on which both the speaker and the hearer share the assumption that the referent is mutually known. In other words, it is assumed that the use of the definite description is referential: in (9), it is about Robert Musil. It can be safely assumed that such an assumption is not consciously processed. Instead, it comes as a ‘shortcut’ through the inferential process, as a default meaning founded on the property of human mental states called their intentionality or aboutness. Mental states such as believing, doubting, or fearing have an object, are about something, and this aboutness normally amounts to knowing the identity of the object thought of: a belief about the author of The Man without Qualities normally contains as much information 9. For a detailed discussion, see Jaszczolt (2005, Chapter 8.1.1).

Communicating about the past 

as the definite noun phrase allows it to contain. We assume that the strongest intentionality is the norm in that it is not weakened by the speaker’s lack of knowledge, misinformation, lack of attention, and other impeding circumstances. The strongest intentionality is the default intentionality and pertains to the referential, correct reading of the description where it stands for Robert Musil. So, the explanation is simple: the default interpretation is an interpretation on which intentionality is the strongest. This type of default is an example of what we call the cognitive default. Needless to say, if intentionality of mental states allows for degrees of strength, then so does the speaker’s intending that accompanies his or her utterance. Normally, the strongest intentionality of a mental state corresponds to the strongest referring by the act of communication that externalizes this mental state. In other words, if A’s belief in (9) is about Robert Musil, then A’s utterance in (9) is also about Robert Musil. The strongest intentionality and the strongest referential intention go hand in hand.10 Cognitive defaults pertain to ample types of language constructions and phenomena. They explain the use of temporal expressions, the construction of anaphoric dependencies, and the use of number terms.11 But they cannot be extended as an explanatory tool to all cases of default meanings. Some defaults are of a very different provenance: they can come from social practices or cultural heritage. It is not difficult to produce examples of salient interpretations that arise due to some cultural or social stereotypes or cultural or social knowledge. In example (10), it is the shared cultural knowledge that almost invariably produces the interpretation (10a) in most (reasonably educated) speakers within the western culture. (10) Leonardo’s women all look similar. a. The women painted by Leonardo da Vinci all look similar. Cultural knowledge allows the addressee to identify ‘Leonardo’ as Leonardo da Vinci, and the possessive as authorship rather than, say, ownership. Similarly, in (11), the salient interpretation (11a) arises due to the shared knowledge that in the relevant society babies are raised by their own parents. (11) The baby cried and the father rocked the cradle. a. The baby cried and the baby’s father rocked the cradle. Following the theory of DS, we tentatively assume that these defaults are normally non-inferential, unless there is evidence that would make us assume otherwise. 10. The argument can also be restated in terms of communicative, informative and referential intentions, or Bach and Harnish’s (1979: 7) communicative-illocutionary intention. See Jaszczolt (2005: 51–52). 11. All of these are discussed in detail in (Jaszczolt 2005, Part II).

 Katarzyna M. Jaszczolt and Jiranthara Srioutai

The interpretation is then automatic, instantaneous and unreflective, in agreement with Recanati’s idea of subdoxastic enrichment.12 It has to be pointed out that in DS defaults are understood as defaults for the speaker, addressee and situation of discourse rather than being predicated of a word of structure tout court. In other words, they are preferred meanings for the situation. They are also considered, as a matter of a methodological assumption, as post-propositional: since at present we have no satisfactory answer to the question as to at what stage exactly default enrichment takes place for various constructions, it is the ‘safe bet’ to assume that they arise after the whole utterance has been processed. A concomitant feature of this post-propositionality is that they are rarely cancelled.13 Assuming that such enrichments are uniformly ‘local’ leads to wrong predictions, as is evident from the experimental testing of Levinson’s (2000) presumptive meanings for example in Noveck & Sperber (2004). Now, it has to be remembered that so far we have only plausible hypotheses, theoretical arguments, and very preliminary attempts at experimental evidence in support of the model proposed in Figure 3. What seems to be certain, though, is that a model on which various sources of meaning information interact is in principle correct. It is also certain that there are shortcuts through conscious processing of information coming from these sources. We recognize two types of such shortcuts that we call default interpretations. In the CD type, one can discern very clear alternative readings, such as referential and attributive readings of descriptions, inclusive or exclusive readings of disjunction, or de re and de dicto readings of propositional attitude reports. On the contrary, in the SCD 1 type, enrichments are culturally and socially salient but they merely give a more finely-grained picture of the situation rather than clear alternatives. Social/cultural defaults are more difficult to pinpoint and to formalize in the theory of meaning. It is not clear how much detail one has to add to the basic meaning of the lexical item or the construction within the confines of the propositional content. In other words, the granularity of SCDs is unclear. For example, ‘nanny’ in (12) may be interpreted as a female nanny or it may be simply left as a concept of a nanny of whatever sex. On the other hand, it may also be interpreted as a female nanny who is young, pretty, musical, strict, or lovable. (12) We advertised for a new nanny. As a result, the boundary between SCD 1 and CPI 1 is not clear-cut since we have no independent evidence concerning the presence or lack of an inferential 12. See Recanati (2004: 38). Incidentally, these defaults are called by Levinson (2000) ‘inferences to a stereotype’! 13. But see, for example, (23) in Section 6.

Communicating about the past 

process. To sum up, while cognitive defaults are well motivated by their property of strong, undispersed intentionality, social/cultural defaults have no such characteristic property to recommend them. They are only motivated by the methodological requirement not to postulate inferential processes beyond necessity. They are hardly distinguishable from cases of conscious pragmatic inference, they are simply on the polar end of an ‘inference + salience’ cline. As is observed in Jaszczolt (2005: 56), one can only assume the boundary between such social/cultural defaults and social/cultural inference. We do not know where exactly to place it but we want to retain the notion of it in the theory because such a distinction is methodologically desirable and psychologically plausible. In what follows, we will have little to say about SCD 1. CDs, on the other hand, will prove to play an important role in the interpretation of utterances as past or modal. The main objective of DS is to construct a plausible algorithm of how all the sources of meaning information indicated by the arrows in Figure 3 interact. To do so, it questions the assumed priority of the WS source, i.e. the combination of word meaning and sentence structure. It allows more power to the sources CPI 1, SCD 1 and CD, which better reflects their interaction with WS than the current truth-conditional, pragmatics-rich semantics does. But this comes at the price of relaxing the boundary between what is said and what is implicated and abandoning the development of the logical form as a defining characteristic of what is said. In other words, the main meaning represented in DS may not always be a development of the logical form of the sentence, but, on some (rare) occasions, it can have an altogether different form. Using Bach’s (1994) example, (13) can now have, say, (13a) as its primary utterance meaning, rather than (13b). (13) Mother to a little boy, crying over a cut finger: “You are not going to die.” a. You shouldn’t worry. b. You are not going to die from this wound. This is a big step, but, it seems, a feasible one. The main meaning modeled in DS need not be partially isomorphic with the linguistic meaning. Pace post-Gricean attempts to draw the boundary, best summarized in Carston (1988) and Recanati (1989), there is no compelling reason to restrict the main modeled meaning to the developments of the logical form provided by WS (to repeat, we will try to avoid the term what is said in order to avoid confusion). The additional strength of this reanalysis is that we obtain a more promising notion of compositionality. We only require the merger representation, rather than the output of grammar (WS), to be compositional. This proposal is compatible with, and can be regarded as an execution of, Recanati’s observation:

 Katarzyna M. Jaszczolt and Jiranthara Srioutai

... the semantics of natural language is not insulationist. (...) [T]he meaning of the whole is not constructed in a purely bottom-up manner from the meanings of the parts. The meaning of the whole is influenced by top-down, pragmatic factors, and through the meaning of the whole the meanings of the parts are also affected. So we need a more ‘interactionist’ or even ‘Gestaltist’ approach to compositionality. (Recanati 2004: 132)14

The proof will lie in constructing an algorithm for this interactive model of meaning and in having a formalized, compositional theory of such mergers.15 The proposed framework attempts to fulfill two main objectives of a theory of discourse meaning, namely providing plausible mental representations of discourse backed by a more formal account of the compositional semantic structure. By locating compositionality on the level of the merger of information about meaning that comes from qualitatively different sources it dissociates formalization from what we can call a ‘syntactic constraint’ (see Sysoeva & Jaszczolt 2007): the representation of the primary, most salient meaning in a discourse need not constitute a development of the logical form of the uttered sentence but may in some cases depart from it and override it, as in example (13a) above.16 Like its parent theory DRT (Kamp & Reyle 1993), it sees formalization as subordinate to the main goal of providing mental representations of discourse. Like DRT, it would subscribe to Jackendoff ’s view that a form of conceptual semantics is a superordinate objective and it only makes use of, rather than being a slave to, formal methods (Jackendoff 2002; Culicover & Jackendoff 2005; Hamm et al 2006): DRT’s claim that there are features of natural language the analysis of which requires a distinct level of discourse representation is consonant with a cognitive perspective on the nature of natural language meaning: Meaning in natural language manifests itself as the semantic competence of the language user... (Hamm, Kamp & van Lambalgen 2006: 5–6)

Freeing representations of discourse meaning (merger representations) from the syntactic constraint, DS makes a further step in the direction of the cognitive perspective. As a result, temporality on the level of merger representation is freed from the constraints on tense and aspect, although, on the level of WS, tense and aspect perform an important role as sources of information. These concepts belong to the sentence level, while in DS all sources of meaning are equally important and therefore merger representation may on occasion override their message. 14. Our emphasis. 15. This formalization for a fragment of English was begun in Jaszczolt (2005), using an extended and amended language of Discourse Representation Theory. 16. This proposal has ample experimental support, see e.g. Nicolle & Clark (1999), Pitts (2005), and Sysoeva & Jaszczolt (2007).

Communicating about the past 

We demonstrate below that this framework is particularly well suited to representing temporality in languages in which there is no formal indication of time on the sentence level. In DS, we start with the assumption that time is a form of modality in that it constitutes a degree of detachment from what would otherwise be ‘timeless truth’ and ‘certainty’. An operator ACC is used to represent modality. ACC, short for acceptability, is an operator on events and states that allows for degrees of strength. ACCΔn reads as ‘it is acceptable, to the degree n, that it is the case that...’ where the object on which it operates is an event (e) or state (s), which are theoretical constructs defined as finely-grained (cf. Kim 1976) and time-independent entities (cf. event types of van Lambalgen & Hamm 2005).17 ACC is loosely modeled on the acceptability operator proposed by Grice (2001). Grice attempted a unified account of practical and alethic modality by means of a sentential modal operator Acc p: ‘it is rationally acceptable that p’. For alethic modalities, he proposed Acc p (‘it is acceptable that it is the case that p’). ACC, our operator on events and states, is intended as a device that fulfils the same purpose as the sentential operator Acc, of providing a unified treatment of modalities, understood broadly as discussed in Section 2. We have extended its use from alethic and practical modality of Grice’s account, to epistemic, deontic and metaphysical modalities as distinguished in our typology.18 This move is fully compatible with Grice’s overall objective of using Acc as an operator by means of which modality in general is explicated. All of the uses are regarded as degrees of certainty to which the speaker of the utterance can subscribe while uttering the sentence. DS uses an adapted and extended language of DRT (Kamp & Reyle 1993; van Eijck & Kamp 1997) and applies it to the merger representations of Figure 3, that is to the merged output of the sources of meaning WS, CPI 1, SCD 1, and CD. One of such extensions is the introduction of the operator ACC. The main adaptation is the use of the language for modeling meaning that derives from a variety of sources, where the grammatical form of the uttered expression is treated on a par with, and hence can be overridden by, the output of any of the other sources, as was discussed in example (13) above. This view will allow us to represent the cases where there is a mismatch between the tense of the sentence and the temporality expressed by the utterance of this sentence, such as the narrative present (called here past of narration) that uses present verb forms for expressing the past.

17. The notion of event used in Default Semantics is described at length in Jaszczolt (2009). 18. And their sub-categories, such as inferential subsumed under epistemic.

 Katarzyna M. Jaszczolt and Jiranthara Srioutai

4. The modality of the past: Evidence from English Merger representations that make use of ACC clearly represent the shared perception of (14) as stronger than (15), and (15) in turn as stronger than (16) with respect to the commitment to the eventuality described in the sentence. (14) (15) (16)

Tom went to London yesterday. (simple past) Tom would have gone to London by then. (epistemic necessity past/inferential evidentiality) Tom may have gone to London yesterday. (epistemic possibility past)

On the other hand, the use of the past of narration as in (17) seems to signal the degree of commitment that is similar to that in (14). (17) This is what happened yesterday. Tom goes to London, meets Mary at the station, and says... (past of narration) We can represent this gradation as a gradation of the informative intention with which the utterance was uttered. The cline will then approximate the one presented in Figure 4, with the proviso that other ways of expressing the past are also possible in English and would have to be included in order to obtain a full typology. Epistemic necessity past and epistemic possibility past would also have to be further spelled out in order to show the variety of constructions they subsume – and, most importantly, a variety of degrees of intention and modality that is represented within each supercategory. We also have to add a disclaimer to this figure in that the placement of the types of expressions on it is only relative. The absolute, or, even, numerical values ascribed to the forms would have to rely on extensive empirical evidence and be determined by clear criteria such as frequency of use, comparison with the main use to which the particular form is put, and so forth. SP stands for ‘simple past’, PN for ‘past of narration’, INP for ‘inferential necessity past’, and EPP for ‘epistemic possibility past’. Similarly, we can represent the relative degree of detachment from the situation described in the proposition expressed, or, in other words, the degree of modality. The strongest informative intention (the strongest commitment) from Figure 4 will now correspond to the weakest modality. Or, in other words, the greatest SP, PN

INP

1

Figure 4.â•‡ The past: Degree of informative intention

EPP 0

Communicating about the past  EPP

INP

1

SP, PN 0

Figure 5.â•‡ The past: Degree of modality

detachment (modality) corresponds to the weakest commitment, and hence the scale will have to be reversed as in Figure 5. All these types of expressing the past can easily be represented in merger representations of DS. To repeat, DS models the merger of information coming from a variety of sources and the representations are not constrained by the grammatical structure of the uttered sentence. For analyzing the semantics of temporal expressions, this feature of the theory has obvious advantages in that the use of a simple present form to express futurity or pastness does not pose a difficulty. It is considered to be a departure from the default way of using the form but this departure can be accommodated through the interaction of WS and CPI, as in Figure 6. Figure 6 is a merger representation for a fragment of (17): ‘Tom goes to London’, with past-time reference. In Figure 6, x and e stand for discourse referents (individual and event respectively), and square brackets mark the material that comes from the source of information indicated by the subscript. In the semantics (see Jaszczolt 2005), Δ was defined as epistemic: Δ = , and the superscript pn stands for ‘past of narration’. In other words, the degree of ACC is represented as that characteristic of pn. The condition [ACCΔpn e]WS, CPI 1, for example, has to be read as ‘It is acceptable to the degree pn that it is the case that e’, where the sources of information are word meaning and sentence structure (WS), that is the structure of the uttered sentence, as well as conscious pragmatic inference1 (CPI 1, see Figure 3) that operates on this output of the lexicon and grammar and ensures that the present tense verb form is understood as used for past-time reference. By comparison, the expression of the past by means of the past tense form of the verb is guaranteed by the grammatical form alone: the default use of simple

xe [Tom]cd (x) [ACCΔpn e]ws, cpi 1 e:

[x go to London]ws

Figure 6.â•‡ Past of narration: Merger representation for ‘Tom goes to London’

 Katarzyna M. Jaszczolt and Jiranthara Srioutai

[Tom]cd (x) yesterday (t)

xte

[ACCΔsp e]ws, cd e:

[x go to London]ws

Figure 7.â•‡ Simple past: Merger representation for (14), ‘Tom went to London yesterday’

past is, naturally, to convey past-time reference. The meaning of sentence (14)19 is given by the merger representation in Figure 7. The subscript CD (‘cognitive default’) signals that ACC of the degree ‘sp’ (‘simple past’) is used in its default sense. We can tentatively conclude on the basis of the analyses of these two examples of a standard and non-standard (CD and CPI) ways of conveying the past in English that DS provides a framework for a unified account for various degrees of modal detachment, whether they are overtly encoded in the lexicon or grammar or conveyed by pragmatic means. The modal operator ACC accounts for them all, rendering them as degrees of departure from full certainty about the event or state at stake. The analysis of the ways of expressing past-time reference in English, combined with the different degrees of modal detachment they can convey, exemplifies the advantages of a theory of meaning that accounts for the interaction of the grammatical and lexical properties of the words and sentences used with the information about meaning that comes from other sources, such as the awareness of the typical, presumed interpretation and the processing of the context and background knowledge. The notion of compositionality is amended accordingly, to provide for this interaction of the outputs of various sources of information about meaning.20 In the following sections we apply this theory to a language in which expressing temporality and modality is much more reliant on pragmatic processing than it is in English. In Thai, formal indicators of time such as markers of tense and aspect are largely optional. Relying on word meaning and sentence structure alone would result in a multitude of possible translations of a Thai sentence into English. In the remainder of this chapter, we concentrate on the marker d1ay1II in Thai, which can be used as an indicator of past tense or modality. We present a 19. Or, strictly speaking, the utterance of sentence (14) in that merger representations model the meaning of utterances. 20. We have accepted that events are arguments of predications. Next, the relational semantics can be built by analogy to that for predication, as was presented in Jaszczolt (2005: 172).

Communicating about the past 

unified analysis of d1ay1II as a modal marker, accounting for its default use (in the sense discussed above, as the contextually preferred sense) as a past-tense indicator and its other non-default modal uses by utilizing the interaction of the sources of meaning information that contribute to the merger representation. 5. Possible uses of d1ay1II In Thai, d1ay1II can have both lexical and functional meanings. It may appear as the only verb in an utterance, where its lexical meaning is ‘receive’, as shown in (18). (18) m3ae:r3i:I d1ay1II c1otm3a:y Mary receive letter ‘Mary received the letter’21 D1ay1II may also behave like an auxiliary. It is this use that is of interest to us in this chapter. It can either immediately precede or immediately follow the verb phrase. In this section, we demonstrate that the semantics of the auxiliary-like marker d1ay1II is sensitive to its verb-initial or verb-final position. When d1ay1II precedes the verb phrase, the sentence can be interpreted in two ways, as shown in (19). ‘Gremlin’ is the name of a cat. (19)

k1r3eml3in d1ay1II c1ap ng3u: Gremlin d1ay1II catch snake a. ‘Gremlin was able to catch a snake (and he caught it)’ b. ‘Gremlin had an opportunity to catch a snake’

The two readings differ in the following ways. In (19a), Gremlin is described as having had an opportunity to catch a snake and having caught one. However, the eventuality of Gremlin catching a snake might not have happened as far as (19b) is concerned. When d1ay1II immediately follows the verb phrase, as in (20), the utterance has two possible interpretations. Note that although the English translations for (19a) and (20a) are identical, the latter is different from the former in that it refers to Gremlin’s physical, mental and circumstantial ability to catch a snake, rather than being restricted to Gremlin’s opportunity, as (19a) is. On the other hand, (20a) indicates that Gremlin did catch a snake while (20b) only speaks of its snake-catching ability, in which case the meaning of deontic or dynamic possibility of d1ay1II is clear. 21. The translation given is the most natural reading in terms of tenses. Other temporal interpretations are also possible, e.g. ‘Mary has received the letter’, ‘Mary is receiving the letter’ and ‘Mary is going to receive the letter’.

 Katarzyna M. Jaszczolt and Jiranthara Srioutai

(20)

k1r3eml3in c1ap ng3u: d1ay1II Gremlin catch snake d1ay1II a. ‘Gremlin was able to catch a snake (and he caught it)’ b. ‘Gremlin can catch a snake (if he wants to)’

It can be seen that the verb-initial or verb-final position of d1ay1II brings about a slight change in its meaning. When d1ay1II occurs immediately before the verb phrase, its meaning focuses on circumstantial ability, or opportunity. When it immediately follows the verb phrase, it is not limited to circumstantial ability but includes physical and mental powers. This semantics of d1ay1II, where its meaning is conditional on its position immediately before or after the verb phrase, is also the case when the proposition is negated, where two interpretations are again possible, as (21) and (22) illustrate. (21)

k1r3eml3in m3ai1I d1ay1II c1ap ng3u: Gremlin neg d1ay1II catch snake a. ‘Gremlin was not able to catch a snake’ b. ‘Gremlin did not have the opportunity to catch a snake’

(22)

k1r3eml3in c1ap ng3u: m3ai1I d1ay1II Gremlin catch snake neg d1ay1II a. ‘Gremlin was not able to catch a snake’ b. ‘Gremlin cannot catch a snake’

While (21) aims to describe Gremlin’s inability to catch a snake in terms of lack of opportunity, only the senses of physical and mental ability of d1ay1II are present in (22). This explains the difference between (21a) and (22a), despite their indistinguishable English translation. What (21a) communicates is the eventuality in which Gremlin was not able to catch a snake because he did not have the opportunity to do so, that is, he did not try to catch a snake at all. But in (22a), Gremlin is depicted as attempting to catch a snake but failing to catch one. Taking into account all of its possible uses mentioned above, our review of previous treatments of d1ay1II, whereby it is analyzed as either a past tense or a modal marker, demonstrates that they do not offer a comprehensive account. In Kanchanawan’s (1978) and Supanvanich’s (1973) approaches, d1ay1II, when immediately preceding the verb phrase, is classified as a past tense marker. In the verbfinal position, Kanchanawan (1978) identifies it as encoding ability or permission. This is in line with Maunsuwan’s (2002) framework, where d1ay1II in the verb-final position is considered similar to the modal verb can in English, which expresses either dynamic modality (i.e. ability) or deontic modality (i.e. permission). In Supanvanich (1973), where only the verb-initial position of d1ay1II is examined and d1ay1II is specified as a tense marker, which may be applicable to (19a) and

Communicating about the past 

(21a), its clearly modal meaning, as in (19b) and (21b), is left unanalyzed. In Muansuwan (2002), where only d1ay1II in the verb-final position is analyzed as a modal marker, its temporal meaning, which seems to be present in (20a) and (22a), is unaccounted for. While Kanchanawan (1978) is apparently the most explanatorily adequate among the three accounts as both verb-initial and verb-final positions of d1ay1II are explored, d1ay1II receives different treatments in different positions. According to Grice’s (1978) Modified Occam’s Razor, which states that senses (linguistic meanings) are not to be multiplied beyond necessity, it is methodologically preferable to come up with a unified account of d1ay1II I both in the verb-initial and verb-final positions. Such an analysis is proposed below. 6. A unified account of d1ay1II In our approach, d1ay1II is classified as a modal marker that comes by default with the past time interpretation. On this analysis both the modal and temporal readings of d1ay1II are accounted for. It is shown below that d1ay1II (i) has modal meaning in both the verb-initial and verb-final positions and (ii) does not encode the past time reference. Pastness is simply its cognitive default, as defined in DS, in that it pertains to stronger speaker intention and at the same time stronger intentionality of the mental state that corresponds to the utterance.22 To repeat, a cognitive default is a salient interpretation that arises without a need for conscious pragmatic inference. In the verb-initial position, as in (19), d1ay1II has the dynamic modality meaning in the sense of opportunity. This analysis is supported by Palmer’s (2002) and Austin’s (1961) observations about this type of modality. According to Palmer (2002), dynamic modality has to be interpreted not only in terms of the subject’s physical and mental powers but also circumstances that affect the subject. Similarly, Austin (1961) contends that to say ‘he can’ in the full sense is to say not only that he has the ability but also that he has the opportunity, and that dynamic can may be used in the restricted senses of merely ability or opportunity. Of the two interpretations of (19), (19a) is more informative than (19b) as it refers to an eventuality where a snake was caught. In terms of DS, it can be said that (19a) is of stronger communicative, informative and referential intentions. In other words, (19a) is the default reading of (19). As for when d1ay1II occurs immediately after the verb phrase, as in (20), the modal meaning it expresses is either deontic or dynamic possibility. Interpretation (20a) is more informative as it refers to an eventuality of Gremlin catching a snake before the utterance time while (20b) 22. For the role of the degree of intentionality in cognitive defaults, see also Jaszczolt (2005).

 Katarzyna M. Jaszczolt and Jiranthara Srioutai

simply refers to Gremlin’s ability to catch one. Like (19), (20) has (20a) as its default interpretation. From (19) and (20), it can be seen that their default interpretation (19a and 20a) is the one that comes with the past time reference and entails the non-default sense (19b and 20b). Comparing (19) and (20), it can be clearly seen that default interpretations are not ‘localized’ to a particular part of the uttered sentence. In (19), where d1ay1II occurs immediately before the verb phrase, in the middle of the utterance, both interpretations have the past time reference. But when d1ay1II immediately follows the verb phrase and is sentence-final, as in (20), both past and present readings are accessible. That is, defaults are neither lexical nor local. Strictly speaking, the marker d1ay1II itself does not give rise to the past time reference by default. If that were the case, d1ay1II in sentence-medial and sentence-final positions would both trigger pastness and the present time interpretation would not be allowed. We now turn to give evidence from the corpus of data to verify that the temporal meaning of d1ay1II is its default and not encoded meaning. The background of this conversation is as follows. Gremlin is a cat with black paws. Speaker A was told that cats with black paws could catch snakes and wanted to convey this information to speaker B by saying (23a). (23a) is exactly the same as (20) above. That is to say, (23a) has two interpretations. D1ay1II can refer to dynamic modality, or ability, as in (23aii), while in (23ai), it gives rise to the past time reference. It was claimed earlier that (23ai) is the default meaning of (23) because it conveys stronger communicative, informative and referential intentions. In other words, (23ai) is more informative than (23aii) in that the speaker is understood to refer to Gremlin being able to catch a snake before the utterance time and catching one. But (23aii) does not refer to any such particular eventuality, only the fact that Gremlin has an ability to catch a snake. (23) a. A: k1r3eml3in c1ap ng3u: d1ay1II Gremlin catch snake d1ay1II i. ‘Gremlin was able to catch a snake (and he caught it)’ [B’s recovered meaning] ii. ‘Gremlin can catch a snake (if he wants to)’ [A’s intended meaning] b. B: n3ai1 ng3u: where snake ‘Where is the snake?’ c. A: k1r3eml3in m3ai1I d1ay1II c1ap ng3u: Gremlin neg d1ay1II catch snake ‘Gremlin did not have the opportunity to catch a snake’

Communicating about the past 

d. A: k1r3eml3in s2a:m3a:t c1ap ng3u: d1ay1II Gremlin can catch snake d1ay1II ‘Gremlin can catch a snake (if he wants to)’ We know from the background information that A’s intended meaning was (23aii). However, Speaker B’s answer in (23b) clearly shows that she interpreted (23a) as (23ai). This in turn shows that (23ai) is a more salient interpretation. The hearer thought the speaker had this interpretation in mind. Next, (23c), which means Gremlin did not have the opportunity to catch a snake, demonstrates that when the proposition is negated, the past time interpretation of d1ay1II is also possible and salient. Lastly, (23d) illustrates that d1ay1II may follow a verb phrase beginning with the modal s2a:m3a:t ‘can’. In the presence of the modal, the past time reference, usually triggered by d1ay1II, does not arise; the temporal interpretation is the time of the utterance. To conclude, while (23abc) show that the past time reference of utterances containing d1ay1II is more salient, (23d) suggests that the temporal meaning is unlikely to be part of its encoded interpretation. The solution provided by DS is to regard the past temporal sense, possibly prompted in the presence of d1ay1II, as the default reading of an utterance containing the marker. Another piece of evidence to support the claim that the past time reference of an utterance containing d1ay1II is not grammatically encoded comes from a real scenario discussed below. Srioutai was Speaker A in the conversation in (23). When she uttered (23a), she had only one interpretation in mind, namely, (23aii), and at that moment did not think about (23ai), the other interpretation, with the temporal default, as an alternative reading of (23). Just when Speaker B asked where the snake was, she realized that there was another possible meaning to the utterance. This goes to show that sometimes the past temporal interpretation of an utterance containing d1ay1II does not arise. Unlike other default accounts where a default reading is said to arise invariably and then undergo cancellation in certain contexts, DS asserts that defaults are situation-based salient meanings. The proposition that the past temporal reading of an utterance containing d1ay1II does not arise in certain contexts also corroborates the claim that the past time reference is not part of the linguistically encoded meaning of d1ay1II. This is why the analysis of d1ay1II as a past time or past tense marker is not a convincing one. However, from the point of view of the hearer, the conversation in (23) represents a rare situation where a default interpretation is retrieved and cancelled. The temporal default of pastness in (23a), captured in (23ai) and recovered by the hearer, i.e. Speaker B, is explicitly withdrawn by the speaker’s denial in (23c). In this particular case, the cancellation is a result of the mismatch between what was intended (what it was in reality) and what was recovered. Typically, cognitive defaults are intended and recovered by the model speaker and hearer; or else, they do not arise.

 Katarzyna M. Jaszczolt and Jiranthara Srioutai

There is still another piece of evidence to support the view that the temporal reading of d1ay1II is not its encoded meaning. Speaker A, who is a competent native speaker of Thai, used (23a) instead of (23d) to convey her intended meaning of dynamic modality. This was the case despite that fact that (23a) could have caused misunderstanding, which it did, and (23d) would have been more effective. This is interesting although it may seem unreasonable to use a d1ay1II construction in a non-default sense. It precisely illustrates that more than one possible interpretation is available and that the past temporal meaning of (23a), captured in (23ai) and recovered by Speaker B, is not always there. If it were, it would be the encoded meaning, not the pragmatic one. Instead, it is pragmatic and we propose that it comes from the source CD of DS in virtue of its strong degree of intentionality (aboutness). In other words, referring to past events conveys more information, pertains to facts, or, in a phenomenological parlance, is more strongly about experiences. Now we will demonstrate how merger representations may be constructed for utterances containing the marker d1ay1II. To recall, a merger representation is a meaning representation in DS that a model hearer is predicted to create in the process of utterance interpretation. It is where the four sources of meaning, sentence structure and word meaning (SW), cognitive default (CD), social/cultural default (SCD 1) and conscious pragmatic inference (CPI 1), are treated on an equal footing. This means that no priority is given to any of these sources of information. Merger representations will be constructed below for d1ay1II utterances in their default interpretation of pastness and in their non-default sense. In this latter circumstance, the interpretation of an utterance with d1ay1II is said to be a departure from the default. The departure from the default in d1ay1II’s case makes use only of its bare encoded meaning without temporality, which is dynamic modality. Let us look at the default case first. Figure 8 shows the merger representation of (23a) when the utterance has its default reading of the past time reference triggered by d1ay1II. In the merger representation in Figure 8 the discourse referent x refers to Gremlin, t to the eventuality time; n is now or the utterance time, and e the eventuality. The subscripts CD and WS indicate the source of the material in the square brackets. The second line reads: the proper name Gremlin by cognitive default refers to an individual x. The modal meaning of d1ay1II is shown by the modal operator ACC. ACC∆ e is to be read as ‘it is acceptable to the speaker that it is the case that e’. The type of modality expressed by d1ay1II in this example is dynamic modality, which may be subsumed under epistemic (see Biber et al. 1999: 465). It can thus be symbolized by ‘it is the case that’, the symbol used by Grice for alethic, and extended by Jaszczolt (2005) to epistemic modality, which is a related, but logically weaker, so to speak, type of modality than alethic. This information is signaled by the presence of d1ay1II in the utterance, which explains the presence of

Communicating about the past 

xtne [k1r3eml3in]cd (x) [ACCΔe]ws [Δ = ˇ]ws [t < n]ws, cd e:

[x c1ap ng3u:]ws

Figure 8.â•‡ The default meaning of an utterance k1r3eml3in c1ap ng3u: d1ay1II (Gremlin catch snake d1ay1II; ‘Gremlin was able to catch a snake (and he caught it)’)23

WS. Also, d1ay1II, by cognitive default, gives rise to the past time reference. This is rendered as ‘t < n’, or ‘the eventuality time t precedes the utterance time n’. The temporal default t < n is specified to be the output of WS and CD. We now turn to a departure from the default meaning of utterances with d1ay1II, which is its bare encoded meaning of modality with no temporality, as shown in (23aii). The merger representation of (23aii), where the past temporal default of d1ay1II does not arise, is provided in Figure 9. Similarly, discourse referent x refers to Gremlin, t to the eventuality time; n is now or the utterance time, and e the eventuality. The proper name Gremlin by cognitive default refers to an individual x. The modal meaning of d1ay1II is shown by ACC∆ e, where ∆ is specified to be of the dynamic modality type by , the same symbol Grice (2001) and Jaszczolt (2005) use for alethic/epistemic. This information is signaled by the presence of d1ay1II in the utterance, which is indicated by WS. The eventuality time is at the same time as the utterance time, so it is expressed as ‘t = n’, or ‘the eventuality time t equals the utterance time n’. This eventuality time is given by WS and CPI 1. That is to say, the present time reference, t = n, is the output of the sources WS and CPI 1. It is represented as a departure from the cognitive default of d1ay1II. WS accounts for the dynamic modality of d1ay1II while CPI 1 produces the inference to the dynamic possibility at present from the WS of d1ay1II and blocks the possible past temporal default of the utterance. 23. Note that the MR of a Thai utterance lacks formal representations of tense such as sp (see Figure 7), reflecting that, unlike in English, there is no grammatical indication of temporality in the language. Instead, ∆ has to be specified for the type of modality by to capture the meaning of the modal marker d1ay1II.

 Katarzyna M. Jaszczolt and Jiranthara Srioutai

xtne [k1r3eml3in]cd (x) [ACCΔe]ws [Δ = ˇ]ws [t = n]ws, cpi 1 e:

[x c1ap ng3u:]ws

Figure 9.â•‡ A departure from the default of an utterance k1r3eml3in c1ap ng3u: d1ay1II (Gremlin catch snake d1ay1II; ‘Gremlin can catch a snake’)

Lastly, Figure 10 illustrates a circumstance in which the temporal default of pastness of an utterance containing d1ay1II is hindered when it co-occurs with the modal s2a:m3a:t ‘can’, as in (23d), in the conversation (23).24

xtne [k1r3eml3in]cd (x) [ACCΔe]ws [Δ = ˇ]ws [t = n]ws, cpi 1 e:

[x c1ap ng3u:]ws

Figure 10.â•‡ A departure from the default of an utterance k1r3eml3in s2a:m3a:t c1ap ng3u: d1ay1II (Gremlin can catch snake d1ay1II; ‘Gremlin can catch a snake’)

24. On condition that (23d) is discourse-initial, two interpretations can be obtained, just like for (23a).

Communicating about the past 

Unlike that in Figure 9, ACC here does not only capture the modality of d1ay1II but also of the modal s2a:m3a:t ‘can’. The past temporal default t < n does not arise, but the departure from the default t = n is accounted for by WS, which is the dynamic possibility of d1ay1II and the modal s2a:m3a:t ‘can’, and CPI 1, which produces the inference to the dynamic possibility at present by combining the WS of d1ay1II with the meaning of the modal s2a:m3a:t ‘can’. 7. Concluding remarks We demonstrated in this chapter that semantic representation of past-time reference in English and in Thai requires a theory that would account for the combination of information about temporality that comes from various sources. There are ways of expressing past-time reference that do not rely on an overt grammatical or lexical marker of temporality but instead use information from standard, presumed meaning (default) or pragmatic inference in order to determine the temporal location of the situation. We used examples from English and from Thai in order to show that, although English sentences normally come with tense and aspect markers while in Thai both grammatical markers of temporality are optional, similar problems with determining the time of eventuality arise in both languages. We hope to have shown that the past-time reference in Thai may be conveyed by the auxiliary-like item d1ay1II and that pastness is not the encoded meaning of the marker because there are cases where the past time interpretation does not arise. We also analyzed selected problematic examples with past-time reference in both languages in the framework of DS and conclude that this framework can easily accommodate sentences without overt grammatical indicators of time in that its founding principle is the representation of the merger of information about meaning that comes from various sources which can add to, or even overcome, the information provided by the grammar and the lexicon. The ‘big picture’ that emerges for a theory of meaning is this. While compositionality is a necessary prerequisite for any theory of meaning, compositionality should not be seen as an aim that would justify complicating formal methods in order to fit natural language into the mould created for formal languages of deductive logic. Instead, we repeat after Jackendoff (2002: 293) that there is no ‘strictly linguistic meaning’ and that constructing a mental representation of discourse has the status of the fundamental objective. But the task of a theory of meaning is not fulfilled until the interaction of sentence structure with the results of pragmatic inference, background knowledge, beliefs, perception, etc. are captured more

 Katarzyna M. Jaszczolt and Jiranthara Srioutai

formally as an algorithm for human communication.25 Temporality in English and Thai is a good example on which this need for merging meaning coming from different sources can be demonstrated and taken forward. References Aikhenvald, Alexandra Y. 2004. Evidentiality. Oxford: Oxford University Press. Austin, John L. 1961. Philosophical Papers. Oxford: Clarendon Press. van der Auwera, Johan & Vladimir A. Plungian. 1998. Modality’s semantic map. Linguistic Typology 2: 79–124. Bach, Kent. 1994. Semantic slack: What is said and more. In Foundations of Speech Act Theory: Philosophical and Linguistic Perspectives. S. L. Tsohatzidis (ed.), 267–291. London: Routledge. Bach, Kent & Robert M. Harnish. 1979. Linguistic Communication and Speech Acts. Cambridge, Mass.: MIT Press. Biber, Douglas, Susan Conrad, & Geoffrey Leech. 1999. Longman Grammar of Spoken and Written English. London: Longman. Carston, Robyn. 1988. Implicature, explicature, and truth-theoretic semantics. In Mental Representations: The Interface between Language and Reality, R. M. Kempson (ed.), 155–181. Cambridge: Cambridge University Press. Carston, Robyn. 2002. Thoughts and Utterances: The Pragmatics of Explicit Communication. Oxford: Blackwell. Culicover, Peter W. & Ray Jackendoff. 2005. Simpler Syntax. Oxford: Oxford University Press. Diller, Anthony 1996. Thai and Lao Writing. In The World’s Writing Systems, P. T. Daniels & W. Bright (eds.), 457–466. New York: Oxford University Press. van Eijck, Jan & Hans Kamp. 1997. Representing discourse in context. In Handbook of Logic and Language, J. van Benthem & A. ter Meulen (eds.), 179–237. Amsterdam: Elsevier Science. Enç, Mürvet 1996. Tense and modality. In The Handbook of Contemporary Semantic Theory, S. Lappin (ed.), 345–358. Oxford: Blackwell. Fleischman, Suzanne. 1982. The Future in Thought and Language: Diachronic Evidence from Romance. Cambridge: Cambridge University Press. Fodor, Jerry A. 1998. Concepts: Where Cognitive Science Went Wrong. Cambridge, MA: MIT Press. Grice, Paul. 1978. Further notes on logic and conversation. In Syntax and Semantics, vol. 9: Pragmatics, P. Cole (ed.), 113–128. New York: Academic Press. Grice, Paul. 2001. Aspects of Reason, Oxford: Clarendon Press. Hamm, Fritz, Hans Kamp, & Michiel van Lambalgen. 2006. There is no opposition between Formal and Cognitive Semantics. Theoretical Linguistics 32: 1–40. 25. It is worth noting that merger representations of Default Semantics schematically given in Figure 3 above are not incompatible with the interfaces of Jackendoff ’s conceptual semantics: see Figure 9.1 in Jackendoff (2002). However, the reliance on intentionality separates Default Semantics from Jackendoff ’s construal and makes it akin to Fodor’s representational semantics (e.g. Fodor 1998).

Communicating about the past  Iwasaki, Shoichi & Preya Ingkaphirom. 2005. A Reference Grammar of Thai. Cambridge: Cambridge University Press. Jackendoff, Ray. 2002. Foundations of Language: Brain, Meaning, Grammar, Evolution. Oxford: Oxford University Press. Jaszczolt, Katarzyna M. 2003. The modality of the future: A Default-Semantics account. In Proceedings from the 14th Amsterdam Colloquium, P. Dekker & R. van Rooy (eds.), 43–48. Amsterdam: ILLC, University of Amsterdam. Jaszczolt, Katarzyna M. 2005. Default Semantics: Foundations of a Compositional Theory of Acts of Communication. Oxford: Oxford University Press. Jaszczolt, Katarzyna M. 2006. Futurity in Default Semantics. In Where Semantics Meets Pragmatics: The Michigan Papers, K. von Heusinger & K. Turner (eds.), 471–492. Oxford: Elsevier. Jaszczolt, Katarzyna M. 2009. Representing Time: An Essay on Temporality as Modality. Oxford: Oxford University Press. Jaszczolt, Katarzyna M. 2010. Default Semantics. In The Oxford Handbook of Linguistic Analysis, B. Heine & H. Narrog (eds.), 193–221. Oxford: Oxford University Press. Kanchanawan, N. 1978. Expression for Time in Thai Verb and its application to Thai-English Machine Translation. PhD Dissertation. University of Texas, United States. Kamp, Hans & Uwe Reyle. 1993. From Discourse to Logic: Introduction to Modeltheoretic Semantics of Natural Language, Formal Logic and Discourse Representation Theory. Dordrecht: Kluwer. Kim, Jaegwon. [1976] 1993. Events as property exemplifications. In Action Theory, M. Brand & D. Walton (eds.), Dordrecht: D. Reidel. Reprinted in Supervenience and Mind: Selected Philosophical Essays, Jaegwon Kim, 33–52. Cambridge: Cambridge University Press. van Lambalgen, Michiel & Fritz Hamm. 2005. The Proper Treatment of Events. Oxford: Blackwell. Levinson, Stephen C. 2000. Presumptive Meanings: The Theory of Generalized Conversational Implicature. Cambridge, Mass.: MIT Press. Ludlow, Peter. 1999. Semantics, Tense, and Time: An Essay in the Metaphysics of Natural Language. Cambridge, Mass.: MIT Press. McTaggart, John E. [1908] 1934. The unreality of time. Mind 17. Reprinted in Philosophical Studies, John E. McTaggart, 110–131. London: Edward Arnold. Mellor, David H. 1993. The unreality of tense. In The Philosophy of Time, R. Le Poidevin & M. MacBeath (eds.), 47–59. Oxford: Oxford University Press. Mellor, David H. 1998. Real Time II. London: Routledge. Moens, Marc & Mark Steedman. [1988] 2005. Temporal ontology and temporal reference. Computational Linguistics 14. Reprinted in The Language of Time: A Reader, I. Mani, J. Pustejovsky, & R. Gaizauskas (eds.), 93–114. Oxford: Oxford University Press. Muansuwan, Nuttanart 2002. Verb Complexes in Thai. PhD Dissertation. University at Buffalo, The State University of New York. Nicolle, Steve & Billy Clark. 1999. Experimental pragmatics and what is said: A response to Gibbs and Moise. Cognition 69. 337–354. Noss, Richard B. 1964. Thai Reference Grammar. Washington, D.C.: Foreign Service Institute. Noveck, Ira A. & Dan Sperber (eds.). 2004. Experimental Pragmatics. Basingstoke: Palgrave Macmillan. Palmer, Frank R. 2002. Mood and Modality. 2nd edition. Cambridge: Cambridge University Press. Parsons, Josh 2002. A-theory for B-theorists. The Philosophical Quarterly 52: 1–20. Parsons, Josh. 2003. A-theory for tense logicians. Analysis 63: 4–6.

 Katarzyna M. Jaszczolt and Jiranthara Srioutai Pitts, Alyson. 2005. Assessing the evidence for intuitions about what is said. Ms, Cambridge: University of Cambridge. Prior, Arthur N. 1957. Time and Modality. Oxford: Clarendon Press. Prior, Arthur N. 1967. Past, Present and Future. Oxford: Clarendon Press. Prior, Arthur N. 1968. Papers on Time and Tense. Oxford: Clarendon Press. Prior, Arthur N. 2003. Papers on Time and Tense: New Edition. Oxford: Oxford University Press. Recanati, François. [1989] 1991. The pragmatics of what is said Mind and Language 4. Reprinted in Pragmatics: A Reader, S. Davis (ed.), 97–120. Oxford: Oxford University Press. Recanati, François. 1994. Contextualism and anti-contextualism in the philosophy of language. In Foundations of Speech Act Theory: Philosophical and Linguistic Perspectives, S. L. Tsohatzidis (ed.), 156–166. London: Routledge. Recanati, François. 2003. Embedded implicatures. http://jeannicod.ccsd.cnrs.fr/documents. Recanati, François. 2004. Literal Meaning. Cambridge: Cambridge University Press. Recanati, François. 2010. Truth-Conditional Pragmatics. Oxford: Oxford University Press. Reichenbach, Hans. 1948. Elements of Symbolic Logic. New York: Macmillan. Sattig, Thomas. 2006. The Language and Reality of Time. Oxford: Oxford University Press. Smith, Quentin. 1993. Language and Time. Oxford: Oxford University Press. Sperber, Dan & Deirdre Wilson. 1986 [1995]. Relevance: Communication and Cognition. 2nd edition. Oxford: Blackwell. Srioutai, Jiranthara. 2004a. The Thai c1a: A marker of tense or modality? Paper presented at Chronos 6: International Colloquium on Syntax, Semantics and Pragmatics of Tense, Mood and Aspect, September 22–24 2004, University of Geneva. Srioutai, Jiranthara. 2004b. The Thai c1a: A marker of tense or modality? In Second CamLing Proceedings, E. Daskalaki et al. (eds.), 273–280. Cambridge: University of Cambridge. Srioutai, Jiranthara. 2006. Time Conceptualization in Thai with Special Reference to d1ay1II, kh3oe:y, k1aml3ang, y3u:I and c1a. PhD Dissertation. University of Cambridge, UK. Steedman, Mark. 1997. Temporality. In Handbook of Logic and Language, Johan van Benthem & Alice ter Meulen (eds.), 895–937. Oxford: Elsevier Science. Supanvanich, I. 1973. Tenses in Thai. Master Thesis. Chulalongkorn University, Thailand. [In Thai] Sysoeva, Anna & Katarzyna Jaszczolt. 2007. Composing utterance meaning: An interface between pragmatics and psychology. Paper presented at the 10th International Pragmatics Conference, July 2007, Göteborg. http://people.pwf.cam.ac.uk/kmj21/pwft.html Thomason, Richmond H. 2002. Combinations of tense and modality. In Handbook of Philosophical Logic 7, D. Gabbay & F. Guenthner (eds.), 205–234. Dordrecht: Kluwer.

The epistemic uses of the English simple past and the French imparfait When temporality conveys modality* Adeline Patard

University of Antwerp This chapter explores the connection between past tense and modality in English and French. After arguing for a temporal definition of past tenses, I reinterpret the classical opposition between temporal uses and modal uses in terms of the speakers’s referential or subjective intentionality. I further distinguish between the epistemic uses – which express the speaker’s assessment of the probability of the denoted situation – and the illocutory uses – which express the speaker’s degree of commitment in her speech act. I finally suggest an analysis of two epistemic uses of the English simple past and the French imperfect, namely their conditional use and optative use, thanks to the notion of dialogism, which refers to the heterogeneity of the enunciative sources of a given utterance. Keywords: past tense, reference point, dialogism, English, French

1. Introduction The observation that past tenses may convey modal meanings is well known and widely documented cross-linguistically.1 This connection between past tense and modality is illustrated in the following uses of the English simple past (now SP) and the French imperfective past (now IP):

* This research is funded by a Marie Curie IEF fellowship granted to the author by the European Commission. The results presented here are also part of research carried out within the context of the GRAMIS project, sponsored by the Belgian Science Policy Office (Interuniversity Attraction Pole programme project P6/44). 1.

Cf. inter alia James 1982, Comrie 1985, Fleischman 1989, Thieroff 1999.

 Adeline Patard

Conditional (hypothetical) use (1) a. And if you left me I would suffer a great deal. (R. Jaffe, After the Reunion) b.

si un jour tu partais sans retour if one day you leave-pst.ipfv without return / les fleurs perdraient leur parfum / the flowers would lose their perfume ‘If someday you left for ever / the flowers would lose their perfume’ (É. Piaf)

Optative use (2) a. I wish I was a punk rocker with flowers in my head (S. Thom) b. Ah! Si j’ étais riche! ah if I be-pst.ipfv rich ‘Ah! If only I was rich!’

(G. de Maupassant, Les bijoux)

Suppositive use (3) a. Suppose you were a rose and I was a whip-poor-will.

(R. Miller)

b. Si c’ était lui, le condé? if it be-pst.ipfv him the cop ‘Suppose that it was him who was the cop?’ (J.-L. Degaudenzi, Zone) Hypothetical comparison (4) a. I’d like to live as if only love mattered b.

chante la vie chante sing the life sing / comme si tu devais mourir demain / as if you must-pst.ipfv die tomorrow ‘Sing life sing / as if you should die tomorrow’

(T. Chapman)

(M. Fugain)

In these examples, the past tense is associated with ‘epistemic distance’ whereby the eventuality described by the verb is presented as ‘distant’ from the speaker’s reality, i.e. as very unlikely (cf. (1a), (1b), (3b), (4a) and (4b)) or unreal (cf. (2a), (2b) and (3a)). These epistemic uses differ from the prototypical ‘temporal’ use of SP and IP exemplified in (5):

(5) a. It was dark all around / there was frost in the ground/when the tiger broke free (Pink Floyd)

blond, il était beau / b. Il était he be-pst.ipfv blonde, he be-pst.ipfv beautiful /

When temporality conveys modality 

Il sentait bon le sable chaud / mon légionnaire he smell- pst.ipfv good the sand warm / my legionnaire ‘He was blonde, he was beautiful / he smelt like warm sand / my legionnaire’ (É. Piaf)

at least in two respects: 1. the time reference: the denoted situation can be present (cf. (2a), (2b), (3a), (3b) and (4a)) or future (cf. (1a), (1b) and (4b)) in the epistemic uses whereas their interpretation is past in the prototypical use (cf. (5)); 2. the epistemic domain: as already noticed, the situation described in the epistemic use is interpreted as counterfactual, that is, their factual status is suspended and called into question (cf. (1), (2), (3)). Contrastively, in the prototypical use, the situation pertains to the domain of reality, for their factuality is asserted (cf. (5)). This general observation about the polysemy of past tenses, as exemplified here in English and in French, has yielded a long-standing debate on the nature of the connection between past tense and epistemic distance and, more broadly, between temporality and modality. We wish to contribute to this debate by offering an original analysis based on the notion of ‘dialogism’. In what follows, we will assume that the SP and the IP also refer to the past in their epistemic uses, but that past reference is covert in such uses due to their ‘dialogic’ interpretation. The chapter will be organized as follows. In Section 2, we argue for a temporal definition of of SP and IP. In Section 3, we introduce distinctions among the different uses of the SP and IP, in order to clarify our definition of the ‘epistemic uses’ at stake in the chapter. Section 4 presents the notion of ‘dialogism’ and its possible application to the description of tenses uses. The last section is dedicated to the analysis of two epistemic uses of the SP and the IP, namely the conditional use and the optative use. 2. A temporal definition of the SP and the IP Verbal tenses are traditionally defined thanks to temporal (and aspectual) features. Following this hypothesis, one usually posits that the SP and the IP encode past reference. This definition of past tense as meaning past is defended by scholars like James (1982), Comrie (1985), Fleischman (1989), Declerck (1991 and 2005), Smith (1991), Thieroff (1999) or Ippolito (2003) for SP, or by Imbs (1960), Vet (1980), Gosselin (1996), Wilmet (2003), Barceló & Bres (2006) or Patard (2007) for IP. This traditional conception may be contrasted with the ‘epistemic’ or ‘inactual’ definition

 Adeline Patard

of past tenses proposed by some linguists (cf. Damourette & Pichon 1911–1936, Langacker 1978 and 1991, Le Goffic 1995, Touratier 1996, Iatridou 2000, De Mulder 2004, De Mulder & Brisard 2006, Brisard 2010). In this second view, past tense morphology does not denote past reference but indicates that the described situation is epistemically distant from the speaker’s present actuality.2 The first subsection will seek to argue briefly for a temporal conception (against the ‘epistemic’ one). 2.1

Past morphology means past

We embrace the traditional view that the past morphology in the SP and the IP serves to refer to a past moment as a basic meaning. The main arguments for adopting this stance are the following. First, as noticed by many authors, the ‘default’ or ‘prototypical’ interpretation of past morphology is clearly past. Hence the temporal reading automatically emerges in the absence of modal markers:

(6) a. She was sad.

b. Il neigeait. it snow-pst.ipfv ‘It snowed/was snowing (in the past)’ These observations seem more easily compatible with a temporal conception of past tense than with an inactual one. Furthermore, cross-linguistic studies show (cf. James 1982, Fleischman 1989) that past tense cannot generally convey modal interpretations on their own, but they always necessitate to be combined with modal markers (such as if and would in English counterfactual conditionals). According to Dahl (1997: 100), this should be a problem for the inactual hypothesis since it is not clear why non-past interpretations of inactuality should be marked in supplementary ways, as opposed to its past interpretations. James (1982), Dahl (1997), and Hogeweg (2009) also emphasize the ‘irregular and idiosyncratic’ character of past tense’s modal readings, as opposed to their temporal interpretations. Indeed, there is cross-linguistically (cf. James 1982) a considerable variation as to the contexts in which a past tense serves to convey counterfactuality. This irregularity of use may also surface in one given language. For instance, in English, the use of the SP in counterfactual environments can be either obligatory (as in counterfactual conditionals or optatives, cf. (1a) and (1b)), optional (7), or impossible (8): 2. Depending on the context, this epistemic distance can be interpreted temporally and give rise to a past reading, or be interpreted modally and bring about counterfactuality.

When temporality conveys modality 

(7) a. Imagine that Johny was coming tomorrow. (adapted from Hogeweg 2009: 185) b. Imagine that Johny is coming tomorrow. (Hogeweg 2009: 185) (8) a. b.

Johny acts like he *was drunk tonight. (adapted from Hogeweg 2009: 186) I have the recurrent dream I *was a gipsy. (adapted from Hogeweg 2009: 186)

On the contrary, the temporal uses of past tenses are ‘fully regular and productive’ and need not be memorized by a user, who can make generalizations to use them properly in such contexts (James 1982: 398). For the authors aforementioned, this asymmetry is not compatible with an inactual definition of past tenses but rather advocates for a past basic meaning. We can quote a last observation by James (1982) and Dahl (1997) who point out that past tenses in their modal uses do not serve to describe situations that are contrary to facts, but to express “a greater degree of distance with reality”. This is typically observed in ‘future less vivid’ conditionals, where the denoted situation refers to a future unlikely event:3 (9) a. If he took that syrup, he would get better. b. If he takes that syrup, he will get better.

(Iatridou 2000: 249)

(10) a.

S’ il venait demain, if he come-pst.ipfv tomorrow je lui donnerais l’ argent. I him would give the money ‘If he came tomorrow, I would give him the money’(James 1982: 388)

b.

S’ il vient demain, if he come-prs tomorrow, je lui donnerai l’ argent. I him will give the money ‘If he comes tomorrow, I will give him the money’

Here, by comparison with the present forms (cf. takes and vient), the past forms (cf. took and venait) present the situation as improbable, but the latter is not completely excluded from the speaker’s (future) reality. For James, the inactual hypothesis fails to explain why past tenses convey lesser probability in these contexts 3. According to us, past tenses also express lesser probability in ‘counterfactual’ conditionals, but due to the interaction with contextual elements (notably world knowledge and actionality, cf. Martin 1991, Gosselin 1999 and Iatridou 2000) the situation is finally interpreted as contrary to facts (cf. Section 4.2).

 Adeline Patard

rather than a complete absence of reality, predicted by the putative inactuality of these forms. Following these observations, we hypothesize that the SP and of the IP have a basic temporal meaning: they encode past time reference. In the next subsection we specify what they exactly refer to in the past. This will also permit us to introduce the notion of (grammatical) aspect. 2.2

Tense and aspect

Since Reichenbach (1947) at least, it is quite trivial to say that tenses do not directly describe a relationship between the denoted situation and the time of speech (now S), but that this relation is always mediated by one or more ‘reference point(s)’. As pointed by Klein (2009: 45), the necessity for a ‘reference point’ to describe the semantics of past tense may appear in an elementary sentence like: (11) Eva’s cat was dead.

(Klein 2009: 45)

If past tenses only encoded the precedence of the situation in relation to S, then the situation [Eva’s cat be dead] would hold before S but not during S, which is of course not what is meant (Eva’s cat is still dead at the present time). That is why linguists usually posit a past ‘reference point’ to which the past tense refers to. Accordingly the situation [Eva’s cat be dead] is said to be true for the past time span denoted by the reference point (at a given past time, it was the case that Eva’s cat is dead), but this does not impede the situation to be also true at other times (Eva’s cat is still dead at the present time). Linguists have suggested many hypotheses as regards the nature of these reference points. Although the question is still a matter of debate, it is reasonable to think that their number varies according to the tense used and that they may play different roles depending on the context.4 In the present study, we distinguish two functions they usually fulfill for any tenses. Thus, the reference point(s) stated by a given tense may serve: 1. as a topic time (Klein 1994): the reference point refers to the time span about which a particular utterance makes an assertion (or ask a question); 2. as a viewpoint5 (Smith 1991, Klein 1994, Gosselin 1996): the reference point functions as a vantage point from which the situation is viewed.

4. For instance, it is generally admitted that conditional past requires at least one more reference point (cf. Declerck 1986, Gosselin 1996). 5.

The term is used by Smith (1991).

When temporality conveys modality 

These two specific functions may be related respectively to the temporal and aspectual semantic content of a tense morpheme: 1. Tense (or temporal location) concerns the relationship between S and the topic time. Thus we hypothesize that past tenses denote a past topic time: they are meant to talk about a past moment. 2. Aspect concerns the relationship between a viewpoint and the situation. One can distinguish three aspectual categories: − imperfective aspect indicates that the viewpoint is embedded within the time of the situation, thus excluding the boundaries of the situation (cf. Smith 1991, Klein 1994, Gosselin 1996); − perfective aspect indicates a strict simultaneity between the viewpoint and the situation, thus including the boundaries of the situation (cf. Gosselin 1996); − neutral aspect encodes no specific relation between the viewpoint and the situation, this relation is then usually determined by the context (mostly the actionality of the verb, but not only) (cf. Smith 1991). Finally, we would like to notice that the topic time equates the viewpoint in most tense forms, as for the SP and the IP, but this is not obligatory the case. We can now define, in the next subsection, the semantics of the SP and IP. 2.3

Definition

We defined the basic meaning of the SP and IP in terms of temporal and aspectual semantic features. Accordingly, the meaning of the SP is characterized by two features: − [past]: it refers to a past topic time, i.e. the SP licenses an assertion about a past moment; the prototypical interpretation of this past feature is that the situation is true at a certain past time (cf. (12) and (13) below); − [neutral]: the viewpoint on the situation (defined by the topic time) is not specified by the SP. Consequently, the SP can give rise to imperfective (12) or perfective interpretations (13) depending on the context: (12) yesterday love was such an easy game to play (13) I left my hand and my heart on the dance floor

(The Beatles) (Lady Gaga)

Likewise, the meaning of the IP is characterized by two features: − [past]: it refers to a past topic time, i.e., the IP licenses an assertion about a past moment (cf. (14) below);

 Adeline Patard

− [imperfective]: the viewpoint (defined by the topic time) is embedded within the time of the situation; the IP thus excludes the boundaries of the situation and is generally associated with an imperfective interpretation of the situation (cf. (14)):6 (14) elle avait quelque chose d’ un ange she have-pst.ipfv some thing of an angel ‘She looked like an angel’

(G. Brassens)

In the next section, we introduce two distinctions concerning the different uses of the SP and IP: ‘temporal’ versus ‘modal’ uses and ‘epistemic’ versus ‘illocutory’ uses. 3. Remarks on the uses of tenses 3.1

‘Temporal’ versus ‘modal’ uses

The first distinction is intended to clarify what we conceive as the ‘modal’ uses of indicative tenses. Crucially, the distinction between ‘temporal uses’ and ‘modal uses’ should not be seen as an exclusive dichotomy between temporality and modality whereby ‘temporal uses’ exclude modal meaning(s) and vice versa. Indeed, indicative tenses convey a modal meaning of ‘realis’ or ‘factuality’ – typically they present the situation described as being the case at a past/present/future moment – which is to be contrasted with the ‘irrealis’, ‘virtual’ or ‘non-factual’ modality of the subjunctive forms (cf. inter alia Martin 1983, Givón 1994, Soutet 2000, Palmer 2001).7 So, as markers of the indicative mood, indicative tenses in English and French have a modal import, which is reflected in their prototypical ‘temporal uses’, where they enable the speaker to ground the situation in the factual world, as past, present or future. Besides, every tense entails specific modal implications due to human experience of the passing of time. The modal dimension of the experienced time is acknowledged since antiquity and may be formulated as follows (at least for Western cultures): we experience time as an ‘irreversible flow’ with the present time corresponding to a modal cut between what is ‘possible’ or ‘indeterminate’ and what is

6. Nevertheless, the IP is not incompatible with a perfective interpretation (cf. its narrative use), but, in that case, it is the context that is responsible for the perfective viewpoint (cf. Vetters & De Mulder 2003, and Bres 2005). 7. See also Gosselin 2009 (Section 8.8) for a discussion of the unreal value of the subjunctive forms.

When temporality conveys modality 

‘irrevocable’ (Gosselin 2005: 89, 2009: 138).8,9 Thus, due to our experience of time, past tenses generally implicate for us that the eventuality is irrevocably factual, and future tenses usually involve that the eventuality remains possible and uncertain. These implicatures explain the acceptability judgments of examples (15): (15) a. It rained yesterday, *but maybe not. b. It will rain tomorrow, but maybe not. In sum, English and French verbal tenses in their ‘temporal uses’ also convey modality, both at the semantic level (as indicative mood markers) and pragmatic level (due to our experience of time). Reciprocally, the ‘modal’ uses of tenses also involve some kind of temporality insofar as they anchor the eventuality in time, albeit not (always) in the factual world. The epistemic uses of the simple past typically convey non-past reference (cf. supra examples (1a), (2a), (3a) and (4a)) that can be contextually interpreted as present or future. And so does the epistemic(cally used) imparfait (supra examples (1b), (2b), (3b) and (4b)). We must conclude from what precedes that temporality and modality do not exclude each other in the so-called ‘temporal’ and ‘modal’ uses of indicatives tenses, but are intricately related in both types of usages. Nevertheless we still use the distinction between ‘temporal’ and ‘modal’ uses, but, in a different sense, as reflecting the intentionality of the speaker, i.e. her communicative purpose when choosing a specific verbal forms (instead of another). If the intentionality is referential and mainly concerned with the temporal anchorage of the situation (or the viewpoint on the situation) in time, we consider the use as ‘temporal’. Thus, when using a past form in examples (5) aforementioned, the speaker only intends to inform the hearer about situations that were the case in the past. Conversely, if the speaker’s intentionality is not primarily concerned with the temporal grounding of the situation (or the viewpoint on the situation) in time, but rather corresponds to a modal attitude conveyed by the tense employed, we call this type of use ‘modal’. Hence, in examples (1)–(4), the speaker does not wish to communicate the pastness of the situations (or of their viewpoint), but the epistemic status, i.e. the unlikelihood or unreality of the situations. For this reason, we consider these uses as ‘modal’.

8. Cf. Plato (Protagoras, 324b), Aristotle (Nichomachean Ethics VII). 9. A similar epistemic conception of time is the ‘branching-futures’ model discussed by Tedeschi (1981) that treats time as having a tree-like structure. In this view, the past (with respect to any point in time) must be regarded as one, while the future is represented as having an infinite set of possible ramifications.

 Adeline Patard

The speaker’s referential or modal intentionality can be revealed by means of substitution tests. If one tense can be replaced by another in a given context and if this substitution does entail a different modal interpretation (e.g., a different epistemic status of the situation), then the use can be considered ‘modal’. To illustrate this, we derive (16) from (1) by replacing the past tense in the protasis by a present tense: (16) a. And if you leave (/left) me I will (/would) suffer a great deal. b.

si un jour tu pars (/partais) sans retour if one day you leave-prs (/pst.ipfv) without return / les fleurs perdront (/perdraient) leur parfum / the flowers will lose (/would lose) their perfume ‘If someday you leave (/left) for ever / the flowers will (/would) lose their perfume’

From an interpretative viewpoint, the contrast between the past forms (left and partais) and the present forms (leave and pars) is epistemic: with the past forms, the situation (the hearer’s leaving) appears to be very unlikely whereas, with the present tenses, it is interpreted as more probable (cf. also James 1982, Iatridou 2000, Ogihara 2000, Dancygier & Sweetser 2005: chap. 3, Patard & Vermeulen 2010). As a result, the past tenses are motivated by a modal intentionality: they are used to convey information about the epistemic status of the situation. Conversely, if the substitution gives rise to a different temporal interpretation (e.g. present-time reference instead of a past-time reference), then the use is seen as ‘temporal’, as in (17) derived from (5): (17) a. It is (/was) dark all around / there is (/was) frost in the ground / when the tiger breaks (/broke) free b.

Il est (/était) blond / il est (/était) beau / he be-prs (/pst.ipfv) blonde, he be-prs (/pst.ipfv) beautiful / il sent (/sentait) bon le sable chaud he smell- prs (/pst.ipfv) good the sand warm ‘He is (/was) blonde / he is (/was) beautiful / he smells (/smelt) like warm sand’

The contrast between the past forms (was, broke, était and sentait) and the present forms (is, breaks, est and sent) results in a different referential interpretation: the past tenses induce a past-time reference and the present tenses a present-time reference. Finally we would like to notice that the proposed distinction does not constitute an exclusive opposition between two different kinds of usage. Indeed, some specific uses, like the IP expressing a ‘thwarted imminence’ (cf. (18)), possess the

When temporality conveys modality 

characteristics of both usage types: they are temporal and modal (from a communicative standpoint).10 Let us consider the following examples: (18) a. Une seconde de plus [le taureau] l’ éventrait. one second of more the bull him gore-pst.ipfv ‘One more second and the bull would have gored (/gored) him’ (Flaubert) b. Une seconde de plus [le taureau] l’ éventra. one second of more the bull him gore-pst.pfv ‘One more second and the bull gored him’ c. Une seconde de plus [le taureau] l’ éventre. one second of more the bull him gore-prs ‘One more second and the bull will gore him’ The difference between the two first examples lies on the use of the IP in (18a) and of the past perfective tense in (18b). This aspectual contrast (between a past imperfective and a past perfective) entails two different epistemic interpretations: with the IP, the situation is primarily interpreted as counterfactual in the past, whereas it is interpreted as past and factual with the past perfective.11 Hence we must conclude that the use of the IP in (18a) is ‘modal’ and motivated by an epistemic intentionality. Now considering the contrast between the use of the IP in (18a) and the use of the present tense in (18c), it brings about two different temporal interpretations, namely a past interpretation with the IP and a non-past interpretation with the present tense.12 As a conclusion, this use of the IP is also temporal (not only modal) since it serves to anchor the situation in the past as well. The proposed definition of ‘temporal’ and ‘modal’ uses in terms of intentionality seeks to clarify the distinction between both usage types and make predictions (using the substitution test) as regards the classification of a given use in one or the other category. It also underlines the fact that temporality and modality do not exclude each other in the usage of verbal tenses, but that a given use can be both temporal and modal in different respects.

10. It is known in the French-speaking literature as the ‘emploi contrefactuel’ or ‘emploi d’imminence contrecarrée’ (cf. Berthonneau & Kleiber 2003, Bres 2006). 11. Nevertheless, the factual reading is not excluded either, though it seems less probable. In the latter case, we would have an instance of ‘imparfait narratif ’ (cf. Berthonneau & Kleiber 2003, Bres 2006). 12. From a modal point of view, the present in the utterance indicates that the eventuality is possible in the near future.

 Adeline Patard

3.2

‘Epistemic’ versus ‘illocutory’ uses

We now introduce a second distinction within the category of the modal uses, namely between ‘epistemic’ uses and ‘illocutory’ uses. We briefly evoked the first category in the introduction. The epistemic uses are motivated by an ‘epistemic’ intentionality of the speaker: by using one or another tense, the speaker expresses some ‘judgment’ regarding the degree of uncertainty or probability of the situation described (cf. van der Auwera & Plungian 1998: 81). Thus, when using the SP and IP (cf. utterances (1)–(4)), the speaker intends to present the situation as improbable. By contrast, the ‘illocutory’ uses are characterized by a distinct communicative purpose. Instead of an epistemic intentionality, the use of a specific tense is motivated by the expression of an illocutory posture of the speaker, i.e. a more or less important commitment in her speech act and a particular attitude towards the hearer. Typically, the illocutory uses of the SP and the IP produce politeness, as in (19): Attenuative use (19) a. I wanted (/want) to ask you to do me a favour. (O. Wilde, Woman’s World) b. Je voulais (/veux) te dire que je t’ attends I want-pst.ipfv (/-prs) you tell that I you wait-prs ‘I wanted (/want) to tell you that I’m waiting for you’ (M. Jonasz) In (18), the speaker uses the SP (wanted) and the IP (voulais) instead of the present tense (want and veux) to mitigate her request (19a) or her assumption (19b), i.e. to attenuate the illocutory force of the speech act. Put differently, the speaker shows a lesser commitment in her speech act (by presenting it indirectly) and, in doing so, she preserves the hearer’s ‘face’ by softening a potentially ‘threatening act’ (Brown & Levinson 1987).13 As for the distinction between ‘temporal’ and ‘modal’ uses, the distinction between epistemic uses and illocutory uses does not delimit two discrete categories of usage, but one single use may exhibit both an epistemic and illocutory intentionality. This is the case in examples (20) and (21): Suggestion (20) a. It’s high time you were (/are) all in bed! (E. Lewis, Alice’s Adventures in Wonderland)

13. See Berthonneau & Kleiber (1994) and Patard & Richard (2011) for a more detailed analysis of the attenuative uses of the IP. To our knowledge, there is no specific study on the attenuative uses of SP in English, though it is mentioned in more general works (cf. Leech 1971, Quirk et al. 1985, Fleischman 1989).

When temporality conveys modality 

b. Si on s’en allait (/va)? if one leave-pst.ipfv (/-prs) ‘Shall we go?’ (21)

(P. Gide, Les faux-monnayeurs)

Preludic use Moi j’ étais (/suis) le papa, me I be-pst.ipfv (/-prs) the daddy et toi, tu étais (/es) la maman. and you you be-pst.ipfv (/-prs) the mommy ‘Me, I’m the daddy and you, you’re the mommy’

(Warnant 1966)

In (20), were and s’en allait produce a specific epistemic interpretation (cf. Declerck 2006 and this volume, Patard 2009): these past forms stress the not-yetfactuality of the situation, as opposed to the present forms (are and s’en va). This epistemic meaning further entails a particular illocutory posture of the speaker. As the non-factuality of the situation is highlighted, the speaker’s suggestion sounds more tentative and polite: the speaker leaves to the hearer the choice to see the situation as factual (or not) and to validate it (or not) in her act. Thus, the past tense also mitigates the proposed suggestion. As a conclusion, the SP and the IP here convey both an epistemic non-factuality and a decreased illocutory force. As regards the preludic (or pretend game) use of tenses (cf. example (21)), the literature on the topic usually describes two different functions for the preludic tense.14,15 These functions may be referred to, following Lodge (1978), as ‘realityswitching’ and ‘self-effacement’. Accordingly the IP in the examples aforementioned (étais and étais) first signals a switch from the real world to the imaginary world of the game, hence producing an epistemic meaning of unreality (cf. Warnant 1966, Fleischman 1989 and 1995, Patard 2010). As a second function, the imparfait allows for the speaker to show an attitude of self-effacement and make less assertive propositions about the game (Patard 2010). This attitude of self-effacement may be seen as a specific illocutory posture whereby the speaker softens propositions that may appear as ‘potentially threatening’ for the hearer’s ‘face’ (Brown & Levinson 1987). In conclusion, the preludic use of the IP must be considered as both ‘modal’ and ‘illocutory’.

14. The label ‘preludic’ is suggested by Warnant (1966) inasmuch as this use typically occurs during the negotiation of the content of the game that prefaces the game itself. 15. See Warnant (1966), Lodge (1978), Musatti & Orsolini (1993), Kauppinen (1996), Patard (2010) for specific studies on the preludic use of tenses in several languages. According to these, the preludic tense(s) may be an imperfective past, a simple past and/or a conditional tense, depending on the language. In French, both the imparfait and the conditional tense may occur in a preludic context (cf. Patard 2010).

 Adeline Patard

The aim of this section was to clarify our definition of ‘epistemic use’. Accordingly an epistemic use of a tense is a type of ‘modal’ use insofar as it is primarily motivated by a modal ‘intentionality’: the speaker wishes to express a certain subjective attitude (about the content of the utterance or towards the hearer). In their epistemic use, tenses express some ‘judgment of the speaker’ regarding the degree of uncertainty or probability of the situation described. By contrast, illocutory uses are concerned with the expression of an ‘illocutory’ attitude towards the hearer in the situation of communication. In the following sections, we focus on the epistemic uses stricto sensu. This means that the illocutory uses (cf. (19)) and epistemic-illocutory uses (cf. (20) and (21)) aforementioned will not be treated in the rest of the study. But, before turning to the epistemic uses of the SP and IP, we first present the notion of dialogism and its possible application to the analysis of verbal tenses. This notion will then enable us (in Section 5) to account for the connection between pastness and counterfactuality in the epistemic uses of the SP and IP. 4. Dialogism and verbal tenses 4.1

The notion of dialogism

This notion, originally introduced by Bakhtin (1977, 1984), gave rise to several linguistic theories of enunciation that refer either to ‘dialogism’ or ‘polyphony’ (cf. Ducrot 1984, Nølke et al. 2004, Bres et al. 2005, Bres & Mellet 2009, Birkelund et al. 2009). In the following analysis, we will adopt the dialogic model developed by Bres (Bres 1999, 2001; Bres & Vérine 2002, Bres & Nowakowska 2005), which he uses to account for the pragmatics of some French verbal tenses (Bres 2003, 2005, 2009). The fundamental idea underlying the notion of dialogism (or ‘polyphony’) is ‘the constituent orientation’ of any utterance towards other utterances (Bres & Nowakowska 2005: 139), that is, no utterance is produced from scratch, but always presupposes previous utterances with which it dialogues. Hence Bres considers dialogism as an ‘inner dialogue’ within an utterance, as opposed to the external dialogue of turn-takings in conversation. More precisely, dialogism refers for him to: la capacité de l’énoncé à faire entendre, outre la voix de l’énonciateur, une (ou plusieurs) autre(s) voix qui le feuillettent énonciativement ‘the capacity of the utterance to imply, besides the voice of the enunciator-speaker, one or several other voices that also contribute to the utterance’ (Bres 2001: 83)

When temporality conveys modality 

The notion of dialogism further emphasizes the different roles of speaker in the utterance act. For Bres (2003: 113), the speaker takes on both the functions of ‘locutor’ and ‘enunciator’: – the ‘locutor’ is responsible for the locutory dimension of an utterance act, i.e. for its material generation; – the ‘enunciator’ is responsible for the enunciative interpretation of the utterance, i.e. he determines, as the subjective source of the utterance, the interpretation of deictic and modal expressions. To describe dialogic utterances, Bres calls for Bally’s distinction between modus and dictum. For Bally (1965: 36–38), each utterance can be analysed in two elements: – the dictum, which is a representation corresponding to the propositional content of an utterance; – the modus,16 which corresponds to the ‘reaction’ or attitude of the ‘modal subject’ (or ‘enunciator’) towards the dictum. Then, for Bres, an utterance is ‘monologic’ when the modus directly applies to a dictum. This definition can be captured by equation (22) and is exemplified in (23): (22) monologic utterance = { modus + [dictum] } (23) Les trois otages des Khmers rouges ont été assassinés. the three hostages of the Khmer rouge have been murdered ‘The three western hostages of the Khmer rouge have been murdered’ (Bres 2001: 85) Here, the dictum [the three western hostages be murdered] is associated with an assertive modus. (23) is therefore monologic. On the contrary, an utterance is ‘dialogic’ when the modus does not directly apply to a dictum, but to an item that already has the status of an utterance, i.e. a dictum that is already assigned a modus. In other words, in a dialogic utterance, what is in the scope of the modus, is not a dictum alone, but a dictum already ‘modalized’ by another modus. Dialogic utterances can be characterized by equation (24) and (25) offers an illustration: (24) dialogic utterance = { modus2 + { modus1 + [dictum] } } (25) Les trois otages des Khmers rouges ont bien été assassinés. the three hostages of the Khmer rouge have indeed been murdered ‘The three hostages of the Khmer rouge have indeed been murdered’ (Bres 2001: 85) 16. Bally’s modus can be compared to Palmer’s propositional modality (2001).

 Adeline Patard

Here, the utterance is dialogic because the adverb bien (‘indeed’) presupposes a previous utterance, corresponding to example (23), which it confirms. In other words, bien conveys an epistemic modality (modus2) – some certainty about the situation – that is not assigned to a dictum, but to a presupposed utterance, i.e. a modus1 – an assertive modality – already attributed to a dictum. Finally Bres (2005: 23) suggests analysing dialogism as an enunciative splitting. Accordingly he discriminates two enunciative sources: – the ‘primary enunciator’ or ‘enunciator-speaker’ called E1 who is responsible for the utterance; the generated utterance (E) is characterized as the ‘embedding utterance’; – one or more ‘secondary enunciators’ called e1 whose enunciations (e) can also contribute to generate E; these enunciations are referred to as the ‘embedded utterances’. Hence, in example (24), the adverb bien (‘indeed’), presupposes besides the utterance E uttered by E1, another utterance e uttered by a secondary enunciator e1, which corresponds to (23). This characterization of dialogic utterances is summarized in Figure 1. We wish to emphasize that the embedded utterance (e) of a dialogic utterance is not necessarily an utterance that has been actually uttered by a locutor. Utterance e is only a presupposed utterance that is distinct from the embedding utterance, though contributing to it. For instance, the embedded utterance (e) can refer to a belief, a thought or an opinion ascribed to the hearer or to a third person (e.g. the doxa). Thus, in the dialogic theory elaborated by Bres, utterance acts are not restricted to acts possessing a locutory dimension, but include any enunciative act corresponding to the application of a modus to a dictum. E1: {E}

modus2 +

e1: {e}

modus1 + [dictum]

Figure 1.â•‡ Enunciative structure of a dialogic utterance

When temporality conveys modality 

4.2

The dialogic uses of tenses

As noticed by some authors (inter alia Donaire 1998, Haillet 2002, Mellet 2000, Vuillaume 2001, Bres 2009, Patard 2007, Patard & Vermeulen 2010), some tenses may have a dialogic interpretation. In these cases, they indicate the presence of a secondary enunciator e1 responsible for an embedded utterance e. This is typically the case for the SP and the IP in sequence of tenses: (26) a. Somebody told me you had a boyfriend who looked like a girlfriend. (The Killers) b. Tu m’ as dit qu’ t’ en avais assez. you me have told that you of it have-pst.ipfv enough ‘You told me that you’d got enough’ (Les cowboys fringuants) Here the SP (had, looked like) and the IP (avais) can be considered dialogic because they signal, thanks to their temporal meaning [past], that a secondary utterance act has taken place in the past. In a context of indirect speech, this secondary utterance act corresponds to the speech act made linguistically explicit by the syntactic embedding and the reporting verbs told and as dit. Put differently, the past reference point denoted by the SP and the IP (cf. supra 1.2), which usually functions as a topic time and aspectual vantage point, here serves to refer to a secondary enunciator e1. Hence, by locating this secondary enunciator e1 in the past, the SP and the IP mark the existence of two distinct enunciative sources. This enunciative splitting may be analysed as follows: – a primary enunciator (or ‘locutor-enunciator’) E1 is responsible for the embedding utterances E (26a) and (26b); – besides this primary enunciator, a secondary enunciator e1 is responsible for an embedded utterance e that can be reconstructed as: {she has a boyfriend who looks like a girlfriend} for (26a) and {j’en ai assez} (‘I’ve got enough’) for (26b); this secondary enunciator e1 is referred to in (26a) by somebody and in (26b) by tu. We may further notice that, what is situated in the past by the dialogic SP or IP, is not the described situation itself, but its utterance by a secondary enunciator e1. As a consequence, the situation is not necessarily the case in the past, but can also hold in the present or in the future, as show examples (27):17 17. Note that the progressive forms is obligatory here to get a dialogic interpretation of the SP. Without the progressive form, the SP entails a relative reading, i.e. locates the situation in the past of the secondary utterance act denoted by the reporting verbs (e.g. John told me that he came (the day before)). This fact will be explained at the end of the section.

 Adeline Patard

(27) a. John told me yesterday that he was coming (today/tomorrow). b. Jean m’a dit hier qu’il venait (aujourd’hui/demain). Finally, we wish to underline that not every tense can exhibit dialogical uses. Indeed, dialogic interpretations of tenses, i.e. the fact that the reference point denoted by a tense refers to a secondary enunciator e1, is determined by temporal and aspectual constraints.18 When the dialogic interpretation presupposes a past utterance act, as is the case with the SP and the IP, the tenses should be able to recreate in the past the same aspectual conditions that hold for a present time utterance. Now, it has been observed that the combination of a present tense with a perfective aspect is functionally infelicitous (Comrie 1976, Bybee et al. 1994: 83, Gosselin 1996: 86, Giorgi and Pianesi 1997, Smith 2007). Consequently, when the utterance act is past, the same incompatibility holds. This predicts that the situation cannot be viewed perfectively from the past reference point corresponding to the secondary utterance time. It ensues that the past tenses producing a perfective interpretation cannot have a dialogic reading. Let us consider examples (27’) derived from utterances (27): (27′) a. John told me yesterday that he came (the day before). b. Jean dit qu’ il vint (le jour précédent).19 John said that he come-pst.pfv (the day preceding) ‘John said that he came (the preceding day)’ In (27’a), the SP conveys a perfective viewpoint with come because eventive verbs generally triggers a perfective reading with the SP (Quirk et al. 1985, Smith 1991, Leech 2004). Since a perfective viewpoint is not compatible with an utterance act (be it present or past), the situation is interpreted as being the case at a preceding time (e.g. the day before). Consequently the past reference point denoted by the SP does not equate anymore the position of the secondary enunciator e1, but is anterior to it. Thus the SP cannot be dialogically interpreted. In order to get the dialogic reading, the SP must be combined with the progressive form (cf. (27a)). Indeed, by presenting the situation as ongoing (from the reference point), the progressive form imposes an imperfective viewpoint on the situation compatible with the dialogic interpretation. Similarly, the French past perfective in (27’b) expresses a perfective viewpoint that cannot be simultaneous with the past utterance act. It follows that the past 18. Cf. Patard (2007) and Bres (2009) for a detailed analysis of the origin of the dialogic interpretation of tenses. 19. We modified a bit utterance (27) to make possible the use of the passé simple by replacing the passé composé in the matrix clause (a dit) by another passé simple (dit). The sentence obtained sounds much more formal than with the passé composé, but is still perfectly correct.

When temporality conveys modality 

reference point denoted by the perfective past is seen as prior to the position of the secondary enunciator e1. As a result, the interpretation is monologic: the situation is said to have occurred before the utterance act. In sum, a past tense may have a dialogic interpretation if the past reference point refers to a secondary enunciator e1. The SP and the IP can have dialogic uses, as opposed to the French past perfective, because they allow for an imperfective viewpoint on the situation that may trigger (under specific conditions) a dialogic interpretation. 5. The conditional and the optative uses of the SP and the IP We focus in this section on two epistemic uses of the SP and the IP, namely the conditional and optative uses. We first point out some facts about the conditional and optative uses in English and French that any account should seek to explain. Then we give a brief description of the conditional and optative constructions in English and French. Finally, thanks to this description, we present a dialogic analysis that suggests an enunciative connection between past reference and counterfactuality in these particular contexts. 5.1

Preliminary observations

a. The first fact to be explained is the exact nature of the epistemic modality conveyed by the SP and the IP in their conditional and optative uses. As previously noticed, the SP and the IP in their conditional use do not necessarily entail that the situation is contrary to facts, they may also present it as possible (though improbable). Examples (29) and (30) illustrate both cases: (29) a. If I had longer arms I would push the clouds away (Task Force) b. Moi si j’ étais un homme, je serais capitaine me if I be-pst.ipfv a man, I would be captain ‘If I was a man, I would be a captain’

(D. Tell)

(30) a. And if you left me I would suffer a great deal. (R. Jaffe, After the Reunion) b.

Si un jour tu partais sans retour / if one day you leave-pst.ipfv without return / les fleurs perdraient leur parfum the flowers would lose their perfume ‘If someday you left for ever / the flowers would lose their perfume’ (É. Piaf)

 Adeline Patard

Some authors (James 1982, Martin 1991, Gosselin 1999, Iatridou 2000) have noticed that the contrary-to-fact reading is not triggered by the past forms, but by other contextual elements, such as the speaker’s knowledge about the situation’s reality and the present or future anchorage of the situation (which is partly determined by the actionality of the verb, cf. Martin 1991, Gosselin 1999). The same observations can be made for the optative usage of the SP and the IP, even though the interpretation is most frequently contrary to facts (due to the same aforementioned parameters): (31) a. If only I could win the lottery! b. Si seulement je gagnais au loto! if only I win-pst.ipfv at the lottery ‘If only I won the lottery!’ One can conclude that the SP and the IP do not involve a complete absence of reality in their conditional and optative uses, but rather convey “a greater degree of distance with reality” (James 1982: 388). b. The second observation concerns the temporal grounding of the situation. In French, the IP can only have a non-past interpretation (Iatridou 2000). Thus, depending on the context (Martin 1991, Gosselin 1999), the situation belongs to the present (cf. (29b)) or to the future (cf. (30b) and (31b)). The same remark can be made for English: the denoted situation can be either present (cf. (29a) and (31a)) or future (cf. (30a)). c. A third set of phenomena has been referred to by Iatridou (2000) as ‘fake aspect’ and concerns the fact that, in some languages like French, the imperfective tense in conditionals and optatives is not necessarily interpreted imperfectively, but can also be associated with perfective readings. For instance, in examples (30b) and (31b), the supposed situations, partir (‘leave’) and gagner (‘win’), are considered as completed. By contrast, as noticed by Iatridou (2000: 257), the English progressive form is never fake in the same contexts, but always presents the situation as ongoing: (32) a. If I was marrying a beautiful girl like Nancy, I would be nervous. (C. Duff Scott, Nancy’s Unconditional Love) b. I wish you were telling the truth. (H. Manton Lodge, Plain Jayne) This contrast between the French imperfective tense and the English progressive form should also be explained. d. The last observation concerns the paradigmatic and syntagmatic relations of the SP and IP with other tense forms. First, we may observe that the SP and the IP cannot be replaced by the other tenses that may also refer to a past situation

When temporality conveys modality 

in English and in French, namely the English present perfect and the French passé simple (past perfective) and passé composé (present perfect). These incompatibilities are exemplified in (30’) and (31’) derived from (30) and (31): (30′) a. And if you *have left me I would suffer a great deal b.

Si un jour tu *partis /*es parti sans retour if one day you leave-pst.pfv /leave-prs.prf without return / les fleurs perdraient leur parfum / the flowers would lose their perfume ‘If someday you had left/have left for ever / the flowers would lose their perfume’

(31’) a. If only I *have won the lottery! b. Si seulement je *gagnai /*ai gagné au loto! if only I win-pst.pfv /win-prs.prf at the lottery ‘If only I had won/have won the lottery!’ Note that the French passé simple becomes possible in counterfactual conditionals if the conditional tense in the apodosis is replaced by a second passé simple: (33)

Si elle acquiesça à sa demande [de mariage], if she acquiesce-pst.pfv to his proposal cf marriage ils furent (à coup sûr) les plus heureux des humains. they be-pst.pfv for sure the happiest of human beings ‘If she accepted his marriage proposal, they were for sure the happiest human beings’ (Voltaire, Candide < Leeman 2001: 223)

One may conclude that the passé simple is not incompatible with conditionals expressing a hypothesis, but with the conditional tense in the apodosis. The same remarks can be made for the English present perfect and the French passé composé. These tenses can occur in the protasis provided that a present (simple or perfect) form substitutes for the would form/conditional tense in the apodosis: (34) a. If you have waited until now to start gift shopping, you probably have missed out on the best deals. (Ebony, December 1998) b.

Si tu as perdu une méchante langue, If you have-prs lose.ptcp a nasty tongue beaucoup gagné. tu as you have-prs a lot win-ptcp ‘If you’ve lost a nasty tongue, you’ve won a lot’ (F. Petrarca De Grenaille, Le sage résolu contre la fortune et contre la mort)

 Adeline Patard

We may deduce from this that the English present perfect and the French passé composé are not incompatible, either with counterfactual conditionals, but with a would form/conditional tense in the apodosis. We may finally remarks that the non-past counterfactual reading of the SP and the IP requires the use of the English would form and the French conditional tense in the apodosis. If, instead of the latter, another verb form occurs, the non-past interpretation is not possible any more. For instance, if a second SP or IP stands in the place of the would form/conditional tense in the apodosis, the situation necessarily belongs to the past: (35) a. If he was spying on me, he didn’t see anything. (E. Payno, The Bandits from Rio Frio) b.

S’ il était l’ ami de d’Artagnan, if he be-pst.ipfv the friend of d’Artagnan il était l’ ennemi du cardinal. he be-pst.ipfv the enemy of the cardinal ‘If he was d’Artagnan’s friend, he was the enemy of the cardinal’ (Dumas, Les trois mousquetaires)

This seems to show that the non-past counterfactual interpretation of conditionals, and so the epistemic reading of the SP and the IP, is to be connected with the use of the would form/conditional tense in the apodosis. In the next section, we seek to provide an explanation for those facts while presenting our analysis of the conditional and optative uses of the SP and the IP. 5.2

A dialogic account

To account for the conditional and optative uses of the SP and the IP, we first need to analyse the semantic contribution of the different elements making up the conditional and optative constructions. By means of this description, we then try to show that the SP and the IP bring about a dialogic interpretation in these uses and that the epistemic effect observed (of lesser probability) is partly derived from this dialogic interpretation. 5.2.1 A brief description of the conditional and optative constructions In the conditional construction, three elements seem to contribute more or less directly to the counterfactual interpretation of the SP and the IP. The first element to consider is the conjunctions if and si. These conjunctions give access to a possible world, either distinct or unrelated to the real world (Declerck & Reed 2001), or, in terms of mental spaces, these conjunctions are space-builders which set up a hypothetical mental space distinct from the speaker’s base space (Fauconnier 1994,

When temporality conveys modality 

Dancygier & Sweetser 2005). So if and si enable the speaker to consider the situation described in the protasis without committing herself to its reality (cf. Vairel 1982, Caudal & Roussarie 2005, Monte 2009). The second element is the conditional construction itself. There is an extensive literature on the subject, but, for the purpose of our analysis, we only retain the following fact: what is asserted by a conditional sentence is the p → q relationship. Following Vairel (1982), Cornulier (1985) and Dancygier & Sweetser (2005), among others, we assume that this link between p and q is fundamentally causal or implicational in nature: the situation denoted by p logically entails the situation denoted by q.20 This link connecting the antecedent p to the consequent q will prove to be crucial in accounting for the epistemic interpretation of the SP and the IP in counterfactual conditionals (cf. Section 4.2.2). The last element is the verb form in the apodosis, namely the would form in English and the conditional tense in French. The conditional tense has been extensively studied in French and most accounts today agree on the dialogic or polyphonic nature of this tense (Abouda 1997, Donaire 1998, Vuillaume 2001, Haillet 2002, Bres 2009, Patard & Vermeulen 2010). According to this view, the conditional tense marks that the denoted situation is not directly asserted by the speaker, but is considered from the viewpoint of a (past) secondary enunciator e1. It follows that the speaker does not commit herself to the reality of the situation but presupposes a secondary enunciator e1 who is actually responsible for the utterance of e. The dialogic or polyphonic meaning of the conditional tense is best illustrated in its evidential use: (36)

Hakimullah Mehsud serait mort Hakimullah Mehsud be-cond dead selon un responsable américain. according to a leader american ‘Hakimullah Mehsud is dead according to an American political leader’ (AFP)

Here the conditional tense serait (‘would be’) indicates that the utterance e {Hakimullah Mehsud be dead} is not that of the speaker (the AFP), but that of another enunciator e1 (un responsable américain, ‘an American political leader’).21 The same analysis may be suggested for the English would form in some of its uses. First, morphologically speaking, the would form exhibits the features of a 20. In addition to connecting two states of affairs, conditional sentences may also link a state of affairs with an utterance (i.e., in ‘Austinian’ conditionals, cf. Corminboeuf 2010) or two utterances (cf. Vairel 1982, Monte 2009). 21. AFP stands for Agence France Presse.

 Adeline Patard

future-in-the-past, like the French conditional tense. Then, some authors have argued that the dialogic nature of the French conditional is directly related to its basic meaning of future-in-the-past (Bres 2009, Patard & Vermeulen 2010), so this could be the same for the English would form. Moreover, the English would form is clearly dialogic in some uses, such as in sequence of tenses: (37) The horse told the boy that he would help him get out of there. (V. Barnouw, Wisconsin Chippewa Myths and Tales) In this example, would signals that the utterance {I will help you get out of there} has been uttered by a past secondary enunciator e1 (the horse). Accordingly, we may hypothesize that the English would form has acquired a dialogic meaning in the uses it has in common with the French conditional tense. In such uses, it thus presupposes a past secondary enunciator e1 responsible for the utterance of the situation. As regards the optative use, it displays a common feature with the conditional use: the reality of the situation is suspended, due to different contextual elements. In the French optatives (cf. (2b) and (31b)) and in the English if-optatives (cf. (31a)), suspended reality is conveyed by conjunctions si and if (cf. supra about the meaning of si and if). In the English wish-optatives (cf. (2a), and (32b), suspended reality is lexically marked by the verb wish. We will now see in the last section how these different elements trigger a dialogic interpretation of the SP and the IP. 4.2.2 The dialogic interpretation of the SP and IP We defend the hypothesis that the SP and the IP are interpreted dialogically in their conditional and optative uses, i.e. the past reference point they denote also serves to refer to a secondary enunciator e1. The dialogical interpretation arises due to the interplay of the contextual elements described in the previous section. a. In the conditional use, three elements interact with the aspectual and temporal meaning of the SP and the IP, but the triggering parameter is the would form/conditional tense in the apodosis. We recall examples (29) for the sake of the analysis: (29) a. If I had longer arms I would push the clouds away (Task Force) b. Moi si j’ étais un homme, je serais capitaine me if I be-pst.ipfv a man I would be captain ‘If I was a man, I would be a captain’

(D. Tell)

As previously seen, we assume that the would form and the conditional tense are dialogic: they presuppose a past enunciator e1. Combined with the conditional

When temporality conveys modality 

construction, these forms call for a dialogic tense in the protasis. Indeed, the conditional construction asserts an implicational p → q relationship, whereby the situation q is presented as the consequent of the situation p (cf. Section 5.2.1). Therefore, if a past enunciator e1 is responsible for the utterance of q, the supposition of p also necessarily falls under the responsibility of e1: in order to make an assertion about p → q, the enunciator of p and q must be the same. In sum, the p → q relationship implies that, if q is dialogic, then p is also dialogic. As a consequence, the would form and the conditional tense in the apodosis require a dialogic tense able to refer to a past enunciator e1, namely a SP and IP. In this interaction, the conjunctions if and si also favour (but without imposing it) a dialogic interpretation of the used tenses insofar as they imply that the speaker does not commit herself to the situation’s reality, and so that the reality of the situation is possibly assumed by an enunciator e1. The dialogism of the SP and the IP in the protasis stems from the following enunciative splitting: – the locutor-enunciator E1 (i.e. the speaker) is responsible for the conditional utterances E corresponding to (29a) and (29b); – a past secondary enunciator e1 is responsible for the utterances e {I have longer arms} and {Je suis un homme} (‘I’m a man’). At this point, we can give an explanation of the first fact noted in Section 5.1, i.e., the sense of lesser probability conveyed by the SP and the IP in the counterfactual conditionals: lesser probability originates in the dialogic interpretation of these tenses. In fact, the speaker, by presupposing a previous utterance of p, is in a way refusing to vouch for the reality of the situation described in the protasis. The speaker thus states that the one who is actually vouching for the situation’s reality is a past enunciator. In doing so, he does not assume the reality of the situation and may appear to be questioning it, so that the situation is interpreted as unlikely. In a nutshell, by leaving the utterance of p to a past enunciator e1, the speaker keeps her distance from the reality of the situation, thus involving an epistemic judgment about the lesser probability of the situation. Several facts support the dialogic analysis of the SP and the IP in their conditional use. First it is confirmed by a linguistic test suggested by Gosselin (1999) to account for the conditional use of the IP (Lewis (1973) suggests a similar paraphrase for English).22 This test consists in using the phrase c’est vrai que (or it is true that in English) in the protasis. We then obtain examples (29’):

22. Nonetheless, Gosselin does not analyse the phenomenon triggered by the test in terms of dialogism.

 Adeline Patard

(29′) a. If it was true that I have longer arms I would push the clouds away b.

Si c’ était vrai que je suis un homme, if it be-pst.ipfv true that I be-prs a man je serais capitaine. I be-cond captain ‘If it was true that I am a man, I would be a captain’

We observe that the SP and the IP do not apply directly to the situation described by the verb, which is in the present (have, suis), but on it is true that and c’est vrai que. In other words, what the SP and the IP locate in the past, is the application of a modus to the dictum (the reality of the situation is asserted), i.e. the utterance of p (cf. supra 4.1). Hence the SP and the IP signal that the utterances {I have longer arms} and {Je suis un homme} are those of a past secondary enunciator e1 distinct from the actual speaker. This dialogical interpretation thus explains the second fact mentioned in Section 5.1, namely the non-past reading of the SP and the IP in the conditional use: what is past, is not the situation itself, but its utterance by an enunciator e1. By contrast, the English present perfect, the French passé simple and passé composé prove to be monologic in the same contexts. Let us apply the test to examples (33) and (34): (33′)

S’ il est vrai qu’ elle acquiesça à sa demande, if it be-prs true that she acquiesce-pst.pfv to his proposal ils furent (à coup sûr) les plus heureux des humains. they be-pst.pfv for sure the happiest of the human beings ‘If it is true that she accepted his marriage proposal, they were for sure the happiest human beings’

(34′) a. If it is true that you have waited until now to start gift shopping, you probably have missed out on the best deals. b.

S’ il est vrai que tu as perdu une méchante if it is true that you have-prs lose.ptcp a nasty langue, tu as beaucoup gagné. tongue you have-prs a lot win-ptcp ‘If it is true that you’ve lost a nasty tongue, you’ve won a lot’

Indeed, the passé simple, the passé composé and the present perfect do not apply to the modus expressed by it is true that and il est vrai que, but to the situation (acquiesça, have waited and as perdu), i.e. to the dictum. Hence, what they locate in the past is the situation itself, not its utterance by a secondary enunciator e1. This confirms that the passé simple, the passé composé and the present perfect cannot

When temporality conveys modality 

be dialogic (cf. supra 4.2), and explains why they are incompatible with a would form or a conditional tense in the apodosis (cf. supra 5.1). The role of the would form and the conditional tense in the dialogic interpretation is corroborated by another fact. As already noticed in 5.1, if one replaces these forms by another tense in the apodosis, e.g. a second SP or IP, then the dialogic interpretation of the protasis also vanishes, as shown by examples (35’) derived from (35): (35′) a. If it is true that he was spying on me, he didn’t see anything. b.

S’ il est vrai qu’ il était l’ ami de d’Artagnan, if it be-prs true that he be-pst.ipfv the friend of d’Artagnan il était l’ ennemi du cardinal. he be-pst.ipfv the enemy of the cardinal ‘If it is true that he was d’Artagnan’s friend, he was the enemy of the cardinal’

The SP and the IP do not apply any more to the modus (it is true that or c’est vrai que), but to the situation that is therefore located in the past. One may conclude that, without a would form or a conditional tense in the apodosis, the SP and the IP cannot be interpreted dialogically. That is why the non-past reading of the SP and the IP requires the use of a would form/conditional tense in the apodosis (cf. Section 5.1). b. We also hypothesize that the SP and the IP are dialogic in their optative use. The dialogic interpretation stems from different elements. First, the reality of the situation is suspended by the linguistic context, by means of the conjunctions if or si in if-optatives, or by the lexical item wish in wish-optatives (cf. supra 5.2). Moreover, the sentences of the form [if only p (!)] and [si (seulement) p (!)] appear in French and English as a conventional means to express wishes, so that the wished situation p is always interpreted as unreal, or at least as very unlikely.23 In wish-optatives, this job is of course done by the verb wish. Consequently, we suggest that the dialogic interpretation is triggered by the fact that the situation’s reality is questioned in the optative context. Thus this type of context requires a tense able to express “a greater degree of distance with reality” (James 1982), i.e. a dialogic SP or IP. The dialogic interpretation of the SP and the IP is confirmed by Gosselin’s test. Let us consider examples (2’): 23. It is reasonable to think that if-optatives derive from hypothetical conditionals in which the apodosis (probably resembling: it would be great/it would be so much better or ce serait tellement bien/mieux) has been elided.

 Adeline Patard

(2′) a. I wish it was true that I’m (/was) a punk rocker with flowers in my head 24 b. Ah! si c’ était vrai que je suis riche! ah if it be-pst.ipfv true that I be-prs rich ‘Ah! If only it was true that I’m rich!’ As for the conditional use, the SP and the IP do not directly apply to the situation (I be a punk rocker and je être riche) but to its modalization by a modus: it was true that and c’était vrai que. Hence the SP and the IP indicate that the utterances {I’m a punk rocker} and {Je suis riche} are those of a past secondary enunciator e1 distinct from the actual speaker. Furthermore, as the speaker refuses to vouch for the situation’s reality, the latter is seen as improbable. The last fact that remains to be explained is the ‘fake aspect’ (Iatridou 2000) to be observed in both the conditional and optative uses of the IP. The IP seems to have a fake aspect here, i.e. it does not necessarily entail an imperfective interpretation, because the aspectual vantage point serves to refer to a secondary enunciator. As a consequence, its function is not to give a certain viewpoint on the situation anymore, but to say something about the utterance of the situation. In other words, what is asserted, is not the situation’s reality (in that case an aspectual viewpoint is required), but its utterance by a past enunciator. Accordingly, since it takes on the function of an enunciator e1, the past reference point no longer plays the role of an aspectual vantage point. Nevertheless, as we saw previously (supra 4.2), the imperfective morphology is not “fake” insofar as it is required by the dialogical interpretation. For English, the picture is a bit different: the imperfective morphology (i.e. the progressive form) is always ‘real’ (supra 5.1). The reason is that the SP does not need to be combined with a progressive form to be interpreted dialogically (supra 4.2). It follows that the interpretation of a past enunciator e1 relies on the sole SP, and that a progressive form can then possibly provide an actual imperfective viewpoint on the situation (cf. (32)). 5. Conclusion The question asked in introduction was: what is the link between pastness and epistemic distance in the epistemic uses of the SP and the IP? Instead of positing a metaphorical 24. With the expression ‘it was true that’, both the simple past and the simple present are possible in the subclause. The former is more ‘standard’ in this context, but the latter may also occur, as illustrated by this attested example: I wish it was true that I do look like her. (www.ilovehateamerica.com) Here, we draw attention to the use of the simple present in the second subclause (‘I wish it was true that I’m’), because this use is revealing of the enunciative structure of the utterance.

When temporality conveys modality 

link as has been argued by some authors (cf. Imbs 1960, James 1982, Vairel 1982, Fleischman 1989), we suggest an enunciative connection by means of the notion of dialogism. Hence we have tried to demonstrate that the epistemic meaning attached to the SP and the IP in counterfactual conditionals and optatives originates in the dialogic reading of their aspecto-temporal basic meaning. Indeed, when interpreted dialogically, the SP and the IP can refer to a past secondary enunciator responsible for the utterance of the situation. In doing so, they enable the speaker to keep a distance from the situation’s reality and therefore to imply a sense of lesser probability. As a final conclusion, the SP and the IP are neither temporal nor modal, but they can combine, in interaction with different contextual parameters, temporality and modality. Such is the case in their epistemic uses where temporality conveys epistemicity. References Abouda, Lotfi. 1997. Recherches sur la syntaxe et la sémantique du conditionnel en français moderne. PhD Dissertation, Université Paris VII, France. Bakhtin, Mikhaïl. 1977. Le Marxisme et la philosophie du langage: Essai d’application de la méthode sociologique en linguistique. Paris: Minuit. Bakhtin, Mikhaïl. 1984. Les genres du discours. In Esthétique de la création verbale, 265–272. Paris: Gallimard. Bally, Charles. 1932/1965. Linguistique générale et linguistique française. Paris: Francke. Barceló, Gérard Joan, Jacques Bres, and Adeline Patard (eds.). 2006. Aspectualité, temporalité et modalité. Montpellier: Université Paul-Valéry [Cahiers de Praxématique 47]. Berthonneau, Anne-Marie & Georges Kleiber. 1994. Imparfaits de politesse: Rupture or cohésion? Travaux de linguistique 29: 59–92. Berthonneau, Anne-Marie & Georges Kleiber. 2003. Un imparfait de plus ... et le train déraillait. Cahiers Chronos 11: 1–24. Birkelund, Merete, Henning Nølke, & Rita Therkelsen (eds.). 2009. La polyphonie linguistique, Langue française 164. Bres, Jacques. 1999. Vous les entendez? Analyse du discours et dialogisme. Modèles linguistiques vol. XX (2): 71–96. Bres, Jacques. 2001. Dialogisme. In Termes et concepts pour l’analyse du discours, C. Détrie, P. Siblot & B. Vérine (eds.), 83–86. Paris: Champion. Bres, Jacques. 2003. Mais oui, il était un joli temps du passé comme les autres, le petit imparfait hypocoristique. Langue française 138: 111–124. Bres, Jacques. 2005. L’imparfait: l’un et/ou le multiple? À propos des imparfaits narratif et d’hypothèse. Cahiers Chronos: 14: 1–32. Bres, Jacques. 2006. «Encore un peu, et l’imparfait était un mode...» L’imparfait et la valeur modale de contrefactualité. Cahiers de praxématique 47: 149–176. Bres, Jacques. 2009. Dialogisme et temps verbaux de l’indicatif. Langue française 163: 21–39. Bres, Jacques, Pierre Patrick Haillet, Sylvie Mellet, Henning Nølke, & Laurence Rosier (eds.). 2005. Dialogisme et polyphonie – Approches linguistiques. Brussels: De Boeck-Duculot. Bres, Jacques & Sylvie Mellet (eds.). 2009. Dialogisme et marqueurs grammaticaux [Langue française 163]. Paris: Armand Colin.

 Adeline Patard Bres, Jacques & Aleksandra Nowakowska. 2005. Dis-moi avec qui tu «dialogues», je te dirai qui tu es... De la pertinence de la notion de dialogisme pour l’analyse du discours. Marges linguistiques 5: 137–153. Bres, Jacques & Bertrand Vérine. 2002. Le bruissement des voix dans le discours: Dialogisme et discours rapporté. Faits de langues 19: 159–169. Brisard, Frank. 2010. Aspects of virtuality in the meaning of the French imparfait. Linguistics 48: 487–524. Brown, Penelope & Stephen Levinson. 1987. Politeness: Some Universals in Language Usage Cambridge: Cambridge University Press. Bybee, Joan, Revere Perkins & William Pagliuca. 1994. The Evolution of Grammar: Tense, Aspect and Modality in the Languages of the World. Chicago & London: University of Chicago Press. Caudal, Patrick & Laurent Roussarie. 2005. Sémantique et pragmatique des propositions en Si. Cahiers Chronos 12: 51–66. Comrie, Bernard. 1976. Aspect. Cambridge: Cambridge University press. Comrie, Bernard. 1985. Tense. Cambridge: Cambridge University press. Corminboeuf, Gilles. 2010. Une description des constructions introduites par un si dit austinien. Travaux de linguistique 60: 127–144. Cornulier de, Benoît. 1985. Effets de sens. Paris: Éditions de Minuit. Dahl, Östen. 1997. The relation between past time reference and counterfactuality: A new look. In On Conditionals Again, A. Athanasiadou & R. Dirven (eds.), 97–114. Amsterdam & Philadelphia: John Benjamins. Damourette, Jean & Édouard Pichon. 1911–1936. Des mots à la pensée, vol. 5. Paris: D’Artrey. Dancygier, Barbara & Eve Sweetser. 2005. Mental Spaces in Grammar. Cambridge: Cambridge University Press. Declerck, Renaat. 1986. From Reichenbach (1947) to Comrie (1985) and beyond: Toward a theory of tense. Lingua 70: 305–364. Declerck, Renaat. 1991. Tense in English: Its Structure and Use in Discourse. London: Routledge. Declerck, Renaat. 2001. English Conditionals: A Comprehensive Empirical Analysis. Berlin: Mouton de Gruyter. Declerck, Renaat. 2006. The relation between temporal and modal uses of indicative verb forms. Cahiers Chronos 13: 215–227. De Mulder, Walter. 2004. Can there be a non temporal definition of the French imparfait? A ‘network’ approach. In Language and Revolution/Language and Time, F. Brisard (ed.), 195– 222. Antwerp: Universiteit Antwerpen. De Mulder, Walter & Frank Brisard. 2006. L’imparfait marqueur de réalité virtuelle. Cahiers de praxématique 47: 97–124. De Mulder, Walter & Carl Vetters. 2002. The French imparfait, determiners and grounding. In Grounding: The Epistemic Footing of Deixis and Reference, F. Brisard (ed.), 113–149. Berlin & New York: Mouton de Gruyter. Donaire, María Luisa. 1998. La mise en scène du conditionnel ou quand le locuteur reste en coulisses. Le français moderne 66: 204–227. Ducrot, Oswald. 1984. Esquisse d’une théorie polyphonique de l’énonciation. In Le dire et le dit, 171–233. Paris: Minuit. Fauconnier, Gilles. 1994. Mental Spaces. Cambridge: Cambridge University Press. Fleischman, Suzanne. 1989. Temporal distance: A basic linguistic metaphor. Studies in Language 13: 1–50.

When temporality conveys modality  Fleischman, Suzanne. 1995. Imperfective and irrealis. In Modality in Grammar and Discourse, J. Bybee & S. Fleischman (eds.), 519–551. Amsterdam & Philadelphia: John Benjamins. Giorgi, Alessandra & Fabio Pianesi. 1997. Tense and Aspect: From Semantics to Morphosyntax. Oxford: Oxford University Press. Givón, Talmy. 1994. Irrealis and the subjunctive. Studies in Language 18: 265–337. Gosselin, Laurent. 1996. Sémantique de la temporalité en français: Un modèle calculatoire et cognitif du temps et de l’aspect. Louvain-la-Neuve: Duculot. Gosselin, Laurent. 1999. Les valeurs de l’imparfait et du conditionnel dans les systèmes hypothétiques. Cahiers Chronos 4: 29–51. Gosselin, Laurent. 2005. Temporalité et modalité. Louvain-la-Neuve: Duculot. Gosselin, Laurent. 2009. Les modalités en français. Amsterdam & New York: Rodopi. Haillet, Pierre Patrick. 2002. Le conditionnel en français: Une approche polyphonique. Gap: Ophrys. Hogeweg, Lotte. 2009. What’s so unreal about the past? In A. Tsangalidis & R. Facchinetti (eds.), Studies on English Modality – in Honour of Frank R. Palmer, 181–208. Bern: Peter Lang. Iatridou, Sabine. 2000. The grammatical ingredients of counterfactuality. Linguistic Inquiry 31: 231–270. Imbs, Paul. 1960. L’emploi des temps verbaux en français moderne. Paris: Klincksieck. Ippolito, Michela. 2003. Presupposition and implicatures in counterfactuals. Natural Language Semantics 11 (2): 145–186. James, Deborah. 1982. Past tense and the hypothetical: A cross-linguistic study. Studies in Language 6: 375–403. Kauppinen, Anneli. 1996. The italian indicativo imperfetto compared to the Finnish conditional verb form – Evidence from child language. Journal of Pragmatics 26: 109–136. Klein, Wolfgang. 1994. Time in Language. London & New York: Routledge. Klein, Wolfgang. 2009. How time is encoded. In The Expression of Time, W. Klein & P. Li (eds.), 39–81. Berlin & New York: Mouton de Gruyter. Landeweerd, Rita & Co Vet. 1996. Tense in (free) indirect discourse in French. In Reported Speech: Forms and Functions of the Verb, Th. A. J. M. Janssen & W. van Wurff (eds.), 141–162. Amsterdam & Philadelphia: John Benjamins. Langacker, Ronald. 1978. The form and the meaning of the English auxiliaries. Language 54: 853–882. Langacker, Ronald. 1991. The auxiliary: Grounding. In Foundations of Cognitive Grammar, volume 2, 40–281. Stanford: Stanford University Press. Leech, Geoffrey. [1971] 2004. Meaning and the English Verb. Harlow: Pearson. Leeman, Danielle. 2001. Pourquoi ne peut-on combiner si et le conditionnel? In Le conditionnel en français, P. Dendale & L. Tasmowski (eds.), 211–230. Paris: Klincksieck. Le Goffic, Pierre. 1995. La double incomplétude de l’imparfait. Modèles linguistiques 16: 133–148. Lewis, David. 1973. Counterfactuals. Cambridge (Mas.): Harvard University Press. Lodge, Ken R. 1978. The use of past tense in games of pretend. Journal of Child Language 6: 365–369. Martin, Robert. 1983. Pour une logique du sens. Paris: PUF. Martin, Robert. 1991. Types de procès et systèmes hypothétiques: De l’aspect de re à l’aspect de dicto. In Les typologies de procès, C. Fuchs (ed.), 87–95. Paris: Klincksieck. Mellet, Sylvie. 2000. À propos de deux marqueurs de bivocalité: Le style indirect libre et ses contextes. Cahiers Chronos 5: 91–106. Monte, Michèle. 2009. Si marqueur d’altérité énonciative dans les si P extraprédicatives non conditionnelles. Langue française 163: 99–119.

 Adeline Patard Musatti, Tullia & Margherita Orsolini. 1993. Uses of past forms in the social pretend play of Italian children. Journal of Child Language 20: 619–639. Nølke, Henning, Kjersti Fløttum, & Coco Norén. 2004. ScaPoLine – La théorie scandinave de la polyphonie linguistique. Paris: Éditions Kimé. Ogihara, Toshiyuki. 2000. Counterfactuals, temporal adverbs, and association with focus. In Proceedings from Semantics and Linguistic Theory X (SALT10), 115–131. Ithaca: CLC Publications. Patard, Adeline. 2007. L’un et le multiple. L’imparfait de l’indicatif en français: valeur en langue et usages en discours. Phd Dissertation, Université Paul Valéry – Montpellier 3, France. Patard, Adeline. 2009. L’imparfait dans le tour [(et) si IMP?]. Revue de Sémantique et de Pragmatique 25/26: 223–242. Patard, Adeline. 2010. L’emploi préludique de l’imparfait entre temporalité et modalité: Éléments d’analyse à partir d’une étude de cas. Journal of French Language Studies 20: 189–211. Patard, Adeline & Arnaud Richard. 2011. Attenuation in French simple tense. Cahiers Chronos 22: 157–178. Patard, Adeline & Céline Vermeulen. 2010. Essai de représentation de la phrase hypothétique de forme [Si P (IMP), Q (COND)]. Cahiers Chronos 21: 215–234. Palmer, Frank. 2001. Mood and Modality. Cambridge: Cambridge University Press. Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech & Jan Svartvik. 1985. A Comprehensive Grammar of the English Language. London: Longman. Reichenbach, Hans. 1947. Elements of Symbolic Logic. New York: Free Press. Smith, Carlota S. 1991. The Parameter of Aspect. Dordrecht: Kluwer Academic Publishers. Smith, Carlota S. 2007. Tense and temporal interpretation. Lingua 117: 419–436. Soutet, Olivier. 2000. Le subjonctif en français. Paris: Ophrys. Tedeschi, Philip. 1981. Some evidence for a branching-futures semantic model. In Tense and Aspect, P. Tedeschi & A. Zaenen (eds.), 239–270. New York: Academic Press. Thieroff, Rolf. 1999. Preterites and imperfects in the languages of Europe. In Tense-Aspect, Transitivity and Causativity, W. Abraham & L. Kulikov (eds.), 141–161. Amsterdam & Philadelphia: John Benjamins. Touratier, Christian. 1996. Le système verbal français. Paris: Armand Colin. Vairel, Hélène. 1982. Les phrases conditionnelles/hypothétiques en français: La valeur de si A, B. L’information grammaticale 14: 5–10. van der Auwera, Johan & Vladimir Plungian. 1998. Modality’s semantic map. Linguistic Typology 2: 79–124 Vet, Co. 1980. Temps, aspects et adverbes de temps en français contemporain. Geneva: Droz. Vetters, Carl. 2001. Le conditionnel: Ultérieur du non-actuel. In Le conditionnel en français, P. Dendale & L. Tasmowski, (eds.), 169–207. Paris: Klincksieck. Vetters, Carl & Walter De Mulder. 2003. Sur la narrativité de l’imparfait. In Mémoire en temps advenir: Hommage à Theo Venckeleer, A. Vanneste, P. De Wilde, S. Kindt & J. Vlemings (eds.), 687–702. Leuven & Paris: Peeters. Vuillaume, Marcel. 2001. L’ expression du futur dans le passé en français et en allemand. In Le conditionnel en français, P. Dendale & L. Tasmowski (eds.) 169–207. Metz: Université de Metz/Paris: Klincksieck. Warnant, Léon. 1966. «Moi, j’étais le papa...»: L’imparfait préludique et quelques remarques relatives à la recherche grammaticale. In Mélanges de grammaire offerts à M. Maurice Grévisse pour le trentième anniversaire du Bon usage, 343–366. Gembloux: Duculot. Wilmet, Marc. 2003. Grammaire critique du français. Brussels: Duculot.

Name index A Abouda, Lotfiâ•‡ 301, 307 Abraham, Wernerâ•‡ 3, 14, 217, 234, 245, 310 Abusch, Doritâ•‡ 224, 245 Aikhenvald, Alexandra Y.â•‡ 117, 129, 130, 132, 253, 276 Aksuâ•‚Koç, Ayhanâ•‡ 110, 120, 123, 130 Anderson, Lloyd B.â•‡ 109, 130 Anstey, Matthew P.â•‡ 114, 130 Austin, John L.â•‡ 59, 84, 269, 276 B Babić, Stjepanâ•‡ 159, 163–165, 167, 179 Bach, Kentâ•‡ 259, 261, 276 Bakhtin, Mikhaïlâ•‡ 13, 14, 292, 307 Bally, Charlesâ•‡ 293, 307 Barbiers, Sjefâ•‡ 217, 245 Barceló, Gérard Joanâ•‡ 2, 3, 14, 148, 157, 281, 307 Barentsen, Adrieâ•‡ 238, 242, 243, 245 Barić, Eugenijaâ•‡ 159, 160, 178 Barsalou, Lawrence W.â•‡ 65, 84 Bergen, Benjaminâ•‡ 65, 84 Berthonneau, Anne-Marieâ•‡ 289, 290, 307 Berthoz, Alainâ•‡ 193, 211 Bhat, Darbhe Narayana Shankaraâ•‡ 118, 130, 235, 245 Biber, Douglasâ•‡ 22, 44, 272, 276 Boogaart, Ronnyâ•‡ 3, 10, 11, 14, 217, 219, 223, 224, 226, 231, 232, 238, 246 Borst, Grégoireâ•‡ 197, 212 Bres, Jacquesâ•‡ 2, 13, 14, 148, 157, 223, 246, 281, 286, 289, 292–296, 301, 302, 307, 308 Bricyn, V. M.â•‡ 240, 246 Brisard, Frankâ•‡ 1, 3–5, 7, 13–16, 45, 46, 65, 66, 75, 84, 85, 90, 91, 97, 106, 107, 109–111, 126, 127, 131, 132, 139, 157, 158, 181, 230, 231, 246, 282, 308

Broccias, Cristianoâ•‡ 182, 211 Brown, Penelopeâ•‡ 290, 291, 308 Bühler, Karlâ•‡ 88, 89, 106 Bulygina,Tatiana.V.â•‡ 236, 246 Bybee, Joanâ•‡ 1, 15, 109, 131, 132, 246, 296, 308, 309 C Caenepeel, Mimoâ•‡ 223, 246 Carston, Robynâ•‡ 255, 261, 276 Caudal, Patrickâ•‡ 301, 308 Chang, Nancyâ•‡ 202, 212 Chung, Sandraâ•‡ 1, 15 Clark, Billyâ•‡ 15, 262, 277 Coates, Jenniferâ•‡ 217, 246 Cocude, Margueriteâ•‡ 192, 197, 212 Cohen, Davidâ•‡ 182, 183, 212 Comrie, Bernardâ•‡ 109, 110, 120, 131, 160, 178, 182, 183, 200, 212, 220, 226, 246, 279, 281, 296, 308 Condoravdi, Cleoâ•‡ 3, 15 Corminboeuf, Gillesâ•‡ 301, 308 Croft, Williamâ•‡ 3, 15 Culicover, Peter W.â•‡ 250, 262, 276 Cutrer, Michelle L.â•‡ 3, 15, 225, 246 D Dahl, Östenâ•‡ 1, 2, 15, 109, 130–132, 160, 178, 179, 282, 283, 308 Dalewska-Greń, Hannaâ•‡ 168, 178 Damourette, Jeanâ•‡ 282, 308 Dancygier, Barbaraâ•‡ 288, 301, 308 de Cornulier, Benoîtâ•‡ 301, 308 Declerck, Renaatâ•‡ 3–5, 15, 21, 30, 33, 35, 44, 224, 246, 281, 284, 291, 300, 308 DeLancey, Scottâ•‡ 7, 15, 110, 117, 129, 131 De Mulder, Walterâ•‡ 246, 282, 286, 308, 310

Denis, Michelâ•‡ 191, 192, 197, 212 Diewald, Gabrieleâ•‡ 94, 102, 103, 106, 143, 146, 147, 157 Dik, Simon C.â•‡ 114, 131 Diller, Anthonyâ•‡ 250, 276 Dirven, Renéâ•‡ 15, 16, 85, 107, 131, 132, 163, 179, 308 Divjak, Dagmarâ•‡ 178, 236, 241, 246 Donaire, María Luisaâ•‡ 295, 301, 308 Donhauser, Karinâ•‡ 100, 106 Dowty, Davidâ•‡ 1, 15 Ducrot, Oswaldâ•‡ 292, 308 E Ehrlich, Susanâ•‡ 223, 246 Eide, Kristin M.â•‡ 232, 246 Eisenberg, Peterâ•‡ 149, 157 Erdal, Marcelâ•‡ 129, 131 Evans, Vyvyanâ•‡ 66, 84, 190, 212 F Fabricius-Hansen, Cathrineâ•‡ 103, 106, 146, 157 Fagan, Sarah M.B.â•‡ 232, 246 Faller, Martinaâ•‡ 119, 131 Fauconnier, Gillesâ•‡ 49, 50, 84, 86, 112, 131, 133, 300, 308 Fillmore, Charlesâ•‡ 89, 106 Fleisch, Henriâ•‡ 182, 212 Fleischman, Suzanneâ•‡ 2, 15, 109, 131, 132, 219–221, 238, 246, 254, 276, 279, 281, 282, 290, 291, 307–309 Fodor, Jerryâ•‡ 276 Fortuin, Egbertâ•‡ 238, 240, 241, 246 Fritz, Thomasâ•‡ 150, 157, 276, 277 G Geld, Renataâ•‡ 8–10, 154, 159–161, 163, 167, 168, 170, 177–179 Giorgi, Alessandraâ•‡ 296, 309 Giridhar, Puttushetra Puttuswamyâ•‡ 118, 131

 Cognitive Approaches to Tense, Aspect, and Epistemic Modality Givón, Talmyâ•‡ 1, 15, 110, 131, 161, 178, 286, 309 Goffman, Ervingâ•‡ 92, 106 Goldsmith, Johnâ•‡ 64, 85, 119, 132 Gosselin, Laurentâ•‡ 3, 13, 15, 239, 246, 281, 283–287, 296, 298, 303, 305, 309 Green, Melanieâ•‡ 187, 190, 206, 212 Grice, Paulâ•‡ 263, 269, 272, 273, 276 H Habermann, Mechthildâ•‡ 147, 157 Haillet, Pierre Patrickâ•‡ 295, 301, 307, 309 Hamm, Fritzâ•‡ 252, 262, 263, 276, 277 Hanks, William F.â•‡ 89, 92, 106 Harnish, Robert M.â•‡ 259, 276 Heine, Berndâ•‡ 217, 246, 277 Hengeveld, Keesâ•‡ 114, 131, 132 Hinrichs, Erhardâ•‡ 223, 247 Hogeweg, Lotteâ•‡ 2, 15, 282, 283, 309 Hollmann, Willem B.â•‡ 182, 211 Hopper, Paul J.â•‡ 130, 133, 222, 247 Hornstein, Norbertâ•‡ 1, 15 Huddleston, Rodneyâ•‡ 22, 26, 44 I Iatridou, Sabineâ•‡ 2, 13, 15, 109, 132, 221, 238, 247, 282, 283, 288, 298, 306, 309 Imbs, Paulâ•‡ 281, 307, 309 Ingkaphirom, Preyaâ•‡ 250, 277 Ippolito, Michelaâ•‡ 281, 309 Itayama, Mayumiâ•‡ 146, 157 Iwasaki, Shoichiâ•‡ 250, 277 J Jackendoff, Rayâ•‡ 250, 262, 275–277 James, Deborahâ•‡ 2, 15, 279, 281–283, 288, 298, 305, 307, 309 Janssen, Theoâ•‡ 3, 14, 93, 95, 96, 106, 220, 247, 309 Jaszczolt, Katarzynaâ•‡ 1, 10–12, 16, 249, 254, 257–259, 261–263, 265, 266, 269, 272, 273, 277, 278 Jespersen, Ottoâ•‡ 28, 44

Johanson, Larsâ•‡ 109, 120, 123, 131, 132 Johnson, Markâ•‡ 49, 65, 85 K Kamp, Hansâ•‡ 11, 16, 222, 226, 242, 247, 250, 262, 263, 276, 277 Kanchanawan, N.â•‡ 250, 268, 269, 277 Kasper, Walterâ•‡ 143, 157 Katičić, Radoslavâ•‡ 159, 160, 164, 165, 173, 179 Kauppinen, Anneliâ•‡ 291, 309 Kim, Jaegwonâ•‡ 38, 263, 277 Kiparsky, Carolâ•‡ 41, 44 Kiparsky, Paulâ•‡ 41, 44 Kleiber, Georgesâ•‡ 289, 290, 307 Klein, Wolfgangâ•‡ 13, 16, 227, 247, 284, 285, 309 Kornfilt, Jaklinâ•‡ 120, 132 Korytkowska, Małgorzataâ•‡ 177, 179 Koseska-Toszewa, Violettaâ•‡ 177, 179 Kosslyn, Stephen M.â•‡ 181, 190–193, 197, 209, 212, 213 Kotin, Michail L.â•‡ 147, 152, 157 L Lakoff, Georgeâ•‡ 49, 85, 190, 212 Landeweerd, Ritaâ•‡ 225, 247, 309 Langacker, Ronald W.â•‡ 2, 3, 5–7, 9, 16, 45, 46, 49, 51, 53, 64, 66–68, 72, 73, 75, 76, 78, 79, 83–85, 88–92, 96–99, 103, 104, 106, 107, 111, 127, 129, 132, 137, 139–142, 157, 158, 165, 173, 179, 181, 182, 190, 193–195, 199–203, 208, 209, 212, 282, 309 Larreya, Paulâ•‡ 2, 16 Laury, Ritvaâ•‡ 92, 107 Lazard, Gilbertâ•‡ 110, 132 Leech, Geoffreyâ•‡ 21, 44, 276, 290, 296, 309, 310 Leeman, Danielleâ•‡ 299, 309 Lee, Paul U.â•‡ 192, 209, 213 Le Goffic, Pierreâ•‡ 282, 309 Levinson, Stephen C.â•‡ 260, 277, 290, 291, 308 Lewis, Davidâ•‡ 120, 132, 290, 303, 309 Lewis, Geoffrey. L.â•‡ 120, 132, 290, 303, 309

Lindstedt, Joukoâ•‡ 160–162, 168, 172, 179 Löbner, Sebastianâ•‡ 228, 247 Lodge, Ken R.â•‡ 291, 298, 309 Ludlow, Peterâ•‡ 2, 16, 251, 254, 277 Lyons, Johnâ•‡ 22, 23, 44, 110, 132 M Martin, Robertâ•‡ 2, 16, 150, 154, 155, 239, 247, 283, 286, 298, 309 Matlock, Teenieâ•‡ 65, 86 Matsumoto, Yoâ•‡ 49, 86 McCarthy, Johnâ•‡ 182, 212, 213 McTaggart, John E.â•‡ 12, 16, 251, 253, 277 Mellet, Sylvieâ•‡ 2, 16, 292, 295, 307, 309 Mellor, David H.â•‡ 251, 252, 277 Menges, Karl Heinrichâ•‡ 123, 132 Michaelis, Lauraâ•‡ 3, 16 Mithun, Marianneâ•‡ 110, 132 Moens, Marcâ•‡ 252, 277 Moguš, Milanâ•‡ 164, 165, 179 Monte, Michèleâ•‡ 301, 309 Mortelmans, Tanjaâ•‡ 8, 16, 97, 100–104, 107, 137, 140, 143, 146, 147, 157, 158, 246 Muansuwan, Nuttanartâ•‡ 250, 269, 277 Mürvet, Ençâ•‡ 276 Musatti, Tulliaâ•‡ 291, 310 N Nichols, Johannaâ•‡ 7, 16, 110, 117, 130, 131, 133 Nicolle, Steveâ•‡ 262, 277 Nølke, Henningâ•‡ 292, 307, 310 Noss, Richard B.â•‡ 250, 277 Noveck, Ira A.â•‡ 260, 277 Nowakowska, Aleksandraâ•‡ 292, 308 Núñez, Rafael E.â•‡ 49, 85 Nuyts, Janâ•‡ 86, 114, 132, 133 O Ogihara, Toshiyukiâ•‡ 288, 310 Orsolini, Margheritaâ•‡ 291, 310 Oversteegen, Eleonoreâ•‡ 223, 247 P Padučeva, Elenaâ•‡ 241, 242, 247 Pagliuca, Williamâ•‡ 15, 109, 131, 308

Name index  Palmer, Frank R.â•‡ 15, 16, 109, 110, 133, 269, 277, 286, 293, 309, 310 Partee, Barbaraâ•‡ 228, 247 Patard, Adelineâ•‡ 1, 2, 10, 13, 14, 16, 181, 279, 281, 288, 290, 291, 295, 296, 301, 302, 307, 310 Perkins, Revereâ•‡ 15, 109, 131, 308 Pianesi, Fabioâ•‡ 296, 309 Pichon, Édouardâ•‡ 282, 308 Pitts, Alysonâ•‡ 262, 278 Plungian, Vladimirâ•‡ 117, 133, 238, 240, 247, 253, 254, 276, 290, 310 Portner, Paulâ•‡ 1, 16 Pranjković, Ivoâ•‡ 159–161, 164, 167, 179 Prior, Arthur N.â•‡ 57, 59, 63, 229, 234, 251, 278, 297 Pullum, Geoffrey K.â•‡ 22, 44 Q Quirk, Randolphâ•‡ 22, 44, 290, 296, 310 R Radden, Günterâ•‡ 107, 159, 163, 179 Raguž, Dragutinâ•‡ 159, 163, 179 Rappaport, Gilbertâ•‡ 240, 247 Recanati, Françoisâ•‡ 255, 256, 260–262, 278 Reed, Susanâ•‡ 21, 30, 35, 44, 224, 247, 300 Reichenbach, Hansâ•‡ 13, 16, 123, 133, 142, 223, 227, 247, 252, 278, 284, 308, 310 Reyle, Uweâ•‡ 11, 16, 222, 226, 242, 247, 250, 262, 263, 277 Rohrer, Christianâ•‡ 11, 16, 226, 247 Rončević, Nikolaâ•‡ 173, 179 Roussarie, Laurentâ•‡ 301, 308 S Salkie, Raphaelâ•‡ 44, 224, 247 Sattig, Thomasâ•‡ 252, 278 Schouten, Edithâ•‡ 221, 239, 247 Silić, Josipâ•‡ 159–161, 164, 167, 179 Slobin, Danâ•‡ 110, 120, 130, 133

Šmelev, Andrej D.â•‡ 236, 246 Smirnova, Elenaâ•‡ 4, 6, 8, 17, 87, 137, 143, 147, 149, 152, 158 Smith, Carlota S.â•‡ 219, 247, 251, 278, 281, 284, 285, 296, 310 Smith, Quentinâ•‡ 219, 247, 251, 278, 281, 284, 285, 296, 310 Soutet, Olivierâ•‡ 286, 310 Sperber, Danâ•‡ 228, 248, 255, 260, 277, 278 Srioutai, Jirantharaâ•‡ 10–12, 249, 254, 271, 278 Stanojević, Mateusz-Milanâ•‡ 8–10, 159–161, 163, 167, 170, 177–179 Steedman, Markâ•‡ 252, 277, 278 Stojićević, Aleksandarâ•‡ 167, 173, 179 Stowell, Tim A.â•‡ 3, 17, 224, 228, 247 Supanvanich, I.â•‡ 250, 268, 278 Sweetser, Eveâ•‡ 49, 68, 78, 84, 86, 111, 112, 116, 133, 288, 301, 308 Sysoeva, Annaâ•‡ 262, 278 T Talmy, Leonardâ•‡ 15, 49, 67, 68, 78, 86, 131, 178, 309 Tedeschi, Philipâ•‡ 287, 310 Temürcü, Ceyhanâ•‡ 4, 7, 109, 111, 114, 126, 133 Težak, Stjepkoâ•‡ 159, 163–165, 167, 179 Thieroff, Rolfâ•‡ 132, 146, 155, 158, 279, 281, 310 Thomason, Richmond H.â•‡ 254, 278 Timberlake, Alanâ•‡ 1, 15, 247 Tošović, Brankoâ•‡ 161, 179 Touratier, Christianâ•‡ 282, 310 Trnavac, Radoslavaâ•‡ 3, 10, 11, 17, 217, 236, 238–240, 246, 247 Tura-Sansa, Sabahatâ•‡ 120, 126, 133 Turner, Markâ•‡ 49, 50, 84–86, 112, 131, 181, 277 Tversky, Barbaraâ•‡ 181, 192, 203, 208, 209, 211–213

U Underhill, Robertâ•‡ 120, 133 V Vairel, Hélèneâ•‡ 301, 307, 310 van der Auwera, Johanâ•‡ 44, 132, 238, 240, 247, 253, 254, 276, 290, 310 van Lambalgen, Michielâ•‡ 252, 262, 263, 276, 277 Vater, Heinzâ•‡ 107, 146, 154, 155, 158, 247 Vendler, Zenoâ•‡ 56, 86 Verhagen, Arieâ•‡ 11, 17, 93, 107, 224, 230, 232, 248 Vérine, Bertrandâ•‡ 14, 292, 307, 308 Vermeulen, Célineâ•‡ 288, 295, 301, 302, 310 Vet, Coâ•‡ 15, 132, 225, 247, 281, 309, 310 Vetters, Carlâ•‡ 286, 308, 310 Vuillaume, Marcelâ•‡ 295, 301, 310 W Wada, Naoakiâ•‡ 220, 227, 248 Warnant, Léonâ•‡ 291, 310 Weinrich, Haraldâ•‡ 149, 158 Wilmet, Marcâ•‡ 281, 310 Wilson, Deirdreâ•‡ 228, 248, 255, 278 Woisetschlaeger, Erichâ•‡ 64, 85, 119, 132 Woodbury, Anthonyâ•‡ 110, 119, 133 Wraga, Maryâ•‡ 190–192, 209, 213 Y Yavaş, Feryaâ•‡ 110, 120, 123, 125, 133 Z Žic Fuchs, Milenaâ•‡ 168, 179 Ziegeler, Debraâ•‡ 148, 158 Zovko Dinković, Irenaâ•‡ 168, 178

Subject index A actualityâ•‡ 112, 115 (see also virtuality) actualizationâ•‡ 22–23 actual planeâ•‡ 174–177 affixationâ•‡ 186–187 Aktionsart (or actionality)â•‡ 55– 57, 183, 200–203, 219, 235 anaphoraâ•‡ 218–219, 225–229, 232, 234–235, 241–244 anchoringâ•‡ 23, 26–27, 111–112, 127, 130 domains ofâ•‡ 111, 127, 130 anchoring categoriesâ•‡ 113 anchoring relationsâ•‡ 111–112, 127, 130 Arabicâ•‡ 182–190, 195–200, 203–211 aspect-marking inâ•‡ 184–190 A-series/B-series oppositionâ•‡ 251–252 aspectâ•‡ 112, 168, 172, 181, 192–203, 208–209, 211, 252, 285 (see also inchoation, continuation, cumulation, duration, gradualness, habituality, iterativity, nearness, partitioning, punctuality, termination aorist)â•‡ (see Croatian aorist) fakeâ•‡ 221, 298, 306 imperfectiveâ•‡ 53–58, 73–75, 182–183, 200–202, 207, 217– 245, 285–286, 289, 296–297, 306 (see also imparfait, past imperfective) neutralâ•‡ 285 perfectâ•‡ 116, 219–223, 227, 232–233, 235 (see also past perfect, present perfect) perfectiveâ•‡ 53, 55–58, 74–75, 168, 172–174, 178, 182–183, 200–203, 207–208, 220– 222, 224–226, 228, 230–231, 233–244, 285 (see also past perfective, present perfective) prospectiveâ•‡ 116, 142

viewpointâ•‡ 284–287, 296–297, 306 attenuation (see tentativeness) B backshiftingâ•‡ 28 double modalâ•‡ 29 modal (or formal distancing)â•‡ 28–29, 33–34, 43 blendingâ•‡ 50 boundednessâ•‡ 54, 203 C certaintyâ•‡ 115, 118–119, 122–124 cognitionâ•‡ 66, 68, 181, 190, (see also Cognitive Grammar, default) Cognitive Grammarâ•‡ 46–56, 65–73, 88–92, 110–111, 126–127, 137, 172–178, 181–182, 193–194, 199–201, 208, 211 conceptual semanticsâ•‡ 46, 50 conditionalsâ•‡ 25, 28, 37, 143, 145, 149–151, 153, 280, 297–298, 300–305 closedâ•‡ 30, 37 counterfactualâ•‡ 30, 32, 81, 220–221, 239, 299–301 factualâ•‡ 31, 37 future less vividâ•‡ 283, 288 imperativeâ•‡ 236, 238–239 neutral theoreticalâ•‡ 30–31, 35, 42 nonneutral theoreticalâ•‡ 30–31, 34, 43 openâ•‡ 30, 32 possible world typology ofâ•‡ 30 tentativeâ•‡ 30, 32 conditional imperativeâ•‡ 236, 238–239 conditionalization (see modal) conditional tenseâ•‡ 299–303 conditions of reasonable assertionâ•‡ 143 construalâ•‡ 46–47, 92

deicticâ•‡ 93, 96 objectiveâ•‡ 140–141 speaker-exclusiveâ•‡ 94 subjectiveâ•‡ 139–141, 177 construal configurationâ•‡ 94, 99, 103, 105, 138, 140–141, 144, 148, 152–153, 155–156 maximally subjectiveâ•‡ 142, 153, 156 objectiveâ•‡ 140, 142 contextualismâ•‡ 256 continuationâ•‡ 110, 121, 204–208, 210–211 contractibilityâ•‡ 54 counterfactualityâ•‡ 25, 36–37, 219, 222, 231, 233–234, 238–239, 281, 289 (see also conditional, S-world) counter-sequentialityâ•‡ 171–172 count/mass oppositionâ•‡ 53–54 Croatian aoristâ•‡ 159–178 (see also counter-sequentiality, narrative, immediate future, generic, recent past, scheduled future) optativeâ•‡ 177 perfect tenseâ•‡ 159–161, 169–172, 177–178 pluperfect tenseâ•‡ 171–172, 177 cumulationâ•‡ 185, 206, 208, 209, 211 current relevanceâ•‡ 159–160, 177 D dative-infinitive constructionâ•‡ 239–241 defaultâ•‡ 49, 52, 58, 73, 83, 93, 101–102, 112, 147, 153, 168, 172, 174, 229, 254, 260, 265–267, 269–275, 282 (see also viewing) arrangement cognitiveâ•‡ 257–259, 261, 266, 269f, 271–273 social/culturalâ•‡ 257, 259–261

 Cognitive Approaches to Tense, Aspect, and Epistemic Modality Default Semanticsâ•‡ 249–250, 254–257, 263f deictic spaceâ•‡ 95 deictic centerâ•‡ 68, 94, 99 commonâ•‡ 93, 94, 99, 100, 102 speaker-exclusiveâ•‡ 93, 102, 105 deixisâ•‡ 88, 92 demonstrativeâ•‡ 90, 95 deonticity (see (root) modal, (root) modality) dialogismâ•‡ 223f, 292–294, 302–307 (see also imparfait, simple past, tense) dictumâ•‡ 293, 304–306 distal/proximal opposition (or immediate/non-immediate)â•‡ 68–70, 73, 82–83, 94, 97, 98, 103 (see also reality) Discourse Representation Theoryâ•‡ 250, 262–263 durationâ•‡ 184–186, 188, 196, 205–206, 208–209–211 Dutchâ•‡ 232–233, 238f dynamic evolutionary modelâ•‡ 104 E Englishâ•‡ 21–22, 26–43, 45–46, 56–68, 71–84, 90–91, 95–100, 104, 110–111, 126–127, 142, 204, 209, 217, 219–224, 226–235, 238, 241–244, 249, 253–255, 264–266, 268–269, 275, 279–283, 285, 287–288, 290–292, 295–307 (see also modal, mood, tense) enunciatorâ•‡ 293 epistemic contingencyâ•‡ 110, 126–129 epistemic controlâ•‡ 68 epistemic distanceâ•‡ 71, 73, 111, 177, 280, 306–307 epistemic domainâ•‡ 94, 100, 103, 104, 281 epistemic evaluation timeâ•‡ 231– 235, 241, 245 epistemic immediacyâ•‡ 46, 65, 127–129, 161, 167–168, 173–174, 176–178 epistemicity (see epistemic contingency, epistemic distance, epistemic immediacy, epistemic necessity, epistemic possibility, modal, modality, tense, world)

epistemic landscapeâ•‡ 69 epistemic modelâ•‡ 67–73 epistemic necessityâ•‡ 253–254, 264 epistemic pathâ•‡ 71 epistemic possibilityâ•‡ 29, 34, 264 (see also possibility) epistemic regionâ•‡ 143 epistemic scaleâ•‡ 34 specifiedâ•‡ 38 epistemic strivingâ•‡ 68 evidentialityâ•‡ 71, 117–118, 252–254, 264 F factualityâ•‡ 36–37, 43, 281, 286 (see also counterfactuality, t-world) absoluteâ•‡ 38 not-yet-â•‡ 28, 33, 33, 36 relativeâ•‡ 37 factuality valueâ•‡ 25, 27, 36, relativeâ•‡ 37 family resemblanceâ•‡ 127 fictionalityâ•‡ 49 fictivity (see fictionality, virtuality) formal distancing (see backshifting) Frenchâ•‡ 142, 220–221, 225, 230–231, 279–292, 295–307 (see also tense) future tenseâ•‡ 28, 238, 254 (see also (prospective) aspect, posteriority, werden + infinitive) future-time referenceâ•‡ 66, 124–125, 237, 254 (see also scheduled future, immediate future) G generic (or general fact)â•‡ 46, 64, 76, 115, 118–119, 122–124, 126, 128, 164 Germanâ•‡ 87–88, 94–106, 138–140, 143–146, 149, 151–154, 156 (see also modal, mood, tense) gradualnessâ•‡ 184 grammatical categoriesâ•‡ 50–52 (see also verb, noun) grammaticalizationâ•‡ 137–138, 140, 147, 152, 250 Greekâ•‡ 221

groundâ•‡ 66, 87–95, 97, 99, 100–103, 105, 138–139, 142–146, 153, 156, 173–176 (see also construal) grounded entityâ•‡ 89–91, 94, 98, 139 groundingâ•‡ 26, 93, 111f, 127, 130, 137–140, 142, 146–147, 152–154 clausalâ•‡ 66, 87 modalâ•‡ 87, 91, 97, 100, 101, 103 temporalâ•‡ 87, 97 grounding predicationâ•‡ 87–91, 95–97, 138–140, 142–144, 146–147, 149, 152–153, 156 modalâ•‡ 90 temporalâ•‡ 90, 98–100 grounding relation(ship)â•‡ 89–92, 98, 101, 104, 139 grounding systemâ•‡ 97 H habitualityâ•‡ 122–123, 189–190, 206–208, 210–211, 228–229, 231 historical present (or narrative present or past of narration)â•‡ 46, 61, 77, 99, 249–250, 263–266 homogeneityâ•‡ 54 hypothetical comparisonâ•‡ 280 hypotheticalityâ•‡ 115, 118, 238–239 I immediacyâ•‡ 112, 115 immediate experienceâ•‡ 69–70 (see also reality) immediate futureâ•‡ 167–170, 172–174 imparfaitâ•‡ 220–221, 225–226, 230–231, 279–283, 285–292, 295–300, 302–307 (see also aspect, past tense) dialogic use ofâ•‡ 295–297 epistemic use ofâ•‡ 280–281, 287–292, 302 (see also conditional, counterfactuality, hypothetical comparison, optativity, supposition, pretend game) illocutionary use ofâ•‡ 290–292 (see also suggestion, tentativeness, pretend game) modal use ofâ•‡ 286–289 temporal use ofâ•‡ 280–281, 286–289 imperative (see mood)

Subject index  inchoationâ•‡ 186, 188, 195–196, 208, 210–211 indicative (see mood) indirect speech (see reported speech) inferenceâ•‡ 117–118, 122, 128, 164, 177, 250–251, 255, 257–258, 261, 265, 269, 272–273, 275, 287 intentionalityâ•‡ 258–259, 261, 264–265, 269–270–272, 276f, 287–290, 292 instantiation, domain ofâ•‡ 53 iteration (or replication)â•‡ 184– 185, 205, 208–211 (see also habituality) L locutorâ•‡ 293 M mental imageâ•‡ 181, 190–193 mental imageryâ•‡ 181–182, 190–192, 197, 209, 211 merger representationâ•‡ 255–267, 272–275 mental spacesâ•‡ 49, 218–219, 300 metaphorâ•‡ 49–50, 307 metonymyâ•‡ 129 mirativityâ•‡ 110, 117 (see also new information) modal (or modal auxiliaries/ verbs)â•‡ 111, 140, 231–233 Englishâ•‡ 21–22, 25–28, 34, 36, 39–40, 66, 68, 78–83, 217, 219–220, 231–233, 253, 268–269 epistemicâ•‡ 34, 40, 78, 103–106, 219, 231–234, 236–237 Germanâ•‡ 96–97, 103–106, 138, 140, 143, 151–152, 156 root (or deontic)â•‡ 39–42, 78, 219, 231–234, 236–237, 267–270 Russianâ•‡ 236–237 tensedâ•‡ 78–82 Thaiâ•‡ 267–275 modal adverbâ•‡ 25, 28, 34, 41–43 modal conditionalizationâ•‡ 28–29, 34, 43 modal stateâ•‡ 26 modalityâ•‡ 21, 27, 43, 45–46, 249–250, 253–255, 263–265

alethicâ•‡ 263, 272–273 dynamicâ•‡ 267–268, 272–273, 275 epistemicâ•‡ 29, 33–34, 36, 39–40, 46, 65, 78, 97, 110, 118, 126–129, 130, 143, 145–148, 153, 161, 167–168, 173–174, 176–178, 217–222, 230–241, 243–245, 253–254, 263–264, 272– 273, 281–282, 287–292, 302–303, 306–307â•‡ (see also epistemic contingency, epistemic distance, epistemic immediacy, epistemic necessity, epistemic possibility) indeterminate epistemicâ•‡ 34, 42 participant-internalâ•‡ 238, 240–241 root (or deontic)â•‡ 34, 38–42, 78, 219, 231–234, 236–240, 254, 267–270 specified epistemicâ•‡ 34, 42 modalized propositionâ•‡ 25–26 modalizerâ•‡ 25–26, 28 modal positionâ•‡ 39 modusâ•‡ 293, 304–306 moodâ•‡ 143, 152, 239, 286 Englishâ•‡ 28, 43, 98, 238 Germanâ•‡ 88, 95, 97–98, 100–104, 106, 138, 143–146, 149, 151–154, 156 imperativeâ•‡ 43, 100, 236 (see also conditional) indicativeâ•‡ 98, 100–102, 106, 143, 152, 156, 286–287 subjunctiveâ•‡ 28, 42–43, 98, 100–103, 104, 106, 143–146, 149, 151–154, 156, 239, 286 N narrativeâ•‡ 162–163, 171, 222–223, 226, 242 nearnessâ•‡ 182, 187, 194, 208, 209, 211 necessityâ•‡ 36 new informationâ•‡ 115, 122–125 non-past (see past) nounâ•‡ 50–51 (see also count/ mass opposition, grammatical categories)

O onstage/offstage elementâ•‡ 48 optativityâ•‡ 221–222, 230–231, 233–235, 280, 298–299, 302, 305–306 (see also Croatian optative) origoâ•‡ 88, 89, 104 P partitioningâ•‡ 205, 207–210 passé composéâ•‡ 299–300, 304–305 passé simpleâ•‡ 225, 296–297, 299, 304–305 past imperfectiveâ•‡ 58, 221, 226– 231, 234 (see also imparfait) past/non-pastâ•‡ 98, 100, 287, 298, 300 past perfectâ•‡ 223, 227, 229–230, 233–234, 243 past perfectiveâ•‡ 228–234, 244 (see also Croatian aorist, passé simple) past tenseâ•‡ 57–58, 66–67, 73–74, 90, 97–98, 100, 128, 144–146, 161, 174, 250, 254–255, 281–285 (see also imparfait, passé simple, past perfect, past perfective, past imperfective, Präteritum, simple past, Croatian aorist) past-time referenceâ•‡ 58, 66, 249–250, 254–255, 266–275 (see also historical present, recent past) performativeâ•‡ 59, 74 perspectiveâ•‡ 47, 218, 223–225, 228–231, 233, 235, 241–243 (see also use of tenses) play-by-play descriptionâ•‡ 60 possibilityâ•‡ 36 (see also potentiality) openâ•‡ 32 posteriorityâ•‡ 28, 32, 34, 43 potentialityâ•‡ 112, 115, 254 pragmatic compositionÂ� alityâ•‡ 249–250, 256–257, 261–262, 266, 272, 275–276 pragmatic enrichmentâ•‡ 255–256, 260–261 (see also inference) Präteritumâ•‡ 97, 144–145, 150–151 present perfectâ•‡ 227, 233, 243, 299–300, 304–305

 Cognitive Approaches to Tense, Aspect, and Epistemic Modality present perfectiveâ•‡ 56–60, 74, 168, 218, 228, 231, 236–238, 242–243 (see also performative, play-by-play description) present tenseâ•‡ 23, 90, 110f, 230 (see also present perfective, present continuous in Turkish) Englishâ•‡ 45–46, 56–57, 59–67, 73–74, 91, 99, 110–111, 231, 235 (see also progressive) Germanâ•‡ 97–100, 147 non-present use ofâ•‡ 60–64, 75–78 (see also generic, historical present, scheduled future, script) present-time referenceâ•‡ 65–66, 110f, 120–121, 127–129, 273 pretend gameâ•‡ 291 preterit (see simple past) probabilityâ•‡ 36, 115, 117, 128 (see also (epistemic) modality) processâ•‡ 51–52, 54 profileâ•‡ 49, 51 profiled process (or event)â•‡ 51, 104–105, 143, 145–146 onstageâ•‡ 144 profiled entityâ•‡ 51, 92, 142 progressiveâ•‡ 56–57, 127, 204, 209, 217, 219, 223, 235 pastâ•‡ 226, 228–230, 233–234, 295f, 298, 306 presentâ•‡ 110–111, 126–127, 226, 230, 234 prominenceâ•‡ 48 prototypeâ•‡ 126–127 proximal (see distal) punctualityâ•‡ 184–185, 198, 203–204, 210 Q quantifierâ•‡ 90 reality (or realis)â•‡ 72, 101–102, 104, 143, 145–146, 148, 153, 286 immediate/non-immediate (see distal) knowledge ofâ•‡ 104–105 potentialâ•‡ 104 projectedâ•‡ 104 virtualâ•‡ 173 reality conceptionâ•‡ 67

recent pastâ•‡ 163, 166–173, 175, 177–178 reduplicationâ•‡ 183–184, 205 reference point (or reference time)â•‡ 88, 93, 96, 99–100, 138, 139, 142–143, 153, 177, 218, 222–223, 225–235, 242, 244–245, 284, 295–297, 302, 306 (see also anaphora possible)â•‡ 144 relevantâ•‡ 141–150, 153, 155–156 relevanceâ•‡ 114, 116 replicabilityâ•‡ 54 reported speechâ•‡ 80, 101–104, 218, 222, 224–225, 229, 233– 235, 243–244, 295–296, 302 residueâ•‡ 26 Russianâ•‡ 218, 235–245 (see also conditional imperative, dative-infinitive construction, modal) aspect marking inâ•‡ 235–236 S scanningâ•‡ 181–182, 193–194, 197, 199 monosequentialâ•‡ 202–204, 208–211 polysequentialâ•‡ 202, 204–211 sequentialâ•‡ 201–202, 207, 211 summaryâ•‡ 182, 199, 201–203, 207–211 scheduled futureâ•‡ 46, 62–63, 75, 99, 164, 167, 172, 240 schematic meaningâ•‡ 126–127 scopeâ•‡ 47–48 immediateâ•‡ 48, 173, 177–178, 208 maximalâ•‡ 48, 208 scriptâ•‡ 63–64 simple past (or preterit)â•‡ 222– 223, 226–228, 233–235, 249, 251, 279–283, 285, 287–288, 290–292, 295–300, 302–307 (see also (neutral) aspect, past tense) dialogic use ofâ•‡ 295–297 epistemic use ofâ•‡ 280–281, 287–292, 302 (see also conditional, counterfactuality, hypothetical comparison, optativity, supposition) illocutionary use of (see also suggestion, tentativeness)

modal use ofâ•‡ 286–289 temporal use ofâ•‡ 280–281, 286–289, 264–266 simulationâ•‡ 65, 181 situationâ•‡ 22 specificityâ•‡ 47 structural generalizationâ•‡ 64–65 subjectificationâ•‡ 96, 129, 138, 140–142, 146–149, 153–154, 177 extremeâ•‡ 91 subjectivityâ•‡ 138–140, 142, 152, 218–219, 222, 226, 228, 230, 232–233, 235, 238–239, 241, 243–244 (see also construal) extreme (or maximal)â•‡ 98, 144, 153, 173 subjunctive (see mood) suggestionâ•‡ 290–291 suppositionâ•‡ 280 S-worldâ•‡ 23 counterfactualâ•‡ 25 theoreticalâ•‡ 25 (see also world) symbolic grammarâ•‡ 50 T temporalityâ•‡ 251–255, 262–263, 275–276, 286–287 (see also A-series/B-series opposition, tense, future-time reference, past-time reference, presenttime reference, temporal coincidence) temporal coincidenceâ•‡ 56 tenseâ•‡ 45, 57, 73, 88, 97, 112, 146, 168, 172, 252–253, 285 (see also future tense, past/non-past, past tense, present tense, use of tenses) Croatianâ•‡ 159–178 Englishâ•‡ 45–46, 56–57, 59–68, 72–74, 91, 97–99, 110–111, 126–127, 204, 209, 279–283, 285, 287–288, 290–292, 295–307 (see also conditional tense, past perfect, present perfect, present tense, progressive, simple past) Frenchâ•‡ 279–283, 285–292, 295–307(see also imparfait, conditional tense, passé composé, passé simple)

Subject index  Germanâ•‡ 97–100, 138 (see also Präteritum, present tense, werden + infinitive, würde + infinitive) Turkishâ•‡ 120–130 tentativenessâ•‡ 290 (see also conditional) terminationâ•‡ 189–190, 195–196, 208, 211 Thaiâ•‡ 249–250, 255, 266–276 (see also modal) time (see temporality) topic timeâ•‡ 227, 284–285 Turkishâ•‡ 110, 120–130 present continuous inâ•‡ 120–130 t-worldâ•‡ 23–25 extendedâ•‡ 24 narrowâ•‡ 24 not-yet-factualâ•‡ 32 objectiveâ•‡ 23 subjectiveâ•‡ 23 U use of tenses dialogicâ•‡ 295–297 epistemicâ•‡ 120–125, 147–148, 151, 228, 240, 242–244, 287– 292, 302–307 (see also conditional, counterfactuality, hypothetical comparison, optativity, supposition, pretend game)

illocutionaryâ•‡ 290–292 (see also suggestion, tentativeness, pretend game) modalâ•‡ 228, 286–289 perspectivizedâ•‡ 225, 228–230, 233, 243–245 subjectiveâ•‡ 218–219, 222, 226, 228, 230, 241, 243–244 temporal (or objective)â•‡ 144– 147, 228–230, 241, 244, 264–266, 286–289 V vantage point (or viewpoint)â•‡ 88, 94, 99, 100, 101, 112, 176, 218, 229–230, 301 (see also (viewpoint) aspect) verbâ•‡ 50–51 (see also grammatical categories, modal) attitudinalâ•‡ 25, 28, 41–43 factiveâ•‡ 41 intensionalâ•‡ 25, 28, 34, 41, 43 world-evoking lexicalâ•‡ 41 viewing arrangement canonicalâ•‡ 141, 173 defaultâ•‡ 59–62, 93 specialâ•‡ 61–62 viewpoint (see aspect, vantage point) virtual planeâ•‡ 161, 174–178 (see also actual plane) virtualityâ•‡ 49, 91, 231, 286 (see also actuality, reality)

vowel lengtheningâ•‡ 185–186 W werden + infinitiveâ•‡ 138, 146–149, 156 würde + infinitiveâ•‡ 138, 149–156 (see also preterit subjunctive) worldâ•‡ 218 (see also S-world, t-world, verb) epistemically danglingâ•‡ 34, 39 factualâ•‡ 24 imaginaryâ•‡ 35 modalâ•‡ 25, 42–43 nonfactualâ•‡ 25, 27, 42 nonneutral theoreticalâ•‡ 30 not-yet-factualâ•‡ 33 possibleâ•‡ 23 theoreticalâ•‡ 28 Z Zeigfeldâ•‡ 88 zoomingâ•‡ 54, 182, 209, 211

Cognitive Approaches to Tense, Aspect, and Epistemic Modality (Human Cognitive Processing)

Epistemic Modality, Language, and Conceptualization: A Cognitive-Pragmatic Perspective (Human Cognitive Processing)

Methods in Cognitive Linguistics (Human Cognitive Processing)

Methods in Cognitive Linguistics (Human Cognitive Processing)

Cognitive Linguistics: Convergence and Expansion (Human Cognitive Processing)

Language Diversity and Cognitive Representations (Human Cognitive Processing)

Cognitive Linguistics: Convergence and Expansion (Human Cognitive Processing)

Epistemic Modality

Epistemic Modality

New Directions in Cognitive Linguistics (Human Cognitive Processing)

New Directions in Cognitive Linguistics (Human Cognitive Processing)

Epistemic Modality

Tense and Aspect Systems

Cognitive and communicative approaches to linguistic analysis

Cognitive and Communicative Approaches to Linguistic Analysis

Cognitive and communicative approaches to linguistic analysis

Modality and Tense: Philosophical Papers

Cognitive Archaeology and Human Evolution

Tense and Aspect in Bantu

Tense and Aspect in Bantu

Analogy as Structure and Process: Approaches in Linguistics, Cognitive Psychology and Philosophy of Science (Human Cognitive Processing, Volume 14)

Tense and Aspect in Bantu

Tense and Aspect in Bantu

Metonymy and Metaphor in Grammar (Human Cognitive Processing)

Discourse, Vision, and Cognition (Human Cognitive Processing, Volume 23)

Transdisciplinary Advancements in Cognitive Mechanisms and Human Information Processing

Motivation in Grammar and the Lexicon (Human Cognitive Processing)

Cognitive Linguistics Investigations: Across Languages, Fields And Philosophical Boundaries (Human Cognitive Processing)

Evidentiality and Epistemic Modality in Spanish (Semi-)Auxiliaries: A Cognitive-Functional Approach (Applications of Congnitive Linguistics)

Evidentiality and Epistemic Modality in Spanish (Semi-)Auxiliaries: A Cognitive-Functional Approach (Applications of Congnitive Linguistics)

Cognitive Approaches to Obsessions and Compulsions: Theory, Assessment, and Treatment

Cognitive Approaches to Tense, Aspect, and Epistemic Modality (Human Cognitive Processing)

Epistemic Modality, Language, and Conceptualization: A Cognitive-Pragmatic Perspective (Human Cognitive Processing)

Methods in Cognitive Linguistics (Human Cognitive Processing)

Methods in Cognitive Linguistics (Human Cognitive Processing)

Cognitive Linguistics: Convergence and Expansion (Human Cognitive Processing)

Language Diversity and Cognitive Representations (Human Cognitive Processing)

Cognitive Linguistics: Convergence and Expansion (Human Cognitive Processing)

Epistemic Modality

Epistemic Modality

New Directions in Cognitive Linguistics (Human Cognitive Processing)

New Directions in Cognitive Linguistics (Human Cognitive Processing)

Epistemic Modality

Tense and Aspect Systems

Cognitive and communicative approaches to linguistic analysis

Cognitive and Communicative Approaches to Linguistic Analysis

Cognitive and communicative approaches to linguistic analysis

Modality and Tense: Philosophical Papers

Cognitive Archaeology and Human Evolution

Tense and Aspect in Bantu

Tense and Aspect in Bantu

Analogy as Structure and Process: Approaches in Linguistics, Cognitive Psychology and Philosophy of Science (Human Cognitive Processing, Volume 14)

Tense and Aspect in Bantu

Tense and Aspect in Bantu

Metonymy and Metaphor in Grammar (Human Cognitive Processing)

Discourse, Vision, and Cognition (Human Cognitive Processing, Volume 23)

Transdisciplinary Advancements in Cognitive Mechanisms and Human Information Processing

Motivation in Grammar and the Lexicon (Human Cognitive Processing)

Cognitive Linguistics Investigations: Across Languages, Fields And Philosophical Boundaries (Human Cognitive Processing)

Evidentiality and Epistemic Modality in Spanish (Semi-)Auxiliaries: A Cognitive-Functional Approach (Applications of Congnitive Linguistics)

Evidentiality and Epistemic Modality in Spanish (Semi-)Auxiliaries: A Cognitive-Functional Approach (Applications of Congnitive Linguistics)

Cognitive Approaches to Obsessions and Compulsions: Theory, Assessment, and Treatment

Recommend Documents