You are on page 1of 308



general editors: Kasia M. Jaszczolt, University of Cambridge and Louis de

Saussure, University of Neuchâtel.

advisory editors: Nicholas Asher, Université Paul Sabatier, Toulouse, Johan van
der Auwera, University of Antwerp, Robert I. Binnick, University of Toronto, Ronny
Boogaart, University of Leiden, Frank Brisard, University of Antwerp, Patrick Cau-
dal, CNRS, Anastasia Giannakidou, University of Chicago, Hans Kronning, Univer-
sity of Uppsala, Ronald Langacker, University of California, San Diego, Alex
Lascarides, University of Edinburgh, Peter Ludlow, Northwestern University, Alice
ter Meulen, University of Geneva, Robin Le Poidevin, University of Leeds, Paul
Portner, Georgetown University, Tim Stowell, University of California, Los Angeles,
Henriëtte de Swart, University of Utrecht.

Language, Cognition, and Reality
edited by Kasia M. Jaszczolt and Louis de Saussure

in preparation
Future Times; Future Tenses
edited by Philippe de Brabanter, Mikhail Kissine, and Saghie Sharifzadeh

Language, Cognition, and Reality

Edited by

Great Clarendon Street, Oxford, ox2 6dp,
United Kingdom
Oxford University Press is a department of the University of Oxford.
It furthers the University’s objective of excellence in research, scholarship,
and education by publishing worldwide. Oxford is a registered trade mark of
Oxford University Press in the UK and in certain other countries
© editorial matter and organization Kasia M. Jaszczolt and Louis de Saussure 2013
© the chapters their several authors 2013
The moral rights of the authors have been asserted
First Edition published in 2013
Impression: 1
All rights reserved. No part of this publication may be reproduced, stored in
a retrieval system, or transmitted, in any form or by any means, without the
prior permission in writing of Oxford University Press, or as expressly permitted
by law, by licence, or under terms agreed with the appropriate reprographics
rights organization. Enquiries concerning reproduction outside the scope of the
above should be sent to the Rights Department, Oxford University Press, at the
address above
You must not circulate this work in any other form
and you must impose this same condition on any acquirer
ISBN 978–0–19–958987–6
Printed in Great Britain by
CPI Group (UK) Ltd, Croydon, CR0 4YY
General Preface vii
Notes on Contributors ix

Introduction: time, temporality, and tense 1

Kasia M. Jaszczolt and Louis de Saussure

Part I. Time, Tense, and Temporal Reference in Discourse

1 Temporal modification 15
Nicholas Asher
2 Temporal reasoning as indexical inference 37
Alice G. B. ter Meulen
3 Perspectival interpretations of tenses 46
Louis de Saussure

Part II. Time and Modality

4 Modal auxiliaries and tense: the case of Dutch 73
Pieter Byloo and Jan Nuyts
5 Semantic and pragmatic aspects of the interaction of time and
modality in French: an interval-based account 98
Laurent Gosselin
6 Modal conversational backgrounds and evidential bases in predictions:
the view from the Italian modals 128
Andrea Rocci

Part III. Cognition and Metaphysics of Time

7 Experience, thought, and the metaphysics of time 157
Simon Prosser
8 Tensism 175
Peter Ludlow
9 Temporality and epistemic commitment: an unresolved question 193
Kasia M. Jaszczolt
vi Contents

10 An account of English tense and aspect in Cognitive Grammar 210

Frank Brisard
11 Frames of reference and the linguistic conceptualization of time:
present and future 236
Paul Chilton

References 259
Index of Names 279
Index of Subjects 286
General Preface
The series Oxford Studies of Time in Language and Thought identifies and promotes
pioneering research on the human concept of time and its representation in
natural language. Representing time in language is one of the most debated issues
in semantic theory and is riddled with unresolved questions, puzzles, and para-
doxes. The series aims to advance the development of adequate accounts
and explanations of such basic matters as: (i) the interaction of the temporal
information conveyed by tense, aspect, temporal adverbials, and context; (ii) the
representation of temporal relations between events and states; (iii) the human
conceptualization of time; (iv) the ontology of time; and (v) the relations between
events and states (eventualities), facts, propositions, sentences, and utterances,
among other topics. The series also seeks to advance time-related research in
such key areas as language modelling in computational linguistics, linguistic
typology, and the linguistic relativity/universalism debate, as well as in theoretical
and applied contrastive studies.
The central questions to be addressed concern the concept of time as it is
lexicalized and grammaticalized in the different languages of the world. But the
scope and the style in which the books are written reflect the fact that the represen-
tation of time interests those in many disciplines besides linguistics, including
philosophy, psychology, sociology, and anthropology.
The current inaugural volume offers a carefully selected sample of analyses of
various aspects of temporal reference. It emphasizes that linguistic means of con-
veying time, be they grammatical or lexical, cannot be considered in isolation from
the semantic factors of sentence compositionality, pragmatic factors such as
contextual relevance, and the process of utterance interpretation (including prag-
matic inference), or philosophical and psychological factors such as the relation of
the concept of time to the ‘reality of time’ on the one hand, and to the expression of
temporality on the other. As such, the collection is able to offer only a flavour of
each of these topics. Part I, Time, Tense, and Temporal Reference in Discourse,
offers a sample of selected problems and solutions pertaining to the semantics and
pragmatics of temporal expressions that stem out of their interaction with other
elements in sentence structure. Part II concerns the interaction of time and
modality. Part III, Cognition and Metaphysics of Time, comprises a selection of
views on the epistemology, metaphysics, and cognitive processing of temporal
Finally, we would like to thank John Davey, Julia Steer, and Vicki Hart of Oxford
University Press for their friendly and professional advice in the process of the
viii General Preface

preparation of this volume; Adrian Stenton, our copy-editor, for his careful reading
of the text and his expert corrections and improvements to the draft, as well as for
collating the bibliography; and finally Nicola Lennon, for preparing the indexes.
Kasia M. Jaszczolt and Louis de Saussure
Cambridge and Neuchâtel
Notes on Contributors
Nicholas Asher received a doctorate in philosophy from Yale University in 1982.
His advisor was Ruth Barcan Marcus. He then spent twenty-four years at the
University of Texas at Austin, first as an assistant professor, then associate professor,
and finally professor of philosophy and of linguistics. In 2006, he became director of
research of the French Centre National de La Recherche Scientifique (CNRS). He is
the author of four books, including two works on the formal theory of discourse
structure and interpretation known as Segmented Discourse Representation Theory
(SDRT), and one book finished in 2011 on lexical semantics and the composition of
meaning. He has also edited two books, written some fifty-four journal articles,
seventy papers for conference proceedings, and numerous book chapters. He has
received several grants from the NSF and ANR, and has been awarded an ERC
advanced researcher grant from 2011 to 2016.
Frank Brisard teaches English grammar and pragmatics at the University of
Antwerp, where he is a member of the Center for Grammar, Cognition, and
Typology. His research interests include the semantics of tense and aspect in English
and other languages, which he studies from a cognitive point of view. Recent
publications focus on modal meanings and uses of tenses, the notion of imperfec-
tivity (e.g. in French), and the ‘present perfective paradox’ (in English, Lingala, and
Pieter Byloo (Ph.D. 2009) is a researcher at the Center for Grammar, Cognition,
and Typology at the University of Antwerp. His main interests include cognitive-
functional semantics, grammaticalization, and corpus research. His dissertation
dealt with the interaction of modality and negation in spoken Dutch and French.
He is now focusing on the grammaticalization and (inter)subjectification of the
Dutch modal auxiliaries.
Paul Chilton is Professor Emeritus in the Department of Linguistics at Lancaster
University. He is a cognitive linguist and discourse analyst whose publications
include studies in linguistics, discourse analysis, literature, and international rela-
Laurent Gosselin is Professor of French Linguistics at the University of Rouen.
His research focuses on the semantics of temporality (tenses, aspectual viewpoints,
adverbials of time, and aspect) and modality in French. He is the author of
Sémantique de la temporalité en français (1996, Duculot), Temporalité et modalité
(2005, De Boeck-Duculot), and Les modalités en français (2010, Rodopi).
x Notes on Contributors

Kasia M. Jaszczolt is Professor of Linguistics and Philosophy of Language at the

University of Cambridge. She has published extensively on topics in semantics and
pragmatics, including propositional attitude ascription, the representation of time,
the semantics/pragmatics interface, and her theory of Default Semantics. One of her
current projects concerns attitudes de se and first-person reference; the other is a
theory of Interactive Semantics (in progress, OUP). Her books include Representing
Time (2009, OUP), Default Semantics (2005, OUP), Semantics and Pragmatics (2002,
Longman), and Discourse, Beliefs and Intentions (1999, Elsevier).
Peter Ludlow is Professor of Philosophy at Northwestern University. He has
published on a number of topics ranging from the philosophy of language and
philosophy of linguistics to issues in metaphysics, epistemology, and the philosophy
of mind. He is author of Semantics, Tense and Time: An Essay in the Metaphysics of
Natural Language (1999, MIT Press).
Alice G. B. ter Meulen (MA, philosophy; MA, linguistics, University of Amster-
dam) obtained her Ph.D. from Stanford’s programme Logic, Philosophy of Lan-
guage, and Philosophy in 1980. After two post-doctoral terms in the Netherlands, she
was appointed Assistant Professor of Linguistics at the University of Washington,
Seattle, in 1984, and gained tenure in 1989. She was subsequently jointly appointed as
associate professor in philosophy and linguistics at Indiana University, Blooming-
ton, and became a full professor 1996. In 1998 she assumed the chair of modern
English linguistics at the University of Gröningen in association with the Behavioral
and Cognitive Neuroscience Research School. Currently her affiliation is with the
linguistics department of the University of Geneva.
Jan Nuyts (Ph.D. 1988, Habilitation 1994) is Professor in the linguistics department
at the University of Antwerp. His major research area is cognitive-functional
semantics, with a focus on the cognitive and functional structure of ‘time–aspect–
modality’ categories—and the modal categories in particular. His most important
books are Aspects of a Cognitive-Pragmatic Theory of Language (1992, Benjamins)
and Epistemic Modality, Language and Conceptualization (2001, Benjamins). He is
currently editing the Oxford Handbook of Mood and Modality (with Johan van der
Simon Prosser is Lecturer in Philosophy at the University of St Andrews. His
research interests lie chiefly in the philosophy of mind and metaphysics. His current
research interests include the nature of conscious perceptual experience, the experi-
ence of time and its relation to the A/B-theory debate in the metaphysics of time,
indexical thought and experience, the nature of concepts, and the metaphysics of
emergent properties.
Notes on Contributors xi

Andrea Rocci is Associate Professor and Director of the Institute of Argumenta-

tion, Linguistics and Semiotics (IALS) at the University of Lugano. His research
interests lie at the intersection between semantics, pragmatics, discourse, and argu-
mentation in different professional contexts, with a special focus on finance and
journalism. He directed a Swiss National Foundation (SNF) research project on
predictions and modality in financial news and currently directs a SNF project on
argumentation in the newsroom.
Louis de Saussure (Ph.D.) is Professor of Linguistics at the University of Neuchâ-
tel, where he co-founded the Centre for Cognitive Science. He was lecturer at the
University of Texas at Austin and at the École des Hautes Études en Sciences
Sociales (Paris). He was a visiting scholar at the French Centre National de La
Recherche Scientifique (CNRS) and at University College London. His research in
cognitive pragmatics focuses primarily on time and modality, but extends to the
cognitive underpinnings of persuasion in discourse.
This page intentionally left blank
Introduction: time,
temporality, and tense


Representing time in language is one of the most debated issues in semantics,

philosophy, and cognitive science and yet it is ridden with unresolved questions,
puzzles, and paradoxes. There is no satisfactory theory of: (i) representing tenses;
(ii) the interaction of temporal information coming from tense, aspect, temporal
adverbial, and context—or any of these that happen to be available in a particular
language; (iii) representing temporal relations between events and states; (iv) human
conceptualization of time; (v) the metaphysics vis-à-vis epistemology of time; or
(vi) relations between events and states (eventualities), facts, propositions, sentences,
and utterances, to name only a few areas. Time also merits more research with respect
to the linguistic relativity/universalism debate, language acquisition, and language
modelling in computational linguistics, as well as theoretical and applied contrastive
studies. The aim of the series inaugurated with this opening volume is to identify
and promote groundbreaking research in any of these, as well as other related,
areas pertaining to the question of what the human concept of time is and how it is
represented in language. We believe that the launch of such a series is particularly timely
in view of the recently intensified interest in the nature of time and the concept of time.
Human representation of time is the subject of research of many disciplines such
as linguistics, philosophy, psychology, sociology, and anthropology. By necessity, the
orientation of the current volume has to be narrower. It is predominantly linguistic
and philosophical in that the main questions addressed here are: (i) temporal
reference in discourse (Part I); (ii) the concept of time vis-à-vis epistemic modality
(Part II); and (iii) the concept of time vis-à-vis the metaphysics of time on the one
hand, and the linguistic expression of temporality on the other (Part III). The topic
of temporal reference is addressed against the background of different theoretical
assumptions and using different research methods. One of the main threads that
2 Kasia M. Jaszczolt and Louis de Saussure

permeates the chapters is the question as to ‘how many systems’ there are: according
to some authors, the conceptual system equals the language system, à la, for example,
Peter Carruthers or Gennaro Chierchia, while according to others, it is more
promising to assume separation and divergence between language system and
conceptual system, à la, for example, Noam Chomsky or Jerry Fodor. Some of the
contributions address this fundamental question directly, while in others the assump-
tion, one way or the other, is tacit throughout the developing argument. Temporal
reference is a particularly rife ground for such debates in that it affords a wide variety
of means across languages, as well as within a system of one natural language. What
is expressed by grammatical tense or aspect in one language can be expressed by
lexical means in another, or left altogether to pragmatic inference. Likewise, the
grammatical or lexical means available for temporal reference are not ‘unshiftable’;
their temporal value may vary depending on the context and perspective. Tense–time
mismatches such as, for example, the use of a present progressive or simple
present form for future time reference (sometimes dubbed ‘futurate progressive’
and ‘tenseless future’) in English, as in (1) and (2), are two of many examples of this
phenomenon (see also de Saussure, this volume).
(1) On Saturday I am giving a party.
(2) On Monday Tom sits an exam.
Pragmatic means are exemplified in the Thai sentence (3), that can convey a wide
variety of temporal reference depending on the context, common ground, inferred
relevance, default assumptions, and other contextual factors, as in (3a–i) (from
Srioutai 2006: 45). The intended temporality, both external and internal, is left to
be catered for by such pragmatic contributions to the overall utterance meaning.
(3) m3ae:r3i:I kh2ian n3iy3ai:
Mary write novel
a. Mary wrote a novel.
b. Mary was writing a novel.
c. Mary started writing a novel but did not finish it.
d. Mary has written a novel.
e. Mary has been writing a novel.
f. Mary writes novels. / Mary is a novelist.
g. Mary is writing a novel.
h. Mary will write a novel.
i. Mary will be writing a novel.
Pragmatic means of conveying temporality are also exemplified in the use of
juxtaposition of the simple sentences in (4) and the use of conjunction in (5). The
order of the sentences, as well as the fact that they both refer to events (e1, e2), allow
Introduction 3

for the pragmatic inference of the temporal order e1 < e2 (see e.g. Asher and
Lascarides 2003 on the so-called rhetorical structure rule of Narration).
(4) James took out a revolver and shot at the masked figure.
(5) James took out a revolver. He shot at the masked figure.
These three parameters (order, presence of a conjunction, and aspectual nature) are,
however, not enough to explain temporal ordering, since it is easy to find examples
where all are satisfied without temporal ordering necessarily occurring:
(6) Paul left for China and Mary bought a new car.
Notions such as the presence of a common topic, common participant(s),
and/or a causal relation do play a role, but they probably still do not suffice to
explain the ambiguity of (7), which can be interpreted as causal–temporal or as
simultaneous–unrelated (if Mary takes the knife to cut the roast and kills the bandit
with the revolver she holds with the other hand—certainly an improbable and funny
context but theoretically possible).
(7) Mary took the knife and killed the bandit.
A number of conceptual parameters, such as the type of relation the eventualities are
likely to exhibit, have to be brought into the picture; they have to pertain to both
coded meaning and to world knowledge. But as (7) shows, still other types of
information intervene, such as background knowledge (whether there is a roast
and whether Mary has a revolver are, for instance, crucial for the interpretation of
(7) ). There is a matter of divergence between formal semantic models, such as Asher
and Lascarides (2003) and pragmatic ones, such as Sperber and Wilson (1995), on
whether an adequate theory should account for time (a) with recourse to general
cognitive processes aiming at discovering a plausible meaningful intention of the
speaker, assuming that inference directed towards meaningfulness has to do with the
discovery of mental states and relies on rapid and frugal heuristics, or (b) by
assuming that temporal relations are due to the management of discursive formal
rules operating at various levels of information structure.
The fact that there are different levels of information which intervene in the
processing of temporal reference implies not only that there has to be a model of
integration of these (or a ‘merger’; cf. Jaszczolt 2005), but also that conflicts may arise
between pieces of information in order for further interpretive effects to obtain. Such
conflicts arise typically between semantic and pragmatic information, thus conse-
quently at the interface between literal and enriched meanings. In particular, the
temporality conveyed by an utterance can be about a time other than the semantic
meaning of the tense would suggest. These time/tense mismatches are often solved
through routines of contextual shift, for example with narrative presents or futures,
but are far more complicated to account for in specific instances such as epistemic
4 Kasia M. Jaszczolt and Louis de Saussure

futures, tenseless presents, or futurate pasts in some languages like, for example,
French. Contextual constraints operate also on the selection of a temporal reference
appropriate to the act of communication: for example, a present perfect can call
either for representation of a remote eventuality (typically a notable experience of
the speaker, as in ‘I have eaten giraffe’; cf. Wilson and Sperber 1993) or, on the
contrary, a recent one, thus forcing either a distal or a proximal interpretation. The
ambiguity or indeterminacy of information provided by the linguistic code only is
massive also in the domain of time, in that temporal reference is extended or
narrowed down depending on the plausible informative intention of the speaker,
or iteration is selected rather than a single occurrence, to list a couple of examples.
Another example of the intervention of pragmatics is when an imperfective
grammatical aspect is combined with an achievement and gets loaded with features
normally excluded from imperfectivity, as in French imperfective pasts, in which
case the representation is modulated, but semantics alone is not able to explain the
representational output which ultimately depends upon the context. To all of this,
one has to add the various types of interpretive outputs of the combination of these
layers of information, connecting time with other notions, such as space and,
probably most of all, epistemic modality. To complicate the picture even more,
tenses are only one among many explicit or implicit ways to communicate inform-
ation about time, and much is still left open for research in the area of time in
language and thought.
Some languages display grammaticalized means of referring to temporal distinc-
tions that are not commonly shared, such as, for example, the so-called ‘consecutive’
tense in Swahili that cuts across the future/present/past distinction (see e.g. Givón
2005). To continue on the diversity, St’àt’imcets (Lillooet Salish) of British Columbia
forms only the future/non-future distinction (according to Matthewson 2006); in
Central Pomo, a Californian language, the future can be conceived of and expressed
as realis or irrealis, where the choice reflects the speaker’s judgement of the degree of
probability of the event (de Haan 2006: 42). Next, in many languages, the orientation
of the subject vis-à-vis the timeline is different from that familiar from English: in
Maori, for instance, the speaker ‘faces the past’, while the future is ‘behind him/her’
(Thornton 1987). The interrelations between time, evidentiality, and epistemic mo-
dality also differ among languages in interesting ways. Such cross-linguistic differ-
ences in expressing temporal reference constitute a vast subject area in itself and are
left for separate volumes in this series. Likewise, the universalism/relativity debate is
largely left for separate enquiries. Important as they are, they are probably the most
frequented topics in the current research on temporality (see e.g. Filipović and
Jaszczolt 2012a, 2012b), but in view of this expanse, are also orthogonal to our
current concern, which is an introduction to the topics in language, cognition, and
the reality of time.
Introduction 5

What we have tried to emphasize in the, by necessity very limited, ‘taster’

represented by this introductory volume are the foundational issues, such as that
linguistic means of conveying time, be they grammatical or lexical, cannot be
considered in isolation from the semantic factors of sentence compositionality;
from pragmatic factors such as contextual relevance and the process of utterance
interpretation (including pragmatic inference); or from philosophical and psycho-
logical factors such as the relation of the concept of time to the ‘reality of time’ on the
one hand, and to the expression of temporality on the other. As such, the collection
is able to offer only a flavour of each of these topics.
Next, in formal semantics, representation of time is a topic of cutting-edge
research, following the tradition of Jespersen, Reichenbach, Montague, and more
recently Dowty, Kamp, Partee, Steedman, and Hornstein, to name a few (see e.g.
Kuhn and Portner 2002 for an overview; or Mani, Pustejovsky, and Gaizauskas 2005
for a selection of core papers; ter Meulen 1995 and this volume). Research in syntax
and formal semantics normally focuses on the question of the logical form of
sentences with time expressions. In contemporary logic, the discussion originated
with the introduction of sentential temporal operators by Prior (e.g. 1957, 1967, 2003)
and temporal logics are now proliferating (for various temporal logics see Øhrstrøm
and Hasle 1995). On the other hand, representing tense as being anaphoric is also
supported by convincing arguments (Partee 1973; Enç 1987; Hornstein 1990). The
debate is far from settled. The topic is also of growing importance in cognitive
linguistics (see e.g. Langacker 2012; Brisard, this volume; Chilton, this volume). Both
orientations are represented in this volume, testifying to the presence of common
interest in the semantic representation of temporal reference, and perhaps, as it is
hoped, also contributing to the search for compatibility and unification of findings
from various theories—a trend that is already noticeable between a formal semantic
framework of Discourse Representation Theory and cognitive linguistics (see
Hamm, Kamp, and van Lambalgen 2006).
A related cutting-edge problem in research on the mental and semantic repre-
sentation of time is the notion of an event. As van Lambalgen and Hamm (2005: 15),
among many others, observe: “[t]he basic building block of the human construction
of time is the event. This is however a fairly ill-defined notion . . . ” Ever since in the
1920s Ramsey distinguished abstract facts (Brutus’s stabbing of Caesar) from spa-
tiotemporally located events (Brutus’s stabbing Caesar at a certain time and in a
certain place), events have become the subject of disputes in philosophy and in
semantic theory. Subsequently, beginning with Davidson’s (1967a) proposal of
events in the logical form of action sentences, quantification over events became
widely discussed (see e.g. Kim 1976; Parsons 1990; ter Meulen 1995; Landman 2000;
Higginbotham, Pianesi, and Varzi 2000; Rothstein 2004; van Lambalgen and Hamm
2005). To give an overview of some central points in the debate: events can be
coarsely grained or finely grained. For Davidson (1967a), events are individuated by
6 Kasia M. Jaszczolt and Louis de Saussure

their causal relations: events are identical if they have the same causes and effects
(see also Taylor 1985). Neo-Davidsonians create theoretical constructs of events that
are more finely grained in reconstructing the semantic roles of event participants
(Parsons 1990; also e.g. Landman 2000). In contrast, for Kim, stabbing Brutus and
stabbing Brutus violently are separate events (Kim 1976; see also Asher 1993; Maien-
born 2005). Next, Schein (2002), for example, argues in favour of a fairly coarsely
grained notion: in the sentence ‘Brutus stabbed Caesar’, the thematic role of the
Agent (Caesar) is no different from the Agent of ‘Brutus insulted Caesar’; absolute
thematic relations then relate to (real, finely grained) scenes to construct sentence
meaning. Van Lambalgen and Hamm (2005) distinguish between event types and
event tokens, where the latter incorporate time boundedness. Rothstein (2004)
similarly proposes a technical construct of an atomic event. Events are also normally
regarded as independent of the linguistic description, as in Discourse Representation
Theory (cf. Asher 2000). In short, we are a long way from reaching a consensus on
what a concept of an event is, but improvements to neo-Davidsonian approaches are
currently being developed (see Asher, this volume).
In the domain of cognitive post-Gricean pragmatics, the models developed to
account for temporal reference mostly focus on the role of ‘procedural’ expressions;
that is, expressions that provide instructions on how to specify and organize
conceptual representations. Since the seminal work of Blakemore (1987), it has
been assumed that a number of linguistic expressions provide rules or instructions
that take representations in their scope. Procedural expressions are often equated
with grammatical morphemes, but this equivalence, which converges with the still
discussed bipartition between declarative memory and procedural system in the
brain (Ullman et al. 1997), and the exact extension of the conceptual and procedural
system, are still a matter of debate (see Escandell-Vidal, Leonetti, and Ahern 2011). In
this line of thought, tenses, but also temporal connectives, are thought of as belong-
ing to the domain of procedural expressions (Nicolle 1996; Moeschler 1998; de
Saussure 2003). Tenses, in this respect, provide instructions about how to saturate
the temporal variables they encode in such a way that their anaphoric functioning is
accounted for. De Saussure (2003) suggests that there is a higher-level procedure of
temporal interpretation at the level of the utterance that manages the temporal
variables coming from various sources: tenses, adverbs and connectives, conceptual
information, and broader contextual information. Concerning other aspects of
temporal ordering, diverging schemes of temporal and causal enrichment in and-
conjuncts have been proposed by both Levinson (2000), in a neo-Gricean frame-
work, and Sperber and Wilson (1995), in a relevance-theoretic framework, jointly
confronting the discussion on default discourse relations in Segmented Discourse
Representation Theory (SDRT; see e.g. Asher and Lascarides 2003) in a debate which
in the end splits the perspectives between a more conventionalized conception of
meaning (neo-Gricean) and a more cognitive one (relevance-theoretic).
Introduction 7

Next, moving to the philosophy of time, most of the current discussions of the
metaphysics and conceptualization of time originate in McTaggart’s (1908) ‘The
Unreality of Time’, which initially polarized the field into the so-called ‘A-theorists’,
who accept the view that the flow of time is real and that events come pre-packaged
with their temporality, and ‘B-theorists’, who accept the earlier-than/later-than
relations between (real) events and the unreality of time (see e.g. Le Poidevin and
MacBeath 1993; Ludlow 1999 and this volume; Jaszczolt 2009 and this volume; Smith
1993; Parsons 2002, 2003; Mellor 1998; Fine 2005; Le Poidevin 2007; Prosser, this
volume). The questions addressed here include the relation between internal and
real time, the concept of time flow, and the indexicality of time, with the overall
objective of shedding more light on how human agents represent time in the mind
and how this representation relates to reality and to language (for an overview see
also the papers in Callender 2011).
In an interdisciplinary volume of this kind, ‘tense’ is a treacherous term in that it
can be used in very different ways. Philosophers writing about the metaphysics of
time use the predicates ‘tensed’ and ‘tenseless’ to talk about the properties of the
world, which stands in clear contrast to linguists’ use of ‘tense’ to talk about the
grammatical system of a language. The first use is represented in the contributions
by Prosser and Ludlow. In metaphysics, ‘tensism’ means a stance according to which
time is real, which is associated with the A-theory mentioned above. This divides
into the stances according to which either (i) the past, present and future are all real;
or (ii) only the present exists (so-called ‘presentism’); or (iii) only the past and the
present exist; or finally, (iv) only the present and the future exist. On the other hand,
a ‘tenseless’ theory depicts reality in a way in which the past, the present, and the
future are equal in status, in that time is not a feature of the world, corresponding to
the views falling under the umbrella of McTaggart’s B-theory.
Another focal thematic area is the relation between time and modality (see Byloo
and Nuyts, this volume; Gosselin, this volume). Theoretical arguments for tempor-
ality as a concept akin to the attitude of acceptance are ample in the literature. For
example, Hume talks about forming the idea of time from the succession of ideas
and impressions. Locke, similarly, claims that it is not duration that makes up the
concept of time but rather events themselves. Kant, Husserl, and Heidegger’s
psychological ideas of time subscribe to a similar concept of retaining images (see
also Kelly 2005a). On the other hand, in the 1920s, William James put forward an
idea of the so-called ‘Specious Present’, according to which we are, at every moment,
in direct contact not only with what there is, but also with what immediately
preceded and what will immediately follow. So the future and the past are part of
what is directly perceived. On these accounts time is a property of its users: it is
a mode of presentation of states of affairs. Similarly, in current formal semantics,
it is often claimed that the English tense/aspect system is founded on the same
conceptual primitives as the system used in languages with evidentials, such as Hopi
8 Kasia M. Jaszczolt and Louis de Saussure

(e.g. Steedman 1997: 932). Evidentials also provide a useful conceptual foundation for
discussing future-oriented speech acts (see Rocci, this volume). The idea of time as
goals is also carefully defended in van Lambalgen and Hamm (2005). Further, in
temporal logic, there are possible-worlds accounts of pastness as historical necessity:
historical possibilities diminish monotonically with the passage of time. The flow of
time is then explained as a ‘loss of possibilities’ (see Thomason 2002: 209), and hence
is modal through and through. Next, Ludlow (1999) argues that all past tense
morphology in Indo-European languages is best analysed as evidential markers:
pastness is evidence available now. Jaszczolt (2009) and Jaszczolt and Srioutai
(2011) also present evidence in favour of the modal character of the past and of
past-tense expressions. Enç (1987), and many others, argued for the modal character
of the future tense as well as futurity as a concept. And, most importantly perhaps, it
has also been argued that time is modal because it is intrinsically tied up with
counterfactuality and it is at least conceptually plausible that time structure of one
possible world may be different from that of another. Since the category of modality
is shown to be instantiated in a wide variety of forms and degrees across the
languages of the world (see e.g. van der Auwera and Plungian 1998), this account
of the present, past, and future also seems to fit well with the current typologies of
As they stand, the issues raised by the notion of time, both external and internal,
and its reflection in language, still appear to be of great complexity and many of
them still await resolution. However, the intense research that this field is undergoing,
together with the related domains of modality and evidentiality, keep opening new
paths of investigations year by year. This volume offers a limited, but, we believe,
accurate, survey of some of these issues and of the kinds of hypotheses that are
currently being proposed to address them.
The content of the volume is as follows. Part I, ‘Time, Tense, and Temporal Reference
in Discourse’, offers a sample of selected problems and solutions pertaining to the
semantics and pragmatics of temporal expressions that stem from their interaction
with other elements in sentence structure. In Chapter 1, ‘Temporal modification’,
Nicholas Asher investigates the question of how temporal adverbials affect the meaning
of verb phrases. He proposes a theory of coercion and a semantics of types. Meaning
shifts that involve coercion are understood as transformations that introduce a special
type structure. Selectional restrictions are understood there as type presuppositions. He
argues for the superiority of this approach over the standard neo-Davidsonian accounts.
Next, in ‘Temporal reasoning as indexical inference’, Alice ter Meulen investigates the
use of English aspectual adverbs with numerical determiners in information structure,
such as ‘There are still/already three students here’. She distinguishes focused from
backgrounded/presupposed/shared information in the common ground. She explains
that the aspectual adverbs induce background assumptions and, on the theoretical level,
Introduction 9

provide the focus alternatives. This is associated with their monotonicity and their
potential to support scalar inferences. She also addresses the question of the behav-
iour of indexical predicates in the inferential process. In Chapter 3, ‘Perspectival
interpretation of tenses’, Louis de Saussure addresses cases where the temporal
reference indicated by tenses cannot be understood literally, that is, it proves to be
irrelevant unless one of the components of the representation (the deictic point, the
reference point, or the eventuality) is modified under the pressure of contextual
consistency or relevance. He suggests that these situations imply a change in the
perspective, or point of view, from which the eventuality is grasped. Three cases are
discussed: (i) narrative (and other non-background) examples of French imperfect-
ive past, which he explains by a point of perspective replacing the reference point;
(ii) futurate composed past (present perfects) in French, which imply a deictic shift
in the future; and (iii) epistemic futures, which rely on a conceptual modulation of
the represented eventuality. In the light of such cases, he envisages in his conclusion
that some metarepresentational meanings may be obtained through simulative
processes in the Theory of Mind.
Part II concerns the interaction of time and modality. In ‘Modal auxiliaries and
tense: the case of Dutch’, Pieter Byloo and Jan Nuyts offer a cognitive-functional
analysis of the interaction between time and modality, focusing on the semantic
effects of the interaction between grammatical means of expressing tense and modal
auxiliaries in a clause. Dutch provides a pertinent case for their analysis in that it has
fully productive preterite and infinite forms and exhibits a flexible organization of
the verbal group. Next, Laurent Gosselin, in ‘Semantic and pragmatic aspects of the
interaction of time and modality in French: an interval-based account’, offers an
insight into the complex relationship between temporal and modal meanings in
French, utilizing a Reichenbach-inspired model in which intervals replace the
traditional coordinates. He suggests that temporal and modal expressions constitute
closely intertwined domains; an eventuality referred to in an utterance is (i) situated
in time and (ii) presented with an aspectual viewpoint and a semantic modality,
where the latter classifies the eventuality as an irrevocable fact or as a mere
possibility. This modality is derived from the aspectual viewpoint. He presents a
model that allows for certain predictions concerning the modal meanings of tenses.
In Chapter 6, ‘Modal conversational backgrounds and evidential bases in predic-
tions: the view from the Italian modals’, Andrea Rocci focuses on the interaction of
future-time reference with modality and evidentiality in predictions. Predictions are
a type of speech act that (a) requires the propositional content to be a future
eventuality and (b) presupposes that the speaker has indirect evidence at his/her
disposal to assert the propositional content. Future reference imposes precise and
finely grained epistemological constraints on the admissible types of evidence. Rocci
discusses in detail the behaviour of the Italian modal verbs potere (‘can’, ‘may’) and
dovere (‘must’, ‘have to’) in a corpus of predictions. He finally analyses conditional
10 Kasia M. Jaszczolt and Louis de Saussure

forms of dovere and points out that they give rise to a limited set of complex
evidence types that support predictions.
Part III, ‘Cognition and Metaphysics of Time’, comprises a selection of views on
the epistemology, metaphysics, and cognitive processing of temporal reference. In
the opening chapter, ‘Experience, thought, and the metaphysics of time’, Simon
Prosser adds a voice to the prominent debate between A- and B-theorists of time,
juxtaposing the metaphysical questions it raises with debates concerning the nature
of temporal experience and the semantics of temporal thought and language. He
discusses some aspects of this interaction, focusing on an argument against the
possibility of perceiving the passage of time. There are characteristics of language,
thought, and experience that make the A-theory appear to be the common-sense
view and in this context he points out the difficulties that face the B-theorist (who
denies that time passes) in explaining these features. Next, Peter Ludlow introduces
‘Tensism’, the view that tensed accounts of the nature of time are not eliminable
and not reducible, in that tensed facts are not reducible to non-tensed facts. He
distinguishes metaphysical tensism and linguistic tensism. The first says that reality
itself is tensed and that these tensed facts are not reducible to untensed facts.
Linguistic tensism offers an analogous view concerning natural language, namely
that language is tensed and that natural-language tense is not reducible to untensed
facts. He assesses the arguments for and against tensism, addressing in the process
some of the pertinent objections, and concludes by making a good case in favour of
tensism, with the help of Fregean Sinn in the semantics. In Chapter 9, ‘Temporality
and epistemic commitment: an Unresolved Question’, Kasia M. Jaszczolt further
develops her argument for the modal supervenience of time by addressing the issue
of the future–past asymmetry. In Representing Time (Jaszczolt 2009), she argued that
on the level of basic concepts, temporality is epistemic modality, where our temporal
concepts of past, present, and future eventualities alike are founded on the degrees of
commitment to the truth of the proposition-like construct. Although the thesis of
the modal supervenience of temporality was strongly supported by theoretical
arguments and by evidence from various languages, there remained an important
Unresolved Question (UQ). It concerned the translatability of what intuitively seems
to be a qualitative difference between the past, the present, and the future into
quantitative differences. In this chapter she presents and assesses two possible
answers to the UQ: that (i) the differences between the past, present, and future
are underlyingly quantitative rather than qualitative (the Direct-Quantitative View);
and that (ii) the differences are qualitative and the value of modal detachment is
contextually established (the Modal-Contextualist View). She concludes with some
remarks on conceptual and naturalistic reductionism. Next, Frank Brisard offers ‘An
account of English tense and aspect in Cognitive Grammar’, beginning with an
overview of the semantic analysis of tense and aspect in this theory and focusing on
the English verb and the systematic contrast between perfectivity and imperfectivity
Introduction 11

as well as the uses of the progressive and the perfect construction. He proposes that
tense/aspect markers reflect the speaker’s/conceptualizer’s modal concepts and that
temporal/modal meanings be analyzed as instantiations of a common epistemic
schema. He concludes that an adequate semantic analysis of tense and aspect should
be concerned with establishing a core schematic meaning for any relevant construc-
tion, and discusses the interaction with contexts of use with the help of the concept
of a semantic map. To continue on a cognitive note, in ‘Frames of reference and the
linguistic conceptualization of time: present and future’, Paul Chilton addresses the
question of how morphologically marked tense relates to conceptual time. In
particular, he focuses on present-tense forms, mostly with reference to English,
and asks how they relate to times that lie outside the speaker’s present. The
explanation invokes the notion of reference frame and the approach of the Deictic
Space Theory (DST), that offers a geometrical model involving three conceptual
dimensions: (i) time, (ii) modality, and (iii) attentional focus. Tense marking is
understood here as inherently deictic, and hence geometrical notions are used to
represent spatial concepts, where the latter underlie more abstract concepts such as
temporal and modal ‘distance’. Next, he shows how the DST framework can be used
to model the use of present tense to refer to the future, as well as the use of the future
tense to communicate modality.
This page intentionally left blank
Part I

Time, Tense, and Temporal

Reference in Discourse
This page intentionally left blank

Temporal modification


1.1 Introduction
It is quite striking how temporal adverbials can affect the readings of verb phrases.
There are well-known examples:
(1) a. John wrote a letter.
b. John wrote a letter in an hour.
c. John wrote a letter for an hour.
d. John kissed Mary at ten in the morning.
e. John kissed Mary for an hour.
According to most linguists who espouse a view of lexical aspect that interacts with
tense, (1a) is a paradigm example of a description of an accomplishment. Letter
writing has a beginning and a natural culmination; the simple past tense applied to
the verb phrase (VP) gives us a completed event of letter writing. The minimal pair
in (1b–c) shows how a different temporal adverbial as an adjunct affects the inter-
pretation of the verb phrase and can change the aspect of the eventuality described.
In (1b), the accomplishment reading is still in full force; the letter is written and
finished within an hour. In (1c), the for adverbial phrase modifies the VP and yields
an entirely different way of thinking about the letter writing: in (1c) the letter-writing
process goes on for an hour without necessarily coming to the culmination of
a finished letter. A similar minimal pair in (1d–e) shows how modification with a
temporal adverbial can shift the VP from the description of an achievement (a
punctual event) to an activity or ongoing process whose termination is not implied,
as one can continue (1e) with In fact he kissed her for a whole three hours.
Another set of examples also motivates this study and shows that the effect of
temporal adverbials on interpretation is not limited simply to temporal localization
and shifts in verbal aspect:
16 Nicholas Asher

(2) a. She left her husband at the airport five minutes ago.
b. She left her husband at the airport five years ago.
c. She left the school five minutes ago.
d. She left the school five years ago.
These examples are quite striking in that the simple span of the temporal adverbial
shifts the sense of leave, and husband or school. In (2a) and (2c), leave is taken by
most speakers to be a physical separation between the woman and her husband or
her school, whereas in (2b) and (2d) the overwhelmingly plausible reading is that the
woman has left the institution represented by her husband (her marriage) or the
school institution. The physical separation reading is still there in (2b) and (2d), but
another sense of leave, school, and husband is clearly also invoked.
How are these shifts possible? The calculation of verbal aspect has a long history
behind it, but still there is no clear understanding of the issue. The last minimal pairs
show that the effects of verbal modification go far beyond those of aspect compos-
ition. There is a general story to be told about coercion.
To tell this story, I have to lay considerable groundwork, much of it developed at
length in Asher (2011). I will have to explore what it is to select a sense of a word like
husband. The story I shall tell, told at greater length in Asher (2011), has to do with
types and meaning composition.

1.2 Type composition logic

At the foundations of the shifts in interpretation is a theory of meaning composition,
that is, a theory of how the meanings of individual words combine to form the
meaning of larger meaningful components of discourse. Meaning composition
proceeds according to function application with a choice of which syntactic categor-
ies function as functors and which as arguments.
But not every word can serve as an argument to any other word as a predicate.
Words come with selectional restrictions, some of which are already encoded in the
syntax; for example, although Isabel is sleeping is a perfectly fine sentence, loves is
sleeping is not a well-formed sentence. But many if not most words come with more
fine-grained, semantic selectional restrictions on their predicates as well. For in-
stance, a verb like hit imposes the restriction that its object or internal argument
must be a physical object. Thus,
(3) Mary hit John’s idea.
is predicted to be difficult to interpret unless the context allows us to interpret John’s
idea as some sort of physical object.
Selectional restrictions of an expression « pertain to the type of object denoted by
the expression with which « must combine. However, this information about the
Temporal modification 17

type of argument is not really asserted. It is rather a type of presupposed content.1

Selectional restrictions resemble presuppositions because their satisfaction seems to
be a prerequisite for any expression’s containing them having a well-defined seman-
tic value. Their demands for satisfaction or justification percolate up the semantic
construction very much in the way that ordinary presuppositions do. That is, (4a–b)
make the same demands on the context of interpretation that the unembedded (3)
does, patterning in a similar way to the presuppositional content of definite descrip-
tions like the present King of France.
(4) a. Mary didn’t hit John’s idea.
b. Did Mary hit John’s idea?
c. John didn’t see the present King of France at the exhibition.
d. Did John see the present King of France?
In dynamic semantics (Kamp and Reyle 1993; Groenendijk and Stokhof 1991),
presuppositions constitute a particular sort of test on the input context. Consider a
sentence like
(5) Jack’s son is bald.
The presupposition generated by a definite noun phrase like Jack’s son (namely, that
Jack has a son) must be satisfied by the input context, if the interpretation of the rest
of the sentence containing the definite is to proceed. One way of satisfying the
presupposition in (5) is for it to be already established in the context of utterance of
(5) that Jack has a son. This can occur, for instance, when (5) is preceded by an
assertion of Jack has a son. Presuppositions can also be satisfied by contexts within
the scope of certain operators, as in (6), even though it has not been established in
the discourse context that Jack has a son:2
(6) If Jack had a son, then Jack’s son would be bald.
In dynamic semantics the content of the antecedent of the conditional is encoded
into the input context for the interpretation of the consequent and the presuppos-
ition it generates. Thus, the satisfaction of the presupposition by the antecedent of
the conditional in (6) means that the presupposition places no requirement on the
input context to the whole conditional, the context of utterance. The satisfaction of
the presupposition by elements of the discourse context entails that the presuppos-
ition does not ‘project out’ as a requirement on the context of utterance. Thus (6) is
consistent with the assertion that in fact Jack has no son. On the other hand,

That selectional restrictions are type presuppositions is a fundamental principle of Asher (2011).
One of the great successes of dynamic semantics has been to show that the behaviour of presuppositions
introduced by material within the consequent of a conditional follows straightforwardly from the conception
of the conditional as a complex test on the input context and thus offers a solution to the so called projection
18 Nicholas Asher

dynamic semantics predicts that if we change (6) just slightly so that the antecedent
does not provide a content that satisfies the presupposition, the presupposition will
project out as a requirement on the input context to the whole conditional:
(7) If Jack were bald, then Jack’s son would be bald too.
Selectional restrictions act in the same way in similar contexts. For instance, to say
something like
(8) The number two is blue.
is to invite at the very least quizzical looks from one’s audience, unless the context
makes clear that the number two refers to some sort of physical object and not the
only even prime. However, a counterfactual with an admittedly bizarre antecedent
can satisfy the type presupposition projected from (8) in much the same way as the
antecendent of (6) satisfies the presupposition of the consequent:
(9) If numbers were physical objects, then the number two would be blue.
What happens when a presupposition cannot be satisfied by the discourse context?
It depends on what sort of presupposition is at issue. Some presuppositions, such as
those introduced by some uses of definite noun phrases, are easily ‘accommodated’.
In dynamic semantic terms this means that the input context is altered in such a
way that the presupposition is satisfied, as long as the result is consistent. Other
presuppositions, such as that generated by the adverbial too, are much less easily
accommodated. Given that operators like conditionals can add ‘intermediate’ con-
texts between the context of utterance and the site where the presupposition is
generated, we need a theory of where and how to accommodate in case the input
context does not satisfy the presupposition. Van der Sandt (1992) stipulates a
particular procedure for handling the binding and accommodation of presupposi-
tions: one tries to bind a presupposition first, and one tries first to bind it locally and
then in a more long-distance way. Then one tries to accommodate at the outermost
context first and if that fails one tries to accommodate at the next outermost context.
The constraints on accommodation are that the addition of the presuppositional
material be consistent with the discourse context. So, for instance, one cannot add
the presupposition that Jack has a son to a context where it is established that Jack
does not have a son.
Something similar happens with selectional restrictions. In effect selectional
restrictions are type presuppositions. Type presuppositions in normal circumstances
are bound or justified in that the type of the argument expression matches the type
presupposition of its predicate. But they can sometimes be accommodated. Consider
the noun water. It can combine with determiners that require either a noun that
denotes something of type mass (10a) or with determiners that are intuitively count
Temporal modification 19

(10) a. some water

b. a water
One way of accounting for this is that water itself does not determine its denotation
to be either of a subtype of type mass or of type count. If that is the case, then we
can accommodate the requirements of the determiner simply by applying the type
count or mass to the type of the expression water—in simplified terms, a water
ends up denoting a property of properties that have a non-empty intersection with
the collections of portions of water. On the other hand, this could also be a form of
coercion by the determiner over its first argument.

1.3 Coercion and presupposition justification

Sometimes the argument of a predicate does not satisfy the type presupposed by the
predicate and cannot be accommodated in the given context. In that case, semantic
composition crashes and there is no well-defined value for the semantic compos-
ition, as in (3). However, it is not obvious what the principles are for accommodating
type presuppositions that sometimes permit the rescue of a predication in those
cases where the argument’s type does not satisfy the predicate’s type presupposi-
tions. Examining this issue takes us to the heart of coercion.
Consider the following example, discussed at length in the literature, in particular
by Pustejovsky (1995).
(11) Julie enjoyed the book.
The intuition of many who work on lexical semantics is that (11) is equivalent in
meaning to:
(12) Julie enjoyed doing something (e.g. reading, writing, . . . ) the book.
The intuition shared by many is this: enjoy requires a noun phrase that denotes an
event as its direct object as in enjoy the spectacle, enjoy the view, enjoy watching TV.
This also happens when enjoy takes a question as its complement, as in enjoy
(hearing) what he said. When the direct object of a transitive use of enjoy does not
denote an event, it is ‘coerced’ to denote some sort of eventuality. It is problematic,
however, to understand what coercion is. Is it, for instance, just the transformation
of the denotation of the noun phrase the book into some sort of eventuality denoting
expression?3 If that is the case, then how can we access in subsequent discourse the
referent of the book?4

The account in Pustejovsky (1995) seems to adopt this line of attack towards the problem.
This problem makes the qualia story as developed in Pustejovsky (1995) a non-starter, as I argue in
Asher (2011).
20 Nicholas Asher

(13) Julie enjoyed the book. It was a mystery.

These observations are familiar but show that we cannot shift the meaning of the
book to some sort of eventuality. Or at least if this happens, whatever process is
responsible for the shift must also allow the book to retain its original, lexical
meaning and its original contribution to logical form.
The other alternative explored in the literature is to shift the predicate, in this case
the verb, to mean something like enjoyed doing something with.5 At the very least,
however, one will need some underspecified form of coercion to handle cases of
gapping coordination like the one below.
(14) Julie enjoyed (reading) a book and Isabel (watching) a movie.
Anaphora tests using ellipsis are instructive here. The meaning of the activity
is predicted on the predicate modification view to shift with the choice of
direct object during semantic composition. But if that is the case, then the
ellipsis predicts a peculiar reading of the second clause of (15) where Isabel
enjoyed doing something to or with the parade, when it doesn’t have that: Isabel
simply enjoyed the parade as an event. Or perhaps she enjoyed being in the
parade. There is a sense that Isabel and Julie did different things with the book
and the parade—and simply shifting the predicate doesn’t do justice to this
(15) Julie enjoyed her book and Isabel her parade.
Things get much worse for the predicate modification view when we look at
coordinations involving normal DPs and gerunds:
(16) a. Julie enjoyed a book and watching a movie.
b. Julie enjoyed a book and Justin watching a movie.
c. I actually enjoyed the book and her writing. (Google)
d. I enjoyed the movie and the story line but of course, I wish that the movie
had not glorified Buddhism so much.
e. I enjoyed the movie and its wry humour.
f. Whilst I enjoyed the book and reading about the mistakes made by
Dan Brown, I think people are taking this too seriously.

Nunberg (1995) and others have pursued this line of attack.
Temporal modification 21

g. I really enjoyed the book and the biological innovations and theories
expressed in it.
h. I really enjoyed the book and how it uses someone’s life in the story
and gives ideas on how to stay unstressed and calm.
While (16a–b) aren’t the best English, they are grammatical; and they, together with
the attested examples above like (16h), spell big trouble for the predicate modifica-
tion view. To handle the DP coordination, the predicate modification view predicts
that Julie enjoyed doing something to watching a movie, which is uninterpretable.
What would that doing be? Did she enjoy watching watching the movie? Similarly,
the sluicing example in (16b) would predict an uninterpretable reading. All this is
strong evidence that coercion does not involve a semantic adjustment to the predi-
cate at the moment of composition.
One might suppose that in fact coercion and type presupposition accommodation
in general is not part of semantics at all, but rather a pragmatic mechanism. The
ellipsis facts seem to point to a pragmatic account according to which the appro-
priate enrichments to predicates and/or arguments occur after semantics has done
its job, and hence phenomena like ellipsis have been resolved. One problem with
pragmatic accounts is they have difficulty accounting for the language-specificity of
many coercions: pragmatic principles are supposed to follow from general principles
of rational interaction between agents, and so they are expected to be universal.
We’ve already seen that coercions are typically language-specific. In addition, we
should expect pragmatic enrichment to allow coercions, say from objects to even-
tualities, whenever the grammar allows for an eventuality reading in the argument.
But this isn’t true. Consider
(17) a. The reading of the book started at 10.
b. #The book started at 10.
c. John started the book.
It’s perfectly straightforward to get the eventuality reading for the object of start (17c)
in an out-of-the-blue context. But it’s impossible to do this when the intransitive form
of start is used with the book in subject position, even though an eventuality reading
for the subject of intranstive start is available, indeed mandatory, as shown by (17a).
So if you can’t shift the meaning of the predicate and you can’t shift the meaning
of the argument and you can’t relegate the problem of coercion to the pragmatics
garbage can, what is left? My answer is that you change the relation of predication
that holds between the predicate and the argument. How would this work? What
happens is that in response to a clash between the type of the argument and the type
presupposition of the predicate, the relationship between the two is licensed. Such a
22 Nicholas Asher

type conflict induces not a type shift in either the argument or the predicate but
rather a type shift on the predication relation itself, which is implemented by
introducing a functor that is inserted around the argument. Thus, the meaning of
the argument does not shift—it remains what it always was; the predicate also retains
its original meaning, but what changes is how they combine. Given that the sluicing
examples recover just the verb’s meaning, this account makes the right predictions
for (16b) or (15).
Such shifts in the predication relation are lexically governed. It is the verb enjoy
that requires an event but also which licenses a change in the predicational glue
between it and its object when it is not of the right sort. In addition, as I argued in
Asher (2011), these licensings are proper to certain arguments of the verb, or rather to
the syntactic realization of other arguments of the predicate, where once again we
are not dealing with a phenomenon of general pragmatic strengthening or repair but
rather with a problem at the syntax/semantics interface, which is how to account for
the differences in (17).
My view requires the background assumptions that first, verbs, and predicates
more generally, distinguish between arguments that denote eventualities and those
that denote, say, physical objects, and second, there is a clear distinction between
physical objects and eventualities. But it seems to me that both of these assumptions
are cogent and well supported linguistically across a wide spectrum of languages. In
the section after the next, I sketch a formal system that works out my view technically.

1.4 A sketch of a formal theory of lexical meaning

1.4.1 Types
Types are semantic objects and they are clearly linked to denotations. In Montague
Grammar (MG), which uses Church’s conception of types (Church 1936), types are
identified with the set of entities in the model of that type. For example, Montague
Grammar countenances two basic types: the type of all entities e, which is identified
in a model with the domain of the model, and the type of truth values, t. Further
types are defined recursively: for instance, all nouns have the same type, which is the
set of all functions from the domain of entities into the set of truth values. But the
type system in MG is far too impoverished to be able to determine the semantic
values of expressions and hence the truth conditions of sentences or discourses.
Types, however, do play a role even in Montague’s system. They have enough
semantic content to check the well-formedness of certain predications.
Given the sort of type checking relevant to examples of coercion that I have
surveyed here, the type system must incorporate many more distinctions. It must
distinguish various distinct basic types that are subtypes of e; for instance, it must
distinguish the type for eventualities from the type for physical objects, as well as
Temporal modification 23

distinguish the type for informational objects from these two. I’ve already indicated
that even more fine-grained distinctions will be necessary in the type system: for
instance, the distinction between states and events must be reflected there, and there
are many other semantic distinctions that end up, in my view, as distinctions
between types.6
Adjectives involve a type that is a function from types to types and whose value
changes given the input type. Consider the adjective flat. It is clear that flat changes
its meaning rather dramatically depending on the noun it combines with:
(18) a. flat country
b. flat curvature
c. flat tyre
d. flat beer
Asher (2011) develops the semantics of types using the framework of category theory.
For example, atomic types are objects in the category, while the type a ) b
describes the category theoretic, exponent object ba, which is an object in category
theory with the right sort of structure. The nice thing about the category theoretic
approach is its representation of the sort of meaning shifts that occur in coercions.
Coercion meaning shifts are understood as natural transformations introducing a
particular type structure into the category when they are licensed. Figure 1.1 provides
a picture of how they work: a coercion involving an argument of type b that is
triggered by a functor of type g a or ga  b involves a natural transformation from
the left-hand side of the BC arrow to the right side, where a new morphism is
introduced from the functor and its arguments into the functor and a new composed
type dependent on the types a and b and some operation B. We’ll look at some
examples of such coercion.
This transformation invokes polymorphic types with underspecification, as the
type determined by B is a function of a and b. We will be looking at particular
‘instances’ of the dependency relative to particular argument types.

1.4.2 Type presuppositions again

The next step is to implement talk of type presuppositions in a formal framework.
Predicates impose type constraints in the context to come on the types of their
arguments of two sorts, one an absolute requirement and occasionally another which
licenses a modification of the predicational environment between the predicate and
its argument. I code these type constraints in the lexical entry, which requires a

This already complicates the task of doing lexical semantics considerably, because the standard
Church/Montague conception of types as sets fails to give a sensible notion of subtyping for higher
functional types. For more discussion, see Asher (2011).
24 Nicholas Asher

g a ×b × a × g g a ×b × a × g

id × id × dep

g a ×b × a × (a , g )

FIGURE 1.1 b Coercions

complication of the lexical entry of all words that includes a context parameter in
which these type presuppositions can be encoded.
How does this parameter differ from the notion of a typing context already found
in some approaches to the l calculus? A typing context in the standard l calculus or
standard typing context can be understood as a function from terms to types. Our
parameter resembles a standard typing context to be sure, but standard typing
contexts are simply a way of encoding types on terms. The context is fixed and
has no particular effect other than the standard one, which is to check the applic-
ability of the rule of application. We need to change this to handle coercion facts. To
do so I will add an argument to terms that is a type parameter and use this to store
type information. This type parameter is an instance of what de Groote (2006) calls a
left context in his reconstruction of the dynamic semantics of Kamp (1981); it can be
an arbitrary data structure on which various operations may be performed separately
from the operations of b reduction in the l calculus. This is the sort of flexibility we
need to handle various operations of presupposition justification at the lexical level.
These include Binding (in which the type presupposition is a super type of the type
of the argument and thus satisfied by subtyping), Simple Accommodation (in which
the two types are compatible in that they have a meet that is consistent, and the
result is to take the meet as the result of the operation), and operations that allow
the readjustment of the predicational relation. These operations together with the
standard lambda calculus comprise the fragment of the Type Composition Logic
(TCL) that is relevant for my purposes here. The function of TCL is to provide a
procedure for how to compose meanings; it extends the ordinary lambda calculus
with operations needed to handle coercion, among other things.
Presupposition parameters pass the type presuppositions to the predicates.
Presupposition parameters range over lists of type declarations on argument posi-
tions for predicates. With verbal predications, it is relatively uncontroversial that the
Temporal modification 25

verb should pass its type requirements to its DP arguments. But what about
modifiers, which are of concern here? Should the noun pass its type requirements
to the modifier or the reverse? Given that the vast majority of adjectives in the
world’s languages are subsective (that is, when an adjective ADJ is subsective, for any
N, an ADJ N is necessarily an N), it seems that the noun’s type requirements are
passed to the adjective. This requires a slightly more complicated lexical entry for a
noun than usual. I assume that modifiers take first-order properties as arguments of
the sort traditionally associated with nouns, but the noun’s lexical entry will also
have a l abstracted variable for the modifier. Thus, letting P be a variable of
modifier type, the functional type from first-order properties to first-order proper-
ties, the lexical entry for tree looks like this:
0 0
(19) 1 :pÞ(x)(lvlp tree(v, p ) )
lP lxlp P p argtree
1 stands for the ‘first’ argument position of the predicate; this argument
position is always associated with the referential or denoting argument; the first
argument position of a transitive verb will be its subject, while the second argument
position will be filled by its direct object or internal argument. p argtree
1 :p says that
whatever fills this first argument position must have a type that justifies the type p,
or physical object. This includes the variable v in the lexical entry (19), but it may
also include a variable contributed by the modifier. So summing up, (19) says that a
noun like tree takes a modifier as an argument in order to pass along its type
presupposition requirements to the modifier. In order for the modifier to combine
with the noun, it must justify the presupposition that is appended via  ( is a
concatenation operation) to its presupposition parameter argument. By integrating
this l term with a determiner and a verb, more type presuppositions will flow onto
the presupposition parameter p. Finally, note that the modifier takes a simple first-
order property, in this case lvlp’ tree(v, p’), as an argument.
Thus, the final entry for tree will be:
0 0
(20) lP: mod lx: e lp P p argtree
1 :pÞ(x)(lvlp tree(v, p ) )

To save on notational clutter, I’ll often just write argi when the predicate is obvious,
or just list the type when the argument position and predicate is obvious. In practice,
I will omit the typings of terms when they are obvious.7 Given these abbreviations,
the general type of nouns shifts from the standard e ) t to mod ) 1, where 1 is the
type of first-order properties.8 For NPs I will have the relatively simple type schema,
a ) (P ) t), where a is a subtype of e.

I suppose that when no adjective is present, P in (20) applies to the trival subsective adjective, lPlx:
elpP(p)(x) and so gives a predictable type for a simple NP. I deal with multiple adjectival modifiers
through type raising to form a complex modifier; the exact logical form for such raised NPs will be a
matter of meaning postulates.
See Asher (2011) for a detailed discussion of first-order properties.
26 Nicholas Asher

In general predicates must pass their typing presuppositions on to their argu-

ments. In the TCL framework, nouns pass their typing requirements to adjectives. A
VP passes its typing presuppositions to its subject, and a transitive verb passes its
typing requirements to its object. For example, in The book is heavy, the type
requirements made by the verb phrase on its subject argument should be satisfied
or accommodated in order for the predication to succeed. In keeping with standard
assumptions about the syntax–semantics interface, this means we must rethink the
entries for our lexical entries to track these presuppositions and to put them in the
right place.
Consider an intransitive verb like fall. We want its type presuppositions to
percolate to the subject. We can do this by assuming that an intransitive verb, or
more generally a VP, is a function from DP denotations to propositions.9 This
function will feed the appropriate type presuppositions to the subject DP in the
same way that nouns feed their presuppositions to their modifiers. In order to do
this properly, we must make our verbal predicates be relatively catholic as to the type
of DP they accept. The specific type requirements that verbs will place on their
arguments will not be placed on the DP itself, but be placed rather in the presup-
position slot so that it may be propagated to a local justification site and justified
there—i.e. on the individual argument term furnished by the DP, if that is possible. If
the higher-order types 1, dp, etc. shift in type, it is usually because of a shift of one of
their constituent types, from one subtype of e to another. It is quite difficult to shift
higher-order terms directly, as we can see from the lack of coercions, say, from
propositions to questions or from propositions to properties.
Thus, for an intransitive verb like fall, we have the following lexical entry where
1 : p specifies the type assignment for fall on its first argument place to be of
type p or of physical type. This first argument place is ultimately filled by the variable
bound by the quantifier that is the translation of the DP in subject position. The
variables that will end up as subject place arguments for the verb can be determined
from the syntax of the l term for the predication. In particular, the variable
introduced by the DP F will be one; and in the representation below, y is one as well.
0 0
(21) lFlp Fðpargfall
1 : pÞ(ly: p lp fall(y, p ) ).

The type of an intransitive verb or a VP, which in Montague Grammar has the type
e ) t, now has the more complex type, dp ) (P ) t), or the functional type from
general DP types to propositional types. Transitive verbs have a similarly trans-
formed type.
The lexical entries chosen determine how presuppositions percolate through the
derivation tree, and they predict that presuppositions will be justified typically

This applies the same trick used to handle transitive verbs taking generalized quantifier arguments—it
makes the latter functions from DP denotations to VP denotations.
Temporal modification 27

locally to the argument’s typing context (the p that determines the typing of the
argument). For instance, when we combine a determiner and an NP to form a DP
that is an argument to another predicate f, then f conveys presuppositions to the
NP or the DP’s restrictor. If such a presupposition is justified in the restrictor of a
DP, it will also be justified in the nuclear scope. The converse, however, does not
hold—leading to a notion of ‘global’ (restrictor) or ‘local’ (nuclear scope) justifica-
tion. Furthermore, just as with clausal presuppositions (see (22) where his father is
bound to the presupposed content of the proper name John), one presupposition can
justify another: the type presuppositions on the nouns car and tree justify, indeed
satisfy, the presupposition of the verb hit, which wants a subject and object argument
of type p in (23).
(22) John’s son loves his father.
(23) A car hit a tree.
We are interested in predicates that license coercions and these involve rather
complex type presuppositions. Consider the verb enjoy; it requires its second
argument to be an eventuality, of type evt, but it licenses a type coercion of its
argument which introduces a second predication relation between its actual
argument and some eventuality. I note this type requirement with the notation
evt – «(hd(F) ), which means that the eventuality coerced is of a type to be
determined by the type of the variable bound by the determiner of the second
argument (F).
We can now look at how a standard coercion interacts with composition. Let’s
look at an example where enjoy applies to a DP like Anna Karenina as in (24). I
assume that book introduces the type book, to avoid the contentious discussion of so
called dual aspect nouns.10 I also will not go into details of the typing restrictions
imposed by determiners of various sorts—again for a discussion see Asher (2011).
(24) George enjoyed Anna Karenina.
Constructing a logical form for the DP and applying it to the entry for enjoy gives us:
(25) lFlp F(pag)lv lQ Q ðparg2 : evt«(hd(F)book)(ak) Þ
(ly1 lp1 (enjoy(v, y1, p1) ^ ag(y1) ¼ v(p1) ) )
Let us assume that polymorphic type presuppositions prefer a local justification,
near the verb. Continuing the reduction and abbreviating our type constraints on x
and y1, we get:
(26) lFlp F(p * ag)(lv enjoy(v, ak, p * evt – «(hd(F), book) )
^ ag(x) ¼ y1(p * evt – «(hd(F), book) ) )

See Pustejovsky (1995); Asher and Pustejovsky (2006); Asher (2011); Luo (2010). For an extended
discussion see Asher (2011).
28 Nicholas Asher

The presuppositions in the nuclear scope of the quantifier cannot be satisfied as they
stand. But enjoy licenses a transformation and the introduction of a polymorphic type
functor with a polymorphic type that will serve to justify the basic type presuppos-
ition. The polymorphic type functor will apply to the l abstract, ly1 lp1 (enjoy(v, y1,
p1) ^ ag(y1) ¼ v(p1) ). For type presuppositions, this is a general procedure for
presupposition justification. The functor introduces a predicate related to the poly-
morphic type. For example if the polymorphic type maps a cigarette to an event of
type smoke(agent, cigarette), then we need to use the predicate smoke(e, x, y).
When the polymorphic type is underspecified and of the form «(a, b) we take the
predicate f«(a,b)(e, x, y). The functor instantiated for this example looks like this:
(27) lPlulp’’ 9z: «(evt, book) 9z1: ag(P(p’’)(z) ^
f«(ag, book)(z, z1, u, p’’) )
Applying the functor on the designated l term within (26) and using the TCL rules
for presupposition justification and the reduction rules of the l calculus, we get:
(28) lFlp F(p * ag)[lv9z9z1 (enjoy(v, z, p) ^ ag(z) ¼ v ^
f«(ag, book)(z, z1, ak, p) )]
We can now integrate the subject into (28) and exploit the fact that ag is a function
to get the finished result:
(29) lp9y(y ¼ g(p) ^ 9z: ag(enjoy(y, z, p) ^ ag(z) ¼ y ^
f«(ag,book)(z, y, ak,p) ) )
The type of functor exploited in (27) suffices to handle all cases of event coercion
with verbs whose type presuppositions are sensitive to both the type of the subject
and object. I’ll call this the E functor. There are many others involved in coercion.
The polymorphic types in event coercion describe a morphism from types of
objects to polymorphic types of eventualities involving those objects, and the «
reflects that morphism from objects to eventualities in logical form. Why should
this transfer principle and type shift from objects to eventualities be sound? The
answer has to do with the presuppositions of the particular words that allow for this
morphism, like, for example, the aspectual verbs and enjoy. Enjoying a thing, for
instance, presupposes having interacted in some way with the thing, and that
interaction is an event. The verb enjoy, however, doesn’t specify what that event is.
The event could be just looking at the object as in enjoy the garden or perhaps some
other activity. Similarly, one can’t finish an object unless one is involved in some
activity with that object, whether it be creating it or engaging in some other activity
towards it. That is why such transformations are lexically based; it is the lexical
semantics of the words that licenses the coercion and that makes the rules sound.
Semantics gets us only so far. We now need to specify the underspecified
eventuality type f«(human,book). This we can do by adding to the type system axioms
Temporal modification 29

that defeasibly specify underspecified types—for instance, the first below says that if
a human engages in doing something with a book, normally he reads it.
. (a v human ^ b v book) > «(a, b) ¼ read(a, b)
. (a v author ^ b v book) > «(a, b) ¼ write(a, b)
. (a v goat ^ b v book) > «(a, b) ¼ eat(a, b)
The type specification logic allows us to write appropriate axioms for specifying
underspecified types. But the specification of the eventuality can also be given by the
discourse context. Consider the following minimal trio.
(30) a. Julie began with the kitchen, proceeded to the living room and finished up
with the bedrooms.
b. Yesterday Julie cleaned her house. Julie began with the kitchen, proceeded
to the living room and finished up with the bedrooms.
c. Last week, Julie painted her house. Julie began with the kitchen, proceeded
to the living room and finished up with the bedrooms.
As I argue in Asher (2011), the discourse in (30a) is not very felicitous because we
don’t know what Julie is doing with the kitchen and the other rooms. It’s like trying
to interpret an utterance of she’s nice in a context with no salient antecedent.
However, once discourse specifies an activity, (30b) and (30c) are completely felici-

1.5 Back to time

With this analysis of coercion in place, we can now tackle coercions linked to
modification of a verbal complex (the interpretation of the VP and its projections
that include tense, modality, and aspect,) and a noun. One of the advantages of the
higher-order system is that we can reflect on the type of entries that we want for
various words. It has been standard practice since Davidson (1968/69) to introduce
additional eventuality arguments to predicates derived from verbs as I have done for
the verbal predicates that characterize coerced eventualities. But one need not do
this. Instead I could have characterized an event object more abstractly via a
realization function. This permits a much more precise characterization of eventu-
alities and requires the theory to be more precise about what requirements lexical
meaning and composition make on the characterization of various types of entities,
in particular eventualities.

1.5.1 Modifications of the verbal complex

Many linguists—Smith (1991), Dowty (1979), Verkuyl (1972), inter alios—have observed
that certain meaning shifts occur to the type of object denoted by the verbal complex
30 Nicholas Asher

when aspect and tense are applied. Since Vendler’s work in the 1950s, it has been
customary to distinguish between different types of denotations of verbs and their
projections.11 Vendler and most of those following him (with the notable exception
of Dowty) have talked of these denotations as types of eventualities, which include
events of different types and states. Combined with Davidson’s (1968/69) treatment
of action sentences, this has led to the received view of verbal modification by
various projections, according to which verbal modification involves predication of
an eventuality introduced by the verb (and bound by tense). If the modification
applies at a node above Tense, for example with an adverb like allegedly or probably,
then the eventuality is no longer available as an argument because it has been bound
by tense and so such modifications are customarily treated as modifications of some
sort of abstract entity like an intension. On the other hand, nominal modification
has been treated in a more heterogeneous fashion but nevertheless still in a David-
sonian fashion.
Davidsonian event semantics is by and large quite successful. But there are cracks
in the edifice. Davidsonian and neo-Davidsonian semantics have a neat way of
explaining verbal modification that crucially involves events. Basically, verbal modi-
fication via syntactic adjunction becomes a simple matter of applying the property
provided by the modifier to an event variable that also serves as the event argument
of the verbal complex:
. VP : VP MOD ! le(kMODk(e) ^ kVPk(e) )
However, there are good reasons to treat some modifiers, in particular modifiers
that describe instruments or goals of the action described by the verb, more like
semantic arguments of the verbal complex. That is the strategy I adopt in Asher
(2011). Not all verbal modifiers saturate optional arguments. Some modifiers simply
take the VP as an argument, as in Montague Grammar. In particular, temporal and
locative modifiers seem to fall into this class. We can have several temporal modi-
fiers that simply narrow down the time at which the event described by the verbal
complex took place. Locative modifiers work similarly.
(31) a. On Monday Isabel talked for two hours in the afternoon between 2 and 4.
b. In Paris John smoked a cigarette on the train in the last second class
compartment in seat number 27.
The fact that temporal and perhaps other modifiers take the VP itself as an argument
and are not arguments of the VP makes predictions in TCL. Predicates pass their
type presuppositions on to their arguments in TCL, not the other way around. So
TCL predicts that temporal adverbials can affect the type of the verbal complex, as

See Vendler (1967), Dowty (1979), inter alios.
Temporal modification 31

seen in the examples in (1). These are examples again of coercion. Consider again
(1c), repeated below:
(1c) John wrote a letter for an hour.
The temporal modifier for an hour takes the verbal complex as an actual argument
but requires a realizer of that verbal complex, and a realizer of a particular type,
namely activity or act.
To be more precise, we need to think about what the verbal complex itself gives us;
well, in the type system, the answer is clear: once the arguments are combined with
the verb, we have a type of proposition, i.e. a subtype of prop. We could of course
postulate an eventuality argument for the verb as in Davidson, but in the higher-
order framework in which TCL is couched, we are not forced to do so. There is a
certain advantage in doing this, because it allows us to have a clearer picture of what
eventualities really are. Mathematical sentences like two and two make four intui-
tively do not have an event argument; they are timelessly true. Though we can apply
the inflectional system within mathematical sentences, this need not introduce an
eventuality, but perhaps only times at which the verbal complexes denoted by such
sentences are true, as in the following:
(32) Two and two always has and always will make four.
Other verbal complexes may indeed under the inflectional system or temporal or
spatial modification give rise to eventuality realizers. And then the semantics makes
clear that these events exist at certain spatiotemporal regions. Eventualities are at
least regions of space–time whose individuation principles may involve their parti-
cipants or the properties they are described by.
So let us assume that for (1c) the verbal complex provides a proposition but that
the adjunct demands an eventuality realizer. A general map or coercion from
propositions to their realizations is clearly part of our conceptual system, and the
temporal adverbial forces the realization to be of activity type. On the other hand,
a temporal adverbial of the form in an hour forces the realization to be of type
accomplishment or achievement. Inflectional systems in most languages, which
include temporal and aspectual markings on verbs, also make demands for an
eventuality realizer and of a particular type. To fill out the picture, we need to
have a clearer picture of the constraints on these coercions. First of all, the type
presuppositions of the inflectional system and of the temporal modifiers have to be
consistent. Second, not all subtypes of prop are equally happy with any temporal
modifier, as many years of linguistic investigation have shown.
(33) Benjamin Franklin invented bifocals for an hour.
TCL accommodates the descriptive work on aspectual type via the type specification
rules. We used the type specification rules before to say how to specify an underspecified
32 Nicholas Asher

eventuality. Now as the type of the eventuality is given as a realizer of the verbal
complex, we use the type specification rules to place negative constraints on coer-
cion: certain subtypes of prop cannot have realizers of a certain type.
One big advantage of such an approach is that it dissolves a problem that has
plagued compositional semantics ever since linguists began to try to formalize the
process of aspectual composition, according to which the type of eventuality realizer
of a verbal complex is not only a matter of the inflectional system but of the lexical
aspect of the verb and the nature of its arguments. To understand the problem,
consider trying to figure out what is the type of the eventuality of:
(34) Those politicians invented excuses to justify their inaction for years.
Given the observation concerning (33), a Davidsonian approach would say that the
eventuality introduced by the verb invent should not be compatible with a temporal
modifier requiring an activity. However, with the plural subject it now appears that
the overall realizing eventuality should be something like an activity. Standard ways
of accounting for such examples postulate an iterative reading and introduce another
eventuality spanning the individual inventions of excuses, an eventuality that lasted
for years. This postulation already looks to be a form of coercion but not at the level
of types and the compositional processes are hence rather obscure or at least quite
stipulative. Furthermore, this large eventuality must presumably be an activity, but
why isn’t a sum of achievement events also an achievement event? Similar puzzles
crop up frequently when one tries to specify the aspect of a sentence realizing
eventuality from the lexical aspect of the verb, and the semantics of temporal
modifiers and inflection.
In TCL, instead of trying to specify the aspect of the sentence denotation during
composition, the compositional process puts constraints on the proposition realizer.
These constraints can make use of all the information in the verbal complex at once,
unlike the compositional calculation of eventuality aspect. For it is only after the
verbal complex has been formed that an eventuality will be introduced into the
logical form and the content of a sentence, if one is introduced at all.
Sometimes the application of a temporal modifier to a verbal complex can affect
the meaning of the predication in the verbal complex. This is a more complex
form of coercion than the simple cases in (1), and more like the one we had with
enjoy. Consider again (2). The verb leave subcategorizes for two logically related
types for its argument: institution, location, and the location argument is
often associated with a physical object. The noun husband, on the other hand,
types its lambda abstracted variable as an agent which is a subtype of physical
object, and this is the type it passes on to the type presupposition parameter. As
it stands, this type does not satisfy the type presupposition of leave, but there is a
natural transformation from physical objects to their locations, which the type
adjustment mechanisms can make use of in a way analogous to that in which we made
Temporal modification 33

use of the transformation from individuals to eventualities they participate in. This
gets us the reading of Mary leaving the location of her husband, which is included
at the airport. To get the other reading of (2), the one prominent in (2b), we need
to look at another map that is part of the type of husband. There is a natural
morphism from the male individual participating in the marriage to the institution
of marriage—i.e. an arrow from the type physical object to institution,
licensed by the word husband. To get the preferences in readings here, we need
to make one more assumption, and that is, that which coercion is licensed is
the one that is more likely—and this could just be a matter of probabilities of
distributions of readings that speakers have come across. As the temporal adverbial
five years ago makes the second coercion more likely, that is the reading selected
for (2b).
This approach to verbal temporal modification and coercion provides a general
mechanism that makes it possible to continue to refer to the individual picked out by
her husband in subsequent discourse, but also the location of the individual or the
institution. The account thus predicts copredications like she left her husband who is
a stockbroker five years ago to be well-formed and can give a compositional analysis
of them using our techniques of local type presupposition justification.

1.5.2 Temporal nominal modification

Tonhauser (2006) investigates temporal modification in the nominal domain. She
claims that nouns of many but not all subtypes of e support temporal modification,
along with similar modal examples.
(35) The president invited the former prisoners to the banquet.
(36) The freed prisoners are now recuperating in the hospital.
(37) #The former lions now lie buried.
(38) Sarah Palin is a possible presidential candidate.
(39) #The possible zebras are coming into view.
Tonhauser claims that predications involving natural kinds and final stage level
predicates like mother and daughter are not subject to temporal modification, which
is not quite correct, as I show below.
What goes on in such modifications? Let’s consider a noun like prisoner. Its
morphology suggest that it is derived from a verb imprison. Deverbal nominals
that end in -er typically denote participants in events or states that are realizers of
propositional complexes involving the verb from which the noun is derived. Tem-
poral adjectives and perhaps modal ones as well are counterparts of the tense and
modal features in the nominal domain, and act in the same way. But interestingly,
they interact not on the denotation of the noun but on the associated realizer. For
instance, former prisoner would have the logical form in (40).
34 Nicholas Asher

(40) lxlp 9y9t(realize(y,^[9z emprison(z, x)], p) ^ y < t ^ t ¼ ?)

What is puzzling is how we get to (40) from the logical form of the noun prisoner,
given in (41) and the entry for former, which is given in (42).

(41)  calPlxylp P p argprisoner
1 : human argprisoner
2 : evtÞ
(realize(y;^[9z emprison(z; x)]), p)
P ranges over adjectival denotations in (40) and takes the entry for former as an
(42) lx 9t P p argformer
1 : evt r(?(Hd(P) ) ) Þ (x) ^ x  t ^ t ¼ ?Þ(x)
In words, this entry says that the temporal modifier requires a realizer as an
argument (a time or an eventuality), but it allows for a transformation taking the
head type of its actual argument into a realizer type r. r is a polymorphic type that
denotes the maximal temporal entity that realizes the proposition resulting from
predicating some underspecified property or relation to the actual argument. This
underspecified property or relation is typically resolved to a property or relation
introduced in the head noun with which the adjective combines—in this case, the
property is that of being emprisoned. But the property may also be contextually
supplied, as we will see shortly. The temporal adjective temporally locates its realizer
type argument prior to some contextually specified time. Typically this is what
linguists call the event time of the sentence introduced by the tense feature in the
grammar, but in principle it could be also the speech time or in fact any salient,
accessible temporal entity introduced in the discourse.
To construct the denotation of the complete noun phrase, we note first that the
argument with type human is incompatible with the type requirements of the
adjective, which demands an argument that is either an event or a time. We now
need to justify these type presuppositions together to construct the logical form in
(40). In this case the justification requires only a simple coercion operator that
makes the realizer argument available to the adjective and so places the effect of the
adjective within the scope of an existential closure of the realizer argument. This is a
relatively trivial operation in the l calculus, but this coercion operator is restricted to
deverbal nominals with -er and -ee suffix morphology, in which both the realizer and
participant arguments appear to be predicable by modifiers.
Not all nouns that accept temporal modification are plausibly understood to be
deverbal nouns. For example, bicycle chain, crew, house, or mother in law accept
temporal modification as in
(43) my former house
(44) my previous bike chain
(45) the previous crew
Temporal modification 35

(46) my previous cat

(47) a future mother in law
but they are not plausibly analysed as deverbal nouns. They are, however, naturally
associated with certain activities that can serve as realizers—living in a house, a
crew’s running of a ship or an aeroplane, or ownership or many other relations
suggested by the Saxon genitive used in conjunction with the temporal modifier.12
Note that with the Saxon genitive construction, even nouns that denote natural
kinds license temporal modification. In fact it is the Saxon genitive that provides an
activity or state associated with the noun and that consequently makes determinate
the morphism that the coercion predicate can exploit to get a temporal entity of a
suitable kind in order for the construction to succeed.
To be more specific, let’s look at how we might compose the meaning of my
previous cat. Cat has roughly the lexical entry of tree:
0 0
(48) lPlxlp P p argcat
1 : pÞ(x)(lvlp cat(v, p ) )

Let’s assume that previous has roughly the same lexical entry as former. The two
terms cannot combine, unless we invoke the natural transformation that takes us
from the head type of the noun p to a realizer. This requires us to specify some
relation or property that holds of the cat. In our example it is the Saxon genitive my
that specifies an activity (that of owning), following the lines of the analysis of Asher
(2011). Because the realizer is maximal, the transformation entails that the time to
which the realizer is prior itself realizes the negation of the contextually specified
(49) lP lp9x(9v (cat(p * x: p * v: evt)(x) ^ R(v, ^ [own(sp, x) ^ cat(x)]) ^ v  t ^
t ¼ ? ^ R(t, ^ :(own(sp, x) ^ cat(x) ) ) ) ^ P(x) )
If we resolve t to the speech time of the main sentence, we get a typical temporal
interpretation of a sentence like my previous cat used to catch birds: prior to the
speech time, a cat, which I no longer have, used to catch birds.
Tonhauser’s observations concerning natural kind terms and final stage level
predicates are partially correct. It’s often hard to figure out what morphism links
an instance of a natural kind to an activity, though with the Saxon genitive
construction the relation is grammatically given. In a particular discourse context
describing a society in which people switch mothers or fathers on a regular basis (in
the institutional sense if not the biological sense) or in which foetuses are able to talk,
it should be all right to say my future mother. The former lions is difficult to interpret

12 For details on the analysis of the Saxon genitive as a construction introducing an eventuality or
realizer, see Asher (2011).
36 Nicholas Asher

precisely because there is no contextually, lexically, or grammatically given relation

that could fill in the underspecified predication entailed by the coercion.

1.6 Conclusions
In sum, there is a lot that coercion mechanisms can do when it comes to temporal
modification. The mechanism is powerful but well-defined and it gives us a very
different view of temporal modification from what Davidsonian or neo-Davidsonian
approaches provide. In this chapter, I’ve sketched out a general theory of coercion
and applied it to some examples of temporal predication. More investigation is
needed to determine the empirical adequacy of the theory, but the initial results
seem quite promising.

Temporal reasoning as indexical



2.1 Introduction
Temporal reasoning in time about time is paradigmatic of situated reasoning in the
first place because the reasoning agent himself situates the past as separated from the
future, when receiving the information from which a conclusion is to be drawn. Any
inflected phrase is in this sense obviously indexical upon this time of reception or
production of the input to the reasoning, since true past tense statements require
reference to events or states preceding the time of their reception/production, and
true present tense statements require reference to an event or state temporally
including it.
For instance, although from (1a) we cannot infer at what time the described event
took place, we do infer that it must have happened before (1a) was produced/received,
as it is past tense input to the present perfect tense conclusion in (1b).
(1) a. John walked to school.
b. John has walked to school.
An inference rule for a system of temporal reasoning in natural language should
accordingly validate the transition from such a simple past tense accomplishment
(1a) to its present perfect tense (1b). But no rule should make an inference from a
past progressive accomplishment to its present perfect progressive tense valid, as (2b)
is not only violating a morphological agreement rule, but also constitutes an
intuitively incorrect conclusion, if it is based on (2a). Without the modifying
temporal adjunct clause, the inference from (2a) with a past progressive accomplish-
ment to (2c) with a present perfect progressive tense seems, however, unproblematic.
(2) a. John was walking to school, when a bus hit him.
b. *John has been walking to school, when a bus hit him.
c. John has been walking to school.
38 Alice G. B. ter Meulen

As is well known from the earliest discussion of the ‘imperfective paradox’ in Dowty
(1979), only divisible activities, sometimes also called ‘atelic eventuality descriptions’,
supports a simple past conclusion on the basis of a past progressive premise,
illustrated in (3a–b).
(3) a. John was walking.
b. John walked.
Accordingly, the inference rules for temporal reasoning in natural language will also
have to take aspectual properties of clauses into account, if they are to do justice
to our intuitions on the validity of inferences with temporal information. The
information that there was the causal interference of the bus accident, possibly
preventing John from completing his walk to school, is altogether independent
from this aspectual information, pace van Lambalgen and Hamm (2005).
It is also well known that the order in which sentences are presented in discourse
may affect the interpretation of their reference to the events they describe. This
constitutes the second reason why temporal reasoning is paradigmatic of situated
reasoning: that is, prior discourse situates the interpretation of a tensed clause by
constraining its temporal reference. In the absence of conflicting information in the
context, an eventuality referred to by a simple past tense clause is temporally
included in the state or eventuality introduced by its preceding clause, only if
it is either stative (including progressive) or a description of an activity. Otherwise,
the order of the presentation of clauses determines the temporal order between the
events they refer to. For instance, from (4a) we infer that John crossed the street
while walking to school, but from (4b) we infer that he saw Mary after completing
his walk to school.
(4) a. John was walking to school. He crossed the street.
b. John walked to school. He saw Mary.
Obviously, the validity of such inferences also depends on the presuppositions
to be accommodated of their premises’ content, on topic continuity, and
compatibility/consistency of descriptive content. The temporal inference toolkit of
the Dynamic Aspect Trees (DAT) in ter Meulen (1994, 1995, 2000) was designed
specifically to account for the validity of such temporal reasoning patterns,
illustrated in (1)  (4), and some additional ones. Constraints on the introduction
and portability of stickers in DATs, representing information contained in the
conclusion, capture most core intuitions regarding the logical properties of temporal
reasoning in English, based on their tense and aspect.
Besides such well-known patterns of situated temporal reasoning in (1)  (4), the
location of the reasoning agent contributes essential situated information in evalu-
ating statements with indexical verbs like come, arrive, and appear, on the one hand,
Temporal reasoning as indexical inference 39

and go, leave, and disappear, on the other.1 Such verbs presuppose that the agent’s
location is fixed once all input has been interpreted and the conclusion must be
evaluated. Since these verbs describe actions of moving towards the agent, or away
from him, they should be considered indexical predicates, as their evaluation requires
a contextually fixed speaker location, just as the classical indexical adverbs here and
now or the first person pronoun do.2 This chapter is intended to contribute some novel,
elementary insights into this rich domain of temporal reasoning patterns, where such
descriptive indexical predicates may appear in the conclusion though none of the
premises contain any occurrence of these indexical predicates. The interaction of
indexical predicates with aspectual adverbs, quantifying over the internal structure
of events, proves to constitute especially interesting patterns of temporal reasoning
that require proper investigation in natural language semantics and pragmatics.

2.2 Indexical predicates and aspectual adverbs

If a hypothetical situation is described statically by (5a), we cannot infer anything
about what is going on, what has happened, or what may happen.
(5) a. There are three students here.
b. There are still three students here.
c. There are already three students here.
However, the aspectual adverbs still and already contribute essential information
about the future or the past of the described situation, assuming constant reference
to the current location with here.3 Lacking any specific assumptions or background
information, we apparently draw the conclusion from (5b) that students must be
leaving and some have left, hence more than three students were here earlier. From
(5c) we infer that students must be arriving or some have arrived, hence that fewer
than three students must have been here before. Doing justice to such inference
patterns introducing descriptive indexical predicates in the conclusion requires the
reasoning agent and his location to remain fixed by the context for the inference, i.e.
interpreting the premises in sequence.
The DRT semantics of the English aspectual adverbs still, already, not yet, no longer
in Smessaert and ter Meulen (2004) primarily captures the interaction between three
truth-functional aspects of their meaning, i.e. their factual or descriptive content. Their
meaning is based on three ingredients specified in (6):
See Barwise and Perry (1983), Perry (1992), and the updated survey of Situation Theory by Moss and
Seligman in van Benthem and ter Meulen (2011: 253–335).
Adjectives and prepositions as left, right, behind, or above should for the same reason also be
considered indexical predicates, but since they do not seem to be directly involved in interesting forms
of temporal reasoning, they are disregarded in this chapter.
Italics are used in this chapter to cite any words occurring in data, to avoid the harrowing questions
regarding the denotation of quoted indexicals.
40 Alice G. B. ter Meulen

(6) (i) the indexical interaction between the speech time, the current state and
its past or future;
(ii) the two polarity transitions START (0/1) and END (1/0);
(iii) temporal dependencies created by the temporal operators SINCE and
UNTIL, with their established semantics.
The stative, factual temporal content of the four basic English aspectual adverbs is
specified in (7), modally introducing past ( ^) or future (^!) polarity reversals,
where D is the domain of entities to which newly introduced entities are added, and
cn the context at stage n of its construction, fixed if aspectual adverbs modify stative
descriptions, otherwise updated to a next stage.
(7) Basic aspectual adverbs
a. D, c0 [VP lx [INFL not yet P(x)]] D’, c0 ¼
D, c0 [ << P’(x,  ) >> & ^![<<END (P’(x,  ))>> & UNTIL
(<<END (P’(x,  )), (P’(x,  )) >>)]
b. D, c0 [VP lx [INFL already P(x)]] D’, c0 ¼
D, c0 [ << P’(x, þ ) >> & ^[<<START (P’(x, þ ))>>
& SINCE (<<START (P’(x, þ )), (P’(x, þ ) ) >>)]
c. D, c0 [VPlx [INFL still P(x)]] D’, c0 ¼
D, c0 [ << P’(x, þ ) >> & ^![<< END (P’(x, þ )) >>
& UNTIL (<<END (P’(x, þ ) ), (P’(x, þ )) >>)]
d. D, c0 [VP lx [INFL not anymore P(x)]] D’, c0 ¼
D, c0 [ << P’(s, x,  ) >> & ^ [<< START (P’(s, x,  )) >>
& SINCE (<< START (P’(s, x,  )), (P’(s, x,  )) >>)]
When adopted as update instructions to add new conditions to the current DRS at
the construction stage cn, these clauses in (7) capture the basic temporal inference
potential of the four aspectual adverbs in English. The since and until clauses serve to
constrain the polarity transition to the last past or the first future one, given the
current state of the context, and hence not to some arbitrary, distant one. This
excludes any possible, intervening polarity transition for the relevant predicate, or
gaps in the event it refers to. Hence these conditions ensure the continuity of the
polarity, but also create an essentially indexical dependency on the current state of
the context cn. It is clear that already and still are veridical, whereas not yet and no
longer are anti-veridical.4

Let Op be a monadic propositional operator. Then:
(i) Op is veridical just in case Op[ p ! p ] is logically valid. Otherwise Op is nonveridical.
(ii) A nonveridical operator Op is anti-veridical just in case Op[ p ! p ] is logically valid.
Still and already are veridical operators, and not yet and no longer are anti-veridical.
Temporal reasoning as indexical inference 41

2.3 Scalar inferences and polarity in aspectual adverbs

The intuitions to be accounted for in developing this theory of temporal reasoning in
English are to a large extent based on the semantic monotonicity properties of the
VPs modified by the aspectual adverbs. For instance, if there are still three students
at t, then at any later time t’ there obviously cannot be more than three students, but
there may be fewer than three students, as long as the presupposition that students
are leaving remains a valid assumption. Similarly, if there are already three
students at t, then at any later time t’ there cannot be fewer than three students,
but there should be more than three students, as long as the presupposition that
students are arriving remains a valid assumption. These presuppositions hence serve
to constrain the set of later times to be considered, excluding times after the event of
students leaving or arriving has ended. The numerical scale, counting in this
example how many students there are, induced by the aspectual adverb still is
hence decreasing into the future, whereas for already it is increasing.
These monotonicity properties explain why (8a) is perfectly fine and coherent, as
the decreasing still shares its monotonicity with the decreasing only at the scale’s
lower end. But (8b) is prima facie unacceptable for semantic reasons, as the increas-
ing already ordinarily cannot modify the decreasing only.5
(8) a. There are still only three students here.
b. *There are already only three students here.
Often in aspectual interpretation specific adjustments in the context may force or
‘coerce’ a special interpretation, overruling the otherwise common assumptions.6
In this way, we may coerce (8b) into expressing the speaker’s judgement that it
is relatively early to have only three students here, if syntactically already modifies
the polarized NP only three students. This syntactic structure would cancel its
presupposition that students are arriving associated with already, retaining only
its temporal meaning concerning its early occurrence. It would project instead the
presupposition, triggered by the decreasing only, that the students must be leaving.
It might constitute an interesting investigation in the semantics/phonology interface
to study the prosodic contours of such coerced interpretations with conflicting
monotonicity properties, as well as its effects, if any, on processing complexity.
From a purely theoretical point of view, one might like to conjecture that their
prosody could be quite different from the prosodic properties of the straightfor-
wardly interpreted (8a), which would have repercussions for their syntactic analysis
as well, if a compositional account is required.

This is strongly reminiscent of the constraint on coordination which requires NPs to have the same
monotonicity properties for conjunction and disjunction; see Partee, ter Meulen, and Wall (1993: 382) for
discussion of this constraint.
See Pickering et al. (2006) and also Asher, this volume.
42 Alice G. B. ter Meulen

The negative polarity data in (9) show that, as one may expect, in simple
indicatives only the anti-veridical aspectual adverbs no longer and not yet license
the polarity any; as their interpretation contains a negative polarity on the descriptive
predicate P. Of course, the downwards entailing interrogative context in (10a–b) and
the downwards entailing antecedent of a conditional (10c–d) themselves also license
any; hence in such contexts still and already contribute merely the subjective timing
information, i.e. the speaker’s judgement that it is relatively late for students to leave or
early for them to arrive, respectively.
(9) a. *There are still any students left. b. *There are already any students here.
c. There are no longer any students left. d. There are not yet any students here.
(10) a. Are there still any students left? b. Are there already any students here?
c. If there are still any students left, . . . d. If there are already any students here, . . .

In the next section such indexical information is reconsidered and structured to

be focus information, where the remaining content of the clause is backgrounded,
together with the presuppositions constraining the polarity reversals.7

2.4 Focus and information structure of aspectual adverbs

Focus semantics is designed to account for the dynamics of the interpretation
process as sequential updates of the context: UPDATE(cn, cn þ 1). The input to the
update is a syntactically parsed phrase, partitioned into given (topic) and new
(focus) information, either possibly empty. Given information is considered to be
either directly available, entailed, or presupposed by the given, current context cn, or,
otherwise, accommodated before the update, and it is phonologically often marked
as deaccented. It is considered to be ‘backgrounded’, i.e. not under discussion,
negotiation or ‘at stake’. The new, or focus, information is the content with which
the current context is updated to the next phase of interpretation cn þ 1. It is
information that provides new values for unfamiliar parameters by extending the
set of assignments to parameters fixed earlier and their alternatives. It is typically
negotiated in a dialogue or interactive situation to be included or excluded as shared
information, assumed to be true by all the participants, even if only for the sake of
To account for focus, the semantic interpretation adds a space of alternatives to
the background, for which the focus provides the current value. This space of
alternatives is most often considered to be a set of propositions, truth value denoting

See Smessaert and ter Meulen (2004) for the semantics of prosodically marked aspectual adverbs as
focus information.
Temporal reasoning as indexical inference 43

semantic objects, but this is by no means a necessary assumption. Lambda abstrac-

tion over a variable of the requisite type creates the space of alternatives of suitable
type in the background, and the focus serves as argument of that lambda abstraction
in the update of the context.
Given the informal discussion of our intuitions on inferences with aspectual
adverbs and numerical determiners, two dimensions of focus alternatives are
required: one varying over the cardinality of the set of students who are here, and
the other over moments of time which could have been the current one, if the event
were happening faster or slower than it actually is.
Accordingly, the interpretation of (5b) requires first accommodating the back-
grounded information in the Common Ground (CG), then the focus structure is
determined to be the set of alternatives to the current situation, and finally the
asserted focus information that the number of students is three is added.
(5) b. D, cn j[There are [FOC still three] students here]jD’, cn þ 1.
CG (i) there are students here ¼> 9 X [ students (X) & loc(X) ¼ lcn]
(ii) students are leaving ¼>
9 Y,e [students(Y) & leave(e,Y) & e  tcn]
Focus structure: l<n, t’> [ n¼ jXj & n # 3 & t’# tcn]
Focus: l<n, t’> [n¼ jXj & n # 3 & t’# tcn] (3, tcn þ 1)
The focus alternatives are lambda-abstracted over ordered pairs consisting of the
cardinality of the set of students and temporal reference points, to provide a set of
two-dimensional alternatives where either there would be fewer than three students or
the number of students was three at an earlier time. The focus or new information is the
actual state in which there are three students. The presupposition triggered by the
decreasing still is (ii) and should be accommodated into the CG before the focus
information is added. Furthermore, the decreasing nature of still entails that at any
later time there are still three or fewer students, while veridicality is guaranteed by (i).
For the interpretation of (5c), a closely analogous information structure is
required, but the cardinality of the set of students had to be smaller than three in
the past and students are arriving, given the increasing monotonicity of already. The
focus structure of alternatives is the set of ordered pairs consisting of numbers
higher than three later, while the focus information itself adds that the current
cardinality of the set of students is three.
(5) c. D, cn j[There are [FOC already three] students here]jD’, cn þ 1.
CG (i) there are students here ¼> 9 X [ students (X) & loc(X) ¼ lcn ]
(ii) students are arriving ¼>
9 Y,e [student(Y) & arrive(e,Y) & e  tcn]
Focus structure: l<n, t’> [ n¼ jXj & n $ 3 & t’$ tcn]
Focus: l<n, t’> [n¼ jXj & n $ 3 & t’$ tcn ] (3, tcn)
44 Alice G. B. ter Meulen

Comparison of (5b) and (5c) clearly brings out that the difference in monotonicity
between still and already is backgrounded, but plays an important role in structuring
the space of alternative values, while the focus information in both statements (5b)
and (5c) is actually the same value applied to a different lambda abstraction to create
the focus structure of alternatives.
Assuming now the lexical semantics of x leave y as describing the event as a
transition between the state of first x’s being at y to later x’s not being at y, cf. (11a),
we infer from (5b) with present perfect tense that some of the students, who were at
the location earlier, have now left. Mutatis mutandis, the inference from (5c) is that
some of the students who were not here yet, have now arrived, exploiting (11b) as
the lexical semantics of arrive.
(11) a. [[leave (e, x, y)]] ¼ lx,e(loc(x) ¼ loc(spc) & move (e,x) ¼> F(loc(x)
6¼ loc(spc)))
b. [[arrive (e, x, y)]] ¼ lx,e(P(loc(x) 6¼ loc(spc)) & move (e,x) ¼> (loc(x)
¼ loc(spc)))
The anti-veridical aspectual adverbs share the common-ground conditions and the
focus structure with their veridical counterpart, but negate the focus information.
Accordingly, not yet three will be interpreted against the same backgrounded
information as already three, but denies that the actual number of students is
three. Ordinarily, no longer three also shares the backgrounded information of still
three, but denies that there currently are three students. It appears also to be possible
with no longer that the (ii) condition concerning the ongoing leaving of the students
may be dropped from the CG without coercion effect. In such diminished CG, no
conclusion is supported regarding the number of students who were here earlier. If
the CG contains the assumption that the students are respectively arriving and
leaving, the inference that there must now be fewer than three students here is
validated in both cases. Further details of this outline of a direction of formalization
may be developed in the various systems of focus semantics currently available.

2.5 Conclusions
Aspectual adverbs are seen to contribute important temporal structure to the
information that is communicated in their use with numerical determiners. They
create cohesion within a context by presupposing a stage level event, even when
subject is generic, serving as topic. The standard semantics of aspectual adverbs
hence needs supplementing in information structure, when the backgrounded,
shared information is partitioned with the focused information about the actual
Aspectual adverbs are also often used with polarity particles like only to
create discourse cohesion by opening up ways to strengthened alternatives based
Temporal reasoning as indexical inference 45

on a presupposed state that remains in force. For instance, in (12) the economy is
assumed to continue to thrive as it did due to export, but now it thrives even more
because of consumer behaviour.
(12) The economy thrives no longer only due to export, but also to consumer
Consumer behaviour is not ordinarily thought of as a clearly stronger alternative to
export as a force of economic health, but the scalar strengthening is induced by this
context in accommodating the ongoing event presupposition of no longer only that
export already makes the economy thrive and continues to do so.8
English may seem to function quite logically with regard to the observed duality of
its system of aspectual adverbs, in comparison to other Germanic languages such as
Dutch and German. They show a richer, more compositional tapestry of aspectual
modification and polarity reversal. In particular, their counterpart of still may be
used, often with marked prosody, as in the Dutch (13), with focus information about
an ongoing monotone increasing event in an indexical predicate, adding more to
whatever is presupposed to be already there, voiding its CG event assumption (ii).
(13) Er komen nog drie studenten aan.
There come still three students on.
‘Three more students are coming.’
The linguistic investigation of universal versus variable properties of aspectual
information structure has never disappeared from the linguistic agenda. Its insights
may have ripened enough now for new dimensions of formalization in terms of
notions of information structure to account for the essential indexicality of temporal
reasoning in natural language beyond mere temporal reference.

See, for some inspiring discussion of the force of not only, Horn (2000).

Perspectival interpretations of tenses


3.1 Introduction
This chapter deals with cases where the temporal reference indicated by tenses
cannot be understood literally, that is, proves irrelevant unless one of the compon-
ents of the representation (the deictic point S, the reference point R, or the eventu-
ality E) is modified under the pressure of contextual consistency or relevance.
I suggest that these situations imply a change in the perspective, or point of view,
from which the eventuality is grasped. I tentatively discuss in the conclusion whether
these linguistic facts call for a simulative explanation.
Linguistics usually views tense as a referential device (typically ‘operators’ or
‘procedural items’)1 pointing on moments associated with the occurrence of even-
tualities (events, states, activities). We will ignore in what follows the various
sophistications on offer in semantics, logic, and philosophy about what time and
tense really are, sticking to the naive conception that moments indicated by tenses
are somehow calculated, that tenses have an indexical component (they denote via
the notion of speech time) (and an anaphorical one, cf. Partee 1973, which we don’t
need to discuss in this chapter), and that accounting for tenses implies, as Recanati
(2006: 64) puts it after preceding philosophers, “the need to capture the primacy of
the internal perspective in dealing with the representation of time in language and
The ‘moments’ associated with tenses are points and intervals bearing temporal
relations, following the seminal work of Reichenbach (1947) and the many develop-
ments and refinements of his initial schematic proposal. The nature of tense as an
indicator of temporal reference has been subject to many discussions, which are

See Recanati (2006: Chapter 6) for a review of philosophical debates on the nature of tense. Recanati
finds an advantage in treating tenses as operators since it allows us to “capture the internal perspective on
time which characterizes our thought and talk” (p. 64). See Nicolle (1996), Moeschler (1998), and de
Saussure (2003, 2011a) for discussions on tenses as procedural (vs. conceptual) expressions.
Perspectival interpretations of tenses 47

widely represented elsewhere in this volume, notably the diachronic primacy of

aspect over tense, of space over time, the possible reducibility of tense to modality
(see Jaszczolt 2009) or to space and distance (Chilton, this volume), and the anyway
complex intrication of temporal and modal meanings (see Gosselin’s, Rocci’s, and
Jaszczolt’s chapters in this volume). I will not focus here on these issues but I will in
the end have to consider some of them incidentally.
In a number of cases, some of them very well known, tenses do not or not only
refer to the time(s) they should refer to according to their intuitive semantics: some
present tense utterances tell about a past or a future; some utterances with a past
tense tell about a present or a future; some future tense utterances are about the
present or the past. Either these cases call for a refinement of the theory, and then
various possibilities are on offer, from the import of notions like modulation or
accommodation to the contribution of complex compositional rules and semantic
constraints; or they suggest the need for a completely other theory where time is not
central to tense. We will not take a final decision on these possibilities, leaving this to
philosophers. Our contribution regards the possibility of accounting for such cases
through general pragmatic principles that take times as inputs, whatever the right
metaphysical theory that explains these times, which are empirically taken to be
understood by speakers.
Indeed, some kind of conflict does often arise between the information conveyed
by the tense as such and either a contextual assumption about the utterance being
about some other time (as in That will be the postman when the doorbell rings) or a
conflicting adverb (as in I’m leaving tomorrow). Such cases raise serious issues. The
questions I am thus going to address are: How is it that such constrained referential
morphemes as tenses allow for switches in their temporal denotation? How do we
cognitively build such switched representations? What exactly is switched in such
cases? What does it tell us about the properties of tense as a morpheme and about
our ability to represent time through their indications?
My claim is that there are three ways to switch the temporal meaning of a tense
under the pressure of other linguistic and non-linguistic, contextual, information,
which I will discuss in the light of three such cases in French, with occasional
comparison with English. I will successively consider narrative and other non-
background uses of the French past imperfective tense, future time reference with
the French composed past (a tense morphologically equivalent to the English present
perfect), and epistemic futures, which look quite similar in French, English, and in a
number of other languages.
An important caveat has still to be made: not all apparently switched uses of tense
are truly such. A typical example of this is when the waiter, having noticed your
calling, utters I’m coming when precisely going away. In that very case, some
might say that the utterance describes a future state of affairs, which implies
either that the present tense is not truly such but something else, for example an
48 Louis de Saussure

ambiguous morpheme between present and other possible times, or a ‘temporally

neutral’ morpheme, as many linguists and philosophers have suggested.2 On the
contrary, one might stick to the simple and naive idea that the present tense is about
the present time, as I do. Then, either the present of “I’m coming” uttered by the
waiter going away represents a future time and has to be accommodated in a similar
vein as Dowty’s (1979: 156) futurate presents (I’m leaving tomorrow) or it represents
indeed the present time, which is counter-intuitive at first sight, but which
I nonetheless suggest. Keeping in mind that verbal communication is about the
pragmatic world, the waiter’s utterance is better explained as representing the present
state of affairs (the waiter is coming), so that all the pragmatic consequences of that
very representation are made accessible to the hearers: the waiter will be here in no
time, the call has been noticed, etc. That the contextual information contradicts the
truth of the representation is here another matter completely: doing so, the waiter
informs that his going away is precisely an event that should not be taken into
account, the only relevant information being that he’s on his way. In such a case,
I claim, the present remains a true present entirely, without calling for anything like
a deictic shift or whatever other kind of change in the representation (nor even for
an extended-now interpretation). Said in Gricean terms: that the speaker violates
ostensively the maxim of quality calls simply for re-establishing cooperation in terms
of implicatures. What the tense itself represents here is in no way changed, switched,
or accommodated.
In what follows, I will turn to utterances whose understanding in context does call
for a change in the tense’s structure, its placement on the line of time, or a
component of its meaning. I will turn first to non-background meanings of the
French imperfective past, which, I claim, imply a modification of the R-point; then
I will discuss future switched present perfects in French, before I turn to epistemic
futures dealing with, I suggest, a modification of the eventuality itself, thus, sche-
matically, a modification operating on E, since E represents the moment where the
expressed eventuality happens.

3.2 Narrative and other non-background use of the

French imperfective past
The French imperfective past tense, imparfait (IMP), differs from the English past
progressive in that it doesn’t imply dynamicity (Molendijk 2005): the IMP is
compatible with any predicate and normally implies a representation for which
the eventuality is conceived of as being in the course of happening (activities and
accomplishments) or ongoing (states) at R; with achievements, the IMP-utterance

See, for example, Recanati (1995, 2006: 74), and Sauerland (2002), among many others.
Perspectival interpretations of tenses 49

normally triggers a representation of the preparatory phase, just as happens with

other types of durative marking. In all these cases, IMP-utterances imply that the
eventuality is unbounded so that no temporal ordering is expected between IMP-
utterances (Kamp and Rohrer 1983). Yet with accomplishments and achievements,
and also with activities (but less easily with non-agentive activities), IMP-utterances
can also trigger an unexpected reading for which the eventuality is bounded,
allowing forward temporal ordering with series of such IMP-utterances, therefore
called ‘narrative’-IMP-utterances; we turn to this case of interest shortly.
Typical IMP-sentences are durative, as exemplified below, and are usually de-
scribed as setting up a background (Weinrich 1964):
(1) Je prenais mon bain lorsque le téléphone sonna.
roughly equivalent to (1’):
(1’) I was taking a bath when the phone rang.
(2) Marie connaissait bien Max.
roughly equivalent to (2)’ but impossible with a past progressive in English (2’’):
(2’) Marie knew Max well (at this time / already . . . )
(2’’) *Marie was knowing Max well.
A simple formula for the IMP which roughly fits with these canonical, standard,
cases, can be something like (3): the reference point R, inherited from some other
sentence with a perfective past or a punctual adverb, is situated inside an ongoing
eventuality E (Reichenbach uses a notion of ‘extended’ time reference for this, which
he applies to the progressive in English as well), and R precedes S:
(3) R  E & R – S
The meaning of IMP-utterances with achievements is more complicated and calls
for an extended discussion that we can’t pursue here. Typical interpretations of
IMP-utterances with achievements are about a preparatory phase (as usual with
durative marking of achievements), as with (4):
(4) Paul atteignait la ligne d’arrivée quand la police arriva.
‘Paul was reaching the finish line when the police arrived.’
Yet when certain contextual constraints are met, IMP-utterances behave in a puz-
zling way, called ‘narrative’, where the completion of the eventuality, and thus
boundedness is interpreted. This is apparently contrary to the formula in (3),
which leaves the eventuality open after the reference point, which in turn blocks
50 Louis de Saussure

temporal progression in discourse. Narrative cases are, however, obviously compat-

ible with temporal progression, and thus aren’t about a descriptive background. They
translate in English not with progressives but with simple pasts, even when the VP is
dynamic. Here, what is represented is an event:
(5) Le lendemain, Paul partait.
‘The next day, Paul left’ (*was leaving).
(6) La clé tourna dans la serrure. Monsieur Chabot entrait, posait son pardessus et
s’asseyait dans son fauteuil d’osier
(adapted from Simenon, quoted by Tasmowski-de Ryck 1985)
‘The key turned in the hole. Mr Chabot came in, hung his coat on the hanger
and sat down’ (*was coming in; *was hanging, *was sitting).
(7) A huit heures, Marie trouvait ses clés et sortait.
‘At eight, Mary found her keys and left’ (*was finding; *was leaving).
(8) Tintin mettait pied à terre sur l’aérolithe et plantait le drapeau du FERS.
‘Tintin stepped on the aerolite and planted the FERS flag’ (*was stepping; *was
It is argued in de Saussure and Sthioul (1999, 2005) that the right way to go into this
puzzle is to assume that IMP gets accommodated when boundedness and/or tem-
poral progression is mandatory because of contextual assumptions (as with (6) and
(7)) or of an adverb ((5) and (7)). The question then is to know into what kind of
representation the IMP gets accommodated, and how, without having to assume
very counter-intuitively two different versions of IMP in the lexicon, one without
boundedness and one with boundedness.
IMP has a number of other meanings which don’t comply with the formula in
(3) either; these are usually called modal in the French literature, although they
rather deal with (among other meanings) conditional counterfactuality, indirect
speech acts, and free indirect speech and thus are not all ‘modal’ in the usual narrow
understanding of ‘modality’:
i) (conditional) counterfactuality, as in (9):
(9) Une minute de plus et le train déraillait.
‘One minute more and the train derail-IMP’.
meaning something like (9’):
(9’) One minute later, the train would have derailed.
Perspectival interpretations of tenses 51

ii) indirect speech acts (similarly as with the English simple past):
(10) J’avais une question.
I have-IMP a question.
‘I had a question.’
iii) the ongoingness in the past of a thought occurring to a third party about some
eventuality (free indirect speech):
(11) La bonne ( . . . ) parlementa quelques instants avec un homme resté en bas, dans
la rue. Il venait chercher le médecin ; il avait une lettre. Nastasie descendit les
marches en grelottant, et alla ouvrir la serrure et les verrous, l’un après l’autre.
(Flaubert, Madame Bovary)
‘The servant talked for a moment with a man downstairs in the street. He look-
IMP [was looking] for the doctor; he have-IMP [had] a letter . . . (rendered as
the other character’s perception in free indirect speech)’
(12) L’existence de Madame de Rênal fut changée. Julien l’aimait donc bien, puisque
de lui-même il avait décidé de la revoir.
(Stendhal, The Red and the Black)
‘Mrs de Rênal’s life was changed. Julien love-IMP [loved] her indeed, since he
himself decided to see her again.’
(13) Mon père, en m’embrassant, fut saisi d’un tressaillement que je crois sentir et
partager encore. “Jean-Jacques, me disait-il, aime ton pays.”
(Rousseau, Confessions)
‘My father, kissing me, had a shiver, which I can still feel and share. “Jean-
Jacques, he tell-IMP [was telling] me, love your country”’ (see de Saussure and
Sthioul 2005).
The recourse to contextual information is often mandatory in order to specify which
reading is relevant, as in (14) (see also Sthioul 1998b: 213):
(14) Le chef de gare actionna le signal. Une minute plus tard, le train partait.
‘The station officer turned the signal (on/off ). One minute later, the train
leave-IMP [left/was to leave/would have left].’
(14) is ambiguous between three readings (the train actually left—narrative reading;
the train was to leave in a minute so that it was the right time to switch on the
signal—free indirect speech; the train was to leave in a minute and luckily the
attendant switched off the signal so that it did not leave—counterfactual reading,
probably best rendered in English with a conditional). Hence we conclude that
general pragmatic constraints of discourse processing are brought in (for example
52 Louis de Saussure

a principle of relevance of the kind proposed by Sperber and Wilson 1995) when it
comes to selecting the intended interpretation.
De Saussure and Sthioul (1999, 2005) suggest that in all cases where the formula
(3) does not hold, the variable which is normally saturated by R is saturated by a
moment at which a third party perceives the action as being in the course of
happening, whatever it may be in terms of Aktionsart. In other words, when a
narrative reading occurs with IMP, that is, when the completion of the eventuality
and/or a temporal progression has to be inferred for contextual reasons of relevance
or consistency, the reference point, which is not perspectival, is exchanged for a
perspectival point viewing the action as ongoing (thus an inner viewpoint, so to
speak).3 Narrative IMP utterances, often described as truth-conditionally equivalent
to simple past utterances, are predicted in fact to be distinct from simple past
utterances in that they imply a projection, or cognitive simulation, anchored on a
third-party viewpoint. A similar line of argument can be pursued with utterances
with the imparfait tense conveying free indirect speech (a third-party thought) and
counterfactuality: in all these cases, we suggest that the IMP does not merely
represent a past situation overlapping an anaphoric R-point but a simulated substitute.
This line of argument may seem quite speculative, although fitting well with
native speakers’ intuitions. But a decisive argument was provided by Sthioul
(1998b), who noticed that narrative IMP cannot always commute with simple pasts
(as previously held in the literature), so that there has to be a semantic difference
between the two; and more precisely, a marker of subjectivity makes narrative-IMP
natural and simple past odd. The argument is the following: the insertion of déjà
(roughly: ‘already’) is possible with IMP as a marker of surprise about the occurrence
of the eventuality (15), whereas it cannot bear this meaning with the simple past (SP)
(16) (examples adapted from Sthioul 1998b: 213):
(15) Le train quitta Londres. Une heure plus tard, il entrait déjà [surprise] en gare
de Birmingham.
‘The train left London. One hour later, it enter-IMP already [surprise] in Bir-
mingham station.’
(16) Le train quitta Londres. Une heure plus tard, il entra déjà [*surprise] en gare de
‘The train left London. One hour later, it enter-SP already [*surprise] in Birming-
ham station.’

I don’t refer here (exactly) to what Recanati (2006) names ‘perspectival thought’, although the
existence of a subjective viewpoint does indeed imply a contextual switch. The notion of ‘perspective’
here doesn’t deal with adopting another person’s assumptions, but rather with the non-propositional
grasping of (here) an eventuality, thus a very much lower-level process than figuring out the conceptual/
propositional thoughts of someone else (which is also an existing and common mental process, of course).
Perspectival interpretations of tenses 53

Both examples would translate in English with a simple past, but the specific notion
of subjectivity with the IMP (and absence thereof with the French SP) is lost in the
(15’) The train left London. One hour later, it already entered Birmingham.
Sthioul’s conclusion is that narrative-IMP sentences license markers of subjectivity;
therefore, he claims, the eventuality is not represented with an R-point anaphorically
obtained (in this case the event of leaving London), which can’t give relevance to the
utterance since entering Birmingham station cannot be seen from the point provided
by ‘leaving London’, but with recourse to a third-party perspective, here that of the
individual feeling surprise, which serves as a substitute for the R-point in formula (3).
Along this line of investigation, a point of representation is still situated inside an
ongoing eventuality, but that very point is the perspective of some third party figured
out as witnessing it. A further consideration worth paying attention to is that the IMP
still bears an imperfective flavour, representing a situation as ongoing even if com-
pleted and bounded. Both parameters—inner and perspectival viewpoint—confirm
the intuition often mentioned in the literature, that narrative-IMP offers a particularly
‘vivid’ rendering of eventualities, which makes it a preferred choice in a number of
cases such as newspaper narratives or focus in literary fiction; ‘vivid’, we claim, is just
an approximate notion of ‘represented as if seen and felt by a subjective perspective’.
Why it is that a perspectival substitute for R allows for boundedness and progression
is still a matter of debate.
This analysis also has the advantage of collapsing all non-background IMP into a
single category of subjective readings, implying that R is toggled with some third-
party viewpoint or ‘perspective’, since a similar analysis can be made of all of them,
while background IMP utterances are successfully explained without this perspec-
tival device. Hence all non-background IMP utterances correspond to (3’) with S’,
a projected non-egocentric, that is allocentric, deictic point, replacing the R-point
of (3):
(3’) S’  E & P – S
Of course, the conclusion of this is that the semantics of IMP involves a variable to
saturate either with R or with S’, so that (3) and (3’) are saturated versions of a basic
arrangement of coordinates; what leads to either (3) or (3’) is a procedure of context-
dependency and relevance-checking (de Saussure 2003; de Saussure and Sthioul 2005).
That the eventuality is represented through a perspective distinct from that of the
speaker at S makes it an allocentric usage of language, assuming that perspectives are
bound to an individual.4 An important question is whether such allocentric usages of

That a perspective point has to be bound to individualities is a complex and controversial issue. That
one can represent an object from the perspective of an inanimate object, as in ‘on the left of the plane’
54 Louis de Saussure

language, triggered by a tense, are similar or different from what happens with forms
of reported speech, for example, irony, if it is admitted that irony is a case of
‘mention’ (Sperber and Wilson 1995), that is, in their terms, an interpretive use of
language; that is, a representation by the speaker at S of another representation,
attributed to another instance, of some fact, thus a metarepresentation. What is
actually at issue is whether the speaker of utterances of type (5) to (8) above is
making a metarepresentation or not. I suggest the latter and will return to this below.
The theoretical questions raised by the narrative IMP are about the nature of the
process that leads to such pragmatic modulations of its meaning. In particular, while
it is clear that a deictic shifting of the temporal reference with, for example, so-called
‘historical’ futures and presents, is somehow ‘just’ pragmatic—it is in no way
inscribed in the semantics of the tense—what happens with the narrative IMP is
unclear. It is in particular unclear whether abandoning the R-point and replacing it
by another ‘perceptive’ deictically shifted point is a mere pragmatic-conceptual
operation, that is, the result of general cognitive pragmatic principles of interpret-
ation, or whether, rather, IMP encodes a less determined semantic description
like ‘some point inside some eventuality’ together with a number of parameters
determining how to saturate that point depending on collocations and contextual
information. If so, IMP does not simply encode a combination of coordinates, but
also imposes a kind of procedure about what to do with these coordinates according
to contextual information, leading to a relevant interpretation. In many cases, the
anaphoric R-point is available and thus used, but under specific contextual con-
straints (completion of the eventuality and/or progression), a S’ switched deictic
point is constructed and used instead.
IMP is not the only tense triggering such perspectival interpretations. Of particu-
lar interest are future uses of the French present perfect (passé composé) and
epistemic futures (in various languages).

3.3 Future reference with the present perfect in French (passé composé)
The French passé composé (PC) is morphologically equivalent to the English present
perfect (PP), but presents a number of known differences with it. In particular, only
the French PC is compatible with repetitive past experience (17), narration (18), and
past adverbials (19) (that the English PP is normally unhappy with adverbials is an
issue known as ‘the perfect puzzle’ since Klein 1992; see also Pancheva and von
Stechow 2004):

meaning on the left side of the plane having a front and a back, that is possibly on the right of the plane
from the speaker’s point of view, is not a decisive argument since one can represent the viewpoint of
virtually anything as if from an individual; furthermore, representing an eventuality from some point of
perspective seems to imply the representation of an individual allocentric conceptualization of that
Perspectival interpretations of tenses 55

(17) J’ai aimé aller au cinéma / J’ai beaucoup aimé ça.

?I’ve enjoyed going to the pictures. / ?I have enjoyed this a lot. (meaning
‘I used to enjoy’)
‘I used to enjoy going to the pictures’ / ‘I used to enjoy this a lot’5
(18) Le concierge est descendu, il a fermé la porte et est parti.
?The attendant has come down (vs. came down), ?has closed (vs. closed) the
door and ?has gone (vs. went away).
‘The attendant came down, closed the door, and went away’.
(19) Marc est arrivé à cinq heures / hier.
*Mark has arrived (vs. arrived) at five / yesterday.
‘Mark arrived at five / yesterday.’
Interestingly, futurate uses of the PC are possible in French, Italian, and in other
varieties of Romance, but impossible with the English PP (maybe simply because
futurate PC (FPC) requires a future adverb while the PP usually rejects adverbs).
Typical examples of FPC are of the following kind:
(20) J’ai bientôt fini / J’ai fini dans 10 minutes.
I finish-PC soon / I finish-PC in 10 minutes.
*I have soon finished / *I have finished in ten minutes.
Tra un attimo ho finito. (Italian)
‘I am done soon, I will be done soon / in ten minutes.’
(21) Dans un an, j’ai fini ma thèse.
*Within a year from now / *In one year, I have finished my thesis.
‘In a year, I will be done with my thesis.’
In a similar vein as with the narrative imparfait, Sthioul (1998a) suggests that FPC
triggers an allocentric representation in the future from which the eventuality is
conceived as past. As a consequence, in such cases, the semantic meaning of the
tense is not modified at all (contrary to what happens with the imparfait when R has
to be exchanged with another point) but wholly deictically transported in the future,
wholly shifted to the right, so to speak.
The French PC has two main realizations: (i) past, as with (17) and (18); and (ii)
perfect resultative. Resultative PC utterances convey the information that a resulting
state of a past eventuality relevantly holds at S, similarly to the English PP, as with
(22), which communicates not only that the keys have been found at some time in
the past, but that the change of state induced by this event still holds in the present
(the speaker still has the keys):

J’ai beaucoup aimé ça can mean either ‘I enjoyed this very much’ or ‘I used to enjoy this very much’.
56 Louis de Saussure

(22) J’ai trouvé mes clés.

I find-PC my keys
‘I found my keys.’
The kind of interpretation triggered by a FPC is of the resultative type (see de
Saussure 2011b for a development): what is communicated is that the resulting
states hold at the considered future point, whether the event itself has yet occurred
at S or not.
Resultative PC can be schematized as follows, where ‘s’ stands for the resulting
state, and ‘O’ for overlaps.
(23) E – R, S & E ! s O S
Formula (23) reads: E is anterior to R simultaneous to S and E implies a new
resulting state (perfect) s overlapping with S.
According to the assumptions above, FPC is nothing more than (23) where S is
simply exchanged with a perspectival point S’ whose position is indicated by a future
(24) S – S’; E – R, S’ & E ! s O S’
S’ is a future projection of S, from which E is represented as past. The position of S is
known only relatively to that of S’, so that nothing is known of the position of E
relative to that of S. This makes FPC practically equivalent to the anterior future
(AF), which also describes a resultative state of affairs true at some point in the
future, although without a S’ perspectival projection. Hence (21) appears similar to
(25) as far as temporal reference is concerned, the deictic switching being the only
difference between the two:
(21) Dans un an, j’ai fini ma thèse.
Within a year, I finish-PC my thesis.
(25) Dans un an, j’aurai fini ma thèse.
Within a year, I finish-AF my thesis.
Both are about a thesis already finished at a point in the future indicated by the
adverb phrase (‘in a year from now’). Thus, the question raised by FPC is about the
difference in meaning achieved by it with regard to AF, since that difference can’t
be about temporal denotation.
A possible line of investigation is that FPC is a perfect type of futurate present.
An obvious motivation for having recourse to futurate presents is to trigger practical
implicatures about the present situation of speech with regard to what is expected
to happen in the future. More specifically, I suggest that indeed the FPC triggers a
deontic-practical modality, that is, information about the conduct or attitude to
adopt in the present, in the light of what is predicted to happen. This meaning is
Perspectival interpretations of tenses 57

obtained through an allocentric representation, that is, a representation which is

attributed to some third-party (the speaker in the future, normally) viewpoint. The
meaning of such utterances, I suggest, is that the sentence will be utterable with
literal meaning at the point in the future indicated by the adverb; thus the eventu-
ality is presented as a reliable prediction as if it were actually uttered at a time when
it will be past, and hence verified as certain. All this is absent from what AF does.
In short, an utterance like (21) means something like (26):
(26) Let’s imagine we are one year ahead: I have finished my thesis.
If this is so, then the temporal meaning of the utterance is deictically shifted by the
temporal adverb, and the semantics of the tense is in no way modified. Under the
pressure of the adverb, it is a whole pragmatic process of deictic shifting that occurs.
Typical restrictions that apply to FPC confirm the assumption of a deontic-practical
modality. Dowty (1979) notes that futurate presents in English—but the same holds
roughly for French—imply that the eventualities enter into some scheme of planned
actions: (27) below is natural because it presupposes that a chain of actions is
planned that leads to the All Blacks playing against France tomorrow, whereas
(28), which is about an unpredictable event, does not enter into a futurate present
(27) Tomorrow the All Blacks play / are playing against France.
(28) *Tomorrow the All Blacks play well.
(adapted from Dowty 1979)
We suggest that the eventuality as planned makes it a good candidate for having
implications on what course of action or attitude to have in the present. What is
planned to happen is going to happen unless a (major) failure occurs, which is not
to be expected, so one has to prepare oneself, or to act, in accordance with the
assumption that the projected event will happen. Here, the present induces a strong
epistemic commitment to the future on the part of the speaker, and, as a conse-
quence, implies a deontic-practical modality about the present. For (27), this might
be about the need to buy tickets, or simply to rejoice in advance about the great
match we’re going to watch tomorrow (or whatever other relevant consequence in
the circumstances).
FPC might then simply be a perfect version of futurate presents (as Vet 1992
claims), therefore implying both ‘planning’, following Dowty, and, we venture to
suggest, a deontic-practical modality in the present on the basis of strong epistemic
Under the hypothesis of ‘planning’, non-agentive predicates are expected to not
enter this construction, hence the contrast between (29) and (30):
58 Louis de Saussure

(29) Le président est bientôt sorti de sa réunion.

The president soon get-PC out from the meeting.
‘The president will soon be done with the meeting.’
(30) ??/*Demain, il a plu.
Tomorrow, it rain-PC.
‘Tomorrow it will rain.’
But things are less clear than with futurate presents. Actually, with FPC, it appears
that (30) does become acceptable in the right context, as we show below; therefore
constraints on planning are weaker, or absent, with it. A number of other observa-
tions made in de Saussure (2011b) show that FPC, being a resultative (perfect),
prevents the future temporal adverb from modifying the temporal reference of the
eventuality E or of the resulting state, but specifies a non-ontological parameter,
which is the positioning of the switched deictic projection S’ (the adverb says that
there is a future moment at which the resulting state is true, regardless of its
leftwards extension). Also, FPC is incompatible with narration (unless a global
resulting state is available) and often appears with aspectual verbs, which are the
best triggers for resulting states. But other verbs are equally available:
(31) Dans une heure, j’ai débouché la baignoire [et tu peux prendre ton bain].
Within an hour, I unclog-PC the bathtub [and you can take a bath].
States (and to a certain extent, activities) raise further problems with the PC in
general as a perfective tense, which I will not address here.
The ‘planning’ criterion is not entirely sufficient if it is not understood as I suggest
(the inference of practical consequences in the present). Examples such as (32) and
(33) explicitly express planning, but are no less odd than is, apparently, (30):
(32) ?Dans dix jours, j’ai acheté des cigarettes.
In ten days, I buy-PC cigarettes.
(33) ?Demain nous nous sommes rendus sur place comme prévu.
Tomorrow we get-PC there as planned.
Let me now briefly elaborate on the hypothesis that futurate PCs are licensed
pragmatically by the possibility of drawing an inference with practical implications
in the present in the perspective of the actual happening of the eventuality. Such an
inference is in effect very salient in the typical contexts of occurrence of this form:
(34) J’ai bientôt fini / le président est bientôt sorti de sa réunion.
I soon finish-PC / the President be-PC soon out of the meeting.
! it’s worth waiting a bit.
Perspectival interpretations of tenses 59

(35) Dans une heure, j’ai terminé la réparation.

In one hour, I finish-PC fixing the car.
! it’s worth a wait in order to get the car back, worth having a coffee in the
nearby café rather than go home and come back to the garage later, etc.
In contrast, (36), with the anterior future, bears no invitation to wait for the fixing to
be done. Of course such an implicature can be made, but it is no way triggered by the
tense, even in this context: a cancellation of such an implicature about what to be
prepared for is easy with AF in (38), whereas it sounds unexpected to have it with the
FPC in (37):
(36) Dans une heure, j’aurai terminé la réparation (the mechanic to the customer).
In one hour, I finish-AF the repair.
‘In one hour, I will have finished the repair’.
(37) Dans une heure, j’ai terminé la réparation, ?mais je ne pourrai vous la rendre
que demain.
In one hour, I finish-AF the repair, but I can-FUT give it back to you only
‘In one hour, I will have finished the repair, but I’ll only be able to give it back
to you tomorrow.’
(38) Dans une heure, j’aurai terminé la réparation, mais je ne pourrai vous la
rendre que demain.
In one hour, I finish-AF the repair, but I can-FUT give it back to you only
‘In one hour, I will have finished the repair, but I’ll only be able to give it back
to you tomorrow.’
This happens because with an FPC the speaker presents a state of affairs as true in
the future, but also invites planning in the present, in view of this future state of
affairs, for all sorts of things that are not communicated by the AF proper. If it is true
that futurate PC communicates practical-deontic modality in the present (triggering
inferences on what attitude should be adopted), the tense as a grammatical marker
here determines specific pragmatic inferences in specific contexts with specific
communicative goals.
What, then, about the apparent oddness of (30)? If the hypothesis is correct, then
(30) should be natural if contextual information is retrieved by the hearer in order to
derive an implicature relevant in the present circumstances. Actually, if we consider
one farmer talking to another about the need to water their lettuces now, the
example looks natural, because of the implications of tomorrow’s rain on the
currently appropriate course of action:
60 Louis de Saussure

(39) Demain il a plu [et tes salades sont sauvées].

Tomorrow it rain-PC [and the lettuces are safe]
! no need to water the lettuces now.
FCP therefore indicates:
i) a deictically shifted interpretation, which is pragmatically constructed under
the pressure of a future time adverb positioning the projected point at which
the utterance is predicted to be utterable literally (allocentric/metarepresen-
tational/interpretive reading), a case which Sperber and Wilson (1995) might
want to treat as ‘interpretive use of language’ and assimilate to irony and free
indirect speech, or any other kind of reported thought or speech (all cases of
‘mention’), that is, a metarepresentation. But we suggest that this cognitive
effect is obtained through access to a third party’s viewpoint in the first place,
a process which is not metarepresentational.
ii) more importantly, FPC entails an epistemic modality of certainty (it is a
prediction presented as certain, as if already verified) and, as a pragmatic
consequence, a deontic-practical modality about how to behave in the actual
present in the perspective of the eventuality happening with certainty in the
FPC implies a ‘pure’ deictic switch: the whole situation described is transported so
that it anchors on an allocentric projection of S in the future; things happen
differently with IMP, and, as we see below, with epistemic futures. What these
differences imply for the notion of allocentric perspective itself is discussed in the
conclusion to this chapter.

3.4 Deontic-practical modality with epistemic futures

Epistemic futures (EF) are those future tense utterances which in fact serve to
communicate the likelihood of the considered state of affairs in the actual present.
Typical examples are:
(40) [doorbell rings] That’ll be the postman.
(41) Ce sera le facteur. (French)
(42) Sarà il postino. (Italian)

FPC is, however, compatible to some extent with expressions of epistemic modality. Very complex
issues of scope intervene in such constructions, such as Dans un an j’ai probablement terminé ma thèse ‘In
one year I finish-FPC probably my thesis’, which I can’t discuss here. See Mari (2012) on the topic of
epistemic modality with perfect tenses and in particular the passé composé.
Perspectival interpretations of tenses 61

Other forms expressing future time are also available, such as anterior future and
periphrastic future:
(43) That’s gonna be the postman.
(44) Ca va être le facteur. (French equivalent with periphrastic future)
(45) That’s gonna be that idiot Max again.
(46) Ca va encore être cet idiot de Max. (French equivalent with periphrastic future)
(47) [The person expected has not arrived] She will have missed her train.
(48) Elle aura raté son train. (French equivalent with anterior future)
French grammarians Damourette and Pichon (1911  1936) were, to my knowledge,
the first to suggest that EF represents a time where the verification of the state of
affairs as being true is performed. In more contemporary terms, this amounts to
saying that epistemic futures also trigger a perspectival effect: they do not denote the
state of affairs but they do represent the event of its acknowledgement in the future.
In our own terms, EF sets up a point of view in the future from which the situation is
grasped as being the case. If this is correct, then the tense in EF-utterances has just
the meaning of the future tense in general (there is no deictic shift or other
accommodation of the tense semantics) and the epistemic reading stems from
pragmatic features through a change in the nature of the eventuality (from P, say,
be the postman, the hearer gets to an event of ‘verify P’). EF triggers a change in the
concept, not in the tense itself. Our assumption is thus twofold:
a) EF represents verification, in the future, of an assumption about the present
state of affairs;
b) EF does this through perspective switching: the state of affairs is represented
as verified by a third party located in the future who can truthfully assert its
We shall call (a) the verification hypothesis and (b) the perspectival hypothesis. We
claim that (a) is actually a result of (b) and will provide arguments to this effect
That EF is best explained by (a) is not consensual: other assumptions are available,
notably if one considers the future as an epistemic modal marker rather than a tense
(for counter-arguments see de Saussure and Morency 2012), or if one suggests that
EF stems from complex compositional parameters (Caudal 2012). The advantage of
(a)—besides intuition—is one of simplicity, keeping the future tense a tense about
future time, dealing with simple coordinates on the timeline, so to speak; the
epistemic meaning with EF is then derived inferentially through general pragmatic
principles of information processing, which affords relevance to potentially incon-
gruent representations (a state of affairs known to be present together with a future
tense description of it).
62 Louis de Saussure

This view, however, raises a series of questions: What are the conditions that allow
or trigger epistemic interpretations of future tense utterances? How do we proceed in
order to move from the representation of a situation in the future to an inference
about the present situation? And most of all: What is the difference between the
meaning brought in by EF and that of explicit modal utterances, such as it is
probably the postman ringing?
We know that the switching of eventualities from ‘P’ to ‘verify P’ arises in contexts
where two conditions are met: (i) it is mutually manifest in the context that a state of
affairs compatible with P is occurring at speech time; and (ii) that the state of affairs
is represented as true in the future, not in the present. The inference resulting from
these two facts is that the speaker is unable to assert the truth of the state of affairs in
the present but nonetheless feels able to assert its truth in some (near) future. From
this then stems a further inference, that the meaning of the utterance is that of an
epistemic attitude that is stronger than mere possibility (usually considered a high
probability). The path of inference explaining epistemic futures is rather simple and
easy to spell out within any pragmatic theory of implicatures that can deal with
representational conflicts and general principles.7 But simple as it is, it is still far
more complicated than the insertion of a simple epistemic adverb in a present tense
clause, which communicates explicitly and quite directly the result of a far more
lengthy path of inference with EF:
(49) It is probably the postman.
Hence the question arises of the difference in meaning obtained with EF as com-
pared with that of (49). It is occasionally suggested in the literature that EF expresses
not only an epistemic meaning but also inferential evidentiality,8 comparable to
epistemic must (although with some particularities); in other words, EF would
express that the communicated modal information, for example it’s most probably
the postman in (49), is presented as being the result of reasoning, not a mere
stipulation. Such reasoning would in turn be based on several premises, among
which are (i) the grasping of a particular situation (for example the doorbell’s
ringing), and (ii) background assumptions (for example if the doorbell rings at this
time, it’s usually the postman). In this view, EF typically explains some event (like
the doorbell’s ringing). But the evidential status of epistemic futures is very much
debatable: in some cases, such as (50) or (51), no inference is available, no situation

In a Relevance-Theoretic framework, the search for relevance would start from the irrelevance of the
utterance in qualifying a present state of affairs and move towards finding a motivation for the future
representation through the assumption that if the speaker talks about the future when he’s expected to talk
about the present, then he doesn’t know it for sure.
See Dendale and van Bogaert (2012) and de Saussure (2012) for a closer view of EF as possibly
conveying an evidential flavour as a consequence of pragmatic processing.
Perspectival interpretations of tenses 63

has to be explained. The epistemic meaning seems here to stem only from unreliable
(50) Call Max: he will know a plumber.
(51) Elles seront sur ce présentoir là-bas (the seller answering about a brand of
‘They will be on that stand over there.’
All these things make inferential evidentiality an optional parameter for EF (there-
fore EF cannot be considered an evidential expression in the sense of Aikhenvald
2004). As far as reference to a mental state is taken as a ground for epistemic
meaning, it is unclear whether EF is different from straight explicit epistemic
formulations as in (49).
In fact, what differs with EF with regard to epistemic adverbs has to do with
practical things. For example, the most natural move following the utterance of (51)
is to go and check, with the assumption that the shoes will be found where indicated.
The seller who would in contrast utter (52) doesn’t seem to engage himself or the
customer very much with a course of action leading to a potential purchase:
(52) They are probably on that stand over there.
He indicates the raw information without any idea of further verification, and, we
would add, of the need to act accordingly (and he is thus a bad salesman: what is
retained is that he doesn’t know the answer, whereas in (51), the speaker is thought of
as being confident that the customer will find the shoes where indicated).
This leads us back to the allocentric projection into the time of verification, in
turn triggering other very pragmatic meanings about the need to take some particu-
lar course of action, something which is unavailable with (49) or (52). Hence, we
suggest that EF implies two components of meaning which can be spelled out as
First, as Sthioul (1998b) claims, future verification is presented as being realized by
an allocentric third party represented in the future, since, we should add, verification
in itself implies a notion of direct perception or mental operations which have to be
represented as held by some subject: verification is not an event that can occur in the
world without a subjective individual being its locus. This component is activated
with EF because the utterance seems rather to convey a prediction about the
propositional content being asserted literally with the present tense in the future,
as in (53) (rather than mere epistemic probability, as in (49)):
(53) In some (near)9 future, it will be possible to utter truthfully ‘It is the postman’.

That is, a future where the current state of affairs still holds.
64 Louis de Saussure

It should be noted that (53) represents epistemic certainty in the future. Logically, if
it is shared knowledge that the situation at S is uncertain, then it is impossible to
assert its very certainty from a future standpoint, since, as a principle, an epistemic
value at S can only decrease at S þ n.10 But here, communication is about predicting
the truth in the future, not predicting what the future may look like starting from an
assumption about a state of affairs at S.
Second, and because of this epistemic load of ‘certainty in the future’, we claim
that EF imposes a representation of a desirable course of action, or attitudes, to be
held by the interlocutors in the very deictic present, an effect that is absent from
explicit modal utterances. This is what we called (de Saussure and Morency 2012) a
deontic-practical modality with EF utterances, just as with FPC, to which we return
EF is not always possible, first because of semantic restrictions based on aspect
(other restrictions to specific verbs and to personal pronouns are presented by Celle
2004 but rejected in de Saussure and Morency 2012; see also Bellahsène 2007 on such
restrictions). These are explained by the (relative) permanence of situations as a
necessary condition for future verification. Hence, in particular, incompatibility with
achievements.11 There are also a number of language-specific features that make EF
more or less available in various situations across languages, in particular when, as
with Italian, EF has become the most standardized form for expressing probability.
In such a case, because of grammaticalization, it is likely that EF triggers less
complex inferences than in languages where EF appears as a form to be justified
in context with regard to the more expected adverbial modal expressions (for EF in
Italian, see Rocci 2000). In French or English, where EF is not the standard way of
expressing probability, a number of truly pragmatic restrictions seem to appear,
without any direct relation to aspect or formal parameters in general.
A first set of pragmatic oddities appears, consistent with the ‘verification’ hypo-
thesis, with utterances which represent situations that cannot be verified in any
accessible future:
(54) ??The Universe will be spherical. [epistemic reading]
(55) ??God will be lenient with him. [epistemic reading]
At the same time, the fact that EF can’t occur here challenges our hypothesis that
verification is represented as assumed by some perspectival point of view, since as a

Caudal (2012) refers to this argument against the notion of verification with EF.
But it becomes possible with achievements in the passive form and with cleft constructions, at least
in French; in de Saussure and Morency (2012), we argue that these cases induce a perfect reading and are
therefore about a resulting state of affairs to be verified, not the eventuality expressed. States, activities, and
even accomplishments are compatible with EF, although accomplishments seem to require a progressive
form in English (I thank here Chris Hart and Paul Chilton for helping me with their native intuitions and
Patrick Caudal for useful discussions), whereas in French non-stative verbs do not allow for an epistemic
reading with an anterior future.
Perspectival interpretations of tenses 65

principle, a perspectival point of view need not be conceived of as actual, so that a

projection in a fictitious future should be possible. In fact such verifications are
perfectly possible in other contexts:
(56) He will have died thinking of his wife.
(56’) Il sera mort en pensant à sa femme. (French equivalent of (56))
Future perspectival verifiability does not imply the factual, concrete, actual verifia-
bility of the events in play. In (56), the speaker projects imaginatively a future
consciousness which knows the thoughts of the deceased; and this is perfectly
acceptable since human representation is not limited by ontological reality. In (56)
the speaker is not merely making a simple conjecture but is adding a particular
component of meaning. She communicates: ‘something similar to a future eventu-
ality (of verification) is expected to prove that my assumption regarding the present
is right.’ Whether this future eventuality is factually possible or not is a different
question entirely. Thus examples like (54) and (55) call for another explanation.
The supplementary hypothesis we wish to put forward is that which we have
already developed regarding the futurate composed past in French: in both cases, we
argue, representing a future truth bearing on the present is relevant only in cases
where the represented eventuality has a causal impact on a fact currently relevant to
the interlocutors, or an impact on the actions to be presently undertaken in relation
to the situation represented. In sum: both FCP and EF trigger a deontic-practical
modality that concerns S.
The oddness in (54) and (55) yields here to another explanation. If the spherical
nature of the universe were relevant for the interlocutors at speech time, then the
epistemic reading would seem quite natural; it is the difficulty of imagining such a
context that restricts this possibility and rules it out. (55) is problematic for similar
reasons; yet one needs only to imagine a play, staged in the afterlife, where certain
beings are surprised not to witness the arrival, in Hell, of a certain individual, a
renowned sinner, thus making God’s indulgence relevant, and hence (55) appear
normal, calling for a change in attitudes (‘it’s not worth waiting for him’). The
oddness of (54) and (55) is only pragmatic: in (quite strange) contexts where the
nature of the universe or God’s lenience bears an impact on what should be done in
the present, (54) and (55) become acceptable.
In sum, besides the representation of a verification of an eventuality by an
allocentric projection, EF prompts for a particular deontic-practical modality.
There is still a consequence of the allocentric projection that remains to be spelled
out in order to get a full picture. So far, we have assumed that EF is about
eventualities and facts, which is of course correct. But the way EF speaks about
these eventualities and facts proceeds through an embedding of representations: the
speaker represents another representation of facts, i.e. that of an allocentric instance.
What is then actually represented by EF is not directly about eventualities and facts
66 Louis de Saussure

but about beliefs: it is a belief about a present state of affairs that is presented as
being verifiable in the future, combined with a consequence at the time of speech.
This makes the epistemic future resemble cases like the French IMP (imparfait) of
politeness ( J’avais une question ‘I had a question’, a case we did not discuss, but see
de Saussure and Sthioul 2005), or the counterfactual IMP (Une minute de plus et le
train déraillait, discussed above in (9)), where a present situation is envisioned in a
different temporality, a different possible world, in order to produce specific inter-
pretive effects. For the epistemic future, these effects would thus be linked not only
to the epistemic nature of a belief that is uttered, but also to a specific degree of
speaker endorsement or commitment, which distinguishes it from explicit modal
formulations like I believe that P or Perhaps P, or even It is likely that P. In turn, the
accuracy of the belief concerned has the deontic-practical consequences we men-
tioned. It is at this level, we think, that the difference is most clearly apparent
between His train will be going through a tunnel, which can typically be imagined
in a context where the utterance prompts one to wait for a better time to call again,
and I think his train is going through a tunnel, which gives raw modal information.
The latter utterance is descriptive of the belief itself, certainly allowing for a
particular attitude or action, but not prompting for it outright. It thus appears, if
we are correct, that what makes such an utterance relevant by comparison with more
explicit modalities is not an evidential inferential meaning, but rather a prospective
action or attitude to be adopted, which is to be undertaken in order to better adapt to
a forthcoming situation which is represented by the future tense utterance as being
verified in the future by an allocentric subject witnessing its occurrence.
That a future representation prompts for adopting an appropriate attitude or
behaviour fitting with the projected state of affairs is not surprising. It should suffice
to explain, in the end, the effect of EF by contrast with other ways of communicating
epistemic modality. What is more complicated is how pragmatic inference works
when achieving a present-time interpretation of a future tense, or any other use of a
tense which departs from its temporal semantics. If a time is semantically coded by a
tense, then modifying this should imply cancelling normally non-cancellable com-
ponents of meaning. Moreover, tenses being operators (Recanati 2006) or proced-
ural expressions (de Saussure 2003, 2010, 2011a), the operation(s) they encode are not
expected to be affected by contextual features. A perspectival account of the kind we
proposed should, however, escape this difficulty through the notion of some inde-
terminacy, not of the time attached to the tense (EF remains primarily about a
future), but reinterpreting the description of the state of affairs itself. Such a change,
from the representation of a state of affairs by the verb, to the representation of the
event of its verification, does not modify the structure of the tense: it is an operation
basically of the same nature as a conceptual modification, realized for the sake of
meaningfulness or relevance. This might not be a complex theoretical issue, as the
contextual information forbids that the concerned state of affairs be conceived of as
Perspectival interpretations of tenses 67

starting in the future, so that it has to be true now. That it is represented as true in
the future only, though, triggers the inference that it is not certain now, hence the
epistemic effect. But figuring out a future verification by means of an allocentric
projection allows to conclude a more finely grained modality than uncertainty, since
it is represented as verified. Hence the conclusion that it is (highly) probable.

3.5 Conclusion: allocentrism as an indicator of simulation in language

We have identified in this chapter three cases which might be successfully explained
through a notion of perspective shift: non-descriptive utterances with the French
imperfective past (unavailable in English), French present perfect (‘composed past’)
utterances with future reference (unavailable in English) and future tense(s) in
utterances with a meaning of epistemic modality about the present (‘epistemic future’,
available in both English and French and, most probably, in many languages).12
Switched imperfective past utterances in French, we claim, instantiate an allo-
centric perspective point of perspective, thus an allocentric deixis, instead of the
usual Reichenbach’s reference point R; the eventuality is then represented as ongoing
from the perspective of a third party distinct from the speaker at S. Composed past
utterances in French with future reference are interpreted as a shifting of the entire
combination of coordinates imposed by the tense so that it doesn’t anchor on S but
on a substitutive allocentric S’ situated at a future time indicated by an adverb
(mandatory); as a consequence, the eventuality is represented as past from the
perspective of a third party at S’ itself in the future. Epistemic future causes a change
in how the eventuality taking place at E is conceived: it is interpreted as verification
of the eventuality or, as we suggested, grasping of the eventuality as being true; this
evaluation is understood, we claim, as performed from a subjective perspective, that
is, again, a third party distinct from the speaker at S.
Switched imperfectives are generated when the achievement of the eventuality has
to be inferred for reasons of consistency or relevance. Such an inference is not
expected to cancel the basic imperfective meaning of the imparfait in French;
therefore R should not be part of the semantics of this tense, but, instead, another
variable, which saturates either with R or with an allocentric point of perception S’,
depending on whether an inference of achievement is performed. Switched com-
posed past (French ‘present perfect’) is another matter completely, since the seman-
tics of the tense is not itself affected by the whole deictic shifting which occurs under
the pressure of the temporal future adverb. Epistemic futures are ultimately not a
matter of a particular tense (it occurs with any future tense) but of a conceptual
modification (from the eventuality to its verification or, better, its grasping). If all of

Since the epistemic interpretation of future tense utterances ultimately relies not on linguistic, but on
conceptual processes, it is likely that it is not bound to specific languages.
68 Louis de Saussure

this is correct, then it is not really that a pragmatic inference interferes with the
semantics of an operator (or ‘procedural expression’), which would be problematic,
but rather that these interpretations are made possible either by variables in the
semantic structure of the tense (as with the French imparfait), or by choosing
another point of origin in the calculus (with doesn’t affect the semantics, as with
futurate composed past), or by reinterpreting a conceptual meaning (as with epi-
stemic futures).
This chapter is not the place to discuss the complex matter of the ways for
cognition to access non-egocentric perspectives, but let me recall that if time always
has to do with the deictic consciousness, or origo (Bühler 1934), then a few conse-
quences are worth noting. If origo is something mental which manages the link
between (i) the perception of the self or proprioception (having to do with con-
sciousness of one’s bodily situation) and (ii) the perception of the external world,
then temporal deixis has to do with the management of this at the level of time. In
simple words, the temporal origo links the knowledge about one being at some
moment of time with other things—events—being at some temporal distance from it
(de Saussure 2008). As a consequence, an allocentric point of view that implies an
inner apprehension of an eventuality, such as that associated with the use of non-
descriptive imperfectives in French, is nothing more than the grasping of a second-
ary origo point associated with an imaginary subjectivity perceiving the eventuality
as ongoing. That this secondary origo gets established as a substitute of the R-point
with past imperfectives in French does not cancel the deictic basis for the calculus of
tense but, so to speak, establishes a secondary now-point associated with a third
party’s presenteness (which is thus represented as distinct from S). The deictic shift
that occurs with futurate uses of the composed past, i.e. the instantiation of the
future point of view for which P is true, does not eliminate the original S-point from
the consciousness of the interlocutors—needless to say—so that consequences in the
very present are ultimately derived. These deontic-practical effects, as we called
them, are ultimately what communication is about: very pragmatic issues about
the practical world and thus appropriate behaviour that is expected to fit with the
world as it is or as it is expected to become. A similar line of explanation is suggested
for epistemic future utterances.
Such a change in origo, brought about by the needs of understanding natural
language, seems better explained as a case of simulation than as a case of applying
tacit theoretical laws of folk psychology (although I will refrain from drawing strong
conclusions here). Language is often seen as an elaborated metarepresentational
device, and clearly the arguments in favour of this view are massive and span various
levels of language structure. According to this widely shared view, linguistic struc-
tures represent other representations (assumptions) which are themselves bound
to speakers and are either about facts or, even, about other representations (as
with reported speech or irony). In this latter case, there are in fact three levels of
Perspectival interpretations of tenses 69

metarepresentational embedding: the utterance is a representation of an assumption,

which is a representation of another assumption (another representation) about a
fact. One can even venture that in John says that Paul wishes that Mary thinks that P,
two more levels are added to the metarepresentational structure. Metarepresenta-
tions are rather easy to deal with within a folk-psychology Theory of Mind, and thus
with a theory–theory kind of model, since the building blocks of metarepresenta-
tions can be assumptions. They are in contrast not easy to deal with through the
weaker technical apparatus of simulation theory, which undergoes critical theoretical
issues, such as lack of theorization and a reliance on stipulative introspection.
Yet perspective shift with tenses seems to activate something far more direct
and low-level in cognition than the deployment of a tacit theory called for by a
theory–theory account: apprehending a situation as if perceived by other eyes in no
way implies the grasping of anything like the assumptions available to that third
party. In turn, but in turn only, such a simulative operation allows for metarepre-
senting; hence the feeling that futurate composed past utterances are about a time
where the composed past sentence will be utterable, that epistemic future utterances
are about a time where the sentence will be utterable with a present tense: all effects
that seem quite metarepresentational.
If we are right, which only further investigations might show, then language
exhibits both types of mindreading process and thus a reconciliation is called for
between simulation theory and theory–theory, or some adaptation to cope with both
types of empirical facts in the realm of the Theory of Mind.
This page intentionally left blank
Part II

Time and Modality

This page intentionally left blank

Modal auxiliaries and tense:

the case of Dutch


4.1 Introduction
Time and (different types of) modality are known to interact in many complex and
often still poorly understood ways. This chapter explores one dimension of this
interaction: that is, what happens semantically when expressions of these categories
meet in a clause. Specifically, we will focus on the semantic effects of the interaction
between tense and the modal auxiliaries in a clause—i.e. the grammatical devices
for expressing these semantic categories. And we will use Dutch as our object of
study, a language in which—unlike in English—the ‘flectional’ possibilities of the
modals (with fully productive preterite and non-finite forms) and the flexibility in
the organization of the verbal group allow an interesting range of combinations of
these forms.
We will approach the matter from a cognitive–functional perspective (Nuyts
2001). We hope, first of all, to demonstrate the relevance of the notion of a
‘qualificational hierarchy’ as assumed in this framework. And second, at a more
basic theoretical level, we hope to illustrate the fact that the form–meaning relation-
ship in language—in this case, the way the semantic qualificational hierarchy is
rendered at the linguistic surface—is complex and determined by several functional
dimensions, and hence requires a dynamic process type model of the kind assumed
in the cognitive–functional perspective (pace tendencies against such an approach in
some branches of cognitive linguistics).
This chapter is organized as follows. In Section 4.2 we introduce some basic concepts
and notions: the cognitive–functional perspective, a definition of the semantic and

This research was supported by an IAP VI project (P6/44) funded by the Department for Science
Policy of the Belgian Federal Government, and by a research project (G.0443.07) funded by FWO-
74 Pieter Byloo and Jan Nuyts

formal categories figuring in the study, and the notion of a ‘qualificational hierarchy’.
Section 4.3 presents the methodology used in the investigation, and Section 4.4 offers a
detailed analysis of our data. Section 4.5 summarizes the findings and discusses them in
the light of the notion of the qualificational hierarchy, and Section 4.6 draws some
conclusions, turning to the broader issue of the meaning–form mapping.

4.2 Basic concepts and notions

4.2.1 The cognitive–functional approach
The central tenet of the cognitive–functional perspective (as discussed in Nuyts 1992,
2001, forthcoming b) is that language is essentially a cognitive system for communi-
cation. The assumptions as such that language is a means for communication, and
that it is a cognitive system, can hardly be called new or controversial: they are at the
core of large parts of language research, including, notably, in traditional functional
linguistics and in cognitive linguistics. Yet each of these strands arguably (see Nuyts
2005b, 2007, 2009b) concentrates primarily on just one of these dimensions (func-
tional linguistics on the communicative dimension and cognitive linguistics on the
cognitive one), and although neither of them will deny the importance of the other
dimension, they do not actively focus on it. The consequence is that they both run
the risk of developing views or concepts which are potentially problematic from the
perspective of the disregarded dimension. The point of the cognitive–functional
perspective, then, is that in order to be fully adequate, a theoretical model of
language has to account for the cognitive and the communicative aspects of language
on an equal basis and in a fully integrated way.
Specifically, this combination of perspectives suggests the following two general
principles as being basic to any concern with grammar—although many approaches
in current cognitive and functional linguistics are not obviously in line with them.
Both principles pertain to the relations between form and meaning in language.
. The principle of ‘depth’. Since language is a system for communication and since
communication is, at least to a considerable extent, a matter of transferring
information about the world, the linguistic systems in cognition do not operate
in isolation, but are unavoidably closely interrelated with the cognitive systems
responsible for storing our knowledge of the world, and for our reasoning with
that knowledge, i.e. what are commonly called the ‘conceptual systems’. There-
fore, a cognitive model of language use cannot afford to concentrate on linguistic
form (and linguistic meaning) only but is forced to also actively deal with these
conceptual systems and how they relate to the linguistic ones.
. The principle of ‘dynamism’. Communication is a complex problem-solving
activity involving several different and often conflicting concerns. And each
communicative situation is different in these terms. Consequently, the linguistic
Modal auxiliaries and tense: the case of Dutch 75

system must be a dynamic, i.e. a context-sensitive, flexible, and adaptive usage

system. And using the system as such is not self-evident either: communicating
is often ‘hard labour’, not only in terms of interpreting the situation correctly
and deciding how to act adequately, but also in terms of getting the ‘shape’ of
the linguistic acts right. Hence coding conceptual meanings into linguistic
forms is something that speakers must work out dynamically, on every com-
municative occasion.
The principle of ‘depth’ is obviously central in cognitive linguistics (cf. the frequent
concern there with the nature of conceptualization), but hardly so in most trad-
itional functional linguistic approaches. And the principle of ‘dynamism’ is not
obviously satisfied in either strand.
Thus, one consequence of both principles—one which figures as a major back-
ground element in the present chapter—is that in any concern with linguistic
phenomena one should clearly distinguish between dimensions of meaning and
dimensions of (linguistic) form, and one should not assume a simple relationship
between them. There are numerous examples in current functional and cognitive
linguistics in which this idea is not obviously met, resulting in potentially debatable
analyses and proposals (see Nuyts forthcoming b for elaborate discussion). One case
in point is the strong tendency in cognitive linguistics to argue against ‘process’
models of grammar—a model type with separate levels of semantic/conceptual and
syntactic representation which are linked by ‘intelligent’ mapping procedures—and
to opt for construction type grammars, which centrally turn around ‘constructions’
as fixed meaning–form units (or ‘symbolic units’) and on the unification of such
constructions as the only type of ‘operation’ in the system. For instance, Langacker
(1987) and Croft (2001) have argued explicitly against the process type of model (and
both obviously adopt a construction type of grammar, even if their actual models are
quite different). Yet process models are obviously in line with the above principles,
while construction grammars are far less so. The question of how process models
and construction-based models differ requires a much more complicated and subtle
answer than is often suggested (many presumed differences are actually purely
notational; see Nuyts 2008b, 2011). But in any case, the concept of a construction
as a fixed pair of a linguistic form and a (conceptual) meaning is potentially
problematic, not only as a theoretical concept, but also in view of the empirical
facts of language. Our present issue, the interaction of time and modality, may offer
another illustration of this point.

4.2.2 Time and modality: the semantic dimension

Let us start with the semantic dimension, and offer definitions of the notions that
will figure in the analyses below. Time and modality are traditionally cornerstones of
76 Pieter Byloo and Jan Nuyts

the system of ‘tense–aspect–modality’ categories or ‘qualifications of states of affairs’.

While the question of how to analyse the semantic concept of ‘time’ is far from
resolved, there is probably fairly little dispute over its basic definition as involving
the situation of a state of affairs (SoA) on/relative to the time axis. There is, however,
much less agreement on how to define ‘modality’ (see Nuyts 2006, forthcoming a,
for surveys of alternative views). In this chapter we follow the analysis in Nuyts
(2005a, forthcoming b; Nuyts, Byloo, and Diepeveen 2010). Space prevents elabor-
ation, but, in a nutshell, we do not consider modality to be one coherent semantic
concept. Rather, we assume a series of separate categories, which we still call ‘modal’
for the sake of recognizability, but which differ in a few fundamental respects, and
each of which figures separately at the same level of analysis as the concept of time.
For our present purpose, the following categories are relevant:
. Dynamic modality, concerning abilities/potentials or needs/necessities. There
are three subtypes: (i) participant-inherent (‘dyn-inh’)—marking the inherent
ability or need of the agent participant to realize the SoA (cf. John can swim;
I must go to the toilet); (ii) participant-imposed (‘dyn-imp’)—marking a possi-
bility or necessity for the agent participant ‘imposed’ by the circumstances (cf.
John can enter, the door is unlocked; you must pass through the kitchen to get
into the garden); (iii) situational (‘dyn-sit’)—marking a potential or inevitability
inherent in the SoA as a whole, rather than in any participant (cf. in an
earthquake even this construction material can get damaged; the driver was so
drunk that this car accident had to happen).
. Deontic modality (‘deo’), unlike in the literature, defined here as indicating the
degree of moral acceptability of the SoA (cf. we cannot fire him just like that, he’s
been our best employee for years—meaning ‘it would be immoral to do so’). (See
Nuyts et al. 2010 for discussion; the traditional ‘deontic’ notions of permission
and obligation are called ‘directive’ here—see below.)
. Epistemic modality (‘epi’), indicating the degree of likelihood of the SoA
(cf. John might be home).
There is a substantial difference between dynamic modality on the one hand, and
deontic and epistemic modality on the other, in the sense that the latter two are
‘attitudinal’ in nature, i.e. they concern, in different ways, the extent to which the
speaker can commit himself to the SoA, while this does not apply to dynamic modality.2
Dynamic modality actually appears closely related to ‘quantitative aspect’, indicating
the frequency or ‘incidence’ of the SoA (semelfactive, iterative, habitual, etc.).

The claim that categories such as deontic or epistemic modality are a matter of ‘degrees of
commitment’, hence are attitudinal in nature, is certainly not uncontroversial, but going into this issue
is beyond the present scope; see Nuyts (2005a, forthcoming b) for discussion.
Modal auxiliaries and tense: the case of Dutch 77

On the other hand, a category usually not labelled ‘modal’, but which can also be
called ‘attitudinal’ and which is also expressed by one of the modals, is:
. Inferential evidentiality (‘evi’), indicating that the SoA has been inferred indir-
ectly from other, directly perceived or generally known, facts. This category is
expressed by must (and its Dutch cognate moeten), for example in someone is
knocking at the door, that must be John.3
Finally, a category also prominent in the modals, but, as indicated—in spite of its
traditional labelling as ‘deontic modal’—of a kind very different from all the above, is:
. Directive (‘dir’), marking a permission, obligation, interdiction, advice, etc.
(cf. you must go now).
This is not an attitudinal or even ‘qualificational’ category, but an illocutionary one
(the same as expressed by the imperative mood; see Nuyts et al. 2010).
Although this will not be our central concern in this chapter, our analyses will
indirectly offer further support for this view of the semantic fields involved.

4.2.3 Tense and the modals: the linguistic dimension

We will be looking at the interactions of the tenses and the modals, as the gram-
matical devices for expressing time and the modal (and related) categories. But tense
and the modals are structural categories, not semantic ones, and although they are
prototypically correlated with the semantic dimensions mentioned, their status in
these terms is not always obvious, and they are occasionally also clearly used in other
This is true to a relatively limited extent for the modals. In our analysis in Section
4.2.2, their directive use is an illustration. A few other non-modal uses will be
mentioned in Section 4.3.
But it is true to a much larger extent for the tenses (see also de Saussure, this
volume). Thus, one obvious issue is the equivocal status of the perfect. In principle, it
expresses ‘completedness’, and this is not a temporal category but a type of phasal
aspect (marking the stage of development of a SoA). It is known, however, that in
many languages, including English and Dutch, the perfect also often functions as a
time marker. This double status of the perfect will be very manifest in the analyses
For some forms one can even dispute whether they are tense forms or modals:
English will and Dutch zullen often invoke an epistemic meaning, next to or even

The traditional notion of evidentiality, as an indication of the source of information about the SoA, is
arguably not a coherent semantic category. While inferentiality may be considered attitudinal, this is
probably not true of, for instance, hearsay or speech reporting; see Nuyts (forthcoming b) for discussion.
This issue is immaterial for our present purposes.
78 Pieter Byloo and Jan Nuyts

instead of a future time reading, and which is most frequent is highly unclear (they
are inextricably ambiguous between the two readings in many of their uses). Hence
it is no surprise that these forms are sometimes considered modals. For the present
purpose, however, we will handle them as tense forms.
Yet other non-temporal uses of tense forms will show up in the course of our
Such observations about the ‘multiple’ mappings between semantic categories and
formal categories are obviously relevant in view of our theoretical point, made in
Section 4.2.1. But also in practical terms, in our analyses below we will have to pay
close attention to the specific meanings involved in individual uses of both the tense
forms and the modals in order to draw conclusions about the semantic effects of
their interaction.

4.2.4 The hierarchy of qualifications of SoAs

Central for our analyses is the concept of the ‘hierarchy of qualificational categories’
as proposed (in different versions) in the functionalist literature, and specifically the
cognitive–functional version of it, rendered in (1) (here confined to the semantic
categories mentioned above; see Nuyts 2001 for discussion).4 ‘Directive’ (see Section
4.2.2) is not in this hierarchy because, as mentioned, it is not a qualificational but an
illocutionary notion, hence has a separate status (see Nuyts 2008a).5
(1) > (inferential) evidentiality
> epistemic modality
> deontic modality
> time
> quantitative (frequency) aspect / dynamic modality
> phasal aspect
> SoA
This hierarchy is meant to account for the relative extension of semantic scope of the
categories involved: higher in the hierarchy means wider scope; categories have

At face value, this kind of hierarchy may seem to resemble Cinque’s (1999) functional hierarchy,
proposed within the generative tradition. The theoretical context of the latter is so radically different from
the present, however, that a simple comparison is impossible—so we will refrain from making it. Likewise,
the interactions between time and types of modality (most notably epistemic modality) in particular,
including their scope relations, as discussed below, have also been the subject of debate in formal
semantics (e.g. Condoravdi 2002; von Fintel and Gillies 2008; see also Kaufmann, Condoravdi, and
Harizanov 2006), but the present conceptual framework is too different to allow a simple comparison,
and space prevents an elaborate discussion here.
Note the quite different position of dynamic modality as compared to deontic and epistemic modality
(cf. Section 4.2.2). Also note that the three dynamic subtypes are not mentioned separately. Dyn-sit might
seem to be more of a sentence-level category than dyn-inh and dyn-imp, and hence to have wider scope
than the latter two. Yet the three subtypes are all equally affected by a time marker (see also Section 4.4).
So for now we do assume that they occupy the same position in (1).
Modal auxiliaries and tense: the case of Dutch 79

scope over those below them, but not over those above them. For example, in John
will probably come home tomorrow, epistemic probably does affect temporal tomor-
row: the probability judgement concerns ‘John’s coming home tomorrow’. But the
epistemic assessment is not situated tomorrow; it applies at speech time. And so it is
probably the case that John will come home tomorrow is fine, but *it is tomorrow the
case that John will probably come home is nonsense. Hence epistemic modality is
higher in the hierarchy than time.
The hierarchy in (1) is meant as a hypothesis about the conceptual position of
these categories, however, and there are factors which complicate the picture at
the linguistic surface. Important for this chapter is the dimension of performa-
tivity vs. descriptivity, which applies to the categories above time. Thus, in (1)
inferential evidentiality is above time—but surely, in, for example, yesterday it still
seemed we could reach the top, but ultimately we haven’t, inferential evidential
seem is in the scope of temporal yesterday, and not vice versa. The point is that the
evidential form here is not performative, that is, does not render the speaker’s
view at the moment of speech (as was the case with probably in the example
above). It is used descriptively, that is without involving speaker commitment—it
reports on a former qualification by the speaker which is no longer applicable at
speech time. Other types of descriptive use of qualificational forms involve
reporting on another speaker’s qualification of a SoA, or the mentioning of a
hypothetical qualification, for example in a question, or a conditional clause. The
‘conceptual’ status of performative and descriptive qualifications is quite different,
though: the hierarchy in (1) captures only the performative versions of the
qualifications above time; the descriptive ones are (to hugely simplify the situ-
ation) part of the SoA at the bottom of the hierarchy (see Nuyts 2001: 39ff., 2009a
for further explanation).6
Still, the qualificational scope hierarchy does have visible effects in linguistic
structure, for example in the linear order of expressions of these categories in a
clause, most clearly so in the range of affixal markers on the verb. In terms of our
own investigation, then, we should expect to see the hierarchy reflected somehow in
the ‘grammatical scope’ of the tense forms and the modals in a clause; that is, in their
grammatical dominancy relations or ordering in the verbal group. Let us see whether
and how this is the case.

4.3 Method and data

As indicated, we focus on Dutch as our object of study, because the language offers
an interesting range of possibilities in the interaction of the modals and the tenses.

The performativity vs. descriptivity distinction actually also applies to illocutionary categories, hence
also to our category of directivity; see Nuyts et al. (2010) and below.
80 Pieter Byloo and Jan Nuyts

Of course, we assume that, at least at the basic level of how the linguistic and the
semantic levels interact, the situation in Dutch is not idiosyncratic, but somehow
representative for the situation in language in general—but whether this assumption
is correct remains to be seen in the light of future comparable studies on other,
preferably typologically substantially different, languages.
Our investigation is supported by corpus data. The modals served as our basis for
selecting instances. We have limited ourselves to the three most central Dutch
modals: kunnen ‘can’, mogen ‘may’, and moeten ‘must’. Four hundred instances
per modal were selected randomly but, for the sake of representativeness, with an
equal share of spoken and written ones, and within each of these categories, of
Belgian and Netherlandic Dutch ones (we will not compare these subsets though).
The spoken data were drawn from the ‘Corpus Gesproken Nederlands’ (Nederlandse
Taalunie 2004), a representative nine million word sample of current spoken Dutch.
The written data were selected from the ConDiv Corpus (Grondelaers et al. 2000) of
newspaper language, complemented by expository texts of different kinds drawn
from carefully selected Internet sources.
For all instances we have determined the precise meaning of the modal (in its
discursive context). Each instance was analysed by between two and four researchers
familiar with the meaning categories. Cases of disagreement were resolved through
discussion. Cases of ambiguity (a pervasive phenomenon in the Dutch modals, even
more so than in the English ones) were not forced into one category, but received a
label for each of their possible meanings.7
As a background for our discussion in Section 4.4, Table 4.1 shows the results of
this analysis. Per modal, and for the three modals together, the ‘abs’ column
presents the absolute number of instances in each meaning category, and the ‘%’
column indicates the share of the number of instances in each meaning category in
the total number of instances (i.e. 400 per modal, 1,200 in total). It is important to
notice that ambiguous instances are included in each of the possible meaning
categories (i.e. Table 4.1 offers what we will henceforth call ‘maximal counts’). For
example, if an instance of moeten is ambiguous between ‘dyn-imp’ and ‘dir’, it is
included in the counts for both meaning categories. Consequently, the sum of all
frequencies in the ‘abs’ column per verb is always more than 400, and for the total
more than 1,200, and the sum of the shares of the different meanings is always
more than 100 per cent. For ‘dyn’, Table 4.1 shows the total and below it the
separate counts for the three subtypes (again in ‘maximal counts’). (Examples will
be given in Section 4.4.)

Multiple labelling of an instance was never the result of disagreement between the raters; it always
reflects a consensus that the alternative readings are possible.
Modal auxiliaries and tense: the case of Dutch 81

Table 4.1 Meanings of the modals

moeten mogen kunnen total

(n ¼ 400) (n ¼ 400) (n ¼ 400) (n ¼ 1200)
abs % abs % abs % abs %

dyn 249 62% 73 18% 389 97% 711 59%

dyn-inh 4 1% 0 0% 121 30% 125 10%
dyn-imp 189 47% 61 15% 237 59% 487 41%
dyn-sit 72 18% 17 4% 82 21% 171 14%
deo 85 21% 85 21% 54 14% 224 19%
epi/evi 12 3% 0 0% 14 4% 26 2%
dir 103 26% 247 62% 20 5% 370 31%
other 54 14% 53 13% 0 0% 107 9%

A few remarks are in order.

. Dynamic modality in mogen is marginal, in spite of the 18 per cent share. The
absence of dyn-inh is no accident; it does not exist (any more) in this modal.
And instances with the other dyn types are idiomatic, frozen relics of a formerly
productive meaning—(2) is an example.8
(2) daar mag u van overtuigd zijn
‘You can be sure about that/be assured’ [C]
Therefore, we will disregard dynamic uses of mogen in the discussion in Section
. Epi and evi are conflated in one row for reasons of simplicity (they also behave
comparably in our present study), but the label ‘epi’ only applies for kunnen,
‘evi’ only for moeten. Mogen does not have an epi meaning any more.
. The category ‘other’ covers a range of special uses. Some may still be ‘modality-
related’, others are clearly not. In many of these it is nearly impossible to assign
a precise meaning because of their purely idiomatic character. One frequent
example is the use of mogen in Netherlandic Dutch and of moeten in Belgian
Dutch to mark a conditional clause (equivalent to English ‘if ’):
(3) Mocht[/Moest] je dat op je website plaatsen, dan heb ik geen bezwaar
‘If you posted that on your website, I would not object’ [C]

For reasons of space we will not provide word-for-word glosses of Dutch examples, unless necessary.
The relevant modal and tense forms are in boldface. Corpus examples are marked by ‘[C]’ after the
English translation. Corpus examples, especially spoken ones, are often simplified by omitting irrelevant
parts, false starts, pause fillers, hearer interruptions, etc. Spoken examples maintain the transcription
conventions of the Corpus Gesproken Nederlands: for example, no capitals, no punctuation (‘?’ indicates
rising intonation, ‘.’ marks the border between transcription units).
82 Pieter Byloo and Jan Nuyts

This use no doubt originates in the dyn meaning, but it makes no sense to still
label it so in present-day Dutch. Given their non-modal and/or unclear status,
all ‘other’ cases are henceforth left out of consideration (and are also no longer
included in the frequency counts in the tables).
Let us now turn to the combination of these modals and the tenses. We will
henceforth not differentiate between the three modals, unless it is relevant.

4.4 Analysis
4.4.1 Modals and the ‘simple’ tenses
The Dutch modals still have productive present, preterite, and non-finite forms.
Table 4.2 shows their frequency for the three modals together—in general, and per
modal meaning separately—in the simple tenses, i.e. without a tense auxiliary.9 For
the sake of clarity, the final double column shows the frequency of the modals in
complex tenses, which will be handled in Section 4.4.2. Table 4.2 offers the absolute
frequencies of each tense form (‘abs’), as well as the share of each (‘%’) in the total
number of instances of the three modals, in general, and split out per meaning
category. The calculation of the share also takes into account the instances with
complex tense, but, for the general count, excludes the instances classified as ‘other’
(see Section 4.3). The frequencies for the separate meaning categories again involve
‘maximal counts’ (see Section 4.3).

Table 4.2 Modals and simple tenses

without tense auxiliary with tense
present preterite non-finite total auxiliary
abs % abs % abs % abs % abs %

general 844 75% 160 14% 24 2% 1028 91% 103 9%

dyn 524 74% 97 14% 20 3% 641 90% 70 10%

dyn-inh 83 66% 24 19% 11 9% 118 94% 7 6%
dyn-imp 377 77% 63 13% 9 2% 449 92% 38 8%
dyn-sit 126 74% 16 9% 1 1% 143 84% 28 16%
deo 179 80% 18 8% 3 1% 200 89% 24 11%
epi/evi 11 42% 0 0% 0 0% 11 42% 15 58%
dir 281 76% 67 18% 5 1% 353 95% 17 5%

Non-finite forms can occur in simple tenses, for example in non-finite subordinate clauses (cf. om te
kunnen winnen heb je een betere conditie nodig ‘to be able to win you need to be in better condition’), or
when preceded by a non-tense auxiliary, for example another modal (cf. dit moet je kunnen oplossen ‘it
must be possible for you to solve this’; or dat moet kunnen ‘that should be possible/acceptable’, with the
non-finite modal as the main verb).
Modal auxiliaries and tense: the case of Dutch 83

The frequencies of the alternative forms as such are difficult to assess, because of
lack of information on the frequency of simple tense forms in general, but there is no
reason to believe they would be abnormal. Also, in terms of the semantic effects of
the combinations, an analysis of the instances reveals nothing surprising. Notably,
the preterite—as compared to the present—straightforwardly situates the modal
meaning in the past. For instances with a deo or dir meaning, this systematically
means that the modal is used descriptively (see Section 4.2.2); performative uses are
restricted to the present form. For dyn and deo, this situation is in line with the
hierarchy in (1): meanings below time—thus also dyn—are sensitive to it, hence are
situated in the past by a preterite; meanings above time—including deo—are not
sensitive to it, hence a preterite changes their status from a qualification in the
hierarchy (‘performative’) to a part of the SoA (‘descriptive’). Illocutionary categories
are also insensitive to time (albeit in a different way), hence it is no surprise that dir
behaves like deo in this regard.
The only remarkable observation is that the category epi/evi only occurs in
present-tense modals, never in preterite or non-finite ones. In fact, also intuitively,
unlike present kan in (4), preterite kon in (5) cannot be read epistemically (see also
Nuyts 2001; the same applies for evidential moeten).
(4) Jan kan thuis zijn
‘John can/may be home’ (dyn and epi possible)
(5) Jan kon thuis zijn
‘It was possible that John was home’ (only dyn possible)
This is still consonant with the hierarchy in (1); all it means is that for some reason
the epi/evi reading of these modals cannot be descriptivized by using the preterite

4.4.2 Modals and the complex tenses Possible combinations If we turn to the complex tenses, the picture becomes
much more complicated.
Table 4.2 shows that combinations of the modals with complex tenses are
rare compared with combinations with simple tenses. Assessing this is again
difficult, for lack of data regarding the global incidence of simple vs. complex
tenses in Dutch. There is one exception to the pattern, though: epi/evi is even
more frequent in the complex than in the simple tenses. The reason will become
clear later.
Table 4.3 below shows in more detail the frequency in our data of combinations of
the different modal meanings with different complex tenses (the key is as for Table
4.2; see Section 4.4.1). The labels in the top row refer to the following pattern types,
presented here according to the ‘grammatical scope’ (or dominancy) relations
84 Pieter Byloo and Jan Nuyts

Table 4.3 Modals and complex tenses

A B C D E F total
abs % abs % abs % abs % abs % abs % abs

general 9 9% 1 1% 13 13% 15 15% 21 20% 44 43% 103

dyn 3 4% 0 0% 9 13% 11 16% 19 27% 28 40% 70

dyn-inh 0 0% 0 0% 0 0% 0 0% 1 14% 6 86% 7
dyn-imp 0 0% 0 0% 7 18% 9 24% 11 29% 11 29% 38
dyn-sit 3 11% 0 0% 2 7% 2 7% 7 25% 14 50% 28
deo 1 4% 1 4% 0 0% 7 29% 0 0% 15 63% 24
epi/evi 5 33% 0 0% 0 0% 0 0% 0 0% 10 67% 15
dir 2 12% 0 0% 5 29% 1 6% 5 29% 4 24% 17

(cf. Section 4.2.4) between the participating forms.10 The examples are constructed.
We provide glosses instead of translations in them since the latter vary depending on
the reading (see below).11
Pattern A: Present modal þ infinitive perfective aux þ past participle main verb
(a) We moeten/mogen/kunnen de bloemen geplukt hebben
We must/may/can the flowers picked have
Pattern B: Preterite modal þ infinitive perfective aux þ past participle main verb
(b) We moesten/mochten/konden de bloemen geplukt hebben
We must/may/can the flowers picked have
Pattern C: Present perfective aux þ infinitive modal þ infinitive main verb
(c) We hebben de bloemen moeten/mogen/kunnen plukken
We have the flowers must/may/can pick
We assume that the present/preterite marking on the finite verb has ‘grammatical scope’ over all
verbal forms (the finite and non-finite ones), and that an auxiliary has ‘grammatical scope’ over all verbal
forms (auxiliary or main) which ‘follow’ it. As will appear in the examples below, the actual ‘surface order’
of the forms in the utterance can sometimes differ from the grammatical dominance order as rendered in
(a)–(f), due to word-order flexibility, but space prevents elaboration of this here.
The rules for when to use an infinitive vs. a past participle are irrelevant here. Most main verbs take
hebben ‘have’ but some take zijn ‘be’ as a perfective auxiliary, but again this is irrelevant here. Combin-
ations of a present/preterite modal þ infinitive future auxiliary þ non-finite verb, i.e. the future
counterparts of Patterns A and B, are impossible in Dutch. A few other patterns are possible, viz.
present/preterite future auxiliary þ infinitive perfective aux þ infinitive modal þ infinitive main verb,
and present/preterite future auxiliary þ infinitive modal þ infinitive perfective aux þ past participle
main verb. These do not occur in our data, though, and for the sake of space and digestibility, we will not
consider them here. In Patterns C–F the modal itself can sometimes serve as the main verb (the ‘infinitive
main verb’ is then lacking, and in some patterns the modal appears as a past participle). We do not handle
these separately, though, since they are rare (three instances in our data) and generally do not affect the
analysis (but it will be mentioned when they do).
Modal auxiliaries and tense: the case of Dutch 85

Pattern D: Preterite perfective aux þ infinitive modal þ infinitive main verb

(d) We hadden de bloemen moeten/mogen/kunnen plukken
We had the flowers must/may/can pick
Pattern E: Present future aux þ infinitive modal þ infinitive main verb
(e) We zullen de bloemen moeten/mogen/kunnen plukken
We will the flowers must/may/can pick
Pattern F: Preterite future aux þ infinitive modal þ infinitive main verb
(f ) We zouden de bloemen moeten/mogen/kunnen plukken
We would the flowers must/may/can pick
In general terms, Patterns A and (especially) B are the least frequent by far. As will
be shown in Section, these two patterns, unlike the other ones, also show
considerable ‘behavioural differences’ between the three modals. Apparently, these
patterns are marked for most uses of the modals. This is a striking difference from
English, in which—considering that real past-tense uses of the preterite forms of
most modals are rare if not fully extinct—Pattern A is basically the only possible
combination. Pattern F, on the other hand, is significantly more frequent than the
others, suggesting that it is special—the qualitative analysis below will show how.
At the level of the individual meanings, however, Table 4.3 should be interpreted
with care, since there are very few instances of most of these combinations in our
data. In fact, some of the missing combinations are actually possible, intuitively. In
most cases, then, one cannot make strong statements about the relative prevalence of
patterns and meanings on the basis of these data. Nevertheless, Table 4.3 is a useful
guide for the following qualitative exploration. Major functions of the tenses in Patterns A–F Before going into the
possibilities of the different modal meanings in Patterns A–F (in Section,
let us first have a look at the most important functions which the tense forms in
them serve, irrespective of the different modal meanings, since quite a few of these
turn out to be ‘special’. These ‘special’ uses usually also exist without a modal, but,
although again we have no comparative information about their frequency in
‘modal-free’ contexts, we have the strong impression that they are relatively more
frequent in the presence of a modal.
Patterns A and B behave comparably. In both, the tense form typically does either
of the following two things: either (a) it situates the completion of the SoA before a
reference time; or (b) it situates the SoA in the past.
In (a), which we will call the ‘anteriority’ (‘ant’) reading, the reference time is
contextually implied or expressed in the clause in a (temporal) adverbial phrase. In
Pattern A this situation remains independent of the speech time, hence the utterance
receives a hypothetical or generic reading. For example, (6) expresses that ‘your mater-
ials should be degreased by the time you use oxygen’, irrespective of the speech time,
hence is not referring to any specific instance of the SoA, but is generically applicable.
86 Pieter Byloo and Jan Nuyts

(6) Indien je met zuurstof werkt dan mag je de verbindingen niet maken met vet en
vlas en alle ander materiaal moet goed ontvet zijn
‘If you work with oxygen, then you shouldn’t secure the joints with grease and
flax, and all other materials should be well degreased’ [C]
In Pattern B, the preterite situates the ‘completion of the SoA anteceding a reference
time’ in the past (‘ant-pst’), i.e. with reference to the speech time, as in (7) (con-
structed because there are no instances in our data).
(7) We moesten de bloemen geplukt hebben tegen 5 uur
‘We had to have the flowers picked by 5 o’clock’
In this ‘anteriority’ use in Patterns A and B, then, the perfect acts as a phasal aspect
marker expressing ‘completedness’, rather than a time marker expressing pastness,
and in Pattern B the preterite serves a purely temporal function.
In (b), however, the perfect does serve as a past (‘pst’) time marker. An example
with Pattern A is offered in (8).
(8) dat moet toch beangstigend geweest zijn
‘That must have been frightening’ [C]
The preterite in Pattern B can then do either of two things (depending on the
circumstances): it creates a past in the past (hence serves as a time marker; ‘pst-pst’),
as in (9), or it marks counterfactuality (instead of time; ‘pst-cf ’), as in (10) (which
indicates that the SoA did happen).12
(9) We hielden er rekening mee dat hij onze list doorzien kon hebben
‘We took into account that he could have fancied our trick’
(10) we mochten dat niet gedaan hebben
‘We shouldn’t have done that’ [C]
Patterns C and D behave in part differently, however. In Pattern C the tense form
typically situates the SoA in the past (‘pst’); the perfect serves as a time marker here.13
(11) we hebben kortbij foto’s mogen maken van de toestellen
‘We were allowed to take close-up pictures of the airplanes’ [C]
This purely temporal use of the tense form also occurs in Pattern D, expressing a
past in the past (‘pst-pst’), i.e. with the preterite adding an extra temporal dimension,
as in (12) (the context makes it obvious that this is the intended reading).

By ‘counterfactuality’ we mean ‘the opposite of what is expressed in the clause’. In (10), for example,
this ‘counterfact’ is actually a factual SoA.
Note that in this case, as in most or all other cases in Patterns A–F, the ‘present’ tense as such, unlike
the preterite, does not do anything in terms of situating the SoA in time.
Modal auxiliaries and tense: the case of Dutch 87

(12) Annemarie had Griekse euro’s moeten meebrengen uit Kreta

‘Annemarie was supposed to have brought Greek Euros from Crete’ [C]
This is, however, by far the least frequent situation in Pattern D. The dominant use
of Pattern D (13 of the 15 instances in our data) is to express counterfactuality of the
SoA. Usually there is still past-time reference (‘pst-cf ’). In that case the perfect still
serves a temporal function, and the counterfactuality is due to the preterite. This is
exemplified in (13) (from a live report of a soccer game).
(13) Van Diemen had sneller kunnen inschuiven
‘Van Diemen could have crossed more quickly’ [C]
But in a different context, even (a variant of) (13), for example, can be interpreted as
a counterfactual referring to a present SoA, i.e. the preterite perfect auxiliary as a
whole serves as a counterfactuality marker (‘cf ’); compare (14).
(14) Als die verdediger daar nu niet in de weg stond dan had VD nu kunnen
‘If that defender had not been standing there, VD could have crossed now’
Also Patterns E and F in part behave differently. Pattern E is used to mark a future
SoA (‘fut’), as in (15).14
(15) Door de symbolen in je dromen te ontcijferen zul je overdag bewuster kunnen
‘By deciphering the symbols in your dreams you will be able to react more
consciously during the day’ [C]
Pattern F is also used to express a future, but a future in the past, i.e. with the
preterite also serving a temporal function (‘fut-pst’). This applies to only one of the
44 instances in our data, however: the one in (16).
(16) Een geschiedenis van de recente Nederlandse literatuur zou nog moeilijk
kunnen geschreven worden.
‘Writing a history of recent Dutch literature would henceforth hardly be
possible any more’ [C]
In the overwhelming majority of instances of Pattern F, however, the preterite future
auxiliary performs a number of entirely ‘non-temporal’ functions (many more than
the past perfect in Pattern D; this no doubt explains the relatively very high
frequency of Pattern F in Table 4.3). This includes marking counterfactuality (‘cf ’),
as in (17) (which implies that journalists do try to influence opinions), or serving as

Zullen here sometimes has an ‘epistemic flavour’ (cf. Section 4.2.3), but it is systematically impossible
to decide whether this is an intended meaning. In any case, all instances do allow a future reading.
88 Pieter Byloo and Jan Nuyts

a mitigator, often of the modal auxiliary (‘mit’), as in (18) (from a live report on
a race).15
(17) journalisten zouden niet aan opiniëring mogen doen
‘Journalists shouldn’t try to influence opinions’ [C]
(18) inmiddels zou ‘t ook wel eens zilver kunnen gaan worden als ‘t tempo nog iets
omhoog kan
‘Meanwhile it might become a silver medal if the speed can go up a bit more’ [C] The modal meanings and Patterns A–F Let us now turn back to Table 4.3
and to the question of how the different meanings of the modals behave in the patterns. Dynamic modality16

In our data, dyn-inh does not occur in any of the perfective Patterns A–D, and also
intuitively this combination is impossible, at least when these patterns express
pastness or anteriority, that is in the vast majority of their uses. For example,
although (19) nearly automatically triggers an ability reading, none of the variants
in (20) (not even Pattern D when expressing past counterfactuality) allows it.17
(19) Jan kan/kon zwemmen
‘John can/could swim’
(20) Jan kan/kon gezwommen hebben [Patterns A–B]
Jan heeft/had kunnen zwemmen [Patterns C–D]
‘John can/could have swum’
When Pattern D in (20) is read as a present counterfactual, however, dyn-inh is possible:
(21) Als Jan het in zijn jeugd geleerd had, had hij nu kunnen zwemmen
‘If John would have learned it in his youth, he would now be able to swim’
The incompatibility between dyn-inh and pastness/anteriority in these patterns is
probably due to the fact that the former presupposes some kind of ‘permanency’
(abilities/needs do not change quickly), which is in conflict with the latter. This
conflict is not present in the simple preterite in (19) because of the different ‘type’
of ‘pastness’ involved: the Dutch perfect is typically used to situate ‘incidental’ SoAs;
the preterite to situate more permanent ones (Haeseryn et al. 1997: 121ff.).
Dyn-imp and dyn-sit, however, are possible in all perfective Patterns A–D (even
in the combinations missing in our data). But, in line with the general tendency, they
It is not always easy to determine precisely what zou does in these non-temporal uses, but we cannot
explore this further here.
As mentioned in Section 4.3, we will leave aside the dynamic uses of mogen here.
There is one remarkable exception to this, i.e. when modal kunnen is the main verb in Patterns C–D:
for example, ik heb dat ooit gekund ‘I once knew how to do this’ does express past ability. We will not go
further into this here, though.
Modal auxiliaries and tense: the case of Dutch 89

are much more common in Patterns C–D than in A–B, and the possibilities in the
latter are restricted and different for the individual modals.
Thus, in patterns A–B, in moeten both dyn-types appear to occur exclusively with
the anteriority reading of the tense form—(6) above illustrates this for dyn-sit in
Pattern A, (7) for dyn-imp in Pattern B. The dyn reading in these is not affected by
the perfect expressing anteriority, but it is by the preterite expressing pastness in
Pattern B (the dynamic necessity is situated in the past). A (past or counterfactuality
in the) past reading of the type in (8)–(10) appears excluded. (22), for example, which
practically excludes an anteriority reading, is impossible.
(22) *Die auto moet/moest wel uit de bocht gegaan zijn met zo’n hoge snelheid
‘That car had to go off the road with such a high speed’
When kunnen is used in these patterns, however, a dyn-imp reading appears
impossible. It is only possible, with an anteriority reading of the tense form, if the
perfective auxiliary is replaced by the aspectual auxiliary krijgen ‘get’; cf. (23):
(23) ??We kunnen/konden de bloemen wel geplukt hebben tegen 5 uur
We kunnen/konden de bloemen wel geplukt krijgen tegen 5 uur
‘We can/could manage to pick the flowers by 5 o’clock’
But dyn-sit with kunnen is possible in these patterns, and not only with an anteriority
reading, but also with a (past in the) past reading. (9) above is an illustration. A
counterfactuality reading in Pattern B appears excluded though. We have no explan-
ation for these seemingly quite ad hoc conditions, but they do underscore the fact that
Patterns A and B are marked for expressing the dynamic readings of the modals.
In Patterns C and D, however, the dyn-imp and dyn-sit readings in both modals
appear unrestricted, with all the readings of the tenses in them. Thus, Pattern C
systematically renders a simple past reading, which affects the dynamic meaning.
(24) illustrates this for dyn-imp moeten.
(24) De twee ‘Swingpaleis’-opnames van deze week heb ik moeten afzeggen
‘I had to cancel both of this week’s ‘Swing Palace’ recordings’ [C]
Pattern D allows a past in the past reading, but our corpus instances are all past
counterfactual. (25) illustrates this for dyn-imp moeten; (26) for dyn-sit kunnen.
(25) ik had eigenlijk van niks moeten gebaren
‘I should have acted as if nothing had happened’ [C]
(26) De opleiding van onze 1.000 computergebruikers had beter gekund
‘The training of our 1,000 computer users could have been better’ [C]
Note that only the SoA is counterfactual, not the potential/necessity. In other words,
the modal is not affected by the preterite-marked counterfactuality. Also, a present
90 Pieter Byloo and Jan Nuyts

counterfactual reading of Pattern D is possible, even in (26), for example, if situated

in an appropriate context: ‘if we had more instructors, then the training as we have it
now could be better.’ But, interestingly, in that case the dynamic meaning is affected
by the counterfactuality marker (‘the training currently cannot be better’).
In general, then, there is hardly any overlap in what is expressed by Patterns A–B
vs. C–D with dynamic kunnen or moeten. The only exception is a (past in the) past
dyn-sit reading in kunnen, but even then the alternatives are not equivalent. Com-
pare (27) featuring Pattern A and (28) featuring Pattern C:
(27) De trein kan ook ontspoord zijn door een steen op de rails
‘It is also possible that the train was derailed by a rock on the track’
(28) De trein is kunnen ontsporen door een steen op de rails
‘The train has derailed [lit.: it has been possible for the train to derail] due to a
rock on the track’
In (28), the SoA (the derailing due to a rock) is factual (it did happen for that
reason), while in (27) its factuality status is open (we do not know whether the
derailing is due to a rock). The reason might be that in (28) the perfect affects the
modal and situates the dynamic value in the past, and due to this the inherently non-
factive nature of dynamic modality is overruled and the SoA is anchored in real time
(by the perfect), while in (27) the modal is not affected by the perfect, hence the
dynamic meaning retains its non-factive effect on the SoA.
Next, all dyn types, in both modals, occur in future Patterns E and F, in all their
uses. The temporal uses are illustrated in (15) and (16) above (with Patterns E and F,
respectively, and both with dyn-imp kunnen). In these the modal is affected by the
tense auxiliary, as in perfective Patterns C–D, but the SoA is nevertheless systemat-
ically non-factual due to the inherently non-factive nature of the future (contrary to
the factive nature of the perfect). The non-temporal use of Pattern F is illustrated in
(18) (which is, however, ambiguous between dyn-sit and epi kunnen), involving a
mitigating use of zou, or in (29) (with dyn-imp moeten), involving counterfactual
zou (it implies that currently the Centre does collaborate).
(29) We moeten er toch niet aan denken dat we die cursussen nog zouden moeten
organiseren zonder de medewerking van het Centrum voor Volwassenenonderwijs
‘Just imagine we would have to organize these courses without the help of the
Centre for Adult Education’ [C] Deontic modality
As Table 4.3 suggests, the possibilities of deo in Patterns A–F are highly restricted. It
is possible in perfective Patterns A and B (though again with variation between the
individual modals), with two readings of the tenses. One is the (past) anteriority
reading. We have one corpus case, (30), with mogen in Pattern A, meaning: ‘it’s OK
Modal auxiliaries and tense: the case of Dutch 91

for El Al pilots to have “forgotten” by the time they arrive here (for security reasons)’
((30) is in fact ambiguous between deo and dir).
(30) als mensen van El Al komen hier dan zijn ze alles vergeten wat ze hebben
gedaan. en je mag alles vergeten zijn
‘When El Al people arrive here then they have forgotten everything they have
done, and you/they may have forgotten everything’ [C]
This use is also perfectly imaginable for moeten, and also for mogen and moeten, in
past Pattern B. In past Pattern B, the deontic meaning then automatically receives a
descriptive reading. Deo kunnen, however, seems impossible in this reading of Pat-
terns A and B. For example, (23) above is also out with a deontic reading, and even
replacing the perfective by the aspectual auxiliary krijgen ‘get’ does not work here.
Also possible is the counterfactuality reading in Pattern B. This is illustrated by
the only deontic instance in this pattern in our data, (10) above, featuring mogen. But
a comparable use is possible with kunnen, as in (31), and also, albeit marginally
(possibly dialectally restricted), with moeten, as in (32).18
(31) Je kon die bloemen nu toch wel even geplukt hebben voor mij
‘You could have picked those flowers for me’
(32) Je moest die bloemen maar geplukt hebben
‘You should have picked those flowers’
Interestingly, in all these cases the deontic reading is performative, i.e. the counter-
factual preterite only affects the SoA, not the deontic meaning.
There are no deontic instances in Patterns C and E in our data, nor in Patterns D
and F with a past or future in the past reading. Intuitively, these do not appear
possible either, not even with deo in a descriptive reading. For example, neither
variant in (33) seems to allow the reading that the SoA is morally necessary.19
(33) We hebben/zullen de bloemen moeten plukken
[?? in this reading:] ‘It was / will be our moral duty to pick the flowers’
As Table 4.3 shows, deo does occur frequently in Patterns D and F, but this is
exclusively in their counterfactual (in D) and non-temporal (in F) uses. In the past
counterfactual reading of Pattern D, the counterfactual SoA is situated in the past,
but the deontic meaning is (in principle)20 performative, that is, not affected by the

This counterfactuality use of Pattern B is generally somewhat marked though; many speakers would
prefer to use Pattern D to express this meaning.
(33) does allow a directive reading, however; see below. This goes to show that deo and dir do behave
differently in a number of respects, which is, of course, in line with our treating them as separate categories
(cf. Section 4.4.2).
A few cases in our data are descriptive, but always due to factors other than the tense form, such as
appearance in a question, in indirect speech, etc. The same applies to a few descriptive deo cases in Pattern F.
92 Pieter Byloo and Jan Nuyts

perfect expressing pastness, nor by the preterite expressing counterfactuality. (13)

above is an example with kunnen (ambiguous with dyn though); (34) features mogen.
(34) Het ging over een in wezen volstrekt triviaal fait divers, dat politiek van geen
enkel belang had mogen zijn
‘It was all about an essentially completely trivial fait divers, which [in the
speaker’s present opinion] politically shouldn’t have played any role’ [C]
Also, in a present counterfactual reading of Pattern D, the deontic meaning is not
affected by the past perfective auxiliary expressing (present) counterfactuality.
The situation is similar in Pattern F: the deontic meaning is (in principle; cf. footnote
19) performative, hence is not affected by zou when it is used as a counterfactuality
marker, as in (17) above, or in (35).21 (In (35) the SoA is non-factual—it is open
whether it will be realized in the future—but this can nevertheless be considered a
counterfactual use in that it implies that at present the opposite of the SoA applies.)
(35) Homeopathische apotheken zouden uit woonzones moeten verwijderd worden
‘Homeopathic pharmacies should be removed from residential areas’ [C] Epistemic modality / evidentiality
For epistemic kunnen and evidential moeten the possibilities are strongly limited.
They do occur in Pattern A, always involving a performative modal affecting a past
SoA. (8) above is an illustration with evi moeten. But they are absent in our data, and
are also intuitively impossible, in Patterns B–E. In Pattern F we do find epi kunnen,
relatively (within this meaning category; cf. Tables 4.2 and 4.3), even very, frequently
(the combination zou kunnen stimulates an epistemic reading of the modal), and
apparently zou here serves exclusively as a mitigator of the modal, modifying its
epistemic value while leaving its performativity intact, as in (18) above, or in (36)
(both are ambiguous with dyn).
(36) Irak zou wel ‘ns bommen met giftige gassen op Israël kunnen gooien
‘Iraq might drop bombs with poison gas on Israel’ [C]
Evi moeten, however, is absent in Pattern F in our data, and is also intuitively
impossible in it, even with zou serving as a mitigator. Inserting zou in (8), for
example, rules out an evi reading, and even makes the utterance unacceptable. Directivity
The directive reading behaves quite like dyn-imp. Dir is possible in Patterns A and B
(in spite of its absence in the latter in our data), but only to a limited extent, and with
variation between the modals. Dir moeten only allows an anteriority reading. (6)
Interestingly, zou in our instances never functions as a mitigator of a deontic modal, in the way it
does function as such in the epistemic use of kunnen (see below).
Modal auxiliaries and tense: the case of Dutch 93

above (ambiguous with dyn, though) illustrates this for Pattern A. We have no
instances in our data for Pattern B, but (37) shows that it is possible, with the
directive situated in the past (descriptive).
(37) We moesten het gebouw voor 5 uur verlaten hebben
‘We had to leave the building before 5 o’clock’
Dir mogen also only allows an anteriority reading in Pattern A, as illustrated in
(ambiguous) instance (30) above. But it appears impossible in Pattern B; cf. (38).
(38) ??We mochten het gebouw voor 5 uur verlaten hebben
‘We were permitted to leave the building before 5 o’clock’
Dir kunnen, finally, does not appear possible at all in Patterns A and B.
Dir occurs freely (in all modals) in all other patterns, however, in all the readings
of the tenses. Thus, Pattern C systematically produces a past reading, with a
descriptive directive; cf. (11) above. The only corpus instance in Pattern D, (12)
above, features a past in the past reading. But intuitively, this pattern also, and more
readily, allows a past counterfactuality reading, as in (39), rendering a past counter-
factual SoA and a descriptive (past) directive.
(39) We hadden de bloemen mogen plukken
‘We were allowed to pick the flowers (but we didn’t do it)’
Also, a present counterfactuality reading of Pattern D is possible, rendering a
counterfactual (hence descriptive) directive.
Pattern E systematically expresses a future directive (i.e. descriptive again), as in (40):
(40) Het ontwerp vaste boekenprijs voorziet dat boekhandelaars geen prijskortin-
gen van meer dan 5% zullen mogen toestaan
‘The draft of the fixed book price regulation stipulates that book sellers will not
be permitted to give a discount of more than 5%’ [C]
And in Pattern F, although a future in the past reading appears possible, in all corpus
instances zou serves non-temporal functions. In (41), for example, it marks counter-
factuality (it implies that currently nothing is censored), and in (42) it mitigates the
speech act (the question).
(41) wat zou er gecensureerd moeten worden?
‘What should be censored?’ [C]
(42) zou ik je nou nog een keer terug mogen bellen
‘Could I call you back once more?’ [C]
Unlike with deo, it appears that dir in this pattern cannot be performative but only
descriptive. The descriptivity is due either to contextual factors, such as its appear-
ance in a question, as in (41) and (42), or to zou itself. (43), for example, can only be
94 Pieter Byloo and Jan Nuyts

read directively if the speaker is reporting on an obligation issued by someone else

but without enforcing it himself on the hearer. Hence zou in a way expresses
counterfactuality again.
(43) Je zou nu moeten vertrekken
‘You are supposed to leave now’

4.5 Discussion
Summarizing the analyses in Section, Table 4.4 shows how modal meanings
and tense meanings (see Section for the abbreviations) combine in the
different patterns. It also indicates whether the modal meaning is affected (shaded)

Table 4.4 Possible combinations of modal and tense meanings

aux aux pret aux aux pret aux aux pret
dyn-inh / / / cf fut fut pst
dyn-imp moeten moeten pst pst pst fut fut pst
ant ant pst pst cf cf/mit
kunnen kunnen cf
/ /
dyn-sit moeten moeten pst pst pst fut fut pst
ant ant pst pst cf cf/mit
kunnen kunnen cf
ant ant pst
pst pst pst
deo mogen mogen / pst cf / cf
ant ant pst cf
pst cf
moeten moeten
ant ant pst
pst cf
kunnen kunnen
/ pst cf
epi/evi pst / / / / moeten
dir mogen mogen pst pst pst fut fut pst
ant / pst cf cf/mit
moeten moeten cf
ant ant pst
kunnen kunnen
/ /
Modal auxiliaries and tense: the case of Dutch 95

or not (not shaded) by the actual reading of the tense auxiliary and (if present) the
preterite marker. Table 4.4 only differentiates between the modals when relevant.
Now, the question arises as to how these observations match with the qualifica-
tional hierarchy in (1). In general, they match according to expectation, but there are
some puzzling elements.
The dyn meanings (when they are possible) are systematically sensitive to the
tense forms when these indicate time and have grammatical scope (cf. Sections 4.2.4
and over the modal, viz. the preterite in Patterns B, D, and F, and the
perfective and future auxiliaries in Patterns C, D, E, and F. Since dyn is below
time in the qualificational hierarchy in (1), the grammatical scope relation here
corresponds to the semantic one, and everything is according to expectation.
Likewise, in Patterns A–B the dynamic modal has grammatical scope over the
perfective auxiliary; hence it is no surprise that the latter can in principle receive only
an anteriority—phasal aspect—reading, which is below dynamic modality in (1). If a
past-time reading of the perfective auxiliary were possible, the grammatical scope
would not correspond to the semantic scope. There is one remarkable and highly
idiosyncratic exception though: dyn-sit kunnen does allow a past reading of the
perfect (unlike dyn-sit moeten, which behaves according to expectations). So, in this
singular case the grammatical scope does not correspond to the semantic one in (1).
But the semantic scope hierarchy in (1) actually does not ‘work’ either: the past
meaning does not affect the dynamic meaning; it does not situate the latter in the
past, unlike in all combinations of dyn-sit with a time referential tense auxiliary in
Patterns C–F, including those with kunnen. Nor is the semantic scope turned
around, such that it violates the hierarchy in (1): the dyn-sit reading does not affect
the temporal one. Rather, the dynamic and temporal meanings are simply juxta-
posed. One can speculate that the inversion of grammatical scope (vis-à-vis the
normal semantic scope) is precisely meant to trigger this semantic effect (i.e. to
neutralize normal semantic scope relations). But if this is so, why is it not possible
with moeten as well, or with dyn-imp? We have no answer to this question.
Next, deo modals generally ‘refuse’ to occur in the grammatical scope of the
perfective or future auxiliaries when these mark time, viz. in Patterns C–F. This is in
accordance with the semantic hierarchy (1)—but it means a deo modal cannot be
descriptivized in this context. There is again one exception: in Pattern D deo does
occur in the grammatical scope of the perfective auxiliary marking pastness, on
condition that the preterite does not mark time as well. The deontic reading still
remains insensitive to the pastness expressed by the perfect, counter to the gram-
matical scope relation but in line with the semantic hierarchy in (1), and in line with
the apparent inability of a tense auxiliary to descriptivize deo.
In Patterns A–B, where the deontic modal has grammatical scope over the
perfective marker, it fairly systematically triggers a phasal aspect reading of the
latter, and does not allow a temporal reading. Why the latter should be the case is
96 Pieter Byloo and Jan Nuyts

unclear since not only phasal aspect but also time is in the semantic scope of deo.
Once more there is one exception: in Pattern B, when the preterite receives a
counterfactual reading, then a past reading of the perfect is possible and, in line
with both the grammatical and the semantic scope, it does not affect the deontic
reading (which is performative).
Deo generally pushes the preterite into a counterfactuality reading in all patterns,
except in mogen and moeten in Pattern B, where it receives a past reading if the
perfective auxiliary expresses phasal aspect. And here, then, this preterite does
descriptivize the deontic reading.
Moving to the next category: that epi and evi should not occur in the scope of any
time marker—be it the preterite or a temporal auxiliary, in Patterns B–F—was to be
expected, since they have semantic scope over time in (1), and since even in the
simple tenses they cannot be descriptivized by a time marker with grammatical
scope over them. Apparently, unlike all other modal meanings, they do not tolerate a
tense form serving as a counterfactuality marker either. This may be so because
epistemic modality (and, possibly, inferential evidentiality) and counterfactuality are
somehow semantically akin. Finally, it is striking that epi/evi only permit a temporal
reading of a perfective auxiliary in their scope, not a phasal aspect reading, although
the latter is lower in the hierarchy in (1), exactly like time.
Finally, dir, despite being a non-qualificational meaning, is nevertheless sensitive to
time markers, in a perfectly straightforward way. In all patterns, both the tense auxiliaries
and the preterite when they express time and have grammatical scope over the directive
modal, trigger a descriptive reading of the directive. This is entirely as may be expected:
illocutions/speech acts function as such (i.e. are performative) only when they are bound
to the speaker’s here and now. On the other hand, in Patterns A and B dir does not permit
a temporal reading of the perfective auxiliary in its grammatical scope, but only a phasal
aspect reading. Again, this is to be expected: one does not/cannot issue directives
(permissions, obligations) regarding a SoA in the past.
One last remarkable observation concerns the counterfactuality uses of tense
forms. The preterite used as a counterfactuality marker systematically affects only
the SoA, but not the modal meaning, and not even dyn. But the preterite perfect and
future auxiliaries had and zou, when used to mark a present (or future) counter-
factual, do affect dyn and dir, but not deo. We have no idea what triggers these
differences, nor what this may mean for the status of counterfactuality as a semantic
category. Exploring this further is in any case beyond the present scope.

4.6 Conclusion
In conclusion, it turns out that the ‘qualificational hierarchy’ introduced in Section
4.2.4 succeeds fairly well in explaining—hence is empirically supported by—the
complex behaviour of modals and tenses in the verbal group in Dutch. The semantic
Modal auxiliaries and tense: the case of Dutch 97

hierarchy is never violated. And, in general, grammatical scope relations do respect

the semantic hierarchy: the latter ‘shines through’ at the linguistic surface. But there
are clearly other elements interfering, since in a few cases the grammatical hierarchy
does not correspond to the semantic hierarchy. Sometimes this does not affect the
semantic hierarchy—for example when a deontic modal remains performative in the
grammatical scope of a time marker. But sometimes it does affect the semantic
hierarchy in the sense that it is neutralized (but not violated, i.e. turned around), as
with dyn-sit kunnen in Patterns A and B. And sometimes extra restrictions apply
which are unexpected in view of the semantic hierarchy, even if they do not violate
it, as we have seen in several cases in Patterns A–B. Presumably, many if not all of
these are explicable on the basis of the inherent semantic properties of the categories
involved: for example of the kind presented above for the incompatibility between
dyn-inh and pastness as expressed by the perfect. Because we are not always able to
formulate them at present, further research will have to show whether this assump-
tion is correct.
In terms of our concept of modal and related notions (cf. Section 4.2.2), our
analyses have shown that dir behaves more like dyn-imp than like deo, which
supports our treating them as separate categories. Also, although dir shows com-
binatorial restrictions with time categories comparable to those in qualificational
modal meanings, we have been able to explain these entirely on the basis of the
inherent properties of illocutionary notions. So our analysis of dir as a type of the
latter, which moreover stands outside the qualificational hierarchy, is not contra-
dicted either. Our analyses do also raise some—again unanswered—questions,
however, most notably the observation (made on several previous occasions; cf.
Nuyts 2001) that the modals do not allow descriptivization of deo or epi/evi in
certain conditions. Here, too, further research is needed to unravel the puzzle.
Finally, in terms of the basic notions of the cognitive–functional approach, as
discussed in Section 4.2.1, the deeper moral behind all of this is that it goes to show
once again that one cannot assume a simple one-to-one relationship between form and
meaning in language. The risk of such a view is alluring in at least some strands in
current linguistics (cf. the basic concept of a construction as a fixed form–meaning pair;
see Nuyts 2008b, 2011). The linguistic surface is shaped by the interaction of many
semantic and functional dimensions, including general principles such as the qualifica-
tional hierarchy, but also by much more ‘local’ elements, such as the inherent semantic
features of singular semantic categories or elements. A process type of model of
language of the kind assumed in the cognitive–functional approach—with separate
levels of conceptual representation and linguistic representation which are linked by
‘intelligent’ procedures sensitive to different kinds of ‘information’—is at least in
principle compatible with this fact (pace arguments that have been formulated against
it; cf. Section 4.2.1). For example, how a model based on the concept of constructions as
fixed form–meaning pairs can handle it (in a ‘parsimonious’ way) is less obvious.

Semantic and pragmatic aspects of

the interaction of time and modality
in French: an interval-based account


5.1 Introduction
This chapter aims at a better understanding of the complexities of the interaction
between the temporal and modal meanings in French by means of a model inspired
by Reichenbach’s (1947) classical apparatus but in which the coordinates are replaced
by intervals (among other amendments). Such a replacement, we suggest, bears a
number of theoretical consequences that allow a model to encompass a large scope
of phenomena dealing with time, modality, and their interaction within a single
framework. French shows a particularly complex variety of items where the complex
interaction of time and modality have long been problematic for the grammatical
tradition. In the argument developed in this chapter, we evoke the notions referred
to by French linguists when tackling these objects of study, observing that the
literature in the field too often struggles to resolve the semantics of temporal and
aspectual markers with the classical but inaccurate assumption that temporal
and modal expressions form separate categories. To the contrary, this chapter
envisages a very different approach, in which each eventuality referred to in an
utterance is not only situated in time and presented with an aspectual viewpoint, but
also conveys a semantic modality that specifies that the eventuality is conceived of
either as a fact or as a mere possibility. We argue that the modality is derived from
the aspectual viewpoint on the basis of the modal asymmetry of linguistic time; that
is, that what precedes the point of reference is irrevocable, and belongs to historical

I am extremely grateful to Louis de Saussure for his helpful comments and suggestions, which greatly
improved this chapter.
The interaction of time and modality in French 99

necessity, whereas what is posterior to the reference point is represented as merely

possible. In this framework, this assumption leads to a set of predictions concerning
the modal meaning of tenses and of the so-called ‘aspectual periphrases’ which are
used in French to express phasal aspect (see Dik 1989).
In Section 5.2, we argue against the classical conception of relations between time
and modality, in particular, in the French tradition. We show that, to the contrary,
temporal and modal expressions cannot be considered as strictly separate. Then we
present the hypothesis of the modal asymmetry of linguistic time (Section 5.3), and
our model using intervals (Section 5.4). The integration of the hypothesis of modal
asymmetry of time within this model leads to a new analysis of modal values of
aspectual viewpoints, and consequently of tenses (Sections 5.5 and 5.6). Finally, we
extend this analysis to a number of ‘aspectual periphrases’ (Sections 5.7–5.9).

5.2 General preliminaries

The French literature usually regards time and modality as strictly distinct,2 that is, it
considers that what is temporal cannot be modal and vice versa. The idea stems from
the grammatical tradition that reasons in terms of ‘modes’, that is, semantic modal-
ities associated with paradigms of tenses. In this line of thought, all ‘modes’ but the
indicative have exclusively non-temporal, modal values, whereas the indicative may
have, depending on the context, either (non-modal) temporal values or (non-
temporal) modal values, as summarized in Figure 5.1.
A particularly interesting case in this respect is the traditional analysis of the
French imperfective past (imparfait, henceforth IMP). This tense is traditionally
seen as bearing a temporal meaning in most cases, as in (1a), but also a modal one,
indicating irrealis, in a number of specific cases, including if-clauses but not
restricted to them; see the examples in (1b–c):3
(1) a. Le 1er janvier 182 . . . , le baron François-Armand de Luizzi était assis au coin
du feu, dans son château de Ronquerolles. (opening sentence of Les Mém-
oires du Diable by F. Soulié)
On January 1st, 182 . . . , Baron François-Armand de Luizzi sit-IMP (‘was
sitting’) by the fire, in his castle at Ronquerolles.4

With the exception of Wilmet’s grammar (1997: §414), according to which “each category is attached
to a formula that combines a modal variable with a temporal or aspectual variable” [my translation]. This
dichotomy can also be found in English linguistics around the question of whether the future will is of a
temporal or modal nature (for a criticism, see Vetters and Skibinska 1998: 255).
One can read, for example: “The indicative imperfect tense denotes a process situated outside the
speaker’s present actuality. It takes on a temporal value when the process is shifted back in time and a
modal value when it is considered possible outside the real universe” (Riegel, Pellat, and Rioul 1994: 305
[my translation]).
We provide full translations only when the gloss is not clear enough; conversely, we provide glosses
only when necessary.
100 Laurent Gosselin


indicative non-indicative

temporal meanings modal meanings


FIGURE 5.1 The classical view on tenses and modality.

b. Si je gagnais au loto, je m’achèterais une voiture.5

If I win-IMP (‘won’) the lottery, I would buy a car.
c. Sans son intervention, je me trompais de chemin.6
Without his intervention, I make-IMP a mistake about the way to follow
‘Without his help, I would have gone the wrong way.’
By comparison, it is assumed that the French simple past (henceforth SP) always
has a temporal value (inscribing the eventuality, so to speak, in the realm of the
real),7 though a more detailed analysis shows that there are in fact usages where the
simple past expresses only a possibility, as (2) demonstrates:
(2) Quant à Fausta, si cette révélation l’émut, si le visage bouleversé de ce père lui
inspira autre chose que de la curiosité, on n’eût pu le savoir. (M. Zévaco, La
As to Fausta, if this revelation move-SP her, if the anguished face of this father
inspire-SP her anything other than curiosity, one could not know it.
‘As to Fausta, there was no way of knowing if this revelation moved her, if the
anguished face of this father inspired anything other than curiosity.’
This alleged dichotomy rests on questionable assumptions, notably that we hold as
temporal that which is certain, proven, known, asserted, but as modal that which is
uncertain, potential, considered, unasserted. It is striking to note that such assumptions
are actually incompatible with the definitions of time and modality that are adopted
even within those studies which posit them. Indeed, the classical view is that tense is
about localization within a time period, determined in relation to the speech point
(‘absolute time’) or to a point of reference (‘relative time’). Modality is itself a concept
that is variously understood from trend to trend, but it is usually conceived of as the
linguistic marking of the speaker’s attitude vis-à-vis the represented content. If we

On these usages, see Martin (1991); Gosselin (1999); Bres (2005); Patard (2007).
On these usages, see Kleiber and Berthonneau (2003); Berthonneau and Kleiber (2006); Bres (2006).
See, among others, Touratier (1996: 101–2); Caudal, Roussarie, and Vetters (2003: 63).
See Brunot (1922: 887–8).
The interaction of time and modality in French 101

adhere to these definitions, the eventualities expressed in examples (1b–c) and (2) bear
not only a temporal meaning but at the same time also a modal one. They are indeed
situated on the timeline with regard to the speech point as anterior (past, examples (1a
and 1c)), simultaneous (present), or posterior (future, example (1b)), but they are
obviously entertained as possible or otherwise unreal, thus bearing a modal meaning.
Two options are then on offer: either we alter the definitions of ‘temporality’ and
‘modality’, which does not seem very sound, or we reject the classical ‘dichotomy’
assumption in favour of a more elaborated account of the interaction of temporal
and modal meanings as empirically observed. This, we suggest, amounts to con-
sidering time, aspect, and modality as essential components of all utterances, each in
its own way, although not implying that their various meanings could not be selected
in various circumstances according to pertinent schemes of interpretation.9 If this
perspective is correct, then all eventualities, be it a state, an activity or an event, are
situated in time—more or less precisely—but also presented under a certain aspect
and according to a specific type of modality. Modality, in turn, should not be
understood in the rather fuzzy way in which it is understood in the grammatical
tradition, but rather as encompassing all possible ways the speaker positions herself
vis-à-vis the propositional content, that is, including presenting the eventuality as
merely true (what Kant calls ‘assertoric modality’).10
Furthermore, concerning these various levels of meaning, we observe the following:
(a) they are not necessarily marked by morphemes specifically dedicated to
expressing them, but generally result from the interaction between several
markers (lexical or grammatical morphemes and constructions);
(b) they have a reciprocal influence upon each other.
This is illustrated in Figure 5.2.


dimensions: time aspect modality

Markers: m1 m2 m3 m4 m5 m6
FIGURE 5.2 Relations between markers and temporal, aspectual, and modal meanings.

This entails in turn a conception where types of meanings are discriminated into, for example,
‘standard’ versus ‘derived’ meanings (or ‘values’), as in Gosselin (1996); ‘prototypical’ versus ‘non-
prototypical’ meanings (De Mulder and Brisard 2006); or ‘descriptive’ versus ‘interpretative’ usages (de
Saussure 2003, 2010 following Sperber and Wilson’s 1995 classic pragmatic distinction).
See Kronning (2004: 44).
102 Laurent Gosselin

In what follows, we will focus on the influence exerted by time and aspect on the
modal status of the eventuality in French.11

5.3 Modal asymmetries in pragmatic and semantic times

In numerous works, time is considered to have what may be called an ‘asymmetrical’
modal structure: the present and the past pertain to the irrevocable12 (‘historical
necessity’),13 while the future is conceived of as open, undetermined, simply possible.
This asymmetrical structure accounts for the fact that time presents itself as an
irreversible flux, which continuously converts the possible into the irrevocable;
this conversion occurs at a specific modal breakpoint involving a modal change.
We represent this structure in Figure 5.3:14

Past Modal breakpoint Future

Irrevocable Possible
FIGURE 5.3 The modal breakpoint.

or, according to branching time theory, in Figure 5.4:



FIGURE 5.4 Structure of branching time.

This modal structure has received various interpretations. In possible-worlds

semantics, the general assumption is that it is of a metaphysical nature,15 that is, it
expresses the objective structure of reality. However, this metaphysical conception is,
contrary to what is sometimes claimed, far from obvious. It corresponds in fact to a
vision of the structure of reality, both occidental and relatively recent, for which
there is no omniscient God, no prophets, no divination, no destiny or determinism.
According to another interpretation, this asymmetrical structure is that of subjective

I have analysed the interaction of temporal and aspectual markers in Gosselin (1996), of modal
markers in Gosselin (2010a), and the interaction of time, aspect, and modality in Gosselin (2005, 2010a).
See von Wright (1984); Vuillemin (1984).
See Thomason (1984).
See Gosselin (2005: 89); Lagerquvist (2009: 52–3).
See Condoravdi (2002); Kaufman, Condoravdi, and Harizanov (2006: 99).
The interaction of time and modality in French 103

time, and the modalities in question thus take on an epistemic value. For Culioli
(1978) and Desclés (1994: 61), the modal breakpoint separates the certain (the past
and present) from the uncertain (the future). In a more complex model,16 Langacker
(1991) opposes known reality, present or past (defined as that which the speaker
accepts as constituting reality), to irreality, which encompasses part of the present
and past and the totality of the future. According to this perspective, we cannot
entertain any certainties about the future, while the past and the present, so long as
they are known, correspond to certainty.
This division can also be correlated with the subject’s cognitive abilities. Accord-
ing to Chafe (1995), an individual accesses the present via perception, the past via
memory, and the future via imagination—and imagination seems incompatible with
certainty. But this analysis does not seem adequate, since in real life we do in fact feel
certain of many things about the future—especially the immediate future. We can
even say that many voluntary movements of our bodies are guided by such a
certainty. And, on the other hand, we do also feel a great deal of uncertainty about
the past17—especially the remote past—even if we have knowledge of it (so long as
we’ve heard about it), and it seems that we access it just as much by the imagination
as we do by memory.
In fact, the modal asymmetry of time is considered obvious only from the
perspective of practical philosophy, from the point of view of deliberation and
action. The argument is a classical one (see Plato, Protagoras 324b; Aristotle, Ethica
Nicomachea VI, 2, 1139b 6–11; Alexander of Aphrodisias, De Fato, 11–12): one cannot
modify the past, one can only deliberate and act upon the future. In this respect, the
modal difference is clear-cut and certainly unquestionable. In so far as this structure
fundamentally concerns action, we will henceforth call this the pragmatic time. The
modal asymmetry of pragmatic time is the fundamental condition for the possibility
of free action (whether physical or verbal), and thus for all that presupposes acting
freely: deliberation, rhetoric, ethics, and so forth. Even the recourse to Theory of
Mind presupposes the modal asymmetry of pragmatic time: speech acts as well as
mental states are themselves oriented towards the irrevocable, and thus towards the
present or the past (e.g. admonition, repentance, or nostalgia), or towards the
possible and thus the future (e.g. suggestions and all directive acts, hope, or fear).
Argumentation itself entails the possibility of acting upon the beliefs of an audience,
insofar as even a philosophy that would refuse to give any kind of credit to the
obvious fact of modal asymmetry would nonetheless need to concede that—from a
pragmatic angle—the future is open. Otherwise, a ‘performative contradiction’
would undermine it.18

For a critical discussion, see Gosselin (2005: 79–82).
See Jaszczolt (2009: 50). On the past–future asymmetry, see also Jaszczolt, this volume.
See Apel (1986).
104 Laurent Gosselin

In this perspective, it is the present moment (being the time of the decision about
a projected action) that brings about the modal change between the possible and the
irrevocable. The present moment thus introduces a modal breakpoint.
The issue then is to assess whether and how much this asymmetrical structure
of pragmatic time affects linguistic time. Since temporal semantics concerns the
way utterances represent eventualities and not what we think we know of the
world we live in, we must first point out that linguistic time mirrors neither
pragmatic time, nor metaphysical time, nor a ‘subjective’ notion of time. Lan-
guage does allow a speaker to evoke past possibilities (i.e. not only facts) and to
present future events otherwise than as merely undetermined, or even as inevit-
able, even when they are not naturally plausible, as in the following ‘prophetic’
usage of the future:19
(3) Aussitôt après la détresse de ces jours-là, le soleil s’obscurcira, la lune perdra de
son éclat, les étoiles tomberont du ciel . . . (The Gospel according to Matthew,
24:29, French version, Bible de Jérusalem)
Immediately after the despair of those days, the sun darken-FUT, the moon
lose-FUT its shining, the stars fall-FUT from heaven . . .
‘Immediately after the tribulation of those days shall the sun be darkened, and
the moon shall not give her light, and the stars shall fall from heaven . . . ’ (Holy
Bible, King James Version)
Some have been able to use the asymmetrical structure to give an account of
temporal semantics with some success.20 Yet this is because such proposals posit a
modal asymmetry comparable to that of pragmatic time, with a difference regarding
the moment taken to create the modal breakpoint. Vet (1981), for example, interfaces
Reichenbach’s model with possible-worlds semantics and takes the reference point R
as the key moment, explaining: “The course of events, which is real up to R, can take
various paths after R” (pp. 112–13). At the level of semantic time, it is thus R which is
assumed as creating the modal breakpoint. We draw upon this assumption within
our own framework, the Interval-semantic model of time (Gosselin 1996, 2005),
which in turn leads to further predictions.

5.4 The Interval-semantic model

ISM—for Interval-semantic model—is a Reichenbach-inspired model which was
initially designed with the aim of accounting for the semantics of tenses in French;21

Contrary to Copley (2009: 11), we observe that although the indeterminacy of the future is unques-
tionable at the pragmatic level, it is not so at the semantic level.
See, among others, Beaver and Condoravdi (2003).
This model has been applied to other languages; see, among others, Hamdani Kadri (2006).
The interaction of time and modality in French 105

its raison d’être comes basically from the identification of two major shortcomings
of the original framework by Reichenbach which (i) is too simplistic as regards
representations of the temporality of eventualities and (ii) suffers from inaccurate
computational principles. Unsurprisingly, then, the ISM model presents two funda-
mental innovations that concern precisely representational format on the one hand
and computational principles on the other.
As regards representation, Reichenbach’s coordinates are replaced by intervals,
which increases the number of potential relations. The ISM model implements four
types of interval positioned along the temporal axis:22
– [B1,B2]: the eventuality’s interval, corresponding to the categorization and
construction of an eventuality;
– [I,II]: the reference-time interval, designating the ‘window of exhibition’, i.e.
what is manifested, or shown, about the eventuality;
– [01,02]: the speech-time interval every utterance implies as a speech act;
– [ct1,ct2]: the circumstantial-time interval, whose construction is set off only by
the presence of adverbials of time (of duration and/or location).
In this framework, absolute time can be defined by the relation between [I,II] and
[01,02].23 We can distinguish three types of absolute time:
– Past: [I,II] previous to [01,02];
– Present: [I,II] simultaneous with [01,02];
– Future: [I,II] subsequent to [01,02].
The aspectual viewpoint24 corresponds to the relation between [I,II] and [B1,B2].
Four basic aspectual viewpoints can be distinguished (in what follows, SP stands for
simple past, IMP for imperfective past, CP for composed past/present perfect, PR for
present, FUT for future, FP for future in the past, PP for past perfect):
(i) Perfective viewpoint: [B1,B2] coincides with [I, II] (Figure 5.5):
(4) Il traversa le carrefour.
He cross-SP the intersection.
‘He crossed the intersection.’

This model is, in certain respects, comparable to Klein’s (1994): [B1,B2] corresponds to TSit (‘time of
situation‘, [I,II] to TT (‘topic time’), and [01,02] to TU (‘time of utterance’).
We conform to the non-standard definition of time proposed by Reichenbach, for which Klein (1994:
21–4) advances decisive arguments.
For elaboration, see Smith (1991).
106 Laurent Gosselin

B1 B2

FIGURE 5.5 Perfective viewpoint.

(ii) Imperfective viewpoint: [B1,B2] includes [I, II] (Figure 5.6):

(5) Il traversait le carrefour.
He cross-IMP the intersection.
‘He was crossing the intersection.’25

B1 I II B2

FIGURE 5.6 Imperfective viewpoint.

(iii) Perfect viewpoint: [B1,B2] precedes [I, II] (Figure 5.7):

(6) Il a traversé le carrefour depuis dix minutes.
He cross-CP the intersection since ten minutes.
‘He crossed the intersection ten minutes ago.’26
B1 B2 I II

FIGURE 5.7 Perfect viewpoint.

(iv) Prospective viewpoint: [B1,B2] follows [I,II] (Figure 5.8):

(7) Il allait traverser le carrefour.
He cross-FP the intersection.
‘He was going to cross the intersection.’
I II B1 B2

FIGURE 5.8 Prospective viewpoint.

The English progressive allows only dynamic aspect, whereas the French IMP also licenses states; see
Molendijk (2010) for a comparison of English and French in this respect.
English normally forbids combining a present perfect with a temporal adverb (a problem known as
the ‘present perfect puzzle’); see Klein (1992). Our English translation does not render the perfect meaning
of (6), and the temporal adverb does not correspond accurately to the French depuis; the schema for (6),
thus, is not accurate for English while being accurate for French, in the resultative meaning of the
composed past.
The interaction of time and modality in French 107

Three examples with their iconic aspectual–temporal representation are discussed

below with further elaborations:
Past tense with imperfective viewpoint (Figure 5.9):
(8) La police recherchait le coupable depuis trois jours.
The police search-IMP for the culprit since three days.
‘The police have been searching for the culprit for three days.’

depuis trois jours

ct1 ct2 B2
B1 I II 01 02

la police rechercher le coupable

FIGURE 5.9 Example (8).

Past tense with perfective viewpoint (Figure 5.10):

(9) La police rechercha le coupable pendant trois jours
The police look-SP for the culprit for three days.
‘The police had sought the culprit for three days.’

pendant trois jours

ct1 ct2
B1 B2
I II 01 02

la police rechercher le coupable

FIGURE 5.10 Example (9).

Past tense with perfect viewpoint (Figure 5.11):

(10) La police avait arrêté le coupable depuis trois jours.
The police arrest-PP the culprit since three days.
‘It’s been three days since the police arrested the culprit.’

Examples are not translated inside the figures. Verbs are indicated in the infinitive.
108 Laurent Gosselin

depuis trois jours

ct1 ct2
B1 B2 I II 01 02

la police arrêter le coupable

FIGURE 5.11 Example (10).

These interval structures are computed from instructions coded by various items
in the utterance in interaction with temporal–aspectual compositional rules and with
background information (encyclopaedic and situational). Technically, a temporal–
aspectual structure results from a network of constraints that apply to interval
variables—a process which the above schemata only sketch for the sake of clarity.
When several eventualities are in relation with each other (for example in a text),
computational rules determine the relations that are set up between intervals. We
represent such relations, still for the sake of clarity, by duplicating the temporal axis.
For instance, when an eventuality with imperfective aspectual viewpoint provides
the background for a foreground perfective one, as in example (11) and Figure 5.12,
we take the latter’s interval (interval of the eventuality) to be established as identical
with the interval of reference associated with the former.28
(11) Il pleuvait. Pierre prit son imperméable.
It rain-IMP. Pierre take-SP his raincoat.
‘It was raining. Pierre took his raincoat.’

B1 I II B2 01 02

B1 B2

FIGURE 5.12 Imperfective background.

We account for the (at least apparent) contextual polysemy of temporal and
aspectual markers by a notion of conflict resolution: the instructions encoded by

For more on background relations and the imperfective aspect, see Weinrich (1973); Reinhart (1986);
Kamp and Rohrer (1983); Asher and Bras (1993).
The interaction of time and modality in French 109

the items may occasionally conflict, either with one another, or with direct compos-
itional rules. These conflicts prompt the regular and predictable conflict-resolution
mechanisms, which, we suggest, have to do with a higher level of compositionality.
A typical case in point is the combination of items deprived of frequentative
meaning that nevertheless lead to an frequentative interpretation. This occurs in
example (12), displaying a famous sentence by Proust (the beginning of his Remem-
brance of Things Past):
(12) Longtemps, je me suis couché de bonne heure.
For a long time, I go-CP to bed early.
‘For a long time, I used to go to bed early.’
The frequentative interpretation of (12) (made explicit in the English translation with
used to) arises despite the fact that no frequentative meaning is associated with the
linguistic items themselves. We suggest that the frequentative interpretation pre-
cisely obtains through a conflict-resolution mechanism. In fact, neither the French
passé composé (roughly a present perfect) nor the verb phrase (se coucher de bonne
heure) bear frequentative meanings. The reading of (13), without the duration adverb
longtemps (‘for a long time’), leaves no doubt that the interpretation is not frequen-
tative but singulative; thus iteration has to do with the insertion of longtemps.
(13) Je me suis couché de bonne heure.
I go-PP to bed early.
‘I went to bed early.’
However, it has to be noted that the adverb longtemps it is not a frequentative
marker either, as the singulative meaning of (14) illustrates:
(14) Longtemps, je t’ai cru mort, perdu! Assassiné! (Villiers de l’Isle-Adam, Le
Prétendant IV, 5)
‘For a long time, I thought you were dead, lost! Assassinated!’
At this point, two options are available. Either (i) we postulate the presence of
an ‘invisible’ (Vlach 1981: 71) or ‘silent’ (van Geenhoven 2004, 2005) frequentative
operator within these structures that is endowed with a semantic content yet
deprived of a morphological and phonetic realization, or (ii) we consider iteration
to be a value that emerges from the interaction between various constitutive markers
in the utterance. The first solution is obviously unfalsifiable, if not circular. The
second type of analysis (adopted by, among others, Vet 1981, Kleiber 1987, and
Moens and Steedman 1988) considers iteration to be a means of resolving the conflict
between discordant elements. In this example, it is the conflict between the duration
adverb longtemps and the isolated eventuality that is resolved by the construction of
an iterative series: the adverb no longer scopes over the eventuality, but, instead, over
the series of eventualities. In such cases, we assume, compositionality is not ‘direct’
110 Laurent Gosselin

(Barker and Jacobson 2007) but ‘holistic’ (Gosselin 1996, 2005), ‘Gestaltist’ (Victorri
and Fuchs 1996), or ‘interactionist’ (Recanati 2004).

5.5 Modal values of aspectual viewpoints

Rephrased in the framework of the ISM model, Vet’s hypothesis (Vet 1981) about the
modal asymmetry of time (saying that the modal change occurs at R) takes the
following form: in the indicative, it is the final bound of the reference interval (II)
which brings about the modal break between the irrevocable and the possible.
A simple test is associated with this hypothesis: if the right-bound context may
indicate that the eventuality did not occur (or wasn’t completed), then the eventu-
ality belongs entirely (or partially if it wasn’t completed) to the realm of the mere
With regard to the types of aspectual viewpoints identified above, this leads to the
following predictions, which relate to what we shall henceforth call ‘aspectual modal
(a) With perfective (il traversa la route ‘he cross-SP the road’) and perfect (il a
traversé la route depuis 5 minutes ‘he cross-CP the road five minutes ago’)
viewpoints, the eventualities are entirely situated within the irrevocable, as in
Figures 5.13 and 5.14, respectively:

B1 B2

FIGURE 5.13 Aspectual modal values with perfective viewpoint.

B1 B2 I II

FIGURE 5.14 Aspectual modal values with perfect viewpoint.

(b) With the imperfective viewpoint (Figure 5.15), the end of the eventuality is
situated in the realm of the possible; hence, with telic eventualities, the
occurrence of the well-known ‘imperfective paradox’29 (it remains unknown,
unless there is further indication, whether the eventuality was completed or
not: il traversait la route (quand soudain . . . ) ‘he was crossing the road (when
suddenly . . . )’):

See Dowty (1979), and, for a recent restatement, Portner (2009: 243–5).
The interaction of time and modality in French 111

B1 I II B2

FIGURE 5.15 Aspectual modal values with imperfective viewpoint.

With atelic eventualities, the same configuration simply leads to considering as

possible only the fact that the imperfective eventuality, taken as a background,
may continue after the completion of the foreground eventuality. For example,
with Les élèves parlaient quand le proviseur entra ‘the students were talking when
the principal entered the room’, whether the students continued to talk remains
unknown in the absence of further indication, as is represented in Figure 5.16.30

B1 I II B2

B1 B2

FIGURE 5.16 Aspectual modal values with imperfective viewpoint.

(c) The prospective viewpoint leaves the entire eventuality within the realm of the
possible (il allait traverser la route (quand soudain . . . ) ‘he was going to cross
the road (when suddenly . . . )’), as in Figure 5.17.

I II B1 B2

FIGURE 5.17 Aspectual modal values with prospective viewpoint.

5.6 Relations between temporal modal values and aspectual modal values
We posited earlier that semantic time is relatively independent of pragmatic time.
However, we suggest that specific meanings arise in the confrontation of these
two temporalities, which are both asymmetrical but for which the modal break-
point occurs at different times. We shall speak of ‘temporal modal values’
when the representation is built upon the speech time (02), corresponding to
The context can imply that the imperfective event is interrupted by the front-stage event (Ils
réveillèrent ceux qui dormaient ‘They woke those who were asleep’) or, conversely—and especially when
interpreted in the discourse as expressing perception (see Bres 2003; De Mulder 2003)—that the imper-
fective event continues beyond the modal breakpoint (e.g. Il s’approcha de la fenêtre. Il neigeait abondam-
ment. ‘He approached the window. It was snowing abundantly.’).
112 Laurent Gosselin

pragmatic time, and of ‘aspectual modal values’, as above, when the representation is
built upon the reference time (II), corresponding to the semantic time. Divergences
between these two types of representations appear in two kinds of situations:
(a) First, the speaker presents a future eventuality as determinate at the semantic
level, that is, pertaining to the possible at the pragmatic level.
(b) Second, the speaker presents a past eventuality, pertaining to the irrevocable
at the pragmatic level, as indeterminate (either partially or totally).
The first case typically corresponds to the simple future, as in (15) and (16), and
Figure 5.18:
(15) La cavalerie attaquera à l’aube (Fauconnier 1979: 13)
‘The cavalry will attack at dawn.’
(16) Et Lady Helena ? Miss Grant ?
Je ne les préviendrai qu’au dernier moment, lorsque tout espoir sera perdu de
tenir la mer. Vous m’avertirez.
Je vous avertirai, mylord. (J. Verne, Les enfants du capitaine Grant, Le Livre de
Poche, 2004: 380–1)
‘And Lady Helena? Miss Grant?’
‘I will only tell them at the last moment, when all hope to stay at sea is lost.
You will warn me.’
‘I will warn you, milord.’

B1 B2
01 I II
modal values
modal values
FIGURE 5.18 Futur simple (simple future).

This clash is usually resolved by the utterance bringing about particular commitment
on the part of the speaker, whether an epistemic one (of certainty) or a pragmatic
one to do something or to make someone do something (as long as the social
situation warrants it). Contrary to what is often said,31 the epistemic value does
not specifically belong to the future tense’s linguistic meaning (otherwise it would
always be attached to it): it is merely inferred starting from the divergence between
semantic and pragmatic time and with regard to contextual factors. Thus, in (15),
depending on whether the speaker is Julius Caesar or a simple observer, the future

See Riegel et al. (1994: 312); Touratier (1996: 176–7).
The interaction of time and modality in French 113

will take on an injunctive or predictive illocutionary force. Likewise, in example (16),

the future alternatively marks the speaker’s commitment to act (je les préviendrai ‘I
will tell them; je vous avertirai ‘I will warn you’), belief (tout espoir sera perdu ‘when
all hope . . . is lost’), and an order (vous m’avertirez ‘You will warn me’). The case of
the ‘prophetic’ meaning with the future tense (cf. example (3) above) is different,
in so far as the prophet and those who consider him such admit that the future is
(pre)determined and non-modifiable, so that there is a convergence, not a diver-
gence, between pragmatic time and semantic time.
The second case, which consists in presenting past events (thus temporally irrevoc-
able) as (partially or totally) possible, is typical of narrative texts. It illustrates the capacity
of the language to represent (that is, rendering as present, to simulate presence)32 past
eventualities, by attributing a characteristic of the present—bringing about the modal
breakpoint in pragmatic time—to a past moment (the Reference point). This device is
particularly used in popular fiction to create suspense, as in (17):
(17) Antoine Laho allait lâcher l’homme qu’il tenait vaincu sous son genou et se relever.
(P. Féval fils, Les chevauchées de Lagardère, Presses de la Cité, 1991: 964)
‘Antoine Laho was going to let go the man he’d pinned down with his knee and
get back on his feet.’
While the eventualities are portrayed in the past, they are presented as possible
because of the prospective aspect which situates the Reference point and thus the
modal breakpoint before the eventuality-intervals. This is why what follows the
utterance may indicate, without any incoherence, that they have in fact not occurred,
as (18) illustrates:
(18) (immediately follows (17)): Il n’en eut pas le temps et tomba la face contre
terre avec un gémissement sourd. Un long couteau catalan était planté entre
ses deux épaules !
‘He didn’t have the time [to do this] and fell face first with a muffled moan.
A long Catalan knife was planted between his shoulders!’
While reading, the reader can thus feel mental states that are normally oriented towards
the future, such as expectation, hope, or fear, despite them being about past events.
We will now confront these general principles in more complex structures in
which certain eventuality-phases are selected. But it is necessary to explain before-
hand how this selection of phases functions in French.

5.7 Phasal aspect

Each eventuality (state or event) is effectively decomposable into five phases:33 three
phases (initial, median, and final) constitutive of ‘internal’ aspect, and two peripheral

32 33
See Gosselin (2005: 11–28). See Dik (1989); Vet (2002); Tournadre (2004: 23).
114 Laurent Gosselin

phases (preparatory and resultant phases) corresponding to the ‘external’ aspect

(Borillo 2005: 67). This phasal structure concerns the eventuality once grasped, and
should not be confused with the internal pre-conceptual structure of changes and
intermediate situations directly related to aspectual categories (states, activities,
achievements, or accomplishments) as such.34 A state, for instance, corresponds to
an absence of change (it is not dynamic) and is thus adequately represented with
the absence of bounds. But what concerns us here is that, once a particular state is
conceptualized in a predicate (for example be ill), it becomes virtually decompos-
able into five phases according to the general mereotopologic35 structure, as in
Figure 5.19.

B1 B2

Phases: preparatory initial median final resultant

internal aspect

external aspect
FIGURE 5.19 The phasal structure of eventualities.

Phases can be picked up through various means. (19)–(21) give examples of phasal
selection by aspectual periphrases:
(19) Il était sur le point d’être malade. (preparatory phase)
‘He was about to be ill.’
(20) Il commençait à être malade. (initial phase)
‘He was starting to become ill.’
(21) Il venait d’être malade. (resultant phase)
‘He’d just been ill.’
The same holds of punctual eventualities, since even if it is true that a punctual
eventuality does not allow for the observation of its internal aspect, all forms of
dilation of these eventualities demonstrate their virtual decomposition into three
internal phases.36
Traditionally, two main ways of selecting the phases of an eventuality in French
are considered: verbal inflection and aspectual periphrases (typically comprising a

See Moens and Steedman (1988); Kamp and Reyle (1993: 558ff.); Gosselin (1996: 50ff.); or Croft (2011).
This aspectual phasal structure varies with each aspectual category.
Mereotopology deals with the disposition of the parts within a whole (see Casati and Varzi 1999).
Victor Hugo especially favours such grossisements fantastiques ‘fantastic zooms‘ (see Gosselin 1996:
68–9 for a full page example we cannot reproduce here).
The interaction of time and modality in French 115

semi-auxiliary37 followed by an infinitive verb or a participle).38 Yet these aspectual

periphrases do not form a homogeneous class (François 1993, 2003). Gosselin (2010b,
2010c, 2011) posits that two classes of French aspectual periphrases must be distin-
guished for their syntactic and semantic behaviour, as specified in what follows.
It is widely acknowledged that the general category of aspect subsumes two distinct
operations: a categorization operation that constructs entities in time (‘conceptual
aspect’) and a operation of ‘monstration’ (‘aspectual viewpoint’) which makes visible
all or part of the entities constructed beforehand by the categorization.39 The categor-
ization constructs eventualities—noted [B1,B2]—while ‘monstration’ determines an
aspectual viewpoint, defined by the position of the Reference interval’s [I,II], relative to
that of the eventuality. We hypothesize that periphrases of the first type select
eventuality-phases through categorization; that is they extract subparts of eventualities
as referents and involve semi-auxiliaries which we label ‘V-MAP’, bearing meanings of
motion,40 mode of action,41 and phase, such as cesser de (roughly ‘stop to’), s’apprêter
à (roughly ‘getting ready to’), partir (‘leave’), or rentrer de X (roughly ‘getting back
from X’). A second type of periphrase, which we label ‘aux-AV’, introduces an
aspectual viewpoint with recourse to aspectual viewpoint auxiliaries; that is, these
define an interval of reference which marks off a visible zone of the eventualities (or a
subpart of them) constructed beforehand. Of this type are periphrastic constructions
such as venir de P (roughly ‘having just P’), être sur le point de (roughly ‘be about to’),
or être en train de (roughly equivalent to progressive forms in English).
With these distinctions and tools, a whole array of converging syntactic and
semantic regularities can be explained; in particular: (a) that aux-AV can scope
over V-MAP but not the reverse; (b) that only V-MAPs are recursive; (c) that only
V-MAPs can be focused and thus occur naturally at the end of an utterance; and
(d) that sub-eventualities (constructed by V-MAPs) can be specified with adverbs of
manner. We elaborate on these regularities below.

Or ‘coverb’—a term used by Roy (1976), Wilmet (1997), and Kronning (2003).
To which certain prefixes can be added, like en considered inchoative with s’endormir or s’envoler
(Martin 1971: 53), as well as ‘aspectual support verbs’ (see Borillo 2006), which also allow for the selection
of an eventuality’s phase (e.g. être en larmes ! fondre en larmes, ‘be in tears ! burst into tears’; faire une
recherche ! entamer/poursuivre une recherche ‘do research ! begin/continue research’). On the other
hand, the presence (under certain conditions) of a temporal adverbial normally incompatible with the
intrinsic structure of the eventuality may bring about conflict resolution consisting of keeping one of the
latter’s phases (the one compatible with the adverbial; see Gosselin 1996: 56–63). Thus the presence of a
duration adverbial scoping over an inherently punctual eventuality can lead to a shift towards the
preparatory phase (e.g. il s’arrêta en dix secondes ‘he stopped in ten seconds’) or towards the eventuality’s
resultant phase (e.g. il s’arrêta pendant dix secondes ‘he stopped for ten seconds’; see Borillo 1986: 138–9).
See Gosselin (1996, 2005). This type of distinction presents affinities not only with the Guillaumian
aspectual tradition (Guillaume 1929) and with the cognitive perspective (Langacker 1987, 1991; Col and Victorri
2007), but also with approaches in formal semantics (see Smith 1991; Caudal and Vetters 2006; Klein 1994;
Demirdache and Uribe-Etxebarria 2002; Laca 2005) and computational semantics (Battistelli 2009).
For arguments in favour of the analysis of motion verbs in French as semi-auxiliaries, see Lamiroy
(1983); Gross (1986); Gosselin (2010b).
See François (2003: Chapter 5).
116 Laurent Gosselin

(a) That aux-AV can have scope over the V-MAP, while the converse is false
(François 2003: Chapter 5; Laca 2005), is easy to explain. Monstration can
scope over a (sub-)eventuality, but one does not see what could count as a
categorization that might affect monstration, given that categorizations, concep-
tually, have to occur before monstration. (22)–(23) give some examples of French
periphrases and their behaviour in this respect.
(22) aux-AVs taking scope over V-MAPs:
venir de ‘just having VP’ cesser de ‘quit’ þ infinitive
être sur le point de ‘be about to . . . ’ finir de ‘finish’
être en train de (progressive) commencer à ‘start’
s’apprêter à ‘get ready to’
partir ‘go’
rentrer de ‘get back from’
(23) V-MAPs (cannot scope over aux-AVs):42
cesser de venir de þ infinitive
finir de être sur le point de
commencer à être en train de
s’apprêter à
rentrer de

(b) Only V-MAPs are recursive (see Laca 2005) insofar as for every V-MAP there
exists at least one subclass of V-MAP capable of scoping over it, while an aux-
AV, to the contrary, can never be under the scope of another aux-AV. This is
because we can conceptually divide a sub-eventuality within another sub-even-
tuality, while we cannot conceive of what a ‘viewpoint holding scope over
another viewpoint’ could look like. (24)–(25) provide examples.
(24) V-MAPs scoping over V-MAPs:
cesser de ‘quit’ s’apprêter à þ infinitive
commencer à ‘start’ hésiter à
continuer de ‘continue to’
s’apprêter à ‘get ready to’ cesser de þ infinitive
hésiter à ‘hesitate to’ commencer à
continuer de

For translations see (22). Let us point out that when a V-MAP has scope over aller/venir de þ
infinitve, the latter are necessarily interpreted as motion verbs (unlike the aux-AVs): “Au moment où
Doremus se disposait à aller remettre sa missive au courrier [ . . . ], Geroul entra dans la pièce.” ‘Just when
Doremus was getting ready to give his letter to the courier, Geroul came in the room.’ (M. Paillet, Les
noyées du grau de Narbonne, 10/18: 13).
The interaction of time and modality in French 117

tarder à ‘put off ’

rentrer de þ infinitive
se hâter de ‘hurry to’
finir de
(25) aux-AVs that cannot scope over aux-AVs:
#43 venir de être sur le point de þ infinitive
être en train de
*être sur le point de venir de þ infinitive
être en train de
(c) Only a sub-eventuality, as opposed to an aspectual viewpoint, has the referential
status that allows focus (without particular intonational or contextual contrast-
ive marking; cf. Nølke 2001: 137). This explains that V-MAPs alone, as opposed
to aux-AVs, may naturally occur in an utterance-final position (subsequent to
either a pronominalization or an ellipsis of the infinitive; see Kronning 2003:
237), as in examples (26)–(27).
(26) V-MAPs focused:
Il commence ‘He’ ‘is starting’
continue ‘is continuing’
finit ‘is finishing’
cesse ‘is stopping’
hésite ‘is hesitating’
s’y met ‘is getting to it’
s’y acharne ‘is working hard to’
s’y apprête ‘is getting ready to’
s’y précipite ‘is hurrying to’
en revient ‘is (just) getting back’
(27) aux-AVs, unavailable for focus:
?*Il (en) est sur le point ‘is about to’
(en) est en train [progressive/dynamic aspect marking (‘being in
the course of’)]
(en) est en passe ‘is about to’
#Il y va ‘is going’
en vient.44 ‘is (just) coming from (there)’
(d) The sub-eventualities (constructed by the V-MAPs), in contrast to aspectual
viewpoints, may happen in various ways, which can be expressed by adverb

This type of construction is possible (Havu 2006 gives an example from Stendhal) though, as long as
venir de is interpreted as temporal and not as aspectual; see Section 5.8 below.
Indeed it does not seem possible to utter il y va/il en vient without it implying some form of
movement (and thus functioning as V-MAPs), as shown by the impossibility of repeating il y va/en vient
after Il va/vient de pleuvoir ‘It’s going to rain/has just rained’.
118 Laurent Gosselin

phrases indicating manner (see Damourette and Pichon 1911–1940, V, §1605;

Kronning 2003: 241–2) as in (28)–(29).
(28) V-MAPs with adverbial complement:
Il commence avec précaution à þ infinitive
continue ‘with precaution to’
Il rentre précipitamment (de/à) þ infinitive
part ‘hurriedly (from/to)’
(29) aux-AVs incompatible with adverbial complements:
Il est en train avec précaution de þ infinitive
est sur le point
#Il vient avec précaution (de) þ infinitive45
We thus confirm the existence of two distinct classes of aspectual periphrases that
allow the selection of an eventuality’s phases via different routes: (i) categorization
and construction of a sub-eventuality versus (ii) an aspectual viewpoint. Here is a
(non-exhaustive) list of French verbal periphrases according to this classification:
aux-AVs: aller, être sur le point / en passe / en voie / en train de, (en) être à, se
prendre à, venir de46
V-MAPs: être prêt à, s’apprêter à, se préparer à, se disposer à, se proposer de, hésiter
à, tarder à, partir, s’installer à, aller, s’arrêter à, s’acharner à, s’efforcer de, se
dépêcher de, mettre du temps à, être long à, commencer à, se mettre à, continuer
à/de, être occupé à, persévérer à, cesser de, (s’)arrêter de, finir de, (re)venir de,
rentrer de, (re)descendre de, sortir de, etc.
The recursiveness of V-MAPs, which follows from the possibility of constructing—
via categorization—a sub-eventuality from another sub-eventuality, also testifies to
the fact that the phasal structure (illustrated in Figure 5.17) holds equally for
eventualities and for parts of eventualities, as well as for parts of parts, as long as
they are categorized as sub-eventualities.

5.8 Classification and representation of aspectual periphrases

We can also classify, from a semantic perspective, the aux-AVs and V-MAPs by the type
of phase they allow the selection of (according to aspectual viewpoint or categorization of
sub-eventualities, respectively) as is summarized in Tables 5.1 and 5.2.

These last examples would, of course, be acceptable if aller and venir were used as motion verbs.
This is but one of the usages of aller and venir.
The interaction of time and modality in French 119

Table 5.1 Classification of the aspectual viewpoint auxiliaries (aux-AV)

phase aspectual viewpoint aux-AV

preparatory prospective aller

être sur le point de
être en passe / voie de
internal aspect, unbounded imperfective être en train de
(en) être à
aller þ present participle
initial inchoative se prendre à
resultant perfect être/avoir þ past participle
venir de

Table 5.2 Semantic classification of V-MAPs47

phase phasal V V of mode of action motion V

preparatory être prêt à se proposer de partir

s’apprêter à hésiter à s’installer à
se préparer à tarder à aller
se disposer à s’arrêter à
internal aspect s’acharner à
s’efforcer de
se dépêcher de
mettre du temps à
être long à
initial commencer à
se mettre à
median continuer à / de
être occupé à
persévérer à
final cesser de
(s’)arrêter de
finir de
resultant (re)venir de
rentrer de
(re)descendre de
sortir de

This list is far from exhaustive; see Wilmet (1997: 79).
120 Laurent Gosselin

This classification calls for several remarks:

(a) The distinction between verbs indicating phase, mode of action, and motion is
not of aspectual nature: the first indicates only the concerned phase, the others
also concern the way they happen.
(b) A given element can represent various usages in context which relate to different
subclasses. This is true, for example, of the verbs aller ‘go’ and venir ‘come’,
which have given rise to an abundant literature (for the periphrastic future, see
especially Vet 1993 and 2001). It appears that these two verbs can function,
among other things, as:
– full verbs (e.g. aller à / venir de Marseille),
– motion verbs (e.g. aller / venir (en voiture) (de) faire les courses),
– temporal auxiliaries with a future meaning (Dans cinq minutes, on va bien
rigoler; example quoted by Sundell 1991 and commented on by Vet 1993: 79)
or a past value (Nous venons récemment de perdre encore une sœur charmante;
Lamartine quoted by Gougenheim 1929: 127),
– aspectual viewpoint auxiliaries with prospective (Je vois que Pierre va se fâcher;
Vet 1993: 74) or perfect meanings (Chatterton venait d’expirer depuis peu de jours
lorsque parurent à la fois un poème burlesque et un pamphlet sur sa mort; Vigny
quoted by Gougenheim 1929: 127; the adverbial [depuis ‘since’ þ duration]
indicates precisely this type of aspectual viewpoint; see Gosselin 1996: 27).48
When there is a succession of two occurrences of aller in preverbal position (tu vas
aller faire les courses), the former works as a temporal auxiliary or an aspectual
viewpoint auxiliary (depending on context), while the latter can only be interpreted
as a motion verb (e.g. Je vais aller me promener).
(c) The V-MAPs may mark:
– the co-occurrence of a sub-eventuality with an eventuality-phase (e.g. com-
mencer à, continuer à, finir de);
– the inclusion of a sub-eventuality inside a phase (e.g. s’apprêter à, hésiter
à . . . ); for instance, rentrer de in the utterance rentrer de faire les courses
marks the inclusion of the sub-eventuality in the resultant phase and not its
co-occurrence with it, so that a prospective viewpoint on this sub-eventuality
does not necessarily entail that the preceding phase still takes place. As a
result, we notice that from (30), one cannot infer (31):

About the two values of venir de: Havu (2005) gives numerous attested examples which show that
venir de sometimes combines with a past-time adverbial, and sometimes with ‘depuis þ duration’,
marking perfection; see also Vetters (2010).
The interaction of time and modality in French 121

(30) Paul était sur le point de rentrer de faire ses courses.

‘Paul was about to come back from grocery shopping.’
(31) Paul était encore en train de (finir de) faire ses courses.
‘Paul was still (finishing his) grocery shopping.’
In effect, it may be the case with (30) that Paul is already in the resultant phase
(i.e. that he has finished grocery shopping, but has not begun going back home);
– an overlap, possibly partial, of several phases (i.e. se hâter de, s’acharner à,
s’efforcer de . . . ).
(d) The verbs indicating a mode of action which are about the internal aspect (e.g.
s’acharner à, se hâter de, se dépêcher de, être long à . . . ) are used to express the
preparatory phase of the eventuality if the latter is punctual (i.e. if the internal
phases are not accessible). We show in Gosselin (1996: 172–6) that this mode of
conflict resolution—by shifting to the preparatory phase—is regular and predict-
able. This explains, among other things, why se dépêcher de becomes equivalent to
ne pas perdre de temps avant de and why être long à is a quasi-synonym of tarder à
when the eventuality is punctual, as in (32)–(33).
(32) Il s’est dépêché de sortir  il n’a pas perdu de temps avant de sortir.
‘He hurried to go out  he didn’t waste time before going out.’
(33) Il est long à sortir  il tarde à sortir.
‘He’s slow coming out  he’s taking his time coming out.’
The V-MAPs select specific phases in constructing sub-eventualities, noted [B’1,
B’2], [B’’1, B’’2], etc. As an example, we may consider the schematic representation
associated with hésiter à commencer à manger ‘hesitate to start eating’ in Figure 5.20.
For clarity we duplicate the temporal axes.

B1 B2

B1 B2

B1 B2 [B1,B2]: manger

[B1,B2]: commencer
[B1,B2]: hésiter

FIGURE 5.20 Hésiter à commencer à manger ‘hesitate to begin eating’.

The sub-eventuality marked by commencer à coincides with the initial phase

of the eventuality manger, while that expressed by hésiter à is included in the
122 Laurent Gosselin

preparatory phase of the sub-eventuality which also corresponds to the preparatory

phase of manger. This structure thus illustrates both the fact that hésiter à commen-
cer à manger is quasi-synonymous with hésiter à manger, and that avoir hésité à
commencer à manger ‘having hesitated to start eating’ does not imply commencer à
manger ‘start to eat’ in so far as B’’2 does not necessarily coincide with B’1 (i.e. the
resultant phase of the sub-eventuality hésiter can begin before the beginning of
the sub-eventuality commencer, which coincides in turn with the beginning of the
eventuality manger).
The aux-AVs, in relation to inflection, serve to indicate the position of the
reference/monstration interval relatively to that of the eventuality, thus defining a
type of aspectual viewpoint. With the utterance in (34) we associate the aspectual
structure in Figure 5.21:

B1 I II B2

FIGURE 5.21 Example (34).

(34) Il était en train de rentrer chez lui.

‘He was going (was in the course of going) back home.’
Sub-eventualities (indicated by V-MAPs) can be combined with aspectual view-
points (indicated by the aux-AVs and/or verbal inflection). The aspectual view-
point corresponds to the position occupied by the reference interval ([I,II]) that is
positioned both in relation to the most embedded sub-eventuality (i.e. the one
corresponding to the most fine-grained stage of decomposition of the (sub-)
eventualities), which we call the direct aspectual viewpoint, and in relation to
the other eventualities and sub-eventualities within the structure, which we call
indirect aspectual viewpoint. Thus, in the utterance in (35), the direct viewpoint,
marked by the French imparfait, is imperfective on the preparatory phase in-
stantiated by se disposer à (B’1 < I < II < B’2), but is prospective on the
eventuality (II < B1).
(35) Au moment où il se disposait à frapper deux coups discrets à la petite porte,
elle s’ouvrit . . . (Ponson du Terrail, Une fille d’Espagne, R. Laffont, 1992: 264)
‘At the instant he was readying himself to discreetly knock twice at the small
door, it opened.’
Figure 5.22 accounts for this:
The interaction of time and modality in French 123

B1 B2

B1 I II B2
[B1,B2]: frapper deux coups ...
[B1,B2]: se disposer à
[B1,B2] / [I,II]: imperfective viewpoint (direct)
[B1,B2] / [I,II]: prospective viewpoint (indirect)

FIGURE 5.22 Example (35).

Likewise, in example (36), the perfect viewpoint, expressed in this context by the past
perfect (plus-que-parfait), targets the sub-eventuality se mettre à ‘begin’, which picks
up the initial phase of the eventuality descendre ‘descend’; thus the direct aspectual
viewpoint is a perfect viewpoint.
(36) . . . un roulement ininterrompu qui venait d’en haut leur fit lever les yeux vers
le plafond. Leurs cheveux se hérissèrent . . . Le plafond s’était mis à descendre
! . . . Il descendait tout d’une pièce, d’un mouvement très lent, mais continu.
(M. Zevaco, L’épopée d’amour, R. Laffont, 1988: 982)
‘ . . . an uninterrupted rolling sound that was coming from upstairs made them
lift their eyes towards the ceiling. Their hair stood on end . . . The ceiling had
begun to descend! It came down as one piece, very slowly, but continuously.
But the indirect viewpoint is imperfective (hence the repetition of descendre in ‘il
descendait’ with the imperfective imparfait), as illustrated in Figure 5.23.

B1 B2

B1 B2 I II [B1,B2]: descendre

[B1,B2]: se mettre à
[B1,B2] / [I,II]: perfective viewpoint
[B1,B2] / [I,II]: imperfective viewpoint
FIGURE 5.23 Example (36).

5.9 Modal interpretation of phasal aspect

The hypothesis according to which the final bound of the reference interval creates—on
the semantic level—a modal breakpoint between the aspectual modal values of
124 Laurent Gosselin

the irrevocable and the possible, also applies to indirect aspectual viewpoints (with a
If the indirect viewpoint on the eventuality is prospective, then the eventuality is
entirely situated within the realm of the possible, whatever the direct viewpoints of
the sub-eventualities (marked by the V-MAP). As a consequence, the right-hand
context may invalidate the eventuality’s achievement (see example (35) above).
When the indirect viewpoint is imperfective (on the eventuality), the initial bound
belongs to the irrevocable, and the final bound to the possible, which implies that
telic eventualities are (at least provisionally) indeterminate as regards their actual
completion. This indirect imperfective viewpoint may result from diverse configur-
ations, such as:
– a direct viewpoint, imperfective or perfective, on a sub-eventuality corre-
sponding to the median phase, e.g. il continuait/continua de faire la vaisselle
‘he continue-IMP/SP to wash the dishes’;
– an imperfective or perfect viewpoint on the initial phase, e.g. il commençait/
avait commencé à faire la vaisselle ‘he begin-IMP/PP . . . ’;
– a prospective or imperfective viewpoint on the final phase, e.g. il allait finir/
finissait de faire la vaisselle ‘he was about to finish/finish-IMP’.
Take, for instance, the following example of an imperfective viewpoint on the initial
phase, with an interruption of the eventuality:
(37) La marquise se replaça auprès du feu, et vers minuit le sommeil commençait à
la gagner, lorsque la porte s’ouvrit tout à coup . . . (Ponson du Terrail, Une fille
d’Espagne, R. Laffont, 1992: 317)
The marquise sat near the fire, and around midnight sleep begin-IMP to
overcome her, when the door suddenly opened.
It is noteworthy that with verbs of initial phase, this configuration, comparable to the
‘imperfective paradox’, occurs even when the initial phase is viewed from a perfect-
ive perspective, since the final bound of the eventuality is not accessible (which
shows, by the way, that the imperfective paradox is not directly tied to progressive
aspect, contrary to what is usually considered), as example (38) and Figure 5.24
(38) Arrivé chez lui, Paul se mit à faire la vaisselle, mais il ne la termina pas.
(Vetters 2003: 123)
‘Once he had arrived home, Paul began to wash the dishes, but didn’t finish.’
When the indirect aspectual viewpoint is ‘terminative’ (i.e. perfective on the final
phase) or perfect, the eventuality is entirely situated within the irrevocable; thus its
occurrence and completion cannot be cancelled, e.g. il acheva de faire les courses/il
The interaction of time and modality in French 125

B1 B2

B1 B2 [B1,B2]: faire la vaiselle
[B1,B2]: se mettre à
[B1,B2] / [I,II]: perfective viewpoint( direct)
[B1,B2] / [I,II]: inchoative viewpoint (indirect)
FIGURE 5.24 Representation of example (38).

rentra de faire les courses ‘he achieve-SP/come-SP back from shopping for groceries’.
On the other hand, if the direct viewpoint on the sub-eventuality is imperfective on
the final phase, again an ‘imperfective paradox’ situation arises, since the indirect
viewpoint on the eventuality is then imperfective, as in example (39) and Figure 5.25.
(39) Il était en train de finir de manger son gâteau (quand soudain . . . ).
He be-IMP in the course of finish eat his cake (when suddenly . . . ).
‘He was finishing eating his cake (when suddenly . . . ).’

B1 B2

[B1,B2]: manger son gâteau

[B1,B2]: finir de B1 I II B2
[B1,B2] / [I,II]: imperfective viewpoint (direct)
[B1,B2] / [I,II]: terminative viewpoint(indirect)

FIGURE 5.25 Representation of example (39).

Let us illustrate this with a final example, (40), taken from journalistic discourse:
(40) Hosni Moubarak venait juste d’arriver à Addis-Abeba hier matin, où il
s’apprêtait à participer au 31e sommet de l’Organisation de l’unité africaine
(OUA). Son cortège traversait les rues de la capitale éthiopienne en direction
du palais des conférences . . . (Libé 27 June 1995)
‘Hosni Mubarak had just arrived in Addis Ababa yesterday morning, where he
was getting ready to participate in the 31st summit of the Organization of
African Unity (OAU). His convoy was crossing through the streets of the
Ethiopian capital toward the conference centre . . . ’
The journalist chose a past reference point (situated by hier matin) to show, through
the use of various aspectual viewpoints, what was happening at that moment, as if
126 Laurent Gosselin

the scene were occurring before the reader’s eyes. All the elements preceding the
final bound of that reference point have irrevocably occurred, while those which
follow are merely possible. Thus arriver à Addis-Abeba is presented as perfect (by the
aspectual auxiliary venir de), thus as having occurred. Then participer au sommet de
l’OUA is viewed prospectively (because of the verb s’apprêter à), and thus as possible.
On the contrary, traverser les rues de la capitale . . . is presented by means of an
imperfective viewpoint (marked by the imparfait), which situates only the beginning
of the eventuality within the irrevocable, its end (the actual arrival at the conference
centre) remaining only possible. The subsequent passage, in (41), indicates precisely
that these possibilities did not occur:
(41) . . . lorsqu’une camionnette est venue se mettre en travers de la route au
niveau de la mission diplomatique palestinienne. Des coups de feu ont éclaté . . .
Le cortège a immédiatement rebroussé chemin et Hosni Moubarak a repris
l’avion pour Le Caire . . . .
‘ . . . when a pick-up truck came to block the road near the Palestinian
Diplomatic Mission. Shots rang out . . . The convoy immediately turned
around and Hosni Mubarak took a plane back to Cairo . . . .
It thus becomes apparent that, contrary to what Reyle, Rossdeutscher, and Kamp
(2007: 630) claim, the aspectual viewpoint plays a decisive role at the truth-condi-
tional level, despite the fact that it has no referential status. More precisely, if an
eventuality occurred at a certain phase, the preceding phases in the phasal structure
necessarily occurred too, whereas those that follow remain open. As a result, the
assertion of an eventuality phase entails the presupposition(s) of the preceding
phases.49 Since in assertive utterances the reference interval is equivalent to the
assertion interval (Klein’s 1994 ‘Topic Time’), the phases on the left-hand side of this
interval are presupposed. Thus from (42) we legitimately infer (43), but not (44):
(42) Pierre a continué de manger son gâteau.
‘Pierre continued eating his cake.’
(43) Pierre a commencé à manger son gâteau.
‘Pierre started eating his cake.’
(44) Pierre a fini de manger son gâteau.
‘Pierre finished eating his cake.’
In this perspective, presuppositions such as Jean used to smoke, conveyed by
utterances like Jean a cessé de fumer ‘Jean quit smoking’, are not due to the specific
properties of the verb cesser ‘quit’ but to general inferences based on the selection of
an eventuality’s terminal phase. The same presupposition would occur with other

For more on the notion of temporal presupposition, see Molendijk (1993); Molendijk and Vet (1995).
The interaction of time and modality in French 127

verbs specifically indicating the terminal phase, such as, in French, arrêter de ‘stop’,
or, with telic eventualities, finir de/achever de/terminer de ‘finish’, ‘achieve’.
This applies to factive constructions as well. A factive presupposition associated
with a phase entails the presupposition of the preceding phases, but not of the
following ones. Utterance (45) entails (42) and (43), but not (44).
(45) Luc ignore que Pierre a continué de manger son gâteau.
‘Luc doesn’t know that Pierre continued to eat his cake.’

5.10 Conclusion
We have attempted to show that the relations between temporality and modality
in French should not be thought of in the perspective of a clear-cut dichotomy. On
the contrary, temporality and modality are two essential, interrelated, dimensions of
the utterance. We suggested that their connection is achieved at the semantic level
by the aspectual viewpoint. Indeed, it is the aspectual viewpoint, expressed mainly by
verbal inflection and aspectual periphrases, which determines the position of the
modal breakpoint that opposes the possible to the irrevocable.
The aspectual viewpoint, defined by the position of the reference interval relative
to that of the eventuality, constitutes the essential element of the representational
semantic–cognitive apparatus which simulates the presentness of a past moment
(see also de Saussure, this volume, for other aspects of temporal simulation with
modal effects), to which is attributed the basic modal property of the present of
hosting the modal breakpoint. In narrative discourse, it is this temporal–modal
organization that enables a reader to have mental states oriented towards the future
(expectations, fear, hope, and so forth), and at the same time tied to past events.
Aspectual viewpoint is no less important for truth-conditional semantics and
inferences; only by taking it into account is it possible to predict legitimate inferences
about the completion of eventualities expressed in utterances.

Modal conversational backgrounds

and evidential bases in predictions:
the view from the Italian modals


6.1 Introduction
This chapter investigates the interaction between reference to future eventualities and
evidentiality via the semantics of a set of modal expressions. During the last two
decades, research in typological and historical linguistics has increasingly considered
the categories of tense, aspect, modality, and evidentiality—the so-called TAME
categories—as one tightly interrelated macro-domain. The meanings traditionally
ascribed to these categories have been tied together by the discovery of cross-
linguistically recurrent grammaticization paths and by the reconstruction of seman-
tic maps aimed at capturing “the sum total of the semantic possibilities of the
category under investigation” (de Haan 2006: 45) in terms of synchronic polysemy
and diachronic semantic change. While the mapping of these polysemic and dia-
chronic connections is not sufficient by itself to provide a semantic or conceptual
analysis of the meanings involved (and even less a fully-fledged language-based
ontology) it does suggest that our reasoning about eventualities in time is deeply
and pervasively intertwined with the way we reason about alternative possibilities
and about evidence.
The interactions between reference to future eventualities and modality have been
an object of interest for philosophy since Antiquity, with a major focus on the
logical–ontological issue of (in)determinism and of the truth conditions of state-
ments about the future inaugurated by Aristotle (De Interpretatione, 9). We perceive
the future as open to different possibilities—certain things, at least, can turn out
otherwise—as opposed to a settled present and past where no real alternative
possibilities are open and the only alternative possibilities that we can envisage are
Modal conversational backgrounds and evidential bases in predictions 129

either counterfactual (abstracting from certain aspects of what actually happened) or

epistemic (relative to our incomplete knowledge of what actually happened). In this
chapter, we will not directly approach this asymmetry between future, present, and
past in our common-sense ontology (cf. Bonomi 1980; Kaufmann, Condoravdi, and
Harizanov 2006). We will look into a distinct, but related, epistemological asymmetry
that concerns the kinds of evidence, reasons, or arguments that may justify taking an
epistemic stance towards a future state of affairs. We do not base our beliefs about
what will be the case on the same kinds of evidence on which we rely for our
knowledge of the present or the past.
To do so we will turn to a study of prediction, a speech act that, according to Searle
and Vanderveken (1985: 186) corresponds to asserting “with the propositional content
condition that the propositional content is future with respect to the time of utterance
and the additional preparatory condition that the speaker has evidence1 in support of
the proposition”. In particular, we will conduct a corpus-based semantic investigation
of the behaviour of the Italian modal verbs potere (‘can’, ‘may’) and dovere (‘must’,
‘have to’) in speech acts of prediction. The main goal is to examine whether and how
these two modal lexemes function as evidential strategies in predictions, manifesting
the kind of evidence the prediction is based on, and also eventually acting as anaphor-
ical or cataphorical pointers towards other discourse utterances where the evidential
basis of the prediction is made explicit. In the latter case the modals function not only
as evidential strategies but also as markers of argumentative discourse relations between
a predictive standpoint and the arguments that support it. This functioning as both a
marker of evidentiality and a marker of argumentative discourse relations has been
observed in the inferential evidential uses of the Italian necessity modal dovere, which
are very similar to those of English must. It has also been attested for other expressions
of epistemic modality in Italian, including the epistemic reading of the Italian future
tense (which is diachronically a modal), as well as for certain epistemic expressions of
possibility such as the impersonal constructions può darsi che / può essere che (‘it may be
that’), or the epistemic adverb forse (‘maybe’, ‘perhaps’), all expressing different nuances
of inferential evidentiality.
(1) La macchina di Giovanni non è nel parcheggio. Dev’essere andato a casa.
John’s car is not in the parking lot. He must-ind-pres-3rd-sing have gone home.
‘John’s car is not in the parking lot. He must have gone home.’
In (1), the presence of the modal deve not only signals that the prejacent—that is the
proposition over which it takes scope—is the result of an inferential process, but also
constrains the interpretation of the immediate co-text, allowing the addressee to
establish an argumentative relation between the utterance in which it appears
and co-textually recoverable evidence. The modal functions as an evidential both

Italics are ours.
130 Andrea Rocci

conceptually (by restricting the source of evidence) and procedurally (by prompting the
addressee to the recovery of evidence of the required kind in the context).2 The thesis
that epistemic dovere—and English must—incorporate an element of inferential evi-
dentiality in their semantics emerges clearly from the fact that they are incompatible
either with direct perceptual evidence (2) or with reportative evidence (3).
(2) *Giovanni dev’essere andato. L’ho visto che usciva.
*‘John must have gone. I saw leaving.’
(3) ?? Giovanni dev’essere andato a casa. Me l’ha detto Laura.
?? ‘John must have gone home. Laura told me so.’
The interpretability of (3) can be saved assuming that the speaker does not trust
Laura completely as a source of information and some further inference is needed to
conclude that John has left. Alternatively, the first sentence in (3) can be read as a
free indirect discourse, where Laura is the subject performing the inference. In both
cases the possibility that the prejacent is accepted on the basis of a simple testimony
is excluded. The fact that dovere functions procedurally as a pointer for the addressee
to recover the evidence from the situational or discourse context can be shown by
comparing dovere or must with other epistemic expressions, such as belief predicates
(I think, I believe, I’m sure that), which are devoid of this procedural element:
(4) a. E’ andato a casa presto. Doveva essere stanco.
He went home early. He must-ind-imperf-3rd-sing be tired.
‘He went home early. He must have been tired.’
b. E’ andato a casa presto. Sono sicuro che era stanco.
‘He went home early. I’m sure he was tired.’
While the modal in (4a) unequivocally points anaphorically to the preceding
utterance as the evidence supporting the conclusion, the belief predicate in (4b) is
more fuzzy: it expresses a subjective state of certainty of the speaker, which might
derive from a variety of sources, and which may or may not include what is observed
in the preceding utterance.
This behaviour of modal verbs as evidential markers fits neatly with a semantic
analysis of the modals as context-dependent relational predicates such as that
proposed by the theory of Relative Modality (RM). The bare bones of this approach,
initiated by Angelika Kratzer (1977, 1981, 2012) are the following. Modals are treated
as relational predicates of the form M (B, p) selecting two arguments: the prejacent p
and a set of propositions, called the conversational background (B). Modal expres-
sions of necessity like dovere can be understood in terms of the logical consequence of
the prejacent from the conversational background, while possibility expressions such
On the distinction between conceptual and procedural encoding in language see the classic paper of
Wilson and Sperber (recently republished as Wilson and Sperber 2012).
Modal conversational backgrounds and evidential bases in predictions 131

as potere ‘can, may’ are to be conceived in terms of the logical compatibility between
the prejacent and the background:
(Def. 1) Dovere ‘must’ (B, p): p is a logical consequence of B (henceforth, symbol-
ically: B & p).
(Def. 2) Potere ‘can’/ ‘may’ (B, p): p is logically compatible with B (symbolically:
B ◊ p).
One advantage of an RM approach is that the variety of modal ‘flavours’ results from
an invariant modal force and a variety of conversational backgrounds reconstructed
in the context of utterance.
In accounting for modal meanings it is useful to distinguish between a few basic
kinds of propositions that may enter modal conversational backgrounds. For the
purposes of this chapter I will use a tripartite distinction between alethic, deontic,
and epistemic conversational backgrounds.3 Alethic conversational backgrounds are
composed of propositions that are facts of a certain kind. These can range from the
basic ontology of the universe, both metaphysical and physical, to very specific sets
circumstances (circumstantial conversational backgrounds). As we will see in the
following sections, social reality can be treated much in the same way as physical
reality and backgrounds including institutional facts or economic laws function
largely in the same way as an alethic background.
Deontic conversational backgrounds are composed of propositions corresponding
to some sort of norm or ideal—states of affairs that are ‘good’ with respect to some
normative system or system of preferences. It can include values, laws and regula-
tions, contracts, and commitments, as well as the simple desires, preferences, and
goals of an agent (teleological conversational backgrounds).4
Epistemic5 conversational backgrounds are composed of a set of beliefs of an
information source, be it an individual, an institutional subject, or an epistemic
community, which may or may not include the speaker and hearer. Often, but not
necessarily, an epistemic background is interpreted deictically as referring to the
belief set of the speaker at the moment of utterance.
While this tripartite distinction can be useful in explaining the behaviour of the
modals, it is important to bear in mind that the contextual determination of the
conversational background is not limited to the selection of one of these three kinds.
First, as we will see, many conversational backgrounds are complex and involve com-
bining propositions of different kinds.6 Second, in actual discourse the set of propositions

See Kronning (1996, 2001) for a similar tripartition and Portner (2009: 140ff.) for a partially
overlapping tripartition.
This admittedly is a very extended sense of the term deontic. Portner (2009: 139) prefers to refer to this
range by the term priority modals.
The proper term for this kind of background should be doxastic, as the term epistemic refers to
knowledge rather than belief. I keep the term epistemic because of its widespread use in linguistics.
Getting formally right what we here loosely call ‘combining’ different conversational backgrounds is
not a trivial task. In ‘classic’ Kratzer-style Relative Modality this work is done through a rather complex
132 Andrea Rocci

selected in context is typically determined in a much more finely grained manner (e.g.
neither deontic, nor legal, obligation, but a particular contextually relevant international
treaty that the country we are speaking about had signed).
Within an RM framework, the evidential component of must or dovere can be
accounted for in a semantic analysis in terms of finer presuppositional restrictions
on the type of propositions that make up an epistemic conversational background,
by formulating restrictions that exclude direct evidence and reports. Conversely,
recent work on languages with fully grammaticalized systems of evidentiality (cf.
Faller 2011 on Cuzco Quechua) has set out to analyse morphemes expressing direct,
inferential, or reportative evidentiality as different kinds of modals which impose
different presuppositions concerning the evidential source on their conversational
backgrounds. Furthermore, in an RM framework the anaphoric properties of the
modals can be accounted for by their very context-dependency. This can be repre-
sented explicitly by a procedural component of the meaning of the modal, instruct-
ing the addressee to recover the propositions making up the B from the discourse
context. For proposals for adding a procedural component to RM semantics see
Rocci (2005b, 2008, in preparation).
Having briefly introduced how modal verbs can function as evidential strategies
and how their functioning can be captured by a semantics along the general lines of
RM, it is worth considering in more detail the context of the speech acts of
prediction, which makes up the object of investigation of this chapter. In other
words, the question is why the context of predictions is interesting for analysing
evidential properties of the modals, and of Italian modal verbs, in particular.
One justification for a descriptive investigation can be simply that the well-known
inferential evidential reading of dovere exemplified in (1), which is available with
several tenses of the indicative mood of this modal verb,7 is blocked when the

possible worlds semantics machinery by introducing two sets of propositions that do different jobs: the
modal base and the ordering source (see Kratzer 1981; Portner 2009). A different formal method,
compatibility restricted union, is proposed by Frank (1996). We will assume the latter in this chapter
without discussing it. Other, more pragmatically oriented, solutions are discussed in Papafragou (2000)
and Rocci (in preparation).
This inferential reading of dovere is accessible in the present, imperfect, and remote past tenses of the
indicative mood. The possibility of this reading in the remote past tense is often omitted in the semantics
literature on modality in Italian as well as in the grammars, but it is perfectly natural and appears rather
frequently in novels as a point of view marking device:
Giovanni dovette prendere una scorciatoia. Perché arrivò là prima di tutti quel giorno.
Giovanni must-ind-rem-pst-3rd-sing take a shortcut. Because he arrived there before everybody else
that day’
‘Arguably, Giovanni took a shortcut. Because he arrived there before everybody else that day.’
In these remote past uses of dovere the inference is never situated in the past and remains anchored to the
origo of the utterance (speaker and speech time). In other words, the past tense morpheme reads like a
raised constituent taking scope on the prejacent alone rather than on the modal. The origo (the starting
reference point for deixis) can indeed shift to the past (or even to the future) when the epistemic modal is
Modal conversational backgrounds and evidential bases in predictions 133

prejacent refers to a future state of affairs and thus is not available in predictive
speech acts. Similarly, the inferential evidential reading of the Italian future tense is
available only when the proposition does not refer to a future state of affairs: when
referring to the future, the future tense loses its evidential constraints (cf. Rocci 2000,
2005b).8 So, it makes sense to look at the modal constructions that do occur in
predictions and see whether they fulfil an evidential function or not, and if they do,
to see what kind of evidentiality they express and whether this evidential function
gives rise to argumentative discourse relations of some kind.
In fact, beyond this descriptive goal—which will dictate the organization of the
present chapter—there are issues of broader import that make the journey intriguing
and worth taking. The first issue concerns what we could call the common-sense
epistemology of predictions, that is, to put it simply, the fact that predictions, being
about future events, impose certain restrictions on the kinds of evidence that might
reasonably bear on them. We will devote a few words to this issue before moving to a
detailed analysis of the corpus of Italian economic–financial news that provides
evidence for our study of prediction and justifies its rationale.

6.2 Evidence for predictions

Let us move, for a while, outside the realm of semantic studies of modality and
evidentiality to examine the idea that what legitimately counts as evidence or
arguments for a statement depends crucially on the basic logical type of the pro-
position that is being put forth. This idea has circulated for a long time in the studies
of argumentation and informal logic in various forms: it is famously, but obscurely,

embedded under indirect reported speech or under an attitude verb. Under appropriate contextual
conditions in narrative discourse the same shift can be caused by free indirect discourse (cf. Rocci
2005b: 239–59; Hacquard 2010). However, these shifted interpretations with embedding of the modal or
with free indirect discourse are only possible with the imperfect form of dovere (doveva). Thus the remote
past cases remain unembeddable and always anchored to the hic et nunc of the utterance.
The lack of inferential evidentiality in the Italian future tense when it refers to future eventualities can
be demonstrated by showing that, contrary to its properly evidential counterpart, it accepts embedding
under a reportative as in Giovanni ha detto che verrà ‘Giovanni said that he will come’. In fact, when
referring to future eventualities the Italian future tense also accepts embedding under clauses such as
Nessuno pensa che ‘Nobody thinks that’, as in Nessuno pensa che verrà ‘Nobody thinks that he will come’,
which suggests that it also lacks a proper epistemic component. When referring to present states or to past
eventualities the future tense rejects both kinds of embedding, as shown by *Nessuno pensa che Giovanni
sarà andato via ‘Nobody thinks that Giovanni must/will have left’, and ??Luigi ha detto che Giovanni sarà
andato via ‘Luigi said that Giovanni must/will have left’. For a more detailed discussion see Rocci (2005b).
These observations on the future reference uses of the future tense should suggest a reconsideration of the
use of Italian data in theoretical discussions of the relationship between tense and modality. For instance,
while we are sympathetic with Ludlow’s (1999) thesis that the Italian future tense morpheme is, not only
etymologically but also semantically, a modal, his statement that in Italian “when the future is used, it is
most likely being used to express possibility or uncertainty” (p. 159) is not entirely warranted. We hope
to soon present an empirical study of the use of the future tense in journalistic economic–financial
predictions conducted with Johanna Miecznikowski.
134 Andrea Rocci

defended in Toulmin (2003 [1958]: 13) and is implied by the whole doctrine of the
status causae in classical rhetorical theorizing. A recent discussion of this idea in
logical terms that can be transferred to a semantic investigation is offered by
Freeman (2005) in his epistemological treatment of the problem of premise accept-
ability in the theory of argumentation. Let us consider one of Freeman’s examples:
(5) a. There was a red apple on the windowsill.
b. Horatio placed that red apple on the windowsill to show his love for Ophelia.
Imagine, as Freeman asks us to, that the two statements9 are put forth as part of
an argumentative exchange. Freeman observes that the addressee of (5a) and (5b)
will proceed rather differently in determining whether these statements are
acceptable—i.e. have an “epistemic presumption” (Freeman 2005: 21–37) in their
favour. While (5a) can be based on direct perceptual evidence by the speaker, and
count as “personal testimony” for the addressee, (5b) is more complex because the
speaker “is not reporting what he has seen but explaining Horatio’s overt act in terms
of Horatio’s dispositions and intentions” (p. 93). For the addressee this would never
count as a simple testimony based on direct perception; rather it could be accepted as
an expert opinion if the addressee is “aware of the proponent’s having expertise or at
least special credentials concerning Horatio’s intimate affairs” (p. 93). For Freeman,
the different conditions of presumption of the two statements are closely connected
with the different logical type of the two propositions. The first corresponds to what
Freeman calls a description, that is a contingent extensional non-evaluative state-
ment: it only contains observational predicates, like “x is an apple, y is a windowsill,
x is on y”. The example in (5b), on the other hand, belongs to the broad class of
interpretations. It invokes non-observational concepts such as Horatio’s love—a
disposition—and the apple being a sign of that love. These are intensional concepts
and make the statement an intensional one whose truth conditions depend not only
on the state of the actual world but also of other possible worlds. Freeman defines
interpretations as contingent, non-evaluative intensional statements. According to
Freeman (p. 108) interpretations either assert or presuppose nomic regularities, that is
“subjunctive conditionals” (i.e. counterfactual in the broad sense) of necessity or
possibility that make a claim about a set of possible worlds of some sort. Physical
causation is, according to Freeman, one such conditional: saying that p causes q entails
that, for all the worlds where certain physical laws hold, wherever p is the case, q is also
the case. Other kinds of nomic regularities that Freeman recognizes include personal
ones, which concern the connections between human goals, beliefs, and action, and
institutional ones, which concern what necessarily follows or is possible given a certain
institutional state of affairs (cf. Searle’s notions of constitutive rule and institutional

Freeman’s discussion (legitimately) ignores illocutionary aspects so that statements become inter-
changeable with propositions.
Modal conversational backgrounds and evidential bases in predictions 135

fact, for instance in Searle 2005). In his classification of statements Freeman further
distinguishes descriptions and interpretations from two other logical kinds: (analytic-
ally) necessary statements and evaluative statements. In this chapter we will not be
concerned with these two further kinds or with the general applicability of Freeman’s
approach as a tool for exploring the relationship between modality on the one hand,
and the kind of evidential support we provide for our statements on the other.10 What
we will take from Freeman for the purpose of this chapter is first, the general idea that
the logical type of the propositional content of an assertive speech act11 constrains the
type of evidence on which it can be based, and second, the distinction between
descriptions and interpretations.
Let us consider how predictions constrain the kinds of relevant evidence by
reviewing first the main evidential distinctions that are typically drawn by gramma-
ticalized systems of evidentiality. Evidential systems (cf. Willet 1988; Aikhenvald
2007) typically distinguish between direct sensory evidence, inference, and reports.
Finer distinctions are often drawn within each of these domains. Some evidential
systems discriminate sight from other sensory evidence within direct evidence and,
within the domain of reports, set apart quotative evidentiality with an overt reference
of the quoted source from hearsay with no reference to the source. Within the realm
of inference there seems to be a cross-linguistically significant distinction between
inferences based on results, or more generally, direct observable circumstances that
function as a sign of non-observable states of affairs, and conjectural inferences,
which are variously characterized as based on logical reasoning, general knowledge,
or simply non-perceptual knowledge (cf. Willett 1988; Aikhenvald 2007; Squartini
2008; Faller 2002a, 2011). In Italian, the evidential use of dovere in the indicative
appears to be, at least preferentially, associated with inferences from observable data.
In this category, the subcategory of actual results exemplified in (6) is the one most
frequently encountered in the corpus.
(6) Il monte Hiei, tradotto da Antonietta Pastore (e purtroppo il suo impegnativo,
accurato, lavoro deve aver subito l’intervento di qualche malaccorto redattore
che ha fatto scempio di nomi e concetti letterari e del buddismo esoterico) è
anch’esso un romanzo autobiografico.
(Il Sole 24 Ore, 23 April 2006)
‘Mount Hiei, translated by Antonietta Pastore (and unfortunately her hard,
accurate work must have suffered the intervention of some clueless editor, who

This will be the object of a subsequent publication (Rocci, in preparation).
We do not consider here the interesting issues of evidence and of argumentative support in relation
to non-assertive speech acts. A generalized approach to evidence in relation to all kinds of illocutionary
acts will have to consider that evidence can be relevant to the truth value of different kinds of felicity
conditions (e.g., in Searlian terms, preparatory conditions, sincerity conditions, etc.). An example of
treatment of inferential evidentiality in non-assertive speech acts is offered in Rocci (2007).
136 Andrea Rocci

wrought havoc with names and concepts, both the literary and as pertaining to
esoteric Buddhism) is also an autobiographical novel.’
With respect to this typology of sources of evidence, predictions exhibit a series of
constraints, the most obvious one being the fact that future events are not directly
observable. Less obvious is the restriction that predictions impose on evidence from
reports: it is possible to base one’s prediction on what other people say about the
occurrence of future events, but in this case the words of others cannot be taken as a
form of testimony as the impossibility of direct evidence holds also for the other
subjects. Reports can merely have the value of expert opinion, presupposing or
embedding an inference of the source. Alternatively, reports may refer to the
verbalized intentions and plans of an agent, again embedding the agent’s reasoning
concerning the feasibility of the planned course of action. Inference is also subject to
restrictions concerning the applicable inference schemes. If we consider just the
linguistically relevant categories mentioned above, we can immediately see how the
category of inference from results—at least in the strict sense—cannot be applied, as
there are no observable results of future events. As we will see in the following
sections, these basic epistemological restrictions on predictions are reflected in the
use of modals in the speech acts of prediction.
Given these constraints on evidence, to what basic logical type of proposition
should we ascribe predictions? Here, assuming that the fourfold typology proposed
by Freeman (2005) is basically correct, we will put forward the hypothesis that
predictions are a form of interpretation; they are non-evaluative intensional state-
ments. Predictions assert that a certain state of affairs will necessarily, probably, or
possibly be the case in view of a set of facts in the actual world and of a nomic
regularity which can be of different types (including the physical, personal, and
institutional ones examined by Freeman). Predictions are thus inherently modal. As
we will see in the following sections, the modals occurring in predictions function as
a diverse and intermittent manifestation of this inherent modality.

6.3 The data: a corpus of financial news

The corpus we use to explore the use of modals as evidential strategies in predictions
was collected as part of a broader research project12 on a unique discourse genre and
consists entirely of economic–financial news reports. The corpus consists of one full
month (April 2006) of three Italian business newspapers (Il Sole 24 Ore, Italia Oggi,
and MF/Milano Finanza) and contains roughly four million words, the strictly
economic and financial sections of the papers amounting to 6,515 texts and

The corpus was collected for a research project entitled Modality in argumentation. A semantic-
argumentative study of predictions in Italian economic-financial newspapers. The project was supported by
the Swiss National Science Foundation (Grant: 100012-120740/1) from 2008 to 2011.
Modal conversational backgrounds and evidential bases in predictions 137

3,087,056 running words. A balanced sample of 200 texts (101,974 running words)
from the financial sections was used to create a manually annotated sub-corpus. The
multi-layer annotation scheme used there includes tags for a variety of formal and
functional categories, amongst them the formal structure of articles, reported speech,
and, most importantly, predictions and future states of affairs.13
The choice of this terrain for studying the interaction of predictions and modality
was motivated by the semantic features of the genre and its socio-pragmatic func-
tioning. Predicting or ‘forecasting’ future events represents a central discourse
activity in economics and especially in finance (Merlini 1983; Bloor and Pindi 1990;
Walsh 2004, 2006; Donohue 2006). While partly delegated to experts—academic or
institutional economists, financial analysts, etc.—it remains a central concern of all
market participants such as managers and investors, as well as journalists. Speech
acts about future events are so central to investment activities that the field of
finance created its own indigenous speech-act label to deal with them metalinguis-
tically: ‘forward looking statements’ (cf. McLaren-Hankin 2008). In the financial
markets, the uncertainty inherent in making statements about the future, and hence
‘unsettled’ events (cf. Kaufmann et al. 2006: 99) is combined with the uncertainty
deriving from an incomplete knowledge of the present situation, which typically
takes the form of the so-called information asymmetry between insiders and out-
siders (Barone-Adesi 2002). The hypothesis behind the corpus selection is that the
social ontology, the epistemic constraints (cf. Cooper and Ebeling 2007) of invest-
ment activities and of financial journalism, and, last but not least, the complex ways
in which they depend on the ontology and epistemology of time, make the financial
markets a particularly favourable ground for exploring the communication of
reasoning about possible future events (and hence for exploring, at a linguistic
level, the relationship between the linguistic representation of futurity, modality,
and evidentiality). Data from the Italian corpus confirm the centrality of predictions
in financial news. As shown in Miecznikowski, Rocci, and Zlatkova (2012), acts of
prediction play a quantitatively and also hierarchically dominant role in the genre of
economic financial news, making up more than one-third of the annotated sub-
corpus in terms of word count, and being present in more than half of the headlines
and in more than one-third of the highlights. An informal survey of the predictions
in the corpus reveals that predictions are typically qualified by modals and accom-
panied by supporting arguments. They are often relative to conditional scenarios and
routinely attributed by the journalist to sources, including financial experts, named
or unnamed insiders, and rumours. Quantitative corpus data also show the import-

The annotation was carried out using UAM Corpus Tool, a freely available open source environment
for the annotation of text corpora created by Mick O’Donnell (; see
O’Donnell (2008).
138 Andrea Rocci

ance of modality in predictions, and in particular of the two modal verbs dovere and
potere. If we extract the statistically more prominent keywords from the speech acts
of prediction (with respect to the rest of the annotated sub-corpus)14 we obtain five
inflected wordforms of the modals among the twenty highest-ranking keywords,
four of them among the ten highest-ranking, as illustrated in Table 6.1.
As Table 6.1 shows, the future and the conditional forms of the modal verbs dovere
and potere are highly prominent in the predictions that are found in the corpus of
Italian economic financial news. If the prominence of the future forms of the modals
is perhaps unsurprising given that we are dealing with predictions, the equal
prominence of the conditional mood forms is perhaps more difficult to explain.
The three main modal uses of the Italian conditional mood are usually considered
to be the following (cf. Miecznikowski and Bazzanella 2007): (i) a hypothetical use,
typically in the consequent of a subjunctive conditional construction, where the
antecedent is evaluated as weakly possible or counterfactual; (ii) a reportative
evidential use, indicating that the prejacent has been reported by a named or
unnamed source; and (iii) an attenuative use, where the conditional appears to
function as an illocutionary force modifier downgrading the illocutionary force of
the speech act. In principle, all three uses of the conditional could be relevant in
financial news predictions.
In the following sections we will present a qualitative analysis of the use of modals
in these utterances assuming an RM semantic analysis of the modals, and look at the
kinds of conversational background that the modals adopt, as well as the way in
which these conversational backgrounds relate to prediction. In the final sections of
the chapter we turn to examining more closely the functioning of the conditional

Table 6.1 Most frequent modals in speech acts of prediction in the corpus
Rank Wordform Gloss Keyness15

4 Potrebbero CAN cond. pres. 3rd plur. 42.53

5 Dovrà MUST ind. fut. 3rd sing. 40.0
7 Dovrebbe MUST cond. pres. 3rd sing. 27.99
8 Potrebbe CAN cond. pres. 3rd sing. 25.61
13 Potrà CAN ind. fut. 3rd sing. 13.37

This means that we calculate the statistical keyness of words occurring in the segments tagged as
‘predictions’ using the remaining text of the tagged sub-corpus as the reference corpus for the calculation.
The keywords of predictions are those words whose relative frequency in the ‘predictions’ text is
significantly higher than their relative frequency in the reference corpus. Keywords are automatically
calculated by the UAM Corpus Tool software.
A keyness value of 100 indicates that the word appears 100 times more in the corpus under study
than in the reference corpus.
Modal conversational backgrounds and evidential bases in predictions 139

forms in order to highlight how they give rise to constructions which impose specific
evidential constraints on their conversational backgrounds.

6.4 Evidential implications of future-oriented alethic and deontic modals

All of the occurrences of the modals pertaining to the indicative mood that we
encountered in predictive speech acts in the corpus—most of them in the future
tense, some in the present tense—can be classified as deontic or alethic circumstan-
tial modals. As we have observed above, the epistemic inferential reading of indica-
tive dovere is blocked with future prejacents, so it is expected that in predictions we
encounter only non-epistemic modals.
As for potere, clearly epistemic readings of the possibility modal with non-future
prejacents are possible, but as shown in Rocci (2005a, 2005b), they are rare and do
not seem to function as inferential evidentials or to be able to license argumentative
discourse relations, except when they are negated and signal an epistemic impossi-
bility. This contrasts with the behaviour of other markers of epistemic possibility in
Italian, such as the adverb forse (‘perhaps’, ‘maybe’) and the impersonal comple-
mentizer constructions può darsi che/può essere che (‘It may be [the case] that’)
which do signal inferential evidentiality and argumentative discourse relations
(cf. Rocci 2005a and Rocci and Zlatkova, in press, respectively). With future pre-
jacents, the distinction between ‘root’ (deontic and alethic) and epistemic readings
becomes apparently more blurred, especially in the area of circumstantial uses of the
modal. Coates (1983, 1995), who observed a similar phenomenon for English can,
referred to these uses as mergers.
Consider the following example:
(7) Può scoppiare un temporale.
‘A storm may/?can break out.’
If (7) is uttered under a darkened cloudy sky it can be taken to mean either that the
breaking out of a storm is compatible with what ‘we’ (or the relevant epistemic
community) know about the world, or that this state of affairs is compatible with the
current atmospheric circumstances and with the nomic regularities of physical (and
specifically meteorological) causality. Thus we can attribute to the modal either an
epistemic or a circumstantial conversational background. Interestingly, when the
modal potere is inflected in the future tense, as in (8), the impression of overlap or
merger between epistemic and circumstantial reading disappears and only the
circumstantial reading remains.
(8) Potrà scoppiare un temporale.
‘It will be possible for a storm to break out.’
140 Andrea Rocci

In (8), the future tense clearly takes scope over the modal and the modal selects a
conversational background of future atmospheric circumstances, rather than of
present knowledge. Similar effects can be achieved if we add temporal adverbials
to (7): the adverbials, as shown in (9), take scope over the modal in a manner quite
unlike what will be expected with epistemic modality:
(9) Può ancora scoppiare un temporale.
‘A storm can still break out.’
These remarks prompt us to hypothesize that, at least in the indicative, for potere too
no genuinely epistemic readings are possible with future prejacents. In fact, in the
corpus of predictions we have not found any likely candidate for epistemic status,
either in the present or in the future tense. What we find, with both potere and
dovere, is a variety of circumstantial and deontic conversational backgrounds. What
is interesting, however, is to look at how these non-epistemic conversational back-
grounds relate to the kinds of evidence on which predictions are based.

6.4.1 Economic causality: necessity and impossibility

Purely alethic circumstantial interpretations of the modal potere in the present and
future tenses are encountered in the corpus. These readings select a conversational
background composed entirely of facts of the world belonging to a certain loosely and
contextually defined kind (and of the associated nomic regularities relevant for the
domain). Most of these alethic circumstantial readings clearly involve causal relations,
but they are not concerned with physical causality. Rather, they signal what we might
call economic causality. They are typical examples of discourse focusing on the eco-
nomy, where events in the markets are not seen as the result of human actions, but as
quasi-natural events causally related by virtue of ‘economic laws’, which are imagined to
operate blindly.16 The negative form of potere is used to express economic impossibility.
The combination of this same possibility modal with only type quantifiers is, remark-
ably, used to express economic necessity, as shown in (10) and (11).
(10) Meno rosee le prospettive per i consumatori: secondo Browne il prezzo della
benzina non potrà che salire data l’impennata del greggio. In Gran Bretagna il
prezzo della benzina ‘potrebbe salire oltre una sterlina al litro’, ha detto
Browne, che ha pero tenuto a sottolineare che BP passa ai clienti una frazione
degli aumenti dei costi reali.
(Il Sole 24 Ore, 14 April 2006)

This kind of discourse, of course, corresponds to the ontology and rhetoric of classical economics, as
wittily pointed out by Searle (2005: 1): “When I was an undergraduate in Oxford, we were taught economics
almost as though it were a natural science. The subject matter of economics might be different from physics,
but only in the way that the subject matter of chemistry or biology is different from physics . . . . At no point
was it ever suggested that the reality described by economic theory was dependent on human beliefs and
other attitudes in a way that was totally unlike the reality described by physics or chemistry.”
Modal conversational backgrounds and evidential bases in predictions 141

‘Less rosy the outlook for consumers: according to Browne gasoline prices can
only rise in view of the surge of the crude oil price. In Great Britain the price
of petrol “could rise to more than one pound per litre”, said Brown, who was
keen to stress, however, that BP passes on to its customers a fraction of the
actual cost increases.’17
(11) Solo se ci fossero segnali di possibili ribassi del costo del denaro in America si
potrebbero giustificare rendimenti decennali così contenuti’ – osserva Holger
Schmieding, co-responsabile settore economico europeo di Bank of America.
‘Ma siccome non ci sono questi segnali, anzi la tendenza è quella opposta, i
rendimenti decennali non possono fare altro che aumentare. Il ragionamento
è semplice: chi investe su una scadenza decennale vuole solitamente interessi
più elevati rispetto a chi acquista obbligazioni a breve scadenza perché il
rischio “temporale” è maggiore.
(Il Sole 24 Ore, 14 April 2006)
‘ “Only if there were signs of possible decline in the cost of money in America
one could justify such limited ten-year yields”, observes Holger Schmieding,
co-head of European economic sector of Bank of America. “As we do not have
these signs, and, in fact, the trend is the opposite, the ten-year yields can only
increase. The reasoning is simple: investors on a ten-year maturity typically
want higher interest rates than those who buy short-term bonds because the
‘temporal’ risk is greater.” ’
The Italian non . . . che constructions in (10) and (11) express an ‘only’ quantifier
taking scope over potere. Thus the modal sentences in (10) and (11) assert that there
is no proposition in a contextually relevant contrast set C other than the prejacent
that is compatible with the conversational background. These quantifiers also
presuppose that the prejacent is indeed compatible with B (cf. Horn 1996 on the
semantics of only). Only, by asserting the impossibility of the alternatives to the
prejacent in C, transforms the modality expressed by potere in an indirectly ex-
pressed alethic circumstantial necessity (¬◊¬ p), under the assumption that the
alternatives in C are exhaustive.
It is worth pausing at these two examples to observe how the functioning of these
two expressions of alethic circumstantial necessity relates to the manifestation of the
evidence on which the predictions are based. The key lies in observing their
behaviour as context-dependent expressions that seek to saturate the conversational
background variable in the discourse context. In examples (10) and (11), the satur-
ation is provided, in part, by the causal clauses that precede the modal (in view of
the surge of the crude oil price, As we do not have these signs, and, in fact, the trend

Edmund John Philip Browne was group Chief Executive of BP (1995–2007).
142 Andrea Rocci

is the opposite) which, in combination with the other circumstances of discourse and
the relevant nomic regularities, yield the causal necessity. This causal chain is at the
same time an argumentative one and constitutes the evidence on which the predic-
tion is based.18 Contrary to what genuine epistemic inferential modal expressions
do, these alethic circumstantial modals do not refer directly to the speaker’s know-
ledge, beliefs, or reasoning processes but to the underlying real-world relations—
causal in this case—that license the inference.
The extended text of example (11) is also noteworthy for another reason: it contains,
in the last sentence, a shift between what we have called economic causality and its main,
underlying, more basic form of connection: economic rationality. The latter concerns
the goal-oriented behaviour and practical reasoning of human agents (investors in this
case) and arguably involves teleological, rather than alethic, modalities (cf. Portner 2009:
185–6). Under the assumptions of classic, non-behavioural, economics, these can be
equated with causal connections because the agent’s goals are considered as given
(utility maximization) and the rationality of the agent is presumed.19

6.4.2 Quantificational readings of the possibility modal

Let us now consider example (12) below. It selects an alethic circumstantial conver-
sational background of economic facts including oil price and rising interest rates
and relates them to consumer spending in a temporally generic statement:
(12) Gli altri rischi che gravano sull’evoluzione del commercio internazionale sono
di natura macroeconomica: prezzo del petrolio e rialzo dei tassi d’interesse
possono influire sui consumi.
(Il Sole 24 Ore, 12 April 2006)
‘The other risks that loom over the evolution of international trade are
macroeconomic in nature: oil price and rising interest rates can influence
consumer spending.’
This reading is close to the so-called “sporadic” (Kleiber 1983) or “quantificational”
(Portner 2009) reading of possibility modals, canonically exemplified by Lions can be
dangerous. It does not directly express a prediction concerning a singular event
(oil price and rising interest rates will possibly influence consumer spending or con-
sumer spending will possibly decrease). Instead, it offers a generalization (possibly

More precisely, we have here an instance of the argumentation scheme from the cause to the effect.
This argumentation scheme exploits the basic entailments of the common-sense ontology of causation to
infer a conclusion. If we admit that ‘p causes q’ entails that, for all the worlds where certain physical laws
hold, wherever p is the case, q is also the case, we can use this entailment to infer q from p. For an in-depth
discussion of contemporary theories of argument schemes and a theoretical proposal making explicit the
connection between argument schemes and commonsense ontologies, see Rigotti and Greco Morasso (2010).
Additionally, the incompleteness of the information available to the agents is ignored as the markets
are presumed to be informationally efficient.
Modal conversational backgrounds and evidential bases in predictions 143

inductively based on previous comparable occurrences) that supports the implicit

prediction with an implicit weak epistemic evaluation.20

6.4.3 Economic circumstances and agent’s goals

In the corpus the necessity modal dovere is never found with a strictly alethic
circumstantial conversational background corresponding to what we have called
economic necessity. This appears to be consistent with the use of the combination of
potere with quantifiers of the type of only to represent this notion that we have seen
in (10) and (11). Certain purely alethic circumstantial readings of dovere are possible,
especially when dealing with physical causality and non-human events, as when
uttering (13) under a cloudy sky:
(13) Deve piovere.
‘It’s going to rain.’
In contexts such as (10) and (11), however, purely alethic circumstantial interpretations of
indicative dovere appear more difficult to access than the deontic ones. We have,
however, a number of occurrences of future tense dovere that come close to expressing
alethic circumstantial necessity, but are subtly different, such as, for example, (14).
(14) Giovedì si è passata per la prima volta in quattro anni la soglia del 5% per i
tassi a dieci anni. La conseguenza più immediata di questo aumento ricade sui
tassi per i mutui immobiliari. Chi aveva contratto mutui a tassi variabili—e
sono stati in molti—si trova alla scadenza del primo periodo e dovrà rinego-
ziare tassi di 200 o 300 punti superiori a quelli di un paio di anni fa. Questo
significa che una famiglia media con un mutuo di 400 mila dollari potrebbe
trovarsi a dover pagare anche fino a mille dollari in più al mese.
(Il Sole 24 Ore, 15 April 2006)
‘On Tuesday the threshold of 5% for fixed ten-year interest rates was crossed for
the first time in four years. The most immediate consequence of this increase will
be on mortgage rates. Those who had subscribed to variable rate mortgages—and
there are many—are at the end of the first period and will have to renegotiate
rates of 200 or 300 points higher as compared with those of a couple of years ago.
This means that an average family with a loan of 400 thousand US dollars could
have to pay even up to a thousand dollars more a month.’
Two factual propositions are anaphorically recovered by the preceding discourse
context and are added to the conversational background: Ten-year fixed interest
rates have climbed over the 5% threshold and Those who had subscribed to variable
rate mortgages are at the end of the first period. These are social, economic facts.
These facts alone, however, are not sufficient to make necessary the renegotiation of
Compare with the inferential chain: Lions can be dangerous and Lambert is a lion, therefore Lambert
may be dangerous.
144 Andrea Rocci

the mortgage with rates of 200 or 300 points higher than those of a couple of years
earlier. Other institutional facts are possible as outcomes. In order to make the
proposition necessary, we need to add a non-factual proposition: a goal like if they
want to keep their houses, or in order to keep their houses. What we have here is a
combination of an alethic circumstantial conversational background and a deontic
teleological conversational background. Deontic conversational backgrounds contain-
ing ideals, goals, values, or laws21 always need to combine22 with actual circumstances
in order to make singular propositions necessary. What makes examples such as (14)
special, however, is the fact that it is taken for granted that the agents will act rationally
to fulfil their goal. This additional premise makes the modal a predictive one,
equivalent to the alethic modal.

6.4.4 Deontic readings of potere

Deontic readings of potere, such as the future tense reading in (15), can also give rise
to predictions:
(15) Europa (e mondo) attenti, a Citigroup sono state tolte le redini. La principale
banca americana e stata perdonata dalla Federal Reserve e, per crescere, potrà
tornare a fagocitare prede sui mercati globali: dopo un anno di forzata
moratoria sulle grandi acquisizioni, imposta davanti alla scoperta di scandali
e inadeguati controlli interni, la Fed ha concluso che il colosso dei servizi
finanziari ha ‘compiuto significativi progressi’ nella governance e nella ges-
tione del rischio, sufficienti a togliere i freni a piani di conquista. Anche se ha
mantenuto un avvertimento: ‘Esamineremo attentamente ogni proposta di
espansione di Citigroup’.
(Il Sole 24 Ore, 5 April 2006)
‘Europe (and the world) beware, Citigroup has been unleashed. The major US
bank was pardoned by the Federal Reserve and, in order to grow, will again be
able/allowed to engulf prey on global markets: after a year of an enforced
moratorium on large acquisitions, set in the wake of the discovery of scandals
and inadequate internal controls, the Fed concluded that the financial services

Not all deontic conversational backgrounds need to combine with circumstantial ones: directive and
commissive ones already originating from singular commands or singular promises do not.
One possibility of formalizing this combination is represented by the set-theoretic operation of
compatibility restricted union (Frank 1996): if BA is the alethic circumstantial conversational background
and BD is the deontic conversational background, the complex conversational background will be BA [! BD.
This operation generates one or more complex conversational backgrounds BA-Dn corresponding to the
union of BA with a maximal subset BDn of BD compatible with BA. The compatibility restricted union may
generate more than one BDn set, as there could be different ways of resolving the incompatibilities between
BA and BD. This can happen, for instance, in practical reasoning, when BD contains desires that are revealed
to be incompatible when confronted with the circumstances in BA. The agent has to choose the desire to
which they will give precedence (cf. Frank 1996: 43–6). In these cases we assume that only one BDn set is
selected pragmatically as relevant for determining the complex conversational background (cf. also Frank
1996: 44–5).
Modal conversational backgrounds and evidential bases in predictions 145

giant has “made significant progress” in the governance and risk management,
sufficient to remove the brakes to plans of conquest. However, the Fed maintain
a warning: “We will carefully review any proposal for expansion of Citigroup.”’
The deontic conversational background of potrà in (15) originates from the directive
power of the Fed, which has eased a prohibition issued against Citigroup. This is
equivalent to permission. In order to understand how this modality works in the
prediction, we have to pause for a moment to consider the nature of the prejacent
which concerns Citigroup taking over other companies. Corporate takeovers are
institutional actions, rather than natural actions, and norms have causal power
over them. This means that the deontic prohibition of the Fed amounts to a special
kind of alethic impossibility: it is not that a takeover without the Fed’s consent would
be illegal; it would be void or null. Thus, the deontic modal ends up counting as a
circumstantial alethic one. The circumstantial possibility is then combined with an
additional premise concerning the intention (in order to grow, plans of conquest, etc.)
to derive an indirect prediction.

6.4.5 Deontic readings of dovere: schedules and plans

The necessity modal dovere is routinely used in predictions in the corpus with
different deontic conversational backgrounds. One type of deontic conversational
background found in predictions consists of the formal or informal commitments
made by institutions, and particularly companies, in the form of plans or schedules.
In (16) below we find a basic example of this recurrent pattern:
(16) Il prossimo 20 aprile gli azionisti dovranno votare sul fatto che sussistano
ancora o meno i requisiti di onorabilità richiesti al manager per presiedere una
(Il Sole 24 Ore, 5 April 2006)
‘On April 20 the shareholders are to vote on whether the manager still
satisfies the requirements of integrity necessary to chair a bank.’
Other researchers working on Romance modal verbs with journalistic corpora
have remarked on these uses and provided divergent interpretations. Kronning
(1996, 2001) insists on the predictive nature of similar uses of French devoir, and
for this reason he considers them a subtype of alethic modality, a “restricted” alethic
modality. According to Kronning, these uses are based on narrative scenarios to
which the future is expected to conform: the prejacent p is true in all the worlds
consistent with the scenario. Squartini (2004) examines similar examples in French
and Italian, coming to a diametrically different conclusion. For Squartini, examples
like (16) represent evidence of a reportative evidential reading of dovere/devoir.
146 Andrea Rocci

In fact, the conversational background B invoked by dovere in examples such as (16)

is neither alethic (propositions that are facts in the world) nor primarily reportative23
(propositions asserted by another speaker), but rather deontic (propositions denoting
norms, commitments, or goals). Plans formulated by companies and other organiza-
tions denote strategic or tactical goals to which certain members are committed. At the
same time, plans take the form of written documents that can be used by external
observers as sources of information. Thus, the deontic and the reportative do overlap
in the journalist’s perspective. The deontic commitment, however, remains primary
with respect to its use as evidence in a prediction by a third party.
Regulative rules, in the form of laws, regulations, and legally binding agreements
(e.g. contracts) can also form the basis of prediction signalled by indicative dovere on
the basis of the assumption that the subject can and will probably abide by the
regulative rules:
(17) Gli statunitensi posseggono il 15% di Lukoil, precisa il gestore, e in base agli
accordi con i russi devono salire al 20% acquistando titoli sul mercato.
(Il Sole 24 Ore, 1 April 2006)
‘The Americans hold 15% of Lukoil, the money manager elaborates, and
according to the agreement with the Russians, they are to climb up to 20%
by buying stocks on the market.’
Neither kind of deontic background warrants an attitude of absolute certainty: in
(17) the Americans might pull out from the deal with the Russians by paying a
penalty; in (16) the board of the bank might just cancel the shareholders’ meeting.
Interestingly, while the addressee may well assess the degree of certainty of the
prediction on the basis of the nature of the deontic background involved, the writer,
by using a deontic modality, does not commit himself directly to a given degree of
epistemic certainty towards the prediction. Should the implicated prediction turn
out to be false, the speaker could always claim that the deontic modality was correct.

In terms of the tripartition of conversational backgrounds that we have adopted here, a reportative
conversational background can be considered a subtype of the epistemic conversational background,
corresponding to the beliefs to which the source is committed (the beliefs of the source’s social persona,
independently of their sincerity). The speaker may or may not associate herself with this belief set. Faller
(2011) argues that certain reportatives fall outside the scope of epistemic modality and should be based on
a separate informational conversational background because they can be used also when the speaker does
not believe the prejacent. This is consistent with her adoption of a strict definition of epistemic modality in
terms of knowledge. Here we maintain that a broader definition of epistemicity in terms of a set of beliefs
associated with an information source can be more advantageous in dealing with issues of evidentiality,
subjectivity vs. intersubjectivity, polyphony, and point of view that arise with epistemic modals (compare
also with von Fintel and Gillies’ 2011 notion of epistemic modality).
Modal conversational backgrounds and evidential bases in predictions 147

6.5 Modals and the conditional mood in predictions: dovrebbe

Having examined how different kinds of alethic and deontic uses of the modal verbs
dovere and potere contribute, indirectly, to relate predictions to recurring types of
evidence and to point to the discursive manifestation of such evidence in the co-text,
we now turn to a construction based on the conditional mood form of the modal
dovere (henceforth DOVREBBE) which, in the literature on Italian modals, is
considered epistemic and has been seen as a marker of both inferential and reporta-
tive evidentiality (cf. Squartini 2004; Pietrandrea 2005, 2007). In Rocci (2006, 2008,
2011), taking inspiration from an analysis of its French analogue devrait by Kronning
(2001), we proposed that the modal and evidential functioning of DOVREBBE, as
well as its contribution to argumentative discourse relations, are best understood if
we relate this construction to circumstantial alethic and deontic uses such as the
ones that we have examined in the preceding sections. More precisely, we proposed
that this epistemic–evidential DOVREBBE is a necessity modal based on a complex
conversational background: one which combines an alethic or deontic conversa-
tional background with a second conversational background, motivated by the
presence of the conditional morphology.
We first present the semantic analysis we propose for the modal and then move to
examining how this analysis accounts for the evidential and discursive properties of
this construction in the predictions found in the corpus. In order to do so we
conducted a systematic analysis of the conversational backgrounds of all the occur-
rences of dovere in the conditional mood in the 200 texts of the sub-corpus.
As shown in Table 6.2, the conditional of dovere occurs in the third person only,
mostly in the singular, and nearly always has a predictive function which can be
traced back to the DOVREBBE construction. The remaining four occurrences are
purely deontic, comparable to English deontic ought (cf. von Fintel and Iatridou
2006), and will not be discussed here.

6.5.1 The semantics of dovrebbe: conversational background

and conditional restriction
According to the hypothesis mentioned above, DOVREBBE involves a double
conversational background Bii, consisting of the combination—in terms of the set-
theoretic operation of compatibility-restricted union—of an alethic or deontic back-
ground Bi and a conditional restriction C (Bii ¼ Bi [! C):

Table 6.2 The conditional of dovere in the sub-corpus

Wordform Morphological gloss Tokens In prediction

dovrebbe cond. pres. 3rd sing. 56 53

dovrebbero cond. pres. 3rd plur. 11 10
Total 67 63
148 Andrea Rocci

(Def. 3) DOVREBBE Bii (Bi, C, p):

a. C is a set of propositions non-factual for the speaker in w0, which are
weak presumptions in the relevant epistemic community.
b. Bi is a set of propositions which are either true in w0 (alethic) or
commitments in w0 (deontic).
Semantic entailment: p follows from Bi [! C
DOVREBBE imposes a presuppositional condition on the saturation of C that is
important to explain the epistemic and evidential functioning of the construction.
First of all, C is composed of propositions that are not known to be facts in the world.
They are non-factual, or ‘counterfactual in a weak sense’: they may turn out to be
false. In this broad respect, the propositions C are like the antecedent of the Italian
subjunctive conditional constructions in whose consequent the conditional mood is
used. However, unlike subjunctive conditional constructions, where the antecedent
is either remotely possible or ‘counterfactual in a strict sense’ (known to be false in
the world), the C of the DOVREBBE construction corresponds to propositions for
which there exist ‘weak presumptions’ within a relevant epistemic community. In
the following two sections, with the help of corpus examples, we consider the alethic
or deontic conversational backgrounds of DOVREBBE before turning to the nature
of its conditional restriction.

6.5.2 Alethic conversational backgrounds: economic causality

Consider example (18) below:
(18) Il dato relativo alla vendita di nuove case negli Usa a febbraio ha fatto
registrare un vero e proprio crollo (–10,5%, a 1,08 milioni di unita), il calo
più forte da nove anni. Aumenta anche il numero degli alloggi invenduti, un
fatto che – se confermato in futuro – dovrebbe riflettersi in una riduzione dei
prezzi degli immobili, con effetti di raffreddamento sulla crescita dell’infla-
zione. Questa statistica ha sostanzialmente ribaltato quella relativa alle case
esistenti, che aveva messo in mostra una crescita del 5,2% a febbraio.
(Il Sole 24 Ore, 3 April 2006)
‘The sale figures for new homes in the US in February showed a real slump
(–10.5%, to 1.08 million units), the strongest decline in nine years. The number
of unsold houses is increasing, a fact that, if confirmed in the future, should be
reflected in reduced house prices, with a cooling effect on the growth of
inflation. This statistic is essentially in reverse to that relating to existing
homes which depicted growth of 5.2% in February.’
In (18), dovrebbe signals a consequence based on economic causality, and can
therefore be interpreted as based on an alethic circumstantial conversational back-
Modal conversational backgrounds and evidential bases in predictions 149

Table 6.3 Subtypes of alethic Bi in the sub-corpus

Saturation of Bi No. of occurrences

Economic causality 32
Complex causality 6
Calculation 2
Human psychology and reasoning 1
Total: 41

ground similar to the causal uses of potere and, to a lesser extent, to the quasi-causal
uses of dovere in the indicative examined in the previous sections. Here, however, the
causal necessity is conditional on the continuation of the slump (which, by the way,
would be normal and expected).
The majority of the occurrences in the 200 texts sample (41/63) have an alethic Bi,
in most cases clearly identifiable with what we have called economic causality, as
demonstrated in Table 6.3.
A small number of predictions (six) are based on complex causal processes which
involve both economic facts (with economic nomic regularities) and other kinds of
propositions (including their relevant nomic regularities). These include physical
causation, the (psychological and organizational) weight of habits, and the economic
rationality of agents. Backgrounds combining economic events with the policy of
central banks also belong here. While economic rationality and policies are, in prin-
ciple, a teleological and a deontic concept, they are not really treated as such.
Economic rationality we have already discussed. As for central bank policies, they
are regarded more as nearly stable elements of the functioning of the system than as
deontic commitments of agents. In two cases, the prejacent of dovrebbe is the result of
a calculation (e.g. short-term interest rates), performed according to a conventionally
accepted method, from a more basic value (e.g. primary interest rates). We should
note here that the role of calculation is not simply epistemic (as it would be in the
natural sciences) but properly causal: banks do not discover short-term interest rates
when they calculate them, they establish them. The interest rates come into being as
institutional facts according to the calculations. Finally, there is a sole example of
psychological causality, which is, in fact, used meta-argumentatively (non dovrebbe
sorprendere ‘it should not come as a surprise’).

6.5.3 Deontic conversational backgrounds

Not all examples of DOVREBBE found in the corpus correspond to a background of
economic circumstances. The other twenty-two of the sixty-three occurrences of
predictive DOVREBBE in the 200-text sample are analogous to (19):
150 Andrea Rocci

(19) Stando a quanto emerso ieri nella riunione del cda Bnl, Bnp sarebbe orientata
a lanciare la prossima settimana l’Opa, che dovrebbe concludersi tra il 15 e il
20 maggio. Secondo indiscrezioni la banca di Parigi avrebbe predisposto tutto
per annunciare già stasera l’ok della Consob e i dettagli dell’operazione, con le
date di inizio e di conclusione.
(Il Sole 24 Ore, 13 April 2006)
‘According to what transpired from yesterday’s meeting of the board of BNL,
BNP would be inclined to launch the takeover bid next week, to be completed
between the 15th and 20th of May. According to rumours, the Paris bank
prepared everything to announce this evening the OK of Consob and the
details of the transaction, with the dates of commencement and conclusion.’
They invariably refer to the plans of corporations and other institutions that play a
role in the markets. In just one case these plans make reference to the precise
calendar scheduling of events. In most occurrences with DOVREBBE, plans are
inside corporate information reported off the record by anonymous insiders (indis-
crezioni). This contrasts with the more public plans and scheduling that we find with
deontic indicative dovere (as well as with the simple future tense). Not surprisingly,
in (19) we find instances of the reportative conditional mood (sarebbe orientata,
avrebbe predisposto) in clauses immediately preceding and following the one with
DOVREBBE. The relationship between dovrebbe and the reportative use of the
Italian conditional mood is better understood if we look more closely at the nature
of the conditional restrictions C, discussed in the following section.

6.5.4 The nature of conditional restrictions

In most of the occurrences of DOVREBBE in the corpus it is relatively easy to identify
propositions in the immediate discourse context that make the conditional restriction
C explicit. These ‘antecedents’ can take the form of: (i) prothases introduced by the
conditional conjunction se ‘if’ with the indicative mood or non-finite verb, as in (18)
(se confermato in futuro ‘if confirmed in the future’); and (ii) other kinds of subor-
dinate clauses; as well as (iii) independent sentences introducing an explicit or implicit
modality to which DOVREBBE points anaphorically, realizing a form of modal
subordination (cf. Roberts 1989). However, DOVREBBE differs from the plain hypo-
thetical conditional mood (and from the English modal would) in that it does not
accept antecedents that are epistemically evaluated as weak possibilities. As a result, we
do not find protases with ‘se þ imperfect subjunctive mood’,24 nor, in the case of
modal subordination, discourse antecedents epistemically modalized with a mere

For instance, if we were to add a finite verb to the verbless protasis in (18) it would rather be an
indicative future form (se verrà confermato in futuro) than an imperfect subjunctive one (se venisse
confermato in futuro).
Modal conversational backgrounds and evidential bases in predictions 151

Table 6.4 Conditional restrictions for the alethic Bi in the sub-corpus

Conditional restriction of alethic Bi Tokens

Continuation of a trend, stability of a situation 5

Hypothesis (including the hypothesis of the stability of a situation) 4
Confirmation of an expectation 4
Reportative (expert predictions, insider predictions) 14
Rebuttal (unless improbable event occurs, risk materializes) 5
Modal subordination (dovrebbe . . . dovrebbe) 2
Generic reference to a margin of uncertainty 7
Total 41

possibility modal (in contrast with the admissibility of modal subordination sequences
‘might . . . would’ in English and ‘potrebbe . . . conditional mood’ in Italian).
If we look at the occurrences in the 200-text sub-corpus, we find that C always
belongs to one of the semantic types from a very restricted inventory. When Bi is
alethic, we find the distribution of conditional restrictions represented in Table 6.4.
As Table 6.4 shows, in a certain number of occurrences C is simply identified with
the continuation of a trend or the stability of a situation, where the latter are taken as
defaults in economic causality, just as they are in naive physics.25 Another group
involves the introduction of an epistemic layer of hypothesizing with antecedent
structures of the type If the hypothesis X is confirmed. Since the hypothesis is
nominalized, it is usually not clear who the subject of the hypothesizing is. The
content of the hypothesis is again always a default. Among these defaults we find
again, for instance, the (hypothesis of) the stability of the situation. Confirmation of
an expectation is another C that contains an explicit epistemic evaluation whose
subject is not spelled out. The most common type of C is reportative. In the case of
an alethic Bi, it is nearly always an expert prediction—in one case it’s an insider one.
So, in the latter uses of DOVREBBE, the prejacent p is the case in all the worlds
where a set of factual economic circumstances with the relevant economic nomic
regularities is the case (Bi) and the experts are right (C). The experts’ being right is
another default, or weak presumption. Defaults can be formulated negatively in the
form of exceptions or rebuttals (Unless improbable, event X happens). Not surpris-
ingly, one of these exceptions is the disappearance of a trend. Finally, the weakly
presumed antecedent can be derived anaphorically via modal subordination as the
truth of the prejacent of a preceding dovrebbe.

We have already mentioned how in our texts economic causality is, in certain respects, treated as if it
were physical causality. The pervasive and sometime subtle use of physical metaphors in economic and
financial discourse addressing experts and laypeople is an important issue (cf. Richardt 2003) that overlaps
significantly with the expression of alethic causal notions in predictions through the use of modal as well
as of aspectual lexicon and morphology. We hope to address this issue in a forthcoming publication.
152 Andrea Rocci

Table 6.5 Conditional restrictions for the deontic Bi in the sub-corpus

Conditional restriction of deontic Bi Tokens

Realization of the plan (including approval by a deliberative body) 13

Reportative (insiders’ disclosures, rumours) 5
Rebuttal (unless plan changes due to circumstance X) 1
Hypothesis (scenario) 1
Generic reference to a margin of uncertainty 1
Unclear, indeterminate 1
Total 22

When Bi is deontic, the saturation of the conditional restriction C is as in

Table 6.5.
With a deontic Bi, the most frequent non-factual default is represented by plan
realization (If everything goes according to plan), which also includes the official
sanctioning of the plan by a deliberative body. The plans that we find with DOV-
REBBE are not the content of an official corporate disclosure or publicly available
schedules. Rather, they correspond to inside information reported from unnamed
sources and to rumour. In some cases this origin is made partially explicit by phrases
such as according to rumours or if rumours are confirmed. Thus, in these occur-
rences, the prejacent follows from a set of rumoured plans (Bi) and the non-factual
assumption that the rumours are true (C)—the truthfulness of what is said being a
(Gricean) normal condition of communication. Furthermore, plans can involve
alternative scenarios so that the prejacent follows from the set of plans (Bi) and
the non-factual assumption that one of the alternative scenarios addressed in the
plans is indeed the case (C). Finally, exceptional circumstances might dictate that the
original plans are abandoned or revised. This gives rise to a conditional restriction in
rebuttal form, similar to those found with alethic Bi.

6.5.5 Conversational backgrounds and evidential bases of predictions: a summary

The exploration of conditional restrictions of DOVREBBE can provide a deeper
insight into the evidential and discourse functioning of this modal construction
in prediction. Both Bi and C point to premises of different kinds that are to be
recovered anaphorically in discourse context, extracted from subordinate clauses of
different types and even from sub-clausal elements such as abstract nominals
(cf. Hobbs 2010 on clause-internal discourse relations). Thanks to the constraints
on the saturation of Bi and C, DOVREBBE ends up pointing to a limited set of
combination of argument schemes (or topoi; cf. Rigotti and Greco Morasso 2010)
supporting the prediction: either you have a causal argument scheme, possibly
combined with an argument from authority based on expert opinion, or you have
Modal conversational backgrounds and evidential bases in predictions 153

an argument based on commitments, that are further grounded in the authority of

the insiders’ position to know. Two distinct evidential roles of reports emerge: the
opinion of the experts and the insiders’ position to know about the plans. In both
cases the recourse to evidence from reports is combined with, rather than alternative to,
the recourse to inference.
It is important to stress that the weak presumptions in C, corresponding to
defaults, normal conditions, do function as premises in the reasoning supporting
the prediction. Contrary to the conversational background of other epistemic mod-
als, these presumptions are not directly associated with the beliefs of the speaker at
the moment of utterance, but rather with a generic epistemic community, a social-
ized doxa from which the speaker can distance himself more or less markedly.26
It could be tempting here to evoke two notions that play an important role in
Aristotelian rhetoric in the characterization of rhetorical premises: the notion of
eikòs, that is what is to be expected because it happens most of the time, and the
notion of endoxon, that is what the majority or the experts believe.

Theorists of linguistic polyphony in the French tradition (cf. Nølke et al. 2004: 39) would call
ON—from the French impersonal third person pronoun on—this kind of collective epistemic subject
distinct by the speaker at the moment of utterance. They distinguish two variants of ON, or rather two
poles to which actual utterances may be more or less close. On the one side there is the viewpoint of a
concrete collective of individuals in the discourse world, while on the other side there is indeterminate ON
of the doxa, of the idées reçues. It appears that the conditions C of DOVREBBEE can be of either type.
This page intentionally left blank
Part III

Cognition and Metaphysics of Time

This page intentionally left blank

Experience, thought, and

the metaphysics of time


7.1 Introduction
In this chapter I shall give what may be a rather unexpected point of view on the
philosophical debate concerning the reality of tense and the related notion that time
passes, and its bearing on the correct semantics for tense. I shall suggest that there
can be no mental representation of objective ‘tensed’ features of reality of the kind
that might be thought to occur when we experience time passing or think of times as
past, present, or future, whether or not such features are part of mind-independent
reality. This, I hold, has important consequences for metaphysics; but (as will be
most relevant to this volume) it is also likely to have important consequences for a
correct semantics for tense. In a nutshell, no correct semantics for tense can treat
what philosophers call ‘A-properties’ (such as real pastness, presentness, or futurity,
as explained below) as semantic values.
The major philosophical debate over the metaphysics of time is between various
version of the A-theory (also known as the tensed theory) of time and the B-theory
(or tenseless theory) of time. The ‘A’ and ‘B’ designations refer to the A-series and
B-series described by John McTaggart (1908, 1927). These are two different kinds of
time series. In an A-series, one time is present while other times are ordered by their
degrees of pastness or futurity. Pastness, presentness, and futurity are seen as irredu-
cible properties of times; they are not relations. I shall follow standard usage and call
these properties A-properties; though (perhaps confusingly) they are sometimes also
referred to by philosophers as tenses, or tensed properties. A B-series, by contrast,
involves no A-properties; times (or events) are ordered by the relations of being
earlier than, being later than, or being simultaneous with, but no time is objectively
past, present, or future (for ease of exposition I shall tend to gloss over the distinction
between times and events in what follows, though for some metaphysical purposes it
does matter).
158 Simon Prosser

It is often held to be part of our ‘common-sense’ view of the world that times have
A-properties, and that these change as time passes. Thus, as I write, the publication
date of this article is future but, by the time you read it, that same publication date
will have been momentarily present and is now becoming ever more past. McTaggart
claimed that such changes led to contradiction; no time can have more than one
A-property, yet if time passes then every time has every A-property. There has been
much debate over whether the apparent contradiction can be removed by appealing
to the fact that no time is simply past, present, and future, but instead can only be
described in more complex ways, for example by saying that my writing this line is
now present, but will be past (i.e. in the future) and was future (i.e. in the past).
McTaggart held that change (understood as above) was essential to time. Conse-
quently his somewhat radical response to his alleged paradox was to conclude that
time is unreal. Few have followed him. Instead, philosophers have divided over
whether change, as construed above, is essential to time. A-theorists answer in the
affirmative; according to the A-theory, there is real, dynamic change (or temporal
becoming) and, at least on some versions of the A-theory, time should be understood
as an A-series that is in constant flux as time passes. The B-theory, by contrast,
denies that there is dynamic change of the kind described by the A-theorist.
According to the B-theory, times are ordered only as a B-series; no time is objectively
past, present, or future, and the apparent passage of time is an illusion. Nothing
really changes in the dynamic, A-theoretic sense; B-theoretic ‘change’ consists
only in the world being in one state at one time and in another state at another
time—much like variation across space, where there is one state of affairs in one
place and another state of affairs in another place.
The A-theory comes in several variants. The traditional ‘moving spotlight’
A-theory, according to which the present is like a spotlight that ‘moves’ along a
time series all parts of which are equally real, is no longer popular, perhaps in part
because of difficulties in responding to McTaggart’s paradox (though this is contro-
versial). An alternative, though still with relatively few advocates, is the ‘growing
block’ theory according to which reality consists of all past and present times, while
the future is seen as ‘open’, and not yet real.1 As future times become present,
more of reality comes into existence; the present is the ‘edge of becoming’. There are
several other variants of the A-theory, and I shall not attempt to list them all. The
most popular version of the A-theory today, however, is presentism, according to
which reality consists entirely of the present, and the past and future do not exist.
Although many B-theorists have regarded presentism as a bizarre view, it must be
said that it has a certain plausibility in so far as it captures the common-sense view

The moving spotlight theory was discussed, though rejected, by C. D. Broad (1923), who instead
advocated a growing block theory. For a more sympathetic recent discussion of the moving spotlight
theory see Skow (2009), and for a more recent growing block theory see Tooley (1997).
Experience, thought, and the metaphysics of time 159

that although Socrates once existed he does not exist now (not even in a remote,
inaccessible region of space–time), and the first child to be born in the twenty-
second century will exist, but does not exist yet (again, not even in a remote,
inaccessible region of space–time).
Presentism gives us a quite different picture of the world from other A-theories;
according to presentism, nothing ‘moves’ along a series of existing times, but rather
there is a single reality whose nature changes. Strictly speaking it is unclear that
presentism should posit A-properties at all; nothing exists that could instantiate past
and future A-properties, and to ascribe presentness to the one reality is thus
redundant (though presentists do nonetheless hold that past- and future-tensed
utterances can be true, and there has been much debate over how this could be
so).2 Presentism is nevertheless usually described as an A-theory because most
versions hold that time passes and there is real, dynamic change.
One can, if one wishes, separate A-properties from change. One could define an
A-theory as any theory that posits A-properties, and a dynamic theory as any theory
according to which time passes and there is dynamic change (it is probably safe to
regard passage and dynamic change as equivalent, however, and I shall do so in what
follows). There is nothing obviously incoherent about a theory according to which
times have objective, irreducible A-properties, yet time does not pass.3 But such
views have rarely been held; it is normally assumed that the A-properties relevant to
the A-theory carry with them the dynamic notions of change and passage (even
though, as presentism shows, the converse need not be true). In any case, I shall
adopt the common practice of using ‘A-theory’ to mean any dynamic theory
according to which time passes and there is dynamic change.
In what follows I shall argue that our apparent experience of the A-theoretic
passage of time is an illusion; we cannot be aware of, or perceive, time passing. I shall
then argue that this makes the A-theory implausible and quite possibly unintelli-
gible; the discovery that the relevant aspects of experience are illusory leaves us with
no way to grasp what it is that the A-theorist is claiming to be true. If I am right, then
any attempt to account for the temporal features of experience, thought, or language
should proceed as though the B-theory were true. After outlining an argument
for this view I shall then discuss the projects that arise from it, along with some
possible steps forward. My discussion will move rather quickly, describing some of
the views and arguments in outline and glossing over many details (though more
detailed versions of some of the arguments can be found in Prosser 2006, 2007,
2012, 2013). Rather than getting into the technicalities, important thought they

For a survey of the latter debates over presentism see Caplan and Sanson (2011).
See Parsons (2002) for one possible development of such a theory. If I understand correctly, the “four-
dimensionalist tensism” described by Peter Ludlow in his chapter in this volume, involving perspectival
properties but no ‘movement’ of the now along the time line, is another.
160 Simon Prosser

are, my main purpose here is to motivate, and begin to outline, a much-neglected

project—the project of accounting for experience, thought, and language from a
B-theoretic point of view.

7.2 Is it possible to perceive the A-theoretic passage of time?

One of the main reasons—perhaps the reason—for accepting the A-theory is that it
seems to best vindicate our experience of the world. It just seems to us as though
time is passing; we seem, in other words, to be aware of time passing (I shall speak
interchangeably of being aware of something and of perceiving it; and these should
be taken broadly, with no restriction to the five standard sensory modalities). This
is perhaps most apparent in the experienced dynamic nature of change and motion,
though there also seems to be an associated sense that one is ‘moving’ through
time toward the future (though clearly ‘moving’ must be a metaphor). Numerous
A-theorists have appealed to this in defence of their view. The A-theory has often
been seen as the obvious, common-sense view, while the B-theory has been seen as
counter-intuitively denying what is manifest; and it has therefore been assumed that
the onus is on the B-theorist to explain away the features of experience that seem to
support the A-theory.
My view, however, is that although the B-theorist must indeed provide a satisfac-
tory account of experience, much of the debate over the A-theory rests on a largely
unexamined yet highly problematic assumption that the passage of time is the kind
of phenomenon that it is possible to perceive. Once this assumption is examined,
however, it becomes clear that the passage of time, or any associated A-properties,
cannot be perceived—they belong to the wrong metaphysical category to be possible
objects of perception. Consequently, contrary to the apparent nature of our experi-
ence, and whether or not time really passes, we are not aware of it passing. I shall
now sketch an argument to that effect.4
It is important to keep in mind that what is being denied is that we are aware of
the passage of time as construed by the A-theory. This does not require the rejection
of everyday thought and talk about the passage of time. The B-theorist does not deny
that everyday statements such as I am aware that a great deal of time has passed since
I first became a student can be true, but insists that they must be interpreted in
accordance with the B-theory. So, for example, a B-theorist might hold that the
everyday utterance just mentioned is true if and only if the speaker is aware that
the time at which the speaker became a student is much earlier than the time at
which the utterance occurs. (Note the use of the B-theoretic ‘earlier than’ in the truth
conditions. This is just a very rough example; different B-theorists will give different

For a fuller defence of the argument sketched below see Prosser (2013).
Experience, thought, and the metaphysics of time 161

specific semantic accounts.) The A-theorist, by contrast, should interpret the same
utterance as asserting an awareness of real passage; something not reducible to the
B-relations of ‘earlier’ or ‘later’.
Before sketching the argument I must introduce some terminology. Let the
phenomenal character of an experience be the subjective nature of the experience,
often described by philosophers as ‘what it is like’ to have the experience. Thus
the phenomenal character of a visual experience of something red is the
subjective quality of the experience that differs from that of a visual experience of
something blue.
Now, whenever one is aware of something, there is an associated phenomenal
character. When one sees a physical object, for example, there is a variety of visual
phenomenal characters corresponding to the object’s colour, size, shape, and so on.
When one is aware that one’s stomach is empty by experiencing a feeling of hunger
there is another phenomenal character, and so on. So we should expect that an
awareness of the passage of time should be associated with a specific phenomenal
character; indeed it would seem baffling if someone were to claim to be consciously
aware of time passing yet claim that there is no phenomenal character, nothing that
it is like for them subjectively, by virtue of which they have this awareness. Indeed,
the literature on temporal experience is full of vivid descriptions of this phenomenal
character.5 It need not be claimed that this phenomenal character can exist apart
from other phenomenal characters, or that one can be aware of passage without
thereby being aware of something else. It may be, for example, that time seems to
pass because of the way in which we experience change (in fact I shall suggest below
that this is at least part of the story).
To perceive something is to stand in a specific kind of relation to it. I shall call this a
perceptual relation. Is it possible for a person to stand in a perceptual relation to the
passage of time? I shall argue that it is not. Consider a person in some typical perceptual
state; they have, for example, visual experiences of a variety of objects of different
shapes and colours, auditory experiences of various different sounds, and so on. For
every feature of the world perceived by a given subject at a given time, the subject’s
experience has a phenomenal character. Different phenomenal characters allow
the subject to perceive different features; if the phenomenal character of a perception
of something red did not differ from the phenomenal character of a perception of
something blue then the subject could not discriminate red from blue, and would thus
perceive neither.
Hence, for a given subject at a given time, there is a unique (one–one) mapping
between phenomenal characters and perceptually discriminated features of the

For just a few of the countless examples of descriptions of this phenomenal character by both A- and
B-theorists, see Le Poidevin (2007: 76), Schlesinger (1991: 427), Davies (1995: 275), and van Inwagen
(2002: 64).
162 Simon Prosser

world. Now, for all perceived features other than the passage of time (or associated
A-properties) we have at least some idea of how this unique mapping can obtain.
When I see a ripe tomato in front of me, an image of a specific shape, colour, and
brightness is projected onto my retina. This stimulates only those retinal cells upon
which the image falls; and stimulates those cells differentially according to the
intensity and wavelength distribution of the light. This, in turn, produces an effect
on the configuration of my brain. A different effect would have occurred if there had
been a different pattern of retinal stimulation. My brain is thus reconfigured by the
perceptual stimulus, and the resulting brain configuration either is, or at least causes,
a perceptual experience with a given phenomenal character. A different brain
configuration would have produced an experience with a different phenomenal
character. Consequently, different perceived features produce different phenomenal
characters, at least for a given subject at a given time. Something similar is true for all
known forms of perception. It is not plausible that unique mappings of this kind are
brute facts; they occur for the kinds of reasons just described.
My argument thus takes the form of a challenge to the A-theorist to explain how
the passage of time could map uniquely onto a specific phenomenal character. In
other words, if the passage of time is experienced, it must be explained what it is that
makes the experience in question an experience of the passage of time, rather than of
something else; and it must be explained why no other experience (i.e. no experience
with a different phenomenal character) constitutes an awareness of the passage of
time. I think that there are good reasons to doubt that this challenge can be met. The
effect of A-theoretic features (A-properties or the passage of time) on the physical
world, in so far as they could be said to have any effect at all, would not be like the
effect of one part of the physical world on another. The ripe tomato in front of me
has a differential effect on the configuration of my brain, as described above. But the
A-theoretic passage of time, or indeed the event of a time acquiring one A-property
rather than another, would not have any comparable effects on the configuration of
my brain; and consequently no comparable, differential influence over the phenom-
enal characters of my experiences. At best it might be argued that passage is an
essential element of all physical causation; but this would apply to all phenomenal
characters equally. In the light of this, it seems very hard to see how an A-theoretic
feature of the world could stand in a unique perceptual relation to an element of my
experience. It is therefore very hard to see how the phenomenal character that we
associate with the awareness of passage could really be an awareness of passage.6
Unless some satisfactory response can be given we must take it that whatever the
subjective nature of experience, it does not constitute an awareness of time passing.

Whether veridical or illusory, the phenomenal character is, of course, an awareness of something, and
the B-theorist should explain what this is. For my own thoughts on this see below, Section 7.4, and also my
Experience, thought, and the metaphysics of time 163

This signals big trouble for the A-theorist; not only does experience fail to provide
any support for the A-theory but, as I shall argue in the next section, there is a
serious danger that the A-theory is rendered unintelligible.

7.3 Can A-theoretic properties figure in thought or language?

There has been much debate about the semantics of temporal indexicals and of
linguistic tense, and sometimes these discussions have been linked to debates over
the metaphysics of time. There are two different kinds of approach that one can take,
in this regard. First, on the more conservative approach, one can consider what the
implications would be for linguistic semantics if one or other metaphysical theory of
time were correct.7 Thus, to mention one rather specific example, one might be
interested in the extent to which I, here, and now are interdefinable. One might, for
example, hold that here is the place where I am now; and I am the person who is now
here. This is true in so far as I can only be in one place at a given time. But it is less
clear whether now can be defined in terms of I and here. According to all theories
other than presentism, I exist at more than one time. In that case, now cannot simply
be the time at which I am here, for I may be here at many times. If I am made up of
temporal parts, then we can at least say that now is the time at which this temporal
part of me is here (though the reference to here would be redundant if my temporal
parts are located at their times essentially). On the other hand, if presentism is true
then I only exist at one time (now), so now is simply the time at which I exist; or,
more succinctly, now is the time at which I am, just as, given presentism, here is the
place at which I am.
The second, bolder approach tries to draw metaphysical conclusions from the
semantics. Thus, historically, debates between A- and B-theorists sometimes focused
on whether B-theorists could give tenseless translations of tensed utterances, though
for the last thirty years or so the emphasis has been more on whether B-theorists can
give tenseless truth conditions, rather than translations, for tensed utterances. This
has been an important part of the defence of the B-theory. Attempts to give positive
arguments for metaphysical theories from semantic considerations have, however,
proven more difficult. Moreover, drawing metaphysical conclusions from linguistic
form strikes many philosophers as a perilous exercise.8
Rather than add directly to these debates, however, I would like to draw attention
to what I think is a questionable assumption behind any approach that either
assumes the A-theory or attempts to defend it in the above ways. This is the

See, for example, Ludlow (1999) for discussion of the implications of presentism for the semantics of
For an example of an A-theorist drawing metaphysical conclusions from the nature of language, see
Smith (1993). For scepticism about arguments of this kind, see Dyke (2007).
164 Simon Prosser

assumption that A-properties, or the passage of time, could figure in the semantic
values of linguistic utterances. If the above argument concerning the perception of
passage is sound then genuine A-theoretic features of time are not reflected in
experience in any way. This raises a problem. To put this in very simple terms: if
the features of time posited by the A-theory cannot be perceived, then this raises a
prima facie concern about whether such features can be referred to. One might
perhaps give what appears, from the point of view of linguistics, to be a perfectly
satisfactory semantics for tensed expressions, or for sentences involving passage
(such as the everyday sentence mentioned above, or expressions such as a year has
gone by . . . ), by assigning A-properties as semantic values. Provided one pays
attention only to the assigning of semantic values to expressions in the construction
of a semantic theory, and one does not think about how such entities could get to be
the semantic values of expressions, the theory might appear acceptable. But, if what
I am suggesting is correct, A-properties could never be semantic values, and any
such semantics would therefore be incorrect.
There are, of course, many imperceptible entities to which we can refer. We refer to
abstract objects (such as shapes and numbers), to future events and objects, and so on.
But arguably in each such case it is possible, at least in principle, to state some
condition that is uniquely met by the entity referred to. Thus, for example, no one
has seen a perfect circle or square, but we know how to define those shapes; and no
one has seen the first film to be shown in the twenty-second century, but we know
what it would take for something to be that film. Hence, when faced with the question
‘Which entity are you talking about?’, we have some idea of how to give an answer.
When faced with the question ‘Which features of time are you talking about?’,
what can the A-theorist say about putative A-theoretic features of time such as
passage or pastness, presentness, or futurity? If the argument given above is sound
then the A-theorist cannot answer that they are the features with which we are all
acquainted in experience. Neither can they be identified as the features referred to by
tensed linguistic expressions, for we wanted to know which features those are.
Neither would it really help to describe the A-series and explain the role of, say,
pastness in this series relative to presentness and futurity. For this would only tell us
that there are some properties of times that can be arranged in a specific temporal
order; it still would not tell us which properties they are. What goes for language
goes for thought; if we cannot refer to A-properties or passage then we cannot think
of them either.
I expect these claims to be met with incredulity by many. It is very tempting to
find it obvious that one knows, in one’s own case, what one is thinking of, or talking
about, when one thinks or talks of A-properties or passage, and consequently it is
tempting to think that arguments such as that given above need not be taken
seriously. If so, however, one should examine the source of one’s supposed under-
standing of these notions. Although I have argued that the A-theoretic passage of
Experience, thought, and the metaphysics of time 165

time cannot be perceived, I have not denied that there are features of experience that
give rise to our illusory sense of time ‘passing’. The danger, I think, is that when one
thinks of passage, or the associated A-properties, one merely performs an off-line
simulation of passage experience (just as, arguably, one neurologically simulates red-
perception when asked to think of the colour red, or cat-perception when asked to
think of a cat). But if passage-experience cannot be a genuine perception of passage,
then putative A-theoretic concepts formed in this way fail to have genuine
A-theoretic properties as semantic values.
If all of this is correct then the A-theory is unintelligible; it cannot even be made
clear what the A-theorist claims about the nature of time. Different ontological
claims, such as the presentist claim that only the present exists, may be clear enough;
but the ontological claims alone, with no notion of passage, do not seem plausible. In
any case, it is the A-theoretic claims that are my target here, rather than the
ontological claims associated with specific versions of the A-theory.

7.4 Experience and the illusion of passage

If the conclusions of the last two sections are correct then the nature of temporal
experience, along with temporal thought and the meanings of utterances containing
terms such as past and future, must all be accounted for without appeal to A-
properties or the passage of time. This is also largely true if the B-theory is accepted
for any other reason. So those who reject the A-theory must embark on a substantial
project, to explain the features of experience, thought, and language that have made
the A-theory seem appealing. This project has been rather neglected by B-theorists,
perhaps because, despite its obvious importance for metaphysics, it is a project
concerning mind and language, not metaphysics. In the rest of this chapter I shall
sketch some possible steps forward in this project. As well as any interest these steps
might have in themselves, however, I hope that what follows will indicate something
of the nature of what is, in my view, an important and much neglected project.
In this section I start by considering temporal experience.
The question before us, then, is why conscious experience has a phenomenal
character that seems to lend support to the A-theory even though, if the above
arguments are correct (or if the B-theory is anyway true), experience does not
involve an awareness of the passage of time. Giving a full answer to this question
is a substantial project; not only must the phenomenal character of temporal
experience be accounted for, but the way in which this phenomenal character
interacts with our thought about time must also be explained. Here I shall do far
less; I shall make a methodological proposal, then I shall put forward a suggestion
regarding just one piece of the puzzle.9

For more details, see Prosser (2012).
166 Simon Prosser

The methodological proposal derives from recent work in the philosophy of

perception. In recent years intentionalism has become an increasingly popular
view. According to intentionalism, all experiences have representational contents
(they represent the world as being one way rather than another, and are correct or
incorrect according to whether the world is as represented). Most intentionalists
hold, further, that for a given subject at a given time the phenomenal character of
the experience is determined wholly by its representational content. Consequently,
I suggest, if we wish to explain why an experience has the phenomenal character that
it has, it will be sufficient to state the representational content associated with that
phenomenal character and to explain why it has that content. This is, in fact, more
or less what cognitive scientists usually do when explaining illusions. If, for example,
one wishes to explain why it is that, in the Hering illusion, one experiences a visual
phenomenal character of the kind most naturally described as being of two curved
lines (which occurs when looking at two straight lines superimposed upon a
background of angled lines of various sorts), then one starts by stating that the
experience represents two curved lines. One then sets out to explain why the lines
are represented as curved, given that they are in fact straight. The assumption that
the illusion is to be explained in terms of representation (or, rather, misrepresenta-
tion) is so automatic as to easily go unnoticed. There is, of course, a deeper question
in the philosophy of mind concerning why an experience with a given representa-
tional content has the phenomenal character that it has, or indeed why it has a
phenomenal character at all; but one need not answer this kind of question in order
to explain illusions such as the Hering illusion, and I suggest that the same is true for
the illusion of time passing. There will, of course, be some mechanism that accounts
for the occurrence of the illusion, just as there is for the Hering illusion; for the latter,
this will concern facts about the way in which straight lines interact with the
background upon which they are superimposed. But before we can consider such
a mechanism for the illusion of passage (which would, in any case, be a job for
empirical cognitive science) we must first know the representational content of the
illusion; just as, before looking for a mechanism for the Hering illusion, we must
know that the illusion consists in the two straight lines being represented as curved.
What, then, might our experiences represent, such that time seems to us to pass?
The obvious answer would seem to be that experience represents that time passes,
but does so falsely. If the B-theory is true, however, then I do not think this answer
can be correct. One reason is that if our thoughts or language cannot have genuine
A-theoretic features as semantic values, as argued above, then it seems implausible
that our experiences could have those semantic values either. But we can cast further
doubt on the experiential representation of A-theoretic features by noting that most
B-theorists hold not merely that the A-theory is false, but that it is necessarily false
(i.e. there is no possible world in which it is true). I haven’t the space for a full
defence of the claim here, but I think that there are reasons for doubting that a
Experience, thought, and the metaphysics of time 167

necessarily uninstantiated but non-compositional feature such as the passage of time

could be represented by an experience.10
Rather than dwell on arguments against the experiential representation of tem-
poral passage, however, I shall instead put forward a tentative positive hypothesis.
There are two main philosophical accounts of the persistence of objects through
time. According to one view, objects persist by perduring; that is, by being made up
of temporal parts that are located at different times. According to the other view,
objects persist by enduring; that is, by existing in their entirety at each time at which
they exist at all, and thus have no temporal parts. Broadly speaking, while there are
certainly exceptions, B-theorists tend to be perdurantists while A-theorists, and
especially presentists, tend to be endurantists.
I think that the association between endurance and the A-theory is a very natural
and intuitive one, and I think it may hold the key to at least one part of our puzzle.
Although obviously metaphorical, our notion of temporal passage seems to involve
ourselves, and perhaps other objects, ‘moving’ through time, and persisting through
changes, while remaining one and the same entity. The idea of having a part in state
S1 at time t1, and a part in state S2 at time t2, somehow falls short of capturing the way
we imagine a change between states S1 and S2. We imagine change as involving the
very same thing being in one state, then another; not merely a succession of different
entities, with different properties, that happen to compose to form a temporally
extended whole. So our intuitive, A-theoretic notion of change seems better captured
in terms of the endurance theory than the perdurance theory. And, similarly, when
we think of time passing or, equivalently, of ourselves or other objects as ‘moving’
through time towards the future, we imagine the very same object being at one time,
then at another. The idea of a single identity being at a succession of different times
seems somehow essential to our notion of ‘moving’ through time.11
My hypothesis, then, is that part of the explanation for our natural inclination
towards the A-theory is that, whether or not objects really endure, our minds have a
tendency to represent objects, perhaps including ourselves, as enduring. I doubt very
much that the hypothesis that I have described provides a complete explanation of
the way we experience time as passing; if nothing else, it tells us very little about the
experienced direction of time. But it may at least help.

Again, for a more detailed argument, see Prosser (2012).
David Velleman (2006) has made a similar suggestion about the importance of our idea of ourselves
as enduring in explaining the illusion of time passing; as he puts it, in order to have a sense of the future
coming towards us, there has to be a ‘fixed point’, whose identity does not change, for the future to come
towards. The idea that something must always persist through a change was put forward by Kant (see
the First Analogy in his 1929 [1781–1787]), though there is no clear reason to think that he had the
endurance/perdurance distinction, or anything like the A-/B-theory debate, in mind. See also Merricks
(1995, 1999) for arguments that presentism and endurance are natural partners; though see also Parsons
(2000) for one way that endurance might be reconciled with the B-theory. For some ideas about
experience related to those described here, see Paul (2010).
168 Simon Prosser

Here I can only speculate about the mechanism that gives rise to the representa-
tion of endurance, and about whether our minds could have represented the world
differently. Perhaps, for the reasons of computational economy that I describe below,
this was the only way our minds could have represented objects over time. Or
perhaps there are possible alien creatures whose minds would not represent objects
in this way, and whose experience of time would therefore be very different from our
own. Perhaps they would experience the world as B-theoretic. It is hard for us to
imagine what this would be like; our difficulty may only reflect our being locked
within our own way of experiencing time due to contingencies of our cognitive
architecture, or it may reflect a deeper necessity to our own way of experiencing
time. There is much more to be said about these issues; I shall not attempt to settle
them here, though I shall make some further tentative suggestions.
Let us suppose that experience does indeed represent objects as enduring rather
than as perduring. It is necessary to make it clearer what this amounts to. What
would be the difference between a mental representation of an object as enduring
and a mental representation of it as perduring? After all, endurantists and perdur-
antists agree that it will be the same object at different times. The two theories do
disagree concerning whether what is represented as occupying a particular time is
the whole object or only a part of the object. So, when an object is seen, for example,
the endurantist should say that the object is seen in its entirety whereas the
perdurantist should say that only a part of the object is seen.
Our question concerns not the real nature of the perceived object, but the nature
that it is represented as having. How could we settle the question of whether
experience represents objects as enduring or perduring? I do not have a knock-
down argument, but offer the following in support of the plausibility of a hypothesis
that should ultimately be judged by its role in an overall explanation of our
experience of time. My suggestion is that human perceptual systems are lazy—they
tend to represent the world in the simplest, most computationally economical way
that helps the perceiver negotiate the perceived environment. An example of this is
the way we perceive solid objects. Solid objects are, as we know, made up of many
small particles (for example, atoms in a crystalline structure); but their surfaces look
solid and continuous. Consider a visual perception of an object with no discernible
parts, with no visible surface texture or patterns. Does visual perception represent
the object as a single individual (with no parts), or as an object with parts, or is it
neutral between these possibilities? We must be careful here; no parts are repre-
sented, but an absence of a representation of parts does not entail a representation
of the absence of parts. Nevertheless I shall argue that something close to this might
be true.
It does not seem correct to say that the object looks (to the subject) as though it is
divisible when no parts are perceived. Provided they had the same macroscopic
properties, a divisible object and an indivisible object need not look any different
Experience, thought, and the metaphysics of time 169

from one another; so there seems no reason to think that a divisible object with no
discernible parts is represented as divisible. One should be careful not to conflate
what is represented by the phenomenology with contents available to the subject in
thought, however; someone who knows that an object is made of a fragile material
may well say that the object looks as though it could be smashed, and therefore looks
divisible. But since an indivisible, unbreakable object need look no different, it is not
clear that the phenomenology represents the object as divisible in such cases.
Perhaps one might focus attention on just one spatial region of an object, treating
this region as an entity distinct from other regions of the object, and one may thus
perceive the object as having parts (the temporal analogy would be that someone
who understands the perdurance theory might focus attention on what they cur-
rently perceive and, understanding that what they see is just a temporal part,
separate this in thought from other temporal parts of the object). But, again, it
does not follow that the object looks as though it has parts (in either the spatial or
temporal case), but only that one can pay selective attention to what one takes to be a
part of the object. Note that an object does not have parts just in virtue of occupying
distinct regions of space; the notion of a spatially extended object with no parts is
perfectly coherent.
None of these considerations seems decisive. But rather than opt for the conclu-
sion that objects are neither represented in perception as having parts, nor repre-
sented as lacking them, I suggest that the most plausible story is that one’s
perceptual system forms a representation of an object by combining simple repre-
sentations of features such as its shape, colour, etc., and that the simplicity of these
representations suggests that the object itself is represented as simple. There has, of
course, been great progress in the empirical study of object perception and it would
be instructive to go into some details of this, but for now it will suffice to note that
plausible accounts tend to describe the visual system as building representations
from simple elements—regions of space are marked as straight edges, or as surfaces,
or a whole two-dimensional plane is simply marked as having a given colour, and so
on. When faced with a surface with no discernible texture, for example, it seems
plausible that the surface is marked as flat, rather than as having an unknown status
between perfect flatness and currently invisible texture. An oversimplified repre-
sentation—the surface is flat and continuous—makes sense from the point of view
of computational economy. So my suggestion is that on the whole, where no
structure is explicitly represented in experience, the object is represented as un-
structured. In the temporal case, I assume that there is just a representation of an
object—perhaps via some kind of mental file, whose contents are updated as
changes are observed—and that since no temporal parts are represented, we should
view this as a representation of an object with a constant identity, and no parts. To
repeat, I do not claim to have shown this decisively, but put it forward as the most
plausible hypothesis.
170 Simon Prosser

I have claimed that one part of the reason why time seems to pass is that objects
(perhaps including ourselves) are represented in experience as enduring through
changes. But perhaps it will not yet be clear why this kind of representation should
make the world seem to us to be an A-theoretic world, incompatible with the B-
theory. If the B-theory allows the possibility of enduring objects (as is sometimes
claimed), then it is unclear why a world of enduring objects should not strike us as
compatible with the B-theory. The answer, I suggest, lies in the way change is
represented. Again, I shall have to gloss over some details and leave many possible
objections unanswered here, but the general idea is as follows.12 When an object, O,
is observed as first having a property F, then no longer having that property, there is
a representational content at t1 of the form ‘O is F’, followed by a representational
content at t2 of the form ‘O is not F’. These representations contradict one another.
Moreover, in order for change to be represented it is not sufficient that there merely
be a change in what is represented. The change itself must be represented; and,
arguably, this requires a single representation that includes both of the conflicting
contents.13 In that case there is a representation with a contradictory content; a single
object is represented as both F and not F. If the world is represented in a way that is
in fact contradictory it is not surprising that that the world should seem to us quite
unlike a B-theory world, which holds no obvious inconsistencies. The world thus
represented does not seem contradictory, of course. I speculate that this is because in
addition to our taking the world to be one in which objects endure, we are also
inclined to take the world to be much as the presentist thinks it is, where each
moment in time constitutes a complete reality, so that different moments in time are
distinct realities which cannot conflict with one another.
We thus arrive at the idea that experience represents the world as one in which
objects endure through changes while retaining their identities (that is, an object
remains the very same object); the object exists though a succession of differing
realities and we, as objects, move through time in the same way. This, I suggest, is an
important part of the explanation of time seeming to pass, though I stress that there
is more to add, at the very least in order to capture the experienced direction of time.
There will, no doubt, be many objections to the above arguments, not least in

For full details of possible objections and the relation of the argument to debates over the problem
of temporary intrinsics, see Prosser (2012).
The idea that a change of representation is not sufficient for a representation of change is one of the
chief motivations for the specious present, the doctrine that our experience of the present encompasses an
extended period of time. For present purposes I need take no view on whether there is a conscious
experience of an extended period or a consciousness of the present moment combined with a short-term
memory (in fact I am sceptical about whether there is a clear distinction between these options, despite
frequent discussion of them in the literature on time consciousness). For more details on the specious
present and its rivals, see James (1890), Broad (1923, 1938), Dainton (2000, 2001, 2008), Gallagher (2003),
Tye (2003), Kelly (2005b), Le Poidevin (2007), and Phillips (2010).
Experience, thought, and the metaphysics of time 171

relation to the claim that the representation of change in enduring objects is

contradictory. There is some considerable debate among metaphysicians over
whether the possibility of change makes endurance contradictory (see Lewis 1986
for an argument for perdurance based on the supposed contradiction), and many of
the moves made there might seem applicable to the question of whether the
experiential representation of change is similarly contradictory. For responses to
some such moves see Prosser (2012).

7.5 Thought, language, and the A-theory

Our project of explaining away the appeal of the A-theory goes beyond the explan-
ation of those aspects of phenomenology that make time seem to pass. It is also
necessary to explain a number of features of our thoughts about time, and of related
linguistic representations. Ultimately, what we say about thought and what we say
about experience must be integrated; the relation between our experiences and our
thoughts must be explained. Here I shall touch on just one important issue relevant
to this.
One of the best-known puzzles concerning temporal thought is Arthur Prior’s
(1959) ‘thank goodness’ argument. Suppose that at time t2 you have an appointment
at the dentist for painful root canal surgery. At time t1, the day before the surgery,
you experience feelings of dread when you think my root canal is tomorrow. But
at time t3, the day after the surgery, you experience feelings of relief, and think
thank goodness my root canal is over! Now, if the B-theory is true then the facts don’t
change. It is eternally true that your utterance of my root canal is tomorrow occurs
one day earlier than the appointment, and it is eternally true that your utterance
of my root canal is over occurs one day after the appointment. The two most popular
B-theoretic accounts of the semantics of temporal indexical terms like tomorrow and
over analyse their contributions to truth conditions in terms of relations to either
tokens or times. Thus, on a token-reflexive account, the utterance my root canal is
tomorrow is true if, and only if, my root canal occurs one day later than my uttered
token of tomorrow; whereas, on a ‘date’ account, my utterance is true if and only
if my root canal occurs one day later than t1. But why, Prior asked, should it
make sense to dread, or thank goodness for, states of affairs such as those? Why
should I thank goodness, at t3, for the fact that my utterance occurs a day
later than the root canal, when this was equally true at t1 (the time at which, in
fact, I dreaded the root canal)? Prior concluded that the B-theorist could provide no
satisfactory answer, and that this phenomenon provided support for his version of
presentism. Given an A-theory, perhaps it makes sense to thank goodness for the
root canal event being past; and it certainly makes sense to thank goodness for
the root canal event not existing (though it must be noted that the presentist who
adopts the latter explanation must then explain why it doesn’t make sense to
172 Simon Prosser

thank goodness prior to the root canal, given that (according to presentism) it
doesn’t exist then either).14
The correct response for the B-theorist lies, I think, in recognizing that truth
conditions can be stated in different ways for the purposes of different explanatory
projects. Consider, by analogy, an utterance U, said or thought by me, of there is a
ferocious tiger near here. Suppose that U occurs at location L. Here are three ways to
state the truth conditions of U:
U is true if, and only if:
1. There is a ferocious tiger near to location L.
2. There is a ferocious tiger near to where U occurs.
3. There is a ferocious tiger near to where I am.
It makes sense for me to fear the state of affairs expressed by U. But clearly the
reason for this cannot be captured by either (1) or (2); neither the fact that a
ferocious tiger is near some particular place, nor the fact that a ferocious tiger is
near to where an utterance occurs, are in themselves reasons for fear. It is only
(3) that really gives me a reason for fear. There may be explanatory purposes for
which (1) or (2) give better accounts of the truth conditions of U; but when we want
to explain the subject’s fear, (3) captures the truth conditions in the right way.
In case it should be objected that it is (1) or (2) that captures the truth conditions
of U and that (3) merely indicates a fact that the subject can infer from ‘I am the
utterer of U’, it should be noted that there are some utterances containing indexical
expressions for which it would be hard to state truth conditions without reference to
the subject. Consider, for example, Uleft: a ferocious tiger is to the left. Since places
and utterance tokens do not have left-hand sides it is impossible to state truth
conditions for this utterance in the same manner as (1) and (2). In many cases it is
the speaker or thinker’s own left-hand side that is relevant to the truth conditions of
Uleft; this would invariably be the case when Uleft is a token thought. So there are
many cases in which the truth conditions are more usefully stated, and some cases in
which they have to be stated, in what we might call person-reflexive terms.
If the B-theory is correct then monadic predicates such as is past or is future are
used in dealing with relations rather than properties. The ‘thank goodness’ problem
only bites if the truth conditions for utterances containing these predicates must be
given by the date or token-reflexive accounts, which are analogous to (1) and (2).
If, instead, we adopt a person-reflexive account a solution is possible. Once again,
however, the metaphysical issues interact with the semantic issues. Consider again
my utterances of my root canal is tomorrow at t1 and my root canal is over at t3. Call
these U1 and U3 respectively. If persons endure then the person-reflexive truth
condition for U1 would have to be along the lines of ‘the root canal occurs the day

See also Jaszczolt (this volume) on the symmetry problem.
Experience, thought, and the metaphysics of time 173

after the time at which I exist’ and for U3 it would have to be along the lines of ‘the
root canal occurs prior to the time at which I exist’. But if the B-theory is true then
these are inadequate, for if I endure, and the B-theory is true, then I exist at many
times. This seems to provide a further reason why representing objects (including
persons) as enduring might make the B-theory seem false, pushing our notion of
time towards an A-theory.
If, instead, we assume that persons have temporal parts, however, then we can give
truth conditions for U1 and U3 in terms of these parts. Suppose I have temporal parts
S1 and S3 at times t1 and t3 respectively. Then U1 is true if and only if the root canal
occurs the day after S1 exists, whereas U3 is true if and only if the root canal occurs
earlier than S3’s existence. S1 and S3 would have to understand these truth conditions
in a first-person manner; otherwise they would state facts that could equally be
accepted by S1 and S3. Given that causation is always directed towards the future,
there might be good reasons for S1 and S3 to differ in their attitudes toward the root
canal. S1, being located prior to the root canal, still has a chance to prevent it;
whereas S3, being located after the root canal, can do nothing about it. So if the truth
conditions are stated in person-reflexive terms that make reference to temporal parts
of persons, then we can make sense of the differing attitudes of the different
temporal parts of a person to the same event. In effect, Prior’s challenge to the
B-theorist was to state a fact (a truth condition) that S1 had reason to dread, and a
fact (a truth condition) that S3 had reason to thank goodness for. We can answer that
challenge by saying that the fact that S1 is located earlier than the root canal gives S1
(but not S3) a reason for dread; whereas the fact that S3 is located later than the root
canal gives S3 (but not S1) a reason for relief.15
Perhaps if words like past and future wore their semantics on their sleeves, such
that we could not fail to recognize that they imply relations to temporal parts
of persons, we would find the B-theory easier to accept. The A-theorist who posits
A-properties thinks that such words do wear their semantics on their sleeves;
according to them the predicate is past ascribes the A-property of pastness. It may
be that the surface form of these predicates naturally helps mislead us into supposing
that there are A-properties of this kind, thus lending further intuitive support to the
A-theory.16 But the B-theorist should hold that the semantic properties of is past are
analogous to those of is to the left; in terms of surface form they are one-place
predicates, but we use them to deal with two-place relations to temporal parts of

See Maclaurin and Dyke (2002) for a similar claim about the differing attitudes of temporal parts.
Smart (e.g. 1967) suggests something of this sort, as do I in Prosser (2006) (where I also discuss
temporal parts of persons as unarticulated constituents of temporal indexical thoughts).
In Prosser (2006) I describe this in terms of John Perry’s (1986) notion of unarticulated constituents.
For an entirely different approach to explaining away A-theoretic intuitions in terms of linguistic
semantics, see Jaszczolt (2009).
174 Simon Prosser

7.6 Conclusions
I hope that the above arguments, brief though they were, suffice to show something
of the different challenges faced by the A-theorist and the B-theorist in giving a
satisfactory account of temporal experience, thought, and language. The major
challenge for the A-theorist, as I see it, is to explain how the features of experience
that are so often used to motivate the A-theory could possibly constitute an
awareness of A-theoretic features of time such as temporal passage. I have briefly
described a reason for doubting that this can be done. Failing this, A-theorists must
explain how to motivate their theory, and indeed how to render it intelligible,
without recourse to the nature of experience; and I have suggested, again, that this
looks like a very hard challenge to meet.
The major challenge for the B-theorist, by contrast, is to explain away the features
of experience to which the A-theorist appeals, and to give a satisfactory B-theoretic
account of those elements of thought and language that might also seem to support
the A-theory, such as the differing attitudes we take to the same events at different
times. These attitudes must be integrated with an account of the contents of
experience. There are many further related questions that have not been touched
on above; for example, the question of whether perceptual contents have to be
present-tensed in order to motivate actions, and the question of how the contents
of perception (whether or not explicitly present-tensed) subsequently come to be
represented in memory with constantly updated degrees of pastness.
My suggestion that objects, including persons, are represented in experience, and
often in thought, as enduring rather than perduring, may offer a small step forward.
But what I hope to have shown overall is something of the variety of interlocking
issues that must be addressed in a satisfactory account of temporal experience,
thought, and language.



Tensism is the doctrine that tense (like past and future tense) is not eliminable and
not reducible to non-tensed facts. We can distinguish two kinds of tensism—
metaphysical tensism and linguistic tensism. Metaphysical tensism is the doctrine
that reality itself is tensed and that these tensed facts are not reducible to more basic
untensed facts. Linguistic tensism is the idea that natural language is tensed and that
natural language tense is not reducible to more basic untensed facts about language
(or anything else).
Sometimes people think that advocates of metaphysical tensism are mistakenly
reading a feature about language into the metaphysics of reality. They might say that
“well, language is tensed, but reality isn’t; you are just projecting a feature of
language onto the world”. But this isn’t what is happening. Different languages of
the world have different ways of expressing tense. Some of them use what we call
tense markers, others use their aspectual system to express tense, some languages use
evidential makers to express tense.1 Some languages use modals to express tense
(English does this to express future tense, for example), and some use spatial
predicates. The point is that tense is not a surfacey feature of the world’s languages;
advocates of tensism (let’s call them tensers) think that tense is a feature of the world
and that different languages find different strategies for expressing claims that
involve this feature of the world. That is, metaphysical tense comes first. Linguistic
tense is what we use to express metaphysical tense.
Accordingly, when we say we are studying tensed features of language, we are not
talking about surfacey properties of language. We are talking about the proper
semantics of tensed expressions—we are talking about how tensed language hooks
up with the world.2 We are attending to the linguistic expression of these claims

For example, an evidential marker that indicates that the speaker has perceptual knowledge of an
event is sometimes used to express past tense.
Could we say that thoughts are tensed but that the world isn’t? I believe this position is no more stable
than saying that language is tensed and the world isn’t. We will want to give a semantics for our thoughts,
176 Peter Ludlow

because doing so lends clarity to the discussion about the metaphysics of tense
and because doing so can expose metaphysical commitments that can otherwise be
smuggled into the debate (I’ll highlight examples of this phenomenon below).
In this chapter I explore the case for tensism, consider some of the objections and
alternatives that have been offered to it, and then try to shore up the position in a
way that answers some of the concerns. In particular, my plan is as follows.
In Section 8.1, I’ll take a look at the initial case for tensism, and I’ll offer that the
initial most plausible analysis of tense is thinking of it along the lines of Fregean
sense. In Section 8.2, I’ll make the case that we can take tense to be a perspectival
property that is consistent with four-dimensionalism in the sense of Sider (2001). In
Section 8.3, I’ll take up a key objection that John Perry has raised against Fregean
approaches to tense, an objection that I call the wrong candidate objection. In Section
8.4, I will offer a response to this objection, leaning on a theory of the cognitive
dynamics of tensed language and how it is used to express tensed thoughts. In
Section 8.5, I’ll take up alternative approaches due to Perry and James Higginbot-
ham, and show that the approaches are either too thin to do the necessary work, or
they smuggle in a notion of Fregean indexical sense. In Section 8.6 I’ll offer some
concluding remarks.

8.1 The initial case for tensism

Consider the following scenario. I am sitting in my office, aware that I have a
meeting with the university president at three o’clock. I may even note to myself
that I have the meeting at three. Let’s say that it is three o’clock but I don’t know it is,
so I do not get up and go to the meeting. But then I look at a clock and see the time.
I realize that my meeting is happening now, so I get up and run to the meeting.
Now consider two possible three o’clock utterances:
(1) I have a meeting with the President at three o’clock.
(2) Oh no, I have a meeting with the President now!3
In the scenario I envisioned, the belief I have when I utter (1) does not get me up and
off to the meeting, but the belief I have when I utter (2) does get me up and off to the
meeting. So clearly there is something different going on in the two cases. The
question is, is that something extra semantic? That is to ask, is the something extra
built into the semantic content of my utterances?

and it will have to hook up to the world in the right way or it will not successfully play a role in our
explanation of actions and emotions.
See de Saussure (this volume), for a discussion of how similar effects hold with epistemic futures and
other cases such as present perfect with future reference in French.
Tensism 177

There are related examples. Following Prior (1959), consider (3) contrasted
with (4), and let’s suppose both are uttered at one o’clock GMT, on 11-11-11.
(3) I am thankful that my root canal is over with.
(4) I am thankful that my root canal is earlier than one o’clock GMT, on 11-11-11.
Prior held that it simply isn’t enough to say someone who utters (3) is grateful the
root canal was complete at one o’clock GMT, on 11-11-11—that just isn’t enough to
explain the relief (thankfulness) an the part of the speaker.4 The speaker is presum-
ably grateful that the root canal is safely in her past—the root canal is temporally
removed from her egocentric position—and is not thankful that the root canal is
earlier than some particular time point.
So again it seems that something more is going on with the tensed version than in
the untensed version, and again the question is whether this is a function of
something extra being built into the semantic content of the utterance. The Fregean
would answer “yes” to this question, but so would a number of non-Fregeans. For
example, in relatively recent work John Perry (certainly not a Fregean) has suggested
that we cannot be cavalier about rejecting a role for cognitive significance in the
I cannot accept that a semantic theory can be correct that does not provide us with an
appropriate interface between what sentences mean, and how we use them to communicate
beliefs in order to motivate and explain action. A theory of linguistic meaning should provide
us with an understanding of the properties sentences have that lead us to produce them under
different circumstances, and react as we do to their utterance by others.
(Perry 2001: 8)

Indeed, Perry offers the following “cognitive constraint on semantics”:

If there is some aspect of meaning, by which an utterance u of S and an utterance u’ of S’
differ, so that a rational person who understood both S and S’ might accept u but not u’, then a
fully adequate semantics should say what it is.
(Perry 2001: 9)

We will get to Perry’s positive proposal in Section 8.5, but let’s take a look at the
Fregean position first. To a first approximation, we could characterize the difference
in sense between these two utterances as follows, where F1 indicates a first gloss on a
Fregean sense analysis.
(1-F1) An utterance u, at three o’clock of ‘I have a meeting with the President
at three o’clock’ is true iff [the individual picked out by the sense of ‘I’ in u]
has a meeting at three o’clock.

See the chapters by Prosser, de Saussure, and Jaszczolt (this volume), for additional discussion of this
178 Peter Ludlow

(2-F1) An utterance u, at three o’clock of ‘I have a meeting with the President now’
is true iff [the individual picked out by the sense of ‘I’ in u] has a meeting at
[the time picked out by the sense of ‘now’ in u].
Setting aside the specifics of how details of the Fregean sense story are going to
work, this looks pretty reasonable, but many philosophers have objected to such a
move. One set of objections is metaphysical, and revolves around an alleged con-
undrum one gets into by wedding tense to a standard four-dimensionalist picture of
time. The second set of objections has to do with puzzles in constructing a coherent
tensed semantics if we think of tense in Fregean terms. I’ll take up these concerns in
the next two sections of the chapter.

8.2 Tensism and four-dimensionalism

Suppose we wanted to add tense to the standard four-dimensionalist picture of time,
as described in Sider (2001) and widely held in philosophy. To make things vivid,
let’s just think about the temporal indexical ‘now’ (the same point can be made with
the basic tenses). Let’s say somebody utters sentence (2) above, containing the word
now, but then they utter it again a week later (let’s say they have a weekly meeting
with the university president). So it seems that the notion of now, whatever that is,
has moved along the time-like dimension of space–time. But wait! Movement is
supposed to take place in time. And we were supposed to be characterizing time. So
how could that movement take place in time?
The same problem occurs if we take future or past tense as well. The future region
of space–time will shrink, and the past will grow. But again, what sense does that
make? Shrinking and growing take place in time.
One possibility here is that we move to a notion of second-order time. The
indexical property does move, not in time proper, but in second-order time. So we
get this: there is a second order event of it being now-at-t1, and it was now-at-t1
before it was now-at-t2.
The problem with this move is that we can quickly generate regress. Why should
we stop with second-order time? Indeed, we will have the same motivations to
introduce tense at the second-order level as we had at the first-order level. To see
this, consider the second-order version of Prior’s example.
(3-2nd order) I’m thankful it’s now-at-t2 now and not yesterday. (said after a visit
to the dentist at t1)
Or suppose there is a meeting scheduled (by a clever philosopher) for when it is
now-at-t2. I realize it is t2 and utter:
(2-2nd order) Oh no, it’s now-at-t2 now!
Tensism 179

We can even generate utterances that seem to cut across levels. So, consider the
(3-cross-level) I’m thankful my root canal was now-at-t1 and not now. (said at t2)
But apart from this there is a question of whether second-order tense is coherent.
Either an expression like now-at-t1 loses all indexical content, in which case it lacks
the cognitive significance to explain our actions and emotions, or if we kept it in it
would have to be a monster in the sense of Kaplan—that is, it would be an indexical
operator the reference of which is not anchored to the time and place of utterance of
the matrix sentence, but can shift according to the linguistic context (for example in
an embedded tensed clause).
As noted by Kamp (1971) and Kaplan (1977, 1979, 1990), real tenses can’t be shifted
to another time index or temporal perspective; thus we have the following examples.
(5) *Yesterday I will be happy.
(6) *Yesterday it was true that I will be happy.
(7) *Yesterday it was true that I’m happy now.
This suggests that the ‘now’ in ‘now-at-t2’ can’t be a genuine temporal indexical—i.e.
not the one we get in a sentence like (2).
I think that there is a way out of this problem, but it involves buying into some
assumptions that I’ve argued for elsewhere—in particular it requires that we adopt
that picture of the dynamic lexicon that I’ve defended in Ludlow (2000, 2006, 2007).
On that view, word meanings are dynamic to the point that they can shift even
within a conversation and even from moment to moment. For example, I’ve applied
this idea to predicates like knows, where the standards for knowledge can shift
with the introduction of a sceptical hypothesis. Applied to the case of tense, the
thought is that not merely the referent of now but the sense of now can shift across
time—or better, now has a different sense at different points in time. One way of
thinking about this would be that there are many words now, and the now is not
moving along a time-like dimension. There are simply different expressions now,
each expressing a different sense at different points along the time-like dimension in
What properties to these nows express? And more to the point, how do we express
what they express at different times? Obviously, on this view, the sense of now in an
utterance tomorrow cannot be the same as the sense of now in an utterance today.
Here is a quick gloss of the concern and how we can answer it.

See Sklar (1977) for a discussion of the nature of space–time.
180 Peter Ludlow

The basic case of using an expression works like this, where we can give the
content of the indexical disquotationally, since we give the content at the moment it
is being used.
(2-T1) ‘I have a meeting now’ is true just in case I have a meeting now.
But what if we want to give the content of ‘I have a meeting now’ as uttered at some
other time? For example, what if it was uttered yesterday? Well, I couldn’t have
meant now, so now must have meant something else then.
(2-T0) An utterance by me yesterday of ‘I have a meeting now’ meant that I had a
meeting then (yesterday).
I’ll say more about this in a bit, but for now suffice it to say that it will involve a
theory of the cognitive dynamics of using expressions to track temporal sense.
If we go with this picture, it is natural to ask what properties in the physical world
we are tracking with these expressions. I believe that each subsequent use of now
expresses a unique perspectival property along a time-like dimension in space–time.
We can even extend this idea to relativistic physics if we take these perspectival
properties to be relative to an inertial frame. They are, in effect, egocentric properties
accessible to an agent who is embedded in time on an inertial frame—in effect,
someone embedded in an arrow on a Minkowski space–time diagram.6
Now you might object that science is the view from nowhere, and there really
should not be perspectival properties in science, and it may well be that such
properties are not of use in physics. Physics, after all, does not need to be interested
in every property. But it may well be that other theories cannot dispense with such
properties—the theory of action, for example. It is also worth noting that even
though general physics may not need such properties, any theory of the experimen-
tal setting in science would, because we want to understand the ways in which we
talk about individual physical events, individual experiments, phenomena, data, etc.
All of these require the use of indexicals in order to fix the items of interest, and in
my view they will all ultimately involve an appeal to indexical sense.
I don’t mean to suggest that four-dimensionalist tensism is the only way out for
the tenser. Prior (1967, 1968), for example, thought the best solution was to opt for
presentism. I don’t think that such a view is a non-starter, but I do believe that there
are technical difficulties not anticipated by Prior and other presentists, particularly
problems centring on the phenomenon of temporal anaphora (for example, as in
‘I turned off the stove then’). In Ludlow (1999), I attempted to address some of these
difficulties on behalf of the presentist, and then, as now, I didn’t believe that the
problems were insurmountable, but I do believe they remain unresolved. My point

See Sklar (1977) for a discussion of Minkowski space–time diagrams.
Tensism 181

here is not to take sides on the best way out. I am simply taking the four-dimen-
sionalist tenser option because it is the version of tensism that offers the minimal
departure from standard views of tense and time.
If I am right, then the best way to understand linguistic tense is that it expresses a
real property of the world—metaphysical tense—and that metaphysical tense can be
understood as an egocentric perspectival property. This property will be crucial in
our semantic theories if we want them to account for why we engage in the actions
and have the attitudes that we do. But what is the best way to characterize these
properties in the semantics?7 Earlier I suggested characterizing them with the help of
Fregean senses, but as we will see, this leads to some objections.

8.3 An objection to the Fregean story

What is the harm of introducing Fregean senses to account for the ‘something extra’
in cases like (2) and (3) above? For my money the most interesting is one introduced
by Perry (1977, 1979) and that turned on the following famous passage from Frege
(1956 [1915–1919]). I will call it the wrong candidate objection.
If someone wants to say the same today as he expressed yesterday using the word ‘today’, he
must replace this word by ‘yesterday’. Although the thought is the same, the verbal expression
must be different so that the sense, which would otherwise be affected by the differing times of
utterance, is readjusted. The case is the same with words like ‘here’ and ‘there’. In all such
cases the mere wording, as it is given in writing, is not the complete expression of the thought,
but the knowledge of certain accompanying conditions of utterance, which are used as means
of expressing the thought, are needed for its correct apprehension. The pointing of fingers,
hand movements, glances may belong here too. The same utterance containing the word
‘I’ will express different thoughts in the mouths of different men, of which some may be true,
others false.
(Frege, 1956 [1915–1919]: 24)

Perry argued that Frege gets into trouble by trying to identify the sense of a sentence
(utterance) with a thought. Why? Well, because ‘yesterday’ and ‘today’ presumably
have different senses, and it therefore follows that ‘Today is a fine day’ and
‘Yesterday is a fine day’ must have different senses (since they are composed of
different senses). But if I can express the same thought today with an utterance of
‘yesterday is a fine day’ that I expressed yesterday with an utterance of ‘today is a fine
day’ then thoughts cannot be associated with senses. Different senses are deployed in
expressing the same thought so thoughts are not in a one-to-one correspondence
with the senses of sentences. Senses are not good candidates for characterizing
cognitive significance.

On a perspectival interpretation of tensed utterances see de Saussure (this volume).
182 Peter Ludlow

Referentialists will say we are better off getting rid of senses and opting for a more
austere theory. Accordingly, referentialists argue that we must keep the something
extra out of the semantics—it will mess up the semantics in crucial cases. They
accordingly suggest that the semantics for (1) and (2) should look like the following,
where the R indicates a referentialist gloss.
(1-R) An utterance u, by s, at t, where t ¼ three o’clock of ‘I have a meeting with
the President at three o’clock’ is true iff s has a meeting with the President at t.
(2-R) An utterance u, by s, at t, where t ¼ three o’clock of ‘I have a meeting with
the President now’ is true iff s has a meeting with the President at t.
The problem is that we now appear to have lost the content that was going to explain
why I got up and got going after I uttered (2) but not after I uttered (1). We have
abandoned the cognitive constraint on semantics that even a non-Fregean like Perry
was happy to endorse. We thus get the following dilemma.
dilemma: If we include the something extra in our semantics—if we make it part of
the truth conditions—we mess up the semantics. If we leave out the something extra,
we don’t seem to have a way of accounting for the difference in cognitive significance
between (1) and (2).
As we will see in a bit, some philosophers have argued that there is a third way here:
that we needn’t go referentialist or Fregean, but there is something in between. For
example, neither Higginbotham nor Perry are Fregeans when it comes to tense—both
offer what they take to be a middle way that includes something extra in the
semantics but not the full Fregean sense content. A bit later I’m going to argue
that the middle way they offer isn’t stable.
Before we get to their positive proposal, however, I want to make the case for the
Fregean position, and before we do this we need to lay out the doctrines that the
Fregean is committed to and see if in fact these doctrines lead to the conundrums
that Perry claims.
For example, in a very helpful mapping of the territory, Heck (2002) argued that
Frege was committed to the following doctrines.
(i) There can be different Thoughts that “concern the same object” and ascribe
the same property to it. For example, the Thought that Superman flies and
the Thought that Clark Kent flies are different, even though Superman is
Clark Kent.
(ii) Sentences of the form ‘N believes that a is F ’ and ‘N believes that b is F ’ can
have different truth values, even if ‘a’ and ‘b’ refer to the same object.
(iii) Sense determines reference.
(iv) The sense of a sentence is what one grasps in understanding it.
(v) The sense of a sentence is a Thought.
Tensism 183

Perry’s argument can be taken as the claim that you can’t really hold all of these
positions simultaneously, so something has to give. Not surprisingly, different
philosophical camps have chosen different doctrines to reject.
The referentialist’s way out is to reject (ii)—that is, to argue that the truth values
are literally the same and that accounting for cognitive significance is not the job of
semantics. Heck’s way out is to reject (iv)—that is, reject the idea that there is a
single thought associated with the understanding of a sentential utterance. Here is
how Heck makes the case for that position.
But why do we want to find something to call the meaning? What we (relatively) uncontro-
versially have are speakers who associate Thoughts with utterances and restrictions upon how
the different Thoughts they associate with a given utterance must be related if they are to
communicate successfully: to put it differently, we have the fact that utterances have cognitive
value for speakers, and we have communicative norms determining how the cognitive values a
given utterance has for different speakers must be related if we are to understand them.
(Heck 2002: 31)

As Heck says in the same paper, “a given utterance can differ in cognitive value for
two speakers without their being unable to communicate successfully”.
I’m not entirely clear on Gareth Evans’ position but I believe that his way out was
to reject (v)—that is, to break the link between senses and thoughts. The case for this
can be found in Evans (1985: 307–8):
. . . there is no headlong collision between Frege’s suggestion that grasping the same thought
on different days may require different things of us, and the fundamental criterion of
difference of thoughts which rests upon the principle that it is not possible coherently to
take different attitudes towards the same thought. For that principle, properly stated, pre-
cludes the possibility of coherently taking different attitudes towards the same thought at the
same time.

Evans appeared to be saying that thoughts are to be identified with senses—but with
different senses at different times.
My own view is that Heck’s characterization of the situation is missing some
details and that there are other assumptions (not necessarily Fregean) required to
generate the problem that Perry claimed to see in Frege’s position. In particular,
I believe that to really generate problems of the sort that Perry envisioned, one needs
to introduce an additional doctrine.
(vi) A sense is intimately tied to its manner of expression, so that the senses of
‘today’ and ‘tomorrow’ remain constant
Perry seems to take it as a given that Frege was committed to this, but I frankly don’t
see it, and this is the premise that is causing the trouble. In response to Perry,
Branquinho (2006) and Ludlow (1999, 2008) have argued that this premise should be
184 Peter Ludlow

rejected—we need to break the link between senses and their forms of expression. As
I argued in Ludlow (2008), indexical expressions like today can be used to express or
display senses, but they do not express the same sense on each occasion of use. This
strategy holds the sense associated with a thought constant, but allows that different
indexical expressions can display the same sense (albeit on different occasions of
use). Branquinho (2006: 2) puts it as follows:
Cases where one is dealing with indexical contents are problematic because they often involve
some realignment in the linguistic means of expression of a thought—on the part of a given
thinker—as time goes by. In other words, there are situations in which the verbal expression of
an indexical thought entertained by a thinker at a given time must, at a later time, be
readjusted in a certain way by the thinker in order for the thought in question to be then
entertained; so that one could presumably say that some attitude held at the earlier time
towards the thought in question has been retained by the thinker at the later time, the very
same thought being the object of the attitude on both occasions. Naturally, such readjustments
are to be thought of as being operated in the linguistic means employed for the expression of
the thoughts. It does not make much sense—at least in the light of the picture of content we
are assuming—to think of the thoughts as being themselves subjected to any sort of change or

For what it’s worth, one can also make the case that Evans had something like this in
mind as well. Here is the passage in Evans (1985) that seems to support the idea:
Frege’s idea is that the same epistemic state may require different things of us at different
times; the changing circumstances force us to change in order to keep hold of a constant
reference and a constant thought—we must run to keep still. From this point of view, the
acceptance on d2 of ‘Yesterday was fine’, given an acceptance on d1 of ‘Today is fine’ can
manifest the persistence of a belief in just the way in which acceptance of difference utterances
of the sentence ‘The sun sets in the West’ can.

Rödl (2007: 196–7) expresses the point as follows, with an emphasis on the idea that
it is really the linguistic expression that is shifting even while the thought or sense
stays constant:
We found that, in suitable cases, “Today . . . ” said yesterday and “Yesterday . . . ” said today
express the same act of thinking. These cases are fundamental in that, without them,
there would be no such thing as an act of thinking expressed by either phrase. And when
“ . . . today . . . ” yesterday and “ . . . yesterday . . . ” today express the same act of thinking, then
they express the same thought. Therefore, it would be misleading to contrast “yesterday”-thoughts
with “today”-thoughts; in the fundamental case, a “yesterday”-thought is a “today”-thought.

8.4 The cognitive dynamics of Fregean accounts of tense

In the previous section I sketched an approach to tense that was largely Fregean. In
this section I want to develop that picture in more detail. To do this, I want to think
Tensism 185

of senses as being displayed by the theorems of a Davidsonian T-theory.8 In doing

so, we need to distinguish between:
(a) what the truth conditions literally state
(b) the way in which the truth conditions are represented
(c) the sense displayed by the truth conditions so represented
Here is an illustration of the idea from McDowell (1980) and Lepore and Loewer
(1987). While theorems (8) and (9) in some sense state the same truth conditions,
they do so in different ways, so that they ‘display’ different senses.
(8) ‘Cicero is bald’ is true iff Cicero is bald.
(9) ‘Cicero is bald’ is true iff Tully is bald.
Similarly, we can think of T-theorems that use indexicals on the right-hand side as
displaying the senses of the indexical expressions on the left-hand side.9
(10) Val(x, ‘I’) iff x ¼ I
(11) Val(T, ‘I walk’) iff I walk
(12) Val(x, ‘today’) iff x is true today
But isn’t this hopelessly naive? Not necessarily. If it is true that we can express the
same sense in different ways, then what we need to do is develop a theory that
affords perspective-sensitive axioms. For example, here is how we might handle the
example introduced by Frege.
(13) if yesterday s left a message m, having the form ‘today is a fine day’, then for
s,m ‘today is a fine day’ is true iff it was a fine day yesterday
Now let me be clear. A semantic theory would not consist of theorems of this form,
but it would be a theory of the capacities that allow us to build theories like this on
the fly. In effect, a theorem like (13) is a theorem generated by a ‘passing theory’ in
the sense of Davidson (1986).
The question is, can this sort of project be made systematic? Branquinho (2006: 3)
offers the following way of envisioning the project:
Suppose that on a certain day, say d, Jones says
[(14)] Today is fine
and believes it. What sentence or sentences should he be disposed to accept on the following
day, d þ 1, so that one could say that he has then retained his previously manifested belief
about the weather on d? Again, a plausible choice would apparently be given in the sentence
[(15)] Yesterday was fine.

See Davidson (1967b). Note that Davidson didn’t think one could lift indexicals into the metalan-
guage, presumably for many of the reasons canvassed here.
See Rumfitt (1993) for a similar proposal.
186 Peter Ludlow

. . . Let us call the above claim, the claim that sentences such as . . . [(14)], taken with respect to
the envisaged contexts of utterance, constitute choices which are appropriate for attitude-
retention the Natural Realignment Claim. . . .
Now the Natural Realignment Claim might be given two different readings. Take Jones’s
case, for instance. On the one hand, it might be held that Jones might be said to have retained
on d þ 1 the belief he held on d (by accepting then [(14)]) only if he would be disposed to
accept [(15)] on d þ 1. That is to say, the claim is that a disposition to accept a token of the
sentence in question on the later occasion is necessary for retention of the belief had on the
earlier occasion. On the other hand, it might be held that Jones’s disposition to accept [(15)] on
d þ 1 is sufficient for retention of the belief he held on d (by accepting then [(14)]). I shall refer
to those distinct versions of the Natural Realignment Claim as the Necessity Claim and the
Sufficiency Claim (respectively).

As Branquinho argues, neither the Necessity Claim nor the Sufficiency Claim
appears to hold, and here we get to the problem of Rip van Winkle.
Rip goes to sleep one day saying to himself ‘today was a fine day’. When he awakes
twenty years later, he may want to express what he expressed by the utterance that he
made before he fell asleep. He may try to do this by saying ‘yesterday was a fine day’,
but in doing so Rip fails to express what he did with his original utterance because he
has lost track of the relative temporal position of his original utterance. This appears to
blow up any chance of systematizing the theory in the way that Branquinho initially
For Evans (speaking of beliefs rather than utterances here), this is a bullet that we
should bite:
I see no more strangeness in the idea that a man who loses track of time cannot retain beliefs
than in the idea that a man who loses track of an object cannot retain the beliefs about it with
which he began.
(Evans 1985: 87n–88n)

Branquinho is not ready to give up so easily and he asks that we consider the less
extreme case where Rip does not realize that midnight has just passed and would not
assent to ‘yesterday was a fine day’.
Thus, if Jones mistracked time in the way described before, then . . . what he would not be in a
position to do at 00:01 a.m. on d þ 1 is to T-retain the particular belief he held at 11:58 p.m. on
d when he accepted [(14)]. In other words, he would not be able to re-express then such a
belief by using a temporal indexical such as ‘yesterday’. But it does not follow that it would be
impossible for him to retain, or even re-express, tout court that belief . . .
Although Jones is ex hypothesi unable on d þ 1 to keep track of d, i.e. to think of d as
yesterday, he might still be said to have retained on d þ 1 his former belief about d in a
certain way, namely by means of memory, and not in virtue of the particular position he
occupies in time or of his knowledge of such a position. One should therefore regard as
Tensism 187

unacceptable the claim that a disposition to accept [(15)] on d þ 1 is necessary for a thinker to
retain or re-express then a belief she had on d by accepting [(14)].
. . . a way by means of which a thinker like Jones could re-express on d þ 1 the belief held on d
(by accepting [(14)] then) would be to accept, or to have a disposition to accept, on d þ 1 a
token of a sentence such as
[(16)] That day was fine.
The demonstrative phrase ‘that day’ would be here taken as expressing in the context a
memory-based demonstrative mode of presentation of d, i.e. a way of thinking of a certain day
anchored upon a memory demonstration of the day in question.

Is it really plausible to think that there can be memory-based demonstrative modes

of presentation? Perhaps, but I want to step back from this problem for just a little
bit and consider it from another perspective. Suppose we don’t want to tie the theory
to what a speaker might assent to. Third parties can report beliefs just as well, after
all. What would underlie our ability to keep track of senses?
Richard Larson and I (1993) developed a theory designed to explain our ability to
know when two that-clauses attribute the same belief or count as saying the same
thing. The challenge was to show how two distinct belief reports using different
linguistic expressions, presented to different hearers, could count as having attrib-
uted the same attitude (thought) to an agent. The key idea was that belief reports are
not designed to offer expressions that mirror the form of some object in the agent’s
head; to the contrary the idea was that a speaker S constructs a report for the benefit
of hearer H, given H’s interests with regard to agent A, shared assumptions about
modelling A’s beliefs, etc.
The components of this theory were as follows:
(i) The tacit theory of belief
(ii) The tacit theory of the goals of belief ascription
(iii) The tacit theory of belief ascription logistics
As we envisioned the project, the theory of belief ascription logistics would state
which expressions should be used in a given context to achieve specific belief
ascription goals. For example, depending upon H’s interests, it was sometimes the
referential component of an expression, and sometimes the syntactic component
that would be important to the goals of ascription—that is, for constructing a
valuable report for H.
In cases where prediction or explanation of behaviour is the goal, we envisioned
the speaker’s choice of syntactic constituents in a belief report to involve a two-stage
process vis-à-vis the hearer H. First, S determined the way in which H models A’s
belief structure. Then S collaborates with H on the expressions to be used in
speaking of the components of that model.
188 Peter Ludlow

In the second stage of selecting a belief report, the idea was that S and H must
agree on expressions used to speak of the components of H’s model of A’s belief
structure. Expressions used in attitude ascriptions would be fixed by tacit entrain-
ment procedures engaged in by participants in the discourse, following general
principles holding of discourses of all kinds (see Ludlow 2006, 2007).
In Ludlow (2008) I expanded on this idea and extended it to temporal indexicals
as follows: in addition to knowing the hearer’s interests, we also need a theory of the
hearer’s spatiotemporal perspective. When reporting what the agent said or thought,
we need to know where the hearer is relative to the agent. For example, if on
Wednesday the agent says ‘It is fine today’, then when we give reports to a hearer
on Thursday we shall want to report the same statement, but relative to the hearer’s
temporal position. Thus we get ‘A said that it was fine yesterday’. Roughly, then, we
will want the theory to keep track of the following:
(i) S’s knowledge of H’s spatiotemporal position
(ii) S’s knowledge of H’s and A’s relative positions
(iii) H’s knowledge that S knows A’s spatiotemporal position (with standard
assumptions about common knowledge)
So what counts as having the same thought or remembering the same thought? That
isn’t a function of what the speaker assents to, but rather the interests of the attitude
reporter and her audience H. Just as there are many ways and languages in which we
might truly report Galileo’s belief that the earth moves (and thus many ways we
might accurately describe his belief), so to there are many ways we, as third parties,
might truly describe Rip van Winkle’s belief on the day he fell asleep.
For example, we might issue a report like ‘Rip thinks it was a fine day the day he
fell asleep’. And surely, this is a true characterization of Rip’s thought. It doesn’t
situate Rip’s thought in our shared egocentric space, however, and here we have a
couple of options. One thing we might say is this: ‘Rip thinks it was a fine day when
he fell asleep twenty years ago.’ Or alternatively, we may choose to situate Rip’s
thought in our egocentric space but without the precision. We might say: ‘Rip thinks
it was a fine day when he fell asleep that day back when.’
The key point here is that these last two formulations can, under the right
circumstances, attribute the same thought to Rip. It uses different expressions to
attribute that thought, but this is really no more bizarre than using yesterday and
today to express the same thought on different days or using both English and Italian
to express Galileo’s thought that the earth moves.
Finally, this gets us back to the case of Rip when he is self-attributing his belief. He
may well incorrectly characterize his belief as follows: ‘I believe it was a fine day
yesterday.’ But this is not the only way Rip has available to him to express that
thought. He might do as we did and simply say this: ‘I believe it was a fine day that
day I fell asleep back when.’ Now he will probably not self-attribute the belief in this
Tensism 189

way, because given his other beliefs he would consider doing it in this way awkward.
But this nevertheless remains a way in which he is capable of expressing his belief
and expressing the belief in this way situates it in his egocentric space. He has not
lost track of his thought (as Evans feared) nor even the capacity to express his
thought. He simply lacks the information needed to consistently express his belief
correctly and to characterize it with quantitative precision.
There might be more exotic cases that foil this approach, such as time travel cases,
but even these don’t clearly create insurmountable problems. Suppose that Rip is
teleported back in time so that he wakes up twenty years before he fell asleep. Just as
we can talk about him falling asleep before he woke up (along his personal time
path), we can, if we wish, characterize his beliefs with respect to his temporal path:
‘Rip believes it was a fine day when he fell asleep’. We needn’t gloss it as ‘Rip believes
it will be a fine day when he will fall asleep’, so why should he be forced to?
All of this remains programmatic, but I believe there is a clear path towards devel-
oping a theory of the cognitive dynamics of index-laden belief attributions—belief
attributions that are sensitive to the egocentric position of both the agent and speaker
and hearer of the attributions.
I want to close by returning to the Perry/Higginbotham alternative, however—the
approach in which they opt for something a bit thicker than the referential account,
and a bit thinner than the Fregean account I have sketched here. It will be my
contention that there is no middle path; their approach either fails to deliver the
appropriate cognitive significance, or it collapses into the Fregean story.

8.5 The Perry/Higginbotham alternative: token reflexive content

To set up our discussion, let’s follow Perry (2001) and his claim that utterances have
at least three kinds of content:
The indexical content (sometime he calls this content-M) of an utterance corres-
ponds to the truth conditions of the utterance given the facts that fix the language of
the utterance, the words involved, their syntax, and their meaning. He might have
called this reflexive content.
The referential content (sometimes he calls this content-C and sometimes the official
content) of an utterance corresponds to the truth conditions given all of these
factors, plus the facts about the context of the utterance that are needed to fix the
designation of indexicals.
The designational content (sometimes he calls this content-D) of an utterance
corresponds to the truth conditions given all of these factors, plus the additional
facts that are needed to fix the designation of the terms that remain (definite
descriptions in particular, but also possessives, etc.).
190 Peter Ludlow

Indexical content is the level of content that we are going to be interested in here,
for that is the content that is supposedly going to provide us with the something
extra that we are going to need to give us the relevant cognitive significance.
Inserting the reflexive content into our analysis of sentence (2) above, we get the
(2-X) This utterance u, at three o’clock of ‘I have a meeting with the President now’
is true iff the utterer of u has a meeting with the President at the time u is
A similar analysis extends to (3), Prior’s root canal case.
(3-X) The utterer of u is thankful that the event of his root canal is earlier than the
time of u.
Higginbotham’s (1995, 2009) approach is not significantly different from this.
Let’s call it (3-H).
(3-H) If e is the event of my affirming to myself with a sense of relief, ‘My root
canal is over’ (or: ‘over now’), then the thought that I think is indicated in
[(A)] (Es) s is the situation of my root canal’s being over & the time of s
includes (the time of ) e.
(1995: 228)

The thought that I have when I affirm with a sense of relief that my root canal is over is a
thought whose very existence depends upon the existence of a certain episode in my mental
life: it contains that episode as a constituent, and could not exist without it.
(1995: 229)

The problem with all of these formulations is that they simply don’t get the reflexive
content into the propositional content. To see this just consider the that-clauses in
all of these examples.
(2-X’) the utterer of u has a meeting with the President at the time u is uttered.
(3-X’) that the event of his root canal is earlier than the time of u.
(3-H’) (Es) s is the situation of my root canal’s being over & the time of s includes
(the time of) e
The problem with all of these is that the relevant indexical information—the
information that situates the event in the agent’s egocentric present or past—simply
does not make it into the propositional content of the thought. But the propositional
content of the thought is what is supposed to be driving our acting the way we do
and having the emotions that we do.
It seems that we need more than this in the propositional content. Shouldn’t the
reflexive content of (3) actually be much closer to (3-XX)?
Tensism 191

(3-XX) The utterer of this utterance u is thankful that the event of his root canal is
earlier than the time of u—this very utterance!
Higginbotham and Perry both feel the pull to put that information into the prop-
ositional content—note Higginbotham’s remark that this “is a thought whose very
existence depends upon the existence of a certain episode in my mental life: it
contains that episode as a constituent, and could not exist without it” (1995: 229).
But this gloss is simply used to fix the thought in question, but when it really
matters—when we look into the content of the thought itself—we simply get the
content in (A) (see (3-H) above). And it doesn’t matter whether we use token
reflexive resources to identify a time or a thought; if those resources don’t make it
into the thought content it is as though we are quantifying into a propositional
attitude context—we only get the referential content.
But now let’s suppose that either Higginbotham or Perry or both conceded this
point. Suppose they opted to put the reflexive content into the propositional content
itself, as along the lines of (3-XX). I’m not saying they would do this, just that they
could, so let’s call the advocates of (3-XX) HigginbothamXX and PerryXX.
Now the question becomes whether someone who makes this move hasn’t slid all
the way over to the Fregean position I advocated above. That is, (3-XX) succeeds
simply because it smuggles an indexical into the analysis. Now it is true that it
doesn’t appear to be a temporal indexical like now, but on closer inspection matters
are not so clear. Exactly what is going on with the demonstrative this? It isn’t a
demonstration in perceptual space. It appears to be indicating the utterance hap-
pening now. Even if it is not a temporal indexical, it is still an indexical that has been
lifted into the metalanguage, and this raises two points. First, if it is legitimate to lift
this indexical into the metalanguage, then why not others? If Perry’s objection to
Frege isn’t an issue here, why would it be an issue when we get to tense? But second,
assuming that this indexical has no tense component, it at least situates the speaker
perspectivally within space–time, and that means that tense-like egocentric perspec-
tival properties are being appealed to in the analysis. And the point of this chapter is
that you can’t do without such properties.
At the beginning of this chapter, I suggested that we would do well to pay
attention to the semantics of natural language when discussing tense and time, not
because we are trying to derive metaphysics from language, but because (among
other things) we want to expose places where metaphysical assumptions have been
smuggled in on the sly. This may well be one of those places, for if we examine the
semantics of the Higginbotham/Perry proposals closely we see that either they
are leaving the critical content out of their analysis of these constructions, or they
(or PerryXX and HigginbothamXX) are smuggling in a form of Fregean sense
content. The move allows them a kind of philosophical sleight of hand—‘now you
see it now you don’t’. Perry and Higginbotham don’t want the Fregean sense in the
192 Peter Ludlow

propositional content of the analysis because of the reasons articulated by Perry. But
they feel pressure to lean on the Fregean sense content because of the constraint on
cognitive significance, so they use it to specify the propositional content. But this
does not yield the necessary results. It is not enough for the sense content to specify
the propositional content; if it is to satisfy the cognitive constraint on semantics, the
sense content must make it into the propositional content.

8.6 Conclusion
I’ve suggested that there are good reasons to endorse tensism. In the first place, tense
seems to play a crucial role in our explanations of action and emotion. But second,
tensism need not commit us to presentism; it works just as well with four-dimen-
sionalism, even on relativistic conceptions of time. On this score I’ve argued that
tense can be seen as corresponding to egocentric perspectival properties. But how do
we characterize these properties in our semantics of tensed language (in a way that
tensed language can be deployed in explanations of action and emotion)? Here I’ve
suggested that our best bet is to introduce Fregean sense into our semantics. While
Perry has argued that this move leads to problems, I’ve argued that a proper
understanding of the Fregean approach to semantics, coupled with a good theory
of the cognitive dynamics of sense tracking, can ameliorate those problems. Fur-
thermore, approaches such as those advocated by Higginbotham and Perry, which
rely on token indexical theories of tense, are not adequate to the demands of a theory
of tense. Either the content they advocate is too thin—in which case they cannot
account for the cognitive significance of tensed thoughts—or it smuggles Fregean
sense content into their analysis. I submit that we are better off being up front about
the need for Fregean sense, and then doing our best to work out a more complete
theory of the cognitive dynamics of temporal sense.

Temporality and epistemic

commitment: an unresolved


9.1 Introduction: the rationale

There is no doubt that the topic of how humans conceptualize time is a complex one.
It is complex for various interrelated reasons that fall into four broad categories:
metaphysical, epistemic, conceptual, and linguistic. In my Representing Time (Jasz-
czolt 2009), I drew on all four of these domains to put forward arguments in support
of the modal conception of time. In brief, I argued that the human concept of time
logically supervenes, that is, is dependent in the sense of its definitional character-
istics, on the gradable concept of epistemic commitment. In other words, on the
underlying level of basic concepts, temporality is epistemic modality, where our
temporal concepts of past, present, and future eventualities alike are founded on the
degrees of commitment to the truth of the proposition (or, as I explain below, the
proposition-like construct) expressing that eventuality. Put differently, these tem-
poral concepts are underlyingly degrees of detachment from the certainty of an
Now, common-sense judgement would easily allow for the construal on which the
past and the future correspond to such non-zero degrees of detachment, where the
‘zero’ degree is to be understood as certainty of the ‘here and now’. However, as
linguistic evidence suggests, even the present yields to such a gradation as long as the
object of inquiry is the concept of time, that is, internal, psychological time. It is so
because just as our memories (the past) and anticipations (the future) can corres-
pond to more, or less, strongly held representations of states of affairs, so the current

See also Ludlow (this volume) on linguistic and metaphysical tensism.
194 Kasia M. Jaszczolt

states of affairs can give rise to more, or less, strongly endorsed representations.
Where this epistemic commitment translates into grammatical differences, we have
the following parallel: just as one can opt for (1a) or (1b) as a linguistic representation
of a past state of affairs, so one can opt for (2a) or (2b) with respect to the present
(1) a. Aly went to Kuala Lumpur last summer.
b. Aly may have gone to Kuala Lumpur last summer.
(2) a. Aly is in Kuala Lumpur now.
b. Aly may be in Kuala Lumpur now.
This basic linguistic evidence is often overlooked by philosophers who discuss the
past and the future as more ‘remote’ than the present.
Having presented in Jaszczolt (2009) a range of arguments in favour of this modal
construal of temporality, I then proceeded to give a semantic representation
of temporal expressions, utilizing for this purpose a radical contextualist theory of
Default Semantics.2 The main idea was that the modal epistemic underpinnings of
temporality can be formally represented by an operator of acceptability (adapted,
with substantial modifications, from Grice 2001), indexed for the degree of epistemic
commitment and, where appropriate, for the semantics/pragmatics of the grammat-
ical category it draws on, and operating on a representation of the intended content
(in the Gricean sense of ‘intended by some abstract model speaker’) pertaining to the
eventuality. For example, ACCDrf ¢ S stands for ‘it is acceptable to the degree
pertaining to the regular future that it is the case that S’, where S (called a ‘merger
representation’) stands for the proposition-like construct, representing the outcome
of processing of the speaker’s intended meaning—a representation that is by defin-
ition compositional, but at the same time drawing on a variety of sources of intended
meaning in addition to the logical form of the uttered sentence.
Although the thesis of the modal supervenience of temporality was well supported
by theoretical arguments and by evidence from various languages, and it also
easily yielded to a Default-Semantic formal analysis, there remained an important
Unresolved Question (henceforth UQ). It concerns the translatability of what
intuitively seems to be a qualitative difference between the past, the present, and
the future into quantitative differences, that is differences represented as the ‘delta
index’ on the acceptability operator (ACCD). In other words, ‘degrees of detachment
from certainty’ and equally ‘degrees of commitment to the (truth of the proposition
representing an) eventuality’ are quantitative concepts, while the past/present/future

Default Semantics (henceforth DS; Jaszczolt e.g. 2005, 2009, 2010) is a radical contextualist theory in
the sense that the representation of discourse meaning is composed out of the material that can be
attributed to various linguistic and non-linguistic sources and merged in the interaction of various
pertinent processes. Readers are referred to the overview of the revised version of the theory in Jaszczolt
Temporality and epistemic commitment 195

distinction intuitively appears to be a qualitative one. If one founds the latter on the
former, one has either to provide a way of correlating one with the other, or explain
away the intuitive qualitative contrast between the past, the present, and the future.
Both have been attempted in the philosophical literature. In what follows, I build on
the extant philosophical arguments, as well as on linguistic evidence, in an attempt
to answer the UQ of the theory of the modal supervenience of time.
The chapter is structured as follows. Section 9.2 presents a brief overview of
the thesis of modal supervenience of time and ends with discussing the UQ of the
past–future symmetry. Section 9.3 discusses temporality and its quantitative under-
pinnings in the form of ACCD against the background of some pertinent tensed and
tenseless theories of time, with a view to addressing the question of gradability of
the underlying concept of epistemic commitment as an explanans for the human
concept of time. It goes back to the proposal of logical supervenience and puts
together evidence and arguments in favour of the ‘apparent qualitative-to-
quantitative’ translation of the concept of time. It also contains two possible answers
to the UQ. Section 9.4 addresses the pertinent question as to whether the tensed or
tenseless theories of time are more adequate supporters of the thesis of super-
venience on modality. Section 9.5 follows with a discussion of the universal building
blocks of the concept of time that lie underneath the cross-linguistic diversity of
means of expressing temporal reference. Section 9.6 concludes with some observa-
tions on the status of the supervenience thesis so revised and strengthened by the
two proposed solutions to the UQ.

9.2 Temporality as epistemic modality: supervenience

or mere correlation?
The relation of supervenience as understood in what follows has the following
A set of properties A supervenes upon another set B just in case no two things can differ with
respect to A-properties without also differing with respect to their B-properties. In slogan
form, “there cannot be an A-difference without a B-difference”.
McLaughlin and Bennett (2005: 1)

I begin with a brief summary of where the proposal was left out. In Representing
Time (Jaszczolt 2009), I put forward a hypothesis of supervenience, understood as
logical and conceptual supervenience, of the concept of time on the concept of
epistemic detachment. I also entertained there and partly supported the idea of
supervenience of the properties of temporality on the properties of space–time. The
concept of time was understood there as being founded on the concepts of
uncertainty, probability, and therefore epistemic detachment on the one hand, and
on probability and relativity of real time on the other. I concluded that both the
196 Kasia M. Jaszczolt

construal of reality and reality itself require modality for their explanation. In other
words again, the thesis is that internal, psychological time supervenes on modality
qua epistemic detachment and also on real time which itself entails various meta-
physical possibilities in which world histories and predictions develop—all of them,
for this purpose, equally real.
Modality was proposed there as an explanans for four related domains: the
metaphysical, the epistemic, the conceptual, and the linguistic. I presented some
compelling arguments for conceptualizing temporality as degrees of epistemic com-
mitment. Such graded commitment (and, equally, graded detachment from the
state of affairs presented in the act of communication) was analysed as determined
by what is encoded, grammaticalized, or pragmatically inferable from linguistic
expressions in various languages. For reference, Figure 9.1 is an example of the
DS-theoretic merger representation (S) pertaining to the reading of (3). It is a
partial representation, focusing on the representation of the future time reference.
The grammar, lexicon (the WS process), and the inferred future reading of the
present-tense verb form (the CPI process)3 interact in producing this content.
The ‘fp’ index on D in ACCDfp stands for ‘futurate progressive’.4
(3) One Direction are giving a concert in Cambridge tomorrow.

x t Σ

[One Direction]CD (x)

tomorrow (t)
[ACCΔfp ⏐− Σ]WS, CPIpm

Σ [x give a concert in Cambridge]WS

FIGURE 9.1 S for the example (3)

See footnote 2.
See also de Saussure (this volume) on a different account of tense–time mismatches.
Temporality and epistemic commitment 197

Speakers can talk about eventualities with a stronger or weaker degree of com-
mitment using various grammatical forms (simple past tense, modal may or might,
evidentials, etc.), various lexical constructions (definitely, certainly, possibly, as far as
I can tell ), and conveying expectations of various inferential processes on the part
of the interlocutor (by communicating indirectly, for example). This gradation of
commitment pertains not only to talking about the past or future, but also the
present. I say more about the status of the present in Section 9.3.
Time was approached from two angles. First, it was approached from the phe-
nomenological perspective where, following Husserl (1928), time was seen as, so to
speak, ‘coming from within’, being imposed on human experience. It consists of
‘objective’ ‘immanent’ time which has the property of duration, as well as ‘pre-
immanent time’, which has the property of flowing and contains the moving
actuality (‘now’), as well as what follows and precedes it. What precedes it for its
part consists of memories, knowledge, and ‘fixed’ experiences; what follows consists
of anticipated experiences. The second perspective was the ‘real time’, ‘time in the
world’. The big question of external vs. internal time was addressed via evidence
from linguistic representations of the human concept of time and it was proposed, to
repeat, that real time of space–time is founded on (real) possibilities for the universe,
while psychological time is founded on an epistemic relation to some ‘stuff from
which reality is made’.
It is clear at this point that any further discussion of the possible modal under-
pinnings of time has to be conducted in terms of the arguments for various versions
of so-called ‘tensism’ (McTaggart’s 1908 A-theory) and the B-theorists’ rebuttals.
For A-theorists, reality is tensed: time flows, and the past, the present, and the future,
or at least the past and the present, or just the present (depending on the view), are
real. B-theorists, on the other hand, claim that reality is tenseless: there is no past,
present, or future out there and time does not flow. Events are real, and so is their
ordering, but that is all: they don’t have pastness, futurity, or present actuality.
In other words, supporters of the B-theory advocate real, metaphysical time without
flow. Internal, psychological time, on the other hand, exhibits the intuitive three-way
distinction into what was, is, and will be.
Approached in this light, we have an illusion of a perfect quantifiability of time.
On the B-series where events are placed according to their relative order of prece-
dence/following, there is nothing to stop us from proposing that if we ‘place’ any
conscious agent at a random point on the scale, the agent will be at a certain fixed
‘distance’ from any chosen event and this ‘distance’ can be measured according to
some agreed conventions. We are here at liberty to superimpose, for example,
‘mental distance’ or ‘active interest distance’, or any other non-temporal measure,
on such a ‘B-series plus an observer’ complex. Equally, on the A-series, quantifiabil-
ity of time can be achieved when we adopt presentism, that is the view according to
which only what is ‘simultaneous with the time of thinking about it’ is real: there is
198 Kasia M. Jaszczolt

no real past and no real future. If the ‘here and now’ is real, then we can perhaps
envisage that moving away from the reality of the hic and nunc gives us degrees that
are measurable by some agreed conventions. I will say more about degrees and
presentism in Section 9.3.
However, there are loose ends and unresolved questions for both construals. And,
likewise, there are unresolved questions for my construal of the graded modality of
time where time supervenes on epistemic modality qua graded (and quantifiable)
detachment. The unresolved questions would not be there if we assumed that the
common-sense view of time, according to which there is past, there is the ‘now’,
however extended, and the future, and they are all dynamic, need not be squared
with some sort of normative philosopher’s concept of temporality. But it would be
unwise to take this route.5 However, if common-sense time and speakers’ intuitions
are to play a part, then one has to answer the question as to how exactly, if at all, the
value associated with the degree of epistemic commitment (or alternatively
the degree of epistemic detachment) identifies the ‘direction’ of detachment, into
the past and into the future. But not only temporal direction is in need of an
explanation: on the modal supervenience account, thoughts and utterances about
the present are also characterized by different degrees of epistemic detachment, as
was exemplified in (2a–b) above. In my DS-theoretic merger representations (Ss),
I presented this correlation as inheritance from language: the grammar, the lexicon,
and pragmatic inference interact in producing the effect of conveying temporal
location of an eventuality. Demonstrating that this can be achieved, using the
sources of information about meaning and processes indentified in DS, gave add-
itional support to the arguments gathered there in favour of supervenience.
But such a proposal did not exclude the possibility that, after all, this is merely a
strong correlation between what we can call speaker’s acceptance6 of a proposition
pertaining to a certain, real or imagined, state of affairs and the eventuality’s
placement in the future, present, or past (henceforth, F, N,7 and P). So, supervenience
was strongly supported but not exactly demonstrated. To repeat, supposing the
difference between F, N, and P is not qualitative after all, one needs somewhat
stronger support than the availability of merger representations or the linguistic
evidence and philosophical arguments amassed there to overcome the overwhelming
intuition that F, N, and P do indeed exhibit a difference of kind. Accumulated cross-
linguistic evidence only demonstrates that there is no clear correlation between

Jackson (2011) entertains this possibility of a philosophers’ concept of knowledge in view of the
common intuitions in Gettier cases. His arguments against such a move, and in favour of experimental
philosophy, are pertinent here.
See Grice’s (2001) unfinished attempt at a unified account of modalities (alethic and ‘practical’) under
a common operator of acceptability that motivated my adaptation (e.g. 2009) of the operator to epistemic
modality, operating on Ss.
For ‘now’.
Temporality and epistemic commitment 199

temporality and tense, while there are clear correlations between modal and eviden-
tial expressions on one hand and temporal reference on the other. It does not
demonstrate that there is a P–F symmetry, nor does it demonstrate that there is an
analogy between degrees of detachment with respect to P, degrees of detachment
with respect to N, and degrees of detachment with respect to F. Swallowing a bitter
pill of dissonance between ‘philosophers’ time’, where time is composed of the
building blocks of degree of possibility (epistemic and metaphysical likewise) and
‘laymen’s time’, that flows from the future into the past, will not do.8 Alternatively,
supposing the difference between F, N, and P is qualitative, we can hypothesize, and
defend, different ‘values’ or ‘ranges of values’ for the three types of temporal
location. Then we will need a good argument or compelling empirical evidence
that F, N, and P always avail different ‘values’ of detachment. We can also take the
route of distinguishing types of values but then supervenience on epistemic modality
would require a further qualification. Be that as it may, the first steps in addressing
the UQ have to be (i) the analysis of the possible P–F symmetry vis-à-vis the tensed/
tenseless reality debates and, consequently, (ii) the analysis of the status of the
degrees of modality vis-à-vis the ‘degrees of reality’ proposed within presentism.
This is attempted in the following section.

9.3 Quantifying time and ‘degrees of reality’

Out of the three categories of temporality, P, N, and F, there is no question that it is
N, the present, that is the best candidate for the status of being real, actual, and
existing. The boundaries of the ‘now’ may vary among the different applications of
the concept but the actuality of ‘now’, when judged by the linguistic expressions
pertaining to current events, is its default characteristic. The truth-makers for state-
ments with present-time reference are relatively unproblematic, while truth-makers
for statements with past- and future-time reference are up for grabs: we remember
some facts about the past, but are memories the same thing as past events? In other
words, is there some objective reality to refer to as far as temporality is concerned, or
are there just representations?9 It has been proposed that the past is only as real as its
present effects (e.g. Łukasiewicz 1961). The past that does not exist in the present in
any form is only a possibility. Equally, arguably, the future exists only in so far as it is
to be derived from the present; as Heidegger (1953) says, the future has to be
envisaged as the person’s own time, in anticipation of the unavoidable finiteness
of life of which one is aware in the expectation of one’s own inevitable death.10

See footnote 5.
I am grateful to Louis de Saussure for his comments on the metaphysical status of these truth-makers.
The literature on this topic is vast. See e.g. Harrison (2003: 142–3) on the role of death and death
rituals in human life and civilization.
200 Kasia M. Jaszczolt

This relative status of F and P is well conveyed in natural languages. Languages

have hierarchies of epistemic modals and hierarchies of evidentials (Faller 2002b).
The speaker’s choice of a construction with a stronger or weaker degree of trust in
the truth of the embedded proposition, or an indicator of the kind of evidence
(in itself weak or strong) can safely be taken as an indicator of the degree of
In possible-worlds semantics, the standard view has it that modality interacts with
time. The modal base is a construct that is a function from world–time pairs to sets
of possible worlds and represents the epistemic state of the speaker. The function is
determined by the context. So, the representation of may and might is as in (4),
where MB stands for modal base, and therefore MB(w,t) is the set of worlds
compatible with what the speaker knows at time t. P stands for property, and [t,_)
for an interval from t to infinity.
(4) lPlwlt9w’ [w’2MB(w,t) & AT([t, _), w’P)]
(after Condoravdi 2002: 71)
In words, in the case of modal statements, there is a world w’ that belongs to the set
of worlds of the speaker’s epistemic state such that the given property is instantiated
in this world at a certain interval. Graded modality can then be explained as an
ordering of the worlds (Kratzer 1981). Considering that the present is represented as
in (5), the interaction of modality with time is evident.
(5) lPw [AT (now, w, P)]
(after Condoravdi 2002: 70)
Equally, the perfect interacts with the modals to produce past time reference with t’
restricted to intervals preceding t. This interaction of modality with the present tense
and the perfect, founded on the scopal relations between the modal and the present
and perfect operators, is, in spite of appearances to the contrary, perfectly compat-
ible with the claim of the decomposition of the concept of time into degrees of
modality. Let us consider the examples in (6).
(6) John may/might be here tomorrow.
John may/might be here now.
John may/might have been here yesterday.

I shall leave aside the question as to whether the categories of evidentiality and epistemic modality
are related and how. See Aikhenvald (2004) for a powerful defence of the dissociation and van der Auwera
and Plungian (1998) and Jaszczolt (2009) for arguments in favour of subsuming evidentiality under
modality. See also de Saussure (in press) on evidentiality and epistemic modality with modal verbs and
the epistemic future in French. He argues that for such constructions, epistemic and evidential effects are
entirely pragmatic and disjoint.
Temporality and epistemic commitment 201

The degrees of dissociation from the eventuality of John’s being at a deictically

specified place at a deictically specified time persist across P, N, and F. They also,
arguably, interact with the concept of temporal reference: may in John may be here
now differs from may in John may be here tomorrow in that the truth-maker for the
first is simple: you, so to speak, ‘go and check’. The truth-maker for the latter either
does not exist, as Aristotelian tradition has it, or exists in the present as a prediction,
or otherwise exists in the form of some atemporal, B-theoretic entity. In sum, it
would in principle be compatible with Kratzer’s (1981) or Condoravdi’s (2002)
construal to suggest that temporality be composed of such diversified modal atomic
concepts as, say, mayn > mayn þ 1 > mayn þ 2 . . . etc, where ‘>’ stands for the ordering
of the strength of expressed commitment. But, to repeat, we can either hypothesize
here either different degrees of detachment, and supplement the theory with a story
concerning the values for the degrees n, n þ 1, n þ 2, etc., or, alternatively, we can
adopt a hypothesis that there are different types of detachment, as long as we avoid
infinite regress: qualitative differences in detachment pertaining to P, N, and F do
not affect the thesis of supervenience as long as we can peg the differences on
linguistic semantic and conceptual distinctions, while maintaining that the core of
graded epistemic commitment is shared. In other words, ‘there is no temporal
difference without a modal difference’, but modal differences themselves are more
than meets the eye of a grammarian. It takes a contextualist approach to meaning,
where the truth-conditional content of the uttered sentence draws on a variety of
sources that are available in the situation of discourse (as e.g. in Recanati 2004, 2010;
Jaszczolt 2005, 2010), to account for modality as the underlying concept for the
representation of time.
To sum up: both options, (i) the direct-quantitative view (henceforth DQ), that is
the direct reliance on values of n, n þ 1, n þ 2, . . . , and (ii) shifting the arguably
qualitative differences between P, N, and F to the also arguably qualitative differ-
ences between modal expressions on their occasions of use (the modal–contextualist
view, henceforth MC), are feasible hypotheses and candidates for answers to the UQ.
DQ and MC both rely on a well-acknowledged property of modals, namely their
malleability in context. Different contextual requirements allow modal constructions
to assume different interpretations. Von Fintel and Gillies (2011: 108) dub them
“context-dependent quantifiers over a domain of possibilities”. This is precisely the
kind of property that contextualist accounts are well equipped to represent in that
the truth-evaluable content does not rely strictly on the logical form of the sentence
but instead incorporates modifications of it—all the way from logical weakening or
strengthening (Recanati 1989) to, on radical accounts such as DS, replacement of the
logical form of the uttered sentence with that pertaining to the content intended by
the speaker when conveyed indirectly. Moreover, it has to be noted that in view of
the perspective adopted here, epistemic modality allows for the scale from full
commitment to its lack, and hence we are talking here not merely of ‘modal
202 Kasia M. Jaszczolt

constructions’ as they are standardly delimited, but of all those acts of communica-
tion that convey an epistemic attitude, including those made in, say, straightforward
present or past tense forms.

9.4 In support of supervenience on modality: tensism or tenseless time?

The supervenience of time on modality seems to fit well with the B-theory of time:
time is not dynamic, it does not pass. Arguably, it would even be difficult to theorize
about the common idea of the passage of time in that the question of ‘how quickly it
passes’ is impossible to answer in quantitative terms.12 Equally, it would arguably be
nonsensical to point out causal links between eventualities if some of them were real
and others unreal, as A-theorists have it, for “in order for there to be a temporal
relation between two events there must be the two events that stand in that relation”
(Oaklander 2002: 74; see also Oaklander and White 2007; Le Poidevin 2007, 2011).
If all there is, is duration and precedence, then the DQ explanation fits the bill:
events are ordered on the scale of later-than/earlier-than, and they occupy certain
metaphysical and conceptual space in a way that allows the thinking agent to form
an attitude to them—an attitude that includes the degree of certainty. This graded
attachment in turn translates into the graded belief that a certain state of affairs is
(tenselessly) real, which in turn produces the illusion (‘time in the mind’) of ‘was / is
/ will be real’.13 Supervenience so construed is in fact perfectly compatible with the
B-theoretic idea of supervenience already put forward in the philosophical literature,
just taking it one step further. For example, Sattig (2006) points to the supervenience
of time on atemporal spatiotemporal location. This pertains to supervenience of
psychological time on real space–time. Equally, real space–time gives rise to
human attitudes towards it that result in the illusion of time passing.
The outcome of this B-theoretic explanation is that we are juxtaposing the
metaphysical domain of space–time with the psychological domain and its concep-
tion of dynamic reality, and adding to it the middle conceptual level of attitudes and
degrees of belief. This shifts P, N, and F even further away from the focus of interest
than standard B-theorists would have it. Time is not only ‘in the mind’, it is in the
mind as a complex rather than primitive concept. That is to say, at some level of
atomic concepts, time passing is not only not real, but P, N, and F are not even in the
mind at all. Instead there are degrees to which states of affairs are accepted by the
agent. The bonus is that if the evidence for past events (memory, written records,
etc.) differs from evidence for present (‘go and check’) and future (planning, fixing,

See for example Olson’s (2009) apt argument questioning the sense of the concept of the rate of
time’s passage.
But see Tallant (2007) on the alleged impossibility to think in terms of B-time and Oaklander and
White’s (2007) rebuttal.
Temporality and epistemic commitment 203

strong prediction through causal links, etc.) events, this is just a problem with the
nature of evidence, not with time itself.14 Time can remain underlyingly modal and
quantifiable as modal.
But is the thesis of modal foundations of internal time also compatible with the
A-theoretic outlook? Intuitively, it should not be compatible in that if reality is
tensed, then one would have to reconcile the human arguably underlyingly tenseless
conception of reality with the flow of real time. However, the stipulation that reality
is tensed comes in various flavours. The most common of them is presentism,
according to which only present states of affairs are real. Despite the objections
mounted by those who point out that past causes have to be real and so do future
effects, presentists claim that P and F are unreal (e.g. Prior 1967, 1968, or, on a
linguistic, modal explanation, Ludlow 1999), or have a lower degree of reality than
N (Smith 2002). It is the latter, ‘degree presentism’, that presents an attractive
analogy to our view of time as degrees of epistemic modality.15 Quentin Smith
(2002: 119) aptly captures the intuitions as follows:
It seems intuitively obvious that what I am doing right now is more real than what I did just
one second ago, and it seems intuitively obvious that what I did just one second ago is more
real than what I did forty years ago.

The problem with this intuitive view is that, if we are to accept that the adjective
‘real’ is gradable, we are committed to degrees of existence and therefore to highly
contentious Meinongian entities. But perhaps the intuition of graded reality can be
preserved when it is properly embedded not in A-theory but instead in B-theory:
time is internal, P, N, and F are in the mind, and so is the intuition of the degrees
of reality. Let us try to superimpose the following passage from Smith (2002) on to
B-theory, indexing it as appropriate. The material in square brackets comes from my
attempt at this shift to the B-theoretic perspective. Although it is clearly incompat-
ible with Smith’s intentions, it produces the right result for my view.
The degree to which [a representation of] an item exists [in the agent’s mind] is proportional
to its [perceived] temporal distance from the [agent’s] present; the [agent’s] present, which
[often] has zero-temporal distance from the [agent’s] present, has the highest (logically)
possible degree of existence [in the agent’s mind].
Smith (2002: 120)

The transfer from the A- to the B-framework clearly removes Smith’s overall
objective. But it preserves the method for the purpose of my theory, and therefore
it would not be right not to employ it while proposing a graded account. It is true
that degrees are allocated elsewhere (into modality), but otherwise the method relies

Ludlow (1999) explains this difference by reducing the past to evidentials and the future to modals.
For a discussion of asymmetric views see also Bourne (2006).
204 Kasia M. Jaszczolt

substantially on Smith’s. In addition to the shift, there is, however, one alteration in
my proposed annotation that makes a qualitative difference even to the method of
graded departure itself. That is the addition of the adverbial often, to produce in
the above passage: “the [agent’s] present, which [often] has zero-temporal distance
from the [agent’s] present.” The alteration is necessary because once we move to
the domain of internal time and to the domain of the epistemic, there are also
degrees of commitment to eventualities reported as present, as was exemplified in
(2a–b) in the use of present-time-referring modals. Clearly, present-time-referring
evidential constructions, even in languages without (grammatical) evidentials such
as English, also testify to a graded classification: compare (2a–b) repeated below with
(2c) that seems to lie in-between on the scale of epistemic commitment of the
speaker (see also Faller 2002b for a well-developed linguistic analysis).
(2) a. Aly is in Kuala Lumpur now.
b. Aly may be in Kuala Lumpur now.
c. Aly seems to be in Kuala Lumpur now.
Finally, degree presentism, so transposed onto a ‘degree internal time’ view, well
supports the DQ solution to our unresolved question as put forward in the previous
section: as Smith (2002: 120) says, we have a difference of degree, rather than a
difference of kind, between P, N, and F, and these degrees are “immediately given in
our phenomenological experience”. But next comes an unexpected bonus: Smith
claims that for him, ‘degrees of existence’ mean how an agent experiences existence
(p. 122). So, losing a limb does not result in a lower degree of existence, but
experiencing an event as past rather than present does. Well, if this is not the shift
to B-theory I have just proposed then it is not clear at all how Smith can reconcile
this claim with his tenacious holding on to presentism. There are degrees, to be sure,
but they are in the conceptualization, not in the world. A fortiori, they are in the
language, not in the world; they are only in the world in the sense in which agents
that hold propositional attitudes, and thereby these agents’ mental states, are in the
world. Smith has to resort here to relational properties: entities exemplify, tense-
lessly, relational properties (‘being alive over 2000 years earlier than now’), produ-
cing tensed facts. This results in a rather weak proposal whereby we have merely a
shift from time as the explanandum to relational properties, the latter apparently
acquired anew as time passes.16 However, all in all, it is evident from the use I
have made of the graded view that the ‘gradation of reality’ is logically independent

On Smith’s proposal, shifting from N to F results in the acquisition of a new property that is “the
past-time version of the presently possessed property” (2002: 135). With every infinitesimally small unit of
time passing, individuals lose present properties and acquire past-time versions of them instead. The
unwelcome regress is diaphanous here and the tenacity with which Smith holds on to tensed facts is only
the more puzzling.
Temporality and epistemic commitment 205

of the tensed/tenseless reality dispute—a fact about which, no doubt, Smith would
have no qualms.
In sum, it seems that the ‘graded reality’ view would fare better if one didn’t
conflate the metaphysical level (reality) with the epistemic one. On B-theory, we can
separate them, tame allegedly tensed reality to the position of the conception of
reality, intuitive time so conceived to internal time, time in the mind; and the bonus
is that we can use linguistic evidence to support the view of the speaker’s graded
commitment, be it on the DQ or MC construal. Linguistic evidence for the modal
foundations of time is what I provided in Representing Time; the question of the
exact mapping between the modal and the temporal is what, without reinventing the
wheel, I am trying to resolve now, having demonstrated with the help of Smith’s
arguments that DQ is a tenable solution.
There is one other version of presentism that needs to be explored in connection
with our question of the location of the epistemic modal view vis-à-vis the question
of reality of time. Presentism can also be held in an ‘asymmetry’ version, whereby
P and N are considered to be real and F unreal.17 ‘Now’ is conceived of as the edge of
real time, and the latter simply keeps growing. Needless to say, to hold water, this
asymmetry view has to be supported by a good argument from causation (Tooley
1997, 1999) or relational concept (‘x (past) is real as of y (present)’; Button 2006,
2007).18 The asymmetric view is also in accord with common intuitions but in
respects different from those supporting the degree view: the past has already
happened and therefore is fixed; the present is fixed in the sense of being actual
and observable; the future on the other hand, is not fixed. As Tooley (1997, 1999)
says, an alleged fact is not a fact if it can still be prevented, so there are no future
facts. Deterministic views aside, common intuitions are well captured by this
asymmetric view. Moreover, in view of our earlier discussion of Jackson (2011) on
the properties of philosophical concepts, intuitions are what should inform the
theory. Causes have to be real at the time when their effects take place, so they
have to be past or present, but the effects do not: they can be in the future.
However, independently of his argument for tensed reality (in that, for Tooley,
accepting causation presupposes accepting a tensed world), Tooley also argues that
tensed concepts can be broken down into, and analysed in terms of, tenseless ones.
He admits that tensed properties cannot be fully explained in tenseless terms, and
hence his position on the pertinent supervenience becomes complicated, but it is
rescued by relational properties: what facts are actual depends on the time at which
they are assessed. Again, analogous to the use we made above of degree presentism,
the asymmetric A-theory advocated by Tooley can be broken down, as far as the
argumentation is concerned, into two independent parts: his supervenience view and

See Gosselin (this volume) for an account of French tenses based on the notion of asymmetry.
See also Tallant’s (2011) defeat of the latter.
206 Kasia M. Jaszczolt

his stance on real time. It is the first that interests us. And, analogous again, his view
on real time is perfectly compatible with the modal picture of the degrees of
epistemic detachment thanks to the adoption of the supervenience of the tensed
on the tenseless. As Tooley observes, one cannot have direct, non-inferential know-
ledge of the future, nor can one have direct perception of the past. One has present
beliefs about the past and present beliefs about the future, and these, qua propos-
itional, epistemic attitudes, exhibit properties that epistemic attitudes do, such as
being more, or less, certain that a state of affairs did happen, will happen, or is
happening. Or, to further promote the DQ hypothesis, in the early stages of language
acquisition, children believe things “this much ( þ a gesture indicating a unit of
length)”—a trait that disappears in early childhood but is indicative of the quanti-
tative tendency in approaching knowing and believing that something is the case.19

9.5 The complex concept of time and its foundations

Languages employ an enormous variety of means to convey temporality. Some
make use of grammatically encoded temporal distinctions; these differ in kinds
and numbers. Others do not grammaticalize temporal reference. Some have a rich
repertoire of aspectual distinctions, reflecting situation-internal time, repetitions,
habits, and perspectives on eventualities. In some, tense and aspect marking is
compulsory in a sentence. In others, it is optional as long as temporal information
can be assumed or inferentially retrieved. Some rely principally on the lexicon,
mostly on temporal adverbials. Some have a rich system of evidential markers that
help with the inference of temporality. If one were to attempt to classify these means
with the aim of giving linguistic support to the A- or B-theories of ‘real’ time, it
would become apparent that both A- and B-theoretic concepts alike are lexicalized,
grammaticalized, and assumed in pragmatic inference. English grammaticalizes
P and N, but F is tightly interwoven with modality through the various grammati-
calization stages of will. Arguably, will originated as a modal, then shifted to a
marker of tense, and now conveys tense with modal colouring (see Fleischman
1982 on bidirectional semantic shift). French grammaticalizes all of them, P, N,
and F. These are A-theoretic terms, or terms that a B-theorist would place in the
category of mental representation of time. Swahili has consecutive tense: the con-
secutive-tense marker ka signals the order in which events occur, independently of
their placement in P, N, or F. It seems then that consecutive tense can be classified as
a B-theoretic device. Thai uses modal particles whose default reading conveys
temporality. For example, the modal auxiliary d1ay1II in Thai by default assumes

This is well attested among pre-schoolers and in early primary-school children but the observation is
in urgent need of a systematic empirical study.
Temporality and epistemic commitment 207

past time reference (Srioutai 2006; Jaszczolt and Srioutai 2011). For Srioutai, modal-
ity is the conceptual foundation, the explanans in the theory of modal supervenience,
while the A-theoretic concept of temporal location is the explanandum.
What emerges from this diversity is that the lexicon and the grammar are not
reliable guides to the conceptualization of time; one can convey the intentions
independently of what is encoded and how. Memories, anticipations, current ex-
periences, as well as their length, ordering, kind of evidence, and certitude are all
employed to interrelated ends. As I have argued elsewhere (Jaszczolt 2012), there is
compelling evidence in support of the view that beneath the linguistic diversity in
conveying temporal reference there is a universal concept. The question is whether
this a basic, primitive concept or a complex one. Now, if B-theorists are right and P,
N, and F are in the mind, then P, N, and F have to be analysable in terms of tenseless
entities. If A-theorists are right, there still can be, or even must be, tenseless
analysans, as Smith’s and Tooley’s reliance on relational properties (discussed in
Section 9.4) suggests.
It seems then that a complex, molecular (as opposed to atomic) concept of time is
supported on all fronts. Now, as I argued above, if the concept is underlyingly modal,
then the DQ solution applies and we can have different values for different eventu-
alities. For example, remembering V-ing (something) will reflect a stronger degree of
epistemic commitment than remembering one’s V-ing, which in turn is stronger
than remembering that one V-ed. For example, in DS, (7a) will correspond to
a higher value of D in ACCD ¢ S than (7b), and (7b) will have a higher value of
D than (7c).
(7) a. Tom remembers PRO saying that time doesn’t flow.
b. Tom remembers his saying that time doesn’t flow.
c. Tom remembers that he said that time didn’t flow.20
As Higginbotham (2003: 504) says, gerundive complements of the verb remember
pertain to events of memory rather than remembering propositions. Concerning (7a)
and (7b), he also makes an apt observation that it is possible to ‘quasi-remember’
someone else’s experience as one’s own (a fact that is well attested empirically), and
therefore (7b) differs in strength from (7a). What Higginbotham calls a distinction
between a memory event and remembering facts on the one hand, and remembering
and quasi-remembering on the other, easily translates into our scale of epistemic
commitment by associating (hypothetical at present) values with D.
What remains is to assess what this diversity and the interaction of various means
of conveying temporality found in different languages implies for the other possible
answer to the UQ, namely the MC view. To repeat, this is a solution in which modal

I am ignoring the sequence of tenses phenomenon as it does not bear on this example.
208 Kasia M. Jaszczolt

expressions, and also any expressions that convey degrees of epistemic commitment
of any degree, acquire their meaning through contextual modifications. The view is
clearly supported by cross-linguistic empirical evidence. Lexical, grammatical, and
pragmatic means of conveying temporal reference interact, and languages employ
different combinations of these means. By the same token, modal expressions can
undergo grammaticalization to markers of tense (discussed in the example of
English will), can convey temporality as their default sense (discussed in the example
of Thai d1ay1II), or can convey temporality in context, via pragmatic inference, as in
the ambiguous use of may in (8).
(8) He may be in London (now/tomorrow).
Adopting Gricean or post-Gricean principles of rational conversational behaviour as
pragmatic universals (see von Fintel and Matthewson 2008; Jaszczolt 2012), we
assume that the speaker would not have uttered (8) if he/she had not been in a
position to assume that the utterance would not be ambiguous to the addressee.21
In a normal conversational situation, temporality can be left to pragmatic inference
or recovered automatically, by default. In DS-theoretic terms, the processes respon-
sible are either CPI (conscious pragmatic inference), or SCWD (social, cultural, or
world-knowledge defaults).22 What matters for the UQ is that on a contextualist
construal of meaning, the temporal location in the past, present, or future can be
conveyed via this interaction of processes of utterance interpretation and the super-
venience on epistemic modality can be left intact.

9.6 Conclusion
In this brief sequel to Representing time (Jaszczolt 2009) I addressed the question
that was left unresolved there, namely:
UQ: If the concept of time is underlyingly modal and supervenes on the degrees of epistemic
commitment to (or detachment from) the narrated eventuality associated with the speech act by
the speaker, then what is the exact correlation between the value of D that represents this degree
of commitment in DS and the type of temporal reference: P, N, or F? In other words, since the
thesis of supervenience makes it necessary that there is no temporal difference without modal
difference, how exactly are they correlated?

I entertained two possibilities in this chapter: (i) that the differences between P, N,
and F are underlyingly quantitative rather than qualitative (which I called the direct-
quantitative, or DQ view), and (ii) that the differences are qualitative and the
value of D is contextually established (the modal–contextualist, or MC view).

Excluding special situations in which ambiguity may be intentional.
See Jaszczolt (2010) for details.
Temporality and epistemic commitment 209

Next, I focused on the DQ view and assessed its compatibility with the tensed and
tenseless theories of time, concluding that both uphold it. Finally, I argued that the
MC view was also tenable as an answer to the UQ, and was independently supported
by cross-linguistic data when the contextualist approach to meaning was adopted.
There is, of course, an independent line of argumentation that comes from the
laws of physics. It goes like this. Time may not flow, but events are ordered in a
unidirectional sequence. A powerful argument for the asymmetry between the past
and the future comes from the second law of thermodynamics, according to which
the disorder of a closed system (its entropy) increases with time. In practice, this
translates into the fact that physical processes are irreversible: a china cup can break
but it cannot reassemble itself back from the pieces to a cup. The arrow of time is
unidirectional; the asymmetry is explained through external, physical evidence. But
we have to remember that real time does not flow: the past and the future have no
objective sense. Put simply:
The labels ‘past’ and ‘future’ may legitimately be applied to temporal directions, just as ‘up’
and ‘down’ may be applied to spatial directions, but talk of the past and the future is as
meaningless as referring to the up and the down.
(Davies 2012: 11)

Moreover, this physical evidence also fuels an argument in support of the asymmetry
of psychological time (the human concept of time), which is exactly what we want
here. Just as entropy is unidirectional, so is memory: memory of an agent increases
and produces the illusion of the flow of time.23 Both the DQ and the MC views
proposed here are compatible with this explanation of the essential asymmetry of
time. And this convergence of a conceptual, epistemic reduction of time with
naturalistic reductionism is precisely the correct way to proceed in order to under-
stand the phenomenon and the concept of time.24

There is also a hypothesis, which will not be pursued here, that the illusion of the flow of time
can be explained by quantum processes in the brain.
Sklar (1981: 126) puts it convincingly as follows. Naturalistic reductionism means “identificatory
reductions of substances and properties to those more ‘scientifically’ fundamental”. On the other hand,
conceptual reductionism involves an analysis of certain propositions in order to determine a set of those
propositions which “could serve as epistemic warrant for them”. As he observes, naturalistic reductionism
owes its success precisely to such conceptual reductions: “much of the transition from space and time to
relativistic spacetime proceeds by just such an epistemically motivated ‘reductionist’ critique”. To put it
crudely, to understand the phenomenon of time one has to ask questions about the concept of time as the
latter may reveal more than just facts about language and the mind.

An account of English tense

and aspect in Cognitive Grammar


10.1 Introduction
In this chapter, I will offer an overview of the theoretical prerequisites underlying
the semantic analysis of tense and aspect marking in Cognitive Grammar (CG), with a
specific focus on the English verb paradigm. In doing so, I will adhere to the ortho-
doxies as presented mainly in Foundations of Cognitive Grammar (Langacker 1991)
and a number of subsequent key publications, supplemented with some original
comments on the interaction between tense and certain aspectual meanings. I will
pay particular attention to the treatment of grammatical aspect in CG, both in terms of
the systematic contrast between perfective vs. imperfective aspect and with regard to
the uses made in English of the progressive and the perfect construction, respectively.
The main thesis defended in this chapter is that tenses, qua ‘grounding predications’,
and, though somewhat less conspicuously, aspectual markers combining with them
refer essentially to modal concerns on the part of the speaker/conceptualizer, and that
the temporal (and modal) meanings conveyed by the resulting constructions should be
analysed as instantiations of a common epistemic schema. Consequently, an exhaust-
ive and unified semantic analysis of tense/aspect marking, in any language, should
ideally be concerned with establishing such a core schematic meaning for a given
construction, to be characterized in modal terms, elaborating on the various meaning
types that emerge in interaction with concrete contexts of use in the form of some
kind of semantic map or network.
In Section 10.2, I will start off with a brief description of the symbolic view of
grammar which is adopted in CG, introducing the concepts and terms needed to

I thank Astrid De Wit and the editors of this volume for their useful comments and suggestions. Of
course, all remaining errors or infelicities are my responsibility.
An account of English tense and aspect in Cognitive Grammar 211

discuss processes at the semantic pole of verbs, including such extremely schematic
ones as tenses, in terms of ‘complex temporal relations’. In Section 10.3, I sketch the
grounding model proposed in CG to deal with various grammatical strategies for
establishing reference in the realms of the nominal and the verbal domain. Section
10.4 will then focus on the basic semantics of English present and past tense marking
as instances of grounding predications. Next, in Section 10.5 I will discuss the
contribution of perfective vs. imperfective grammatical aspect: though these markers
are not really part of the English grounding system, their interaction with tense
markers is pervasive and systematic and should be included in an account of English
tenses. A fuller analysis of present and past time reference in English grammar,
considering dedicated markers for the expression of progressive and perfect aspect
(and their intrinsic polysemy), will be given in Section 10.6. I will end with some
concluding remarks in Section 10.7.

10.2 A symbolic alternative

CG adopts a subjectivist or conceptualist view of meaning, in which linguistic
semantics is equated with conceptual structure. This is done within the larger effort
to present a linguistic theory in which grammatical “description requires nothing
more than symbolic elements (i.e. pairings between semantic and phonological
structures)” (Langacker 1995: 89; emphasis in the original). Such a reduction
of grammar to symbolic relationships, involving structured conceptualizations,
provides a radical alternative to theories upholding both the autonomy of grammar
in relation to meaning and the objectivist (truth-conditional) basis of semantics, and
has some noteworthy implications for those areas of grammar where highly abstract
meanings are posited—as opposed to no meaning at all, a view espoused by most
standard modern theories.
Considering the semantic pole of linguistic expressions, a few initial clarifications
are in order with regard to some basic assumptions that are made. First of all, a
semantic representation can always be considered at various levels of abstraction or
schematization. Thus, a beagle can alternately be described as a dog, mammal,
animal, or even thing (with the latter conceivably constituting the most schematic
description, of any noun for that matter),2 and verbs like be, do, and have decidedly
represent the least specific of verbs in the English lexicon, which makes them
all the more eligible for performing certain grammatical functions, like acting
as auxiliaries, after due processes of grammaticalization. Similarly, one could say,

If properly defined, that is. The definition of a thing as a bounded region in some cognitive domain
(prototypically space, but possibly others like time, colour, the visual field, etc.) allows CG to propose this
concept as the notional, thus universal, basis for the grammatical category of nouns. This is a contention
that many other theories would firmly contest.
212 Frank Brisard

a grammatical morpheme like a (periphrastic or inflectional) tense marker can

be characterized as a schematic verb: designating, that is, a schematic process
whose details are filled in or ‘elaborated’ by the main verb with which the marker
Next, we need to define what a process is. A process is a complex relation that
saliently involves the basic cognitive domain of time. It presupposes at least one
nominal participant (argument) but focuses on the interconnection between that
participant and either itself, some configuration in a relevant domain, or some other
participant, rather than on the participant(s)/configuration proper. This intercon-
nection is then said to develop through time, showing either a change of state
(dynamic predicates) or none (stative predicates). For a verb, both entities and
their interconnections are semantically relevant, but it is only the interconnections
and their development that are actually attended to, while nouns do the job of
specifying the entities participating in them. This—the contrast between nouns and
verbs—is explained in CG by referring to the notion of construal: the words
explosion and explode, for instance, may refer to the same objective content, but
construe that content differently. The noun construes it as a region (the conception
of a set of interconnected entities involved in the exploding event), while the verb
focuses on these developing interconnections themselves. One important facet of
construal is the relative prominence that a conceptualizer attaches to an element in a
complex configuration. Again, with verbs, we can say that they designate a series of
relations or states and presuppose the associated nominal entities, which are,
however, not actually attended to. These presupposed entities as well as the cognitive
domain that frames the whole conceptualization belong to the base of a certain
expression and are consequently not in focus when that expression is conjured up.
What is conceptually designated, or ‘profiled’, receives a special kind of prominence,
which at once determines the grammatical category of an expression. With verbs, the
profile is a complex temporal relation (a process), and with nouns, it amounts to a
thing (i.e. the conception of a bounded region). Profiling is indicated by heavy lines
in the CG diagrams that follow. At the level of the clause, then, the profiles of the
central nominal participants (receiving the traditional grammatical roles like subject
and object, etc.) and that of the finite verb form are integrated into one complex
relational representation of the scene evoked by the clause, pitched against some
conceptual background. Upon this representation conceptualizers invariably impose
a figure/ground3 organization: the figure or nominal entity accorded the most
prominent status is called the trajector, and a prominent entity other than the
trajector is referred to as a landmark. Thus, in Jude gave the money to me, the

The concept is well known from Gestalt psychology and reflects the general inspiration found by CG
in psychological studies of cognitive processing, most notably in the domain of (visual) perception.
An account of English tense and aspect in Cognitive Grammar 213

subject acts as the trajector and is by definition presented as the scene’s most salient
participant. In I was given the money by Judy, in contrast, the same scene is described
from the perspective, so to speak, of the recipient, who accordingly gets assigned
primary prominence (subject/trajector status), thereby demoting the agent and
patient/mover of this transaction to the status of secondary and tertiary landmark.
Similar observations hold for sentences like You look like your mother vs. Your
mother looks like you, which are truth-conditionally equivalent, but not conceptually
Finally, we need to consider the grammatical relevance of matters of scope, since
this will be important in our discussion of grammatical aspect in particular. If we say
that an expression profiles a given entity, this means that this entity constitutes
the ‘immediate scope’ of attention upon any particular usage event. Given an
encyclopaedic conception of semantics (i.e. of how much of a cognitive domain is
evoked), no sharp boundary can be drawn between knowledge that is and that is
not linguistically relevant for the use of an expression. Still, certain elements in a
semantic configuration will be more directly relevant than others, and these will
then be included in different types of scope. For instance, the conception of a
finger constitutes the immediate scope for the characterization of nail, while a
conception of the human body makes up its ‘maximal scope’ (observe that we do
not say *handnail, but rather fingernail). One can also say that the immediate
scope contains elements which are put ‘onstage’ as focused objects of conception,
while those which remain ‘offstage’ are again included in the maximal scope.
Specifically for the following discussion of grounding (the act of relating an entity
to the speaker/conceptualizer’s position and thereby identifying it), these notions
(immediate/maximal scope, onstage/offstage) also correlate with the contrast
between ‘objective’ and ‘subjective’ construal: when conceiving of an object, the
features of that object are objective while everything associated with the act of
conception itself, such as a conceptualizer to do the conceiving, is said to be
subjective. By definition, an expression’s profile is construed with a high degree of
objectivity, being the focus of attention within its immediate scope. Consequently, if
a tense marker is a schematic verb, then it must profile a schematic process instead
of the temporal relation which it also expresses but does not designate (i.e. which is
construed subjectively, offstage, as part of the maximal scope).

10.3 Grounding
The grounding model in CG is primarily meant to capture the conceptual analogies
between various grammatical strategies speakers have at their disposal to establish
reference to nominal and verbal (clausal) entities (see also Langacker 2002a, 2002b).
Thus, ‘grounding predications’ are defined as designating a grounded instance of a
214 Frank Brisard

(a) Grounding Element (b) Grounding Element

(Nominal) (Finite Clause)

tr lm

Rg Rg


Figure. 10.1 Nominal and clausal grounding (Langacker 2002a: 13).

nominal or clausal type conception, whereby the notion of the ground is used to
refer to the speech event, its participants, and its immediate circumstances (includ-
ing, for clausal grounding, the time of speaking). Importantly, grounding predica-
tions do not by themselves indicate the relation between a grounded instance and
the ground (G), since their cognitive import lies in identifying (or profiling) the
entity that is at issue. Rather, the grounding relation (Rg) figures in the background
conception of any true grounding element, as illustrated in Figure 10.1, which
represents the semantic pole of nominal and clausal grounding predications.
In Figure 10.1, profiling is rendered in bold and establishes the immediate scope of
an expression, whereas other semantically relevant elements feature in the back-
ground, the expression’s maximal scope. A nominal grounding predication like this
turns a noun, which refers to a type of thing, into a nominal, which refers to a
particular instance of the type labelled by the noun. In the nominal this pen, the way
this reference comes about is due to elements in the relational meaning of this
particular demonstrative, namely that the speaker is referring to a pen near the
speaker and hearer and known to the speaker. But this meaning, Rg, is not what
the nominal actually refers to: the contribution of this in particular gets the nominal
to refer to the actual targeted object of conception, indicated in bold as the profile,
whereas the way this is done (by characterizing the relation with the ground in a
certain way) is subjectively construed, because that is not what a demonstrative like
this is ‘about’. With clausal grounding (Figure 10.1b), what speakers attend to when
they utter a finite (i.e. tensed) clause is the conception of a specific instance of
interaction between a trajector (tr) and a landmark (lm) (the prototype of a verbal
conception: say, a girl kissing a boy), which may be identified on the basis of the
relation between that conception and the current ground at the time of speaking. For
example, in A girl kissed a boy, that relation is presented as one of precedence
between the designated event and the time of speaking (the ground). As with the use
of this, the use of past tense marking is not motivated by the speaker wanting
to talk ‘about’ pastness per se, but rather by the fact that indicating this relation will
An account of English tense and aspect in Cognitive Grammar 215


Unknown Reality

Known reality

Figure. 10.2 Elaborated Epistemic Model (Langacker 1991: 244).

help the hearer identify which particular (if non-specific) instance of two entities
kissing the speaker intends to refer to.
Clausal grounding predications (in English, tenses and the core modal auxiliaries)
implicitly conjure up an ‘epistemic model’ that represents the Conceptualizer’s (C)
knowledge about the reality status of certain situations at a given point in time.
While modals locate situations in various portions or Irreality (consisting of Non-
and Unknown Reality), tenses invariably portray a designated situation as Real. This
‘Elaborated Epistemic Model’ is depicted in Figure 10.2. Within Reality, a further
distinction can be made between situations coinciding with the ground/time of
speaking and therefore, from an epistemic perspective, directly available to the
conceptualizer (constituting Immediate Reality) on the one hand, and, on the
other, those situations that are considered real/known yet removed from the ground
and thus lacking certain phenomenal qualities. The most salient of these qualities,
perceptual access, is closely associated with our folk notion of the Present, which, for
all analytical purposes, will be identified with the ground/time of speaking (and
consequently has a certain temporal depth, in contrast with many other, including
formal, approaches to time in grammar). If we include this reference to the ground,
then we can superimpose a ‘Time-Line Model’, as in Figure 10.3, upon the more
schematic epistemic model, whereby the squiggly line in the front part of the
growing ‘cylinder’ of reality represents the utterance of a clause (and the time it
takes to do so).
In this model, the Past is the region that does not coincide with the ground. It is
still considered part of reality, and the temporal relation to the ground is one of
216 Frank Brisard

Past Present Future

Figure. 10.3 Time-Line Model (Langacker 1991: 244).

anteriority/precedence. The Future does, strictly speaking, not belong to reality but
should be characterized as possible non-reality.4

10.4 Tense in CG
Bearing in mind the concepts introduced above, it is now possible to provide a
schematic definition of the meanings of all of the clausal grounding predications that
may be distinguished in a language. In English, there is a basic formal opposition
between the presence vs. the absence of a modal and, within the latter category,
between the presence vs. the absence of the ‘past tense’ morpheme. The first
opposition signals the modal contrast between Irreality and Reality, and the second
differentiates between regions within reality that are distal (Non-Immediate) vs.
proximal (Immediate) to the ground. These regions are typically, but not always,
interpreted in temporal terms as representing Past vs. Present. The grounding
predications that refer to them are called past and present tenses, and their respect-
ive semantic configurations are rendered in Figure 10.4.5
Present tenses situate an objective proposition P within the conceptualizer’s
Immediate Reality, i.e. as directly coinciding with it. This (grounding) relation is
indicated by the dashed arrow. The focus lies on the designation of P, whose

See also Prosser (this volume) and Ludlow (this volume) on the metaphysics of the past and the
In this and the following figures, MS indicates an expression’s maximal scope of predication
(cf. Section 10.2), i.e. all of the conceptual content that is evoked, ‘on-’ or ‘offstage’, in the use of the
expression and that has some potential linguistic (semantic or pragmatic) relevance.
An account of English tense and aspect in Cognitive Grammar 217

(a) Immediate (‘Present’) (b) Non-Immediate (‘Past’)




Figure. 10.4 The English present and past (Langacker 2011: 74).

semantic content is provided by the lexical verb, and which constitutes the expres-
sion’s profile or immediate scope. Everything else relevant to the characterization of
this tense (i.e. an epistemic model of reality and the grounding relation proper) is
considered part of its maximal scope and as such construed subjectively. In contrast,
with past tenses P is situated outside of Immediate Reality or, in temporal terms,
preceding it. It is still, though, accepted as real by the conceptualizer.
A matter complicating this rather simple picture of tense meaning in English is
the interaction of tense with lexical aspect. In particular, in the English present tense
paradigm it should be noted that not all verbs may take simple present marking to
refer to an ongoing situation in the present. In order to explain this, a basic
opposition is assumed in CG between two verb types, those denoting dynamic
processes and those denoting stative processes.6 A dynamic predicate like kick the
ball indicates a change of state and is thus internally heterogeneous, and it has a
bounded temporal profile, as can be seen in Figure 10.5a. Boundedness is indicated
by the full inclusion of the verb profile inside the immediate scope. Notice that
Figure 10.5 contains representations of the meanings of non-grounded verb types,
which explains the absence of any reference to the ground.7

The labels in Figure 10.5 below reflect the standard CG terminology for talking about lexical verb
classes. This terminology presupposes a unidimensional approach to lexical and grammatical aspect.
However, to avoid confusion with the more general tradition of restricting the use of the terms ‘perfective’
and ‘imperfective’ to grammatical aspect only, I will use the terms ‘dynamic’ and ‘stative’ to refer to
lexically perfective and imperfective verb types, respectively.
Strictly speaking, for a language like English it is impossible to talk about a verb’s lexical aspect in
isolation, i.e. as divorced from considerations of grammatical aspect. This would only work for languages
with a relatively poor system of grammatical aspect, like German (where the interpretation of a sentence
as, e.g., telic or not depends on the verb’s lexical aspect). Most other languages distribute aspect over the
lexicon and grammar, such that verb types systematically interact with aspectual markers and vice versa.
The diagrams in Figure 10.5 should be read accordingly: for instance, Figure 10.5a shows that dynamic
verbs are conceived in their entirety and as bounded in interaction with perfective aspect, while stative
verbs (Figure 10.5b) are unbounded, again in interaction with perfective aspect. Of course, boundedness,
telicity, and stativity in a way count as intrinsic qualities of a given process type and are as such given in
the lexicon, but typically the only way to find out which lexical aspect they give rise to is by looking at what
happens with a verb in interaction with grammatical aspect. For instance, in Pichi the verb tinap ‘stand
218 Frank Brisard

(a) Perfective Verb (b) Imperfective Verb


t t

Figure. 10.5 Verb classes (Langacker 2001: 257).

A stative verb like love (Figure 10.5b) indicates a state that is unbounded within
some imposed immediate scope, extending beyond the actual profile in both tem-
poral directions, and internally homogeneous (i.e. any instance of loving over the
relevant time period will be considered part of the same state). In example (1), we
notice that a verb denoting a dynamic process cannot take the simple present to refer
to an ongoing instance of that process (the utterance may be used in certain marked
contexts, such as play-by-play reporting in sportscasts, to refer to present events, but
these are rather marginal), while this is not a problem with verbs denoting stative
processes, as in (2):
(1) ? She kicks the ball.
(2) He loves her.
CG offers an explanation for this in terms of the conflict between a verb’s dynami-
city/boundedness and CG’s definition of the (English) present tense: “PRES indicates
the occurrence of a full instantiation of the profiled process that precisely coincides with
the time of speaking” (Langacker 1991: 250; emphases mine). Canonically, the second
requirement of precise coincidence poses a temporal problem with dynamic verbs in
that the events denoted can rarely be made to align exactly with the time it takes to
describe them. The first requirement of full instantiation causes an additional epi-
stemic problem, since a conceptualizer, given the typical heterogeneity of dynamic
processes,8 needs to have access to at least part of an event before she can start to

(up)’ can be used both in an inchoative–stative and in a dynamic (‘begin to stand (of a toddler)’) way,
depending on the aspectual construction it combines with and other contextual factors like temporal
adverbials: verbs like these “appear to vacillate in their lexical aspect between an inchoative-stative and a
dynamic sense” (Yakpo 2009: 189). Samoan (Sasse 1991) represents the most extreme case in this respect,
not displaying any type of lexical aspect and only working with vague concepts whose verbal status is even
There are also dynamic processes which may be internally homogeneous, such as the homogeneous
activities or ‘episodes’ noted in Michaelis (2005: 63): “Verification of a homogeneous activity, e.g., holding
a broom, standing in a corner or sleeping, requires access to points of inception and termination, as well as
several contiguous frames between those endpoints. Sleeping is distinct both from being comatose and
from nodding off for a second, and staying at one’s sister’s house is distinct both from popping in on one’s
sister and living with her.” These differ from states, which endure in the same way but which do not have
to end before they can be identified: any subinterval of a state counts as an instance of that same state. This
property is referred to in CG as the ‘contractibility’ of states.
An account of English tense and aspect in Cognitive Grammar 219

(a) Past Perfective (b) Present Perfective


t t

(c) Past Imperfective (d) Present Imperfective


t t

Figure. 10.6 Combinations of tense and lexical aspect (Langacker 2001: 262).

describe it. This is obviously not a problem with states in the present: they have a
stable character and are thus highly predictable, and an arbitrary instance of them
can easily be made to co-align with the time of speaking. As a result, when describing
the English simple tenses, one should allow for four distinct possibilities, each time
detailing the interaction between lexical aspect and tense, as in Figure 10.6. In the
past tense paradigm, no problems of alignment with the time of speaking occur, and
thus both dynamic and stative verbs can take simple past marking to refer to a past
instance of an event or state.
The difficulty, diagrammed in Figure 10.6b, of having a perfective process coincide
with the ground, and especially its modal origin, raises questions about whether a
purely temporal account (e.g. of the present tense as indicating present time) will
suffice to explain the intricacies of when this tense may and may not be used, and
what it means when it is used in a large variety of contexts. Moreover, it also forces
us to address the issue of grammatical (next to lexical) aspect, because a definition of
the present tense like the one given above certainly makes it sound as if more could
be said about how strict or tolerant a language can be in allowing less than precise
coincidence between a process and the ground—languages might, for instance,
define the immediate scope imposed by the present tense less rigidly and conse-
quently allow simple present forms of dynamic verbs to refer to ongoing events,
unlike in English. This suggests a basic dichotomy between languages with a simple
present that is grammatically perfective and those whose simple present is ambigu-
ous between a grammatically perfective and imperfective reading, and the test for
finding out which type a particular language belongs to is exactly the behaviour of
the simple present with dynamic verbs. All of this will be the topic of the next
220 Frank Brisard

10.5 Grammatical aspect in CG

Predications of grammatical aspect are not considered true grounding elements, and
understandably so: markers of grammatical aspect like the progressive or the perfect
construction may combine with past and present tenses and therefore do not
constitute the final, criterial step in the assembly of a fully referential representation
of a finite clause. Instead, they provide information about the perspective or view-
point that is adopted on a given process. However, in combination with the English
tenses, such markers of grammatical aspect may be said to form constructions whose
semantics essentially relates to grounding concerns, that is, to indicating the reality
status of a proposition. In other words, viewpoint may be inextricably bound with
every use of a tense form, whether or not that viewpoint will be formally marked. More
particularly, I will say that a tense form is always grammatically perfective and/or9
imperfective in itself (whereby progressive marking is considered a subcategory of
imperfectivity), regardless of whether this is explicitly signalled; in some languages it
may be, in others it may not. But even with those lacking certain explicit markings,
conceptual differences may still be detected between uses of the simple present form
to refer to an ongoing event in the present and those where the internal structure of
the event is simply not attended to: thus, in Dutch the same sentence in the simple
present may, given the appropriate context, be interpreted as indicating an imper-
fective or a perfective construal (cf. below).
Let us now see where this leads us in analysing the main uses of simple tenses in
the domain of the present and that of the past. In De Wit & Brisard (to appear a), it
is suggested, in line with standard CG practice, that a present imperfective construc-
tion (like the English present progressive) should be rendered in terms of two
separate immediate scopes that are imposed. The first, IS1, symbolizes the contribu-
tion of the aspectual marker, whereas the second, IS2, does the actual grounding
work in relating a process to the ground. This is represented in Figure 10.7; the figure
is neutral in terms of the lexical aspect of the process type.
A present progressive zooms in on a designated situation and relegates its
boundaries, if any, to the maximal scope, while attending exclusively to an inner
portion of the process that is conceptually homogenized (i.e. turned into something
state-like) and co-aligned with the ground. Diagrammatically, this is rendered by
positing two immediate scopes which take care of two different jobs. But what if we
posited the relevance of both these scopes in all tense forms, also the simple ones? In
that case, we would have to say that a ‘perfective’ (simple) present tense as in English
comprises two immediate scopes that coincide (with the ground). IS1, the ‘aspectual’

The ‘and’ here captures cases whereby a tense marker is ambiguous between the two readings, such as
the simple present in Dutch and French.
An account of English tense and aspect in Cognitive Grammar 221


Figure. 10.7 Present imperfective.

scope, offers a complete, non-internal view of a process, and the problem in English
is to reconcile the perfective nature of the simple present with dynamic, i.e. non-
contractible (see footnote 8) verbs. Other languages, however, might be more lenient
and could therefore be described in terms of having an ambiguous present tense:
one which requires full and exact coincidence with the ground for stative verbs,
and another (grammatically imperfective) which imperfectivizes dynamic verbs
and makes them alignable with the ground, just like the progressive does in English.
Here, IS2, the tense scope, is included in IS1, the aspectual scope, without coinciding
with it. However, in a language like Dutch this imperfectivization is not made visible
and needs to be assumed in cases like (3):
(3) Ze stampt tegen de bal.
she kicks against the ball.
‘She’s kicking the ball.’
A correlate of this property of Dutch grammar is that there is no heavily gramma-
ticalized progressive construction as in English (where it is virtually obligatory with
dynamic verbs in the present): the one serious candidate, the aan het V-inf con-
struction (ze is aan het spelen ‘she’s playing’), is significantly less frequent than its
English counterpart and optional rather than necessary to convey ongoingness in the
present. On the basis of these (partly semantic, partly formal) considerations, then,
we can posit the existence of present perfective markers of the type represented in
Figure 10.8. In English, this is the only possible interpretation that can be ascribed to
the simple present, whereas in Dutch the simple present form is in principle
ambiguous between a perfective and an imperfective reading.


Figure. 10.8 Present perfective (two scopes).

222 Frank Brisard

(a) IS1=IS2 MS (b) MS


Figure. 10.9 Simple past in English.

Similar observations can be made with respect to the past tense, though of course
the grammatical repercussions will differ due to the absence of alignment problems
in the past. A language generally has both an imperfective and a perfective past
tense, although the two are not always formally distinct: they are in Romance
languages like French and Spanish, which feature separate inflectional paradigms
for marking grammatical aspect in the past (traditionally referred to as ‘imperfect(ive)’
vs. ‘simple’ or ‘indefinite’ past tenses), but in languages which only have periphrastic
means to mark past imperfectivity explicitly, like most Germanic ones, those con-
structions generally occur significantly less frequently and are certainly not obligatory
for the expression of imperfective aspect. As in the present tense paradigm, the degree
of grammaticalization of a corresponding past imperfective construction may serve as
an indication of how ambiguous a simple past form is in this respect. Thus, the
English progressive construction is far less central to the description of the past tense
paradigm than it is in the present tense paradigm, and, unlike with the simple present,
English simple past forms may get imperfective readings on certain occasions (e.g.
with achievement verbs; cf. the example Thick smoke filled the corridor [In a matter of
minutes, we could no longer see the exit signs.] from Michaelis 2004: 53, where fill has
an inchoative interpretation in which smoke diffuses through the corridor; see also
Boogaart 1999: 190, 193). In these contexts, the past progressive may serve as an
acceptable paraphrase for explicating the meaning of the simple past.10 We can
therefore assume, on analogy with our analysis of the present tense paradigm, that
there are two possible configurations corresponding to the use of the English simple
past, both of which are given in Figure 10.9.
The usability of a diagram like Figure 10.9b becomes apparent when tackling the
semantics of formally marked imperfective past tenses in Romance languages, such
as the French imparfait or the Spanish pretérito imperfecto (Doiz Bienzobas 2002).
The specific range of contexts in which these are used (invariably definite pasts)
suggests that there is something about non-present imperfective aspect that makes it

I’m not suggesting here that the French imparfait, as an example of a dedicated marker of past
imperfectivity, and a form like the English past progressive are semantically identical. For one, Molendijk
(2005) has shown that, unlike with the imparfait, the English past progressive cannot be used to suggest
past ongoingness if the event referred to is an instant: quand Pierre entra, 1 heure sonnait vs. ?when Peter
came in, the clock was striking 1 (vs. when Peter came in, the clock was striking 9).
An account of English tense and aspect in Cognitive Grammar 223

t G=IS2


Figure. 10.10 Imperfective past tense (shifted ground; adapted from Brisard 2010: 516).

evoke some type of shifted reference point (other than the actual ground): indeed,
adopting an internal viewpoint on a process that is not actually present seems to
necessitate a mental operation of displacement, so that the conceptualizer finds
herself virtually ‘inside’ the designated past situation (or, conversely, that the past
situation is virtually represented as if it were coinciding with the actual ground).
Both takes on the same configuration are represented in Figure 10.10. The imper-
fectivizing work of IS1, when situated in a non-present context, seems to suggest
the presence of an alternative viewpoint to the present one (the ground), internal
to the non-present situation and virtually represented in the conceptualizer’s present
as G’.
Often, such imperfective past tenses are used to evoke a feeling of vividness in the
description of non-present situations, which is one of the effects of the virtual
construal that sets this tense apart from the more detached narrative mode typically
associated with the perfective past tense. Moreover, imperfective past tenses also
serve in many different modal contexts, expressing hypothetical and other non-real
states of affairs. In those cases, what is missing is a correspondence link between the
virtual representation conjured up in the conceptualizer’s present and some real past
situation, leaving us with something like Figure 10.11. What these uses express is that
a situation is considered real, as with all other uses of a past tense, but this time not
in actual reality (e.g. hypothetically real, as in the protasis of a conditional).
It is no surprise that such modal readings never come about with perfective past
tenses in the languages that have them, since there is no element of virtuality to be
detected in these tenses’ semantic make-up: all they indicate is that a past situation is
viewed, as a whole, from the perspective of the present, without intimating any
process of displacement on the part of the conceptualizer. This is the configuration
represented in Figures 10.6a and 10.9a.
224 Frank Brisard

t G’= IS2


Figure. 10.11 Modal uses of an imperfective past tense.

10.6 Progressive and perfect marking

Having established that all tense forms, including simple ones, have an aspectual
(perfective vs. imperfective) signature, two things remain to be done. One is to
demonstrate how each tense–aspect combination yields various (temporal and
other) meaning types that ultimately go back to a single underlying modal schema,
in line with CG’s basic characterization of grounding predications as epistemic in
nature. Just as temporal and modal uses of imperfective past tenses share a notion of
displaced reality that perfective ones lack, the region of Immediate Reality can be cut
up into different domains. These domains do not differ so much as to the presence
vs. the absence of virtual meaning elements, as to the degree to which a proposition’s
status as real is considered consolidated at the time of speaking. Simple vs. progres-
sive marking in the paradigm of the present in English addresses exactly this issue,
which is the topic of Section 10.6.1.
The other goal, discussed in Section 10.6.2, is to illustrate how relations of scope
within an expression’s semantic configuration are not stable and are subject to
diachronic evolution and synchronic variation. The case of the have perfects in
English and other Germanic languages exemplifies how the same construction can
go from expressing perfect/anterior aspect to indicating past tense, but above all,
how this is a gradual process that depends entirely on altering the relative promin-
ence of one immediate scope over another.

10.6.1 Structural vs. contingent reality in English

A verb’s ability to take the progressive in the present is widely acknowledged
as a diagnostic for its lexically dynamic (i.e. non-stative) nature in English. Thus,
example (4) is perfectly acceptable as a description of a dynamic process that is going
on at the time of speaking, while (5), involving a lexically stative verb, comes across
as a marked way of describing a state holding at the time of speaking, perhaps
An account of English tense and aspect in Cognitive Grammar 225

suggesting something like an extraordinary effort that goes into maintaining the
state at issue:
(4) She’s kicking the ball.
(5) He’s hating her.
These observations, together with the range of special contexts in which English
perfective verbs can take the simple present, have led Brisard (2002) and Langacker
(2011) to reconsider the ‘naive’ characterization of the English present tense in purely
temporal terms (of full and exact coincidence with the ground; cf. above) and add a
modal component to this definition covering all of the meaning types that can be
distinguished for the simple and progressive present forms in English. So let us start
with the simple present: given the absence of an imperfective reading for the English
simple present (in contrast with, say, Dutch or French), the form always indicates a
situation that is fully predictable (i.e. treated as given knowledge) at the time of
speaking. This is not a problem with states, since these are stable and are generally
already holding before the time of speaking. But with dynamic events, the simple
present can only be used if the event in question is in one way or another accepted by
the conceptualizer as a structural or necessary, and thus predictable, part of reality.
And in any case, the simple present can never be used with dynamic verbs to refer to
an actually ongoing event, except in performative contexts (example (6)), repre-
sented in Figure 10.12, where the speech event and the profiled process are one and
the same, as a matter of definition. Here, no epistemic problem arises, since the
conceptualizer has prior knowledge about what is going to happen at the time of
speaking: “Because the speaker is not merely reporting on what happens but is rather
in control of the event described, he does not have to observe the event in order to
identify and describe it” (Langacker 2001: 263):
(6) I promise I’ll come.
In all other cases, the use of the simple present with dynamic verbs implies a
virtual representation of the designated event, which is mentally made to coincide
with the ground even though in reality the overlap with the ground is, in the best of
cases, less than complete. This holds true, for instance, for the play-by-play reporting
mentioned in Section 10.4, a convenient illusion used to create the impression of


Figure. 10.12 Performative (Langacker 2001: 264).

226 Frank Brisard

complete simultaneity, as well as in similar situations where the conceptualizer is in

control of an event and its duration is approximately that of the ground—these include
the concurrent, script-like description of bodily actions, step-by-step demonstrations,
and so on. What these uses share with more radical departures from the description of
an ongoing event (including in recipes, manuals, stage and other directions, photo
captions, futurate uses of the simple present, and historical presents) is that they all
represent the designated event as virtually coinciding with the ground; in other words,
it is the conceptualizer who locates the event in a virtual plane, knowing very well that
the actual event does not coincide with the present (cf. Langacker 2011: 59–64). This is
depicted in Figure 10.13, representing the meaning of a futurate use as in (7):
(7) Your train leaves in five minutes.
The virtual plane evoked may contain representations of events which the speaker
expects will happen, knows to have happened, or otherwise has some form of
epistemic control over. To the speaker, in other words, these events are necessarily
part of reality, and at any time she can choose to evoke them as if they were the case

Event1 Event2 Event3


Events t

Figure. 10.13 Scheduled future (Langacker 2001: 268).


m w

... m w m w m w m w ...

Figure. 10.14 Habitual (Langacker 2011: 65).

An account of English tense and aspect in Cognitive Grammar 227

at that moment; in this sense, their construal corresponds to what Goldsmith and
Woisetschlaeger (1982) have called “structural knowledge types”. Some uses of the
simple present may even go further and conjure up a virtual representation that does
not stand for one, but for many instances of the same event type. In those cases
(habitual and generic uses of the simple present, as in example (8)), the virtual plane is
to be interpreted as representing structural aspects of reality, and as purely mental
entities they can be manipulated at will by the conceptualizer and thus are not subject
to the alignment problems that occur with representations of actually ongoing events.
This situation is illustrated in Figure 10.14, which profiles a virtual instance of a
relation (of straight drinking) between mother (m) and whisky (w) at a structural level:
(8) My mother drinks her whisky straight.
We can conclude that the English simple present is always used to express a
structural aspect of reality: with states, their very stability ensures their categoriza-
tion as structural (at least within the scope of Immediate Reality), and with dynamic
events, only those events are referred to that are felt by the speaker at the time of
speaking to be an accepted part of her reality, and not just an incidental description
of it. This is the result of the strictly perfective character of this tense in English.
If the meaning of the English simple present may be rendered in terms of
structural necessity, then we might expect the present progressive to express the
opposite, namely contingency or the absence of necessity. Remember that the present
progressive is still part of the present tense paradigm and refers to situations located
in Immediate Reality. But this epistemic status has not been consolidated at the time
of speaking; it remains “phenomenal” rather than structural, in the words of
Goldsmith and Woisetschlaeger (1982). Once again, this modal characterization
results from the temporal (i.e. imperfective) profile that the progressive construction
imposes (see Figure 10.7); we might add that, in contrast with other possible
imperfective construals, the progressive specifically imposes boundaries on the
designated event in MS, as given in Figure 10.15. This could explain, for instance,
the effect this construction has on lexically stative verbs, which are coerced into a
more dynamic reading and imply the temporary (extraordinary, etc.) nature of the
designated situation (cf. example (5), He’s hating her).

Present Progressive

Figure. 10.15 Present progressive (Langacker 2001: 261).

228 Frank Brisard

At the same time, a modal description of the semantics of the present progressive
is (more) basic (than a temporal one) in that it is relevant for all of the many
meaning types that can be distinguished for this construction in English. De Wit and
Brisard (to appear a) therefore introduce a semantic network that starts from a
common schema of “Contingency in Immediate Reality” to develop into a number
of more or less connected temporal and modal meaning types, the latter including
uses of the present progressive to express surprise, atypicality, and so on. Some of
these uses also involve virtual constructs, but above all the network shows how the
basic semantic configuration for expressing current ongoingness (the least marked
or default use of the present progressive) may be exploited to produce other meaning
types differing only in the slightest detail, such as the relative salience of one or
more of the configuration’s components. For instance, from Figure 10.15 it is easy
to go to Figure 10.16, representing the meaning of temporary validity for which the
present progressive can sometimes be used. An example is given in (9):
(9) He’s having a hell of a time.
With these uses, the fact that the designated event has boundaries (located in MS)
is being emphasized, rather than downplayed as with other usage types. This results
in a higher degree of salience ascribed to the process’s boundaries, without these
boundaries actually becoming part of the expression’s profile; let’s just say that their
salience is higher than what would be expected from elements in an expression’s
maximal scope. In cases where the element of duration, not boundedness, is
exploited, the opposite happens. Starting from the same Figure 10.15, we now end
up with a configuration in which the boundaries of the process are (even more)
downplayed, symbolizing the speaker’s judgement in such cases that these bound-
aries are semantically irrelevant. This is what happens for utterances like (10),
diagrammed in Figure 10.17:
(10) He’s still making noises with the paper.
Many other modifications are possible, of course, including virtualizing uses (for
habitual imperfectives; a modification of Figure 10.14) as well as manipulations of the


Figure. 10.16 Temporary-validity reading of present progressive.

An account of English tense and aspect in Cognitive Grammar 229


Figure. 10.17 Durational/‘continuative’ reading of present progressive.

temporal position, rather than the salience, of (one of) the temporal boundaries: in
inchoative uses, for instance, the left boundary is explicitly made to coincide with IS1.
The major advantage of the network description referred to here is that it unifies
all the different usage types, including ones that have pretty much escaped previous
attention and that would in any case be hard to characterize in purely temporal
terms, since their use does not seem temporally motivated (in other words, the
reason for using a progressive rather than a simple form here does not primarily
seem to involve the expression of ongoingness or duration). That is why, in De
Wit and Brisard (to appear a), they are called modal usage types. Here are some
(11) Well, I’m telling you, withholding goodies works.
(12) I’m not dancing with guys!
(13) He’s smart, he talks about minorities. But he’s really talking about African
None of the observations made in this section may simply be transferred to
(i) other uses of simple vs. progressive forms in non-present paradigms (e.g. the
past) or (ii) uses of simple and progressive present tenses in other languages. Modal
usage types do occur in other languages as well, such as Dutch and French, even
though the progressive is not so grammaticalized in these languages (see De Wit and
Patard, in press). In fact, they were already present in earlier stages of English, too
(notably, from Middle English onwards). This shows that the rise of modal usage
types does not entirely go hand in hand with increased grammaticalization.11 Still,
the peculiar properties of the English present tense paradigm are deemed to be a
consequence of the strictly perfective character of the simple present and the highly

Modal interpretations of the progressive construction appear to be part and parcel of its semantics.
Though progressives represent a type of imperfectivizing construction, this does not mean that these
modal meanings are necessarily shared by other imperfective constructions. What is special about
progressives in contrast with other imperfectives is the former’s systematic association with dynamic
verbs, suggesting the presence of boundaries in the maximal scope (cf. Figure 10.15), and it is this property
which may be deemed responsible for the modal usage types.
230 Frank Brisard

grammaticalized status of the present progressive construction. On the other hand,

the description of the English present tense paradigm proposed here does make use
of principles of cognitive semantics that are universal, and, as a model, it can easily
be applied to tense paradigms in languages with fairly diverging semantic properties
(see, e.g., Brisard and Meeuwis 2009; De Wit and Brisard, to appear b).

10.6.2 From perfect aspect to past tense

To end this section, let us now consider how the current framework facilitates the
description of semantic variation (in time, or over different varieties of a language)
for the same construction. Given that semantic configurations in CG are made up of
some essentially gradient notions (scope, salience), the use of a particular construc-
tion may evolve to exploit one or the other meaning element, so that it gradually
develops various related meaning types and may eventually even move from one
grammatical category to another. To describe this evolution, and some of the in-
between configurations that result from it, we will have a look at the perfect
construction formed with auxiliary have, which is prevalent in many European
languages but may differ significantly in individual languages in terms of the
contexts in which it can be used.
I will base this discussion, which is necessarily brief, partly on the semantic
classification of have perfects proposed by McCawley (1971). We begin by consider-
ing the use that is closest to the original construction in Old English, where it
denoted a present state that is the result of a prior action. In example (14), a
resultative use, the present perfect also serves to focus on the result of a previous
(14) Don’t come near me, I’ve caught the flu.
Having the flu is the result of a previous event of catching it, but this event is
relegated to the background, just like the boundaries of a dynamic process marked
by the progressive. The reference of the past participle is to the state resulting from



Figure. 10.18 Present perfect: resultative use.

An account of English tense and aspect in Cognitive Grammar 231

the past event situated in MS. In line with our previous characterization of the
English simple present (Figure 10.8), which is the tense of the auxiliary, IS1 and IS2
coincide, reflecting the idea that a perfective perspective is adopted on a present
state. (The scope of the participle per se, which would be IS3, is not indicated in the
diagram, but it would coincide with the entire state resulting from a completed
process.) However, one might argue that this configuration is already one step
removed from its Old English ancestor, in that there is no literal state of ‘having’
in the sense of possession that is actually profiled. Moreover, the profile, in subse-
quent uses of the perfect, seems to shift gradually away from the present end-state to
the prior event itself. Several factors are involved here.
Prototypical resultative uses—and other uses, for example in Irish English, that
are retentions of the older syntactic pattern, as in she’s nearly her course finished
(Harris 1984: 307)—involve a present state that can still be seen as profiled against
the background of an immediately prior, typically dynamic and transitive event. It
usually denotes an objective transfer of energy between two participants, trajector
and landmark, resulting in a salient end-state that one of the participants (the
landmark) finds itself in. The focus lies on that end-state, which holds in the present
and thus justifies the present tense form of have. Changes in this original configur-
ation will be situated at the level of the subjective construal of objective temporal
configurations involving a past event, and they will be shown to relate primarily to
two cognitive parameters, namely perspective and scope (or the relative prominence
of one situation over another in a complex scene). What seems to happen in general
is that the perfectivity of the original construction forces the tense to shift so that the
construction as a whole behaves as a perfective past tense (rather than a perfective
present tense þ perfect aspect), even though formally of course the tense of the
auxiliary remains present. Thus, in subsequent diagrams IS1, the perfective aspectual
scope, becomes dominant, while IS2, the present tense scope, is the recessive one and
will ultimately disappear to make room for a past IS2 that is associated with the same
perfective IS1.12
The following three usage types (examples (15)–(17)) manifest a weakening of the
objective features just described for the resultative uses or, in other words, an
increasing subjectification of the configuration. Historically, they illustrate the pro-
gressive loss of objective factors motivating the association between a past event and
a resultant state in the present:
(15) Osama Bin Laden has just been assassinated. (hot news)
(16) I’ve known Max since 1969. (universal)
(17) I’ve read ‘Principia Mathematica’ five times. (existential (‘indefinite anterior’))

The terms ‘dominant’ and ‘recessive’ are borrowed from Malchukov (2009), though he talks about
grammemes rather than constructions.
232 Frank Brisard



Figure. 10.19 Present perfect: extended now.

If we want to move away from the configuration in Figure 10.18, then we must first
relax the condition of temporal contiguity between the past event and the present
state. This creates a type of ‘extended now’ that will remain relevant for the
characterization of most uses of the perfect in English.
In Figure 10.19, the conceptualizer’s attention is distributed between, on the one
hand, an end-state that one participant is in and that is simultaneous with the position
at the time of speaking of the other participant (i.e. the original configuration: an agent
that ‘has’ or, more abstractly, relates to something in a state as the result of the agent’s
action), and on the other the actual past action itself, which becomes more profiled in
the entire expression and loses the requirement of direct contact with the present (i.e.
can be located in a more distant past and thus acquires some conceptual autonomy of
its own, independently of the end-state that it leads to). One could also say that the
tense profile, defined by IS2, recedes or loses in salience and that the prominence of the
aspectual IS1 gradually grows in dominance, creating a more balanced distribution of
prominence over both ISs (a formalization of the feeling of an extended now) and
thereby increasingly problematizing the categorization of the construction as a mere
aspectual variant of the present tense. In other words, one could say that the
construction becomes more and more perfective, and less and less present. In more
evolved cases, the requirement that the preceding situation needs to be a dynamic/
transitive event is lost altogether (generalization), as seen in Figure 10.20.
At the same time, the ‘presentness’ of the resultant state, which is still part of
the background conception (in its maximal scope), is more and more a matter
of the speaker’s subjective appreciation, rather than of a participant’s objective
condition, until we reach the point where the mere “current relevance” (Comrie
1976) of a past situation is left as the sole relic of the construction’s initial end-
state focus. This might be captured by the idea of a present perspective, already
immanent in the conception of a participant’s current end-state (e.g. in resulta-
tive uses), being gradually more attributable to the speaker. That is to say that
the speaker, in using perfects that do not strictly speaking refer to a present
state, portrays a past situation from her own temporal viewpoint, and that the
An account of English tense and aspect in Cognitive Grammar 233



Figure. 10.20 Present perfect: current relevance.

role of the non-subject participant, if any, is restricted to the (profiled) past action
proper; this fits the reanalysis responsible for the auxiliary status of have and the
promotion of the participial form to ‘head of VP’ in the grammaticalization process
of the perfect construction. Figure 10.20 is, obviously, the same configuration as in
Figure 10.9a, representing a ‘perfective past’ tense.
In its most extreme form, as in existential or indefinite anterior uses, the perfect
might best be described as having the value of a perfective past, ending up as a marker
of tense rather than aspect (cf. the passé simple and pretérito indefinido/perfecto
in Romance languages). The motivation for keeping the perspective in the present
even in such perfective past tense contexts is the obligatory use of what I would call
‘ground-including’ time adverbials (like ever or already) or constructions (like since
and for constructions) with the present perfect, rather than real past-time adverbials.
More conservative dialects, such as Hiberno-English, Standard American English, and
Black English Vernacular, represent intermediate stages of this evolution and have not
(yet) gone all the way. For instance, they usually do not feature the present perfect in
these indefinite contexts, for which the simple past is used.
Other Germanic languages, finally, have moved beyond this point and have
started to use the have construction in contexts where, at least from the perspective
of English, one would expect a true past tense. This is generally referred to as
‘preterite loss’, and it occurs in contexts where a definite point in the past is explicitly
or implicitly given. In such languages as Dutch, German, and Afrikaans, an utter-
ance like (18) is entirely grammatical:
(18) Ik heb gisteren spaghetti gegeten. (Dutch)
I have yesterday spaghetti eaten.
‘Yesterday I ate spaghetti.’
In French, too, this is possible; the difference with an utterance containing an
imparfait form would be that the latter implied more of a feeling of ongoingness,
typical of narrative background uses of this tense:
234 Frank Brisard

(19) Hier j’ai mangé des spaghettis.

yesterday I-have eaten spaghettis
‘Yesterday I ate spaghetti.’
(20) Hier je mangeais (quand Pierre est entré/entra).
yesterday I was-eating when Peter has entered/entered
‘Yesterday I was eating when Peter came in.’
The configuration for these uses is given in Figure 10.10. Here, the have construction
has moved away entirely from its aspectual origins and has come to cover (both
perfective and) imperfective readings in the past traditionally associated with
real past tense forms. In Afrikaans, for instance, the het ge-V construction can
even be used to express modal meanings (e.g. in optative contexts) which, in English,
Dutch, or German still call for the morphological past tense.

10.7 Conclusion
The tools of CG, including the grounding model and universal cognitive-semantic
principles such as predicational scope (as a function of prominence of attention on
the part of a conceptualizer), allow the systematic description of tense–aspect
constructions in a language’s verbal paradigm, stressing the common modal mean-
ing elements in individual constructions’ usage types and thus unifying the analysis
in a domain well known for its considerable constructional polysemy. The assump-
tion of temporal depth assigned to the time of speaking has a number of important
grammatical consequences in the paradigm of the present, which English deals with
differently from other related languages. The durational and epistemic problems
arising from this feature have prompted the postulation of virtual entities that are
relevant to the characterization of certain configurations (i.e. those in which no
reference is being made to an actually ongoing event). I have also described the
peculiar properties of present tense marking on lexically dynamic verbs as a conse-
quence of the strictly (grammatically) perfective nature of the simple present in
English, on the further assumption that all tense forms, including simple ones, are
semantically specified for grammatical (im)perfectivity; simple tense forms can then
be considered ambiguous or not in this respect, depending on the language one is
looking at. The combination of tense and aspect in a single verb form necessitates
positing two immediate scopes in all instances, even if these may coincide (as with
the English simple present or a perfective past tense).
The present progressive in English, as an imperfectivizing construction involving
two clearly distinct scopes, is a good illustration of how a basic semantic configur-
ation can be tweaked to produce various usage types sometimes only subtly differing
in meaning. The cognitive mechanisms implicated relate to notions like bounded-
ness and its relative prominence, or to the exploitation of a configuration’s modal
An account of English tense and aspect in Cognitive Grammar 235

signature (to the detriment of temporal motivations for using a progressive form).
The case of the English present perfect, finally, has been presented to exemplify
the possible evolution of an originally aspectual construction in the direction of a
pure tense marker. Crucial in this evolution is the rising prominence of one
immediate scope over another, coupled with an increasingly subjectified construal
of the original configuration (involving a past event and a present end-state).

Frames of reference and the linguistic

conceptualization of time: present
and future


11.1 Introduction: times, tenses, and frames of reference

Reference frames are fundamental to our understanding of time and space, whether
in the domain of physics or of ordinary human communication. In physics, the
principle of relativity is built on the theoretical notion of frames of reference:
space–time is relative to the observer’s frame of reference. In ordinary human
linguistic communication, which draws on the non-linguistic human cognition of
space–time, phenomena that are variously called frames of reference, viewpoint, and
perspective are ubiquitous. The relationship between the physicist’s frame of refer-
ence and frames of reference in the human mind–brain is a matter for speculation. It
is possible that we can speak of a supervenience relation (as Jaszczolt 2009: 14 does in
relation to other aspects of the problem of the representation of time; see also
Jaszczolt, this volume). It is possible that the kinds of cognition that are recruited
by languages for communication ultimately depend on physical space–time, but are
not reducible to it. However, it may well be that we can use a similar or related
theoretical framework for understanding both realms. That is why this chapter
approaches the linguistic conceptualization of time—the kind of conceptualization
connected with tense markers in languages—by drawing on elementary geometrical
notions, the most general of which is frames of reference or coordinate systems.
Of course, to describe linguistic conceptualizations, the geometrical models have to
be of a relevant special kind. The starting point is to treat tense marking as essentially
deictic and then to model deixis geometrically. Deixis is understood not only as spatial
and temporal but also as modal (cf. Frawley 1992; Chilton 2005). This approach is not
incompatible with Reichenbach’s foundational approach, which remains influential
Frames of reference and the linguistic conceptualization of time 237

in some of the recent advanced models of time and tense (e.g. de Saussure 2003, and
this volume; Evans 2004; Jaszczolt 2009).
The problem for linguists in accounting for tense has been to explain, or to model,
how it is possible to speak of, for example, a present tense marker when such forms
frequently appear to have interpretations that are not to do with the present time of the
speaker. There is no doubt that there is interplay between conventionalized ‘schematic’
or ‘default’ conceptualizations associated with tense markers. In this chapter I will
attempt to show how we can firstly model what I shall refer to as ‘schematic’ tense
conceptualizations in terms of relative positioning in frames of reference, and secondly,
crucially, model the variation in time interpretations of tenses in terms of operations on
the basic schematic meanings—specifically, embedding of frames of reference one
inside another and the transformation of frames of reference.
A further kind of conceptual interplay that has been of great interest to seman-
ticists is that between time and modality (cf. Kaufmann, Condoravdi, and Harizanov.
2006: 90–102). It has been noted many times that while the past is ‘settled’, the future
is not, given some point in time at which such judgements are made. In addition,
times, whether past or future, can seem closer or more distant in time, and distance
has a cognitive connection with epistemic certainty. The framework of Deictic Space
Theory (DST), which I summarize in Section 11.2 below, inherently reflects this idea.
This chapter will thus focus on two classic areas of concern: how linguistically
marked tense is related to conceptual time and whether linguistically marked tense
relates to modality. The following analyses, like those of a number of other scholars,
build on Reichenbach’s (1947: 287–98) framework and uses insights from Langacker
(1991, 2011). Description of English tenses requires time points: the point of speech,
event point, and reference point (Reichenbach 1947: 287–8), with reference point often
being defined in the discourse or pragmatically rather than explicitly in the sentence.
The English tenses appear to correspond to different alignments of the three points. In
DST the speech point is intrinsic in the deictic space. It follows Reichenbach in
treating the three points in terms of relationships and ordering on the timeline.
However, because the deictic space includes direction as well as distance, DST says
more about the nature of the relationships, and in particular it is able to incorporate
‘viewpoint’ (cf. Langacker 1995; Michaelis 1998, 2006; amongst others). The most
important feature of DST is indeed the notion of ‘frames of reference’, treated as copies
of the base coordinate system that can be nested within that base system, which is
termed R, the ‘reality’ conceptualized by the speaker (self, subject) S. These copies are
‘virtual realities’ that can be anchored at different deictic points in S’s R. It will be argued
that this approach provides a new and adequate way of modelling the complexities of
conceptualizations that are associated with the English present tense.1

The tenses of other languages are equally complex, of course. But each language varies in respect of
tense semantics and deserves individual treatment, as well as cross-linguistic typological treatment.
238 Paul Chilton

There are thus two conceptual operations that will be explained below:
(i) operations linked to the conceptualizations induced by English simple present
and progressive, and (ii) temporal frame-shifting. These require no ad hoc machin-
ery if the fundamental geometrical architecture is adopted: they are given in very
simple spatial geometry and may indeed have their origins in spatial cognition. The
emphasis is thus placed on conceptual operations, which we may call ‘perspectivis-
ing’. Thus, it is not that the English present tense constructions are polysemous:
rather, they are associated with core schemas that can be cognitively manipulated in
combination with the conceptual structures associated with the multiplicity of event
types that the lexicon conventionalizes.2

11.2 DST in brief

11.2.1 Three dimensions and an integrated abstract space
The deictic space theory postulates a conceptual space consisting of three dimen-
sions that are taken to be specific to the kinds of conceptualization drawn on and
induced by linguistic forms. While the morphological and syntactic means of
languages vary, the hypothesis is that all languages will use the same conceptual
space. This is a deictic space in the sense that its three dimensions are relative to S
(for speaker, subject, or self ) and it has the configuration shown in Figure 11.1. The
particular geometry of this space is explained and justified in more detail in Chilton
(2005, 2007, 2009) and the following is a necessarily curtailed summary.
The three dimensions of Figure 11.1 are not to be confused with the three spatial
dimensions of perceived physical space. Rather, the idea is to represent a conceptual
space representing an integration of discourse referents (the d-axis), conceived time
(t-axis), and epistemic modality (m-axis). The basic conceptual components are relative
distance and direction. The three axes are not the usual ‘x, y, z’ axes of spatial geometry
but abstract dimensions: d is attentional distance of discourse referents from S (fore-
grounding, middle ground, background), t is deictic time, and m is epistemic distance
from S. ‘Distance’ here is relative cognitive distance, not a metric. Linguistic construc-
tions give rise to conceptual representations, the fundamental structure of which is the
deictic space. Diagrams like Figure 11.1 are intended to sketch a conceptual space,
centred on S, whose zero point is the conscious now-here-real, which can be located at
any point in real space–time, and perhaps in any possible world, should we wish to do
so from a universal vantage point. This is not to say there is no connection between real
space–time and the cognitive deictic space, but it is not the concern of this chapter to
explore that deep issue. Figure 11.1 is the template that will be used for particular
discourse space models (DSMs) of tense and time in the remainder of this chapter.

For a Cognitive Grammar account of the present in English see also Brisard (this volume).
Frames of reference and the linguistic conceptualization of time 239



–t proximal

pluperfect peripersonal
had been simple region
past pres
was perf S
has now periph-
been rastic future
future will be +t
is future
going perfect
to be will
Figure 11.1 The fundamental deictic space.

The relative distances on each axis are motivated by the kind of relative concep-
tual ‘distancing’ communicated in demonstrative expressions (cf. Fillmore 1982;
Diessel 1999, 2006). There are three relative positions: proximal, medial, and distal,
in addition to the zero point (i.e. deictic and geometric origin of the three dimen-
sions). However, the linguistic and psychological evidence suggests that proximal is
not simply a point but a region, and a region of a special kind, rather unfamiliar in
the linguistics literature, known as peripersonal space. This is the psychological space
that is defined in part by the reach of an organism’s limbs and which includes
personal space, the space occupied by the body.3 We shall speculate, for the purpose
of investigating tense conceptualizations, that peripersonal space is transposed on to
the time axis: peripersonal time. It is important to note that psychological periper-

Peripersonal space is widely studied in psychology and cognitive science. Of particular relevance for
present purposes are: Gallese (2003); Weiss et al. (2000); Berti and Rizzolatti (2002); and on their relevance
for spatial language, Kemmerer (1999, 2010) and Coventry and Garrod (2004).
240 Paul Chilton

sonal space can be extended (e.g. by the use of tools, which extend personal reach),
and I speculate that in linguistic communication this effect may also arise by way of
linguistic and pragmatic cues. If it is reasonable to speak of peripersonal time,
I suggest an analogous effect may help understand some of the conceptual effects
of tense semantics. What would a peripersonal temporal space be like? A reasonable
guess is that it consists of memory of the recent past and anticipation and planning
for the immediate future: memories of actions just performed and experienced
simultaneously with intentions for one’s next actions.
Just as DST proposes that peripersonal space projects on to the time axis, so it also
proposes that it projects on to the m-axis. What is within our peripersonal space is
not only spatially close, it is epistemically ‘close’—it is what we can ‘grasp’ in both
senses of the term, and what we can grasp is epistemically most certain. The spatial
foundation of the three dimensions of DST thus gives us an integrated correlation of
the attentional, the temporal, and the epistemic.
Jaszczolt (2009) argues that tense forms in languages are in effect modal. It seems
intuitively plausible to think that future events are, to the human mind, inherently
less certain than present ones. However, DST does not attempt to build this idea
directly into its architecture. If we were to revise the DST geometry in the light of the
claim that time equals modality, one axis would disappear. Another possibility
would be to design a geometry that would make the t-axis not orthogonal but a
curve such that t þ curved asymptotically towards distal counterfactuality on m.
And since it might well be argued that the past too becomes epistemically less certain
the more distant it is, the same could be the case for t  . Such a geometry might
capture the conceptual structure of the human mind but for present purposes it
would immensely complicate the theory. I shall simply assume that the relative
distances on t þ and t  may indeed correspond to possible degrees of epistemic
certainty. Another reason for not revising the fundamental geometry in this way is
the following. We are concerned with conceptual representations brought on by
linguistic constructions, and the fact remains that languages generally combine the
indication of time location relative to ‘now’ with modal representation of degrees of
certainty. Thus we can say, referring to a future event, Mary will write the report next
week, Mary might write the report next week, it is probable Mary will write the report
next week, etc. We need to be able to model such combinations and this can be done
in the three-dimensional deictic space of DST, even if we were to make the t-axis
curved. In Jaszczolt’s model (Jaszczolt 2009: 140–54) there is a cline of epistemic
certainty from most to least certain tense forms: tenseless future (Peter goes to
London tomorrow), futurative progressive (Peter is going to London tomorrow),
periphrastic future (Peter is going to go to London tomorrow), regular future (Peter
will go to London tomorrow), epistemic necessity future (Peter must be going to
London tomorrow), epistemic possibility future (Peter might be going to London
tomorrow). Apart from the fact that the intuitive semantic judgements involved here
Frames of reference and the linguistic conceptualization of time 241

are debatable, it seems to me that placing all these expressions on a single scale loses
distinctions with regard to the futurative use of epistemic modals.
The intersection of the axes, the origin, defines the viewpoint of the speaker S. On
each of the three axes we postulate components corresponding to peripersonal space,
and define them as unit vectors. The coordinate system described in this way
corresponds to the speaker’s self, in the sense that it corresponds to the speaker’s
cognizance of what is here (the graspable in primary peripersonal space defined
physically), what is now (what is temporally within reach, that is, peripersonal space
projected onto time), and what is real (what can be ‘grasped’ cognitively and in some
possibly non-linear fashion correlating with spatial and temporal distance). What
this definition of a proximal spatiotemporal–modal space encapsulates is the intu-
ition, reflected in language and discourse structure, that things are more likely to be
real for the speaker in the here and now. What is distal, either perceptually or
temporally, is less actual.

11.2.2 Frames of reference and geometrical transformations

The abstract meaning space that has just been outlined is the frame of reference R,
the assumed ‘reality’ of the self–speaker–subject S, as already noted. DST aims to
make explicit some fundamental elements of linguistic meaning by using this space
to describe the conceptual effects of lexical and grammatical structures. The
discourse referents that constitute arguments in predications are represented
by ‘labels’ that have coordinates on the d-axis and relations between them—
predicates—are represented as unit vectors. It is unnecessary to go into further
detail on this particular aspect (see Chilton 2007) but what is important is a further
piece of standard apparatus from simple geometry—transformations. Geometric-
ally, transformations move the base axis system around, invert it, reflect it, and so
on. In DST we use this geometrical idea of transformation to deal with linguistic
conceptualizations, specifically temporal ones, that are equivalent to shifts of
reference frame.
This is important cognitively. We can think of these reference frames as
virtual realities (a similar notion has been found necessary by Langacker 1999,
2001, Brisard 2002, and other cognitive grammarians, who speak of ‘virtual
planes’), generally under the inspiration of Fauconnier’s ‘mental spaces’ (Fauconnier
1994). These virtual reference frames are ubiquitous in the conceptualizations
involved in language and can be located at different points in the basic R space.
They are of great relevance in helping to understand the complexities of the
meanings that are associated with present-tense forms. They are not invented
ad hoc, just to explain tense meanings, but are already implicit in the geometry
of DST.
242 Paul Chilton

It might be objected that this type of model implies the following problem. It may
appear that the three-dimensional space is conceptually prior to the insertion of
discourse referents and the relations between them—that the human mind has a
conceptual space of the kind shown in Figure 11.1, a kind of tabula rasa on which
referents and events are drawn. This may be the way we have to proceed in
theoretical exposition but it is not intended to claim that the three-dimensional
space represents something innate. Whatever hard-wired systems may be involved
in our conceptualization of time, it is likely that such conceptualization also involves
the experience of events in their varying duration and also the communicative
conceptualization of events as referents of different types and varying ‘distances’
from the utterance moment. I am suggesting there is a hard-wired component and
also a culturally developed one that integrates the needs of communication—thus
that t in the basic DST coordinate system represents a conceptualization of some-
thing that has emerged from cognitive–communicative evolution in which events of
varying duration are reported to conspecifics. As for the cognitive systems involved,
memory provides evidence for, and is the basis of our conceptualization of, the past.
Future conceptualizations presumably emerge from the ability of humans to plan
ahead, an adaptive cooperative social mechanism.
What I am putting forward is not rule-based (DRT, SDRT) or procedural (Jasz-
czolt 2009; de Saussure 2003). DST does not aim to model mental processing of
linguistic input with or without pragmatic contextual input. The aim is (i) to propose
models of the schematic default semantics associated with tense marking and (ii) to
model the cognitive operations that can switch the default time interpretations,
given pragmatic input. How the pragmatic input is processed in detail is not my
concern here. The former (i.e. (i) ) are models of cognitive material held in long-
term memory and resulting from conventional meaning–form pairings that develop
in and are acquired in a socio-cultural setting. The latter (i.e. (ii) ) arise in the same
way and are learned in the same way but are not involved in communicative
cognition unless pragmatic considerations require them in order to make sense of
the default semantic schemas. Such pragmatic material and the principles governing
them may be of the kind outlined by Jaszczolt (2009) or by de Saussure (2003) and
plausibly involve the cognitive relevance principle (Sperber and Wilson 1995).

11.2.3 Now
A conceptualization of now emerges from the three-dimensional abstract deictic
space, which I have described above. Geometrically the point (0,0,0), the origin,
defines now for S. What does this mean in conceptual terms? For the moment I will
not consider the notion of speech time, which is important for Cognitive Grammar
accounts of the present, and will focus on conceptualization as distinct from the
physical act of speaking. Thus (0,0,0) is a point in consciousness, but not only a
Frames of reference and the linguistic conceptualization of time 243

temporal one. It integrates consciousness of the present, maximal epistemic cer-

tainty, and maximal attention. This is perhaps equivalent to what is called ‘imme-
diacy’ by cognitive grammarians.
In using the deictic space to model a given sentence, the DSM will place labels for
discourse entities at points on d greater than 0, and the vector relating them may
appear at various points on m, depending on how the sentence modalizes the verb.
Wherever such a vector is placed on m, we are concerned here with the case where
its t coordinate is t ¼ 0. This would be the case for the DSM of a sentence such as
Carbon emissions cause climate change. The crucial point is that in the DST model,
this type of expression has no dimension on the t-axis, i.e. no temporal duration.
This is a useful feature and corresponds to what Dummett, following Frege, calls the
“tense of timelessness” (Dummett 2006: 108). Frege in fact thought of the simple
present tense in terms of timelessness and time-setting:
The present tense is used in two ways: first, in order to give a date, second, in order to
eliminate any temporal restriction where timelessness or eternity is part of the thought.
(Frege 1956 [1915–1919]: 296)

What Frege means by ‘giving a date’ is the use of the present to point to the time
of utterance. In DST terms this amounts to locating R on the objective arrow of time,
or perhaps rather the calendric timeline in a relevant cultural context. This is not our
concern here. The important point is that treating the present tense as expressing
temporal presentness (as opposed to temporal distance) is integral with epistemic
certainty. And epistemic certainty adheres both in perceptual (therefore temporal)
immediacy and in the conceptual transcendence of temporal boundedness. This
characterization emerges naturally from DST’s geometric approach; it also seems
phenomenologically natural. This way of understanding the meaning of what it is
that the simple present tense corresponds to is essentially similar to revised ideas
about the English present tense found in Cognitive Grammar (Brisard 2002: 262ff.
and this volume; Langacker 1995, 2001).
The zero-point concept of time does not, however, correspond to all the concep-
tualizations that are set off in the mind when the English present tenses (whether
simple present or progressive) are used. As has been seen, the DST space includes a
temporal peripersonal region, which ‘surrounds’ the point (0,0,0). Like peripersonal
space, I take peripersonal time to have a basis in neurological and physical processes,
and to have a basic minimal span. However, both spatial and temporal spheres
around the self can be extended depending on context and activity. The peripersonal
time span consists of points, of which one is the origin, the zero point; the
peripersonal time span contains the now instant located at the origin. These two
notions of now are thus not separate and it is not surprising that both are needed to
give an account of the conceptualizations prompted by the present tense forms.
244 Paul Chilton

11.3 Aktionsart schemas, tense forms, and cognitive operators

The linguistics literature has for a long time been concerned with the differences
between stative and non-stative verbs. The existence of two different grammatical
forms in English, simple present and progressive present, has been predominantly
treated as a way of detecting a verb’s membership of the stative or the non-stative
category. More recently, in Cognitive Grammar, the tendency has been to treat the
two tense forms as also having semantic, or rather conceptual, content. And this is
the perspective adopted here. Simple present and progressive forms are conceptual
phenomena: they are linguistic phenomena that reveal underlying cognitive phe-
nomena. The two linguistic forms (simple present and progressive) have evolved, by
cultural convention, to communicate two ways of conceptualizing events.
Events are conceptual constructs arising in human minds from the point of view
of human inexperience of the physical world. Languages select and stabilize con-
ventional conceptualizations in conceptual frames associated with particular lexical
predicates. So we have a set of lexicalized conceptualizations of event types (Aktion-
sarten) on the one hand, and on the other, two key cognitive operations that can be
applied to them. To some of these event types the conceptualization associated with
present tense applies more readily than that associated with the progressive. This is
what gives rise to the categorization by analysts into stative (Langacker’s imperfective)
and non-stative (or dynamic, or Langacker’s perfective) categories. But as is well
known, there are exceptional uses: for example Mary believes in ghosts, Mary is
believing in ghosts at the moment. It is therefore preferable to speak of a particular
verb as being prototypically stative or non-stative and to allow for either simple
present or progressive forms to potentially operate on all event types.
Many attempts have been made to characterize the conceptualizations associated
by linguistic convention with simple present and progressive forms in English (and
other languages, although there are important differences cross-linguistically). Such
characterizations have been attempted in the tradition of truth-conditional seman-
tics, but I shall not consider them here. They have also been carried out in Cognitive
Grammar in much descriptive detail, partly by means of pictorial diagrams and
partly through meta-description in ordinary English (Langacker, e.g. 1987, 1991, and
Brisard 2002 are important examples). The approach I am putting forward is
different and doubtless controversial. As we have seen, the DST model is stated in
terms of conventional elementary geometry. I want to see whether and how much
this geometric modelling can tell us about, and make more precise, the semantics of
the English simple present and the English progressive forms. In the next sections,
I adopt an approach that combines geometric modelling with a cognitive interpret-
ation. In Chilton (2007) I outlined a geometric way of modelling the conceptual
schemas that accompany different eventuality types (Aktionsarten), statives and
Frames of reference and the linguistic conceptualization of time 245

processes (including their subtypes activities, accomplishments, and achievements).

The cognitive operators, instancing (induced by simple present) and windowing
(induced by the progressive), act on these schemas and are induced by the linguistic
tense forms.
The geometric frame for stative event types provides a basis for setting up a model
of instancing operators. In outline, DST models stative predicates as sets of parallel
vectors filling a space, any one of which is linearly dependent on the others. This
corresponds to the widely held view of state predicates, namely that they are
homogeneous over time and that any subpart of a state is equivalent to any other.
The instancing operator is a cognitive operation that picks one vector as an instance,
or one instant, from the set at S’s now, i.e. 0 in terms of the geometry of Figure 11.1.
The important implication of this kind of modelling is that such an instancing vector
has no time dimension, considered geometrically. Such representations occur when
certain grammatical constructions view events as completed timeless wholes, when
properties are treated as timeless and essential (Earth revolves around the sun),
and also when activities are represented as permanent properties of individuals, as
in habituals.
Non-stative Aktionsart schema predicates are modelled in DST as predicates that
progress over time. The tail of the vector has coordinates in the deictic space: the tail
of the vector arrow has a coordinate on the d-axis for (say) the agent of a predicate,
and a coordinate ti on the t-axis for the start of the action; the tip of the vector has
a more distal coordinate on the d-axis for (say) the patient, and a coordinate tj on
the t-axis, such that tj is later than ti (see Chilton 2007 for detail). The English
progressive tense forms, which normally only apply to non-statives, give rise to a
cognitive effect that I call ‘windowing’. In effect the peripersonal temporal region
occludes the beginning and end of an action and simultaneously moves the view-
point ‘closer’ to a virtual viewpoint. These features correspond with many informal
descriptions of the progressive or imperfective.

11.4 Present and future

Jaszczolt (2009: 38–45, 50–5, 140–1) argues strongly for a modal view of the expres-
sion of deictic future times and proposes a scale of future time markers correspond-
ing to a scale expressing “a certain degree of detachment from the certainty of now”.
Following Jaszczolt’s scheme, but excluding be to, be about to, and be on the point of,
we might arrange the following sentences in the following order:
Henry visits Calais this Thursday
Henry is visiting Calais this Thursday
Henry is going to visit Calais this Thursday
Henry will visit Calais this Thursday
246 Paul Chilton

According to Jaszczolt, these are graded from highest certainty (least detachment) to
least certainty (highest detachment). While this scale (cf. Jaszczolt 2009: 140) seems
intuitively satisfactory, it is not completely clear how these particular tense forms
encode graded modal meanings. Jaszczolt outlines in some detail how Default
Semantics, including its revised version, explains meaning as an emergent phenom-
enon arising from societal, cultural, and other pragmatic information (p. 132) but the
tense form must have core meaning, and what is not clear is whether and how these
core meanings encode modal gradations. I shall therefore leave aside the question of
whether the semantic structure of the different tense forms has inherent modal
meaning that enables us to place them on a modal scale, and instead focus on the
semantic structure itself. By semantic structure here I mean the most abstract con-
ceptualizations that can be postulated for context-less meanings of the different
constructions that are used to communicate about deictically future events and states.
We shall consider the following examples:
(1) a. Henry visits Calais this Thursday
b. *Henry visits Calais [future time reference, no pragmatic or lexical indicator]
(2) a. Henry is visiting Calais this Thursday.
b. *Henry is visiting Calais [future time reference, no pragmatic or lexical
These examples suggest that present-tense forms cannot refer to a subjectively
future time without further cognitive input. This means either some explicit lexical
indicator such as this Thursday, next year, in a few days’ time. Note that a vague
indication such as in the future is probably not enough:
(3) a. ??Henry visits Calais in the future
b. ?Henry is visiting Calais in the future
It seems that some more precise time-indicating expression is required. This is not
necessarily a lexical time expression in the clause, however. It may be a mutually
shared piece of knowledge in the conversational context.4 These considerations will
lead us to a detailed model of present-for-future expressions that is somewhat more
detailed than the conventional accounts, and which is primarily temporal rather
than modal. I turn now to simple present and progressive present forms used to
conceptualize times that are in the future relative to the speaker and hearer.

11.4.1 Simple present and deictic future

In order to account for the ‘scheduling’ use of the simple present in (1), Langacker
(2001) makes use of a notion of ‘virtual entities’, understood by him in terms of a
Default Semantics also takes account of such information in explaining semantic representations
(merger representations) of utterances.
Frames of reference and the linguistic conceptualization of time 247

“non-canonical viewing arrangement”. Brisard deals with the problem primarily by

emphasizing a modal meaning of the simple present that he calls “immediate
givenness” (2002: 263–8). Brisard retains Langacker’s idea that the simple present
has a double meaning (immediate phenomenal experience and general structural
knowledge of the world), but offers a different description, in which the processual
verb is “projected” (by use of the simple present) on to a “virtual plane” equated with
representation of “structural aspects of the world” (p. 274). The DST approach,
however, provides a way of dealing with such matters within the overall theory of
frames of reference. This is not simply a question of a convenient ad hoc formal-
ization for describing the conceptual effects of language structure but one that has
ramifications into the wider philosophical arena. Bernard Lonergan (1957) proposes
three kinds of reference frames: the personal, the public, and the special. The third of
these is mathematical and physical and need not concern us here. The first corres-
ponds to personal cognition of three-dimensional physical space:
. . . everyone has his [and her] personal reference frame. It moves when he moves, turns when
he turns, and keeps its ‘now’ synchronized with his psychological present . . .
(1957: 144)

It is the case that DST does not work with the physical space but with a personal
conceptual space of a kind that specifically underlies language. To that extent DST
seeks to advance beyond Lonergan’s idea of reference frames. Nonetheless, Loner-
gan’s general point and the link he makes with temporal as well as spatial deictic
expressions is consistent with DST’s deictic space. What is of special concern in the
present context is his second class of reference frame, ‘public reference frames’.
These are both spatial and temporal. In the following, Lonergan is primarily
describing temporal reference frames:
[People] are familiar with alternations of night and day, with the succession of weeks and
months, with the uses of clocks and calendars. Now such relational schemes knit together
extensions and durations. But they are not personal reference frames that shift about with an
individual’s movements. On the contrary, they are public, common to many individuals, and
employed to represent the here and now of the personal reference frame into generally
intelligible locations and dates . . .
(1957: 144)

Cognitive linguistics, coming from a different angle, has a complementary point to

make concerning the nature of this publicly shared knowledge of times, drawing on
Fillmore’s work on cognitive frames (e.g. 1985). Lakoff describes what English calls a
week as an “idealized cognitive model”—we could equally call it an idealized public
reference frame—that contrasts with, for example, Geertz’s description of the Bali-
nese calendar (Geertz 1973: 392–3, cited by Lakoff 1987: 68). Names of days, such as
Tuesday, can only be defined relative to a reference frame, and this is clearly a
conceptual construct:
248 Paul Chilton

Tuesday can be defined only relative to an idealized model that includes the natural cycle
defined by the movement of the sun, the standard means of characterizing the end of one day
and the beginning of the next, and a larger seven-day calendric cycle—the week. In the
idealized model, the week is a whole with seven parts organized in a linear sequence . . .
(Lakoff 1987: 67)

Such frames (or ‘models’) do not exist in nature but are conceptualizations that
are agreed upon by long processes of cultural coordination.
Some formal semantic accounts of scheduling present take it that such cases arise
because they have the ‘connotation’ that they are predetermined with respect to the
present (Kaufmann et al. 2006: 90). I suggest that this ‘connotation’ arises because of
the insertion of a cognitive frame such as the week frame, where the days of the week
and the cyclic recurrence of weeks is in a sense ‘predetermined’.
I am proposing that the notion of idealized frames can be usefully thought of in
terms of reference frames. Fillmore’s frames and Lakoff ’s models, at least those that
concern time, can usefully be understood as reference frames that enter into relations
with personal reference frames of the kind that DST models in its particular
geometry. The crucial point is that now is a personal cognitive experience, not a
socially agreed-upon time in a public frame of reference The personal and the public
frames of reference have to be ‘coordinated’—their separate coordinate systems have
to be brought into some sort of alignment, the one relative to the other. We shall make
use of this idea in looking at the meanings of English present tense morphology.
In DST we already have the apparatus to model time-related frames. The first step
in considering how to use them to model the ‘scheduling’ use of the simple present
tense form is to note that its use suggests a secondary set of axes R’ whose origin is
located at some time t in the future relative to S. Further, we may say that the origin
of the scheduling axes R’ is not at S’, but some other referent such as a timetable or
some pragmatically given ‘public reference frame’ (PRF), which might include a
shared mental representation of a written or verbal agreement, or shared cognitive
frame representing a time cycle such as the calendar or the seasons, in the encyclo-
paedic memory of the interlocutors.
In my use of the term ‘schedule’ here, one should not be misled into importing
contemporary associations, since modern scheduling is merely a kind of cooperative
planning for the future that is characteristic of all human societies from early
beginnings that may very well be closely tied to the emergence of human language.
More needs to be said about how DST can geometrically model PRFs, but consider
first the diagram in Figure 11.2. In this and subsequent figures, the peripersonal
temporal region is shown as a shaded plane.
Because the PRF is public there is no S defining a subjective now, so there is also
no m-axis representing the modal component of S’s conceptualization, and no d-axis
on which S ‘locates’ discourse referents at subjectively relative ‘distances’. It is simply
Frames of reference and the linguistic conceptualization of time 249


time region


Tue S

Figure 11.2 (1) Henry visits Calais this Thursday.

an ordered set of seven arbitrarily named time periods, each of which corresponds to
the diurnal cycle. These intervals lie on a finite directed timeline that defines a one-
dimensional plane. Though inherently it is not deictically centred, it is directed in
the direction of ontological time. In common with other individuals in the culture, S
holds PRF in long-term memory but also aligns it, in virtue of socially shared
knowledge about ‘which day it is’, with his subjective conceptualization of periper-
sonal now.
Consider Figure 11.2. First the base axes R are aligned with the public reference
frame PRF so that S’s now is aligned with ‘Tuesday’ in the week frame. Next set up
R’, a copy of S’s base R, a virtual frame of reference in which S’s now is aligned with
Thursday in the PRF. It is in R’ that Henry’s sailing to Calais is located, a reference
frame located at a specific future that has both subjective and public reality for S,
as the diagram shows.
Now this future event is expressed by the simple present, and the implications of
this tense as distinct from the progressive have to be explained. The vector ‘visits’ is
located at 0 in PRF. Either this means the event of leaving on Thursday is a timeless
property, or it means that it is viewed as a timeless instant—the relevance here is that
the event is viewed instantaneously as a whole, in contrast with the windowing
conceptualization cued by the progressive form, which excludes beginning and end.
250 Paul Chilton

11.4.2 Present progressive and deictic future: window on the future

Neither simple present nor present progressive can refer to the future without extra
lexical or pragmatic specification and presumably therefore additional cognitive
structure. Consider the following:
(4) Henry is visiting Calais
(2) Henry is visiting Calais this Thursday
In (4) the default understanding is that Henry is visiting Calais now, at the time of
speaking. In (2) the speaker sets up a mental space R’. But this future is located in a
PRF, as shown in Figure 11.3. The deictic space model for the conceptualization
induced by this use of the progressive combined with a lexical (or pragmatic) time
indicator proceeds as for the case of the simple present (Figure 11.2) above.
In this new space, represented by the dashed axes in Figure 11.3, S (and the hearer)
have a present now located at a deictically future time, indicated by a temporal
deictic this Thursday, defined relative both to the PRF and to base R. Future now and
present this Thursday are conceptually co-located. Within the embedded frame, i.e.
relative to it, we have the windowing operation due to the progressive form is
visiting. The ‘up close right now’ element is represented by the conceptual window-
ing operation within that frame. Sentence (2) seems to bring us ‘close’ to the action,
to ‘put us in the picture’. It does not seem necessary to consider this effect ‘modal’.
Moreover, it is often noted that the intuited sense of such examples is that the future
event is somehow viewed as rooted in the speaker’s present—according to the
present model, this is accounted for within the shifted reality frame R’.
How should we characterize the difference between the simple present and the
progressive present referring to a future time? The elements of a description are
made explicit in the notions of coordinate geometry, vectors, and shifts of reference
frame. The temporal component is the same for both tense forms (see Figures 11.2
and 11.3). The difference is the structure of the instancing operator and the window-
ing operator (induced by the simple present form and progressive form, respect-
ively). In the former, the whole event is viewed as a completed instant, but since the
simple present also serves to conceptualize timeless generic properties, it carries a
potential implication of high subjective certitude. These elements are not all neces-
sarily present in the case of the progressive form (windowing operator), which has
the of close-up viewing and incompleteness (as the ‘ends’ of the action are not ‘in
view’). This effect may or may not result in a sense of greater or lesser certitude
compared with the simple present. It is thus not clear to me that these forms should
be viewed as primordially ‘modal’, though modal effects may arise by implication
under the influence of various contextual factors.
Frames of reference and the linguistic conceptualization of time 251


Sun S S
Tue Henry
Figure 11.3 Frame shift for (2) Henry is visiting Calais this Thursday.

11.4.3 Moving into the future

We are concerned here with the ‘periphrastic future’ tense forms be going to/gonna,
as in (5):
(5) Henry is going to/gonna visit Calais this Thursday
Such forms are discussed in terms of grammaticalization by Hopper and Traugott
(2003). In the case of gonna, which is only possible with a following verb (not an
NP), the fact that going assimilates to indicates that we are dealing with a coalesced
concept of directed motion; that is, we are not dealing with an infinitive form ‘to þ
verb’, e.g. to visit. In terms of cognitive metaphor theory, the underlying metaphor,
as is well known, is purpose is directed movement (cf. Lakoff and Johnson 1980).
Hopper and Traugott’s explanation of the grammaticalization of going to as an
auxiliary expressing future time rests on a pragmatic inference of futurity from
purposive action as expressed in the lexical verb go. However, this fails to generalize
over the symmetrical pair found in a language like French: venir de V (literally ‘to
come from V’, ‘to have just done’) and aller V (‘go V’). While Hopper and Traugott’s
account is not irrelevant, the French example suggests that this kind of temporal
auxiliary basically results from a direct conceptual transfer from space to time,
preserving deictic relations. The DST account makes this assumption and the
coordinate and vector components of that theory are well placed to model what is
252 Paul Chilton




this Thursday

Figure 11.4 (5) Henry is going to/gonna visit Calais this Thursday.

at issue. Spatial translation to/from a location is directly mapped onto temporal

translation to/from a time point.
As suggested by the grammatical construction in question, the model in Figure 11.4
shows two vectors: a component for be going follows the modelling already discussed
for progressive forms (windowing operations); this corresponds to the reported sense
that be going to futures are closely connected to the present, to some process already
in train at S’s now. The main verb (here the visit component vector) appears as a
timeless event vector at some point ti > t0 that might be specified pragmatically, or
lexically by a PP (here this Thursday).
Windowing operations usually leave the culmination of activities ‘out of view’, but
in the case of the periphrastic future construction that uses them the preposition to
specifies the endpoint, namely the eventuality expressed in the main verb (here visit).
Nonetheless, the model also captures the fact that from S’s present viewpoint, the
culmination is ‘outside’ her peripersonal space. And as for all progressive forms, the
windowing operator also leaves ‘out of view’ the beginning point of the activity,
which equally seems to be appropriate for the conceptual effects of the periphrastic
future that are often described.
Frames of reference and the linguistic conceptualization of time 253

These features seem to correspond well with the sense that periphrastic futures are
used for future situations that are predictable from the present moment and that the
speaker has a close experiential contact with evidence of in the present of action in
the future. Another way of putting this (Jaszczolt 2009: 65, reporting Eckardt 2006),
is to say, in Reichenbachian terms, that event point E is in the future and reference
point R is in the present and that be going to sentences are about the present time
and what is imminent in it, and perhaps immanent.
Note that there can be a subtle conceptual difference in Henry is going to visit
Calais next week. The speaker S may be imagining Henry’s intention, or S may be
communicating her/his own prediction. DST modelling handles this. In Figure 11.4
we have the case where S projects her own world R’ in which Henry visits Calais at a
future time. To model the meaning in which S is communicating Henry’s intention,
the origin of R’ has Henry’s coordinate on the d-axis.5

11.4.4 The putative future: a reference frame solution

Across languages it is not uncommon to find that a future tense form occurs in
contexts that yield a cognitive effect generally described as modal, or epistemic, or as
we shall prefer here, ‘putative’. I am not concerned with the precise pragmatic factors
that allow such understandings, merely with their conceptual structure when they do
occur. As Jaszczolt argues, the fact that the conventional future form will þ V can
express present probability may be evidence of the close cognitive connection between
future time reference and the inherent cognitive (or even metaphysical) uncertainty of
the future, but this is not necessarily a reason to conflate modality and temporality when
there is a good case, for the purpose of semantic modelling, for thinking that they are, or
can be, separate in linguistically expressed cognition. The English putative future
meaning of will þ V is not just a conventionalized expression of high probability.
The meaning of Mary is probably writing the report is not the same as Mary will be
writing the report (now) and is unlikely to be used in the same pragmatic contexts, i.e.
have the same pragmatic potential effects.
Furthermore, the meaning of will þ V seems to have a different internal
conceptual structure from must, may, and might þ V. Linguists working on French
have noted that the equivalent French form, futur putatif, presupposes a future point
of view at which the present eventuality is predicted to be verified (cf. Damourette
and Pichon 1911–1940; Sthioul 1998b; de Saussure and Morency 2012; de Saussure,
this volume). This is an insight that I adopt here.
The conceptualizations we have to consider arise in the context of certain kinds of
prompts for deictic times:

A DSM for venir de constructions can be built along similar spatiotemporal lines, allowing for the fact
that this French construction does not involve the progressive.
254 Paul Chilton

(6) Henry will visit Calais this Thursday

(7) Henry will be visiting Calais this Thursday
(8) Henry will be visiting Calais (now)
A simple present version of (8), Henry will visit Calais now, in the putative sense is
not valid because of the regularities governing of non-statives.
Sentence (6) is the ‘regular future’ and (7) is a ‘regular future’ use of will combined
with the progressive form (windowing) that is represented by the insertion of a new
viewpoint in a new R’ located at some time in the future relative to S—this is shown
in Figure 11.5. What do we have for (8), where we have a future progressive form
understood as referring to the present now? We can simply put the visit verb vector
at high probability on the m-axis. But this is not explanatory. We want to see if the
frame-shift model of DST can account for this use of the future form with auxiliary
will. The approach adopted here (Damourette and Pichon 1911–1940; Sthioul 1998b;
de Saussure and Morency 2012) is based on the idea of a represented verification of a
current eventuality (current, that is, for S in base R). The authors of this approach
speak of an ‘imaginary’ future ‘perspective’ or ‘viewpoint’. They also insist that this
viewpoint is allocentric, in the sense that it is distinct from the current speaker in the
actual world, a claim that DST frames in a slightly different way.

PRF Henry

this Thursday

Figure 11.5 (7) Henry will be visiting Calais this Thursday [non-putative].
Frames of reference and the linguistic conceptualization of time 255

Figures 11.5 and 11.6 show the different conceptual structures, in DST terms, of the
different potential meanings of the sentence Henry will be visiting Calais; Figure 11.5
modelling the non-putative ‘regular future’ meaning, and Figure 11.6 modelling the
putative use of regular future to refer to a present eventuality.
Unlike the future tense form will þ V, the progressive form introduces the
viewing frame (windowing operator), which involves an embedded set of coordin-
ates, transposing a represented or ‘imaginary’ S to the future time point in the base
coordinates indicated pragmatically by will at the deictic time t represented by this
Thursday. The cognitive effect seems to be one in which the utterer of this sentence
is ‘closer’ to the projected future event than in the sentence Henry will visit Calais
next week. Next we have to consider whether this modelling approach can handle the
conceptual effects triggered when the adverbial this Thursday is replaced by, for
example, now, at this moment, etc. or by some equivalent but unexpressed pragmatic
indicator, as in (8).
The frame-shifting principles of DST seem to be able to accommodate this
putative future use of the future. The putative future is the reverse of (2). Whereas
in the latter we have a present tense form referring to a future time, in (8) we have a
future tense referring to present time, always relative of course to the speaker’s now.


S Henry
t0 now

Figure 11.6 (8) Henry will be visiting Calais (now) [putative].

256 Paul Chilton

Broadly following the proposals of Sthioul (1998b) and de Saussure and Morency
(2012), we have in Figure 11.6 a meta-represented point of view at some ti > t0. This
additionally represented point of view does not have to be an unspecified allocentric
other in the sense of a different individual Si. Rather, this other point of view is that
of a cognitive avatar of S, S’, and can still be regarded as ‘allocentric’ in that sense.
This is also consistent with examples from contexts in which it is pragmatically
manifest that the speaker herself will be in a position to verify, in the future, a
current situation. In Figure 11.6, from the viewpoint of S’ the vector for ‘visiting’ is in
the past and it is true (located in the plane m¼0); from the viewpoint of S ‘visiting’ is
in the present. The use of coordinate systems as reference frames makes it possible to
model this property of the putative future construction.
Just as ‘regular future’ tense forms can, in the putative construction, refer to S’s
present time, so ‘regular future perfect’ tense forms can refer to eventualities in S’s
past time. Such a reading is available for (9):
(9) Henry will have visited/will have been visiting Calais
In the account proposed by Sthioul and de Saussure and Morency, sentence (9) may be
said to take a view of a past event from a future vantage point at which it will be verified.
The sentence is of course ambiguous, since it can also be read in a simply temporal
sense as referring to a time ti in S’s future but before some future vantage point (i.e. tj >
ti). One may indeed say that it is doubly ambiguous, since for some speakers this
temporal reading of (9) may be either simply temporal or putative, again in the sense of
Sthioul and de Saussure and Morency. If this is so, the ambiguity indicates not that
future time is inherently modal but that there are two possible conceptualizations of the
future, one predictive and the other something short of predictive.
The geometric DST model for these sentences will have the same structure as in
Figure 11.6, except that the vector representing visit will be located at some time prior
to S’s now, as in Figure 11.7, which illustrates a model for the non-progressive in
order to simplify, and for a reading that is putative.
In this model for the putative reading of (9) the embedded axes are still in the
future relative to S, since, in the account of the putative meaning that we are
adopting, verification is still in the future relative to S. What is different between
(9) and (8) is the location of the eventuality in the past relative to S, i.e. at some
ti < t0. The diagram can show how the have been component of the construction,
referring to the past relative to S, is conceptually represented simultaneously with the
will component associated with future reference relative to S.6

A similar basic structure models the progressive tense version of (9), but with some additional
complexities that we shall not attempt to deal with here.
Frames of reference and the linguistic conceptualization of time 257


S Henry

Figure 11.7 (9) Henry will have visited Calais [putative reading].

11.5 Concluding remarks

In this chapter I have paid attention only to a small number of temporal expressions,
in particular those that refer to the future, that is, the deictic future relative to some
speaker S: simple present for future, progressive present for future, periphrastic
present going to, with brief mention of the ‘regular’ will future. The overall aim
has been to demonstrate how a geometrical approach formulated in DST can,
perhaps surprisingly, elucidate temporal relations and in particular bring out the
role of shifting reference frames. This approach separates the conceptual dimensions
of modality and time and one of the ancillary aims has been to characterize the
distinctions between the different constructional means of referring to the future in
terms of shifting point of view rather than modal scales. This is not to deny that
modal effects may arise, but the present model does not treat them as inherently
conflated with time. If and when they arise, they arise through contextual factors
interacting with the conceptual structures built from reference frames.
Finally it should be noted that the geometrical structures I have proposed are
primarily heuristic. They are a hypothesis about what I suspect is the fundamentally
perspectival and deictic scaffolding of core linguistic structures. Of course the
coordinate-and-vector models do not, and are not intended to, capture all the semantic
content of lexical elements. In attempting to shed light on the nature of the linguistic
258 Paul Chilton

conceptualization of temporal relations, the borrowing of geometrical notions may

seem perverse. However, if language uses spatial representations to communicate
about time, as is widely claimed by cognitive linguists, then geometry is appropriate.
Moreover, if perspectival alternation is a crucial feature of linguistic conceptualiza-
tion, then the essentially geometrical notion of reference frames may also provide a
fruitful means of description and analysis. Reference frames are indeed essential to
any account of space–time and may also be essential for human cognition.
Aikhenvald, Alexandra Y. (2004). Evidentiality. Oxford: Oxford University Press.
Aikhenvald, Alexandra Y. (2007). ‘Information source and evidentiality: what can we
conclude?’, Italian Journal of Linguistics 19(1): 209–27.
Apel, Karl-Otto (1986). ‘Le problème de l’évidence phénoménologique à la lumière d’une
sémiotique transcendantale’, trans. J. Poulain, Critique 464–465: 79–113.
Asher, Nicholas (1993). Reference to abstract objects in discourse. Dordrecht: Kluwer.
Asher, Nicholas (2000). ‘Events, facts, propositions, and evolutive anaphora’, in
J. Higginbotham, F. Pianesi, and A. C. Varzi (eds.), Speaking of events. Oxford: Oxford
University Press, 123–50.
Asher, Nicholas (2011). Lexical meaning in context: a web of words. Cambridge: Cambridge
University Press.
Asher, Nicholas and Myriam Bras (1993). ‘The temporal structure of French texts within
Segmented Discourse Representation Theory’, in M. Aurnague, A. Borillo, M. Borillo, and
M. Bras (eds.), Semantics of time, space, movement and spatio-temporal reasoning (Working
Papers of the 4th International Workshop, University Toulouse-Le Mirail), 203–17.
Asher, Nicholas and Alex Lascarides (2003). Logics of conversation. Cambridge: Cambridge
University Press.
Asher, Nicholas and James Pustejovsky (2006). ‘A type composition logic for generative
lexicon’, Journal of Cognitive Science 6: 1–38.
van der Auwera, Johan and Vladimir A. Plungian (1998). ‘Modality’s semantic map’, Linguistic
Typology 2: 79–124.
Bally, Charles (1932). Linguistique générale et linguistique française. Paris: Leroux.
Barker, Chris, and Pauline Jacobson (eds.) (2007). Direct compositionality. Oxford: Oxford
University Press.
Barone-Adesi, Giovanni (2002). ‘The role of inside information. Financial disclosure and value
creation’, in S. Russ-Mohl (ed.), Business journalism, corporate communications, and
newsroom management. Lugano: Università della Svizzera Italiana, 63–8.
Barwise, Jon and John Perry (1983). Situations and attitudes. Cambridge, MA: MIT Press.
Battistelli, Delphine (2009). ‘La temporalité linguistique’, HDR dissertation, University Paris
Beaver, David and Cleo Condoravdi (2003). ‘A uniform analysis of before and after’,
in R. Young and Y. Zhou (eds.), Proceedings of SALT XIII. Cornell: CLC Publications,
Bellahsène, Linda (2007). ‘L’expression de la conjecture: le cas du futur en français’ in
D. Bouchard, I. Evrard, and E. Vocaj (eds.), Représentations du sens linguistique. Brussels:
De Boeck, 253–66.
van Benthem, Johan and Alice G. B. ter Meulen (eds.) (2011). Handbook of logic and language,
2nd edn. Berlin: Springer.
260 References

Berthonneau, Anne-Marie and Georges Kleiber (2006). ‘Sur l’imparfait contrefactuel’,

Travaux de linguistique 53: 7–65.
Berti, Anna and Giacomo Rizzolatti (2002). ‘Coding near and far space’, in H.-O. Karnath,
D. Milner and G. Vallar (eds.), The cognitive and neural bases of spatial neglect. Oxford:
Oxford University Press, 119–30.
Blakemore, Diane (1987). Semantic constraints on relevance. Oxford: Blackwell.
Bloor, Thomas and Makaya Pindi (1990). ‘Schematic structures in economic forecasts’, in
T. Dudley-Evans and W. Henderson (eds.), The language of economics: the analysis of
economics discourse (ELT Documents: 134). London: Modern English Publications in
association with the British Council, 55–66.
Bonomi, Andrea (1980). ‘Determinismo e semantiche per logiche temporali’, in S. Bernini
(ed.), Atti del convegno nazionale di logica (1–5 ottobre 1979). Naples: Bibliopolis, 351–79.
Boogaart, Ronny (1999). ‘Aspect and temporal ordering: a contrastive analysis of Dutch and
English’, Ph.D. dissertation, VU University Amsterdam.
Borillo, Andrée (1986). ‘La quantification temporelle: durée et itérativité en français’, Cahiers
de grammaire 11: 117–56.
Borillo, Andrée (2005). ‘Peut-on identifier et caractériser les formes lexicales de l’aspect en
français?’, in H. Bat-Zeev Shyldkrot and N. Le Querler (eds.), Les périphrases verbales.
Amsterdam: Benjamins, 67–82.
Borillo, Andrée (2006). ‘Quelques formes lexicales de l’aspect adnominal en français’, Cahiers
de grammaire 30: 21–38.
Bourne, Craig (2006). A future for presentism. Oxford: Clarendon Press.
Branquinho, João (2006). ‘On the persistence of indexical belief ’. Paper given to the APA
Eastern Division Meeting, December, 2005, revised 2006. Available online at <http://www.>.
Bres, Jacques (2003). ‘Temps verbal, aspect et point de vue: de la langue au discours’, Cahiers
de praxématique 41: 55–84.
Bres, Jacques (2005). L’imparfait dit narratif. Paris: CNRS Editions.
Bres, Jacques (2006). ‘Encore un peu, et l’imparfait était un mode . . . L’imparfait et la valeur
modale de contrefactualité’, Cahiers de praxématique 47: 149–76.
Brisard, Frank (2002). ‘The English present’, in F. Brisard (ed.), Grounding: the epistemic
footing of deixis and reference. Berlin: Mouton de Gruyter, 251–97.
Brisard, Frank (2010). ‘Aspects of virtuality in the meaning of the French imparfait’, Linguis-
tics 48: 487–524.
Brisard, Frank and Michael Meeuwis (2009). ‘Present and perfect in Bantu: the case of
Lingála’, Journal of African Languages and Linguistics 30: 21–43.
Broad, Charlie Dunbar (1923). Scientific thought. London: Kegan Paul.
Broad, Charlie Dunbar (1938). An examination of McTaggart’s philosophy. Cambridge:
Cambridge University Press.
Brunot, Ferdinand (1922). La pensée et la langue. Paris: Masson.
Bühler, Karl (1990 [1934]). Theory of language. The representational function of language.
Amsterdam/Philadelphia: Benjamins.
Button, Tim (2006). ‘There’s no time like the present’, Analysis 66: 130–35.
References 261

Button, Tim (2007). ‘Every now and then, no-futurism faces no sceptical problems’, Analysis
67: 325–32.
Callender, Craig (ed.) (2011). The Oxford handbook of philosophy of time. Oxford: Oxford
University Press.
Caplan, Ben and David Sanson (2011). ‘Presentism and truthmaking’, Philosophy Compass 6:
Casati, Roberto and Achille Varzi (1999). Parts and places. The structure of spatial represen-
tation. Cambridge, MA: MIT Press.
Caudal, Patrick (2012). ‘Relations entre temps, aspect, modalité et évidentialité dans le système
du français’, Langue française 173: 115–29.
Caudal, Patrick and Carl Vetters (2006). ‘Les temps verbaux: des connecteurs temporels qui
s’ignorent ?’, Cahiers Chronos 15: 105–37.
Caudal, Patrick, Laurent Roussarie, and Carl Vetters (2003). ‘L’imparfait, un temps inconsé-
quent’, Langue française 138: 61–74.
Celle, Agnès (2004). ‘The French future tense and English will as markers of epistemic
modality’, Languages in Contrast 5(2): 181–218.
Chafe, Wallace (1995). ‘The realis–irrealis distinction in Caddo, the Northern Iroquoian
languages, and English’, in J. Bybee and S. Fleischman (eds.), Modality in grammar and
discourse. Amsterdam: Benjamins, 349–65.
Chilton, Paul (2005). ‘Vectors, viewpoint and viewpoint shift’, Annual Review of Cognitive
Linguistics 3(1): 78–116.
Chilton, Paul (2007). ‘Geometrical concepts at the interface of formal and cognitive models:
Aktionsart and aspect in English’, Pragmatics and Cognition 15(1): 91–114.
Chilton, Paul (2009). ‘From mind to grammar: coordinate systems, prepositions, construc-
tions’, in V. Evans and P. Chilton (eds.), Language, cognition and space. London: Equinox,
Church, Alonzo (1936). ‘An unsolvable problem of elementary number theory’, American
Journal of Mathematics 58: 354–63.
Cinque, Guglielmo (1999). Adverbs and functional heads. Oxford: Oxford University Press.
Coates, Jennifer (1983). The semantics of the modal auxiliaries. London: Croom Helm.
Coates, Jennifer (1995). ‘The expression of root and epistemic possibility in English’, in
J. Bybee and S. Fleischman (eds.), Modality in grammar and discourse. Amsterdam:
Benjamins, 55–66.
Col, Gilles and Bernard Victorri (2007). ‘Comment formaliser en linguistique cognitive?
Opération de fenêtrage et calcul du sens temporel’, in G. Achard-Bayle and M.-A. Paveau
(eds.), Cognition, discours, contextes. Poitiers: Corela (online publication).
Comrie, Bernard (1976). Aspect: an introduction to the study of verbal aspect and related
problems. Cambridge: Cambridge University Press.
Condoravdi, Cleo (2002). ‘Temporal interpretation of modals: modals for the present and for
the past’, in D. Beaver, S. Kaufmann, B. Clark, and L. Casillas (eds.), The construction of
meaning. Stanford: CSLI, 59–88.
Cooper, Geoff and Mary Ebeling (2007). ‘Epistemology, structure and urgency: the sociology
of financial and scientific journalists’, Sociological Research Online 12(3).
Copley, Bridget (2009). The semantics of the future. New York: Routledge.
262 References

Coventry, Kenny R. and Simon C. Garrod (2004). Saying, seeing and acting: the psychological
semantics of spatial prepositions. London: Taylor and Francis.
Croft, William (2001). Radical Construction Grammar. Oxford: Oxford University Press.
Croft, William (2011). Verbs: aspect and causal structure. Oxford: Oxford University Press.
Culioli, Antoine (1978). ‘Valeurs modales et opérations énonciatives’, Le français moderne 46
(4): 300–17.
Dainton, Barry (2000). Stream of consciousness. London: Routledge.
Dainton, Barry (2001). Time and space. Chesham: Acumen.
Dainton, Barry (2008). ‘Sensing change’, Philosophical Issues 18: 362–84.
Damourette, Jacques and Edouard Pichon (1911–1940). Des mots à la pensée. Essai de gram-
maire de la langue française. Paris: d’Artrey.
Davidson, Donald (1967a). ‘The logical form of action sentences’, in N. Rescher (ed.), The logic
of decision and action. Pittsburgh, PA: University of Pittsburgh Press, 81–120. Reprinted in
D. Davidson (1980) Essays on actions and events. Oxford: Clarendon Press, 105–122.
Davidson, Donald (1967b). ‘Truth and meaning’, Synthese 17: 304–23. Reprinted in
D. Davidson (ed.), (1984) Inquiries into truth and interpretation. Oxford: Oxford University
Press, 17–36.
Davidson, Donald (1968/69). ‘The logical form of action sentences’, in P. Ludlow (ed.),
Readings in the philosophy of language. Cambridge, MA: MIT Press, 337–46.
Davidson, Donald (1986). ‘A nice derangement of epitaphs’, in E. Lepore (ed.), Truth
and interpretation: perspectives on the philosophy of Donald Davidson. Oxford: Blackwell, 433–46.
Davies, Paul (1995). About time: Einstein’s unfinished revolution. Harmondsworth: Penguin.
Davies, Paul (2012). ‘That mysterious flow’, Scientific American Special 21(1): 8–13.
Demirdache, Hamida and Myriam Uribe-Etxebarria (2002). ‘La grammaire des prédicats
spatiotemporels: temps, aspects et adverbes de temps’, in B. Laca (ed.), Temps et aspect.
De la morphologie à l’interprétation. Saint-Denis: P.U.V., 125–76.
De Mulder, Walter (2003). ‘Imparfait et “point de vue” à la recherche d’une théorie
“cognitive” ’, in J. Guéron and L. Tasmowski (eds.), Temps et point de vue. Université
Paris X-Nanterre: Publidix, 81–102.
De Mulder, Walter and Franck Brisard (2006). ‘L’imparfait marqueur de réalité virtuelle’,
Cahiers de praxématique 47: 97–124.
Dendale, Patrick and Julie van Bogaert (2012). ‘Réflexion sur les critères de définition et
les problèmes d’identification des marqueurs évidentiels en français’, Langue française 173: 13–29.
Desclés, Jean-Pierre (1994). ‘Quelques concepts relatifs au temps et à l’aspect pour l’analyse des
textes’, Etudes cognitives 1: 57–88. Warsaw: Polish Academy of Sciences, Institute of Slavic Studies.
De Wit, Astrid and Frank Brisard (to appear a). ‘A Cognitive Grammar account of the
semantics of the English present progressive’. Manuscript.
De Wit, Astrid and Frank Brisard (to appear b). ‘Zero verb marking in Sranan’. Journal of
Pidgin and Creole Languages.
De Wit, Astrid and Adeline Patard (to appear). ‘Modality, aspect and the progressive: the semantics
of the present progressive in French, in comparison with English’, Languages in Contrast.
Diessel, Holger (1999). Demonstratives: form, function and grammaticalization. Amsterdam:
Diessel, Holger (2006). ‘Demonstratives, joint attention, and the emergence of grammar’,
Cognitive Linguistics 17(4): 463–89.
Dik, Simon (1989). The theory of Functional Grammar I. Dordrecht: Foris.
References 263

Doiz Bienzobas, Aintzane (2002). ‘The preterite and the imperfect as grounding predications’,
in F. Brisard (ed.), Grounding: the epistemic footing of deixis and reference. Berlin: Mouton
de Gruyter, 299–347.
Donohue, James P. (2006). ‘How to support a one-handed economist: the role of modalisation
in economic forecasting’, English for Specific Purposes 25(2): 200–16.
Dowty, David R. (1979). Word meaning and Montague Grammar: the semantics of verbs
and times in generative semantics and Montague’s PTQ (Studies in Linguistics and
Philosophy 7). Dordrecht: Kluwer.
Dummett, Michael (2006). Thought and reality. Oxford: Clarendon Press.
Dyke, Heather (2007). Metaphysics and the representational fallacy. London: Routledge.
Eckardt, Regine (2006). Meaning change in grammaticalization: an enquiry into semantic
reanalysis. Oxford: Oxford University Press.
Enç, Mürvet (1987). ‘Anchoring conditions for tense’, Linguistic Inquiry 18: 633–57.
Escandell-Vidal, Victoria, Manuel Leonetti, and Aoife Ahern (eds.) (2011). Procedural mean-
ing. Problems and perspectives. Bingley: Emerald (CRISPI 25), 55–79.
Evans, Gareth (1985). ‘Understanding demonstratives’, in G. Evans (ed.), Collected papers.
Oxford: Oxford University Press, 291–321.
Evans, Vyvyan (2004). The structure of time: language, meaning and temporal cognition.
Amsterdam: Benjamins.
Faller, Martina T. (2002a). ‘Semantics and pragmatics of evidentials in Cuzco Quechua’, Ph.D.
thesis, Stanford University.
Faller, Martina T. (2002b). ‘Remarks on evidential hierarchies’, in D. I. Beaver, L. D. Casillas
Martínez, B. Z. Clark, and S. Kaufmann (eds.), The construction of meaning. Stanford, CA:
CSLI Publications, 89–111.
Faller, Martina T. (2011). ‘A possible worlds semantics for Cuzco Quechua evidentials’, in N. Li
and D. Lutz (eds.), Proceedings of SALT 20. Ithaca, NY: Cornell University Press, 660–83.
Fauconnier, Gilles (1979). ‘Comment contrôler la vérité. Remarques illustrées par des asser-
tions dangereuses et pernicieuses en tout genre’, Actes de la recherche en sciences sociales 25:
Fauconnier, Gilles (1994). Mental spaces. Cambridge: Cambridge University Press.
Filipović, Luna and Kasia M. Jaszczolt (eds.) (2012a). Space and time in languages and cultures:
linguistic diversity (Human Cognitive Processing Series 36). Amsterdam/Philadelphia:
Filipović, Luna and Kasia M. Jaszczolt (eds.) (2012b). Space and time in languages and cultures:
language, culture, and cognition (Human Cognitive Processing Series 37). Amsterdam/
Philadelphia: Benjamins.
Fillmore, Charles (1982). ‘Towards a descriptive framework for spatial deixis’, in R. J. Jarvella
and W. Klein (eds.), Speech, place and action. Chichester: John Wiley, 31–59.
Fillmore, Charles (1985). ‘Frames and the semantics of understanding’, Quaderni di semantica
6(2): 222–53.
Fine, Kit (2005). ‘Tense and reality’, in K. Fine. Modality and tense: philosophical papers.
Oxford: Clarendon Press, 261–320.
von Fintel, Kai and Anthony S. Gillies (2008). ‘CIA leaks’. Philosophical Review 117: 77–98.
264 References

von Fintel, Kai and Anthony S. Gillies (2011). ‘ “Might” made right’, in A. Egan and
B. Weatherson (eds.), Epistemic modality. Oxford: Oxford University Press, 108–30.
von Fintel, Kai and Sabine Iatridou (2006). ‘How to say ought in foreign: the composition of
weak necessity modals’, in J. Guéron and J. Lecarme (eds.), Time and modality. Dordrecht/
Heidelberg/London/New York: Springer, 115–41.
von Fintel, Kai and Lisa Matthewson (2008). ‘Universals in semantics’, Linguistic Review 25:
Fleischman, Suzanne (1982). The future in thought and language: diachronic evidence from
Romance. Cambridge: Cambridge University Press.
Fleischman, Suzanne (1995). ‘Imperfective and Irrealis’, in J. Bybee and S. Fleischman (eds.),
Modality in grammar and discourse. Amsterdam: Benjamins, 519–51.
François, Jacques (1993). ‘La compréhension de l’articulation chronologique des séquences
narratives: un exercice de linguistique cognitive’, Verbum 1993(4): 49–70.
François, Jacques (2003). La prédication verbale et les cadres prédicatifs. Louvain/Paris: Peeters.
Frank, Anette (1996). ‘Context dependence in modal constructions’, Ph.D. thesis, Institut für
Maschinelle Sprachverarbeitung der Universität Stuttgart.
Frawley, William (1992). Linguistic semantics. Hillsdale, NJ: Erlbaum.
Frawley, William (ed.) (2006). The expression of modality. Berlin: Mouton de Gruyter.
Freeman, James B. (2005). Acceptable premises: an epistemic approach to an informal logic
problem. Cambridge: Cambridge University Press.
Frege, Gottlob (1956 [1915–1919]). ‘The thought: a logical enquiry’, Mind, New Series, 65(259):
289–311 [First published in the Beiträge zur Philosophie des Deutschen Idealismus 1915–1919;
reprinted in P. F. Strawson (ed.) (1967), Philosophical logic, Oxford: Oxford University
Press, 17–38].
Gallagher, Sean (2003). ‘Sync-ing in the stream of experience: time-consciousness in Broad,
Husserl, and Dainton’, Psyche 9(10). Available online at <
Gallese, Vittorio (2003). ‘A neuroscientific grasp of concepts: from control to representation’,
Philosophical Transactions: Biological Sciences 358(1435): 1231–40.
van Geenhoven, Veerle (2004). ‘For-adverbials, frequentative aspect and pluractionality’,
Natural Language Semantics 12: 135–90.
van Geenhoven, Veerle (2005). ‘Atelicity, pluractionality and adverbial quantification’, in
H. Verkuyl, H. de Swart, and A. van Hout (eds.), Perspectives on aspect. Dordrecht: Kluwer,
Geertz, Clifford (1973). The interpretation of cultures. New York: Basic Books.
Givón, Talmy (2005). Context as other minds: the pragmatics of sociality, cognition and
communication. Amsterdam/Philadelphia: Benjamins.
Goldsmith, John and Erich Woisetschlaeger (1982). ‘The logic of the English progressive’.
Linguistic Inquiry 13: 79–89.
Gosselin, Laurent (1996). Sémantique de la temporalité en français. Louvain-la-Neuve:
Gosselin, Laurent (1999). ‘La valeur de l’imparfait et du conditionnel dans les systèmes
hypothétiques’, Cahiers Chronos 4: 29–51.
Gosselin, Laurent (2005). Temporalité et modalité. Bruxelles: De Boeck-Duculot.
Gosselin, Laurent (2010a). Les modalités en français. Amsterdam/New York: Rodopi.
References 265

Gosselin, Laurent (2010b). ‘Entre lexique et grammaire: les périphrases aspectuelles du fran-
çais’, Cahiers de lexicologie 96: 67–95.
Gosselin, Laurent (2010c). ‘Les relations entre périphrases aspectuelles et conjugaisons
en français’, in N. Flaux, D. Stosic, and C. Vet (eds.), Interpréter les temps verbaux.
Berne: Peter Lang, 36–61.
Gosselin, Laurent (2011). ‘L’aspect de phase en français: le rôle des périphrases verbales’,
Journal of French Language Studies 21(3): 149–71.
Gosselin, Laurent and Cédric Person (2005). ‘Temporalité’, in P. Enjalbert (ed.), Sémantique et
traitement automatique du langage naturel. Paris: Hermès-Lavoisier, 173–213.
Gougenheim, Georges (1929). Etude sur les périphrases verbales de la langue française. Paris:
Les Belles Lettres.
Granger, Gilles-Gaston (1976). La théorie aristotélicienne de la science. Paris: Aubier Mon-
Grice, Paul (2001). Aspects of reason (ed. R. Warner). Oxford: Clarendon Press.
Groenendijk, Jeroen and Martin Stokhof (1991). ‘Dynamic predicate logic’, Linguistics and
Philosophy 14(1): 39–100.
Grondelaers, S., K. Deygers, H. van Aken, V. van den Heede, and D. Speelman (2000). ‘Het
ConDiv-Corpus Geschreven Nederlands’, Nederlandse Taalkunde 5: 356–63.
de Groote, Philippe (2006). ‘Towards a Montagovian account of dynamics’, in M. Gibson and
J. Howell (eds.), Proceedings of SALT 16. Ithaca, NY: CLC Publications, 148–55.
Gross, Maurice (1986). Grammaire transformationnelle du français 1. Syntaxe du verbe. Paris:
Guillaume, Gustave (1929). Temps et verbe. Paris: Nizet.
de Haan, Ferdinand (2006). ‘Typological approaches to modality’, in W. Frawley (ed.), The
expression of modality. Berlin/New York: Mouton de Gruyter, 27–69.
Hacquard, Valentine (2010). ‘On the event relativity of modal auxiliaries’, Natural Language
Semantics 18: 79–114.
Haeseryn, W., K. Romijn, G. Geerts, J. de Rooij, and M. C. van den Toorn (1997). Algemene
Nederlandse Spraakkunst, 2nd rev. edn. Groningen: Nijhoff.
Hamdani Kadri, Djaouida (2006). Sémantique de la temporalité en arabe parlé d’Alger. Berne:
Peter Lang.
Hamm, Fritz, Hans Kamp, and Michiel van Lambalgen (2006). ‘There is no opposition
between formal and cognitive semantics’, Theoretical Linguistics 32: 1–40.
Harris, John (1984). ‘Syntactic variation and dialect divergence’, Journal of Linguistics 20:
Harrison, Robert P. (2003). The dominion of the dead. Chicago, IL: University of Chicago Press.
Havu, Jukka (2005). ‘L’expression du passé récent en français. Observations sur l’emploi de
la périphrase venir de þ infinitif ’, in H. Bat-Zeev Shyldkrot and N. Le Querler (eds.),
Les périphrases verbales. Amsterdam: Benjamins, 279–92.
Havu, Jukka (2006). ‘La grammaticalisation des périphrases aspectuelles et temporelles’.
Available at: <
Heck, Richard G. (2002). ‘Do demonstratives have senses?’, Philosophers’ Imprint 2(2). Avail-
able at: <>.
266 References

Heidegger, Martin (1953). Sein und Zeit. Tübingen: Max Niemeyer. [Trans. J. Stambaugh as
Being and time (1996). Albany, NY: State University of New York Press.]
Higginbotham, James (1995). ‘Tensed thoughts’, Mind and Language 10: 226–49.
Higginbotham, James (2003). ‘Remembering, imagining, and the first person’, in A. Barber
(ed.). Epistemology of language. Oxford: Oxford University Press, 496–533.
Higginbotham, James (2009). Tense, aspect and indexicality. Oxford: Oxford University Press.
Higginbotham, James, Fabio Pianesi, and Achille C. Varzi (eds.) (2000). Speaking of events.
Oxford: Oxford University Press.
Hobbs, Jerry R. (2010). ‘Clause internal coherence’, in P. Kühnlein, A. Benz, and C. L. Snider
(eds.), Constraints in discourse 2. Amsterdam/Philadelphia: Benjamins, 15–34.
Hopper, Paul J. and Elizabeth C. Traugott (2003). Grammaticalization, 2nd edn. Cambridge:
Cambridge University Press.
Horn, Laurence R. (1996). ‘Exclusive company: only and the dynamics of vertical inference’,
Journal of Semantics 13: 1–40.
Horn, Laurence R. (2000). ‘Pick a theory (not just any theory). Indiscriminatives and the free-
choice indefinite’, in L. Horn and Y. Kato (eds.), Negation and polarity: syntactic and
semantic perspectives. Oxford: Oxford University Press, 147–92.
Hornstein, Norbert (1990). As time goes by: Tense and Universal Grammar. Cambridge, MA:
MIT Press.
Husserl, Edmund (1928). Vorlesungen zur Phänomenologie des inneren Zeitbewusstseins.
Jahrbuch für Philosophie und phänomenologische Forschung IX. Halle: Max Niemeyer.
[Trans. J. B. Brough as Lectures on the phenomenology of the consciousness of internal
time in On the phenomenology of the consciousness of internal time (1893–1917), Part A
(1991). Dordrecht: Kluwer.]
van Inwagen, Peter (2002). Metaphysics, 2nd edn. Boulder, CO/Oxford: Westview Press.
Jackson, Frank (2011). ‘On Gettier holdouts’, Mind and Language 26: 468–81.
James, William (1890). The principles of psychology. New York: Henry Holt. [Reprinted
Cambridge, MA: Harvard University Press (1983).]
Jaszczolt, Kasia M. (2005). Default semantics: foundations of a compositional theory of acts of
communication. Oxford: Oxford University Press.
Jaszczolt, Kasia M. (2009). Representing time: an essay on temporality as modality. Oxford:
Oxford University Press.
Jaszczolt, Kasia M. (2010). ‘Default semantics’, in B. Heine and H. Narrog (eds.), The Oxford
handbook of linguistic analysis. Oxford: Oxford University Press, 193–221.
Jaszczolt, Kasia M. (2012). ‘Cross-linguistic differences in expressing time and universal
principles of utterance interpretation’, in L. Filipović and K. M. Jaszczolt (eds.), Space
and time in languages and cultures. Vol. 1. Linguistic diversity. Amsterdam/Philadelphia:
Benjamins, 95–121.
Jaszczolt, Kasia M. and Jiranthara Srioutai (2011). ‘Communicating about the past through
modality in English and Thai’, in A. Patard and F. Brisard (eds.), Cognitive approaches to
tense, aspect, and epistemic modality. Amsterdam/Philadelphia: Benjamins, 249–78.
Kamp, Hans (1981). ‘A theory of truth and semantic representation’, in J. A. G. Groenendijk,
T. M. V. Janssen, and M. B. J. Stokhof (eds.), Formal methods in the study of language.
Amsterdam: Mathematical Centre Tracts, 277–322.
References 267

Kamp, Hans and Uwe Reyle (1993). From discourse to logic. Dordrecht: Kluwer.
Kamp, Hans and Christian Rohrer (1983). ‘Tense in texts’, in R. Bauerle, C. Schwarze, and
A. von Stechow (eds.), Meaning, use and interpretation of language. Berlin/New York:
Mouton de Gruyter, 250–69.
Kamp, J. A. W. (1971). ‘Formal properties of “now” ’, Theoria 37: 227–73.
Kant, Immanuel (1929 [1781–1787]). Critique of pure reason, trans. Norman Kemp Smith.
London: Macmillan.
Kaplan, David (1977). ‘Demonstratives’. Unpublished ms, UCLA. Reprinted in J. Almog,
J. Perry, and H. Wettstein (eds.) (1989) Themes from Kaplan. Ithaca: Cornell University
Press, 481–564.
Kaplan, David (1979). ‘On the logic of demonstratives’, Journal of Philosophical Logic 8: 81–98.
Kaplan, David (1990). ‘Thoughts on demonstratives’, in P. Yourgrau (ed.), Demonstratives.
Oxford: Oxford University Press, 34–49.
Kaufmann, Stefan, Cleo Condoravdi, and Valentina Harizanov (2006). ‘Formal approaches
to modality’, in W. Frawley (ed.), The expression of modality. Berlin: Mouton/Hillsdale, NJ:
Erlbaum, 71–106.
Kelly, Sean D. (2005a). ‘Temporal awareness’, in D. W. Smith and A. L. Thomasson (eds.),
Phenomenology and philosophy of mind. Oxford: Clarendon Press, 222–34.
Kelly, Sean D. (2005b). ‘the puzzle of temporal experience’, in A. Brook and K. Akins (eds.),
Cognition and the brain: the philosophy and neuroscience movement. Cambridge: Cam-
bridge University Press, 208–38.
Kemmerer, David (1999). ‘ “Near” and “far” in language and perception’, Cognition 73: 35–63.
Kemmerer, David (2010). ‘A neuroscientific perspective on the linguistic encoding of categor-
ical spatial relations’, in V. Evans and P. Chilton (eds.), Language, cognition and space.
London: Equinox, 139–68.
Kim, Jaegwon (1976). ‘Events as property exemplifications’, in M. Brand and D. Walton (eds.),
Action theory. Dordrecht: Reidel, 159–77. [Reprinted in J. Kim (1993) Supervenience and
mind: selected philosophical essays. Cambridge: Cambridge University Press, 33–52.]
Kleiber, Georges (1983). ‘L’emploi sporadique de pouvoir’, in J. David and G. Kleiber (eds.), La
notion sémantico-logique de modalité. Metz: Centre d’Analyse Syntaxique de l’Université de
Metz, 183–203.
Kleiber, Georges (1987). Du côté de la référence verbale. Les phrases habituelles. Berne: Peter Lang.
Kleiber, Georges and Anne-Marie Berthonneau (2003). ‘Un imparfait de plus . . . et le train
déraillait’, Cahiers Chronos 11: 1–24.
Klein, Wolfgang (1992). ‘The present perfect puzzle’, Language 68(3): 525–52.
Klein, Wolfgang (1994). Time in language. London: Routledge.
Kratzer, Angelika (1977). ‘What “must” and “can” must and can mean’, Linguistics and
Philosophy 1: 337–55. [Revised version in A. Kratzer (2012). Modals and conditionals.
Oxford: Oxford University Press, 1–26.]
Kratzer, Angelika (1981). ‘The notional category of modality’, in H.-J. Eikmeyer and H. Rieser
(eds.), Words, worlds and contexts—new approaches to word semantics. Berlin/New York:
W. de Gruyter, 38–74. [Reprinted in P. Portner and B. H. Partee (eds.) (2002). Formal
semantics: the essential readings. Oxford: Blackwell, 289–323; revised version in A. Kratzer
(2012). Modals and conditionals. Oxford: Oxford University Press, 27–69.]
Kratzer, Angelika (2012). Modals and conditionals. Oxford: Oxford University Press.
268 References

Kronning, Hans (1996). Modalité, cognition et polysémie: sémantique du verbe modal devoir.
Uppsala/Stockholm: Acta Universitatis Uppsaliensis and Almqvist & Wiksell International.
Kronning, Hans (2001). ‘Nécessité et hypothèse: devoir non déontique au conditionnel’, in
P. Dendale and L. Tasmowski (eds.), Le conditionnel en français. Metz: Centre d’études
linguistiques des textes et des discours de l’Université de Metz, 251–76.
Kronning, Hans (2003). ‘Auxiliarité, énonciation et rhématicité’, Cahiers Chronos 11: 231–49.
Kronning, Hans (2004). ‘Modalité et médiation épistémiques’, in R. Delamotte-Legrand (ed.),
Les médiations langagières. Des faits de langue aux discours. Mont Saint-Aignan: Publica-
tions de l’Université de Rouen, 35–66.
Kuhn, Steven T. and Paul Portner (2002). ‘Tense and time’, in D. Gabbay and F. Guenthner
(eds.), Handbook of philosophical logic. Vol. 7. Dordrecht: Kluwer, 277–346.
Laca, Brenda (2005). ‘Périphrases aspectuelles et temps grammatical dans les langues ro-
manes’, in H. Bat-Zeev Shyldkrot and N. Le Querler (eds.), Les périphrases verbales.
Amsterdam: Benjamins, 47–66.
Lagerqvist, Hans (2009). Le subjonctif en français moderne. Esquisse d’une théorie modale.
Paris: P.U.P.S.
Lakoff, George (1987). Women, fire, and dangerous things. Chicago, IL: Chicago University Press.
Lakoff, George and Mark Johnson (1980). Metaphors we live by. Chicago, IL: Chicago
University Press.
van Lambalgen, Michiel and Fritz Hamm (2005). The proper treatment of events. Oxford:
Lamiroy, Béatrice (1983). Les verbes de mouvement en français et en espagnol (Lingvisticae
Investigationes: Supplementa 11). Amsterdam: Benjamins.
Lamiroy, Béatrice (1999). ‘Auxiliaires, langues romanes et grammaticalisation’, Langages 135: 33–45.
Landman, Fred (2000). Events and plurality: the Jerusalem lectures. Dordrecht: Kluwer.
Langacker, Ronald W. (1987). Foundations of Cognitive Grammar, Vol. 1: theoretical prerequis-
ites. Stanford, CA: Stanford University Press.
Langacker, Ronald W. (1991). Foundations of Cognitive Grammar, Vol. 2: descriptive applica-
tion. Stanford, CA: Stanford University Press.
Langacker, Ronald W. (1995). ‘The symbolic alternative’, in H. Kardela and G. Persson (eds.),
New trends in semantics and lexicography. Umeå: Swedish Science Press, 89–118.
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:
Langacker, Ronald W. (2002a). ‘Deixis and subjectivity’, in F. Brisard (ed.), Grounding: the
epistemic footing of deixis and reference. Berlin: Mouton de Gruyter, 1–28.
Langacker, Ronald W. (2002b). ‘Remarks on the English grounding systems’, in F. Brisard (ed.),
Grounding: the epistemic footing of deixis and reference. Berlin: Mouton de Gruyter, 29–38.
Langacker, Ronald W. (2011). ‘The English present: temporal coincidence vs. epistemic
immediacy’, in A. Patard and F. Brisard (eds.), Cognitive approaches to tense, aspect, and
epistemic modality. Amsterdam: Benjamins, 45–86.
Langacker, Ronald W. (2012). ‘Linguistic manifestations of the space–time (dis)analogy’, in
L. Filipović and K. M. Jaszczolt (eds.), Space and time in languages and cultures: language,
culture, and cognition. Amsterdam/Philadelphia: Benjamins, 191–215.
References 269

Larson, Richard K. and Peter Ludlow (1993). ‘Interpreted logical forms’, Synthese 96: 305–55.
Le Poidevin, Robin (2007). The images of time: an essay on temporal representation. Oxford:
Oxford University Press.
Le Poidevin, Robin (2011). ‘The temporal prison’, Analysis 71: 456–65.
Le Poidevin, Robin and Murray MacBeath (eds.) (1993). The philosophy of time. Oxford:
Oxford University Press.
Lepore, Ernie and Barry Loewer (1987). ‘Dual aspect semantics’, in: E. LePore (ed.),
New directions in semantics. London: Academic Press, 83–112.
Levinson, Steven C. (2000). Presumptive meaning: the theory of generalized conversational
implicature. Cambridge, MA: MIT Press.
Lewis, David (1986). On the plurality of worlds. Oxford: Blackwell.
Lonergan, Bernard. (1957). Insight: a study of human understanding. London: Longmans,
Green & Co.
Ludlow, Peter (1999). Semantics, tense, and time: an essay in the metaphysics of natural
language. Cambridge, MA: MIT Press.
Ludlow, Peter (2000). ‘Interpreted logical forms, belief attribution, and the dynamic lexicon’,
in K. M. Jaszczolt (ed.), The pragmatics of propositional attitude reports. Oxford: Elsevier
Science, 31–42.
Ludlow, Peter (2006). ‘The myth of language’, Croatian Journal of Philosophy 18: 385–400.
Ludlow, Peter (2007). ‘Cheap contextualism’, in E. Sosa and E. Villanueva (eds.), Noûs.
Philosophical Issues 16: Annual Supplement, 104–29.
Ludlow, Peter (2008). ‘Understanding temporal indexicals’, in M. O’Rourke and C. Washing-
ton (eds.), Situating semantics: essays on the philosophy of John Perry. Cambridge, CA: MIT
Press, 155–78.
Łukasiewicz, Jan (1961). ‘O determinizmie’, in J. Słupecki (ed.), Jan Łukasiewicz. Z zagadnień
logiki i filozofii. Pisma wybrane. Warszawa: Państwowe Wydawnictwo Naukowe, 114–26.
[Trans. Z. Jordan as ‘On determinism’, in L. Borkowski (ed.) (1970). Jan Łukasiewicz:
selected works. Amsterdam: North-Holland, 110–28.]
Luo, Zhaohui (2010). ‘Type-theoretical semantics with coercive subtyping’, in N. Li and
D. Lutz (eds.), Proceedings of SALT 20. Ithaca, NY: Cornell University Press, 38–56.
Maclaurin, James and Heather Dyke (2002). ‘ “Thank goodness that’s over”: the evolutionary
story’, Ratio 15: 276–92.
Maienborn, Claudia (2005). ‘On the limits of the Davidsonian approach’, Theoretical Linguis-
tics 31: 275–316.
Malchukov, Andrej (2009). ‘Incompatible categories: resolving the “present perfective
paradox” ’, in L. Hogeweg, H. de Hoop, and A. Malchukov (eds.), Cross-linguistic semantics
of tense, aspect, and modality. Amsterdam: Benjamins, 13–32.
Mani, Inderjeet, James Pustejovsky, and Robert Gaizauskas (eds.) (2005). The language of time:
a reader. Oxford: Oxford University Press.
Mari, Alda (2012). ‘Pouvoir au passé composé: effet épistémique et lecture habilitative’, in L. de
Saussure and A. Rihs (eds.), Etudes de sémantique et pragmatique françaises. Berne: Peter
Lang, 67–99.
Martin, Robert (1971). Temps et aspect. Essai sur l’emploi des temps narratifs en moyen français.
Paris: Klincksieck.
270 References

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.
Matthewson, Lisa (2006). ‘Temporal semantics in a superficially tenseless language’, Linguis-
tics and Philosophy 29: 673–713.
McCawley, James D. (1971). ‘Tense and time reference in English’, in C. J. Fillmore and
D. T. Langendoen (eds.), Studies in linguistic semantics. New York: Holt, Rinehart and
Winston, 96–113.
McDowell, John (1980). ‘On the sense and reference of a proper name’, in M. Platts (ed.),
Truth, reality and reference. London: Routledge and Keegan Paul, 111–30.
McLaren-Hankin, Yvonne (2008). ‘We expect to report on significant progress in our product
pipeline in the coming year: hedging forward-looking statements in corporate press
releases’, Discourse Studies 10: 635–54.
McLaughlin, Brian and Karen Bennett (2005). ‘Supervenience’, in E. Zalta (ed.), Stanford
encyclopedia of philosophy. Available at: <>.
McTaggart, John M. E. (1908). ‘The unreality of time’, Mind 17(4): 457–74. [Reprinted in
J. M. E. McTaggart (1934). Philosophical studies. London: E. Arnold, 110–31.]
McTaggart, John M. E. (1927). The nature of existence, vol. II. Cambridge: Cambridge
University Press.
Mellor, D. H. (1998). Real time II. London: Routledge.
Merlini, Lavinia (1983). Gli atti del discorso economico. La previsione status illocutorio e modelli
linguistici nel testo inglese. Parma: Edizioni Zara, Università di Parma.
Merricks, Trenton (1995). ‘On the incompatibility of enduring and perduring entities’, Mind
104: 523–31.
Merricks, Trenton (1999). ‘Persistence, parts, and presentism’, Noûs 33: 421–38.
ter Meulen, Alice G. B. (1994). ‘Dynamic aspect trees’ (with J. Seligman), in L. Polos and
M. Masuch (eds.), Applied logic: how, what and why. Logical approaches to natural language
(Synthese Library, vol. 247). Dordrecht: Kluwer, 287–320.
ter Meulen, Alice G. B. (1995). Representing time in natural language: the dynamic interpret-
ation of tense and aspect. Cambridge, MA: MIT Press.
ter Meulen, Alice G. B. (2000). ‘Chronoscopes: the dynamic representation of facts and
events’, in J. Higginbotham, F. Pianesi, and A. Varzi (eds.), Speaking of events. New
York/Oxford: Oxford University Press, 151–68.
ter Meulen, Alice G. B. (2005). ‘La quantification dans les contextes temporels: la logique des
adverbes aspectuels’, in P. Joray (ed.), La quantification dans la logique moderne (coll.
Epistémologie et Philosophie des Sciences). Paris: L’Harmattan, 343–63.
Michaelis, Laura A. (1998). Aspectual grammar and past time reference. London: Routledge.
Michaelis, Laura A. (2004). ‘Type shifting in Construction Grammar: an integrated approach
to aspectual coercion’, Cognitive Linguistics 15: 1–67.
Michaelis, Laura A. (2005). ‘Entity and event coercion in a symbolic theory of syntax’, in
J.-O. Östman and M. Fried (eds.), Construction Grammars: cognitive grounding and
theoretical extensions. Amsterdam: Benjamins, 45–88.
Michaelis, Laura A. (2006). ‘Time and tense’, in B. Aarts and A. McMahon (eds.), The
handbook of English linguistics. Oxford: Blackwell, 220–34.
References 271

Miecznikowski, Johanna and Carla Bazzanella (2007). ‘The attenuative conditional: context,
appropriateness and interaction’, in A. Fetzer (ed.), Context and appropriateness: micro
meets macro. Amsterdam/Philadelphia: Benjamins, 203–33.
Miecznikowski, Johanna, Andrea Rocci, and Gergana Zlatkova (2012). ‘L’argumentation dans
la presse économique et financière italienne’, in L. Gautier (ed.), Les discours de la bourse et
de la finance. Berlin: Frank und Timme, 65–83.
Moens, Marc and Mark Steedman (1988). ‘Temporal ontology and temporal reference’,
Computational Linguistics 14(2), 15–27.
Moeschler, Jacques (1998). ‘Les relations entre événements et l’interprétation des énoncés’, in
Jacques Moeschler (ed.), Le temps des événements. Paris: Kimé, 293–321.
Molendijk, Arie (1993). ‘Présuppositions, implications, structure temporelle’, in C. Vetters
(ed.), Le temps: de la phrase au texte. Lille: Presses Universitaires de Lille, 167–90.
Molendijk, Arie (2005). ‘The imparfait of French and the past progressive of English’, in
B. Hollebrandse, A. van Hout and C. Vet (eds.), Crosslinguistic perspectives on tense, aspect
and modality: selected papers from Chronos 5, Groningen 2002. Amsterdam: Rodopi, 119–29.
Molendijk, Arie (2010). ‘La simultanéité globale neutre et la simultanéité progressive en
français et en anglais’, in N. Flaux, D. Stosic, and C. Vet (eds.), Interpréter les temps
verbaux. Berne: Peter Lang, 63–82.
Molendijk, Arie and Co Vet (1995). ‘Interprétation, référence et cohésion’, Sémiotiques 9:
Nederlandse Taalunie (2004). Corpus Gesproken Nederlands, Version 2.0. Leiden: TST-
Centrale INL.
Nicolle, Steve (1996). ‘Conceptual and procedural encoding in relevance theory: a study with
reference to English and Kiswahili’, unpublished dissertation, University of York.
Nølke, Henning (2001). Le regard du locuteur 2. Paris: Kimé.
Nølke, Henning, Kjersti Fløttum, and Coco Norén (2004). ScaPoline. La théorie scandinave de
la polyphonie linguistique. Paris: Kimé.
Nunberg, Geoffrey (1995). ‘Transfers of meaning’, Journal of Semantics 12: 109–32.
Nuyts, Jan (1992). Aspects of a cognitive-pragmatic theory of language. Amsterdam: Benjamins.
Nuyts, Jan (2001). Epistemic modality, language, and conceptualization. Amsterdam:
Nuyts, Jan (2005a). ‘The modal confusion’, in A. Klinge and H. Müller (eds.), Modality.
London: Equinox, 5–38.
Nuyts, Jan (2005b). ‘Brothers in arms?’, in F. Ruiz de Mendoza and S. Peña (eds.), Cognitive
linguistics. Berlin: Mouton, 69–100.
Nuyts, Jan (2006). ‘Modality’, in W. Frawley (ed.), The expression of nodality. Berlin: Mouton,
Nuyts, Jan (2007). ‘Cognitive linguistics and functional linguistics’, in D. Geeraerts and
H. Cuyckens (eds.), Oxford handbook of cognitive linguistics. Oxford: Oxford University
Press, 543–65.
Nuyts, Jan (2008a). ‘Qualificational meanings, illocutionary signals, and the cognitive
planning of language use’, Annual Review of Cognitive Linguistics 6: 185–207.
Nuyts, Jan (2008b). ‘Pattern vs. process concepts of grammar and mind’, Jezikoslovlje 9:
272 References

Nuyts, Jan (2009a). ‘The “one-commitment-per-clause” principle and the cognitive status of
qualificational categories’, Linguistics 47: 141–71.
Nuyts, Jan (2009b). ‘Language, conceptualization, and TAM marking: a cognitive-functional
perspective’, Wai guo yu [Journal of Foreign Languages, Shanghai] 32(1): 2–43.
Nuyts, Jan (2011). ‘Pattern versus process concepts of grammar and mind: a cognitive-
functional perspective’, in M. Brdar, S. T. Gries, and M. Žic Fuchs (eds.), Cognitive
linguistics. Amsterdam: Benjamins, 47–66.
Nuyts, Jan (forthcoming a). ‘Analyses of the modal meanings’, in J. Nuyts and J. van der
Auwera (eds.), The Oxford handbook of mood and modality. Oxford: Oxford University
Nuyts, Jan (forthcoming b). Modal and other qualifications in language and thought.
Nuyts, Jan, Pieter Byloo, and Janneke Diepeveen (2010). ‘On deontic modality, directivity, and
mood’, Journal of Pragmatics 42: 16–34.
Oaklander, L. Nathan (2002). ‘Presentism, ontology and temporal experience’, in C. Callender
(ed.), Time, reality and experience. Cambridge: Cambridge University Press, 73–90.
Oaklander, L. Nathan and V. Alan White (2007). ‘B-time: a reply to Tallant’, Analysis 67:
O’Donnell, Mick (2008). ‘The UAM CorpusTool: software for corpus annotation and explor-
ation’, in La lingüística aplicada actual/Applied Linguistics Now. Proceedings of the XXVI
Congreso De AESLA: AESLA, 1433–47. Available online at
Øhrstrøm, Peter and Per F. V. Hasle (1995). Temporal logic: from ancient ideas to artificial
intelligence. Dordrecht: Kluwer.
Olson, Eric T. (2009). ‘The rate of time’s passage’, Analysis 69: 3–9.
Pancheva, Roumyana and Armin von Stechow (2004). ‘On the present perfect puzzle’, NELS
34: 469–83.
Papafragou, Anna (2000). Modality: issues in the semantics–pragmatics interface. Amsterdam:
Papafragou, Anna (2006). ‘Epistemic modality and truth-conditions’, Lingua 116: 1688–702.
Parsons, Josh (2000). ‘Must a four-dimensionalist believe in temporal parts?, Monist 83:
Parsons, Josh (2002). ‘A-theory for B-theorists’, Philosophical Quarterly 52: 1–20.
Parsons, Josh (2003). ‘A-theory for tense logicians’, Analysis 63: 4–6.
Parsons, Terence (1990). Events in the semantics of English: a study in subatomic semantics.
Cambridge, MA: MIT Press.
Partee, Barbara H. (1973). ‘Nominal and temporal anaphora’, Linguistics and Philosophy 7:
Partee, Barbara H., Alice G. B. ter Meulen, and Robert E. Wall (1993). Mathematical methods
in linguistics. Dordrecht: Kluwer.
Patard, Adeline (2007). ‘L’un et le multiple. L’imparfait de l’indicatif en français. Valeur en
langue et usages en discours’, Ph.D. dissertation, Université Paul-Valery-Montpellier.
Paul, Laurie A. (2010). ‘Temporal experience’, Journal of Philosophy 107: 333–59.
Perry, John (1977). ‘Frege on demonstratives’, Philosophical Review 86: 474–97.
Perry, John (1979). ‘The problem of the essential indexical’, Noûs 13: 3–21.
References 273

Perry, John (1986). ‘Thought without representation’, Aristotelian Society Supplementary

Volume 60: 137–52.
Perry, John (1992). The problem of the essential indexical and other essays. Oxford: Oxford
University Press.
Perry, John (1997). ‘Rip van Winkle and other characters’, European Review of Philosophy 2:
13–40. [Reprinted in John Perry (2000). The problem of the essential indexical. Stanford, CA:
CSLI, Chapter 21.]
Perry, John (2000). The problem of the essential indexical. Stanford, CA: CSLI.
Perry, John (2001). Reference and reflexivity. Stanford, CA: CSLI.
Person, Cédric (2004). ‘Traitement automatique de la temporalité du récit: implementation du
modèle linguistique SdT’, Ph.D. dissertation, Université de Caen-Basse-Normandie.
Phillips, I. (2010). ‘Perceiving temporal properties’, European Journal of Philosophy 18:
Pickering, Martin J., Brian McElree, Steven Frisson, Lillian Chen, and Matthew Traxler (2006).
‘Underspecification and aspectual coercion’, Discourse Processes 42(2): 131–55.
Pietrandrea, Paola (2005). Epistemic modality. Functional properties and the Italian system.
Amsterdam/Philadelphia: Benjamins.
Pietrandrea, Paola (2007). ‘The grammatical nature of some epistemic–evidential adverbs in
Italian’, Italian Journal of Linguistics 19(1): 39–63.
Portner, Paul (2009). Modality. Oxford: Oxford University Press.
Prior, Arthur N. (1957). Time and modality. Oxford: Clarendon Press.
Prior, Arthur N. (1959). ‘Thank goodness that’s over’, Philosophy 34: 12–17.
Prior, Arthur N. (1967). Past, present and future. Oxford: Clarendon Press.
Prior, Arthur N. (1968). Papers on time and tense. Oxford: Clarendon Press. [Reprinted (2003)
as Papers on time and tense: new edition. Oxford: Oxford University Press.]
Prior, Arthur N. (2003). Papers on time and tense: new edition. Oxford: Oxford University
Prosser, Simon (2006). ‘Temporal metaphysics in Z-Land’, Synthese 149: 77–96.
Prosser, Simon (2007). ‘Could we experience the passage of time?’, Ratio 20: 75–90. [Reprinted
in L. Nathan Oaklander (ed.) (2008) Philosophy of time: critical concepts in philosophy,
vol. 3. New York/London: Routledge.
Prosser, Simon (2012). ‘Why does time seem to pass?’, Philosophy and Phenomenological
Research, 85: 92–116.
Prosser, Simon (2013). ‘Passage and perception’, Noûs 47: 69–84.
Pustejovsky, James (1995). The generative lexicon. Cambridge, MA: MIT Press.
Recanati, François (1989). ‘The pragmatics of what is said’, Mind and Language 4: 295–329.
[Reprinted in S. Davis (ed.), (1991). Pragmatics: a reader. Oxford: Oxford University Press,
Recanati, François (1995). ‘Le présent épistolaire: une perspective cognitive’, L’information
grammaticale 66: 38–44.
Recanati, François (2004). Literal meaning. Cambridge: Cambridge University Press.
Recanati, François (2006). Perspectival thought: a plea for (moderate) relativism. Oxford:
Oxford University Press.
Recanati, François (2010). Truth-conditional pragmatics. Oxford: Clarendon Press.
274 References

Reichenbach, Hans-Georg (1947). Elements of symbolic logic. New York: Macmillan.

Reinhart, Tanya (1986). ‘Principes de perception des formes et organisation temporelle des
textes narratifs’, trans. P. Pica, Recherches linguistiques de Vincennes 14/15: 45–92.
Reyle, Uwe, Antje Rossdeutscher, and Hans Kamp (2007). ‘Ups and downs in the theory of
temporal reference’, Linguistics and Philosophy 30: 565–635.
Richardt, Susanne (2003). ‘Metaphors in expert and common-sense reasoning’, in C. Zelinsky-
Wibbelt (ed.), Text, context, concepts. Berlin/New York: Mouton de Gruyter, 243–98.
Riegel, Martin, Jean-Christophe Pellat, and René Rioul (1994). Grammaire méthodique du
français, Paris: PUF.
Rigotti, Eddo and Sara Greco Morasso (2010). ‘Comparing the argumentum model of topics
to other contemporary approaches to argument schemes: the procedural and material
components’, Argumentation 24(4): 489–512.
Roberts, Craig (1989). ‘Modal subordination and pronominal anaphora in discourse’, Linguis-
tics and Philosophy 12(6): 683–721.
Rocci, Andrea (2000). ‘L’interprétation épistémique du futur en italien et en français: une
analyse procédurale’, Cahiers de linguistique française 22: 241–74.
Rocci, Andrea (2005a). ‘Epistemic readings of modal verbs in Italian: the relationship
between propositionality, theme–rheme articulation and inferential discourse relations’,
in B. Hollebrandse, A. van Hout and C. Vet (eds.), Crosslinguistic views on tense, aspect and
modality. Amsterdam/New York: Rodopi, 229–46.
Rocci, Andrea (2005b). La modalità epistemica tra semantica e argomentazione. Milano:
Pubblicazioni dell’ISU-Università Cattolica.
Rocci, Andrea (2006). ‘Le modal italien dovere au conditionnel: évidentialité et contraintes sur
l’inférence des relations de discours argumentatives’, in L. de Saussure (ed.), Temps,
description, interprétation. Travaux neuchâtelois de linguistique (TRANEL), 45: 71–98.
Rocci, Andrea (2007). ‘Epistemic modality and questions in dialogue. The case of Italian
interrogative constructions in the subjunctive mood’, in L. de Saussure, J. Moeschler,
and G.Puska (eds.), Tense, mood and aspect: Theoretical and descriptive issues.
Amsterdam/New York: Rodopi, 129–53.
Rocci, Andrea (2008). ‘Modality and its conversational backgrounds in the reconstruction of
argumentation’, Argumentation 22(2): 165–89.
Rocci, Andrea (2011). ‘The Italian modal dovere in the conditional: Future reference, evidenti-
ality and argumentation’, in T. Mortelmans, J. Mortelmans, and W. De Mulder (eds.), In the
mood for mood. Amsterdam/New York: Rodopi, 61–79.
Rocci, Andrea (in preparation). Modality in argumentation.
Rocci, Andrea and Gergana Zlatkova (in press). ‘The Italian epistemic modal adverb forse:
from semantics to argumentation’, in S. Cantarini, G. Gobber, and F. Hundsnuscher (eds.),
Dialogue. Studies in memory of Sorin Stati. München: LINCOM Academic Publishers.
Rödl, Sebastian (2007). Self-consciousness. Cambridge, MA: Harvard University Press.
Rothstein, Susan (2004). Structuring events: a study in the semantics of lexical aspect. Oxford:
Roy, Gérard-Raymond (1976). Contribution à l’analyse du syntagme verbal: étude morpho-
syntaxique et statistique des coverbes. Paris: Klincksieck.
References 275

Rumfitt, Ian (1993). ‘Content and context: the paratactic theory revisited and revised’, Mind
102: 429–53.
van der Sandt, Rob (1992). ‘Presupposition projection as anaphora resolution’, Journal of
Semantics 9: 333–77.
Sasse, Hans-Jürgen (1991). ‘Aspect and Aktionsart: a reconciliation’, in C. Vetters and
W. Vandeweghe (eds.), Perspectives on aspect and Aktionsart. Amsterdam: Benjamins, 31–45.
Sattig, Thomas (2006). The language and reality of time. Oxford: Clarendon Press.
Sauerland, Uli (2002). ‘The present tense is vacuous’, Snippets 6. Available at: <http://
de Saussure, Louis (2003). Temps et pertinence. Eléments de pragmatique cognitive du temps.
Bruxelles: De Boeck.
de Saussure, Louis (2008). ‘Maintenant: présent cognitif et enrichissement pragmatique’, in
M. Vuillaume (ed.), Ici et maintenant. Amsterdam/New York: Rodopi, 53–76.
de Saussure, Louis (2010). ‘Pragmatique procédurale des temps verbaux: la question des usages
interprétatifs’, in N. Flaux, D. Stosic, and C. Vet (eds.), Interpréter les temps verbaux. Berne:
Peter Lang, 129–59.
de Saussure, Louis (2011a). ‘On some methodological issues in the conceptual/procedural
distinction’, in V. Escandell-Vidal, M. Leonetti, and A. Ahern (eds.), Procedural meaning.
Problems and perspectives. Bingley: Emerald (CRISPI 25), 55–79.
de Saussure, Louis (2011b). ‘Le passé composé en usage futur et la pertinence au présent’,
Journal of French Language Studies 22(2): 189–206.
de Saussure, Louis (2012). ‘Modalité épistémique, évidentialité et dépendance contextuelle’,
Langue française 173: 131–43.
de Saussure, Louis and Patrick Morency (2012). ‘A cognitive-pragmatic view of the French
epistemic future’, Journal of French Language Studies 22: 207–23.
de Saussure, Louis and Bertrand Sthioul (1999). ‘L’imparfait narratif: point de vue (et images
du monde)’, Cahiers de praxématique 32: 167–88.
de Saussure, Louis and Bertrand Sthioul (2005). ‘Imparfait et enrichissement pragmatique’, in
P. Larrivée and E. Labeau (eds.), Nouveaux développements de l’imparfait. Amsterdam:
Rodopi, 103–20.
Schein, Barry (2002). ‘Events and the semantic content of thematic relations’, in G. Preyer and
G. Peter (eds.). Logical form and language. Oxford: Clarendon Press, 263–344.
Schlesinger, George N. (1991). ‘E pur si muove’, Philosophical Quarterly 41: 427–41.
Searle, John R. (2005). ‘What is an institution?’, Journal of Institutional Economics 1(1):
Searle, John R. and Daniel Vanderveken (1985). Foundations of illocutionary logic. Cambridge:
Cambridge University Press.
Seligman, Jeremy and Lawrence S. Moss (2011). ‘Situation theory’, in Johan van Benthem and
Alice G. B. ter Meulen (eds.), Handbook of logic and language, 2nd edn. Berlin: Springer,
Sider, Theodore (2001). Four-dimensionalism. Oxford: Oxford University Press.
Sklar, Lawrence (1977). Space, time, and spacetime. Berkeley, CA: University of California Press.
Sklar, Lawrence (1981). ‘Up and down, left and right, past and future’, Noûs 15: 111–29.
276 References

Skow, Bradford (2009). ‘Relativity and the moving spotlight’, Journal of Philosophy 106: 666–78.
Smart, J[ohn] J. C. (1967). ‘Time’, in P. Edwards (ed.), The encyclopedia of philosophy, vol. 8.
New York: Macmillan, 126–34.
Smessaert, Hans and Alice G. B. ter Meulen (2004). ‘Temporal reasoning with aspectual
adverbs’, Linguistics and Philosophy 27(2): 209–61.
Smith, Carlota (1991). The parameter of aspect. Dordrecht: Kluwer.
Smith, Quentin (1993). Language and time. Oxford: Oxford University Press.
Smith, Quentin (2002). ‘Time and degrees of existence: a theory of “degree presentism” ’, in
C. Callender (ed.), Time, reality and experience. Cambridge: Cambridge University Press,
Sperber, Dan and Deirdre Wilson (1995). Relevance: communication and cognition, 2nd ed.
Oxford: Blackwell.
Squartini, Mario (2004). ‘Disentangling evidentiality and epistemic modality in Romance’,
Lingua 114: 873–95.
Squartini, Mario (2008). ‘Lexical vs. grammatical evidentiality in French and Italian’, Linguis-
tics 46: 917–47.
Srioutai, Jiranthara (2006). ‘Time conceptualization in Thai with special reference to D1ay1II,
Kh3oe:y, K1aml3ang, Y3u:I and C1a’, Ph.D. thesis, University of Cambridge.
Steedman, Mark (1997). ‘Temporality’, in J. van Benthem and A. ter Meulen (eds.), Handbook
of logic and language. Amsterdam: Elsevier Science, 895–938.
Sthioul, Bertrand (1998a). ‘Le passé composé: une approche instructionnelle’, in S. Vogeleer,
A. Borillo, C. Vetters, and M. Vuillaume (eds.), Temps et discours. Louvain-la-Neuve:
Peeters, 79–94.
Sthioul, Bertrand (1998b). ‘Temps verbaux et points de vue’, in J. Moeschler, J. Jayez,
M. Koslowska, J.-M. Luscher, B. Sthioul, and L. de Saussure (eds.), Le temps des événements.
Paris: Kimé, 197–220.
Sundell, Lars-Göran (1991). Le temps futur en français moderne. Stockholm: Almqvist &
Tallant, Jonathan (2007). ‘What is B-time?’Analysis 67: 147–56.
Tallant, Jonathan (2011). ‘There’s no future in no-futurism’, Erkenntnis 74: 37–52.
Tasmowski, Liliane and Patrick Dendale (1998). ‘Must/will and doit/futur simple as epistemic
modal markers. Semantic value and restrictions of use’, in J. E. van der Auwera (ed.), English
as a human language. To honour Louis Goossens. München: Lincom Europa, 325–36.
Tasmowski-de Ryck, Liliane (1985). ‘L’imparfait avec et sans rupture’, Langue française 67:
Taylor, Barry (1985). Modes of occurrence: verbs, adverbs and events. Oxford: Basil Blackwell.
Thomason, Richmond H. (1984). ‘Combinations of tense and modality’, in D. M. Gabbay and
F. Guenthner (eds.), Handbook of philosophical logic: extensions of classical logic, vol. II.
Dordrecht: Reidel, 135–65.
Thomason, Richmond H. (2002). ‘Combinations of tense and modality’, in D. Gabbay and
F. Guenthner (eds.), Handbook of philosophical logic, 2nd edn, vol. 7. Dordrecht: Kluwer,
Thornton, Agathe (1987). Maori oral literature as seen by a classicist. Dunedin: University of
Otago Press.
References 277

Tonhauser, Judith (2006). ‘The temporal semantics of noun phrases: evidence from Gurani’,
unpublished dissertation, Stanford University.
Tooley, Michael (1997). Time, tense, and causation. Oxford: Clarendon Press.
Tooley, Michael (1999). ‘The metaphysics of time’, in J. Butterfield (ed.), The arguments of
time. Oxford: Oxford University Press, 21–42.
Toulmin, Stephen E. (2003 [1958]). The uses of argument. Cambridge: Cambridge University
Touratier, Christian (1996). Le système verbal français. Paris: Armand Colin.
Tournadre, Nicolas (2004). ‘Typologie des aspects verbaux et intégration à une théorie du
TAM’, Bulletin de la Société Linguistique de Paris 99(1): 7–68.
Tye, M. (2003). Consciousness and persons. Cambridge, MA: MIT Press.
Ullman, Michael T., Suzanne Corkin, Marie Coppola, Gregory Hickok, John H. Growdon,
Walter J. Koroshetz, and Steven Pinker (1997). ‘A neural dissociation within language:
evidence that the mental dictionary is part of declarative memory, and that grammatical
rules are processed by the procedural system’, Journal of Cognitive Neuroscience 9(2):
Velleman, J. David (2006). ‘So it goes’, Amherst lecture in philosophy, 2006. Available at:
Vendler, Zeno (1967). Linguistics in philosophy. Ithaca, NY: Cornell University Press.
Verkuyl, Henk (1972). On the compositional nature of the aspects. Dordrecht: Kluwer.
Vet, Co (1981). ‘La notion de “monde possible” et le système temporel et aspectuel du français’,
Langages 64: 109–24.
Vet, Co (1992). ‘Le passé composé, contextes d’emploi et interprétations’, Cahiers de prax-
ématique 19: 37–59.
Vet, Co (1993). ‘Conditions d’emploi et interprétation des temps futurs du français’. Verbum
1993(4): 71–84.
Vet, Co (2001). ‘Deux cas de polysémie: le passé composé et le futur périphrastique’, in
H. Kronning, C. Norén, B. Novén, G. Ransbo, L.-G. Sundell, and B. Svane (eds.), Langage
et référence (Mélanges offerts à K. Jonasson). Uppsala: Acta Universitatis Upsaliensis,
Vet, Co (2002). ‘Les adverbes de temps: décomposition lexicale et coercion’, in B. Laca (ed.),
Temps et aspect. De la morphologie à l’interprétation. Saint-Denis: P.U. de Vincennes,
Vetters, Carl (2003). ‘L’aspect global: un effet secondaire d’un contenu procédural?’, Cahiers
Chronos 11: 113–31.
Vetters, Carl (2010). ‘Développement et évolution des temps du passé en français: passé
simple, passé composé et venir de þ infintif’, Cahiers Chronos 21: 277–98.
Vetters, Carl and Elzbieta Skibinska (1998). ‘Le futur: une question de temps ou de mode?
Remarques générales et analyse du “présent-futur” perfectif polonais’, Cahiers Chronos 2:
Victorri, Bernard and Catherine Fuchs (1996). La polysémie. Construction dynamique du sens.
Paris: Hermès.
Vlach, Frank (1981). ‘La sémantique du temps et de l’aspect en anglais’, Langages 64: 65–79.
278 References

Vuillaume, Marcel (2000). ‘Heureusement que Pierre n’est pas venu demain!’, Cahiers
Chronos 6: 107–16.
Vuillemin, Jules (1984). Nécessité ou contingence. Paris: Minuit.
Walsh, Polly (2004). ‘Investigating prediction in financial and business news articles’, in
R. Facchinetti and F. R. Palmer (eds.), English modality in perspective: genre analysis and
contrastive studies. Frankfurt am Main: Peter Lang, 83–100.
Walsh, Polly (2006). ‘Playing safe? A closer look at hedging, conditions and attribution in
economic forecasting’, in V. K. Bhatia and M. Gotti (eds.), Explorations in specialized
genres. Bern: Peter Lang, 135–54.
Weinrich, Harald (1964). Tempus. Besprochene und erzählte Welt. Stuttgart: Kohlhammer.
Weinrich, Harald (1973). Le temps. Paris: Seuil.
Weiss, Peter H., John C. Marshall, Gilbert Wunderlich, Lutz Tellmann, Peter W. Halligan,
Hans-Joachim Freund, Karl Zilles, and Gereon R. Fink (2000). ‘Neural consequences of
acting in near versus far space: a physiological basis for clinical dissociations’, Brain 123:
Willet, Thomas L. (1988). ‘A cross-linguistic survey of the grammaticalization of evidentiality’,
Studies in Language 12: 51–97.
Wilmet, Marc (1997). Grammaire critique du français. Louvain-la Neuve: Duculot-Hachette.
Wilson, Deirdre and Dan Sperber (1993). ‘Pragmatics and time’, UCL Working Papers in
Linguistics 5: 277–300.
Wilson, Deirdre and Dan Sperber (2012). ‘Linguistic form and relevance’, in D. Wilson and
D. Sperber, Meaning and relevance. Cambridge: Cambridge University Press, 149–68.
von Wright, Georg Henrik (1984). Truth, knowledge, and modality, Oxford: Blackwell.
Yakpo, Kofi (2009). A grammar of Pichi. Berlin: Isimu Media.
Index of Names
Ahern, A. van Bogaert, J.
and Escandell-Vidal, L. M. 6 and Dendale, P. 62 n8
Aikhenvald, A. Y. 63, 135, 200 n11 Bonomi, A. 129
van Aken, H. 80 Boogaart, R. 222
Alexander of Aphrodisias 103 Borillo, A. 114, 115 n38
Apel, K-O. 103 n18 Bourne, C. 203 n15
Aristotle 103, 128 Branquinho, J. 183–6
Asher, N. 6, 8, 16, 17 n1, 19 n4, 22–3, 25 n8, Bras, M.
27, 29–30, 35, 41 n6 and Asher, N. 108 n28
and Bras, M. 108 n28 Bres, J. 100 n5
and Lascarides, A. 3, 6 Brisard, F. 5, 10, 223, 225, 228–9, 238 n2, 241,
and Pustejovsky, J. 27 n10 243–4, 247
van der Auwera, J. and Meeuwis, M. 230
and Plungian, V. A. 8, 200 n11 and De Mulder, W. 101 n9
and De Wit, A. 220, 228–30
Barker, C. Broad, C. D. 158 n1, 170 n13
and Jacobson, P. 110 Browne, E. J. P. 141 n17
Barone-Adesi, G. 137 Brunot, F. 100 n8
Barwise, J. Bühler, K. 68
and Perry, J. 39 n1 Button, T. 205
Bazzanella, C. Byloo, P.
and Miecznikowski, J. 138 and Diepeveen, J. and Nuyts, J. 76–7,
Beaver, D. 79 n6
and Condoravdi, C. 104 n20
Bellahsène, L. 64 Callender, C. 7
Bennett, K. Caplan, B.
and McLaughlin, B. 195 and Sanson, D. 159 n2
Berthonneau, A.-M. Carruthers, P. 2
and Kleiber, G. 100 n6 Casati, R.
van Benthem, J. and Varzi, A. 114 n35
and ter Meulen, A. G. B. 39 n1 Caudal, P. 61, 64 n10–n11
Berthonneau, A.-M. and Vetters, C. 115 n39
and Kleiber, G. 100 n6 and Roussarie, L., and Vetters, C.
Berti, A. 100 n7
and Rizzolatti, G. 239 n3 Celle, A. 64
Blakemore, D. 6 Chafe, W. 103
Bloor, T. Chen, L. 41 n6
and Pindi, M. 137 Chierchia, G. 2
280 Index of Names

Chilton, P. 5, 11, 47, 64 n11, 236, 238, 241, Donohue, J. P. 137

244–5 Dowty, D. R. 5, 29–30, 38, 48, 57, 110 n29
Chomsky, N. 2 Dummett, M. 243
Church, A. 22, 23 n6 Dyke, H. 163 n8
Cinque, G. 78 n4 and Maclaurin, J. 173 n15
Coates, J. 139
Col, G. Ebeling, M.
and Victorri, B. 115 n39 and Cooper, G. 137
Comrie, B. 232 Eckardt, R. 253
Condoravdi, C. 78 n4, 102 n15, 200–1 Enç, M. 5, 8
and Beaver, D. 104 n20 Escandell-Vidal, L. M.
and Harizanov, V. and Kaufmann, S. and Ahern, A. 6
78 n4, 129, 137, 237, 248 Evans, G. 183–6, 189
Cooper, G. Evans, V. 237
and Ebeling, M. 137
Copley, B. 104 n19 Faller, M. T. 132, 135, 146 n23, 200, 204
Coppola, M. 6 Fauconnier, G. 112, 241
Corkin, S. 6 Filipović, L.
Coventry, K. R. and Jaszczolt, K. M. 4
and Garrod, S. C. 239 n3 Fillmore, C. 239, 247–8
Croft, W. 75, 114 n34 Fine, K. 7
Culioli, A. 103 Fink, G. R. 239 n3
von Fintel, K.
Dainton, B. 170 n13 and Gillies, A. S. 78 n4, 146 n23, 201
Damourette, J. and Iatridou, S. 147
and Pichon, E. 61, 118, 253–4 and Matthewson, L. 208
Davidson, D. 5–6, 8, 29–32, 36, 185 Fleischman, S. 206
Davies, P. 161 n5, 209 Fløttum, K.
De Mulder, W. 111 n30 and Nølke, H. and Norén, C. 153 n26
and Brisard, F. 101 n9 Fodor, J. 2
De Wit, A. 210 n1 François, J. 115–16
and Brisard, F. 220, 228–30 Frank, A. 132 n6, 144 n22
and Patard, A. 229 Frawley, W. 236
Demirdache, H. Freeman, J. B. 134–6
and Uribe-Etxebarria, M. 115 n39 Frege, G. 10, 176–8, 181–5, 189, 191–2, 243
Dendale, P. Freund, H.-J. 239 n3
and van Bogaert, J. 62 n8 Frisson, S. 41 n6
Desclés, J.-P. 103 Fuchs, C.
Deygers, K. 80 and Victorri, B. 110
Diepeveen, J.
and Byloo, P. and Nuyts, J. 76–7, 79 n6 Gaizauskas, R.
Diessel, H. 239 and Mani, I. and Pustejovsky, J. 5
Dik, S. 99, 113 n33 Galileo 188
Doiz Bienzobas, A. 222 Gallagher, S. 170 n13
Index of Names 281

Gallese, V. 239 n3 Hopper, P. J.

Garrod, S. C. and Traugott, E. C. 251
and Coventry, K. R. 239 n3 Horn, L. R. 45 n8, 141
van Geenhoven, V. 109 Hornstein, N. 5
Geertz, C. 247 Hume, D. 7
Gillies, A. S. Husserl, E. 7, 197
and von Fintel, K. 78 n4, 146 n23, 201
Givón, T. 4 Iatridou, S.
Goldsmith, J. and von Fintel, K. 147
and Woisetschlaeger, E. 227 Inwagen, P. van 161 n5
Gosselin, L. 7, 9, 47, 100 n5, 101 n9, 102 n11,
102 n14, 103 n16, 104, 110, 113 n32, Jackson, F. 198 n5, 205
114 n34, 114 n36, 115, 120–1, 205 n17 Jacobson, P.
Gougenheim, G. 120 and Barker, C. 110
Grice, P. 6, 48, 152, 194, 198 n6, 208 James, W. 7, 170 n13
Groenendijk, J. Jaszczolt, K. M. 7–8, 10, 47, 103 n17, 172 n14,
and Stokhof, M. 17 173 n17, 177 n4, 193–5, 200 n11, 201,
Grondelaers, S. et al. 207–8, 236–7, 240, 242, 245–6, 253
de Groote, P. 24 and Filipović, L. 4
Gross, M. 115 n40 Jespersen, B. 5
Growdon, J. H. 6 Johnson, M.
Guillaume, G. 115 n39 and Lakoff, G. 247–8
and Srioutai, J. 8, 207
de Haan, F. 4, 128
Hacquard, V. 133 n7 Kamp, H. 5, 24, 126
Haeseryn, W. et al. and van Lambalgen, M. 5
Halligan, P. W. 239 n3 and Reyle, U. 17, 114 n34
Hamdani Kadri, D. 104 n21 and Rossdeutscher, A. 126
Hamm, F. 88 and Rohrer, C. 49, 108 n28
and van Lambalgen, M. 5–6, 8, 38 Kamp, J. A. W. 179
Harizanov, V. Kant, I. 7, 101, 167 n11
and Condoravdi, C. and Kaufmann, S. Kaplan, D. 179
78 n4, 129, 137, 237, 248 Kaufmann, S.
Harris, J. 231 and Condoravdi, C., and Harizanov, V.
Harrison, R. P. 199 n10 78 n4, 129, 137, 237, 248
Havu, J. 117 n43, 120 n48 Kelly, S. D. 7, 170 n13
Heck, R. G. 182–3 Kemmerer, D. 239 n3
van den Heede, V. 80 Kim, J. 5–6
Heidegger, M. 7, 199 Kleiber, G. 109, 142
Higginbotham, J. 176, 182, 189–92, 207 and Berthonneau, A.-M. 100 n6
and Pianesi, F. and Varzi, A. C. 5 Klein, W. 54, 105 n22–n23, 115 n39, 126
Hickok, G. 6 Koroshetz, W. J. 6
Hobbs, J. R. 152 Kratzer, A. 130, 131 n6, 200–1
282 Index of Names

Kronning, H. 101 n10, 115 n37, 117–18, Mani, I.

131 n3, 145, 147 and Pustejovsky, J., and Gaizauskas, R. 5
Kuhn, S. T. Mari, A. 60 n6
and Portner, P. 5 Marshall, J. C. 239 n3
Martin, R. 100 n5, 115 n38
Laca, B. 115 n39, 116 Matthewson, L. 4, 208
Lagerqvist, H. 102 n14 and von Fintel, K. 208
Lakoff, G. 247–8 McCawley, J. D. 230
and Johnson, M. 251 McDowell, J. 185
van Lambalgen, M. McLaren-Hankin, Y. 137
and Hamm, F. 5–6, 8, 38 McLaughlin, B.
and Kamp, H. 88 and Bennett, K. 195
McTaggart, J. M. E. 7, 157–8, 197
Lamiroy, B. 115 n40 Mellor, D. H. 7
Landman, F. 5–6 Merlini, L. 137
Langacker, R. W. 5, 75, 103, 115 n39, Merricks, T. 167 n11
210–11, 213–19, 225–7, 237, 241, Michaelis, L. A. 218 n8, 222, 237
243–4, 246–7 Miecznikowski, J. 133 n8
Larson, R. K. and Bazzanella, C. 138
and Ludlow, P. 187 and Rocci, A., and Zlatkova, G. 137
Lascarides, A. Moens, M.
and Asher, N. 3, 6 and Steedman, M. 109, 114 n34
Le Poidevin, R. 7, 161 n5, 170 n13, 202 Moeschler, J. 6, 46 n1
and MacBeath, M. 7 Molendijk, A. 48, 106 n25, 126 n49, 222 n10
Lepore, E. and Vet, C. 126 n49
and Loewer, B. 185
Levinson, S. C. 6 Nicolle, S. 6, 46 n1
Lewis, D. 171 Nølke, H. 117
Locke, J. 7 and Fløttum, K., and Norén, C. 153 n26
Loewer, B. Norén, C.
and Lepore, E. 185 and Fløttum, K. and Nølke, H. 153 n26
Lonergan, B. 247 Nunberg, G. 20 n5
Ludlow, P. 7–8, 10, 133 n8, 159 n3, 163 n7, Nuyts, J. 73–9, 83, 97
179–80, 183–4, 188, 193 n1, 203, 216 n4 and Byloo, P., and Diepeveen, J. 76–7, 79 n6
and Larson, R. K. 187
Luo, Z. 27 n10 O’Donnell, M. 137 n13
Łukasiewicz, J. 199 Oaklander, L. N. 202
and White, V. A. 202
MacBeath, M. Olson, E. T. 202 n12
and Le Poidevin, R. 7, 161 n5, 170 n13, 202 Øhrstrøm, P. and Hasle, P. F. V. 5
Maclaurin, J.
and Dyke, H. 173 n15 McElree, B. 41 n6
Maienborn, C. 6 Meeuwis, M.
Malchukov, A. 231 n12 and Brisard, F. 230
Index of Names 283

Miecznikowski, J. 133 n8 Recanati, F. 46, 48 n2, 52 n3, 66, 110, 201

Montague, R. 5 Reichenbach, H.-G. 5, 9, 46, 49, 67, 98,
Morasso, S. G. 104–5, 236–7, 253
and Rigotti, E. 142 n18, 152 Reinhart, T. 108 n28
Morency, P. Reyle, U.
and de Saussure, L. 61, 64, 253–4, 256 and Kamp, H. 17, 114 n34
Moss, L. S. and Rossdeutscher, A., and Kamp, H. 126
and Seligman, J. 39 n1 Richardt, S. 151 n25
Riegel, M.
Pancheva, R. and Pellat, J-C., and Rioul, R. 99 n3,
and von Stechow, A. 54 112 n31
Papafragou, A. 132 n6 Rigotti, E.
Parsons, J. 7, 159 n3, 167 n11 and Morasso, S. G. 142 n18, 152
Parsons, T. 5–6 Rioul, R.
Partee, B. H. 5, 46 and Pellat, J-C. and Riegel, M. 99 n3,
and ter Meulen, A. G. B. and Wall, 112 n31
R. E. 41 n5 Rizzolatti, G.
Patard, A. 100 n5, 229 and Berti, A. 239 n3
and de Wit, A. 229 Roberts, C. 150
Paul, L. A. 167 n11 Rocci, A. 8, 9, 47, 64, 132–3, 135 n10–n11,
Perry, J. 39 n1, 173 n17, 176–7, 181–4, 189–92 139, 147
and Barwise, J. 39 n1 and Miecznikowski, J. and Zlatkova,
Phillips, I. 170 n13 G. 137
Pianesi, F. and Zlatkova, G. 139
and Higginbotham, J. and Varzi, A. C. 5 Rödl, S. 184
Pichon, E. Rohrer, C.
and Damourette, J. 61, 118, 253–4 and Kamp, H. 49, 108 n28
Pickering, M. J. et al. 41 n6 Romijn, K. 88
Pietrandrea, P. 147 de Rooij, J. 88
Pindi, M. Rossdeutscher, A.
and Bloor, T. 137 and Reyle, U. and Kamp, H. 126
Pinker, S. 6 Rothstein, S. 5–6
Plato 103 Roussarie, L.
Plungian, V. A. and Caudal, P. and Vetters, C. 100 n7
and van der Auwera, J. 8, 200 n11 Roy, G.-R. 115 n37
Portner, P. 110 n29, 131 n3–n4, 132 n6, 142 Rumfitt, I. 185 n9
and Kuhn, S. T. 5
Prior, A. N. 5, 171, 173, 177–8, 180, 190, 203 van der Sandt, R. 18
Prosser, S. 7, 10, 159, 160 n4, 165 n9, 167 n10, Sanson, D.
170 n12, 171, 173 n16–n17, 177 n4, 216 n4 and Caplan, B. 159 n2
Pustejovsky, J. 5, 19, 27 n10 Sasse, H. J. 218 n7
and Asher, N. 27 n10 Sattig, T. 202
and Gaizauskas, R. and Mani, I. 5 Sauerland, U. 48 n2
284 Index of Names

de Saussure, L. 2, 6, 9, 46 n1, 53, 56, 58, Taylor, B. 6

62 n8, 66, 68, 77, 98 n1, 101 n9, 127, ter Meulen, A. G. B. 5, 8, 38–9
176 n3, 181 n7, 196 n4, 199 n9, and van Benthem, J. 39 n1
200 n11, 237, 242, 253 and Partee, B. H.and Wall. R. E. 41 n5
and Morency, P. 61, 64, 253–4, 256 and Smessaert, H. 39, 42 n7
and Sthioul, B. 50–2, 66 Tellman, L. 239 n3
Schein, B. 6 Thomason, R. H. 8, 102 n13
Schlesinger, G. N. 161 n5 Thornton, A. 4
Searle, J. R. 134–5, 140 n16 Tonhauser, J. 33, 35
and Vanderveken, D. 129 Tooley, M. 158 n1, 205–7
Seligman, J. van den Toorn, M. C. 88
and Moss, L. S. 39 n1 Toulmin, S. E. 134
Sider, T. 176, 178 Touratier, C. 100 n7, 112 n31
Sklar, L. 179 n5, 180 n6, 209 n24 Tournadre, N. 113 n33
Skow, B. 158 n1 Traugott, E. C.
Skibinska, E. and Hopper, P. J. 251
and Vetters, C. 99 n2 Traxler, M. 41 n6
Smart, J. J. C. 173 n16 Tye, M. 170 n13
Smessaert, H.
and ter Meulen, A. G. B. 39, 42 n7 Ullman, M. T. et al. 6
Smith, C. 29, 105 n24, 115 n39 Uribe-Etxebarria, M.
Smith, Q. 7, 163 n8, 203–5, 207 and Demirdache, H. 115 n39
Socrates 159
Speelman, D. 80 Vanderveken, D.
Sperber, D. and Searle, J. R. 129
and Wilson, D. 3, 6, 52, 54, 60, 101 n9, Varzi, A.
131 n2, 242 and Casati, R. 114 n35
Squartini, M. 135, 145, 147 and Higginbotham, J. and Pianesi, F.
Srioutai, J. 2, 207 Velleman, J. D. 167 n11
and Jaszczolt, K. M. 8, 207 Vendler, Z. 30
von Stechow, A. Verkuyl, H. 29
and Pancheva, R. 54 Vet, C. 57, 104, 109–10, 113 n33, 120
Steedman, M. 5, 8 and Molendijk, A. 126 n49
and Moens, M. 109, 114 n34 Vetters, C. 120 n48, 124
Sthioul, B. 51–3, 55, 63, 253–4, 256 and Caudal, P. 115 n39
and de Saussure, L. 50–2, 66 and Caudal, P. and Roussarie, L. 100 n7
Stokhof, M. and Skibinska, E. 99 n2
and Groenendijk, J. 17 Victorri, B.
Sundell, L.-G. 120 and Col, G. 115 n39
and Fuchs, C. 110
Tallant, J. 202 n13, 205 n18 Vlach, F. 109
Tasmowski-de Ryck, L. 50 Vuillemin, J. 102 n12
Index of Names 285

Wall, R. E. Woisetschlaeger, E.
and Partee, B. H. and ter Meulen, and Goldsmith, J. 227
A. G. B. 41 n5 von Wright, G. H. 102 n12
Walsh, P. 137 Wunderlich, G. 239 n3
White, V. A.
and Oaklander, L. N. 202
Yakpo, K. 218 n7
Weinrich, H. 49, 108 n28
Weiss, P. H. et al. 239 n3
Willet, T. L. 135 Zilles, K. 239 n3
Wilmet, M. 99 n2, 115 n37, 119 n47 Zlatkova, G.
Wilson, D. and Miecznikowski, J. and
and Sperber, D. 3–4, 6, 52, 54, 60, 101 n9, Rocci, A. 137
130 n2, 131 n2, 242 and Rocci, A. 139
Index of Subjects
absolute time 105 A-theoretic properties in thought and
accommodation 47 language 163–5
accomplishment 15, 31, 37, 48–9, 64 n11, thought, language and A-theory 171–3
114, 245 evaluation of 174
achievement 4, 15, 31–2, 48–9, 64, 67, 114, 124, experience and illusion of passage
222, 245 165–71
activity 218 n8 perceptual relation 161
actual 223–4, 226 phenomenal character 161–2
adjectives 23, 25 presentism 157–60
Afrikaans 233–4 auxiliaries, see modal auxiliaries
agent’s goals, see Italian
Aktionsart schemas 52 Belgian 80
see also frames of reference belief ascription theory 187
alethic conversational backgrounds, see biology 140 n16
conversational backgrounds; Italian Black English Vernacular 233
allocentrism 256 see also English
allocentric projection 60, 63, 65, 67 bounded region 211 n2
allocentric usage 53–4 boundedness 217, 228, 234
see also perspectival interpretations of tenses branching time theory 102
ambiguity 208 n21 British Colombia, see St’àt’imcets
anaphora (Lillooet Salish)
temporal 5–6, 132, 180 B-theory, see A-theory
anterior future (AF) 56, 59, 61, 64 n11 calendars
anthropology 1 Balinese 247–8
argumentation theory 134, 142 n18 category theory 23
aspect, see English Central Pomo 4
aspectual adverbs, see temporal reasoning chemistry 140 n16
aspectual periphrases, see French: time and coercion, see temporal modification
modality in cognition 10, 236
aspectual viewpoint 105 cognitive dynamics, see tensism
assertoric modality 101 Cognitive Grammar (CG), see English
asymmetry 203 n15 205 n17–n18, 209 cognitive linguistics 5, 74, 247
of psychological time 209 ‘cognitive constraint on semantics’ 177
see also French: time and modality in cognitive metaphor theory 251
A-theory 7, 157–74, 197, 205–7 see also metaphor
‘A-properties’ 157 cognitive operators, see frames of reference
A-theoretic passage of time, perceptions cognitive science
of 160–3 peripersonal space, study of 239 n3
Index of Subjects 287

cognitive–functional approach, see modal corpus linguistics

auxiliaries corpora:
Common Ground (CG) 43–4 ConDiv Corpus of newspaper
compatibility restricted union 132 n6 language 80
compositionality 5, 109–10 Corpus Gesproken Nederlands 80–1
computational linguistics Italian financial news 136–9
language modelling 1 corpus techniques:
see also corpus linguistics keyness 138
computational semantics 115 n39 corpus tools:
conceptual reduction 209 n24 UAM Corpus Tool 137 n13, 138 n14
conceptual systems 74–5 see also computational linguistics
conceptualization 204, 207, 211–12 counterfactuality 86–92, 129
see also frames of reference Cuzco Quechua 132
conditional mood, see Italian
conditional restrictions, see Italian Davidsonian semantics
conjunction 2–3, 6 event semantics 30, 32
conscious pragmatic inference T-theory 185
(CPI) 208 death 199
consecutive tense 4, 206 Default Semantics (DS) 194, 196, 246
content DS-theoretic merger representation
designational 189 196, 198
referential 189, 191 degrees of reality, see temporality
reflexive 189–92 deictic future, see frames of reference
contingent reality deictic space theory (DST) 11, 237–43
Contingency in Immediate Reality 228 3D and integrated abstract space 238–41
see English frame shifting principles 251, 255
conversational backgrounds frames of reference and geometrical
alethic 131, 139–46 transformations 241–2
conditional restrictions for the alethic fundamental deictic space 239
B 151 non-putative future 254
economic causality 148–9 now 242–3, 255
circumstantial 131, 139–40 peripersonal region 243, 249
combining 131 n6 peripersonal space 239
conditional restriction and 147–8 peripersonal time 239–40, 243
deontic 131, 139–46, 149–50 putative future 253–7
conditional restrictions for the windowing 245, 250–2
deontic B 152 see also frames of reference
epistemic 131, 139 deixis 236
evidential bases of predictions and see also deictic space theory (DST)
152–3 deontic modality, see modal auxiliaries
reportative 146 deontic readings, see Italian
teleological 131, 142 deontic-practical modality, see perspectival
see also Italian interpretations of tenses
288 Index of Subjects

depth, principle of 74–5 grounding 213–16

determinism 102, 205 habitual 226
directivity, see modal auxiliaries imperfective past tense 223
Direct-Quantitative View (DQ) 10, 201, imperfective past tense: modal uses
203–6, 208–9 of 224
Discourse Representation Theory (DRT) modal usage types 229
5–6, 39, 242 perfect aspect and past tense 230–4
Segmented (SDRT) 6, 242 performative 225
discourse space models (DSMs) 238 present and past tenses 217, 239 n2, 243–4
downwards entailing context 42 present imperfective 221
Dutch 45, 233–4 present perfect: current relevance 233
cognitive-functional analysis of 9 present perfect: extended now 232
ConDiv Corpus 80 present perfect: resultative use 230
Corpus Gesproken Nederlands 80–1 present perfective 221
imperfectivization 221 present progressive 227
modal auxiliaries and tense 73–97 present tense 218
modal usage types 229 present tense morphology 248
simple present 220, 225 progressive and perfect marking 224–34
see also modal auxiliaries scheduled future 226
Dynamic Aspect Trees (DAT) 38 simple past 222
dynamic modality, see modal auxiliaries stativity 244
dynamic semantics 17–18 structural knowledge types 227
dynamism, principle of 74–5 structural vs contingent reality in
English 224–30
economic causality 151 symbolism 211–13
see also Italian temporary-validity reading of present
economic circumstances, see Italian progressive 228
economic rationality 142, 149 tense in CG 216–19
economics 137, 140 n16 theory 210–11
see also corpus linguistics; Italian verb classes 218
egocentric properties 177, 180–1, 188–91 conditionals 51
eikòs 153 epistemic future 67
emotion 176 n2, 179, 190, 192 future ‘will’ 99 n2
endoxon 153 grammaticalization of temporal
English reference 206
Cognitive Grammar (CG), tense and PPs and adverbials 54–5
aspect in 8–11, 210–35 present perfect puzzle 47, 106 n26
combinations of tense and lexical progressives 106 n25
aspect 219 tense-time mismatches 2, 196 n4
durational/‘continuative’ reading of see also Black English Vernacular; Irish
present progressive 229 English; Middle English; Old English;
evaluation of toolkit 234–5 Standard American English;
grammatical aspect in CG 220–4 temporal modification
Index of Subjects 289

epistemic commitment, see temporality with perfective viewpoint 110

epistemic detachment 195, 246 with prospective viewpoint 111
epistemic distance 238 composed past 47
epistemic futures (EF), see French contextual shift 3–4
epistemic modality 4, 10, 146 n23, 193 epistemic futures (EF) 9, 200 n11
evidentiality and 200 n11 grammaticalization of temporal
temporality as 195–9 reference 206
see also modal auxiliaries have-construction 233–4
epistemology of time 10, 137 imperfective past: 222
events 6, 244 narrative and non-background use
evidential bases, see Italian of 9, 48–54
evidentiality, see modal auxiliaries imperfective pasts 4
experience, see A-theory modal usage types 229
experimental philosophy 198 n5 present perfect with future
reference 176 n3
focus, see temporal reasoning present perfect:
folk psychology 68–9 future reference with 9, 54–60
formal semantics 115 n39 simple present 220 n9, 225
four-dimensionalism, see tensism time and modality in 9, 98–127
frames of reference 9, 236–58 aspectual periphrases, classification and
Aktionsart schemas 244–5 representation of 118–23
cognitive operators 244–5 aspectual support verbs’ 115 n38
evaluation of 257–8 aspectual viewpoint auxiliaries (aux-AV),
geometrical transformations and 241–2 classification of 119
present and future: 245–57 evaluation of 127
‘moving into’ the future 251–3 general preliminaries 99–102
present progressive and deictic intervals 98–9
future 250 interval-semantic model 104–10
putative future as reference frame markers and temporal, aspectual, and
solution 253–7 modal meanings 101
simple present and deictic future modal asymmetries in pragmatic and
246–51 semantic times 102–4
tense forms 244–5 modal values (temporal/aspectual),
times, tenses and 236–8 relations between 111–13
see also deictic space theory (DST) modal values of aspectual
free indirect discourse 133 n7 viewpoints 110–11
Fregean perspective, see tensism monstration 116
French phasal aspect 113–18
aspectual modal values 110–11 phasal aspect: modal interpretation
temporal modal values, relations of 123–7
with 111–13 phasal aspect: phasal structure of
with imperfective viewpoint 111 eventualities 114
with perfect viewpoint 110 semi-auxiliaries 115 n40
290 Index of Subjects

French (cont.) imperfective paradox 38, 110, 124–5

simple future 112 imperfective past, see French
tenses and modality, classical view on 100 indexical inference, see temporal reasoning
V-MAPs 115–22, 124 indexical predicates, see temporal reasoning
V-MAPs: semantic classification of 119 indicative mood 132 n7
variants of ON 153 n26 see also Italian
functional linguistics 74 Indo-European languages 8
see also cognitive-functional approach; inference rule 37–8
modal auxiliaries information structure, see temporal
geometry 236, 238, 240 insider/outsider status 137
geometrical transformations 241–2 institutional states of affairs 134–5
see also deictic space theory (DST); frames instructions 6, 40, 108
of reference intentionalism 166
German 45, 233–4 internal aspect 114, 119, 121
lexical aspect 217 n7 interpretation process 42
Germanic languages 45 interpretive reading 60
have-perfects 224 interval-based account, see French: time and
imperfectivity 222 modality in
Gestalt psychology 212 n3 intervals, see French: time and
Gettier cases 198 n5 modality in
grammatical aspect, see English interval-semantic model 104–10
Gricean pragmatics 6, 48, 194 absolute time 105
account of modalities 198 n6 aspectual viewpoint 105
rational conversational behaviour 208 compositionality 109–10
grounding 213–16 conflict resolution 108–9
Elaborated Epistemic Model 215 imperfective background 108
nominal and clausal grounding 214 imperfective viewpoint 106
profiling 214, 217 past tense with imperfective
Time-Line Model 215–16 viewpoint 107
see also English past tense with perfect viewpoint
growing block theory 158 n1 107–8
past tense with perfective
Hering illusion 166 viewpoint 107
heuristics 3, 257 perfect viewpoint 106
Hiberno-English, see Irish English perfective viewpoint 106
historic necessity 8, 98–9, 102 prospective viewpoint 106
Hopi 7 see also French: time and modality in
intuitions on inference 43
Idealized Cognitive Model (ICM) 247 Irish English 231, 233
idiom 81 see also English
illocutionary acts 135 irony 68
‘imparfait’ 99, 122–3, 126, 222, 233 ‘irrevocable’ 9, 98, 102–4, 110, 112–14,
see also French 124, 126–7
Index of Subjects 291

Italian 55 lexical meaning, see temporal modification

epistemic futures (EF) 60, 64 linguistic communication 236
verification hypothesis 64
future-oriented alethic and deontic Maori 4
modals 139–46 mathematical sentences 31
deontic readings of dovere 9–10, 130–1, maximal scope 213, 216 n5, 229 n11
135, 145–6 memory 207
deontic readings of potere 144–5 mental states 3
economic causality: necessity and mereotopology 114 n35
impossibility 140–2 metaphor 151 n25, 160, 251
economic circumstances and agent’s see also cognitive metaphor theory
goals 143–4 metaphysics
possibility modal, quantificational of time 7, 10, 102, 165, 216 n4
readings of 142–3 metarepresentation 9, 54, 60, 68–9
inferential evidentiality 129, 133 n8 Middle English 229
modals 128–53 see also English
conversational backgrounds and mitigation 88–93
evidential bases 128–33, 152–3 modal asymmetry, see French: time and
evidence for predictions 133–6 modality in
financial news corpus 136–9 modal auxiliaries
predictions 9 analysis 82–94
TAME categories 128 deontic modality 76, 90–2
modals and the conditional mood in directivity 92–4
predictions: dovrebbe 147–53 dynamic modality 76, 81, 88–90
alethic conversational backgrounds: epistemic modality 76, 92
economic causality 148–9 evidentiality 92
conditional restrictions, nature of functions of tenses 85–8
150–2 modal meanings 88–94
deontic conversational modals and complex tenses 83–94
backgrounds 149–50 modals and ‘simple’ tenses 82–3
semantics of dovrebbe: conversational possible combinations 83–5
background and conditional basic concepts and notions 74–9
restriction 147–8 cognitive–functional approach 74–5
speech acts of prediction: hierarchy of qualifications of SoAs
most frequent modals 138 78–9
linguistic dimension 77–8
knowledge semantic dimension 75–7
concept of 198 n5 cognitive–functional approach 73, 97
combinations of modal and tense
lambda abstraction 43–4 meanings 94–6
language acquisition 1 evaluation of 96–7
language processing 3, 10, 61–2, 194, method and data 79–82
212 n3, 242 meanings of modals 81
292 Index of Subjects

modal auxiliaries (cont.) perspectival interpretations of

modal and tense meanings: tenses 46–69
possible combinations of 94 allocentrism 53, 60, 67–9
qualificational hierarchy 73–4 contextual understanding 46–8
tense and 73–96 deontic-practical modality:
see also Dutch; French: time and with epistemic futures 60–7
modality in epistemic futures (EF) 60–7
Modal-Contextualist View (MC) 10, 201, French imperfective past:
203–6, 208–9 narrative and non-background use
modality, see Dutch; French; modal of 48–54
auxiliaries; perspectival French present perfect:
interpretations of tenses: deontic- future reference with 54–60
practical modality phasal aspect, see French: time and
modulation 47 modality in
monotonicity 41 phenomenology 169
Montague Grammar (MG) 22, 26, 30 philosophy 1, 202
moving spotlight theory 158 physics 140 n16, 180, 209, 236
thermodynamics 209
narration 3 Pichi 217 n7
see also French polarity
natural language 39, 45, 191 polarity particle 44
Natural Realignment Claim 186 polarity transition 40
natural science 140 n16 see also negative polarity; temporal
naturalistic reductionism 209 reasoning
Necessity Claim 186 possibility readings, see Italian
negative polarity 42 possible worlds semantics 132 n6, 200
see also polarity pragmatic inference 5
newspaper discourse, see corpus linguistics predictions
nominal modification 30 evidence for, see Italian
see also temporal modification present perfect, see French
now, see frames of reference present progressive, see frames of reference
numerical determiners 8, 43–4 presentism 7, 158–9, 163, 192, 199, 204–5
degree 203
Old English 230–1 see also tensism
see also English presupposition justification, see temporal
ontology of time 137 modification
origo 68 132 n7 presuppositions 17–18, 126–7
temporal 126 n49
past tense, see English see also type presuppositions
perceptual systems 168–9 preterite loss 233
visual 168, 212 n3 priority modals 131 n4
perfect aspect, see English probability 195
perfect marking, see English ‘problem of temporary intrinsics’ 170 n12
Perry/Higginbotham alternative, see tensism procedural expressions 6
Index of Subjects 293

processes 212 space-time 179–80

progressive marking, see English Minkowski space–time diagram 180
projection problem 17 n2 Spanish
proprioception 68 imperfective past tenses 222
prosody 42 n7, 45 specious present 7, 170 n13
psychology 1 speech acts 8–9, 135, 137
human 149 most frequent modals (Italian) 138
peripersonal space, study of 239 n3 St’àt’imcets (Lillooet Salish) 4
see also folk psychology Standard American English 233
public reference frame (PRF) 248 see also English
putative future, see frames of reference state of affairs (SoA) 76–7
hierarchy of qualifications of 78–9
qualificational hierarchy, see modal see also modal auxiliaries
auxiliaries states 218 n8
quantification of time, see temporality statistics, see corpus linguistics
quantificational readings, see Italian stativity 217 n7, 218, 244
status causae doctrine 134
radical contextualism, see Default Semantics structural reality, see English
rate of time 202 n12 subjectification 231
reductionism 10 Sufficiency Claim 186
reference, see frames of reference supervenience 10, 194, 205–6, 236
referentialism 182–3 support for 202–6
Relative Modality (RM) theory 130–2 temporality as epistemic modality
relativity 195, 236 195–9
universalism, debate with 1 see also temporality
relevance theory 6, 62 n7, 242 Swahili
reported speech 68 consecutive tense 206
representing time 205 symbolism, see English
Romance languages 55, 222 symmetry problem 172
see also asymmetry
Samoan 218 n7
Saxon genitive 35 TAME categories, see Italian
scalar inferences, see temporal reasoning telicity 217 n7
schedules 248 temporal adverbials, see temporal
SCWD (social, cultural, or world-knowledge modification
defaults) 208 temporal modification 8, 15–36
semantic modalities 99 b coercions 24
sense coercion and presupposition
indexical 176, 180 justification 19–22
tracking 192 evaluation of 36
see also tensism lexical meaning, formal theory of 22–9
simple present, see frames of reference type presuppositions 23–9
sociology 1 types 22–3
294 Index of Subjects

temporal modification (cont.) tensism 7, 175–92

temporal adverbials 15–16 definition 175–6
type composition logic 16–19 evaluation of 192
verbal complex, modifications of 29–36 four-dimensionalism and 159 n3,
modifications 29–33 178–81
temporal nominal modification 33–6 Fregean accounts of tense:
temporal parts 173 cognitive dynamics of 184–9
temporal reasoning Fregean senses 181–4
as indexical inference 8, 37–45 initial case for 176–8
aspectual adverbs 39–40 metaphysical vs linguistic 10
basic 40 Perry/Higginbotham alternative:
evaluation of 44–5 designational content 189
focus and information structure of 42–4 indexical content 189–90
prosodically marked 42 n7 referential content 189
scalar inferences and polarity in 41–2 token reflexive content 189–92
indexical predicates 39–40 tenseless time vs 202–6
past progressive 37–8 testimony 130, 134, 136
present perfect progressive 37–8 Thai
simple past 37–8 modal auxiliaries 206–7
situated reasoning 37–9 pragmatic means 2
temporality ‘thank goodness’ argument 171–2
epistemic commitment and 193–209 Theory of Mind 9, 69
quantifying time and ‘degrees of thought, see A-theory
reality’ 199–202 time
rationale 193–5 complex concept of 193, 206–9
temporality as epistemic modality and:
modality 195–9 in French 98–127
evaluation of 208–9 temporality, tense and 1–10
time, tense and 1–10 tenseless 202–6
see also supervenience; time see also temporality
tense token reflexive content, see tensism
in Cognitive Grammar (CG), see English truth-conditional semantics 244
modal auxiliaries and 73–97 type composition logic, see temporal
philosophical debates 46 n1 modification
as procedural (vs. conceptual) Type Composition Logic (TCL)
expressions 46 n1 24, 31–2
time, temporality and 1–10 type presuppositions 18, 23–9
see also English; modal auxiliaries; nuclear scope 27
perspectival interpretations polymorphic type functor 28
of tenses Simple Accommodation 24
tense forms, see frames of reference standard typing context 24
tensed theory, see A-theory type parameter 24
tensers 175 type system axioms 28–9
Index of Subjects 295

see also presuppositions; temporal vector 243–56

modification verbal complex, see temporal
types, see temporal modification modification
veridicality 43
unarticulated constituents 173 n17 ‘view from nowhere’ 180
uncertainty 195 virtual realities 241
relativity, debate with 1
wrong candidate objection 176, 181
unresolved question (UQ) 10, 193–209
temporality and epistemic commitment
see also temporality update 40, 42–3 zero-point concept 243