Connectives As Discourse Landmarks (Pragmatics and Beyond New Series)

Connectives as Discourse Landmarks Pragmatics & Beyond New Series (P&BNS) Pragmatics & Beyond New Series is a continu...

Author: Agnes Celle | Ruth Huart

60 downloads 1389 Views 2MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

Connectives as Discourse Landmarks

Pragmatics & Beyond New Series (P&BNS) Pragmatics & Beyond New Series is a continuation of Pragmatics & Beyond and its Companion Series. The New Series offers a selection of high quality work covering the full richness of Pragmatics as an interdisciplinary field, within language sciences.

Editor Andreas H. Jucker

University of Zurich, English Department Plattenstrasse 47, CH-8032 Zurich, Switzerland e-mail: [email protected]

Associate Editors Jacob L. Mey

University of Southern Denmark

Herman Parret

Jef Verschueren

Susan C. Herring

Emanuel A. Schegloff

Belgian National Science Foundation, Universities of Louvain and Antwerp

Belgian National Science Foundation, University of Antwerp

Editorial Board Shoshana Blum-Kulka Hebrew University of Jerusalem

Jean Caron

Université de Poitiers

Indiana University

Masako K. Hiraga

St.Paul’s (Rikkyo) University

University of California at Los Angeles

Deborah Schiffrin

David Holdcroft

Georgetown University

Sachiko Ide

Kobe City University of Foreign Studies

Sandra A. Thompson

Thorstein Fretheim

Catherine KerbratOrecchioni

John C. Heritage

Claudia de Lemos

Teun A. van Dijk

Marina Sbisà

Richard J. Watts

Robyn Carston

University College London

Bruce Fraser

Boston University University of Trondheim University of California at Los Angeles

University of Leeds Japan Women’s University

University of Lyon 2 University of Campinas, Brazil University of Trieste

Volume 161 Connectives as Discourse Landmarks Edited by Agnès Celle and Ruth Huart

Paul Osamu Takahara

University of California at Santa Barbara Pompeu Fabra, Barcelona University of Berne

Connectives as Discourse Landmarks

Edited by

Agnès Celle and Ruth Huart University of Paris-Diderot

John Benjamins Publishing Company Amsterdam / Philadelphia

8

TM

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

Library of Congress Cataloging-in-Publication Data Connectives as discourse landmarks / edited by Agnes Celle and Ruth Huart. p. cm. (Pragmatics & Beyond New Series, issn 0922-842X ; v. 161) Includes bibliographical references and index. 1. Discourse markers. 2. Grammar, Comparative and general--Connectives. I. Celle, Agnès. II. Huart, Ruth. P302.35.C66 2007 415--dc22 isbn 978 90 272 5404 7 (Hb; alk. paper)

2007014007

© 2007 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Co. · P.O. Box 36224 · 1020 me Amsterdam · The Netherlands John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa

TSL[v.20020404] Prn:10/05/2007; 10:34

F: PB161CO.tex / p.1 (v)

Table of contents

List of contributors Connectives as discourse landmarks: Introduction Agnès Celle and Ruth Huart

vii 1

Part I. Connectives and modality Connectives, modals and prototypes: A study of rather Raphael Salkie

15

The interface between discourse and grammar: The fact is that Karin Aijmer

31

Part II. From syntax to pragmatics And as an aspectual connective in the event structure of pseudo-coordinative constructions Mark de Vos

49

‘Are you a good which or a bad which?’ The relative pronoun as a plain connective Rudy Loock

71

From temporal to contrastive and causal: The emergence of connective after all Diana M. Lewis

89

Part III. Discourse strategies Orchestrating conversation: The multifunctionality of well and you know in the joint construction of a verbal interaction Barbara Le Lan

103

TSL[v.20020404] Prn:10/05/2007; 10:34



F: PB161CO.tex / p.2 (vi)

Table of contents

A because B so A’: Circularity and discourse progression in conversational English Frédérique Passot Not that . . . versus It’s not that . . . Ruth Huart

117 135

Part IV. In search of operations ‘He’s a cop but he isn’t a bastard’: An enunciative approach to some pragmatic effects of the coordinator but Martine Sekali Continuity and discontinuity in discourse: Notes on yet and still Graham Ranger

155 177

Reconsidering the discourse marking hypothesis. Even, even though, even if, etc. as morpheme/construction pairs François Nemo

195

Index

211

TSL[v.20020404] Prn:9/05/2007; 15:41

F: PB161LC.tex / p.1 (vii)

List of contributors

Karin Aijmer English Department Göteborg University Box 200 SE 405 30 Göteborg Sweden [email protected] Agnès Celle UFR d’Etudes anglophones University of Paris-Diderot 8-10 rue Charles V 75 004 Paris France [email protected] Mark de Vos Department of English Language and Linguistics Rhodes University Grahamstown 6140 South Africa [email protected] [email protected] Ruth Huart UFR d’Etudes anglophones University of Paris-Diderot 8-10 rue Charles V 75 004 Paris France [email protected] Barbara Le Lan UFR d’anglais Université Paris IV-Sorbonne 1 rue Victor-Cousin 75 230 Paris Cedex 05 France [email protected]

Diana M. Lewis Département d’Etudes du Monde Anglophone Faculté des Langues, 74 rue Pasteur Université Lumière Lyon 2 69 365 Lyon Cedex 07 France [email protected] Rudy Loock Université de Lille III – U.F.R. Angellier U.M.R. 8528 SILEX B.P. 149 59 653 Villeneuve d’Ascq Cedex France [email protected] François Nemo UFR Lettres Langues Sciences Humaines – Sciences du Langage Université d’Orléans 10 Rue de Tours – BP 46527 45 072 ORLEANS CEDEX 2 France [email protected] Frédérique Passot UFR du Monde anglophone University of Paris 3 – Sorbonne Nouvelle, 5, rue de l’Ecole de Médecine 75 006 Paris France [email protected] Graham Ranger Université d’Avignon et des Pays de Vaucluse Faculté des Lettres et Sciences Humaines –

TSL[v.20020404] Prn:9/05/2007; 15:41

F: PB161LC.tex / p.2 (viii)

 List of contributors

Département d’anglais 74 rue Louis Pasteur 84 029 Avignon Cedex France [email protected] Raphael Salkie School of Languages University of Brighton Falmer, Brighton E. Sussex, BN1 9PH

United Kingdom [email protected] Martine Sekali UFR d’Etudes anglo-américaines University of Paris X, Nanterre. 200 avenue de la République 92 001 Nanterre Cedex France [email protected]

TSL[v.20020404] Prn:30/05/2007; 10:09

F: PB16101.tex / p.1 (1)

Connectives as discourse landmarks Introduction Agnès Celle and Ruth Huart University of Paris-Diderot

The present volume includes a selection of eleven papers delivered at the international conference “Connectives as Discourse Landmarks”, held at University of Paris-Diderot in May 2005. These studies explore the syntax and semantics of connectives in English with respect to their discourse function. The title of the collection “Pragmatics and Beyond” summarizes accurately the editors’ purpose in uniting these particular papers. Discourse, i.e. language in use, is at the heart of the authors’ preoccupations. Yet, pragmatic considerations are treated here neither as a starting point nor as an aim, but rather as integrated into a comprehensive linguistic treatment of various types of units.

Background Discourse analysis, which investigates units larger than the sentence, as well as speaker/hearer (or writer/reader) relationships, is traditionally considered as belonging primarily to the field of pragmatics. At a period when syntax, and therefore sentence structure, dominated most linguistic investigations, a few researchers, recognizing the need to look beyond the sentence, began studying elements which did not fit neatly into syntactic patterns. The models available at the time were either semantic – how do the markers studied influence the truth conditions of utterances? (Warner 1985) – or pragmatic, in terms of coherence and Gricean maxims (Schiffrin 1987). Another pioneer, J.-O. Östman, looked closely at different discourse types to discover recurring patterns characteristic of each. These authors studied samples of authentic interactive speech and tried to classify into functional categories what they called, respectively, “discourse connectives” (Warner), “discourse markers” (Schiffrin), and, in the Germanic tradition, “pragmatic particles”. The definitions which unite the various items show two basic concerns: the rela-

TSL[v.20020404] Prn:30/05/2007; 10:09



F: PB16101.tex / p.2 (2)

Agnès Celle and Ruth Huart

tions between sentences, on the one hand, and the relations between speakers, on the other.

Terminology In the twenty years which have intervened since these initial studies, other terms have been proposed, with more or less restrictive senses, in particular “discourse particles” (Hansen 1998; Aijmer 2002). Hansen defines these as “nonpropositional linguistic items whose primary function is connective and whose scope is variable” providing “instruction on how to integrate the host unit into a coherent mental picture of the discourse” (1998: 75). Aijmer (2002) also underlines the non-propositional character of particles, and adds a criterion of grammaticalization: in other words such markers are treated as distinct from the lexical items which gave birth to them. For example, according to Schiffrin (1987), “y’know”, which is obviously made up of the second person pronoun subject and the verb of cognition know, now functions as an indivisible unit, used in discourse to attribute to the speaker the role of information-provider, contingent upon hearer reception. It is to be noted that what these authors call “particles” range from monosyllabic interjections (oh, well) to full prepositional or noun phrases (as a matter of fact, and all that sort of stuff.) In the present volume, we have chosen the term “connectives” with no particular theoretical connotations intended. In particular, we are not restricting the field to extra-syntactic material. Alongside markers with obvious pragmatic function (well, you know), the items under study range from traditional conjunctions (and, but, because. . .), whose role is by nature connective, to full sentential frames (the fact is that. . . , it’s not that. . .). In other cases, adverbs which may be found in various positions (still, yet, rather, even. . .) can serve to link chunks of discourse when sentence initial. The term connective is thus to be taken in a loose, nontechnical sense to mean element used for linking, whether or not it meets certain syntactic criteria. As for the semantic criterion of “grammaticalization”, as developed for example in Hopper and Traugott (1993), we wish to leave the question open. Some of our markers are obviously wholly grammatical from the outset (and, not that, so. . .). For others (after all, fact is), the notion of “semantic bleaching” is central. However, as several of the studies show, it is not so much the progression from lexical content to grammatical function that characterises connectives, as the interaction of semantics, syntax and pragmatics in producing new syntactic configurations with corresponding pragmatic functions. R. Loock’s paper on which is particularly revealing in this respect: the anaphoric value inherent in the relative pronoun takes on a new colouring as the pronominal use is attenuated.

TSL[v.20020404] Prn:30/05/2007; 10:09

F: PB16101.tex / p.3 (3)

Connectives as discourse landmarks

Invariance Among the questions that have been debated in the literature is that of “core meaning”. Is there a basic semantic “content” that prevails throughout the uses of each term and can therefore be brought to the surface through systematic investigation of what varies in the context (Hansen 1998; Schiffrin 1987)? Or would any such search yield such abstract, vague definitions that they would be of no use in differentiating the various markers (König 1991; Aijmer 2002)? K. Aijmer is representative of those who believe that it is more useful to consider that a central meaning traceable to etymological origins has given rise to related meanings which have come to be associated through usage, but that the unifying factor behind a given discourse particle is functional, rather than semantic, consisting in instructions for interpretation. At the opposite end of the spectrum, current trends in lexical semantics, as represented, for example, by C. Rossari (2000; Rossari et al. 2004) support the view that invariant properties characterise each term, independently of context and situation, so that all elements of interpretation which can be traced to pragmatic factors should be stripped away in order to determine the contribution of the markers themselves. On this view, an initial contention by Warner (1985) that since the same relations between adjacent sentences are often found with or without connectives, the paratactic and hypotactic forms are simply two surface manifestations of a single underlying structure, is untenable. In spite of the varied approaches reflected in this book, the authors hold the view that to a difference in form corresponds a difference in value.

An integrative approach The papers united here reflect a variety of approaches to the question of invariant properties. What they have in common is the refusal to isolate any particular component of the language. Not only are syntax and semantics inseparable, both through the evolution of forms and the fact that position influences meaning, but also prosody, as a reflection of speaker-hearer relations, is an essential factor (Le Lan, Passot). Whether the emphasis is on grammaticalization in progress (Aijmer, Lewis, Loock) or on the fundamental operations underlying various uses of a common grammatical item (De Vos, Nemo, Sekali), the traditional distinction between “lexical items” and “grammatical function words” is blurred. While the connectives under consideration may be viewed as having primarily “instructional” implications, involving the role a speaker attributes to the addressee, these pragmatic properties are not detached from the semantico-syntactic primitives associated with the forms.



TSL[v.20020404] Prn:30/05/2007; 10:09



F: PB16101.tex / p.4 (4)

Agnès Celle and Ruth Huart

This effort to analyse the formal manifestations of language in terms of enunciative operations has characterized much of English linguistics in France for over thirty years, in the wake of G. Guillaume and E. Benveniste. One of the features of this volume is to make such approaches accessible to linguists working within other frameworks, through application to authentic stretches of discourse. Another significant aspect of discourse studies is the collection of data. Most specialists agree that context is of utmost importance, so that relatively extensive extracts from authentic sources are desirable. For this purpose, use of computerized corpora of both written and spoken material has proved invaluable (Aijmer 2002; Aijmer & Stenström 2004). Nearly all the papers in this book base their findings either on standard corpora such as the BNC, or on samples of speech collected by the authors, sometimes supplemented by the former. This practice allows minute analysis of the conditions that give rise to the choice of each marker and description of the patterns that can only emerge when numerous occurrences are observed. In exploiting corpus data, frequency counts can be valuable guides to analysis, but are not, in the present case, the main focus of attention. When statistics are given, they serve to consolidate hypotheses about evolution in progress, for example. While some of the connectives under scrutiny here have already been studied within a discourse framework (and, but, well, so, because: Schiffrin 1987; you know: Östman 1981; still, yet, even: König 1991), each contribution adds a new dimension, for example by integrating the discourse marking function into a larger framework (Sekali, Nemo) or by showing the interaction between two terms which are not normally seen as related (Passot, LeLan). In other cases, little or no attention has been paid in the literature to the development of new functions for familiar terms (Aijmer, Loock). Thus, in spite of the variety in the theoretical frameworks applied, these papers can be seen to echo each other and contribute collectively to the integration of various facets of linguistic analysis, in which the speaker’s role is central. The four sections provide a dynamic progression in spiral form. Part I: Connectives and Modality sets the spirit of the volume by illustrating, through two very different sorts of markers – a single polyvalent lexical item (rather) and a sentential frame (the N is that) – the inextricable interaction between syntax and semantics and the expression of modality. The three papers in Part II: From Syntax to Pragmatics show how slight variations in structure develop in parallel with special interpretations. The following section on Discourse Strategies deals with pairs of markers that have complementary functions in dialogue, and are examined jointly to discover the precise conditions of occurrence and subtleties of meaning associated with them. In the last section: In Search of Operations, the various dimensions studied are shown to derive from basic linguistic properties, thus making it unnec-

TSL[v.20020404] Prn:30/05/2007; 10:09

F: PB16101.tex / p.5 (5)

Connectives as discourse landmarks

essary to distinguish pragmatic “effects”, semantic “core meanings”, or syntactic position when discussing connectives.

Connectives and modality Connectives are known to have different uses and meanings, which raises the issue of the relationship between semantics and pragmatics. In this respect both the problem of polysemy and the ways to tackle it are similar to what can be observed in the treatment of modals. By assigning a single meaning to connectives and modals alike, unitary analyses tend to rely on the context and on pragmatics for the interpretation of meaning. Following Récanati (2004), Raphael Salkie argues that a prototype approach helps to clarify the semantics-pragmatics distinction and applies it to the study of rather. Rather can be used in the modal phrase would rather; it can also be used as a discourse connective or a degree modifier. Whereas previous analyses (such as Bolinger (1972), Thompson (1972) and Gilbert (1989)) have failed to convincingly connect all these uses, a prototype-based analysis can help to make a distinction between primary pragmatic processes and secondary pragmatic processes. Like modals (Huddleston & Pullum 2002), connectives have to meet a set of criteria (Fraser 1999 and Rouchota 1996). The first criterion set out by Fraser, namely that connectives “make explicit a connection between p and the context (linguistic or non-linguistic)”, refers to a function that connectives and modals have in common. Crucially, connectives “narrow down the range of possible interpretations of a proposition, or the relationship between two propositions”. Other shades of meaning, such as the contrast sense that may arise when an assumption is denied, result not from rather, but from the relationship between the two propositions contrasted. Karin Aijmer discusses the rhetorical function of the fact is that and compares this structure with other “shell noun phrases”, as Schmid calls them, such as the truth is that, the thing is and the trouble is. Using spoken corpus data taken from the BNC, she shows that all these phrases belong to the same semantic category, signalling the speaker’s position vis-a-vis the preceding context or the hearer. These phrases can have several variants, including ungrammatical ones (fact is). They should therefore be regarded as instances of a “collocational framework” (Renouf & Sinclair 1991). As pre-front field constituents, these phrases may have either a connective framing function or a metapragmatic – possibly modal – one. These functions go “beyond shelling a propositional content”, which supports the view that the complement clause is not semantically subordinate and conversely that the main clause – i.e. the fact is – is pragmatically subordinate to the that-clause. Given that the lexical meaning of this shell noun has also undergone semantic bleaching, allowing a process of both routinization and subjectification to take place, there



TSL[v.20020404] Prn:30/05/2007; 10:09



F: PB16101.tex / p.6 (6)

Agnès Celle and Ruth Huart

is reason to believe that this phrase is becoming grammaticalized, unlike other phrases such as the reason is that. Aijmer thus shows that syntactic and semantic adjustments take place jointly, affecting certain lexical items but not others. New pragmatic functions arise out of these variations, as speakers condense specific patterns into new units of “meaning”.

From syntax to pragmatics The sort of subtle transformation just described is studied in more detail in Part 2 with respect to three grammatical items whose “canonical” uses have given way to special speaker-determined functions in some contexts. Approaching the markers from different angles – syntax for and, stylistics for which and diachronic evolution for after all – the authors show how each has developed a specific role in structuring texts. In Mark De Vos’ article, it is shown that connectives can enhance cohesion within event structure. By defending the idea that grammatical cohesion is performed at a syntactic level (Gutwinsky 1976; Martin 1992), this approach runs against the view that cohesive connectives operate at a supra-syntactic, textual level (Halliday 1973; Schiffrin 1987). De Vos focuses on connective and used in pseudocoordination (sit and watch) and in reduplicative constructions (he laughed and laughed). Several syntactic tests make it clear that pseudo-coordinative constructions differ from ‘ordinary’ coordination, while behaving in a similar way to reduplicative coordination. Although it is to some extent semantically bleached, the pseudo-coordinative verb (go, sit etc.) retains “a specific flavour of meaning” which as such characterizes the nature of the event referred to by the lexical verb. Aspectual sub-stages of event structure are thus brought into focus, as in reduplicative coordination. Pseudo-coordination may therefore be viewed as involving a true coordinative connective within a complex predicate. Typical properties of coordination such as additive and ordering functions serve here to encode coherence within event structure. Conversely, pragmatics is shown to be able to constrain syntactic structure in reduplicative coordination. For Rudy Loock, it is the relative pronoun which that is taking on a new function as it loses its pronominal nature, along with some of its strictly anaphorical value in “non-standard” relative clauses, which Loock prefers to call “atypical” in order to avoid value judgements which some linguists proffer (Biber et al. 1999; Huddleston & Pullum 2002). A characteristic feature of relative clauses is that the relative pronoun can assume different functions within the relative clause where a gap is left by WH- movement. In atypical appositive relative clauses, however, two structures are possible which both deviate from this syntactic norm. In the gap-filled structure, either a resumptive pronoun or even a ‘resumptive phrase’

TSL[v.20020404] Prn:30/05/2007; 10:09

F: PB16101.tex / p.7 (7)

Connectives as discourse landmarks

is used in place of the gap in the relative clause, in spite of WH- movement. In the gapless structure, no gap is apparent in the appositive relative clause, and the relative pronoun does not seem to have an antecedent. In standard appositive relative clauses, the relative pronoun has both an anaphoric and a connective role (Jespersen 1970), whereas it only has a connective function in non standard appositive relative clauses, the anaphoric function being either irrelevant in gapless structures or fulfilled by the resumptive item in resumptive structures. Interestingly, which is the only relative pronoun to have this solely connective function, which raises the question of the nature of the link between the main clause and the appositive relative clause. Following Arnold (2004), Loock argues that appositive relative clauses function syntactically like subordinate clauses, but are semantically interpreted like independent clauses. The use of which is nonetheless motivated by discourse: by using a subordinate clause, the speaker adopts a ‘floor-keeping strategy’ to avoid interruption, as if his/her argument needed further elaboration. In this way, the speaker may surreptitiously add a new argument. The last paper of this section addresses the issue of the syntactic weakening of the prepositional phrase after all along with its diachronic emergence as a connective. After all has three distinct senses in present-day English: a temporal sense, a counterexpectation sense, and a justification connective sense, claimed by Diana Lewis to result from a metonymic evolution. Although this expression developed from a temporal usage, it is worth noting that it collocated with contrastive contexts as early as the 16th century. As a result, it easily acquired a contrastive sense over the following centuries, while simultaneously lexicalizing into an expression meaning ‘in the end’. At the end of the 17th century, the contrastive rhetorical pattern and the conclusive one seem to have been existing side by side. It is only recently that initial after all came to act as a solely conclusive connective marker. There is no evidence that older senses metaphorically developed into newer ones. Instead, Lewis argues that the counter-expectation marker and the justification connective gradually developed from related senses and contexts.

Discourse strategies The importance of context for the interpretation of connectives, leading to their specialisation in certain argumentative functions, is illustrated in the next three papers, which examine language in interaction. In written language, clause-combining connectives mainly have a logical function. Corpora of conversational English provide new insights into the role of connectives in spoken language. In spontaneous conversation, connectives seem to depart from their primary semantic meaning and shift to a discourse function. Connectives may be seen as part of a discourse strategy used by the speaker to put



TSL[v.20020404] Prn:30/05/2007; 10:09



F: PB16101.tex / p.8 (8)

Agnès Celle and Ruth Huart

forward arguments and symbolically negotiate with the hearer. As discourse markers, connectives may build on or even modify expectations formed in the previous context and at the same time assist in moving the text forward. In her analysis of well and you know, Barbara Le Lan defends the idea that these markers are “antilinear” in that they allow a speaker to introduce subjective comments into an otherwise linear message in a structured way. In the wake of Schiffrin (1987) and Aijmer (2002), she argues that the indexicality of discourse markers accounts for their multifunctionality. The discourse function of you know and well can be traced to the semantic and syntactic behaviour of these markers. Both discourse markers implicitly refer to views other than the speaker’s and at the same time reinforce the speaker’s involvement, thus creating an interaction between the context and the utterance they occur in as well as between speaker and hearer. The dynamic character of this interaction may be construed as a mental movement from the hearer’s point of view to the speaker’s in the case of well or from the speaker’s point of view to the hearer’s in the case of you know. Frédérique Passot, analysing the apparently circular pattern A because B so A’, is also concerned with the use of connectives to move discourse forward. Although this pattern – be it binary or ternary – involves segments which are redundant in terms of informational import, it is shown to have a spiral configuration which has an effect on discourse. The main clause precedes the subordinate clause introduced by because or cos in about 95% of the utterances provided by Passot’s corpus data, which indicates that a broad scope relationship tends to obtain between the subordinate clause and the main clause. Given that these clauses are separated by a long pause and function as independent information units, it is even questionable whether the syntactic relationship is still one of subordination, let alone causality. In like fashion, the relationship between because B and so A’ is not necessarily logical. The outcome of this loop is not a relationship of identity between A and A’, but rather a qualitative shift on the semantic, referential or modal level, which gives credence to the spiral hypothesis. Although the structure uses markers of connection, it has features in common with paratactic structures and as such contributes to the organisation of discourse. Indeed, the spiral pattern provides discourse with a dynamic principle whereby shared knowledge is increased as new developments are turned into background knowledge. These connectives function as discourse landmarks since they allow the conversation to be carried forward by constantly updating speakers’ consensus. Ruth Huart discusses the discourse function of initial not that with respect to the preceding context and compares this structure with the apparently more “complete” it’s not that. She shows that not that cannot be considered as a true concessive marker, but rather as a metalinguistic rectifying marker by which the speaker refutes the inference that the interlocutor might make from the preceding utterance. It is the justification for this speech act that is inferred and denied by

TSL[v.20020404] Prn:30/05/2007; 10:09

F: PB16101.tex / p.9 (9)

Connectives as discourse landmarks

not that. In terms of discourse strategy, this structure may indicate that semantic adjustments are necessary before the right choice of words can be endorsed by the speaker. The structure may also be used as a transition to move on to another topic. By contrast, it’s not that refers to the state of affairs that triggered the previous utterance. As with initial not that, an inference is denied, but this inference concerns the propositional content of the that-clause and not the motive for the previous utterance. The three papers do more than just describe the pragmatic effects of the forms: they attempt to offer explanations that stem from the basic values of the markers, giving each combination a specific function in verbal exchanges.

In search of operations This effort is carried a step further toward abstraction in the final section, where argumentative uses of several common connectives are shown to correspond to underlying linguistic operations. While the multifunctionality of connectives is a well-established fact, there is no consensus on how to account for it. The respective roles of the connective and of the context it occurs in need to be closely examined, and may be viewed differently according to the theoretical approach adopted. Using the Theory of Enunciative Operations, Martine Sekali shows that indirect meanings ‘are not encoded in the context prior to the connection, but are linguistically constructed’ in a dynamic process through the association of the connective with the connected utterances. The meaning of the connective itself then depends on the connected relations. The coordinator but is shown to establish not a binary relation, but a three-term relation by simultaneously creating and rejecting an implicit relation of subordination. Hence it may be taken as a counter-subordinator opposing speaker’s and addressee’s points of view in an intersubjective modal relation. Sekali further argues that the argumentative force of the utterance containing but results from this core linguistic operation of countersubordination, thus integrating into linguistic analysis what is often considered as a pragmatic effect. Within the same theoretical framework, Graham Ranger shows how the different uses of yet and still are linked. According to the properties of the notional domains connected, aspectual, focal or argumentative meanings may be derived. In the case of yet, an offline position is constructed in opposition with a preconstructed notional representation, whereas still constructs an occurrence continuous with a preconstructed domain existing prior to utterance. Here again, the argumentative use is not viewed as a separate pragmatic effect, but as the result of linguistic operations. As in the aspectual and focal uses, a concessive utterance



TSL[v.20020404] Prn:30/05/2007; 10:09



F: PB16101.tex / p.10 (10)

Agnès Celle and Ruth Huart

starting with still constructs a position continuous with a preconstructed position, while a concessive utterance starting with yet is notionally discontinuous with a preconstructed position. The tools used in this description may be applied to other apparently heterogeous linguistic phenomena, the theoretical framework being potentially generalisable. In an attempt to account for the multifunctionality of discourse connectives, François Nemo rejects the claim that discourse connectives function as discourse markers which as such suffice to reveal the nature of the relation between discourse units. Instead, he holds the view that two distinct connective links interact, one emerging from the utterance itself, and the other resulting from the encoded meaning of the morpheme associated with the connective. This theoretical hypothesis is then tested against the analysis of connective and non-connective even. Using an Indicational-Indexical Semantics approach, Nemo captures the encoded meaning of the morpheme even, which he considers to remain the same whatever its uses (as an adjective, a verb or a connective). According to him, even always indicates that there is no difference between two points. When using even as a connective, the speaker simply flattens any difference between two alternatives. The search for abstract representations of the operations underlying all uses of a given marker and making each one distinct from any other should eventually obviate a number of controversial questions, such as those concerning polysemy and grammaticalization already mentioned, the distinction between semantics and pragmatics, and even perhaps traditional divisions into “parts of speech”.

References Aijmer, K. 2002. English Discourse Particles. Evidence from a corpus [Studies in Corpus Linguistics 10]. Amsterdam/ Philadelphia: John Benjamins. Aijmer, K & Stenström, A.-B. (eds.) 2004. Discourse Patterns in Spoken and Written Corpora. Amsterdam/ Philadelphia: John Benjamins. Arnold, D. 2004. “Non-Restrictive Relative Clauses in Construction-Based HPSG”, in Proceedings of the 11th International Conference on Head-Driven Phrase Structure Grammar, Stefan Müller, ed., CSLI Publications, Stanford, CA. Benveniste, E. 1966, 1974. Problèmes de linguistique générale, vol. 1 & 2. Paris: Gallimard. Biber, D., Johansson, S., Leech, G., Conrad, S. & E. Finegan. 1999. Grammar of spoken and written English. London: Longman. Bolinger, D. 1972. Degree Words. The Hague: Mouton. Fraser, B. 1999. “What are discourse markers?” Journal of Pragmatics 31: 931–952. Gilbert, E. 1989. “Quite, rather.” Cahiers de recherche, grammaire anglaise 4: 5–61. Guillaume, G. 1964. Langage et science du langage. Paris: Nizet. Halliday, M. 1973. Explorations in the Functions of Language. London: Edward Arnold. Hansen, M.-B. Mosegaard. 1998. The Function of Discourse Particles. A Study with Special Reference to Spoken Standard French. Amsterdam/Philadelphia: John Benjamins.

TSL[v.20020404] Prn:30/05/2007; 10:09

F: PB16101.tex / p.11 (11)

Connectives as discourse landmarks

Hopper, P.J. & E. C. Traugott. 1993. Grammaticalization. Cambridge University Press. Huddleston, R. & G. Pullum. 2002. The Cambridge grammar of the English language. Cambridge: Cambridge University Press. Jespersen, O. [1927] 1970. A Modern English Grammar on Historical Principles. Vol. III. Londres: George Allen & Unwin. König, E. 1991. The Meaning of Focus Particles. A comparative perspective. London: Routledge. Östman, J.-O. 1981. You Know: a discourse-functional study. Amsterdam/Philadelphia: John Benjamins. Östman, J.-O. 1982. “The symbiotic relationship between pragmatic particles and impromptu speech” in Impromptu Speech: A Symposium, N. E. Enkvist, ed., Turku: Publications of Abo Akademi Foundation, 78, 147–177. Renouf, A. & J. McH. Sinclair. 1991. “Collocational frameworks in English.” In English corpus linguistics. Studies in Honour of Jan Svartvik, K. Aijmer and B. Altenberg (eds.), 128–44. London: Longman. Rossari, C. 2000. Connecteurs et relations de discours: des liens entre cognition et signification [Collection Langage-Cognition-Interaction]. Nancy : Presses universitaires de Nancy. Rossari, C. et al. 2004. Autour des connecteurs. Bern: Peter Lang. Rouchota, V. 1990. “But: contradiction and relevance.” UCL Working Papers in Linguistics 2: 65–81. Schiffrin, D. 1987. Discourse Markers [Studies in Interactional Sociolinguistics 5]. Cambridge: Cambridge University Press. Thompson, S. 1972. “Instead of and rather than clauses in English.” Journal of Linguistics 8:237– 249. Warner, R. G. 1985. Discourse Connectives in English. New York/London: Garland Publishing, Inc.



TSL[v.20020404] Prn:26/03/2007; 14:47

 

Connectives and modality

F: PB161P1.tex / p.1 (13)

TSL[v.20020404] Prn:9/05/2007; 15:59

F: PB16102.tex / p.1 (15)

Connectives, modals and prototypes A study of rather Raphael Salkie University of Brighton, England

Few words both function as a connective and express modality, but rather seems to do so; it can also be used as a degree modifier. We show that the different uses of rather each have their own puzzling properties, using examples from BNC Baby. Previous attempts to explain what the different uses have in common are few and limited in scope. Using a prototype approach to connectives and modality, we argue that in all its uses, rather serves to narrow down the possible interpretations of an utterance. It is this pragmatic function, rather than a shared underlying meaning, which links the different uses. We also illustrate the development of rather into a verb in some varieties of English.

.

Similarities between connectives and modals

One of the central problems in the analysis of discourse connectives is the relationship between their meaning and their use – that is, where to draw the line between semantics and pragmatics. The meaning–use issue is particularly difficult if a connective has different uses (for example, contrast but [1a] vs. denial of expectation but [1b]); or if a word can be used as a connective but also in other ways, such as contrast still (2a) and yet (3a) vs. temporal uses (2b) and (3b): (1) a.

The relation between them, therefore, is not oppositional, nor limitrophe, but one of alterity. b. It all sounds rather complicated, but it does serve a useful function in ordering the almost countless number of species in the most economical way.

(2) a.

There is an immediate paradox here though: attempted murder may not involve the infliction of any harm at all, since a person who shoots at another and misses may still be held guilty of attempted murder. b. While the obvious colonial ties have been broken, the power still lies elsewhere.

TSL[v.20020404] Prn:9/05/2007; 15:59



F: PB16102.tex / p.2 (16)

Raphael Salkie

(3) a.

There is nothing to suggest that the innumerable fossil brachiopods had any other method of feeding. Yet it is obvious that there are a great variety of shapes (see above) among the fossil forms, many of them unmatched in living species. b. I haven’t eaten your dinner, it’s not ready yet.

(Except where indicated otherwise, all examples are taken from BNC Baby, a four million word subset of the British National Corpus, divided equally into fiction, newspapers, academic writing and spoken conversation). Any proposal that assigns a single underlying meaning (such as Rouchota (1990) and Blakemore (2000) on but; Crupi (2004) on still and yet) has to argue that more of the interpretive burden is carried by the context – that is, by pragmatics. Similar issues arise with modals, where each item typically has many different uses. Analyses which propose a single underlying meaning (e.g. Papafragou 2000) have the same consequence of requiring extra pragmatics. In a recent paper (Salkie, to appear), I argue that a prototype approach to modality helps to clarify some of these issues. Following a suggestion by Recanati (2004: 461) that the semantics-pragmatics distinction itself displays “prototypicality effects”, I argue here that a similar approach to connectives is also enlightening, and apply it here to a little-studied word, rather, which (uniquely to my knowledge) can be both part of a modal and a discourse connective. We start by distinguishing different uses of rather, and then investigate what the connective and modal uses of rather have in common.

. Uses of rather We can distinguish three uses: as a connective, with various sub-types; as a degree modifier; and the phrase would rather. The distribution of these three uses in BNC Baby is shown in Table 1. . Rather as a connective Examples (4)–(12) illustrate the different uses of rather as a connective – on its own (4)–(6), preceded by or (7), preceded by but (8)–(9), and followed by than (10)– (15): (4) Interestingly, ‘very few’ employers saw examinations as essential for specific jobs (ibid., p. 23). Rather, they were demanded most rigorously where further training involved FE or professional courses where the educational qualification was an entry requirement.

TSL[v.20020404] Prn:9/05/2007; 15:59

F: PB16102.tex / p.3 (17)

Connectives, modals and prototypes

Table 1. Occurrences of rather Occurrences

%

Connective rather or rather but rather rather than All connective:

(86) (44) (65) (804) 999

(5.1) (2.6) (3.9) (47.5) 59.1

Degree modifier Would rather might rather Total:

566 125 1 1691

33.5 7.4 <1 100

(5) Notably absent from the CBI’s understanding of the problems of inner cities was any notion that they should be met by national economic policy. The assumption, rather, was one of special, even isolated, blackspots that can be dealt with by localized attention. (6) A: You forgot to put new tapes in the recorder last night. B: What new tapes? A: New batteries rather. (7) These phrases do, however, have the ring of truth in an aesthetic sense – or rather there was an aesthetic process involved which had to some extent its own story. (8) The overall shift has not been so much from agriculture to manufacturing but rather a steady rise in the service industries, and a fall in primary industries. (9) It may not be low social class in itself which is associated with crime, but rather that low social class is liable to be linked with other factors, perhaps parental criminality, deprivation or inadequacy of some kind, which might predispose individuals brought up in certain social classes toward criminality. (10) In both, Gironella is exploring the characters of the sitters as suggested in their portraits rather than the formal, painterly qualities. (11) It is in such mediating contexts, rather than in any direct tie to capital through laissez faire, that the social history of Darwin’s science should be sought. (12) Indeed, overall women referred by GPs showed greater evidence of neurotic, rather than psychotic, problems and a greater predominance of social problems. (13) The 1983 general election result fueled rather than ended the controversy. (14) In other words, rather than set these antithetical approaches against each other as one might have expected, Foucault suggests that they are part of the same mutation.



TSL[v.20020404] Prn:9/05/2007; 15:59



F: PB16102.tex / p.4 (18)

Raphael Salkie

(15) In this context, we may recall that Althusser, rather than simply criticizing the notion of history as a totality as Foucault often tended to do, argued for the rearticulation of different histories within a decentred totality, on the assumption that history cannot do without one.

Sentence (4) shows rather sentence-initially, in (5) it occurs medially, and in (6) finally. The corpus contains few instances of sentence-final rather, all of them in spoken data (recall that a quarter of BNC Baby is spoken language). On its own, connective rather indicates a contrast which can either be absolute, as in (5) and (6), or nearly absolute as in (4). The effect of or rather, as in (7), is a less strong contrast. With but rather (8)–(9), the contrast is always with a negative element in the preceding text, as is normal with contrastive but (this is the but that corresponds to Spanish sino or German sondern). Here it is rather which seems to reduce the absolute contrast that would be expressed by but on its own. The expression rather than is the most common use of rather in the corpus, accounting for 804 (47.5%) of 1691 occurrences of the word. It is one of two rather expressions which raise a puzzling question about speaker choice between two constructions. We find rather than before a range of constituents, including a noun phrase (10), a prepositional phrase (11), an adjective (12), just the verb (13), or the entire predicate as in (14) and (15). Where the entire predicate is contrasted, speakers have a choice between using the base form of the verb (set in [14]) or the -ing form (criticizing in [15]). The choice seems to be partly influenced by the form of the contrasting verb: if this is an -ing form, then we usually find an -ing form after rather than, as in these examples: (16) For example, the manual labourer is unlikely to take up an energetic hobby such as cycling and he is likely to spend his holidays relaxing at the seaside rather than walking in the mountains. (17) So we have a story that may or may not be a story (it had already appeared in The People back in January, when it raised few eyebrows), and a news reporter who suddenly finds himself giving press conferences rather than attending them. (18) Another and at least equally powerful strand says no, because it wants to emphasise the distinctive value of women’s insights, and also the special bond of sisterhood which is seen as binding women together, rather than letting them be absorbed and assimilated into the wider human group. (19) Even at this early stage of contact between white and black in Britain, it is clear that she would have been building on already existing racial caricature and prejudice rather than creating an artificial tide of hostility against black people.

TSL[v.20020404] Prn:9/05/2007; 15:59

F: PB16102.tex / p.5 (19)

Connectives, modals and prototypes

The -ing form may be a gerund, as in (16)–(18), or part of the progressive be+ -ing construction (cf. 19). Speakers who tend to classify rather than as a phrasal preposition (perhaps by analogy with instead of, which is similar in meaning, is clearly a phrasal preposition, and normally takes the -ing form), presumably tend to prefer the -ing form. In (14) and (15) the contexts are similar, however, and it would be perfectly acceptable to use setting in (14) and criticize in (15).1 . Rather as a degree modifier When rather is used in this way, as in (20)–(25), another choice often faces the speaker: (20) It is this rather elusive second aspect which gives the book its definitive character. (21) She had made, circumstances had made, rather a mess of this love. (22) Huh, he’s a good, good lad, but he’s er he drinks rather a lot. (23) She is a piece in an international game of chess; she is gloomy and melancholic; she is also, rather more strongly than in English, unlucky, wretched, doomed. (24) When this is done, a rather different picture emerges. (25) But subsequent inquiries revealed rather a different story.

In (20)–(23), there is no choice about where to put rather: it has to directly precede the adjective elusive in (20), the entire noun phrase a mess in (21), the quantifier a lot in (22), or the adverb phrase more strongly in (23). On the other hand, where the noun phrase contains an indefinite article and an adjective as in (24)–(25), the speaker can put rather before the adjective (24) or before the article (25): the corpus contains 72 relevant instances of a rather, and 54 of rather a. As with our earlier discussion of rather than, it is hard to explain why speakers choose one

. In BNC Baby, rather than is tagged for word class in two different ways: either PRP (preposition) or CJS (subordinating conjunction). The intention seems to have been to distinguish cases where the phrase after rather than is nominal, as in (10), from those such as (13) where it is verbal. Unfortunately the practice does not seem consistent: example (11) with rather than in any direct tie to capital is classed as a CJS; and whereas in example (15), rather than simply criticizing is labelled a PRP, a similar example such as this one is identified as a CJS: (i)

Poststructuralism, which in its own way also takes part in that history of Western Marxism, differs only insofar as it foregrounds the implications of the theoretical difficulties involved rather than repressing them in pursuit of the unrealized ideal.



TSL[v.20020404] Prn:9/05/2007; 15:59



F: PB16102.tex / p.6 (20)

Raphael Salkie

alternative over the other, particularly when the same adjective is used as in (24)– (25). Table 2 shows the most frequent collocates to the right of degree modifier rather, excluding function words: Table 2. Most frequent words occurring after rather in BNC Baby collocate

instances

more like different less nice good similar rare

48 28 20 14 10 7 6 5

. Rather as a part of a modal The expression would rather has various structures: (26) She says she would rather die than do that again. (27) Shamed Souness said: ‘I’d rather not talk about the goals we gave away and concentrate instead on the pluses .’ (28) Alex Mair said: ‘I’d rather you waited. (29) And I’d rather you eat it than make a mess of it eh? (30) Would you rather me do it on my own?

This use of rather lacks the stylistic markedness of the connective and degree modifier: the football manager Graham Souness sounds plausible using I’d rather in (27), but would be unlikely to say “that was rather a good goal”, which has an effete quality to it.2 In examples (26) and (27), we find the bare infinitive after would rather, a characteristic restricted to the modal auxiliaries in English. . A referee did not agree that this usage is effete. No doubt there is some social variation in such judgements. As a small test, I looked for examples of rather in columns written by the comedian and left-wing activist Mark Steel for the Independent newspaper. Mark Steel tries to maintain a working-class London accent in his speech, and to write accordingly. Of 36 instances found by Google, 25 (69.5%) were rather than, 7 (19.5%) were would rather, and 4 (11%) were the degree modifier. Compare this last figure with 33.5% for the degree modifier in BNC Baby. What’s more, three of the four degree modifier uses were in invented quotes from “posh” sources which Steel was deliberately parodying: Lord Hutton, Tony Benn, and very old-fashioned books about sex. Only in one instance was Steel writing as himself:

TSL[v.20020404] Prn:9/05/2007; 15:59

F: PB16102.tex / p.7 (21)

Connectives, modals and prototypes

Only two other expressions behave in this way: had better and may/might as well. It is interesting that both of these other marginal modal expressions, like would rather, involve a comparison – clearly the case in had better, and arguably so with may/might as well (cf. Mitchell 2003). A point of contrast is that would rather and may/might as well both involve modal auxiliaries which take a bare infinitive on their own: this is not the case for had better. One thing that sets would rather apart from the other two is its ability to take finite clause complements as in (28), and infinitives with an overt subject (29)–(30). The verb waited in the complement of (28) is a modally remote past tense, which makes would rather part of the small group of expressions like it is high time, and some uses of wish, which normally require this type of past tense. Here are some more examples: (31) I’m sorry, Mother, I’d rather you said yes, because in any case I’m going. (32) I’d rather you went after it with my support than behind my back. (33) You’d rather I died, I know. (34) ‘Was there someone else around you would rather she had married?’ he asked bluntly, deciding it was that sort of conversation. (35) She didn’t want to hear any more. Of course she did not believe a word Sean O’Farrell was saying. It was nothing more than servants’ gossip. Even so, she would rather it remained unspoken. And even supposing – just supposing – that any of it was true; well, it would be quite different in her own case.

Examples (34)–(35) are different from the others because their context is a narrative past time one. The verbs in (29) and (30) do not use the modally remote past, though eat and do could be treated as base forms rather than present tense forms. There is a similar example which appears to be a base form (or a subjunctive): (36) A: Yeah, they like to roam about and explore and and, and er, I’m not sure I wanna cope with litter trays for ever, would you? B: No. A: Would rather a cat be out cos it wants to be out. B: Yeah. A: Does as it wants to do when it goes outside rather than inside.

(i)

And Widgery did not even have to make his script work dramatically for television. Which is rather a shame. . .

A more thorough sociolinguistic investigation is clearly needed here. Notice also that the other connective uses – rather, but rather, and or rather – are absent from Steel’s columns, which may indicate that they too are felt by some speakers as effete.



TSL[v.20020404] Prn:9/05/2007; 15:59



F: PB16102.tex / p.8 (22)

Raphael Salkie

The verb prefer, close in meaning to would rather, can also take a modally remote past tense, though only one instance occurs in the corpus: (37) I’m conscious of not abiding by rules, in saying all this. You’d much prefer that I didn’t.

There was one occurrence of rather with the modal might: (38) This suggests he is worried that those who can might rather walk the dog or paint the house than work for minimal benefit.

This seems to be calqued on the use of than after would rather, as in: (39) Eating spaghetti, Clare began to suspect that James would rather talk about a relationship than have one.

The next example shows that the choice of which verb form to use after rather than also applies to would rather . . . than: (40) they’d rather do that than having to piece twelve quarter then you’ve got to have join the bit then all the pattern that will be matching exactly er you’ll use more

. Would sooner An alternative to would rather, which appears to share the same meaning and use, is would sooner: (41) ‘I’d sooner lie down,’ Charlotte muttered. (42) I don’t like it, I’d sooner it be like this you know I mean

In non-standard varieties we find other expressions too: (43) I’d just as soon stay in a small pub. (44) “Being so,” said he, “I would as lief they didnae see me”. (Stevenson, Kidnapped)

For our present purposes we can note that sooner shows no tendency to turn into a connective or a degree modifier in the way that rather has: we could not substitute sooner in examples (4)–(25). In BNC Baby we can distinguish six different uses of sooner as shown in Table 3. Apart from examples like (45), all these uses are to a greater or lesser extent multi-word expressions. (45) If he had been alert to everyday things he knew he would have heard it sooner. (46) ‘By next weekend, if not sooner.’

TSL[v.20020404] Prn:9/05/2007; 15:59

F: PB16102.tex / p.9 (23)

Connectives, modals and prototypes

Table 3. Uses of sooner in BNC Baby Use

Frequency

Examples

time adverbial sooner or later the sooner . . . the better the sooner . . . the X no sooner had / was / would would sooner Total frequency

9 22 10 5 9 9 64

45, 46, 47 48, 49 50 51, 52 53, 54 43, 44

(47) Foreign Office chiefs last night said the latest UN crackdown would happen ‘sooner rather than later’. (48) But sooner or later writers ask whether geography really is a human study at all, possibly because some writers on human geography appear to have studied almost everything except the people. (49) ‘It’ll have to happen sooner or later. (50) ‘I know you are only trying to help but the sooner we get this over with the better. (51) ‘In which case, Master, the sooner we begin, the sooner we finish!’ (52) At the price it is, the sooner I can find a buyer for it the happier I’ll be.’ (53) No sooner had Edward attended to one lot than he was seized upon by another. (54) But no sooner was Mr Major off his battlebus than a couple of eggs took flight.

. What links the different uses of rather? Several scholars have suggested connections between the various uses of rather. Interesting though these suggestions are, none is fully convincing. . Bolinger on degree words Bolinger (1972) is primarily concerned with the use of rather as a degree modifier. He makes a useful distinction between the intensifying use of degree modifiers (cf. rather elusive in (20) above), and the identifying use of such words (as in the connective uses of rather on its own in examples like (6)). Bolinger notes that many English words have both uses: for instance, such and so can be used as intensifiers (so elusive / He is such an idiot) or to identify (I told you so / I don’t know such a person). In the intensifying use of such, the speaker indicates the degree to which



TSL[v.20020404] Prn:9/05/2007; 15:59



F: PB16102.tex / p.10 (24)

Raphael Salkie

“he is an idiot”; but as an identifier, it is not the degree of personhood that is at stake but the mere fact of being a person of a particular type. The “preference” sense of would rather can be linked to the intensifying use of degree modifier rather. The same is true of the preference sense of rather than in examples like (55) from Thompson (1972: 242): (55) I went to the movies rather than study.

Thompson points out that this has the sense “I went to the movies so I wouldn’t have to study”, a sense which we do not find if we substitute instead of studying. Bolinger suggests that intensification has to do with quantity, whereas identification relates to quality, but he does not develop this suggestion; and nor does he attempt to explain the full range of uses of rather as instances of the intensifying – identifying relationship. . Denial of assumption rather Thompson (1972) contrasts examples like (55), where the verb after rather than is non-finite, with examples like (56) which have a finite verb: (56) It rained rather than snowed.

She notes that “these sentences are generally used to deny an earlier assumption” (1972: 243). Such examples are rare in the BNC (cf. example (13) above), but a few more can be found: (57) I sensed, rather than heard, my mum come in. (58) In its external elements then, the war widened the Unionists’ confidence and appeal rather than created it, lining up the nation with the party rather than the reverse. (59) And in the second version he introduces a three-dimensional hand not, as in the Velázquez with the fingers straight and limp, but drawn up into an uneasy gesture, that claws at her skirt rather than rests against it. (60) Miranda taller, with her bushy hair and colouring that the Italians whose paintings she’d been looking at in the Louvre rendered by priming the canvas with a copper-based green paint, creating a complexion that draws light in rather than gives it out; Xanthe beside her with her candy radiance of pink and gold, and rounder too, more neatly assembled, wrist to hand, neck to shoulders, ankle to foot.

Thompson notes that in (55), but not in (56), the rather than phrase can be preposed: (55’) Rather than study, I went to the movies.

TSL[v.20020404] Prn:9/05/2007; 15:59

F: PB16102.tex / p.11 (25)

Connectives, modals and prototypes

(56’) *Rather than snowed, it rained.

She proposes a “tense replacement” transformation which applies differently to (55) and (56). Dieterich and Napoli (1982) suggest a different approach, arguing that the rather than construction is a sentence modifier in (56) but a VP modifier in (55). They treat would rather constructions like (61), where there is an explicit than, as similar to (55) and as part of the VP: (61) Hal would rather go to the movies than study.

Most of their paper deals with the syntactic properties of rather, not with its semantics. They do claim, however, that denial of assumption rather than (56) has the truth functional properties of “and not” (as in “p and not q”). . Rather as an underlying comparative Gilbert (1989) is the most comprehensive study of rather in the literature. He argues that rather always marks the same operation: “étant donné les deux termes d’une alternative, rather indique la prééminence de l’un sur l’autre” (1989: 38).

Gilbert starts with the different structures with rather where an explicit than makes this analysis plausible; he then argues that the same operation is at play even when there is no apparent contrast between two alternatives, notably in the degree modifier use. The suggestion is that even in these cases, the speaker operates on two alternatives I and E, the interior and exterior of the notional domain expressed by whatever rather modifies. So in (21), reproduced here, I is “a mess” and E is “not a mess”: (21) She had made, circumstances had made, rather a mess of this love.

Gilbert notes that degree modifier rather can either have the sense of “somewhat” or be equivalent to “very”, as in (62) [his (122), (1989: 48)]: (62) “Don’t you love to feel the fire on your skin?” he said. “It is rather pleasant,” said Gerald.

His proposal is that in (62), the alternatives are not I and E but “du positif et de l’encore plus positif ”; in this case, I is “une valeur attendue, mais dans le sens de préconstruite et non de désirable” (1989: 48). The sense of “very” results from confirming that the presupposed or assumed sense is in fact correct. This is a subtle analysis, but it remains unclear why rather can both confirm an existing assumption as in (62), and deny it as in (56), repeated here: (56) It rained rather than snowed.



TSL[v.20020404] Prn:9/05/2007; 15:59



F: PB16102.tex / p.12 (26)

Raphael Salkie

. Using prototypes It is commonly accepted by linguists of various theoretical persuasions that the notion of “prototype” is a useful way of thinking about grammatical categories. In typological studies this idea has taken firm root, as noted by standard handbooks such as Comrie (1989: 37–38 and 106–110) and Croft (1990: 66–67 and 124–127). The prototype strategy was explicitly proposed for English grammar by Huddleston (1984: 72), and is adopted in a number of recent grammars of English, such as Greenbaum (1996: 92) and Biber et al. (1999: 59). Notions like “connective” and “modality” are not straightforward grammatical categories, but I believe that a similar strategy is useful in these areas too. For modality, this view is shared by Huddleston and Pullum (2002), who say that “modality . . . is not sharply delimited or subdivided, so that we shall need to make reference to the concept of prototypical features and to allow for indeterminacy at the boundaries of the categories” (2002: 172). In Salkie (to appear) I propose a set of criteria which are intended to be met in full by expressions which are core or prototypically modal. These are intended as universal criteria and are primarily semantic, as one would expect for crosslinguistic generalisations. A language may also have its own additional morphosyntactic criteria for modal expressions. A. Express possibility or necessity. B. Epistemic or deontic. C. Subjective, involving (i) commitment by the speaker. (ii) primary pragmatic processes. (iii) a sharp distinction between the modal expression and the propositional content. D. Located at one of the extremes of a modal scale.

Criterion C requires some comment. The term “subjectivity” has been used in many different ways in studies of modality, notably Lyons (1977) and Verstraete (2001). In Salkie (forthcoming) I suggest that Recanati’s (2004) work on the semantics-pragmatic interface is helpful in distinguishing between primary pragmatic processes which are obligatory and narrow down the range of possible meanings, and secondary pragmatic processes which are optional and broaden the range of possible meanings. Consider this sentence: (63) I work at Brighton University.

A primary process enables the hearer to assign a referent to the subject pronoun. A secondary process is responsible for interpreting “work”, if the sentence is uttered by me, as “lecture / teach” rather than “serve in the canteen”. The first process is

TSL[v.20020404] Prn:9/05/2007; 15:59

F: PB16102.tex / p.13 (27)

Connectives, modals and prototypes

obligatory, whereas the second one is optional and, indeed, defeasible. My claim is that the interpretation of core modals involves only primary processes. Turning now to connectives, we can start with the relatively uncontroversial criteria set out in Fraser (1999), who proposes that connectives:3 A. Make explicit a connection between p and the context (linguistic or nonlinguistic). B. Are outside the main grammatical structure of sentences. C. Express contrast, inference, elaboration, topic relations, etc. D. Are formed from words which have other meanings or uses.

Some more contentious criteria are suggested in particular by Rouchota (1996): E. F. G. H.

Encode procedural meaning, not conceptual meaning. Link an utterance and a particular inferential process. Subjectivity. Metarepresentational.

This is not the place to unpack the contentious issues, but criterion G is relevant here. The subjectivity of connectives, I would argue, is connected to Fraser’s criterion A: by making explicit the link between text and context, connectives perform a similar function to modals. The crucial point is that connectives narrow down the range of possible interpretations of a proposition, or the relationship between two propositions, by directing the hearer to the one that the speaker intended. This gives us a promising way to analyse rather. I propose that it is this narrowing down of possible interpretations which rather explicitly signals. As a degree modifier, this may involve narrowing to a limited amount of the quality expressed by the item modified (21), or it may involve narrowing to an area where the quality is strongly in evidence (62). Bolinger’s distinction between identification and intensification results from a narrowing of quantity or of quality. As a connective, there must be some relationship between the two situations contrasted when rather is used: but this is a property of all contrast relationships, not a property specific to rather, as Asher & Lascarides note (2003: 168). The denial of assumption sense of (56) arises for this reason. As for preference, which we saw in (55), this can be seen as narrowing when there is a human agent as subject. In the expression would rather, the preferential narrowing of rather combines with the residual volition sense of would. Although volition and preference are not part of the core of modality according to the criteria proposed above, the element of subjectivity which they express is similar enough to the subjectivity of rather to make it possible for them to combine. My proposal, then, is that it is their similar . Fraser uses the term discourse marker, as does Rouchota (1996). The label is not a significant issue here, however.



TSL[v.20020404] Prn:9/05/2007; 15:59



F: PB16102.tex / p.14 (28)

Raphael Salkie

pragmatics which links would and rather, rather than their meaning. This is not to say that the emergence of would rather in English was necessary, but it does suggest the factors that made it possible.

. The future of rather To round off this study of rather, it is worth noting some changes that appear to be taking place in current English. In some American varieties, the word has been used as a noun for some time, usually in the plural: (64) But he chose to start and try to cover future of his own And he got his rathers because now his younger brother Is his father’s and his mother’s only son www.lyricsdownload.com/ray-price-jesse-younger-lyrics.html (65) I guess my rathers was that she should, considerin’; but I did wish it had been Anne, for she ain’t got nothin’ better in her than just to live genteel.” ... www.worldwideschool.org/library/books/lit/drama/OldportDays/chap7.html

The form druthers is also attested as a dialect word: (66) If I had my druthers, MS would stay to their own course. They are just too big a monkey to share the bathwater. static.userland.com/userLandDiscussArchive/msg011453.html

More recently, we find rather used as a verb, as noted by Juge (2002) who gives several examples of rathered such as: (67) LBJ much rathered to fight a war against poverty . . . but performing in the shadow of an assassinated president did not afford him that luxury. www.coe.uh.edu/hypergroups/courses/hist3322/0043.html

Other verbal forms can also be found on the web: (68) My 2 year old likes this movie, but even she rathers to watch the original. www.uniquegiftsale.com/index.php?Operation=CustomerReviews&ItemId= B00005JMB3&ReviewPage=2 (69) what shes tryin2 say rich.. is she rathers me www.dragonninja.co.uk/printthread.php?t=8809&page=114&pp=25 (70) Throughout her 20s, she mostly played in independent bands, and focused on songwriting. She has also written and performed music for children’s television shows, but remains secretive as to which shows in particular, rathering her fans to play the guessing game. en.wikipedia.org/wiki/KT_Tunstall [The latest version of this Wikipedia en-

TSL[v.20020404] Prn:9/05/2007; 15:59

F: PB16102.tex / p.15 (29)

Connectives, modals and prototypes

try has changed the last sentence so that it ends “preferring her fans to play the guessing game”] (71) Justine’s been married for seven years to the oafish Phil (John C.Reilly), but he doesn’t seem to cherish his time with wife anymore, instead rathering to veg out on the coach all day and smoke weed with pal, Bubba (Tim Blake Nelson). www.moviehole.net/reviews/487.html

It is noticeable that none of these examples shows the modal-like properties of would rather – any more than the non-modal verb prefer to which they are similar.

Conclusion To summarise the picture of rather that emerges from this paper, it is useful to compare our account with the one presented in the Oxford Dictionary of English (ODE) (Soanes & Stevenson 2003). The ODE is highly regarded, corpus-based, and organised so that each entry starts with the most important sense in the view of its compilers. The senses of rather are listed in this order: 1. Would rather 2. Degree modifier 3. Connective. The use on its own is given first, followed by or rather and rather than. From a statistical point of view, the data from BNC Baby in Table 1 suggest that this is exactly the wrong way round: rather than should come first instead of last, and would rather has the smallest number of instances. Conceptually, the core sense of “narrowing down” that we have proposed could be clearly indicated by putting rather than first, followed by the “limited quality” sense of the degree modifier. The final sense would be the “preference” meaning of would rather, which could be cross-referenced to the volition use of would and the connective use of rather. This would give an underlying unity to the entry for rather, which is something that ODE tries to do wherever possible. I have tried to show that the varied uses of the word rather repay close investigation, and that a prototype-based analysis can shed some light on what the different uses have in common. There is certainly more research to be done on this rather mysterious word, which I rather hope will receive from linguists more of the attention which it surely deserves.



TSL[v.20020404] Prn:9/05/2007; 15:59



F: PB16102.tex / p.16 (30)

Raphael Salkie

References Asher, N. & A. Lascarides. 2003. Logics of conversation. Cambridge: Cambridge University Press. Biber, D., S. Johansson, G. Leech, S. Conrad and E. Finegan. 1999. Longman Grammar of Spoken and Written English. London: Longman. Blakemore, D. 2000. “Indicators and procedures: nevertheless and but.” Journal of Linguistics 36: 463–486. Bolinger, D. 1972. Degree Words. The Hague: Mouton. Comrie, B. 1989. Language Universals and Linguistic Typology (2nd ed.). Oxford: Blackwell. Croft, W. 1990. Typology and Universals. Cambridge: Cambridge University Press. Crupi, C. D. 2004. But Still a Yet: The Quest For a Constant Semantic Value For English yet. Ed Doc dissertation, Graduate School of Education, Rutgers University. Dieterich, T. & D. J. Napoli. 1982. “Comparative rather.” Journal of Linguistics 18: 137–65. Fraser, B. 1999. “What are discourse markers?” Journal of Pragmatics 31: 931–952 Gilbert, E. 1989. “Quite, rather.” Cahiers de recherche, grammaire anglaise 4: 5–61. Greenbaum, S. 1996 The Oxford English Grammar. Oxford: Oxford University Press. Huddleston, R. 1984. Introduction to the Grammar of English. Cambridge: Cambridge University Press. Huddleston, R. & G. Pullum. 2002. The Cambridge grammar of the English language. Cambridge: Cambridge University Press. Juge, M. 2002. “Unidirectionality in grammaticalization and lexical shift: the case of English rather.” Berkeley Linguistic Society Proceedings 28: 147–154. Lyons, J. 1977. Semantics. Cambridge: Cambridge University Press. Mitchell, K. 2003. “Had better and might as well: on the margins of modality?” In Modality in Contemporary English, R. Facchinetti, M. Krug, & F. Palmer (eds.), 129–49. Berlin: Mouton de Gruyter. Papafragou, A. 2000. Modality: Issues in the Semantics-Pragmatics Interface. Amsterdam: Elsevier. Recanati, F. 2004. “Pragmatics and semantics.” In The Handbook of Pragmatics, L. Horn & G. Ward (eds.), 442–462. Oxford: Blackwell. [Also available on the web: http://jeannicod.ccsd. cnrs.fr/documents/disk0/00/00/00/91/index.html] Rouchota, V. 1990. “But: contradiction and relevance.” UCL Working Papers in Linguistics 2: 65–81. Rouchota, V. 1996. “Discourse markers: what do they link?” UCL Working Papers in Linguistics 8: 199–214. Salkie, R. (To appear). “Degrees of modality.” Submitted to Proceedings of ModE2. [Prepublication version available on the web: http://www.brighton.ac.uk/languages/research/ Salkie_Degrees_of_modality.pdf] Soanes, C. & A. Stevenson. 2003. Oxford Dictionary of English. Oxford: Oxford University Press. Thompson, S. 1972. “Instead of and rather than clauses in English.” Journal of Linguistics 8:237– 249. Verstraete, J.-C. 2001. “Subjective and objective modality: Interpersonal and ideational functions in the English modal auxiliary system”. Journal of Pragmatics 33: 1505–1528.

TSL[v.20020404] Prn:1/06/2007; 15:29

F: PB16103.tex / p.1 (31)

The interface between discourse and grammar* The fact is that Karin Aijmer Göteborg University

The fact is has variants such as fact is which are grammatically irregular and used with particular functions in spoken language. It is argued that the new structures should be analysed as emergent ‘syntactic gestalts’ shaped by their use in discourse rather than as grammatically compositional. In the analysis proposed the different variants belong to a collocational framework licensing some forms while disallowing other variants. The examples which are taken from the British National Corpus are analysed with regard to meaning, discourse function and prosody.

.

Introduction

This paper will be about patterns such as e.g. the fact is that, the thing is that, the evidence is that with different head nouns followed by is and a that-clause. The nouns occurring in this structural pattern have special properties. For example unlike ‘full-content’ nouns such as tree or book which refer to objects in the real world their reference is unclear and abstract. Schmid (2000: 4) calls the nouns occurring in this structure ‘shell nouns’. They function as ‘labels for their content’, ‘as indicators of what they contain’ and ‘supply propositions with conceptual shells which allow speakers to grab them and carry them along as they move on in discourse’ (2000: 13). For example, in the pattern ‘the fact is that’, the fact provides a shell for the content in the that-clause. Such expressions are sometimes regarded as awkward or inconvenient because they have variants which are grammatically irregular and they are used with pragmatic and discourse functions in speech. Biber et al. (1999: 1073–1076) for example describe them as utterance launchers or ‘ouvertures’ since they mark a new utterance or a new topic. * I am much indebted to Göran Kjellmer for commenting on an earlier version of the text.

TSL[v.20020404] Prn:1/06/2007; 15:29



F: PB16103.tex / p.2 (32)

Karin Aijmer

In this paper I want to argue that the fact is and related structures can be described as formulas and function as a rhetorical resource in interaction negotiating a position vis-à-vis the preceding context or the hearer. The relationship between the fact is that and (the) fact is + main clause will be analysed with regard to grammaticalization, although the perspective in this article is mainly synchronic. ‘The fact is that’ and ‘the truth is that’ are generally interchangeable. Other nouns in the same structure are evaluative rather than or in addition to modal (the trouble is, the thing is). The paper will focus on these phrases although other phrases belonging to the same lexical field could have been included as well. It will be shown that they have certain features in common suggesting that they are appearing as a semantic category. A large corpus is valuable because it gives information about variant forms and their distribution as well as frequencies. The material for this study is taken from the British National Corpus (BNC) mainly from spoken language (10 million words). There were 775 examples of the fact is, of which 120 tokens were from the spoken part. The truth is was less frequent (36 examples in the spoken part of BNC). The evaluative phrases were also frequent (e.g. (the) thing is that 161 examples; (the) trouble is that 20 examples). The paper is organised as follows. To begin with, I discuss structural patterns such as the fact is that, the truth is that, the thing is that and variations of the patterns. Along with the description of the forms I will analyse meanings, prosody, and the contexts in which the phrases occur.

. The formulaic ‘the fact is that’ The fact is (that) is formulaic, i.e. it has a fairly fixed form with a copula in the present tense. Shell noun phrases can also be analysed from the point of view of the nouns which are part of the structure. Fact belongs to the class of ‘nouns of simple fact’ (Halliday 1994: 267) and structures such as ‘the fact is that’ would be related to ordinary modalized propositions such as ‘it is a fact that’ or ‘it is the case that’ where the informationally rich structure comes at the end. The fact is draws attention to the that-clause which contains the main message. As Schmid (2000: 9) points out, ‘the noun phrases introducing such clauses ((the problem is (that), the trouble is (that), the fact is (that), the truth is (that)) are more or less redundant from a purely propositional point of view’ and the important work is being done in the complement clause. A paraphrase is ‘What is important is that. . .’ where ‘what is important’ marks the focus for extraordinary attention (Schmid 2000: 333). All shell nouns have not developed a focalizing function (Schmid, ibid). The reason is that only shares some features with the fact is that.

TSL[v.20020404] Prn:1/06/2007; 15:29

F: PB16103.tex / p.3 (33)

The interface between discourse and grammar

(1) A print of the right kind is not simply something that is worth money. It is more like actual currency. The reason is that while it may be in limited supply, it isn’t unique (A 39)

In (1) a causal relation is expressed paratactically between the propositions. It would be easy to use a hypotactic relation instead, e.g. to say ‘it is more like actual currency since it isn’t unique’. The nouns can have meanings such as cause, result or explanation rather than modality (fact, truth) or evaluation (thing, trouble). We can conclude that there are different types of shell nouns with different lexical properties and therefore different uses. There is not a sharp distinction between the two types however.

. Collocational frameworks The fact is that occurs in different shapes. (2) exemplifies the type ‘the fact is that’ with a full clause: (2) Yet the fact is that such attitudes were not shared by the conformist cultural establishment. (A 7M 1086)

When that is present the definite article must also be present. There were no examples such as *fact is that (a single example was found in the whole of the BNC). ‘The fact is’ can be followed by a zero-connective: (3) But the fact is I have done it all, up till about the first six months there and then I started getting well, well out of my league, you start getting the machining and stuff like that, I haven’t done much on them at all. (KDA 6752)

The comma indicates a pause: (4) Erm, the fact is, she didn’t play the dominant part in the standard tradition, although she was one of the editors of course (HUN 464)

When the is missing, fact is not usually preceded by an adjective (*Simple fact is, *simple fact is that). The structure in (3) and (4) can also be characterised prosodically. As pointed out by Schmid (2000: 338), ‘is’ receives the tone (falling tone) suggesting that the copula is no longer experienced as a copula. The collocation fact is is a reduced form: (5) Fact is, it’s precisely the too-thorough internalization of punk’s dread of the hippy and the resulting strategy of self-limitation that has led to a wholly different kind of dire stasis. (AB3 503)



TSL[v.20020404] Prn:1/06/2007; 15:29



F: PB16103.tex / p.4 (34)

Karin Aijmer

Fact is can be compared with the pragmatic markers you know, you see, I think. However unlike these markers it must occupy the first position in the utterance. ‘Truth’ can be used as an ordinary referential noun: (6) The truth is bland: Mr Major will call an election if the opinion polls suggest he cannot lose (ABE 1869)

‘Fact’ is a referential noun only in certain idioms: (7) That would make me some kind of accessory after the fact, wouldn’t it?” (FAP 3465)

The shell noun phrase ‘the truth is that’ is illustrated in (8): (8) The truth, the truth is, the truth is that more and more people and organisations are realising the benefits of local control of N H S hospitals, as in the Trust hospitals. (JJD 356)

As with ‘the fact is that’ we find variants of the truth is that. With a zero-connective: (9) the truth is they’re rushing through the debate because they know that in the budget the next month, the chancellor will announce further defence cuts. (K6D 16)

Truth is as a pragmatic marker is illustrated in (10): (10) Yes Chairer er I think Vince’s report is a very fair summary of the er the the issues which face us erm truth is, is what we’ve been as in this proposed consortium which no doubt will go through with it’s most of the reforms of the N H S despite the er consultation mechanism which they’ve gone through you know, but largely these are beforehand given the unaccountable of the N H S management! (F7V436)

A wider range of structural variation is permitted when we look at the nouns thing or trouble. Without the definite article: (11) Thing is that the furniture trade was really booming then, begin that was in the thir you know in the th er early thirties that was the (KGL 537) (12) thing is that men’s dicks should be as big as their hands. (KCE5243)

Without the definite article and with a modifying adjective: (13) Silly thing is that they actually believe it don’t they? (KDA 4524)

Trouble is occurred with or without a preceding adjective: (14) Trouble is, competition for passengers has never been so fierce and Virgin is still a tiny airline that’s losing money. (HE 624)

TSL[v.20020404] Prn:1/06/2007; 15:29

F: PB16103.tex / p.5 (35)

The interface between discourse and grammar

(15) Only trouble is, it’s me that pays four quid, you know, a quid here, quid back to lend him it, and then I’ve (KCX 1460) (16) They must have somewhere where we can play, but erm perhaps I ought to trouble is that you get fobbed off you know. (KBK6118)

The fact is that and related phrases can be described grammatically but many variants of this structure would be ungrammatical. In the analysis I will propose shell noun phrases such as the fact is that, the truth is that and their synchronic variants are examples of what Renouf and Sinclair refer to as the collocational framework phenomenon rather than an ad hoc combination of words. A ‘collocational framework’ (Renouf & Sinclair 1991: 128) is defined as a discontinuous sequencing of (grammatical) words enclosing a characteristic grouping of words. Renouf and Sinclair discuss for example the framework be+? +to which can ‘generate’ be able to, be expected to, be allowed to, etc. when the empty slot is filled. In the same way ‘Det+N+ is (that)’ is not a matrix clause but a semi-fixed unit or collocational framework consisting of grammatical words with a slot for a (shell) noun. Below I give examples of the collocational framework for ‘the fact is that’ and ‘the truth is that’. The (adj) truth is (that) S fact thing trouble etc.

It is clear that there is a large number of nouns which fit into this framework, for example suggestion, consensus, tragedy, difficulty. The latter group of nouns contribute some characterisation of what comes in the that-clause, for example that it is tragic, or difficult (cf. also ‘it is a tragic thing that’). However they do not have the focalizing function to draw attention to what comes next. The copula must be in the present tense. The trouble is points forwards to something new: (17) Going away next week. You are? Where you going? Oh south Wales. Mm, The trouble is there’s a hundred and one things to do before we go. teach you to go on holiday. (G5Y58)



TSL[v.20020404] Prn:1/06/2007; 15:29



F: PB16103.tex / p.6 (36)

Karin Aijmer

If the copula is in the past tense the noun is used anaphorically to refer back to the preceding text. We find for instance ‘the reason was’, ‘the problem was’, ‘the trouble was’ but not normally ‘the fact was’ or ‘the truth was’. (18) Anyway You see the trouble was that the twenty five percent business was falling, falling fast. People were getting educated to endowment assurance you see. (FY1679)

A collocational framework allows specific local transformations, deletions, additions. There is a slot in which a particular (evaluating) adjective can be inserted before the noun) (the BNC contains the simple fact is, the plain fact is, the salient fact is, the interesting fact is, the sad fact is, the most disturbing fact is) allowing a more fine-grained evaluation of the shell content. The frame can have an anomalous constituent structure (cf. Bybee & Scheibman 1999). As Tuggy suggests (1996: 725) the atypical intonation pattern with a stressed copula would favour [Subject [be X]] rather than [Subject be] X]. The formulaic character is particularly obvious in the reduced form ‘fact is’ where the pattern diverges from a grammatical structure and cannot be related to ‘the fact is that’ by a grammatical transformation. Collocations are processed as units. They are stored in the speakers’ long-term memory and produced and interpreted as units. A large number of patterns are derived from the frame rather than created compositionally in the communication situation. In this way, both high frequency regular forms such as ‘the fact is that’ and grammatically irregular forms would be represented in memory. The processing units which are stored are for example the fact is that, the fact is ø, fact is, the simple fact is. Related frames can be members of the same family. The frame below ‘generates’ the clefts ‘it is a fact that’, ‘it is a problem that’, ‘it is the case that’ which can be compared with the frame where the fact is the topic: It is a fact the case a joke etc.

that S

or with adjectives: It is ironic tragic etc.

that S

To sum up, the collocational framework represents different ways of saying the same thing but also imposes restrictions on the potential variation. It permits the fact is that or fact is but not (sad) fact is that. On the other hand, trouble is that should be acceptable as well as only trouble is that. We can assume that there is a

TSL[v.20020404] Prn:1/06/2007; 15:29

F: PB16103.tex / p.7 (37)

The interface between discourse and grammar

rationale behind different forms and that the ‘different ways of saying the same thing’ can be further described for example with regard to function, text type or social situation.

. The fact is and emergent grammar Just as verbs or adjectives can be followed by complements it can be assumed that nouns take a complement completing or specifying the meaning relationship implied by the preceding (matrix) clause (Quirk et al. 1985: 1231). My analysis of ‘the fact is that’ is based on the assumption that ‘the fact is that’ is a fixed or a semi-fixed unit which can be represented as a collocational framework. The framework provides us with a collection of ready-made forms each with a particular function and linked to the situational context. The fact is that is a matrix sentence followed by a that-clause but it also occurs loosely integrated in the rest of the sentence and in a reduced form. This is illustrated in Figure 1. The figure gives a simplified description of the topology of the declarative sentence (following Auer (1996) and the German tradition of distinguishing different topological fields). ‘a new sentence’ pre-front field

front field

middle field

end field

e.g. the fact is the truth is the facts of the matter are the truth of the matter is the thing is fact is, etc.

Figure 1. The pre-front field and the declarative sentence

The pre-front field constituents are pragmatic markers, i.e. they have a signposting or framing function. They are for instance used to encode information about how propositions are linked to each other. This is the connective framing function. Equally important is the metapragmatic function. This refers to the interpersonal (modal and evaluative) and rhetorical functions. For example, when the speaker uses the fact is this involves some emotional commitment which is balanced against possible opposition or counterargument. Here are a few examples of ‘things that look like a main complement-taking clause’. which could be regarded as pragmatic markers. In addition to the modal and evaluative shell noun phrases they include pseudo-clefts:



TSL[v.20020404] Prn:1/06/2007; 15:29



F: PB16103.tex / p.8 (38)

Karin Aijmer

Modal the fact is the truth is Evaluative The trouble is The problem is Pseudo-clefts (emphatics) What I mean is What I would like to say is

There are cognitive and iconic motivations for placing ‘the fact is’ and similar expressions in the pre-front field. For example, what looks like a matrix clause is separated from the rest of the sentence and occupies the same rank-level as the sentence which it points forwards to. From a cognitive perspective the function of these expressions goes beyond ‘shelling’ the propositional content. The shell noun phrases ensure that the context in which the utterance should be interpreted is present from the start by means of the framing or signposting function. Some additional support for this analysis comes from the ‘fragment analysis’ of I think that by Thompson (2002) described by Boye and Harder (unpublished) as ‘a usage-based attack on the classical subordination analysis’ since the structure is regarded as a by-product of usage and frequency. As in Auer’s analysis the structure is ‘flat’, i.e. there is emphasis on the linear arrangement of forms.. The suggested analysis reverses the hierarchical relationship between the clause containing that and the clause containing the content (the ‘complement clause’). Thompson argues for example that what is normally considered as a complement clause can be shown not to be ‘semantically’ subordinate and that the complementtaking phrase has been weakened. She distinguishes the following fragments in the analysis of a sentence introduced by I think (that): – a modal (epistemic/evidential/evaluative) fragment – a declarative clause which can be introduced by the linking that.

Evidence for this analysis comes from the development of the prosodically and grammatically integrated I think into a parenthetical clause by weakening of the epistemic meaning (Thompson & Mulac 1991). When we analyse the fact is that strengthening of the epistemic meaning is involved. The similarity between the two types of phenomena is that the main message is conveyed in what looks like a subordinate clause. The difference is that I think can be parenthetical and placed in medial and final position with regard to the complement clause. The fact is that on the other hand always occupies initial position. In the analysis chosen here the fact is that is analysed as a main clause on the structural level but is described pragmatically as weakened or as subordinate in relation to the that-clause. However

TSL[v.20020404] Prn:1/06/2007; 15:29

F: PB16103.tex / p.9 (39)

The interface between discourse and grammar

grammaticalization may achieve better isomorphism between form and pragmatic or discourse function. As a result ‘the fact is’ can occupy the pre-front field which is only loosely integrated with the rest of the sentence. The exact lexical and syntactic conditions for the use of a shell-noun phrase as a pragmatic marker in pre-front field position have been the topic for much discussion. For example the meaning of the noun is of major importance. The pragmatic marker has modal or evaluative meaning. We do not find the reason is, the explanation is, the evidence is as markers. The syntactic conditions are more difficult to describe. The presence of that shows that we have subordination. However the fact is no longer contains the main message. Consider sentences with a tag question. We can say for instance ‘the fact is that Uncle Tom is a famous character isn’t he’ where the tag question refers to the proposition and not to the matrix clause. To sum up, the fact is that is much less grammatical than other structures we analyse as main clauses. It has some features of a main clause with a shell noun. It is flexible (‘the simple fact is’) and it has the form of a clause. ‘The fact is that’ is on its way to becoming grammaticalized as shown by the fact that the information in the matrix clause is backgrounded.

. ‘The fact is that’ and grammaticalization The pre-front field is the end-point for grammaticalization, i.e. the elements which are found there can be diagnosed as showing symptoms of grammaticalization. It is evident that this applies to the fact is that but not to all shell noun phrases. One argument for grammaticalization is the lexicalization of ‘the fact is’ but also semantic and pragmatic factors are involved. The diachronic scenario strongly supports the hypothesis of ongoing grammaticalization. Fact as a lexical noun with concrete meanings such as ‘action’, ‘deeds’, etc. is obsolete (OED fact 1) except in phrases such as ‘after the fact’. However the noun is much alive in structures such as ‘the fact is that’ or ‘it is a fact that’ where its semantic meaning has been bleached. Diachronically grammaticalization operates on clines guiding the development of new meanings and forms. The development of ‘the fact is that’ can be described on the basis of Figure 1 in Section 5 which shows that grammaticalization implies a movement towards a loosely attached pre-front field constituent: Matrix clause → pre-front field constituent → pragmatic marker (cf. Auer 1996: 313)

The main clause moves out of the sentence. It is semantically weakened and ends up in the pre-front field slot. The new pre-front field constituent is pragmatically enriched which identifies it as a pragmatic marker. Concomitant processes associated with grammaticalization are increased bonding between contiguous el-



TSL[v.20020404] Prn:1/06/2007; 15:29



F: PB16103.tex / p.10 (40)

Karin Aijmer

ements (‘fact is’) and deletion (‘fact is that’) and routinization (Lehmann 1995). Routinization is the fading of existential-situational meaning that occurs when any action is constantly repeated until it becomes routine’ . . . . Our speech is a vast collection of hand-me-downs that reaches back in time to the beginnings of language. The aggregation of changes and adjustments that are made to this inheritance on each individual occasion of use results in a constant erosion and replacement of the sediment of usage that is called grammar.’ (Hopper 1987: 159; cf. Haiman 1991)

As a result of grammaticalization we find many structures which are not grammatical in the sense that they have a normal constituent structure (cf. fact is, trouble is that, only thing is). They have grammatical features which we associate with pragmatic markers such as the absence of integration in the following clause. Other features such as increased subjectivity are pragmatic. According to Traugott and Dasher, meanings become more subjective in the sense that ‘SP/Ws come over time to develop meanings for Ls [lexemes] that encode or externalize their perspectives and attitudes as constrained by the communicative world of the speech event rather than by the so-called “real-world” characteristics of the event or situation referred to’ (Traugott & Dasher 2002: 30).

. Fixed phrases and meaning Fixed or semi-fixed phrases can have meanings which are closely linked to form. However the interpretations may be ‘awkward’ because they have little to do with the literal meaning of the component lexical expressions. For example the fact is that can have interpersonal or rhetorical meanings such as rejection, elaboration or justification and strengthening. Nevertheless we seem to have a clear idea of what these expressions mean and how they are used. For example, (but) the fact is is used with a challenging and rejecting function and and the fact is that introduces elaboration. Pragmatic meanings are generally the result of conversational implicatures. However there are different types of implicature. Generalized conversational implicatures (GCIs) (Grice 1975; Levinson 2000). are closer to coded expressions than (particularized) conversational implicatures which are not encoded at all but emerge spontaneously in the context. Generalized conversational implicatures are associated with idioms, fixed expressions, routines, etc. (Levinson 2000). We find for example conversational routines with a high degree of fixedness and conventional meaning (thank you, sorry) but also collocational patterns such as (but) the fact is that, (and) the fact is that, the fact is, fact is, the thing is that, the thing is, etc. are linked to particular meanings.

TSL[v.20020404] Prn:1/06/2007; 15:29

F: PB16103.tex / p.11 (41)

The interface between discourse and grammar

The conventionalisation of a certain meaning or implicature can be explained by general expectations or heuristics adhered to in discourse. For example, there is an association between the fact is and the so-called M(anner)-principle (Levinson 2000: 38). According to Levinson’s M(anner) principle or the better known Grice’s maxim of manner (Be perspicuous, Avoid prolixity, avoid obscurity) (Levinson 2000: 38), what is said simply and briefly picks up the stereotypical interpretation. If, on the other hand, a marked expression is used (such as the fact is) it is suggested that the situation is also marked and non-stereotypical. An interpretation providing a reason for the non-stereotypical the fact is for example that the speaker rejects a preceding proposal. In what follows I will look in more detail at these senses and how they are associated with particular forms. There is a close relationship between the meaning of ‘the fact is’ as emphasis and rhetorical or interactive functions which we can explain in Peter White’s Bakhtin-inspired theory of engagement (White 2000, 2003). White offers a framework where elements of modal and evaluative stance are used interactively by speakers/writers ‘to take a stance towards the various points-of-view and value positions being referenced by the text and thereby align themselves vis-à-vis those who hold, or are represented as holding, these positions’ (White 2003: 260). The derived meanings range from those where the literal meaning is present to meanings which are more opaque. The meaning of emphasis (a high degree of certainty) is found after but (65 examples in the BNC). The unreduced form is used in the epistemic meaning (cf. *but fact is): (19) He may regard me as an amateur but the fact is that, due to physical difficulties, I cannot pull and twist in a horizontal mode and so I invoked gravity to assist. (AOX 291)

And in (20) the speaker rejects a hypothetical view-point as incorrect or false: (20) say we didn’t get on but the fact is, we hardly ever met.

In (21), the contrastive interpretation is signalled by yet rather than by but: (21) Yet the fact is that such attitudes were not shared by the conformist cultural establishment (A 7M 1086)

In many examples the fact is that is adversative (without but). It is disclaiming rather than proclaiming since it involves the ‘direct rejection or countering of a dialogically contrary position’ (White 2003: 271). It signals the speaker’s refutation of a dialogically contrary position to something which is expressed in the earlier context (or an implicature derived from what has been said). In the following example the fact is that is used with the function of ‘disclaiming’.



TSL[v.20020404] Prn:1/06/2007; 15:29



F: PB16103.tex / p.12 (42)

Karin Aijmer

(22) It didn’t arrive, even after had given me assurance every household in would get a leaflet. They got them in . I don’t know why, it’s outside the area. We didn’t get em. So you can consult people in about what’s happening in and wider afield. The fact is, there’s been a complete lack of consultation. It’s only this Authority that the consultation process in this district and you’ve had the answer, a resounding no to the Trusts. (JJD 406)

In (22) the fact is represents the speaker or ‘textual voice’ as countering an expectation which has arisen in context as the result of the announcement that you can consult people in the district since everyone had been given a leaflet. The expectation is not fulfilled since there’s been a complete lack of consultation. In (23) and (24) the fact is adds justification or elaborates on a preceding statement. The fact is that has interactive or argumentative function since it presents the speaker as the persuader: (23) “We have reviewed the results and the fact is we lost as many as we have won over the last two seasons,” said Bates.(K5A1688) (24) She is really a grand person and the fact is that she really is very often right. (COF1162)

In (25) on the other hand, the fact is that can be compared with I would contend that (cf. White 2003: 270). The textual voice conveys a heightened emotional investment and thereby confronts a contrary position. The speaker’s ‘contending’ is directed towards the opponent’s view ‘there is a natural progression from being in love to something less exciting’. The implication is that the speaker’s argument is more appropriate or relevant: (25) Holmes from Oxford Homes Relate, the old Marriage Guidance Council, says a year is a very short time, she says couples may be confused by the natural progression from the state of being in love to something less emotionally exciting. The fact is you can’t stay on top of a cloud indefinitely, however long you put it off, there’s going to be a corresponding second phase of disappointment and disenchantment when each partner becomes aware of the bits that they’re not so keen on. (KRT 4645)

Just like in fact which is discussed by Schwenter and Traugott (2000: 21) the function of the fact is is ‘to rank q as a stronger rhetorical argument than p, which is nevertheless parallel in orientation to q.’ Similarly in (26), the fact is that is not ad-

TSL[v.20020404] Prn:1/06/2007; 15:29

F: PB16103.tex / p.13 (43)

The interface between discourse and grammar

versative but rhetorical: An argument is proposed which is stronger, more relevant and appropriate: You can’t stay on top of a cloud indefinitely. (26) The debate on fox hunting has become unnecessarily complex <pause> and that can only suit the pro hunters because the basic objections to hunting are obscured. The more complicated you make the issues, the more the basic objections become obscured. The fact is that any so called sport which involves the exhaustion, distress and eventual death of an animal has got to be obscene in the eyes of any human being. (JNB 549)

Further semantic bleaching is involved in (27). The fact is expresses a number of pragmatic functions e.g. as a hesitator and as an organiser of the text (introducing a subtopic): (27) it’s not all negative, of course, I mean the fact is we’ve got low underlying inflation at the moment, and that means that i a seven percent return and if the inflation rate’s less than three, you’re actually, it’s, it’s really reasonably profitable. (GH4 156)

. Thing is and the trouble is Thing is has the form and function of a pragmatic marker. Like other markers it situates the utterance in relation to other utterances in the discourse. In (28), it signals a new (sub)topic: (28) X Chris X Chris X

Chris

I mean, don’t worry So are you, are you be going to seminar? Yes. Yeah. Yeah, I have to <end of voice quality>. Thing is I, I just I can’t work <end of voice quality>. No. (HUN 852)

The thing is that has conventionalised the rhetorical meaning of explanation or justification for a point of view (cause, ground for an opinion expressed in the preceding context): (29) MH:

Occasionally I see something that you’ve missed and think that it’s a lot better than you do. The thing is that you can’t always judge your own work, this is the value of working together. (CCO 339)



TSL[v.20020404] Prn:1/06/2007; 15:29



F: PB16103.tex / p.14 (44)

Karin Aijmer

When ‘the thing is’ has epistemic meaning it can combine with but to express opposition or rejection: (30) But the thing is that it seems that we’re a cultural the way the system is set up (HUJ 364)

The phrase can be used rhetorically to add a new argument: (31) And the thing is that when of the, I haven’t seen anywhere any kind of poll system, you know, I think it’s a, I’ve always thought it outrageous as a woman actually, that women have had to pay for toilets, and erm, I mean if we, you know, if we, we might, I mean, I, I would be in favour of a system where, you know, you, say a local Council issues a pass, which you pay for, have differential rates, but it’s like people paying up front for that toilet service. (D95684)

Compare also the trouble is. Besides the emphatic or contrastive use (but the trouble is) the phrase conveys rhetorical framing: a new, stronger argument is proposed: (32) John:

yeah well we don’t want it brushing on your The trouble is I’m quite sure that the plasterers never did that. (JT1420)

To sum up, to every pre-fabricated unit which is licensed by the collocational framework there is a description of what meaning it performs in the interaction. New meanings emerge in the interaction starting out as ad hoc implicatures. Some implicatures can become conventionalised because they are frequent and useful for particular purposes. As a result we can explain that ‘the fact is that X’ although meaning what it actually says can also have the dialogic meaning of rejection (disclaiming) or can be used to strengthen a position (proclaiming) or elaborate on it for argumentative purposes. The more routinised phrases (most typically ‘thing is’) are the most expendable and have primarily textual function.

. Conclusion I have argued that the fact is that represents an emergent ‘syntactic gestalt’ which is shaped by discourse in a way reflecting the speakers’ past experiences of the form and assessment of the present context (cf. Hopper 1987: 156). Such an emergent structure may ‘look like grammar’ but it can also diverge from the general grammatical pattern. In my analysis ‘the fact is that’ and its variations can be more or less lexicalized as predicted by the collocational frame to which they belong. The frame describes the synchronic relationship between the fact is that and pragmatic markers such as fact is which do not have the form of a clause. The pre-front

TSL[v.20020404] Prn:1/06/2007; 15:29

F: PB16103.tex / p.15 (45)

The interface between discourse and grammar

field is especially interesting as an emergent syntactic structure which is specialized for rhetorical and pragmatic functions. The fact is that is syntactically a matrix clause but it shows symptoms of grammaticalization. For example the clause no longer contains the main content. The weakly grammatical ‘the fact is that’ can be contrasted with the strongly grammatical ‘fact is’ which has a reduced form. The relation between syntax and pragmatics is however complex and involves the lexical meaning of the shell noun as well. As we have seen only some structures with shell nouns can undergo syntactic and semantic change. In particular, structures such as ‘the fact is that’ but not ‘the reason is that’ can develop new pragmatic functions in the interaction. Just as forms can vary there are several levels of meaning. I have suggested that we need to posit a level of meaning where we account for patterns or routines in terms of conventionalised implicatures. The functions of the fact is and related items are argumentative or rhetorical rather than modal. This implies that their meanings go beyond truth-claiming and provide the means for speakers to take up contentive positions which involve refutation, elaboration, justification and rhetorical strengthening.

References Auer, P. 1996. “The pre-front field in spoken German and its relevance as a grammaticalization position.”Pragmatics 6(3): 295–322. Biber, D., Johansson, S., Leech, G., Conrad, S. & E. Finegan. 1999 Longman Grammar of Spoken and Written English. London: Longman. Boye, K. & P. Harder. Unpublished. “Complement-taking predicates: usage and linguistic structure.” Bybee, J. and J. Scheibman. 1999. “The effect of usage on degrees of constituency: the reduction of don’t in English.” Linguistics 37(4): 575–596. Grice, H. P. 1975. “Logic and conversation.” In Syntax and Semantics 3: Speech Acts, P. Cole, and J. Morgan (eds.), 41–58. New York: Academic Press. Haiman, J. 1991. “Motivation, repetition and emancipation: the bureaucratization of language.” In Linguistic Studies Presented to John Finley. Memoire 8, H. C. Wofart (ed.), 45–70. Algonquian and Iroquoian Linguistics (Winnipeg, Manitoba). Halliday, M. A. K. 1994. An Introduction to Functional Grammar. 2nd ed. London: Edward Arnold. Hopper, P. 1987. “Emergent grammar.” In J. Aske, N. Beery, L. Michaelis and H. Filip (eds.), Berkeley Linguistic Society 13: 139–157. Lehmann, C. 1995. Thoughts on Grammaticalization. München-Newcastle: Lincom Europa. Levinson, S. 2000. Presumptive Meanings. Cambridge, Mass & London: The MIT Press. Quirk, R., S. Greenbaum. G. Leech and J. Svartvik. 1985. A Comprehensive Grammar of the English Language. London: Longman.



TSL[v.20020404] Prn:1/06/2007; 15:29



F: PB16103.tex / p.16 (46)

Karin Aijmer

Renouf, A. & J. McH. Sinclair. 1991. “Collocational frameworks in English.” In English Corpus Linguistics. Studies in Honour of Jan Svartvik, K. Aijmer and B. Altenberg (eds.), 128–44. London: Longman. Schmid, H.-J. 2000. English Abstract Nouns as Conceptual Shells. From Corpus to Cognition. Berlin and New York: Mouton de Gruyter. Schwenter, S. A. & E. C. Traugott. 2000. “Invoking scalarity: The development of in fact.” Journal of Historical Pragmatics 1(1): 7–25. Thompson, S. A. 2002. “‘Object complements” and conversation towards a realistic account.’ Studies in Language 26(1): 125–64. Thompson, S. A. & A. Mulac. 1991. “A quantitative perspective on the grammaticization of epistemic parentheticals in English.” In Approaches to Grammaticalization. (2 Vols), E. C. Traugott and B. Heine (eds.), 313–31. Amsterdam/Philadelphia: John Benjamin Publishers. Traugott, E. C. & R. B. Dasher. 2002. Regularity in Semantic Change. Cambridge: Cambridge University Press. Tuggy, D. 1996. The thing is that people talk that way. The question is Why? In Cognitive Linguistics in the Redwoods. The Expansion of a New Paradigm in Linguistics, E. H. Casad (ed.), 713–752. Berlin /New York: Mouton de Gruyter. White, P. 2000. “Dialogue and inter-subjectivity: reinterpreting the semantics of modality and hedging.” In Working with Dialogue, M. Coulthard, J. Cotterill & F. Rock (eds.), 67–80. Tübingen: Niemeyer. White, P. 2003. “Beyond modality and hedging: A dialogic view of the language of intersubjective stance.” Text 23(2): 259–284.

TSL[v.20020404] Prn:29/03/2007; 10:10

 

From syntax to pragmatics

F: PB161P2.tex / p.1 (47)

TSL[v.20020404] Prn:9/05/2007; 16:48

F: PB16104.tex / p.1 (49)

And as an aspectual connective in the event structure of pseudo-coordinative constructions Mark de Vos Rhodes University

Semantic events – usually encoded by a syntactic structure akin to a verb phrase – are complex constructs. They include a variety of aspectual information such as agentivity, duration, boundedness etc. While some languages utilise specific aspectual markers to encode this information, English does not have the morphological resources to do so. It is argued that English can use connectives to encode subtle aspectual meanings within complex events. I draw parallels between English pseudo-coordinative connectives and reduplicative coordination, arguing that these constructions shed light on the nature of the aspectual system in English. The implication is that certain strategies to create cohesion can also be used at a semantic and syntactic level to combine and order parts of event structure.

.

Connecting events

Communication through language inherently involves breaking world knowledge into organized chunks of information. Connectives are a valuable means of organizing these chunks, identifying their interrelationships and encoding coherence (Braunwald 1985). In doing so, isolated sentences are transformed into a text. Connectives can also occur within sentences as part of grammatical cohesive strategies.1 This paper will argue that the English connective and is used to encode coherence within event structure and bring out nuances in its meaning. The English conjunctive connective and is highly under-specified with respect to its function. Connectives in general, and and in particular are used as markers of cohesion (Halliday & Hasan 1976; Martin 1992; Schiffrin 1987, 2001). In . This runs against a trend within the field that suggests that cohesive connectives primarily occur at a supra-syntactic, textual level (Halliday 1973; Schiffrin 1987) rather than grammatical cohesion at a syntactic level (Gutwinsky 1976; Martin 1992). This is what Martin (2001: 36) refers to as ‘a territorial dispute over how much work the grammar is expected to do in discourse analysis’.

TSL[v.20020404] Prn:9/05/2007; 16:48



F: PB16104.tex / p.2 (50)

Mark de Vos

the following examples, coordination assists in encoding a temporal relationship between events.2 (1) a. John fell down the stairs and John broke his leg b. John fell down the stairs and he broke his leg c. John fell down the stairs and – broke his leg

Example (1a) illustrates one of the most fundamental functions of the connective and: the additive function. The sentence simply consists of two independent propositions. There is no relationship of ordering or causality between the conjuncts. Interestingly, the greater the level of syntactic cohesion, the greater the sense of temporal ordering and causality. Example (1b) has a stronger sense of cohesion (assisted by the pronoun) and the sense of temporal ordering is quite pronounced. Finally, example (1c) has a subject gap in the second conjunct (another indicator of cohesion) and the sense of temporal ordering is highly salient. It is not possible to construe (1c) as involving separate, unrelated events. In other words, this example involves a complex event consisting of two sub-stages of the main event. Here and assists in encoding a complex event. The temporal dependency is supported by the fact that the two conjuncts are non-commutable. Examples (1b, c) illustrate a second function of and, namely segment ordering (Evers-Vermeul 2005; Sanders 1992). Importantly, additive connectives such as and are often compatible with, and supportive of, a causal function although they do not necessarily encode causality in and of themselves (Evers-Vermeul 2005: 14). . Connectives and pseudo-coordination The examples in (2) are instances of ‘ordinary’, Boolean coordination. The presence of a subject in both conjuncts demonstrates that each conjunct is at least a tensed clause. The events referred to in each conjunct are distinct from each other. For example, in (2a), there is a literal event of Caesar going across the Rubicon and another, independent event of Caesar conquering Gaul. The temporal ordering of the conjuncts is irrelevant and is not necessarily reflected in the ordering of the conjuncts. . For the sake of explanatory simplicity, unless otherwise indicated, examples are constructed and judged grammatical or ungrammatical using the well-known introspective method. This method is used because some sentences require methods of controlling interpretation (e.g. using WH-movement) which makes it more difficult to find representative examples from corpora. Where necessary, naturalistic examples (generally from the British National Corpus) have been provided. Readers wishing to confirm the empirical basis for the results of this paper may choose to read articles based on corpora which bear it out (Stefanowitsch 1999; Koops 2004; Wulff 2006 and references therein).

TSL[v.20020404] Prn:9/05/2007; 16:48

F: PB16104.tex / p.3 (51)

Connectives and event structure

(2) a. Caesar went across the Rubicon and he conquered Gaul b. Caesar sat in his chair and he read the parchment

Pseudo-coordination in English typically involves a lexical verb, the connective and and a verb such as go or sit (3a, b).3 I would also like to consider instances of reduplicative coordination (3c) under the heading of pseudo-coordination (De Vos 2004).4 (3) a. b. c. d.

e.

Caesar went and read the parchment! Caesar sat and read the parchment Caesar read and read in his tent all night David Crystal, er actually is probably the only person in the world to have sat and read through from cover to cover four other encyclopaedias (BNC, KRT 6612) She read and read and read while her aunt mended the socks of her husband and brothers or sewed innumerable buttons on their shirts (BNC, FRC 1509)

The coordination in these types of examples is quite distinct from those in (2). The coordinated verbs and the connective are superficially contiguous and there is no overt subject in the second conjunct. In addition, there are syntactic and semantic differences between (2) and (3) that will be explored in the following sections. These properties have led many researchers to analyze and in this context as a subordinative connective (among others Carden & Pesetsky 1977; Gleitman 1965; Johannessen 1998; Wiklund 2004, 1996). . Structure of this paper Section 2 explains why examples like (3) can be considered pseudo-coordinative in nature. It is argued that the subset of English pseudo-coordinative structures il. There are also some instances of come being used fairly frequently, with other marginal usages with verbs like lie and run (Pullum 1990). In addition some authors include try in the same category (Carden & Pesetsky 1977; Stefanowitsch 1999). While try can indeed be regarded as pseudo-coordinative, there are very good reasons why pseudo-coordinative constructions with try are quite distinct from those with go and sit (Pullum 1990). They will not be considered in this paper. In addition there is a subset of pseudo-coordinative constructions named Scenesetting coordination (De Vos 2005). For reasons of space, these will also not be considered in this paper, although they might conceivably be amenable to a similar treatment. . Reduplicative coordination is also called augmentative coordination (Haspelmath 2007) and includes non-verbal examples like up and up, more and more and higher and higher. To claim that there are valid instances of pseudo-coordinative reduplicative coordination is not to deny the existence of sentences with coordinated IPs: Caesar read and he read and he read!



TSL[v.20020404] Prn:9/05/2007; 16:48



F: PB16104.tex / p.4 (52)

Mark de Vos

lustrated by (3) can be analyzed as complex predicate heads. This opens up ways of exploring the aspectual meanings associated with pseudo-coordination in Section 3. This leads to a discussion of the Aktionsart properties of pseudo-coordination and Section 4 develops the idea that pseudo-coordination plays a connective role within the event structure of the complex predicate.

. Pseudo-coordination The following section briefly outlines arguments to show that examples like those in (3) are considered pseudo-coordinative in nature. . Extraction One of the defining characteristics of pseudo-coordination is that it freely allows extraction from the second conjunct. In other words, filler-gap dependencies can be formed by associating a fronted WH-word with a corresponding gap in nonsubject position. This observation goes back at least to Ross (1967) who noted that pseudo-coordinative structures differed in this regard from ‘ordinary’ coordinative structures. Coordinative structures are subject to the Coordinate Structure Constraint (CSC) and the Across-the-Board (ATB) exception to it (Ross 1967; Williams 1978). (4) a.

CSC: In a coordinate structure, no conjunct may be moved, nor may any element contained in a conjunct be moved out of that conjunct (Ross 1967: 89). b. ATB: In a coordinate structure, the same constituent may be extracted from within all the conjuncts simultaneously (Ross 1967; Williams 1978).5

In the following illustrations of the CSC, example (5a) is the base sentence. In subsequent examples the gap is marked by t. Examples (5b, c) contain coordinated clauses, from which an NP has been extracted from the first and second conjuncts respectively. The result is ungrammatical and is an example of a CSC violation. Example (5d) shows that when the same constituent is extracted from both conjuncts, the result is grammatical. This is an example of the ATB exception to the CSC. (5) a.

Elizabeth admired Sir Robert and Paris had the hots for Helen

. The extracted constituent must perform the same general semantic function in both conjuncts e.g. it must be a deep subject in both or an object in both etc.

TSL[v.20020404] Prn:9/05/2007; 16:48

F: PB16104.tex / p.5 (53)

Connectives and event structure

b. *Who did Elizabeth admire t and Paris have the hots for Helen? [CSC] c. *Who did Elizabeth admire Sir Robert and Paris have the hots for t? [CSC] d. Who did Elizabeth admire t and Paris have the hots for t? [ATB]

Pseudo-coordinative constructions are different in this respect. It has often been noted that pseudo-coordinative constructions in English and other languages license extraction in non-ATB fashion (among others Carden & Pesetsky 1977; De Vos 2004; Gleitman 1965; Goldsmith 1985; Lakoff 1986; Ross 1967; Schmerling 1975; Wiklund 1996). (6) a. What was the parchment that Caesar went and read t? b. What was the parchment that Caesar sat and read t? c. What was the parchment that Caesar read and read t all night?

It is important to note that in subsequent examples, extraction of this type will be used to filter out the unwanted ‘ordinary’ coordinative readings and ensure pseudo-coordinative readings.6 . Coordinator substitution Another indicator of pseudo-coordination is the inability to substitute the coordinator and with another such as or (Schmerling 1975). Ordinary coordination allows one coordinator to be substituted for another and retain grammaticality (at the expense of a semantic change). (7) a. John both ate some cake and drank some tea b. John either ate some cake or drank some tea

Coordinator substitution is not possible with pseudo-coordination as the following contrasts indicate. (8) a. What was the parchment that Caesar went and read t? b. *What was the parchment that Caesar went or read t? (9) a. What was the parchment that Caesar sat and read t? b. *What was the parchment that Caesar sat or read t? (10) a. What was the parchment that Caesar read and read t? b. *What was the parchment that Caesar read or read t?

. This is because coordinative structures can be ambiguous between ‘ordinary’ coordination and pseudo-coordination. In fact, it has been shown by De Vos (2005) that, in English, pseudocoordination itself is not a unitary phenomenon, but can be subdivided into structures that allow extraction of any element and structures that only allow extraction of arguments. It is the former that are the primary focus of this paper.



TSL[v.20020404] Prn:9/05/2007; 16:48



F: PB16104.tex / p.6 (54)

Mark de Vos

. Distributivity Ordinary coordination can support both modification of the coordinated verbs. Since both is a distributive operator over two separate events, it is not at odds with ordinary coordination where two independent propositions are coordinated. (11) John both ate some cake and drank some tea

Pseudo-coordination constructions do not permit modification by both. Example (12a) is an ordinary coordination construction that superficially looks like pseudocoordination. However, examples (12a, b) show, that when extraction is used to force a pseudo-coordination reading, then modification by both is not possible. (12) a. John both went and read the book b. *What did John both go and read t?

Intuitively, the reason for this is that pseudo-coordination instantiates only a single event. Since both requires distribution over two events, it is not able to occur in pseudo-coordinative contexts. . A morphological argument Morphological facts also suggest that the pseudo-coordinative examples like (3) are different from ordinary coordination. In cases of ordinary coordination, verbs with different morphological forms can be coordinated. Example (13) shows that a clause with a participle can be coordinated with a clause containing a future modal. (13) Caesar has eaten some carpaccio and he will probably feel ill tomorrow

However, in pseudo-coordinative structures such as (3), the morphological specifications of both verbs must be identical in accordance with the following generalization (see also Pullum 1990). (14) Morphological ‘Sameness’ Condition (MSC): Both verbs of a pseudo-coordinative construction must have the same type of morphological specification i.e. both verbs must be either bare or morphologically marked with present, past, participle or similar.

This is illustrated with respect to the past tense in the following examples. The same point can be demonstrated with the present tense as well as with participle and future forms (De Vos 2005). (15) a. I wonder what Caesar went and ate t? b. *I wonder how Caesar went and eats t? c. *I wonder what Caesar goes and ate t?

TSL[v.20020404] Prn:9/05/2007; 16:48

F: PB16104.tex / p.7 (55)

Connectives and event structure

(16) a. I wonder what Caesar sat and ate t? b. *I wonder what Caesar sat and eats t? c. *I wonder what Caesar sits and ate t? (17) a. I wonder how long Caesar ate and ate for t? b. *I wonder how long Caesar ate and eats for t? c. *I wonder how long Caesar eats and ate t?

. Summary This section outlined some arguments showing that pseudo-coordination is distinct from ordinary coordination. Evidence included the ability to extract a constituent in non-ATB fashion from the second conjunct, inability to substitute the coordinator with another, ungrammaticality of distributive operators, and the fact that the morphological specifications of the verbs must be identical. These data also illustrated that pseudo-coordinative constructions with sit and go behave identically to reduplicative coordination. Henceforth, reduplicative coordination and the more traditional types of pseudo-coordination will be considered as instantiations of the same, broad type of construction.

. Towards a circumscription of the meaning of pseudo-coordination Having discussed some of the syntactic properties of pseudo-coordination, the following section explores the meaning of different types of pseudo-coordinative constructions. Three main points will be made. First, the pseudo-coordinative verb often appears to be semantically bleached. Second, it appears that each pseudo-coordinative construction has a specific ‘flavour’ of meaning which is conferred by the properties of go, sit etc. respectively. These ‘flavours’ are determined by the properties of these verbs independently of their occurrence in pseudocoordinative contexts. Third, the pseudo-coordinative verb focuses on internal sub-stages of the event. . Semantic bleaching of the pseudo-coordinative verb The meaning of pseudo-coordinative structures like those in (3a, b) is quite subtle. First of all, it seems that the first verb (i.e. go or sit) does not necessarily contribute a literal semantic interpretation. This is illustrated by example (18a) where a weather verb occurs in a pseudo-coordinative context. Clearly there is no literal interpretation of go. Similarly, in (18b), there is no literal interpretation intended for go.



TSL[v.20020404] Prn:9/05/2007; 16:48



F: PB16104.tex / p.8 (56)

Mark de Vos

(18) a. It went and rained b. Um Thomas Carlisle um wen was very much impressed by the work of people like Wordsworth and Coleridge because they were just a bit before him and he actually went and met Coleridge (BNC, KGW 569)

Similarly, in the following examples, it appears that the verb sit does not contribute a literal meaning of sitting insofar as the helicopter is hovering in the air. Similarly, it is not pragmatically plausible that people were physically sitting for ten years. In this context, sit implies a sense of location with corresponding ‘lack of dynamicity’ (Koops 2004) on the part of the subject. (19) a.

These helicopters are piloted with a computer control panel” which enables them to fly and sit and hover,” Fischel tells MassNews. I didn’t believe it until I saw it myself,” he says (http://www.massnews.com/2002_editions/12_Dec/ 122302_mn_uss_constell.shtml (14.07.2004)) b. He said: ‘For ten years we have sat and watched Diana being destroyed’ (BNC, ECM 80)

The fact that the posture verb is bleached is driven home with the following contrast from Koops (2004: 20). The negation of a pseudo-coordinative construction serves to negate the lexical verb and not the posture verb. (20) a. I’m not going to sit and read War and Peace if I can rent the movie! b. ?I’m not going to sit and read War and Peace if I can stand and read it! (Koops 2004: 20)

Importantly, semantic bleaching is not a necessary property of pseudo-coordinative structures. This is illustrated quite simply with a reduplicative example like (3c) reprinted here. (21) a. Caesar read and read in his tent all night b. She read and read and read while her aunt mended the socks of her husband and brothers or sewed innumerable buttons on their shirts (BNC, FRC 1509)

Since read is a lexical verb, it is not particularly grammaticalized. In this example, it contributes its full lexical meaning. Similarly, many pseudo-coordinative constructions with go and sit are compatible with literal interpretations when the context is consistent with such a reading. . Sit: focus on lack of dynamicity Pseudo-coordination with true posture verbs such as sit and stand tends to focus on the lack of dynamicity of the action involved (Koops 2004).

TSL[v.20020404] Prn:9/05/2007; 16:48

F: PB16104.tex / p.9 (57)

Connectives and event structure

(22) a. Why did you stand and watch while the thugs beat up the old lady? b. Why did you sit and watch while the thugs beat up the old lady? c. The ordinary German people could only sit and watch while he reduced their homes, possessions and friends to rubble and a tangled, indistinguishable mess (BNC, HPG 213).

In these examples, the focus is not necessarily on whether the observer actually stood or sat. Instead, it is a question about why the observer was passive during the confrontation. Such an interpretation is consistent with cross-linguistic correlations between posture verbs. Cross-linguistically, posture verbs instantiate a cline of activeness and potential power independently of their occurrence in pseudocoordinative constructions (Newman & Rice 2001). Thus, lie can be seen as a position of very low power and is associated with passivity, sickness, death etc. In contrast, stand is a posture of relatively more power, although it still encodes lack of dynamicity. (23) less active lie → sit → stand more active

This cline can also be seen in non-pseudo-coordinative contexts (Newman & Rice 2001). Thus, it is not pseudo-coordination construction per se that is responsible for this ‘passive’ interpretation, but rather the posture verb. In effect, posture verbs encode relative lack of dynamicity; let us refer to this as their having a [dynamicity] feature. Thus, given a particular event (indicated by the continuous time-line), the lexical verb indicates the type of action that characterizes the event. The pseudocoordinative verb sit, in conjunction with pseudo-coordination places focus on a manner component within the event, emphasizing the relative lack of dynamicity involved in the action. (24)

. Go: Focus on prospective nature of event The verb go brings another flavour to the constructions it occurs in. Examples like the following ones often encode a sense of counter-expectationality, surprise and agentivity. (25) a. The gladiator got killed b. The gladiator went and got killed c. If you was anyone else, I’d say, get divorced, but you did that once and then you went and got married again (BNC, G1D 1006).



TSL[v.20020404] Prn:9/05/2007; 16:48



F: PB16104.tex / p.10 (58)

Mark de Vos

The sentence in (25a) is a neutral statement about the murder of the gladiator. In contrast, (25b) implies that the gladiator was actively involved in the build-up to his death; that he perhaps did something (stupid) that resulted in his own death. While such a reading is of course consistent with a neutral sentence (25a), it is explicitly coded through pseudo-coordination in example (25b). Thus, pseudocoordination seems to bring into focus, existing aspects of the event structure of the verb. In this case, it is the build-up to – the incipient nature of – the main part of the event which is brought into focus. Thus, in a pseudo-coordinative construction, while the lexical verb indicates the type of action characterizing the event, the verb go, in combination with pseudo-coordination, places focus on the build-up to the action. (26)

In focussing on the incipient nature of the event, there is the implication that the subject brings the event into being; that the subject is an active agent in the initiation of that event. I will call this the pseudo-causative reading associated with go. The reason why go has this property is that this verb can be characterized as having a [prospective] feature in its lexical specification. Evidence for a prospective feature associated with go independently of pseudo-coordination is provided by the following example which is consistent with a reading where although everybody will eventually die, it is not the case that they have all already started dying. The sentence merely makes a claim about the prospect of eventually dying. (27) It is a fact that everybody is going to die.

. Reduplicative coordination: Focus on the event itself Reduplicative coordination can be described in very similar terms to the previous examples. Like pseudo-coordination with sit and go, reduplicative coordination focuses sub-stages of event structure. In particular, it is the event itself that is brought into focus, yielding interpretations consistent with intensity and/or a protracted nature. In the following example, there is the reading that the event – namely reading – is carried out to an excessive or intensive degree. (28) a. Caesar read and read in his tent all night b. She read and read and read while her aunt mended the socks of her husband and brothers or sewed innumerable buttons on their shirts (BNC, FRC 1509)

In this case, the second verb indicates the type of action involved in the event i.e. it is a reading event. The first verb, in combination with pseudo-coordination places

TSL[v.20020404] Prn:9/05/2007; 16:48

F: PB16104.tex / p.11 (59)

Connectives and event structure

focus on that part of the event which is concerned with the actual reading. Thus, pseudo-coordination serves to focus the nature of the event itself. (29)

. Summary The meanings associated with pseudo-coordination are subtle. In this section, these meanings have been described and explained. It has been suggested that, depending on which verb is involved, the interpretative ‘flavour’ of the construction is altered. – sit: focus on manner: lack of dynamicity of the activity – go: focus on the preparatory stage of the activity – Reduplicative: focus on the activity itself (intensification)

It has also been shown that these ‘flavours’ are not a unique property of pseudocoordination itself but follow from the general properties of the verbs involved. Instead, what is specific to pseudo-coordination is that the focus is placed on a sub-stage of the main event itself. In order to do so, pseudo-coordinative constructions necessarily require complex events with internal structure; by definition this is a durative event and pseudo-coordinative constructions are thus inherently aspectual.

. Connecting Aktionsarten The previous section explored the meanings encoded by pseudo-coordinative verbs. In the following section, I will focus on the nature of the connective itself. I will argue that the connective, and, can be analyzed as a true coordinator within the argument structure of the event. It will also be suggested that its role is remarkably similar to its function at discourse level, namely that it has additive and ordering functions that make the complex event cohesive and place focus on sub-stages of that event. . Aspect, Aktionsart and event structure Until this point, I have referred to the fact that pseudo-coordinative constructions are aspectual and focus various aspectual sub-stages of the complex predicate. Although aspect and Aktionsart are often referred to collectively as aspect, it is important to distinguish them. By the term ‘Aktionsart’, I refer to situation as-



TSL[v.20020404] Prn:9/05/2007; 16:48



F: PB16104.tex / p.12 (60)

Mark de Vos

Table 1. Vendler Classes Asp. Class

Vendler Class

Notation

States Achievements Accomplishments Activities

–process, –definite –process, +definite +process, +definite +process, –definite

[–] [τ] [φ, τ] [φ]

pect (Smith 1997), an inherent property of verbs whereby they are specified as being bounded or unbounded. This reduces to the Vendlerian distinction between states, activities, achievements and accomplishments. (30) Caesar resembled Marcus Maximus [State] (31) Ben Hur won a race [Achievement] (32) Caesar sent the captives back home [Accomplishment] (33) Hannibal’s legions trudged through the snow for hours [Activity]

Every event may have a starting point, initium, a process, cursus or an ending point, finis (Dowty 1979; Tenny 1987; Vendler 1957; Verkuyl 1972, 1993). This is what Johanson (1996) calls the ‘Internal Phase Structure’ which reduces to the distinction between telic and atelic predicates. The cursus can be represented as a phase φ, a non-punctual stretch of time corresponding to Vendler’s [+process], and the finis can be represented as a telos τ, a point of punctual change corresponding to Vendler’s [+definite]. The resultant classification is basically that of Vendler (1957). φ can be subdivided into subparts whereas τ, being punctual, cannot be subdivided any further. States, having no apparent internal structure, cannot be subdivided. It is important to note that the Aktionsart of the verb is generally lexically specified and is not a property of clauses. It is simply a lexical fact that some verbs such as ‘wander’ are activities whereas other verbs such as ‘shoot’ are punctual. This is not to deny that there are interactions between the lexical Aktionsart of a verb and other entities within the clause, such as the direct object. Thus, an unbounded activity verb can be provided with an endpoint by an appropriate DP. Aktionsart should be contrasted with viewpoint aspect, which, although it has commonalities with Aktionsart, is an external view of an event as to whether it is starting, progressing, completed etc. regardless of its Vendlerian class (Comrie 1976). Henceforth, when the term ‘aspect’ is used, it refers to viewpoint aspect. . Coordination of heads and event structure Given the preceding discussion that pseudo-coordination involves processes within event structure, I follow the intuitions of Cormack and Breheney (1994),

TSL[v.20020404] Prn:9/05/2007; 16:48

F: PB16104.tex / p.13 (61)

Connectives and event structure

Koops (2004), Pollock (1994), Postma (1995), Stefanowitsch (1999) in claiming that pseudo-coordination involves a complex head. This has been formalized by De Vos (2004, 2005) as involving ‘true’ coordination at the level of the head to form a complex predicate.7 The implication of this is that the so-called pseudocoordinative connective and is, in fact, a true, coordinative connective within the event structure. Consequently, it is expected that it will display properties of coordination including additive and ordering functions in the service of the coherence of the event itself. (34)

Since coordination is central to the following discussion it is necessary to outline some fundamental assumptions about the nature of coordination. Two sets of assumptions will be taken for granted: (i) the Coordinate Structure Constraint and (ii) the Law of Coordination of Likes. The previous discussion already touched on the subject of the Coordinate Structure Constraint (4a). In addition to this, coordination in natural language almost always coordinates (at least two) similar entities.8 The level of similarity is not necessarily restricted to syntactic category but is also related to semantic function (Dik 1968; Haspelmath 2007; Munn 1993; Peterson 2004; Sag et al. 1985; Schachter 1977). In the literature, this property is known as the Law of Coordination of Likes. Notwithstanding a number of wellknown exceptions to this principle (Bayer 1996; Dik 1968; Lakoff 1986; Postal 1998; Progovac 1998a, b; Sag et al. 1985; Zoerner 1995), I assume it to be operative and to be ultimately derived from a deep property of the additive function of coordination. These conservative assumptions about coordination lead to two . See De Vos (2005) for a detailed discussion. . Dik (1968) traces this concept back to antiquity.



TSL[v.20020404] Prn:9/05/2007; 16:48



F: PB16104.tex / p.14 (62)

Mark de Vos

converging predictions. Given the preceding discussion, coordination should potentially be able to coordinate Aktionsart features (i.e. φ and τ) present within the event structure of the predicate. By virtue of the Law of Coordination of Likes, such features must be equivalent. (35) a. Prediction 1: Pseudo-coordination interacts with Aktionsarten. b. Prediction 2: Pseudo-coordination involves coordination of ‘like’ features.

. Pseudo-coordination as a system of Aktionsarten This section explores the types of verbs that pseudo-coordinative constructions can co-occur with. The following examples show that verbs like go are the least restricted in their distribution, occurring with Activities, Accomplishments and Achievements (examples (36) to (39)). Sit is more restricted, occurring only with Activities and Accomplishments (examples (40) to (43)). Reduplicative coordination has the most restricted distribution, occurring only with Activities (examples (44) to (47)). Go: (36) *John went and resembled his father? [State] (37) a. Which board-game did John go and win? [Achievement] b. But if the president says hey, you know, go and win the war in Vietnam for me, you know, that’s a bit more tricky, a bit more tricky (BNC, JSM 176) (38) a. Who did John go and drive back home in two hours? [Accomplishment] b. But, still the same, Fabia felt hot by the time she’d made it to her room, so went and took a shower and changed her clothes (BNC, JYF 1417) (39) a. Which board-game did John go and play for hours? [Activity] b. Go and play on the slide and let the girls get on with their game (BNC, H0F 4089)

Sit: (40) *Who did John sit and resemble? [State] (41) *?Which board-game did John sit and win? [Achievement] (42) a. What did John sit and eat 43 of in only 30 minutes? [Accomplishment] b. Yes, it was nice to sit and eat my lunch in peace, with no-one to interrupt me (BNC, ANM 746) (43) a. Which board-game did John sit and play for hours? [Activity] b. And rather than sit and eat chocolate I can eat apples and fruit (BNC, KPD 217)

TSL[v.20020404] Prn:9/05/2007; 16:48

F: PB16104.tex / p.15 (63)

Connectives and event structure

ReCo: (44) *John resembled and resembled his father [State] (45) *John won and won the race [Achievement] (46) *John ate and ate 46 hamburgers in only 2 hours [Accomplishment] (47) a. John walked and walked for hours [Activity] b. He walked and walked, in fair and foul weather, pretending that he was training the young hounds to stay at heel; and did so (BNC, APW 3258)

These examples show that the pseudo-coordinative constructions utilizing go, sit and reduplicative coordination form a gradually more restrictive system of meanings.9 These are illustrated in (48). Prediction one (35a) has been borne out: it has been demonstrated that pseudo-coordination does interact with Aktionsarten. The other half of the prediction – that only ‘like’ Aktionsart features can be coordinated – will be discussed in the following section. (48) The aspectual nature of pseudo-coordination: Go Sit Verb & verb

State * * *

Achievement

Accomplishment

Activity

J

J J

J J J

* *

*

. Aktionsarten and coordination of likes It has been proposed that pseudo-coordination involves coordination within event structure. Events can be decomposed into punctual occurrences τ and nonpunctual stretches φ. τ are discrete and cannot be subdivided further, whereas φ can, by definition, be subdivided into smaller instances of φ. These are the fundamental units of event structure, and it can be demonstrated that it is precisely these units that are coordinated. .. Sit The verb sit implies a static location, which is an activity of sitting, or being at a certain location. The verb sit thus has the Aktionsart specification of φ. Similarly, play is an activity: φ When both these non-punctual sub-stages of the playing event are coordinated, the LCL is satisfied. This is a corroboration of the second prediction in (35b). . See Wulff (2006) for a corpus study of constructions with go which corroborates the findings for this verb.



TSL[v.20020404] Prn:9/05/2007; 16:48



F: PB16104.tex / p.16 (64)

Mark de Vos

(49) a.

Which board-game did John sit and play for hours? [φ] and [φ]

[Activity]

b.

The verb eat combines with a direct object to create an accomplishment: [φ,τ]. Since there is still some part of the event that is in common with the specification of sit, coordination of φ can still satisfy the LCL. However, not all the Aktionsart features are within the scope of coordination as illustrated in (50b). (50) a.

What did John sit and eat 43 of in only 30 minutes? [Accomplishment] [φ] and [φ,τ]

b.

It is, however, not possible for coordination of sub-stages to occur when the substages are of a fundamentally different type. This would be a violation of the LCL. The fact that examples like (51a) are ungrammatical corroborates the idea that the LCL is indeed operative at sub-stage level. This is confirmation that pseudocoordination really does involve true coordination. (51) a. *?Which board-game will John sit and win? [φ] and [τ] b.

[Achievement]

States cannot occur in pseudo-coordinative contexts because states have no internal structure. In other words, there are neither τ nor φ to coordinate. Since there are no sub-stages, there can be no coordination of sub-stages. (52) a. *John will sit and resemble his father [φ] and [–] b.

[State]

.. Go The verb go has the least restricted distribution of any of the pseudo-coordinative predicates. This is also because go is the most grammaticalized of the verbs under discussion. The following examples demonstrate that the specification of go, independently of pseudo-coordinative contexts, can be either φ or τ. (53) a. Alexander went to India for 10 days [φ] b. Alexander went across the sea to India in 10 days [φ,τ] c. Alexander went ballistic [τ]

TSL[v.20020404] Prn:9/05/2007; 16:48

F: PB16104.tex / p.17 (65)

Connectives and event structure

Example (53a) shows go as an activity: [φ]. However, go can also be construed as an accomplishment – a bounded activity (53b): [φ,τ]. Finally, it is also possible for go to imply a punctual change as in (53c): [τ]. Given the underspecified Aktionsart of go, it is expected that this affects the kinds of predicates it may be combined with. In the following example, the φ features of go allow it to combine with activities. (54) a.

Which board-game did John go and play for hours? [φ] and [φ]

[Activity]

b.

The verb win is an achievement specified as τ. The specification of go as τ enables it to combine with achievements. (55) a.

Which board-game did John go and win? [τ] and [τ]

[Achievement]

b.

The verb go can also be combined with accomplishments in the same way that sit can. The Law of Coordination of Likes ensures that two features of the ‘same’ kind are coordinated, the remaining features remaining beyond the scope of coordination as shown in (56b). (56) a.

Who did John go and drive back home in two hours? [Accomplishment] [φ] and [φ,τ]

b.

Finally, go cannot be combined with states for the same reason that sit cannot. States simply do not contain any internal structure which can be modified. (57) a. *John went and resembled his father [φ] and [–] b.

[State]

.. ReCo In reduplicative coordination, the same verb appears in both conjuncts. As discussed in Section 3.4, the role of the second verb is to determine the nature of the action carried out. The first verb determines what part of the event will be brought into focus. Since both verbs are identical, the effect is to place focus on the event itself.



TSL[v.20020404] Prn:9/05/2007; 16:48



F: PB16104.tex / p.18 (66)

Mark de Vos

A verb like read is an activity. Its event structure is thus composed of φ which, by definition, can be subdivided into further instances of φ. Since both verbs are identical, when sub-stages of the event are coordinated, φ is in both conjuncts. This is schematically illustrated. The LCL is respected. (58) a.

John read and read all day long [φ] and [φ]

[Activity]

b.

It is also possible for reduplicative coordination to occur with punctual or bounded predicates as long as they are construed as being durative. A punctual predicate consists of τ which cannot be subdivided. There are three possible ways of construing a punctual predicate as being durative: by means of (i) an iterative interpretation, (ii) a serial interpretation and (iii) by coercion of the event structure. These will each be dealt with in turn.

Iterative readings. One way of construing τ as a durative event is to interpret a sequence of punctual sub-stages of an event as being part of a larger durative event. This yields an iterative reading. (59) a.

John shot and shot at the rabbit [τ] and [τ]

b.

Example (59) has a reading where John repeatedly, and excessively, pumps bullets into the rabbit. In other words, the punctual predicate shoot is interpreted as being a serial activity. In terms of the proposed structure, one verb determines what kind of event it is: an event of shooting with the internal structure τ. The second verb provides an additional τ, thus allowing an iterative reading with ordered substages consisting of τ. It is in this context that the temporal ordering function of the connective becomes apparent.

Serial readings. A second way of construing a punctual event as being durative is by means of a serial reading. A serial reading is frequently licensed by a plural subject and differs from the iterative reading insofar as a series of events are distributed across a plurality of subjects. (60) a. b.

The police shot and shot at the protesters [τ] and [τ]

TSL[v.20020404] Prn:9/05/2007; 16:48

F: PB16104.tex / p.19 (67)

Connectives and event structure

In example (60), a reading is possible whereby many shots were fired at the protesters, but each police officer need only have shot once. Thus, a series of shooting sub-stages of a larger, complex event of riot control are distributed across a plurality of subjects. In this way, a punctual predicate is conceived of as being durative.

Coerced readings. The third way of construing a bounded predicate as being durative is to coerce its Aktionsart properties. Not all verbs are able to be coerced and this is at best a marginal property. However, it does provide an intriguing insight into the way pragmatics can constrain syntactic structure (for a similar view, see Bickel 1997; Ward & Birner 2001). The verb drown is inherently an accomplishment insofar as it involves a cumulative process of drowning (φ ) followed by a bounding point (τ) which necessarily implies the death, by drowning, of the subject. Thus (61) necessarily implies that Shelly died by drowning. (61) Shelly drowned

However, in the case that a subject is singular, there are a very limited number of cases where inherent endpoints can be deaccented in the context of reduplicative coordination. (62) a.

And he just drowned and drowned and I saw his head go under [φ]τ and [φ]τ (http://www.abc.net.au/austory/transcripts/s418 748.htm (14.07.04))

b.

But note that the entailments of this example are very different from those of (61). Whereas (61) entails that Shelly necessarily dies by drowning, (62) does not entail that death follows immediately after the first sub-stage of the drowning event or indeed at all. The only possible reading for (62) is that drowning is a durative event and that each drown is actually a sub-stage of the larger drowning event. It is not even necessary that the subject eventually dies in this example, in contrast to normal usage of this verb. For instance, it would be perfectly felicitous to continue the dialogue in the following way. (63) . . . But suddenly a lifeguard put an arm around him and lifted him to safety

Thus, it is possible, depending on context, that the endpoint inherent in drown is deaccented. In other words, it is φ which is being coordinated in (62) at the expense of τ. This is compelling evidence for the LCL within the event structure as well as for the role of pragmatics in influencing syntactic structure.



TSL[v.20020404] Prn:9/05/2007; 16:48



F: PB16104.tex / p.20 (68)

Mark de Vos

. Conclusion This paper has provided a comprehensive account of the role of connectives in English pseudo-coordinative constructions. (i) Pseudo-coordination is shown to involve the manipulation of the internal structure of the event. (ii) It is shown that the subtle but precise meanings of pseudo- coordinative constructions are provided by the nature of the first verb, namely sit or go or a reduplicated verb. In each case, the meanings of these verbs independently of their occurrence in pseudo-coordinative constructions determine their behaviour in pseudocoordinative contexts. (iii) The role of the connective is surprisingly quite similar to the connective functions of and in other contexts. In other words, and has an additive function and may serve to encode ordering of sub-stages of events in a way that enhances the cohesion of the complex event.

References Bayer, S. 1996. “The Coordination of Unlike Categories.” Language 72: 579–616. Bickel, B. 1997. “Aspectual Scope and the Difference Between Logical and Semantic Representation.” Lingua 102: 115–131. Braunwald, S. 1985. “The Development of Connectives.” Journal of pragmatics 9: 513–525. British National Corpus Consortium. 2000. British National Corpus Online Edition ( Oxford: Humanities Computing Unit of Oxford University, Published to Web by The British Library). Carden, G., & D. Pesetsky. 1977. “Double-Verb Constructions, Markedness and a Fake Coordination.” In Papers from the 13th Regional Meeting, Chicago Linguistic Society. Chicago: University of Chicago, 82–92. Comrie, B. 1976. Aspect: An Introduction to the Study of Verbal Aspect and Related Problems. Cambridge: Cambridge University Press. Cormack, A., & R. Breheney. 1994. “Projections for Functional Categories.” UCL working papers in linguistics 6: 35–62. De Vos, M. 2004. “Pseudo Coordination is Not Subordination.” In Linguistics in the Netherlands 21, L. Cornips, and J. Doetjes (eds.), 181–192, Amsterdam/Philadelphia: John Benjamins. De Vos, M. 2005. The Syntax of Verbal Pseudo-Coordination in English and Afrikaans. Ph.D. thesis, Leiden University Centre for Linguistics. Dik, S. 1968. Coordination: Its Implications for a Theory of General Linguistics. Amsterdam: North Holland Publishing Company. Dowty, D. 1979. Word Meaning and Montague Grammar: The Semantics of Verbs and Times in Generative Semantics and in Montague’s PTQ. Dordrecht: Reidel. Evers-Vermeul, J. 2005. The Development of Dutch Connectives: Change and Acquisition as Windows on Form-Function Relations. Ph.D. thesis, Universiteit Utrecht. Gleitman, L. 1965. “Coordinating Conjunctions in English.” Language 41: 260–293. Goldsmith, J. 1985. “A Principled Exception to the Coordinate Structure Constraint.” In CLS 21, Part 1, Papers from the General Session at the Twenty-First Regional Meeting, W. Eilfort, P. Kroeber, and K. Peterson (eds.), 133–143. Chicago: Chicago Linguistic Society.

TSL[v.20020404] Prn:9/05/2007; 16:48

F: PB16104.tex / p.21 (69)

Connectives and event structure

Gutwinsky, W. 1976. Cohesion in Literary Texts: A Study of some Grammatical and Lexical Features of English Discourse. The Hague: Mouton. Halliday, M. 1973. Explorations in the Functions of Language. London: Edward Arnold. Halliday, M., and R. Hasan. 1976. Cohesion in Language. London: Longman. Haspelmath, M. 2007. “Coordination.” To appear in: Shopen, T. (ed.) Language typology and syntactic description. 2nd ed. Cambridge. Cambridge University Press. Johannessen, J. 1998. Coordination. New York: Oxford University Press. Johanson, L. 1996. “Terminality Operators and their Hierarchical Status.” In Complex Structures: A Functionalist Perspective, B. Devriendt, L. Goossens, and J. Van der Auwera (eds.), 229– 258. Berlin:Mouton de Gruyter. Koops, C. 2004. “Emergent Aspect Constructions in Present-Day English.” In Studies in Linguistic Motivation, G. Radden, and K-U. Panther (eds.), 1–34, Berlin: Mouton de Gruyter. Lakoff, G. 1986. “Frame Semantic Control of the Coordinate Structure Constraint.” In CLS 22, Part 2, Papers from the Para session on Pragmatics and Grammatical Theory, A. Farley, P. Farley, and K-E. McCullough (eds.), 152–167. Chicago: Chicago Linguistic Society. Martin, J. 1992. English Text: System and Structure. Amsterdam: Benjamins. Martin, J. 2001 “Cohesion and Texture.” In The Handbook of Discourse Analysis, D. Schiffrin, D. Tannen, and H. Hamilton (eds.), 35–53. Malden MA: Blackwell. Munn, A. 1993. Topics in the Syntax and Semantics of Coordinate Structures. Ph.D. thesis, University of Maryland. Newman, J., & S. Rice. 2001. “English sit, stand and lie in Small and Large Corpora.” In International computer archive of modern and mediaeval English 25: 109–134. Peterson, P. 2004. “Coordination: Consequences of a Lexical-Functional Account.” Natural language and linguistic theory 22: 643–679. Pollock, J-Y. 1994. “Checking Theory and Bare Verbs.” In Paths Towards Universal Grammar: Studies in Honour of Richard S. Kayne, G. Cinque (ed.), 293–310. Washington D.C.: Georgetown University Press. Postal, P. 1998. Three Investigations of Extraction. Cambridge MA: MIT Press. Postma, G. 1995. “Zero Semantics: The Syntactic Encoding of Quantificational Meaning.” In Linguistics in the Netherlands 12, M. Den Dikken, and K. Hengeveld (eds.), 175–190. Amsterdam: John Benjamins. Progovac, L. 1998a “Structure for Coordination (Part I).” GLOT International 3 (7): 3–6. Progovac, L. 1998b “Structure for Coordination (Part II).” GLOT International 3(8): 3–9. Pullum, J. 1990. “Constraints on Intransitive Quasi-Serial Verb Constructions in Modern Colloquial English.” In When Verbs Collide: Papers from the 1990 Ohio State Mini-Conference on Serial Verbs, B. Joseph, and A. Zwicky (eds.) 218–239. Columbus, Ohio: Ohio State University, 1990, number 39 in Ohio State University Working Papers in Linguistics. Ross, J. 1967. Constraints on Variables in Syntax. Ph.D. thesis, Massachusetts Institute of Technology. Sag, I., G. Gazdar, T. Wasow, and S. Weisler. 1985. “Coordination and How to Distinguish Categories.” Natural language and linguistic theory 3:117–171. Sanders, T. 1992. Discourse Structure and Discourse Coherence: Aspects of a Cognitive Theory of Discourse Representation. Ph.D. thesis, Katholieke Universiteit Brabant. Schachter, P. 1977. “Constraints on Coordination.” Language 53: 86–103. Schiffrin, D. 1987. Discourse Markers. Cambridge: Cambridge University Press.



TSL[v.20020404] Prn:9/05/2007; 16:48



F: PB16104.tex / p.22 (70)

Mark de Vos

Schiffrin, D. 2001. “Discourse Markers: Language, Meaning and Context.” In The Handbook of Discourse Analysis, D. Schiffrin, D. Tannen, and H. Hamilton (eds.), 54–75. Malden MA: Blackwell. Schmerling, S. 1975. “Asymmetric Conjunction and Rules of Conversation.” In Speech Acts, P. Cole, and J. Morgan (eds.), 211–231. New York: Academic Press. Smith, C. 1997. The Parameter of Aspect. Dordrecht: Kluwer Academic Publishers. Stefanowitsch, A. 1999. “The Go-and-Verb Construction in a Cross-Linguistic Perspective: Image-Schema Blending and the Construal of Events.” In Proceedings of the Second Annual High Desert Linguistics Society Conference, D. Nordquist, and C. Berkenfield (eds.), 1–12. Alburquerque: High Desert Linguistics Society. Tenny, C. 1987. Grammaticalizing Aspect and Affectedness. Cambridge MA: MIT Press. Vendler, Z. 1957. “Verbs and Times.” The Philosophical Review 66: 143–160. Verkuyl, H. 1972. On the Compositional Nature of the Aspects. Dordrecht: Reidel. Verkuyl, H. 1993. A Theory of Aspectuality: The Interaction Between Temporal and Atemporal Structure. Cambridge, UK: Cambridge University Press. Ward, G., and B. Birner. 2001. “Discourse and Information Structure.” In The Handbook of Discourse Analysis, D. Schiffrin, D. Tannen, and H. Hamilton (eds.), 119–137. Malden MA: Blackwell. Wiklund, A-L. 2004. The syntax of tenselessness: on copying constructions in Swedish. Department of Philosophy and Linguistics, Umeaa, Sweden. Wiklund, A-L. 1996. “Pseudo-coordination is Subordination.” Working papers in Scandinavian syntax 58: 29–54. Williams, E. 1978. “Across the Board Rule Application.” Linguistic Inquiry 9: 31–43. Wulff, S. 2006. Go-V vs. Go-and-V in English: A Case of Constructional Synonymy?” In Corpora in Cognitive Linguistics, S. Gries, and A. Stefanowitsch (eds.), 101–126. Berlin, New York: Mouton de Gruyter. Zoerner, E. 1995. Coordination: The Syntax of &P. Ph.D. thesis, University of California, Irvine.

TSL[v.20020404] Prn:30/05/2007; 10:12

F: PB16105.tex / p.1 (71)

‘Are you a good which or a bad which?’ The relative pronoun as a plain connective Rudy Loock Université de Lille III – U.F.R. Angellier U.M.R. 8528 SILEX du CNRS

This paper offers an analysis of appositive relative clauses in spontaneous English that deviate from traditional syntactic schemas, particularly in relation to the presence of a gap. The data to be considered here consists of ‘gap-filled’ (presence of a resumptive pronoun) and ‘gapless’ examples, traditionally considered ungrammatical yet found in English nonetheless. The analysis proposed here will start with traditional ARCs, analysing the complex role of the relative pronoun as fulfilling two functions at the same time, anaphoric and connective. We will then show that these functions can be dissociated, which results in the occurrence of ARCs that seem to deviate from the norm, with a specific, purely connective role attributable to the relative pronoun.

.

Introduction

The starting point of this study was the compilation of a corpus of ARCs (Appositive Relative Clauses) in spontaneous speech (i.e. non-restrictive, in opposition to determinative/restrictive RCs, henceforth DRCs, following a long-established distinction): (1) The people of Oz, who were scared of the Witch of the East, were relieved when Dorothy’s porch crushed her to death. (ARC) (2) The people of Oz who were scared of the Witch of the East were relieved when Dorothy’s porch crushed her to death. (DRC)

While compiling spontaneous utterances with ARCs, we came across relative clauses whose structure seemed to deviate from the “norm” established by grammar books. After considering them as performance errors, we realized that their * I particularly would also like to thank Mai Kuha, from Ball State University, for providing me with her self-collected corpus. I also would like to thank Ruth Huart, Liliane Haegeman, Philip Miller, Chad Langford and Kathleen O’Connor for their useful comments and suggestions.

TSL[v.20020404] Prn:30/05/2007; 10:12



F: PB16105.tex / p.2 (72)

Rudy Loock

frequency of occurrence was too important for these deviant structures to be simply a random phenomenon. And by analyzing them, we realized that the relative pronoun seemed to be fulfilling a role quite different from its traditional role. Interestingly too, these examples as we are to analyze them here in this paper also shed new light on ARCs as a whole, as ARCs have for long been a problem for syntacticians, seeming to be on a border between coordination and subordination, and having a syntactically and semantically ambiguous behaviour. Our corpus consists of examples we personally collected from spontaneous English (RAD, TEL, CONV) and also from Mai Kuha from Ball State University in the United States, who worked on speakers’ attitude towards such examples from a sociolinguistic point of view. She kindly transmitted her corpus to me, whose examples are used (MK) alongside mine. We also collected examples from the British National Corpus (spoken section, from which we extracted a subcorpus made of spontaneous speech only, BNC).

. The data . Typical and atypical ARCs In a traditional ARC a gap is left by WH- movement, which generative grammar would call a trace (t) or copy, following Chomsky (1995): (3) This law, which some lawyers would regard ti as tendentious, prohibits employers from HIV testing. (4) Dennis, whoi we were talking about ti last night, says you’re absolutely mental.

Atypical ARCs, on the other hand, consist of two different kinds of examples, those in which the gap is filled (gap-filled examples) and those in which no gap can be found (gapless examples)

(i) With a resumptive/shadow pronoun (5) (BNC_ARC50) My nickname is “Pan” whichi I don’t like iti so much. (6) (BNC_ARC26) Well I’d like, Dave Girt, Leeds City, I’d like s some recognition of West Yorkshire’s problems to be evident in er the deliberations, whichi er at the moment iti ’s iti ’s absent, it may it may have been taken into account but it’s absent in the exposition. (7) (MK_ARC4) My foot is narrow in the arch area, whichi I would’ve expected iti to widen.

TSL[v.20020404] Prn:30/05/2007; 10:12

F: PB16105.tex / p.3 (73)

‘Are you a good which or a bad which?’

(8) (RAD15_ARC2) If you do wash your hair then you’re washing away some of the goodness out of you. And whichi you don’t want to do thati because you’re carrying a child inside you. (9) They were just towed across the Midway onto the bridle path, wherei they were just sitting therei peacefully. (10) (CONV_ARC3) That’s a suggestion of yours which I followed, whichi I didn’t even want to do thati . (11) (MK_ARC15) . . . and I got a new change of clothes, which I wanted you to know that.

Examples with resumptive pronouns are well-known, as is shown by the existence of an extensive generative literature on the subject. They are often considered as a last-resort solution because of syntactic constraints. That is to say, when the distance between the trace/copy and the extraction site is too important (island constraint) or because the trace is not governed (empty category principle, ECP), then a resumptive pronoun appears. The two cases are illustrated in (12) and (13) respectively. (12) *This man, who I wonder whether Mary saw t last week, is a liar. a. ?This man, who I wonder whether Mary saw him last week, is a liar. (13) *These symptoms, which the doctors don’t know what t are, should not be ignored. a. ?These symptoms, which the doctors don’t know what they are, should not be ignored.

The appearance of a resumptive pronoun is then linked to the idea that without it the sentence would be totally ungrammatical whereas with it the sentence is at least interpretable (Kroch 1981). To sum up this point of view, Prince (1990: 2) explains: What mention there has been in the literature of English resumptive pronouns has generally taken the approach that they are ungrammatical but are ways of salvaging a sentence that a speaker has started without realizing that it is impossible or at least difficult to finish it grammatically, ‘making the best of a bad job’. . .

However, not all examples with resumptive pronouns are due to syntactic constraints, as examples (5)–(11) show. Prince (1990) studied such examples where no syntactic constraints are involved, and found that, in both Yiddish and English, resumptives are more frequent with ARCs and DRCs which do not intervene in the construction of the reference of the NP, like DRCs with an indefinite antecedent with a specific interpretation: (14) He bought a house which he’ll move into it in June. (Prince 1990: 14)



TSL[v.20020404] Prn:30/05/2007; 10:12



F: PB16105.tex / p.4 (74)

Rudy Loock

There is therefore a correlation between the referential independence of the socalled antecedent on the one hand and the potential for a resumptive pronoun on the other hand. It is also interesting to notice that Alexopoulou and Keller (2002) have shown that resumptive pronouns do not improve the acceptability of sentences. The examples that we have here can be rephrased without the pronoun, and this time, removing the pronoun does not worsen the acceptability and interpretability of the sentences, contrary to the examples with syntactic constraints. From a purely prescriptive point of view, they are actually better than the versions with the pronoun: (5) a.

(m1 BNC_ARC50) My nickname is “Pan” which I don’t like so much.

(6) a.

(mBNC_ARC26) Well I’d like, Dave Girt, Leeds City, I’d like s some recognition of West Yorkshire’s problems to be evident in the deliberations, which at the moment is absent, it may it may have been taken into account but it’s absent in the exposition . . .

(7) a.

(mMK_ARC4) My foot is narrow in the arch area, which I would’ve expected to widen.

(8) a.

(mRAD15_ARC2) If you do wash your hair then you’re washing away some of the goodness out of you. And which you don’t want to do because you’re carrying a child inside you.

(9) a.

They were just towed across the Midway onto the bridle path, where they were just sitting peacefully.

(10) a.

(mCONV_ARC3) That’s a suggestion of yours which I followed, which I didn’t even want to do.

(11) a.

(mMK_ARC15) ...and I got a new change of clothes, which I wanted you to know.

Prince concludes that in such occurrences the pronoun then acts like an ordinary pronoun, as in an independent clause. However she does not address the question of the role of the relative pronoun, which we shall do here.

(ii) With a resumptive ‘phrase’ (15) (RAD01_ARC1) I’d also recommend er er our website er the Environmental Resources Trust wherei there’s information about my solar home and how to build one [ on the website]i it’s www.ert.net.

. Whenever ‘m’ precedes the reference number of an example, it means that the example has been modified for our purposes.

TSL[v.20020404] Prn:30/05/2007; 10:12

F: PB16105.tex / p.5 (75)

‘Are you a good which or a bad which?’

These examples are hardly ever mentioned, but they do exist. Instead of a resumptive pronoun, the original phrase remains in its position in spite of Wh-movement.

(iii) Without any gap and apparently no antecedent for the relative pronoun (16) (MK_ARC12) She gained a half pound which they were predicting she’d gain five pounds. (17) (MK_ARC7) She won’t do it on the carpet, which I don’t blame her. (18) (MK_ARC1) Those are warm jammies, which it’s been real warm lately anyway. (19) (MK_ARC21) I’m gonna have someone there, just so books aren’t stolen, which I think one was taken last year. (20) (CONV_ARC4) And you got a side dish with it which I had a gratin dauphinois. (21) (CONV_ARC2) And she decided to move out which I think she’s crazy. (22) (TEL03_ARC10) So there’s really actually nothing to be said except what John Bunyan said at the beginning of Paradise Lost – Part II which when he’s sending his book out into the world and the book’s complaining because it feels frightened John Bunyan gives it a pep talk and tells it to pull itself together.

These examples are problematic, as no gap is apparent in the ARC and also, no antecedent seems to be systematically retrievable for the relative pronoun. This can be shown through the impossibility to reformulate (16)–(22) with two separate sentences, which we can easily do with typical examples (2)–(4), and even gapfilled examples (a few examples below): (2) a.

The people of Oz were relieved when Dorothy’s porch crushed the Witch of the East to death. They were scared of her.

(3) a.

This law prohibits employers from HIV testing. Some lawyers would regard it as tendentious.

(4) a.

Dennis says you’re absolutely mental. We were talking about him last night.

(5) b. (mBNC_ARC50) My nickname is “Pan”. I don’t like it so much. (6) b. (mBNC_ARC26) Well I’d like, Dave Girt, Leeds City, I’d like s some recognition of West Yorkshire’s problems to be evident in er the deliberations. At the moment it’s absent, it may it may have been taken into account but it’s absent in the exposition . . . (7) b. (mMK_ARC4) My foot is narrow in the arch area. I would’ve expected it to widen. (8) b. (mRAD15_ARC2) If you do wash your hair then you’re washing away some of the goodness out of you.(And) you don’t want to do that because you’re carrying a child inside you.



TSL[v.20020404] Prn:30/05/2007; 10:12



F: PB16105.tex / p.6 (76)

Rudy Loock Structures with a resumptive pronoun (I)

GAP-FILLED STRUCTURES

Presence of the NP/PP itself (‘resumptive phrase’) (II)

NON-STANDARD ARCS GAPLESS STRUCTURES (III)

Figure 1. Non-standard ARCs

(9) b. They were just towed across the Midway onto the bridle path. They were just sitting there peacefully.

They are never mentioned in the literature, except in Kjellmer (1988) and an unpublished paper by Kuha (1994), who provided me with her examples. To sum up, Figure 1 distinguishes the different kinds of non-standard/atypical ARCs. . Performance error or innovative syntax? Of course one may wonder about the legitimacy of such examples, especially gapless examples: do they represent a real phenomenon of the English language or are they plain performance errors? All grammars of English, including even the most recent such as Biber et al. (1999) and Huddleston & Pullum (2002) consider such examples as ungrammatical. Only the examples with resumptive pronouns are mentioned, generally in footnotes, and are systematically labelled as performance errors: These characteristics are associated with the difficulties of online production, sometimes resulting in constructions that might be considered non-standard or even a disfluency. (Biber et al. 1999: 622) Pronouns used in place of a gap in relative clauses are known as ‘resumptive pronouns’. In some languages they represent a regular feature of relative clause formation, but in English they are ungrammatical . . . (Huddleston & Pullum 2002: 1091n)

As for type II and especially type III examples, they are never mentioned, as it is considered obvious that they are not part of the system. We shall not go into too much detail here, but their frequency of occurrence, especially for resumptive pronouns (although we do not have precise statistics), their existence in writing, especially in Dickens’s Great Expectations as a typical linguistic characteristic of the character of Joe Gargery (26)–(28), and their exis-

TSL[v.20020404] Prn:30/05/2007; 10:12

F: PB16105.tex / p.7 (77)

‘Are you a good which or a bad which?’

tence in other dialects of English (29) seem to be serious clues as to the significance of the phenomenon. (23) “At the funeral we barely talked. So you avoided me a little, which I understand it, believe me, and can sympathize.” (The Body Artist, Don Delillo. 2001. Picador: 58–59) (24) Oh, I dreamed of a true class reunion of our family. Just imagine their faces, if they saw this place. Which, I might add, none of them came. (The Poisonwood Bible, B. Kingslover. 1999. Harper Perennial: 463) (25) She and her friends raised the money and bought a good Land Rover with a rebuilt engine in Atlanta. Which, by the way, Mother’s group has never raised one red cent for me, to help put in upstairs plumbing at the Equatorial, for example. (The Poisonwood Bible: 475). (26) ‘Yes, Joe; but what I wanted to say, was, that as we are rather slack just now, if you would give me a half-holiday to-morrow, I think I would go up-town and make a call on Miss Est – Havisham.’ ‘Which her name,’ said Joe, gravely, ‘ain’t Estavisham, Pip, unless she have been rechris’ened.’ (Great Expectations, C. Dickens. 1861) (27) For, Joe had actually laid his head down on the pillow at my side and put his arm round my neck, in his joy that I knew him. ‘Which dear old Pip, old chap,’ said Joe, ‘you and me was ever friends. And when you’re well enough to go out for a ride – what larks!’ (Great Expectations, C. Dickens. 1861) (28) At last, one day, I took courage, and said, ‘Is it Joe’? And the dear old home-voice answered, ‘Which it air, old chap’. (Great Expectations, C. Dickens. 1861) (29) You can leave at Christmas if your birthday’s in December to February which I think is wrong like my birthday’s March and I have to stay on to May which when I’m 16 in March I could be looking for a job. (Miller 1988)

Kuha (1994) claims that speakers do produce this kind of RC, although they have a tendency to reject them when confronted with such examples. She found such productions from speakers of all ages, from different areas and educational backgrounds: It seems that users include men and women, high school graduates and university professors, natives of California, Indiana, Michigan, and Ohio; their ages range from 20 to 50.; [t]he use of this construction is not necessarily limited to informal situations (. . .) [but] is found more frequently in speech than writing. (Kuha 1994: 1)



TSL[v.20020404] Prn:30/05/2007; 10:12



F: PB16105.tex / p.8 (78)

Rudy Loock

We shall therefore here put aside the performance error theory and consider these examples as part of the English system.2 Their analysis is to bring up a specific role for the relative pronoun.

. The role of the relative pronoun . The double role of a standard relative pronoun It is well-known that the role of a standard relative pronoun is twofold – see e.g. Jespersen (1970), but the idea is already present in Port Royal grammar: –

–

First, as a pronoun, it stands for a referent previously mentioned in the main clause. It clearly has a role as an anaphoric marker. In the case of an ARC, the relative pronoun refers to a nominal antecedent but also, contrary to a DRC, a PP, an AdjP, an AdvP, and also a sentential antecedent in the case of sentential relatives (predicate, whole sentence or even whole paragraph). Second, it relates, or links two clauses, one of which is subordinate to the other.

This double role – anaphoric and connective – gives the pronoun its complexity, since the two functions are synthesized in one marker. Evidence for the two functions of the WH- relative pronoun can be brought to the surface if we consider examples from diachronic (30)–(31) and dialectal English (32)–(33), where the two roles are dissociated. In Old English in particular, the invariant relativiser þe could be followed by a pronoun, as shown by the following examples (CHELVol. 1: 229): (30) . . .& ic gehwam wille þærto tæcan þe hiene (acc) his and I whomever shall thereto direct Rel him of-it lyst ma to witanne would-please more to know ‘and I shall direct anyone to it who would like to know more about it.’ (31) Swa bið eac þam treowum þe him (dat) gecynde bið up heah to So is also to-those trees Rel to-them natural is up high to standanne stand ‘So it is also with trees to which it is natural to stand up straight.’

. This does not mean that no examples can be explained by the occurrence of performance errors, but we are setting aside the hypothesis that performance errors are the sole explanation for the appearance of these seemingly deviant structures.

TSL[v.20020404] Prn:30/05/2007; 10:12

F: PB16105.tex / p.9 (79)

‘Are you a good which or a bad which?’

In some dialects of English, the dissociation of the two functions of the relative pronoun can also be found, such as in Scottish English, in particular for the genitive case (Chevillet 1991: 135; Milroy & Milroy 1993: 111–3): (32) the girl that her eighteenth birthday was on that day the girl whose eighteenth birthday was on that day (33) that’s the man that his house was burnt that’s the man whose house was burnt

. The role of a non standard relative pronoun In our atypical examples, from resumptives to gapless examples, the role of the relative pronoun seems to be reduced to only one function: connective, as the anaphoric role is fulfilled either by the resumptive pronouns or simply unnecessary in the case of gapless examples. As we shall see, it is not coincidental that it is the pronoun which that appears in these examples. Suggesting that which has a peculiar connective function means wondering about the kind of complex inter-clausal relations that exist in such examples, both from a semantic point of view but also from a syntactic point of view. From a semantic point of view, these range from concession to causality or plain coordination, just as in standard ARCs, which then resemble adverbial clauses (Quirk et al. 1985; Cornilescu 1981). For instance, (34) and (35) can be paraphrased, in the world as we know it, as (34a) and (35a). The semantic link is then to be inferred. (34) John, who passed his exam, is unhappy. a. John, as he failed his exam, is unhappy. (35) John, who passed his exam, is unhappy. a. John, although he passed his exam, is unhappy.

The relation can also be plain coordination, as is shown by the frequent possibility to rephrase ARCs, especially when they are in final position, with a coordinated clause. This predominance of the connective function appears through the reformulation of our examples with the replacement of which with a connective word in the broad sense of the word, therefore re-introducing explicitly the implicit semantic link between the two clauses, both for gap-filled and gapless examples:

Gap-filled (5) c.

(mBNC_ARC50) My nickname is “Pan” but/although I don’t like it so much. .



TSL[v.20020404] Prn:30/05/2007; 10:12



F: PB16105.tex / p.10 (80)

Rudy Loock

(6) c.

(mBNC_ARC26) Well I’d like, Dave Girt, Leeds City, I’d like s some recognition of West Yorkshire’s problems to be evident in er the deliberations, for/as at the moment it’s absent, it may it may have been taken into account but it’s absent in the exposition . . .

(7) c.

(mMK_ARC4) My foot is narrow in the arch area, but/yet/while I would1 ve expected it to widen.

(8) c.

(mRAD15_ARC2) If you do wash your hair then you’re washing away some of the goodness out of you. But you don’t want to do that because you’re carrying a child inside you.

(9) c.

They were just towed across the Midway onto the bridle path, and they were just sitting there peacefully.

(10) b. (mCONV_ARC3) That’s a suggestion of yours which I followed, but I didn’t even want to do that. (11) b. (mMK_ARC15) ...and I got a new change of clothes, and I wanted you to know that. (15) a.

(mRAD01_ARC1) I’d also recommend er er our website er the Environmental Resources Trust as/since there’s information about my solar home and how to build one on the website it’s www.ert.net

Gapless (16) a.

(mMK_ARC12) She gained a half pound, although/while they were predicting she’d gain five pounds.

(17) a.

(mMK_ARC7) She won’t do it on the carpet, so/and I don’t blame her.

(18) a.

(mMK_ARC1) Those are warm jammies, and it’s been real warm lately anyway.

(19) a.

(MK_ARC21) I’m gonna have someone there, just so books aren’t stolen, because/as/for I think one was taken last year.

(20) a.

(mCONV_ARC2) She’s decided to move out, and/but I think she’s crazy because. . .

(21) a.

(mCONV_ARC5) And you got a side dish with it, and I had a gratin dauphinois. . .

(22) a.

(mTEL03_ARC10) So there’s really actually nothing to be said except what John Bunyan said at the beginning of Paradise Lost – Part II and when he’s sending his book out into the world and the book’s complaining because it feels frightened John Bunyan gives it a pep talk and tells it to pull itself together

The sentences thus obtained are perfectly acceptable, and even better from a purely prescriptive point of view.

TSL[v.20020404] Prn:30/05/2007; 10:12

F: PB16105.tex / p.11 (81)

‘Are you a good which or a bad which?’

a.

Typical ARC CLAUSE 1

WHICH

CLAUSE 2

b. Atypical ARC CLAUSE 1

WHICH

CLAUSE 2

Figure 2. The role of which in typical and atypical ARCs

In the gap-filled examples, the anaphoric function is fulfilled by the resumptive pronoun/phrase while the connective function is fulfilled by the relative pronoun which. In the gapless examples, only a connective function is fulfilled, by the relative pronoun; no anaphora is needed between the RC and the main clause. The only role that is then left to the relative pronoun is the connection between the MC and the so-called RC. The two schemas (Figure 2) symbolise the role of the relative pronoun in (a) standard ARCs and (b) atypical, gapless, ARCs. . The predominance of which A word needs to be said about the predominance of which, as nearly all our examples here involve the relative pronoun which. It is no coincidence that the relative pronoun which should be the most represented in our corpus of atypical ARCs. In opposition to who(m), where, when. . ., it can have, and only in an ARC, a greater variety of antecedents, including what we call sentential antecedents: a predicate, a whole sentence or even a whole paragraph, with ARCs representing then an independent sentence. These sentential RCs with sentential antecedents have a reputation for being easily rephrased with and this/that/it, although such a transformation is not systematically possible and equivalent to the utterance with an RC. (36) Tennis is possibly the least popular sport at the Olympics, which is quite something when it is up against Greco-Roman wrestling and walking. a. Tennis is possibly the least popular sport at the Olympics, and this is quite something when it is up against Greco-Roman wrestling and walking.



TSL[v.20020404] Prn:30/05/2007; 10:12



F: PB16105.tex / p.12 (82)

Rudy Loock

(37) He went to Paris, which Mary never did. a. He went to Paris, and this Mary never did. (38) The consequences of America’s obsession with domestic television ratings is that the site for the Olympics remains at the mercy of NBC – what Ebersol, to his credit, calls ‘the quintessential example of ugly Americanism’. And if you’re wondering why they wanted to come to Sydney, they didn’t. The only reason the Games are here is that, in the aftermath of the 1989 Tiananmen Square uprising, the IOC had to postpone their long-term plan to take the Olympics to China. That would have suited NBC – and the corporate sponsors who can’t wait to get their hands on the Chinese market. Sydney was considered a viable option, facilities-wise, and won the next vote in 1993 but Australia, with fewer than 20 million people, does not provide the sort of selling bonanza that China would have done. As the AFR puts it, they ‘would love to go to Beijing to foster the spirit of Olympic marketing.’ Which brings us to Athens. For several weeks now, rumours have grown that the city will not be able to meet infrastructure and financial deadlines for the next Games – which could be switched to Seoul, where time is on the Americans’ side, not to mention old business partners. a. And this brings us to Athens. For several weeks now, rumours have grown that the city will not be able to meet infrastructure and financial deadlines for the next Games – which could be switched to Seoul, where time is on the Americans’ side, not to mention old business partners.

The specificity of which is that its antecedent can be vaguer, looser, as is the case for this or that for example. This possibility makes which an anaphoric marker that is linked to its antecedent in a less constrained way. This can explain the attenuation of its anaphoric role to the benefit of its connective role, especially in an ARC which is linked to the MC in a looser way too, compared to a DRC.

. Subordination or coordination? What we notice in our atypical examples is that which is sometimes replaced with a subordinator (because, although, since, whereas. . .) but sometimes also with a coordinator (and, but, yet). This specific use of the relative pronoun which as a connective therefore brings up the question of the nature of the syntactic link between the main clause and the so-called relative clause. Whether ARCs as a whole are derived from a coordinated sentence or not has been debated for a long time. Several theories indeed exist concerning the syntactic relationship between an ARC and its MC:

TSL[v.20020404] Prn:30/05/2007; 10:12

F: PB16105.tex / p.13 (83)

‘Are you a good which or a bad which?’

– – –

There is no syntactic relationship between the two clauses, which are not linked syntactically (e.g. Fabb 1990) An ARC is part of the NP, just like a DRC is. The RC is embedded into the main clause and a subordination link exists between the two (e.g. Jackendoff 1977). An ARC is derived from a coordinated independent sentence (e.g. Emonds 1979). An ARC therefore has the characteristics and behaviour of a main clause.

This last point of view is of particular interest as it sheds light on a specific kind of ARCs, i.e. continuative ARCs (e.g. Jespersen 1970; Cornilescu 1981; Loock 2003, 2005, 2007), whose properties seem to distinguish them clearly from other ARCs: (39) I gave the book to John, who gave it to Mary. (40) She was airlifted to hospital, where she died hours later.

These ARCs are particular, as they make narrative time move forward, a phenomenon traditionally specific to independent clauses (Depraetere 1996: 699). The ARC, like the MC, describes an extra-linguistic event and the two events are on the same level. The semantic behaviour of an ARC is then similar to that of independent clauses. This is why some linguists (Daalder 1989 and Kuntsman 1994) have postulated that the relative pronoun takes on the role of a coordinating conjunction, in spite of its retaining an anaphoric role. Emonds (1989) puts forward several arguments in favour of a derivation of ARCs as a whole from coordinated sentences, like root phenomena, the presence of adverbs typical of MCs, the fact that ARCs can convey an independent speech act, most of which can be discussed. But an ARC, whether continuative or not, still retains the traditional characteristics of a subordinate clause. In languages like Dutch, for which the verb is in final position in subordinate clauses, the verb will still be found in final position in continuative ARCs: (41) Ik heb het boek aan Jan gegeven, die het dan heeft doorgegeven aan Marie. I have the book to John given, which it then has given to Mary I gave the book to John, who then gave it to Mary.

Arnold’s (2004) proposal that ARCs behave syntactically like subordinate clauses, but are interpreted like independent clauses represents a compromise between the subordination and coordination hypotheses. This would explain why in Dutch for example an ARC has the syntactic behaviour of a subordinate clause but can from a semantic point of view be interpreted like a main clause (independent speech act, presence of adverbs typical of main clauses. . .). It is of course well beyond the scope of this paper to answer that thorny question. All that can be said here is that some of our atypical ARCs can be paraphrased with a coordinated sentence, which is an argument in favour of a syntactic relationship sometimes closer to coordination than subordination.



TSL[v.20020404] Prn:30/05/2007; 10:12



F: PB16105.tex / p.14 (84)

Rudy Loock

. Role of the relative pronoun in the organization of discourse One may wonder about the role of the relative pronoun in such atypical ARCs, as opposed to coordinators/subordinators which do exist but are rejected in favour of a relative pronoun outside its traditional role, and also as opposed to a juxtaposed clause. One could of course say that this has to do with the constraints on speakers’ performance, but it seems legitimate to think that a very specific role can be attributed to such a use of the relative pronoun. And this use seems to be supported by prosodic evidence as well. According to Kjellmer (1988), who discussed gapless examples, the relative pronoun – which he calls adverbial or conjunctional (Kjellmer 1988: 161), the use of the relative pronoun in such atypical examples corresponds to the necessity of oral English sentences to be introduced rather than juxtaposed without any link between them. His corpus study indeed shows that most clauses in oral English are introduced by an element in order to avoid abruptness. The relative pronoun which then appears to take on that role in what he calls ‘substandard’ oral English: The use of clause-opening which will be thought of as a mark of correctness generally. It is reasonable to suppose that clause-opening relatives thus come to be regarded as practically redundant elements, to be used at the speaker’s discretion for stylistic purposes. (Kjellmer 1988: 161)

This idea is interesting but we would like to go further by positing that such a use of the relative pronoun in discourse is a floor-keeping manoeuvre. The idea is that when starting a clause with a relative pronoun, a speaker has fewer chances to be interrupted than when uttering an independent clause. The speaker then starts with a relative pronoun to show his/her intention to proceed, but actually utters an independent clause, in which no gap appears, either because it is filled with a resumptive pronoun or because it simply does not exist. In both cases the part following the relative pronoun is an independent clause which is linked to the main clause thanks to the so-called relative pronoun. This strategy is supported by prosodic evidence. It is generally assumed that in the case of ARCs, a pause or a break in the melody exists between the MC and the RC, specifically between the antecedent and the relative pronoun – although this is not systematic. However the study of our corpus of ARCs as a whole shows that pauses sometimes occur at an unexpected place, i.e. after the relative pronoun. This is something we noticed too when atypical examples came up. Although this does not mean that the intonation unit is closed after instead of before the relative pronoun, there is a delaying of the pause. The speaker therefore interrupts his/her speech after the relative pronoun, thus leaving a clause unfinished with the relative pronoun in “mid-air”, clearly indicating that s/he has not finished speaking. The addressees are therefore invited not to interrupt.

TSL[v.20020404] Prn:30/05/2007; 10:12

F: PB16105.tex / p.15 (85)

‘Are you a good which or a bad which?’

We unfortunately have no recorded examples of atypical relative clause attesting such a phenomenon (they are hard to collect), but we do have examples of typical ARCs in which such a delaying of the pause occurs: (42) (RAD03_ARC4) I think we’d be looking at a different situation where | {1030ms} companies would be dropping from record profits to er to breaking even in a in a month (43) (RAD07_ARC1) You know everything. . . it’s an entire roller coaster for the audience | and the audience just don’t sit there and watch there | but become actually engulfed in what they’re seeing which | {480ms} is why it’s won you know Pulitzer Prize for drama

This kind of strategy is called floor-keeping manoeuvre and is well-known as a conversational strategy, mentioned by Biber (1988) and Wennerstrom (2001) in particular. As is emphasized by Wennerstrom, “it is safer to pause in midphrase than at a phrase boundary if one wishes to avoid interruption.” (Wennerstrom 2001: 173). Wichmann (2000) shows that such a configuration is very frequent in spontaneous speech, the boundary of a clause occurring not before but after a conjunction. Interestingly enough, as Kuha (1994) notes, Schleppegrell (1991) found that a similar phenomenon occurs with the subordinator because, which can go beyond the expression of a semantic link of causality, playing a role in the organization of discourse, as in the following example: (44) The fifth position break is in a lot of dances. Especially in a lot of Latin dances. Because this is the fifth position break. (Demonstrates). (Schleppegrell 1991)

The role of because is then not to introduce an explanation of what has just been said but to express the speaker’s intention to proceed, which Schleppergrell calls “further elaboration to hold the floor.” The clear advantage of leaving the relative pronoun in mid-air, used according to the analysis suggested here as a connector between two clauses, instead of a coordinator like and for instance, is that the addressee(s) are less likely to interrupt the speaker after what sounds like a relative pronoun, even if the rest of the sentence shows no gap as a traditional RC would but instead represents an independent clause.

. Conclusion In this paper, we hope to have shown that there is a specific use of the so-called relative pronoun which in these atypical examples. As its role is exclusively connective, it plays a role in the organization of the sentence and of discourse as a whole.



TSL[v.20020404] Prn:30/05/2007; 10:12



F: PB16105.tex / p.16 (86)

Rudy Loock

The Wh- pronoun fulfils a particular role in atypical ARCs: – –

–

It has a connective function only. Contrary to what happens in standard ARCs, it fulfils no anaphoric function because what follows is a gap-filled/gapless clause, in other words a clause with the linear order of an independent clause. Regarding the organization of discourse, the pronoun is used by speakers to keep the floor and show their intention to proceed, which is why atypical ARCs are not to be found in writing, except in fictitious dialogues.

References Alexopoulou, T. & F. Keller. 2002. “Resumption and Locality: a Crosslinguistic Experimental Study.” In Papers from the 38th Meeting of the Chicago Linguistic Society Vol. 1. Chicago Arnold, D. 2004. “Non-Restrictive Relative Clauses in Construction-Based HPSG.” In Proceedings of the 11th International Conference on Head-Driven Phrase Structure Grammar, Stefan Müller, ed., CSLI Publications, Stanford, CA. Biber, D. 1988. Variation across speech and writing. Cambridge: CUP. Biber, D., Johansson, S., Leech, G., Conrad, S. and E. Finegan. 1999. Grammar of spoken and written English. London: Longman. The Cambridge History of the English Language (CHEL). Cambridge University Press. Vol. 1 and 2. Chevillet, F. 1991. Les Variétés de l’anglais. Nathan Université. Chomsky, N. 1995. The Minimalist Program. Cambridge: MIT Press Cornilescu, A. 1981. “Non-restrictive Relative Clauses, an Essay in Semantic Description.” In Revue roumaine de linguistique XXVI, 1: 41–67. Daalder, S. 1989. “Continuative Relative Clauses.” In Sprechen und Hören: 195–207. Depraetere, I. 1996. “Foregrounding in English Relative Clauses.” In Linguistics 34: 699–731. Emonds, J. 1979. “Appositive Relatives Have No Properties.” In Linguistic Inquiry 10, 2: 241–3. Fabb, N. 1990. “The Difference between English Restrictive and Nonrestrictive Relative Clauses.” In Journal of Linguistics 26: 57–78. Huddleston, R. and G. K. Pullum. 2002. The Cambridge Grammar of the English Language. Cambridge: Cambridge University Press. Jackendoff, R. 1977. X-bar Syntax: A Study of Phrase Structure. Cambridge, MA: MIT Press. Jespersen, O. [1927] 1970. A Modern English Grammar on Historical Principles. Vol III. Londres: George Allen & Unwin. Kjellmer, G. 1988. “Conjunctional/adverbial which in substandard English.” In Studia Anglica Posnaniensia 21: 125–37. Kroch, A. ?1987. “Limits on the Human Brain Generator.” In TINLAP 87–1043, http://acl.ldc.upenn.edu/T/T87-1043.pdf Kroch, A. 1981. “On the role of resumptive pronouns in amnestying island constraints violations.” In CLS 17: 125–35. Chicago: University of Chicago. Kuha, M. 1994. “Attitudes towards users of coordinate which.” Ms.

TSL[v.20020404] Prn:30/05/2007; 10:12

F: PB16105.tex / p.17 (87)

‘Are you a good which or a bad which?’

Kuntsman, P. 1994. “Relatif et liaison: le cas du relatif dit de <>.” In Le Moyen Français. Philologie et linguistique. Approches du texte et du discours. Actes du VIII◦ colloque international sur le moyen français, edited by B. Combettes et S. Monsonego. Paris: Didier Erudition: 517–27. Loock, R. 2003. “Les Fonctions des propositions subordonnées relatives <> en discours.” In Anglophonia 12: 113–31. Loock, R. 2005. Appositive Relative Clauses in Contemporary Written and Spoken English: Discourse Functions and Competitive Structures. Ph.D. Dissertation, Lille III University and UMR 8528 SILEX, CNRS, France. Loock, R. 2007. “Appositive Relative Clauses and their Functions in Discourse.” In Journal of Pragmatics 39: 336–62. Miller, J. 1988. “That: A Relative Pronoun? Sociolinguistics and Syntactic Analysis.” In Anderson, J. M. et Macleod, N. (eds): Edinburgh Studies in the English Language. Edinburgh: John Donald Publishers Ltd: 113–19. Milroy, J. & L. Milroy (eds). 1993 . Real English: The Grammar of English Dialects in the British Isles. London, New York: Longman. Prince, E. 1990. “Syntax and discourse: a look at resumptive pronouns.” In Hall, K. et al. (eds), Proceedings of the Sixteenth Annual Meeting of the Berkeley Society: 482–97. Prince, E. 1995. “On kind-sentences, resumptive pronouns, and relative clauses.” In Guy, G., Baugh, J. et D. Schiffrin (eds), Towards a social science of language – a festschrift for William Labov. Cambridge: CUP: 223–35. Quirk, R. et al. 1985. A Comprehensive Grammar of the English Language. London: Longman: 1238–62. Schleppegrell, M. J. 1991. “Paratactic because.” In Journal of Pragmatics 16: 323–37. Wennerstrom, A. 2001. The Music of Everyday Speech: Prosody and Discourse Analysis. Oxford University Press. Wichmann, A. 2000. Intonation in Text and Discourse: Beginnings, Middles and Ends. Edinburgh: Pearson Education Limited.



TSL[v.20020404] Prn:30/05/2007; 10:13

F: PB16106.tex / p.1 (89)

From temporal to contrastive and causal The emergence of connective after all Diana M. Lewis Université Lumière Lyon 2

This paper considers the historical evidence on after all for the light it can shed on the mechanisms of change at work in the development of connectives. The paper describes usage of after all in present-day English and traces the emergence of a connective from the prepositional phrase as part of a growing polysemy. It argues that the connective sense has not resulted from any metaphorical extension of the temporal sense, nor from an ad hoc innovation, but rather has been a metonymic evolution. The analysis is based on diachronic corpora of written English, and on corpora of present-day written and spoken English.

.

Introduction

One of the fascinating aspects of connective expressions is their characteristic multifunctionality: expressions used as connectives typically occur as non-connective expressions too, and sometimes can be used with more than one connective function. This multifunctionality and its origins have been the subject of much recent work on discourse connectives and discourse particles (e.g. Aijmer 2002; Brinton 1990, 1996; Fischer 2006; Mosegaard-Hansen 1998; Traugott & Dasher 2002). This paper addresses the question of how the various senses develop diachronically. It argues that they develop with the persistent use of particular rhetorical patterns containing the expression. The English connective after all provides an example of an expression which seems to have three distinct polysemies: a temporal sense, a counter-expectation sense and a justificative connective sense. Examples (1)–(3) illustrate these senses. In example (1), after all functions as an adverbial of time with scope over VP.

TSL[v.20020404] Prn:30/05/2007; 10:13



F: PB16106.tex / p.2 (90)

Diana M. Lewis

(1) – Oh god I’m tired. You woke me up this morning I tell you Sid. – What time did you get to bed after all? (BNC-Conv-KDY)1

Example (2) shows after all in a counter-expectation context. It is arguably temporal, since the expectation that the book will be a history of the footnote must precede the discovery that it is not. But the main idea is the contrast between expectation and reality rather than between two points in time. (2) Another dismaying early discovery is that the book is not, after all, going to be a history of the footnote, as the title implies. (Book Reviews Corpus, Sunday Times 31.08.97)

In example (3), after all has a fully connective sense, linking the second idea to the first idea in a relation of justification2 (3) – Well do it on the canvas, after all that was what I bought it for (BNC-ConvKBB)

It has scope over the whole clause ‘that was what I bought it for’ and therefore functions as, and seems to have lexicalized as, a discourse-connective sentence adverb.

. Data Data are taken from both full-text and non-full-text sources. The full-text presentday English (PDE) sources are the periodicals section of the British National Corpus (BNC) (420 tokens of after all), the 4-million word demographically-sampled section of the BNC (65 tokens), a 1-million word book reviews corpus (91 tokens) and a 1.5-million word corpus of written-to-be-read speeches (63 tokens). Historical data are taken from the Helsinki Corpus, the Lampeter Corpus, the Corpus of Early English Correspondence Sampler (CEECS), and a 2.8-million word collection of texts from the sixteenth to the early twentieth centuries.

. BNC-Per: periodicals section of the British National Corpus; BNC-Conv: demographicallysampled section of the British National Corpus; final codes are BNC filenames. . ‘Justification’ is used here in the sense of Mann and Thompson (1987). It refers to a causative connective relation where the reader’s understanding of the idea introduced by after all increases his/her readiness to accept the writer’s right to present the previous idea.

TSL[v.20020404] Prn:30/05/2007; 10:13

F: PB16106.tex / p.3 (91)

From temporal to contrastive and causal

100

Frequency pmw

80 60 Justificative Contrastive Temporal

40 20

s Co nv er sa tio n

Sp

ee

ch e

iew s Re v

Pe rio

di

ca

ls

0

Figure 1. The distribution of after all across four PDE genres

. After all in PDE . Distribution Data from the BNC suggest that the overall frequency of after all is around 40 per million words (pmw). Figure 1 shows the distribution of after all in the PDE fulltext data. The figure is based on all occurrences of after all, whether Adv, PP, or P + predeterminer. In formal registers involving argumentation, therefore, after all is primarily used as a justificative. The proportion of justificative tokens is very similar across Periodicals, Reviews, and Speeches, at around 75%. Conversation, however, shows a very different usage pattern. Not only is the overall frequency lower than in written language, but fewer than 30% of occurrences are justificative. More than half the occurrences are temporal and consist of the construction after + NP where the NP takes the form all + Det. + N. . Counterexpectation after all In these temporal after + NP constructions, however, a contrast can almost always be inferred due to a contrast in the context between expectation and reality, as in (4): (4) cos I thought I had to have a tetanus .. and she goes ..oh you don’t need it . . . after all that psyching myself up for it (BNC-Conv-KP5)



TSL[v.20020404] Prn:30/05/2007; 10:13



F: PB16106.tex / p.4 (92)

Diana M. Lewis

In (4), the expectation (I thought I had to have a tetanus), the reality (you don’t need it) and the event pointed to by after (psyching myself up for it) are all expressed. But the expectation can be implicit, especially in sequences like after all that time, after all that work, after all that, and above all, of course, in the adverb after all. Example (5) is the opening sentence of a newspaper article about cricket. (5) David Gower will tour India after all - as part of Sky Sports’ commentary team. (BNC-Per-CBG)

It can only be understood as presupposing a generalized expectation or belief (that David Gower would not tour India) and indicating that the expectation is unfulfilled. This counter-expectation after all is characterized in present-day standard British English by a stress on the last syllable. . Justificative after all The justificative sense of after all requires a previous idea to link back to; it is always connective.3 Blakemore (1996: 337) notes that (justificative) after all can be used in non-linguistic contexts. Her example is (6) “[the speaker takes an extremely large slice of cake] After all, it is my birthday.” (Blakemore 1996: 338)

But even in this context, after all is connective. An immediately previous idea is required for after all to refer back to. The fact that in certain circumstances, such as those of Blakemore’s example, that idea can be an event, rather than a linguistic reference to an event, does not alter the requirement: a previous idea must be present in the context. (By contrast, counter-expectation after all requires no special prior event or statement: it contains a presupposition that the hearer retrieves – the notion that the antithesis was expected – and it can therefore begin a discourse.) The idea in the scope of justificative after all is presented neither as the reason for which the first event is carried out (though it may be the reason), nor as circumstantial evidence for the prior claim. Rather, it is a justification consisting of a generalization, truism or undisputed fact which makes the claim seem reasonable. (Nevertheless, the distinction between causative, evidential and justificative relations is not always clear-cut in practice.) The main argument is foregrounded,

. As Traugott (1997) points out, it is doubtful whether “there is a real distinction to be based on whether a DM [discourse connective] can or cannot initiate a turn, despite efforts to find such a distinction” (1997: 7). The evidence of after all, however, suggests there is a distinction between a marker that allows a presupposition to be retrieved and one which does not.

TSL[v.20020404] Prn:30/05/2007; 10:13

F: PB16106.tex / p.5 (93)

From temporal to contrastive and causal

Table 1. After all in counter-argument pattern Argument

After all justification

Counter-argument

We’re sure Moby’s behaviour is simply a phase.

After all, he’s only been experiencing the outside world since his vaccination course was completed a few weeks ago.

But, with him already so large and strong, we’re concerned . . . (BNC-Per-A17)

It may seem misguided and unnecessarily pedantic to take issue with the White Paper’s notion of punishment.

Punishment is just a word after all: nothing more than convenient political rhetoric ...

However, this fails to acknowledge a fundamental axiom of social work practice; namely that language not only reflects but also shapes social reality. (BNC-Per-ALP)

Panama’s President Francisco Rodríguez should have been there, too

(it is a Group of Eight meeting, after all),

but his country is in the doghouse . . . (BNC-Per-A3U)

and the idea introduced by after all is backgrounded (example (3) above and examples (7)). (7) a.

. . . effective disciplines to ensure that quality of service targets are met . . . are an important consumer safeguard . . . A reduction in quality of service is, after all, a price rise by another name (Speeches Corpus, Beckett 30.06.97) b. Hitler’s generals were not as hopeful as the Führer, but they trusted his star: he had, after all, proved them wrong, again and again (Book Reviews Corpus, Sunday Times 03.05.98)

The idea in the scope of justificative after all thus supports a main claim of the speaker/writer. Further, after all is commonly used in a rhetorical pattern of counter-argument consisting of [[Argument + [after all + Justification]] + Counter-argument]. This pattern is illustrated in Table 1. In this pattern after all is used to support an uncertain idea – one that cannot be verified or is not the case - and justifies the belief that it is or might have been the case. The idea is therefore always modalised (for example by we’re sure, may seem, should in Table 1). One sixth of occurrences of justificative after all in the periodicals sub-corpus conform to this pattern.



TSL[v.20020404] Prn:30/05/2007; 10:13



F: PB16106.tex / p.6 (94)

Diana M. Lewis

. The development of after all . From PP to Adverb Then seek this path that I to thee presage, Which after all to heaven shall thee send. Spenser, The Faerie Queen, 1590

After all became a connective relatively recently (not earlier than the eighteenth century, and arguably later). It comes from the temporal after plus a NP. It is a somewhat unusual connective in that it comes from uses of prepositional after occurring predominantly clause-initially and only later moves to final position. From at least the mid sixteenth century, after all can be observed in the temporal sense of ’after some unspecified events that can be inferred from the context’, as in examples (9a, b): (8) a.

. . . and doctur Whyt bysshope of Lynkolne dyd pryche at the sam masse; and after all they whent to his plasse to dener (1559, Helsinki, Machyn, Diary) b. there he made me right welcome, both with varietie of fare, and after all, hee commanded three of his men to direct mee to see his most admirable Colemines (1630, Helsinki, Taylor)

After all here is a PP functioning as an adverbial but is arguably already lexicalizing into an adverb. Over the seventeenth and eighteenth centuries, two changes seem to occur, apparently simultaneously. One is that after all acquires contrastive implicatures as a result of its collocation with contrastive contexts. This use results in due course in the counter-expectation after all of PDE. The other change is that the sense of after all as the semi-lexicalized adverbial expression deriving from P+N, broadens and becomes more abstract, to mean ‘in the end’, and finally to introduce a conclusion or generalization.4 It is this more abstract use that develops into the PDE connective. . Development of counter-expectation after all From the early seventeenth century on, a pattern of [[After all N] + [contrasting idea]] can be observed; this pattern continues into PDE. Common collocations are expressions of endeavour such as pains, labour, travail, effort. Examples (8) show occurrences from the sixteenth century up to PDE. . The entry for after all in Samuel Johnson’s dictionary of 1755 is good evidence of lexicalization as a ‘conclusive’ adverb: “after all: When all has been taken into the view; when there remains nothing more to be added; at last; in fine; in conclusion; upon the whole; at the most.”

TSL[v.20020404] Prn:30/05/2007; 10:13

F: PB16106.tex / p.7 (95)

From temporal to contrastive and causal

(9) a.

Yet after all heuynesse, penaunce, and dysconfyture, She reioysed in soule. (1513, OED) b. we thought good to set down one example more to solace your mindes with mirth after all these scholasticall preceptes, which can not but bring with them . . . much tediousnesse (1569, Puttenham) c. .. if after all the Labour and Pains of Acquiring that Knowledge .. their bread shall be taken off their Trenchers .. by every Illiterate Pretender (1712, Lampeter, Science) d. It must be a shock to find that, after all your effort and in spite of your qualifications, Price Waterhouse does not have any work for you. (BNCPer-AJ2)

All these examples contain a contrast, in that the later of two events is quite unexpected, given the earlier event (rejoicing after suffering, mirth after scholastical precepts, absence of reward after effort, etc.). So although the after + NP sequence is temporal, the contextual interpretation includes a contrastive implicature. From about the mid seventeenth century, the adverbial phrase or adverb after all occurs in contrastive contexts where the temporal succession idea is less important, and the earlier events are presupposed rather than mentioned. The temporal sense is thereby considerably weakened, as in examples (10a, b). (10) a.

. . . my Lord avoided speaking with him, and made him and many others stay expecting him, while I walked up and down [with my Lord] above an hour . . . And yet, after all, there has been so little ground for his jealousy of me, that I am sometimes afraid that he do this only in policy to bring me to his side by scaring me (1663, Pepys, Diary) b. I have revolved this Sentence in my Mind till I have quite tired myself, but cannot, after all, find any Meaning in it. (1739, Anon, Review of Hume)

The contrastive relationship between the two ideas is marked by a conjunction (yet and but in example 10a,b). The contrast is not always marked by such a conjunction, however. Examples (11) show clearly contrastive contexts (but without an overt connective) where after all itself is likely to be interpreted as contrastive. (11) a.

. . . they thought themselves past danger & were after all Cast away [at sea] (1688, Newdigate) b. A man who has studied logic all his life may be, after all, only a petulant wrangler. (1774, OED)

Regular occurrence in contrastive contexts is hypothesized to have strengthened the contrastive implicature in after all, so that it has come to indicate counterexpectation when there is no but present, and when there is no explicitly mentioned contrasting event. It becomes a signal of counter-expectation, meaning ‘de-



TSL[v.20020404] Prn:30/05/2007; 10:13



F: PB16106.tex / p.8 (96)

Diana M. Lewis

spite what might have been thought’. It does not become a connective. Contrastive conjunctions continue to collocate with after all in PDE. Adverbial after all, as mentioned above, comes to mean simply ‘in the end’, when all stands for unspecified events. It occurs both with overtly contrastive contexts and with contexts that are not obviously contrastive. Over the seventeenth and eighteenth centuries, the contexts of after all tend to be compatible with both a counter-expectation reading and an ‘in the end’ reading, as in example (12). (12) We do not now again talk so much as formerly of Don Juan’s journey into Italy, which, after all, it’s believed his Highness is not thoroughly disposed to (1675, The London Gazette, Nov 4–8)

The contexts are often polar or scalar, insofar as the idea in the scope of after all is typically something smaller than, or less than, expected, or is the contrary of what was expected. . Emergence of connective after all “His form is ungainly – his intellect small – ” (So the Bellman would often remark) “But his courage is perfect! And that, after all, Is the thing that one needs with a Snark.” Lewis Carroll, The Hunting of the Snark: an Agony in Eight Fits, 1876

The development of ‘conclusive’ after all thus overlaps with that of the counterexpectation marker. In fact, the counter-expectation and conclusive readings of after all, one or the other more salient depending on context, continue to co-exist well into the nineteenth century. It is the conclusive use that develops into justificative after all. ‘In the end’ naturally occurs at the end of an argument, as in the case of Lewis Carroll’s snark, and especially to introduce a conclusion or a generalization (examples 13a, b). (13) a.

they, perhaps, loved, where we only pity; and were stern and inexorable, where we are not merciful, only irresolute. After all, the merit of a man is determined by his candour and generosity to his associates,. not by moderation alone. (1767, Ferguson) b. both ladies thanked me a very great deal more than I deserved; for after all it was a business matter, and a resident patient was the very thing that I needed. (1894, Conan Doyle)

In these examples, after all no longer has any temporal reference, but refers to a final position in a line of argument: it can be paraphrased by ‘in sum’ or ‘after considering everything’ (see footnote 4). Since a conclusion often follows necessar-

TSL[v.20020404] Prn:30/05/2007; 10:13

F: PB16106.tex / p.9 (97)

From temporal to contrastive and causal

ily from what precedes, the notion of causation may be present. But for examples (13a, b) there is no reason to assume that causation has yet semanticised in after all. Rather, to ensure a causative interpretation of (13b), the causative marker for is needed. The salient sense of after all in these examples is simply the conclusive ‘in the end’.5 In PDE, initial after all, as in example (3), is unambiguously a connective of justification. As an introduction to a conclusion or generalization, after all comes to be associated with speaker commitment to the truth of the proposition. The claim presented in the scope of after all is an acknowledged one, that can be taken for granted, so that after all comes to mark an accepted, backgrounded idea which is mentioned in order to provide support for a new and more salient idea. (This backgrounding function is, of course, already present in the PP temporal adverbial after all NP: after usually introduces an event that has occurred and that is expressed as a given (e.g. ‘all their pains’ presupposing ‘they took pains’)). . Summary of historical development Following the development of the adverb, around the end of the seventeenth century, after all tends to occur in two main rhetorical patterns: (i) where the end is an unexpected outcome, giving rise to counter-expectation after all; (ii) where an argument is summed up, giving rise to justificative after all, which is connective. This functional split is sketched in Fig. 2.

temporal after + all → ‘in the end’

→ despite predictions/beliefs → Counterexpectation after all → to sum up → ‘as is known’ → Justificative (connective) after all

Figure 2. Functional split of after all

There is no evidence in the data of any sudden metaphorical shift or innovation. Rather, the data suggest long periods of slow crystallization of counter-expectation implicatures, and later causative implicatures, due to the recurrence of specific rhetorical patterns of argumentation and information structuring. This functional split is relatively recent. In PDE, initial after all is always interpreted as connective, so that it is no longer possible for other types of after all to . Traugott (1997) claims that the justificative use dates back to the early eighteenth century, but in her example after all (Berkeley 1713) follows the causative connective for. It seems more likely from the data that at that period the causal notion was carried mainly by connectives such as for and as. However, causal implicatures would surely have attached to the apparently frequent sequence for after all.



TSL[v.20020404] Prn:30/05/2007; 10:13



F: PB16106.tex / p.10 (98)

Diana M. Lewis

occur in initial position. But until the early twentieth century, initial after all could still be interpreted as conclusive or as counter-expectation (14). (14) To his horror, Mr Lloyd George . . . discovered that he could not in five days persuade the President of error in what it had taken five months to prove to him to be just and right. After all, it was harder to de-bamboozle this old Presbyterian than it had been to bamboozle him (1919, Keynes)

The fact that this construction is no longer possible in PDE shows that the connective sense has thoroughly semanticised. (By contrast with after all, the idiom after all’s said and done survives in PDE in initial position for introducing a conclusion.)

. Conclusion Although justificative after all may appear retrospectively like a metaphorical use of an originally temporal after, there is no evidence of a metaphorical leap in the history of the expression. Nor is there evidence for any ad hoc innovation. The contexts in which the connective develops are not quirky or innovative ones, but regular, established ones where the expression has its prototypical sense. The adverbialization of after all is therefore best seen as an evolved shorthand, a compression of information, or a form of economy. The outcome is a subjectification in terms of semantic change: from marking a temporal relation, after all comes to express a speaker attitude, both as a counter-expectation marker and as a connective expressing the rhetorical relation of justification. There are also consequences in terms of information structure, with connective after all acquiring a backgrounding function, so that in PDE new information can no longer be presented in the scope of initial after all. While PDE usage of after all as part of an adverbial PP has changed little since the seventeenth century, new polysemies have developed alongside. Older senses did not change into newer ones; rather, there has been a gradual reweighting of implicatures in regular sequences that favoured holistic interpretations, leading to the polysemy we see in PDE. Although after all is fairly clearly split into three polysemies, the relatedness of the three uses is perceptible. In around half the cases of temporal after all in the PDE data, for instance, there is a contrast between the earlier and the later event, as in (4) above. This paper has argued that the polysemies of after all have developed slowly and gradually, as a result of co-occurrence with particular types of contrastive and conclusive contexts, and in particular recurrent rhetorical sequences. It is the inferred relation between sequential arguments and/or counterarguments that gradually led to the emergence of after all as a full discourse connective.

TSL[v.20020404] Prn:30/05/2007; 10:13

F: PB16106.tex / p.11 (99)

From temporal to contrastive and causal

References Aijmer, K. 2002. English Discourse Particles. Evidence from a corpus. Amsterdam: John Benjamins. Blakemore, D. 1996. “Are apposition markers discourse markers?” Journal of Linguistics Vol. 32(2), 325–347. Brinton, L. J. 1990. “The development of discourse markers in English”. In J. Fisiak (ed.) Historical Linguistics and Philology, 45–71. Berlin and New York: Mouton de Gruyter. Brinton, L. J. 1996. Pragmatic Markers in English: Grammaticalization and Discourse Functions. Berlin and New York: Mouton de Gruyter. Fischer, K (ed.) 2006. Approaches to Discourse Particles. Amsterdam: Elsevier. Johnson, S. 1996. Dictionary of the English Language (First edition of 1775 and Fourth edition of 1773, on CD-ROM), edited by A. McDermott. Cambridge: Cambridge University Press. Mann, W. C. and Thompson, S. A. 1987. Rhetorical Structure Theory. A Theory of Text Organization. ISI Report RS-87-190. Marina del Rey, CA: Information Sciences Institute, University of Southern California. Mosegaard Hansen, M.-B. 1998. The Function of Discourse Particles. A Study with Special Reference to Spoken Standard French. Amsterdam: John Benjamins. Traugott, E. C. 1997. ”The discourse connective after all: a historical pragmatic account”. Paper presented at the Sixteenth International Congress of Linguists, Paris, July 1997. Traugott, E. C. & Dasher, R. 2002. Regularity in Semantic Change. Cambridge: Cambridge University Press.



TSL[v.20020404] Prn:26/03/2007; 15:04

 

Discourse strategies

F: PB161P3.tex / p.1 (101)

TSL[v.20020404] Prn:10/05/2007; 13:43

F: PB16107.tex / p.1 (103)

Orchestrating conversation The multifunctionality of well and you know in the joint construction of a verbal interaction Barbara Le Lan Université Paris IV – Sorbonne

This paper investigates the multifunctionality of well and you know in authentic spontaneous conversations. In the wake of Östman (1981), Schiffrin (1987), Aijmer (2002), Aijmer and Simon-Vandenbergen (2003), I define well and you know as metacommunicative, deictic and heteroglossic markers. These pragmatic properties can be related to their semantics and syntax as non DMs (well adjective/adverb and you know as a clause). This brief semantic analysis enables us to explain the frequent co-occurrences of both markers with specific types of utterances like personal evaluations, generic or hyperbolic statements, and allows a more thorough description of well and you know in discourse as the signals of mental movements where the actual landmark is the addressee.

Introduction Among the numerous names that have been given to these particular “connectives”, perhaps the best is “mystery particles” (Longacre 1976), not only because it is a recognition in itself of their elusive and multifunctional character, but also because it conveys the idea that there is something going on underneath the semantic and rational surface of language every time these particles are used. But when these small words are examined more closely, they rapidly cease to belong to the underworld of conversation and the main question quickly becomes: how is it possible that these prefabricated forms, or prefabs, which seem to do so little, actually enable speakers to do so much? This article presents the provisional conclusions I have come to after a functionalist and corpus-based analysis of 400 wells and 200 you knows as they are used in the authentic spontaneous conversations that were transcribed in A Corpus of English Conversation (CEC from now on, Quirk & Svartvik 1980). Östman recognizes three major functions shared by all discourse markers (DMs from now on): “discourse organizing”, “interactionsignalling”, and “involvement signalling” (1995: 99). This threefold distinction will

TSL[v.20020404] Prn:10/05/2007; 13:43

F: PB16107.tex / p.2 (104)

 Barbara Le Lan

be further looked into, and corresponds roughly to that made by Erman (2001) between “textual monitors”, “social monitors”, and “metalinguistic monitors”. The aim is not to enumerate the 14 different functions that have been found for well, and the 11 functions that have been found for you know so far. Rather, I will suggest a unified account of how well and you know function in spontaneous conversation and propose an explanation of some of the 14 different uses I have distinguished in my corpus analysis of well and some of the 11 different uses I have made out in my analysis of you know, keeping in mind that most of these uses overlap in actual discourse. The terms ‘discourse’, ‘speech’, ‘interaction’ and ‘conversation’ will be used interchangeably. The following analyses of well and you know will be the starting points of the present study. More particularly, I will discuss the dynamic nature of both markers in verbal exchanges that these analyses emphasize: “[when using you know,] the speaker strives towards getting the addressee to cooperate and/or to accept the propositional content of his utterance as mutual background knowledge” (Östman 1981: 17). “[T]he general function which we propose for well [...is] modal in a broad sense, that is, to turn the utterance into a heteroglossic one, signalling the speaker’s awareness of the heterogeneity of views, positioning the utterance in the context of preceding and following texts. By doing this, speakers at the same time connect their utterance to other utterances and orient to addressees’ expectations.” (Aijmer & Simon-Vandenbergen 2003: 1155). I will first focus on the discourse organizational role of well and you know and propose a semantic analysis of their ‘original’ meaning as non DMs that might account for their role as discourse connectives. In the wake of this brief semantic analysis, the interactional function of both markers will be investigated and finally, I will show that the interpretation of well and you know as expressions of mental movements can be highly significant and revealing.

.

Well and you know as discourse organizers: indexicality and semantic ‘origin’

. Well and you know as focalization cues In the course of this research, it turned out that the deictic nature of discourse markers was crucial in accounting for their multifunctionality, as many linguists have shown: “[w]e need another dimension of analysis if we are to go further in understanding the contribution of DMs to coherence: I suggest that this dimension is deixis and that all markers have indexical functions”. (Schiffrin 1987: 322, author’s emphasis). “The most important property of discourse particles is their indexicality. This property explains that they are linked to attitudes, evaluation, types of speakers and other dimensions of the communication situation”. (Aijmer

TSL[v.20020404] Prn:10/05/2007; 13:43

F: PB16107.tex / p.3 (105)

Orchestrating conversation 

2002: 5). As indicated by Schiffrin and Aijmer, DMs are used to point out particular stretches of discourse and thus work as metacommunicative and focalizing devices. P. Cotte1 (2005) explains that focalization is meaningless and irrelevant if the totality or the whole context that the focused utterance belongs to is not kept in the focalizer’s memory. By definition, something is always focalized within and as part and parcel of a particular situation or a larger semantic unit. So reference to an explicitly focused utterance presupposes implicit reference to the whole semantic unit it is embedded in and I suggest that it is precisely what markers like well and you know are meant to indicate in discourse.2 As Östman puts it, “pragmatic particles IMPLICITLY anchor an utterance to a situation” (1981: 6, author’s emphasis). This has a series of important consequences as to the discourse scope and organizational function of both markers. . Organizing oral discourse: well and you know as antilinear operators In spontaneous conversations, well and you know often function as discourse frames, so that one could almost describe them as antilinear operators. This is implied in Östman’s own analysis of the discourse function of you know: “the speaker steps out of his propositional frame, and metacommunicates his attitudes and feelings” (1981: 16, my emphasis). Indeed, when producing one or several utterances, the speaker is often tempted to enter into details and theoretically runs the risk of losing sight of his/her initial meaning. Well and you know can be resorted to so that the speaker does not lose sight of his/her own meaning before or after diving into the linearity of language. The speaker can introduce or round off his/her turn with well and/or you know to signal to the hearer that one overall intention is at the origin of the diversity of the information data that are expressed in a necessarily linear order. Well and you know thus indicate that the speaker has one general semantic intention despite the apparent multiplicity of information and comments that s/he gives. Consider the following example: (1) A: I acquired an absolutely magnificent s\ewing machine by foul means did I t\ell you th/at# B: n\o# . This idea was communicated in Cotte’s research seminar of 14 May, 2005. See . . The title of this section refers to Gumperz’s notion of “contextualization cues”, which are defined as follows: “the means by which speakers signal and listeners interpret what the activity is, how semantic content is to be understood and how each sentence relates to what precedes or follows” (1982: 131). Contextualization cues can thus be seen as another term for discourse markers.

TSL[v.20020404] Prn:10/05/2007; 13:43

F: PB16107.tex / p.4 (106)

 Barbara Le Lan

A: well ^when I was doing freelance advertising [...]. (CEC, 1.3.1000).3

Then a “three-page” narrative ensues. Well, used as a narrative frame, composes 6.5% of all occurrences of well in the corpus analysed. In this example, well clearly opens a new space for the anecdote that has been announced in speaker A’s previous utterance and signals that every single utterance to come is part and parcel of the intended narrative. It is interesting to note that well, which functions as a metacommunicative frame, occurs just before a temporal subordinate clause, which functions as a referential frame. The following is an example of you know used in a similar way: speaker A’s general aim is to show the humorous contrast she observed between the cold and awe-inspiring atmosphere of the common room in a women’s college and the topics of conversation of the people in the room: (2) Everybody made their contrib\ution# from all over the senior c\ommon room# about their point of v\iew about eggs# they were. . .some would rather have them !m=uch too soft# than much too h=ard# and some people would !rather not have an egg at \ /all# and s\ome people# um.. thought the !th\ing to do# was just put them in the water and take them \out again# and th\en let them go on cooking# without cracking their h\eads# you kn/ow# you got every p\ossible# point of v\iew# about boiled eggs# (CEC, 1.3.48.8210).

The role of you know as an antilinear operator is clear here. The speaker “steps out” (Östman 1981: 16) of her various utterances and pragmatically subordinates all her individual utterances under the unifying marker you know. The proof of this pragmatic phenomenon lies in the utterance that follows you know, “you got every possible point of view about eggs”, which can be interpreted as the repetition in word form of you know’s unifying function in this turn. The expression “pragmatically subordinates” is chosen on purpose because it can be proved that the ability of you know and well to organize discourse and the interaction at large is due to their ‘original’ semantics and syntax. . Semantics and syntax of well and you know as non DMs Despite appearances, the discourse functions of well and you know can be at least partially accounted for when one refers to their explicit, referential meaning and . The capital letters (A, B..) indicate that these speakers have been surreptitiously recorded. “b, c”..signal that these speakers know that they are being recorded. The essential intonative data have been reproduced in the examples because utterances were felt to lose much of their meaning without them. “/” transcribes a Rise, “\” is a Fall, “/ \” is a Rise Fall, “\ /” a Fall Rise, “=” is a Level Tone, “#” marks the end of a tone unit, “!” indicates strong emphasis, “ *. . . . . ..*” indicate an overlapped or overlapping stretch of discourse. “(1.3.1000)” refers to the text number and the tone unit where the DM occurs in the CEC.

TSL[v.20020404] Prn:10/05/2007; 13:43

F: PB16107.tex / p.5 (107)

Orchestrating conversation 

syntactic identity as non DMs. Let us start with you know. To “know” someone or something is “to be familiar with or have experience and understanding of something or someone” (Cambridge Advanced Learner’s Dictionary 2003). So “know” means acquiring knowledge and experience of something (thus gaining cognitive control) as a result of one or several actual encounters with this thing or person (that is, after some kind of personal involvement). But you know as an ex main clause also has relevance here: as a transitive verb, “know” requires a direct object, i.e. a semantic unit, and we can reasonably think that the object of you know, used as a DM, is the theme of the speaker’s discourse; you know enables the speaker to encompass his verbal contribution to the exchange. And the ultimate goal of the speaker’s turn is. . . “you”, that is the hearer’s understanding of and sympathizing with what s/he says. This “you” is interesting in more than one respect, because in referring to the hearer in his/her own discourse, the speaker may very well mentally adopt the hearer’s exterior stance from his/her oral production in “stepping out” of it, which is another way of gaining a unified view of it. Well as an adjective or adverb means “according to one’s wish” (Cambridge Advanced Learner’s Dictionary). It refers to the notion of positive judgement, so well also signals the speaker’s personal involvement. Now the act of judging implies the existence and finiteness of the reality that is judged, i.e. a semantic unit, and also a minimal mental distance from that reality, that is, a kind of cognitive control, otherwise any judgement would be impossible. And as an adjective or adverb, well presupposes the existence of a clause or noun (i.e. a semantic unit) to which it is applied. Hence, we realize that the same ideas (semantic whole, personal involvement and cognitive control) are repeated at the semantic and at the syntactic levels. Thus, both well and you know have retained their semantic and syntactic past, which enables them to delineate discourse units, and also to acquire interpersonal functions in spontaneous conversations.

. Well and you know or how to contribute to the making of an interaction . Well and you know as heteroglossic markers4 Here, I build on Schiffrin’s notion of “interactive focus” (1987: 287), which she uses in her study of you know, in order to show how well and you know enable the speaker to go beyond the individual nature of his/her utterance and make it gain the status of a real contribution to the ongoing verbal exchange. Let us come . The notion of heteroglossia as relates to well is brought out in Aijmer and SimonVandenbergen (2003).

TSL[v.20020404] Prn:10/05/2007; 13:43

F: PB16107.tex / p.6 (108)

 Barbara Le Lan

back briefly to the meaning of well as a non DM and to the notion of “positive judgement”. By definition, any person exercising judgement keeps the thing being judged in his/her memory for at least a short period of time. This might account for the function of the DM well as a marker of anaphoric continuity: well is a very effective means of indicating that one is keeping in mind what has just been said. But well as a non DM, as we have seen in the previous section, is uttered by a speaker who mentally stands at a distance from the thing being judged. Therefore, well as a DM can also be interpreted as an effective means of stressing the otherness of one’s own contribution to the exchange. In other words, well is able to convey continuity and separation at the same time.5 In Aijmer’s and SimonVandenbergen’s words, the speaker signals his “awareness of the heterogeneity of views” (2003: 1155). You know is more explicit than well in its appeal to the hearer’s attention and sympathy, and achieves this in a different way. As an ex governing clause, you know signals that the speaker is taking a kind of bird’s eye view at a particular moment of his own oral production in order to submit a unified semantic whole to the addressee. . A case in point: well and you know with generic statements The co-occurrence of both markers with generic statements illustrates their dialogic nature. In the following example, A and B are sharing anecdotes to show how bizarre, inaccessible and cruel academics can be. B tells an anecdote and A comments: (3) B: and on one occ=asion#.. as I.. went out p\ast him# ..at the end of the l/ecture# an\other undergr/aduate# walked \ /over to him# and said something rather like !th\is# . . .excuse me s/ir# would you !clarify a !point in your !l\ecture# which I !couldn’t under!st\and# . . .n\o# it’s perfectly clear from the n\otes# !![pæng]# A: you !kn/ow# th\is is# what’s so \awful about academics# \isn’t it# this is the !w\orst side of them# (CEC, 1.6.123.11300)

Not only has speaker B’s contribution been taken into account in speaker A’s speech through the latter’s double use of “this”, but it has been recategorized and reinterpreted as a symptom of a more general tendency in the behaviour of academics; it is as though B’s turn were pragmatically subordinated to A’s generic point of view. I suggest that you know, together with the proform ‘this’ at the referential level, is meant to signal the speaker’s bird’s eye view and inclusion of B’s . To put it differently, well echoes the addressee’s words and announces the speaker’s words at the same time. See also Svartvik (1980: 177) for a similar approach of well as a marker of both separation and continuity.

TSL[v.20020404] Prn:10/05/2007; 13:43

F: PB16107.tex / p.7 (109)

Orchestrating conversation 

contribution into a cognitively larger frame. Well has the same ability of including the hearer’s previous remark into the speaker’s discourse and of recategorizing it as the symptom of a more general tendency, as example (4) shows: (4) A: people didn’t think you were going to be in today /actually# B: well they n\ever think I’m in today /actually# (CEC, 1.5.47.7080).

The dialogic role of well is particularly salient in B’s utterance: he almost literally includes A’s message in his own and recontextualizes it as the habit that most of his colleagues have of assuming that he is not in when they are. The co-occurrence of never with today has a humorous effect, which could be jeopardized if well were deleted from B’s utterance. In fact, the whole sentence would even have overtones of resentment as B would then reproach his colleagues for not paying attention to his presence at the workplace. With the same intonation as the original, the utterance without well is likely to be considered as B’s blunt correction of A’s meaning, and as a direct contradiction of A’s interpretation of the world: such an utterance would create the unpleasant impression that B ventriloquizes A’s words for ironic purposes. With well, the effect is almost the reverse: B does not contradict A; he just implies that his interlocutor had only partial access to the reality he was referring to, which did not only happen “today”, but “actually” happens all the time. In other words, B’s utterance in example (4) suggests that A does not know how right he is. Both well and you know in general signify this movement of inclusion of the interlocutor’s previously expressed point of view into the speaker’s own discourse, which is a highly cooperative way of expressing individual views. Well and you know are thus very often used to involve the hearer into one’s own speech, and in this respect they not only achieve discourse coherence: they also have the intriguing power of expressing personal comments and emotions, and do so in an indirect way, together with the intonation used. . Connecting beyond words: Implicit heteroglossia and expressivity Here, I shall develop the notion of personal involvement mentioned above in the semantic analyses of well and you know as non DMs, and Aijmer and SimonVandenbergen’s notion of “intersubjective positioning” which they use in their analysis of well. The frequent collocations of well and you know with personal judgements are worth studying in this respect: 12% of the wells analysed in the corpus occurred with personal evaluations, and the figure increases to 32% for you know. Consider the following example, where A relates the formal dinner ritual she went through in a women’s college: (5) um..the gh\astly th/ing# was the..!senior common-room convers\ation# that one had to conduct .. which was just !fr\ightful# and the. . .!\absolutely

TSL[v.20020404] Prn:10/05/2007; 13:43



F: PB16107.tex / p.8 (110)

Barbara Le Lan

grot\esque# r\ /itual# of high table formal !d\inner# in the \evening# which started with sh\erry# in the senior c\ommon room with the. . . !pr\esident# sitting there like G\od .. you kn/ow#. talking !absolute in\anity# on a very pro!found l\evel# (b and c laugh) (CEC, 1.3.32.5460).

In the wake of Foxtree and Schrok’s description of you know as a marker of “congenial atmosphere” (2002: 727), I suggest that A uses a kind of “conveying you know” that focuses on a hyperbole, which itself refers to the most telling and emblematic element of the whole description. What is also noteworthy in this example is the slight change of tone adopted by the speaker as soon as she utters the word “president”: she then lowers her voice, sounds dignified and self-important. The hearer of the recording has no doubt that A is actually mimicking the god-like face and body position of the aforesaid president. You know helps highlight the theatricality of this caricature: in appealing to the hearer’s faculty of projecting him/herself into this specific related scene, it indicates that speaker and hearer are communicating on a more personal level. It is indeed probable that you know is often used with hyperboles because, when using a hyperbole, the speaker seeks to share personal experience through individual and creative wording. Hyperboles in spontaneous conversations are thus both speaker-oriented (the speaker gives his/her personal version of the facts) and hearer-oriented (hyperboles are meant to trigger the addressee’s reaction or some form of response to the topic under discussion), just as you know is. Well can also emphasize the speaker’s personal involvement in what s/he is saying, as in example (6), where A is talking about a student who is in the seventh year of her PhD, and speaker c says that she hopes this student will find employment of some kind. A replies: (6) well !I would have th\ought# after seven y=ears# they ought to ch\ /uck her out in the world# and say go and do some t\eaching or something# you kn/ow#... (CEC, 1.3.17.2890)

In this kind of use, where well and you know indirectly convey the speaker’s personal involvement in what she is saying, these markers can often be described as intonation-enhancers. Together with intonation (which of course merits an indepth analysis which has not been undertaken here), these DMs are not only context-dependent, but they are also context-creating devices: they “overflow” the referential meaning of words, so to speak, in order to emphasize and amplify the emotional load of the message and convey it all to the interlocutor. Well and you know’s interpersonal role in conversation has so far been considered from a static point of view. If the analysis stopped here, it is probable that we would miss the inherent nature of well and you know as dynamic markers, as I will show in the last part of this study.

TSL[v.20020404] Prn:10/05/2007; 13:43

F: PB16107.tex / p.9 (111)

Orchestrating conversation

. Afference and efference: Well and you know as mental movements . Well, you know, and their host utterance: Hedging or amplifying Both markers modify the illocutionary force of the utterance they occur in by contextualizing it. As was seen in the first section, well and you know have the ability of giving pragmatic indications of how their host utterance interacts with its context of occurrence. Nikula (1996) chose to highlight this major aspect by calling discourse markers in general “pragmatic force modifiers”. In general, researchers tend to make a broad distinction between two categories of modifiers: hedges and intensifiers, the first being the most widely studied.6 When they are used as hedges, the very existence as well as the intonation contours of discourse markers create an extra stage in the process of utterance making and function like ‘buffer’ words between speaker and hearer. But DMs can also amplify the illocutionary force of the utterance, as I have shown in the previous section.7 So welland you know-utterances are generally either context-tamed or context-inflated and this is the first aspect of their dynamicity. But there is another important aspect, which might give the analyst further insight into the cognitive processes underlying the use of well and you know: both markers are also dynamic markers in the interpersonal field. . Well, you know, and the addressee: afference or efference Here, I bring out what is implied not only in Östman’s notion of “striv[ing]” (see the introduction), but in many other analyses where the dynamic aspect of well and you know has been pointed out. For example, Svartvik defines the general meaning of well in the following terms: “[well] signals that the speaker is going to shift ground, i.e. that he is going to modify one or more assumptions or expectations which have formed the basis of the discourse so far.” (1980: 177, my emphasis). According to Aijmer, you know signals “the switch from one deictic centre to the other” (2002: 228, my emphasis), and at the end of their study of you know, Erman and Kotsinas conclude that “[t]he overall function of discourse markers at the tex-

. Hedges enable the speaker to qualify the illocutionary force of his/her utterance by making things clearer or more obscure (Lakoff 1972). I do not dwell on the relationship between DMs and their host utterance, as this aspect has been extensively studied. See Fernandez (1994) and Hübler (1983) for an extensive bibliography on hedges. . See also Nikula’s section on “modifiying devices as signals of involvement” as well as the references she gives on this particular aspect (1996: 20), and Tannen (1992: 14–15) on the dialectic relation between involvement and independence in conversation.



TSL[v.20020404] Prn:10/05/2007; 13:43



F: PB16107.tex / p.10 (112)

Barbara Le Lan

tual level is to move the text forwards, thus functioning as moving markers” (Erman & Kotsinas 1993: 85, my emphasis). The dynamicity of well and you know can best be compared when they occur together in the same tone unit. In the CEC, there are 44 such co-occurrences. Note that the order of occurrence is always “well you know” when they are in the same tone unit, and never the reverse (?? “you know well”). In trying to account for such an absence of occurrence, I propose a comparative and partial descriptive analysis of well and you know: if the hearer is the actual landmark (i.e. the mental starting point) in the speaker’s choice of things to say and relate, then well in turn-initial position seems to signal a mental movement from the hearer’s to the speaker’s own thoughts, i.e. an efferent movement. You know seems to signal the opposite mental movement, that which goes from the speaker’s thoughts to the representation of the hearer’s thoughts, i.e. an afferent movement. It can be hypothesized that the opposite order of occurrence (“you know, well”) is not possible because a speaker can first create a sort of distance from his interlocutor with well in order to create mental space for speaking and then indirectly ask for the hearer’s point of view and/or sympathy with the use of you know (and a rising intonation), whereas s/he cannot first probe for potential common views with the interlocutor with a you know and then immediately afterwards distance him/herself from the latter with a well.8 The following example is taken from the beginning of an interaction where A explains that it was stupid of him to make himself a cup of tea shortly before leaving his office if he wanted to be here on time. The addressee seems astonished: (7) a: A: a: A: a: A:

you made yourself a cup of tea y\eah you mean before you came out ^w\ell# ^y\/ou know# ten ^m/\inutes ag/o# oh I see yes so ^that’s why I’m about five minutes !l\/ate you see#

This example deserves special attention for it is teeming with implicit meaning whose presence is only palpable through the use of well and you know. Speaker ‘a’ is surprised at first, and even laughs when he utters the third line of the example because he understands that A had just made himself a cup of tea and left the office without drinking it. A realizes that he did not convey the idea he wanted to express and mentally leaves the field of presuppositions just delineated by speaker ‘a’ with the help of well in order to build a new field of presuppositions. And you know contributes to the delineation of such a field by immediately marking the opposite . . . . unless the speaker takes a distance from himself, hence the need for long pauses (to signal self-correction or hesitation) to make the sequence “you know well’ acceptable but even there, you know and well are not part of the same tone unit.

TSL[v.20020404] Prn:10/05/2007; 13:43

F: PB16107.tex / p.11 (113)

Orchestrating conversation

mental movement, from speaker to hearer this time, which enables A to point to the implicit, shared knowledge that when someone “makes him/herself ” a cup of tea, they generally drink it afterwards, hence his lateness. Note how the existence of this implicit general statement, which is meant to rebuild a shared mental space, is inferred from the sequence “ten minutes ago” only through the use of you know. The existence of such a general statement on the implicit side of meaning is due to the metalinguistic nature of you know, which points to how speaker A defines the expression “made myself a cup of tea” and what he actually meant by using it. You know also shows that A’s main concern is addressee-oriented: he wants to go on in the shared field of knowledge and reference he thought he and his addressee had never left. The utterance of well is the moment when A suddenly realizes the existence and width of the gap that separates him from his addressee (the efferent moment), and the utterance of you know mentally builds the bridge that enables A to interact in a shared field again (the afferent moment), and the order of the two markers cannot be reversed: the speaker cannot bridge a gap before digging it. The pause just before well and the strong accent on both DMs make the double mental movement particularly salient and palpable. It is in this kind of linguistic manipulation that the true nature of well and you know seems to be the most salient: they verbalize mental movements (sometimes mental meanderings) within the process of spontaneous utterance production, hence the use of “well you know” in transitions and hesitating moments. But when each marker is taken separately, one of the most cogent arguments in favour of this explanatory hypothesis is the fact that one of you know’s most frequent contexts of occurrence is that of justification of one’s own position (about 21% of the cases analysed in the CEC) and one of well’s most frequent collocations is with utterances that bring out the consequences or implications of a previous utterance (15%). Let us briefly consider you know first. In the following example, A is telling B about an academic meeting where the teachers present tried to understand the various professional titles that were on a candidate’s CV. A is now commenting on one of these titles: (8) A: this this this !str\uck me# as a kind of / \odd !t\itle# you s/ee#...uh. B: *\m# . . .\m#*. A: ‘recognized teacher *in applied lingu\istics’#* uh. . . you kn/ow# with ‘applied’ in br/ackets# (CEC, 1.2a.6.9710).

You know is right in the middle of a clarification and a factual justification of the speaker’s personal opinion on a specific matter. It points to the particular detail that gave rise to A’s suspicion about the actual proficiency of the person who submitted this CV. Together with you see in this case, you know signals the speaker’s intention of sharing his personal conclusions with the hearer and thus operates



TSL[v.20020404] Prn:10/05/2007; 13:43



F: PB16107.tex / p.12 (114)

Barbara Le Lan

an afferent movement. The afferent movement is confirmed and reinforced by the laugh A gives when uttering this you know. How can we account for the frequent co-occurrence of you know with justifications? One could suggest that one of the best ways to share personal views and judgments with others is to justify them. When I give reasons for what I think or feel, I make my stance more legitimate and more acceptable to the addressee, and the individuality of my views is immediately less salient. You know thus conveys the speaker’s wish to erase or to qualify the personal nature of his utterance in order to make it gain the status of a verbal contribution to and component of the ongoing exchange. I now focus on the co-occurrence of well with utterances that bring out the consequences or implications of what has just been said. In example (9), A and B are teachers and they have just started a new conversational topic: (9) B: \I had a seminar tod/ay# in which.. people !hadn’t read the !st\uff# because of !s\essionals# hadn’t !read the !pl\ay# so we had to !spin it \out# it’s always a bit /awkward# A: ..^well I I uh I m\ean# if they haven’t r\ead it# you can’t !really !h\ave a seminar# (CEC, 1.4.71.10870).

When well is used in these contexts, the speaker builds on the interlocutor’s just completed turn and verbalizes what is implied in the latter’s words. In this example, well, together with I mean, the probably deliberate hesitations following well, and the if -clause, materializes the passage and the logical continuity from the interlocutor’s to the speaker’s own point of view, and thus operates an efferent movement, which is also signified through the short pause preceding this well. This pause is the silent counterpart of the small mental gap dug by well. The interpretation of well and you know as mental movements may account for their impressive multifunctionality, which goes as far as being used for radically “different, sometimes incompatible, purposes. In one context, well may express enthusiasm and in another context it means reluctance. [. . .] It may express agreement or acceptance but also disagreement if the context is one of verbal conflict” (Aijmer & Simon-Vandenbergen 2003: 1129). Thus, well and you know are multifunctional because they enable speakers to start or end their turn with whatever stance was previously expressed by the hearer. The description of well and you know as signals of mental movements might also explain why they are metacommunicative rather than strictly metalinguistic (*this is well and *you know what I say are incorrect glosses of well and you know). The description of well and you know as strictly metalinguistic words betrays a static approach to DMs: these markers do not merely refer to the linguistic sequence within their scope: well and you know point to a process, i.e. the utterance in the making and to its result at the same time, that is, when uttering well and/or you know, the speaker signifies his personal involvement in and cognitive control of what he says at the same time.

TSL[v.20020404] Prn:10/05/2007; 13:43

F: PB16107.tex / p.13 (115)

Orchestrating conversation

. Conclusion The present study has attempted to show that the ‘mystery’ of well and you know can partially be unveiled by the semantic analysis of their non DM equivalents. More particularly, the semantic components of “memory” or “cognitive control” (i.e. ‘keeping something in one’s mind’, ‘being familiar with something’) as well as “personal involvement” or “subjectivity” (positive judgment with well, and the “you” in you know, which is necessarily uttered by an “I”) proved decisive in accounting for the connective and interpersonal roles of well and you know in many of their uses. Indeed, when one combines the two semantic traits of “memory” and “subjectivity”, one realizes that well and you know make the subjective truly intersubjective in conversation. The description of these DMs as antilinear operators showed that they were both useful in helping the speaker to constitute a semantic whole in spite of the necessary linearity of language: well and you know act as boundaries between the said and the implied or the yet unsaid, and between showing and telling when we focus on their interactional functions. In this respect, they team up with intonation in conversation, which is also of an indexical nature, in that it shows more often than it tells. I further investigated the overarching role of well and you know by describing them as the signals of mental movements from the speaker’s thoughts to a representation of the hearer’s thoughts (with you know), or the reverse (with well), which seems to reintroduce some kind of linearity into the picture. But this apparent paradox is in fact another way of seeing the ability that both these markers have of signalling personal involvement and cognitive control at the same time. Indeed, as indexicals, they can refer at the same time to the person showing, the object shown and the addressee of the whole process of showing, in other words: speaker, discourse and hearer. The musical metaphor in the title of this article was meant to capture the elusive nature of the organizational and interactional work that is actually performed by well and you know in authentic spontaneous conversations. Although the present research is still in progress, I hope to have shown that well and you know are indeed the self-effacing conductors of the complex polyphony that is performed when speakers interact. As indexicals, both DMs function as regulators and organizers of the fundamental triad (speaker, hearer, discourse) and therefore, it is no surprise that both markers should occur at key moments in the ongoing verbal interaction, as indicated, on the one hand, by the intonation as well as the short stare (and sometimes the short pause) that often accompany you know, and as indicated on the other hand by the existence of other well-known expressions such as welcome and . . . farewell.



TSL[v.20020404] Prn:10/05/2007; 13:43



F: PB16107.tex / p.14 (116)

Barbara Le Lan

References Aijmer, K. 2002. English Discourse Particles: Evidence From a Corpus. Amsterdam: John Benjamins. Aijmer, K. and Simon-Vandenbergen, A. M. 2003. “The Discourse Particle Well and Its Equivalents in Swedish and Dutch.” Linguistics 41 (6): 1125–1161. Cambridge Advanced Learner’s Dictionary. 2003. Cambridge: Cambridge University Press. Cotte, P. 1992. “Réflexions sur la linéarité.” Travaux du CIEREC 76: 53–76. Cotte, P. (2004–2007). Séminaires de recherche en linguistique anglaise du Professeur Pierre Cotte (cours de Master 2) Erman, B. 2001. “Pragmatic Markers Revisited with a Focus on You Know in Adult and Adolescent Talk.” Journal of Pragmatics 33: 1337–1359. Erman, B. and Kotsinas, U. B. 1993. “Pragmaticalization: the Case of Ba’ and You Know”. Studier I Modern Språkvetenskap 10: 76–93. Fernandez, M. M. J. 1994. Les Particules énonciatives. Paris: Presses Universitaires de France. Foxtree, J. E. and Schrock, J. C. 2002. “Basic Meanings of You Know and I Mean.” Journal of Pragmatics 34 (6): 727–747. Gumperz, J. J. 1982. Discourse Strategies. Cambridge: Cambridge University Press. Hübler, A. 1983. Understatements and Hedges in English. Amsterdam: John Benjamins. Lakoff, G. 1972. “Hedges : a Study in Meaning Criteria and the Logic of Fuzzy Concepts.” Chicago Linguistic Society 8: 183–228. Longacre, R. 1976. “ ‘Mystery’ Particles and Affixes.” In Papers from the Twelfth Regional Meeting of the Chicago Linguistics Society, S. S. Mufwene et al. (eds.), 468–475. Chicago: Chicago Linguistics Society. Nikula, T. 1996. Pragmatic Force Modifiers: a Study in Interlanguage Pragmatics. Academic Dissertation, Jyväskylä: University of Jyväskylä. Östman, J. O. 1981. You Know: a Discourse Functional View. Amsterdam: John Benjamins. Östman, J. O. 1995. “Pragmatic Particles Twenty Years After.” In Organization in Discourse. Proceedings from the Turku Conference, B. WÅRVIK et al. (eds.), 95–108. Turku: University of Turku. Quirk, R. and Svartvik, J. (eds.). 1980. A Corpus of English Conversation. Lund: Lund Studies in English. Schiffrin, D. 1987. Discourse Markers. Cambridge: Cambridge University Press. Schourup, L. 1999. “Discourse Markers.” Lingua 107: 227–265. Schourup, L. 2001. “Rethinking Well.” Journal of Pragmatics 33: 1025–1060. Svartvik, J. 1980. “Well in Conversation.” In Studies in Linguistics for Randolph Quirk, S. Greebaum, et al. (eds.), 167–177. London: Longman. Tannen, D. 1992. That’s Not What I Meant! London: Virago Press.

TSL[v.20020404] Prn:10/05/2007; 9:59

F: PB16108.tex / p.1 (117)

A because B so A’ Circularity and discourse progression in conversational English Frédérique Passot University of Paris 3 – Sorbonne Nouvelle, France

This paper is concerned with an apparently circular pattern in naturally occurring conversational English involving because and so. We analyze the pattern, schematized as A because B so A’, with respect to discourse progression and the construction of the symbolic space in which views are exchanged and negotiated. Semantically weakened compared to their written counterparts, the markers readily depart from their traditional logical roles to adopt a behavior closer to that of discourse markers. Through the analysis of the role of the connectives and of the relationship between the three segments of the schema, we show that this pattern, far from being a sterile infinite loop, is indeed a fertile spiral that brings essential dynamism into conversational speech.

.

Introduction

This paper is concerned with the role of the markers because/cos1 and so in spontaneous conversation in standard British English. More specifically, it focuses on an apparently circular pattern involving both markers, which can be described under the basic form A because B so A’. We wish to show that this pattern, far from being a sterile infinite loop, is indeed a fertile spiral that brings essential dynamism into conversational speech. Our study is based on the analysis of an original corpus of naturally occurring one-to-one conversations. Four pairs of native speakers of standard British English (aged 20 to 50) were recorded using separate high quality recorders and clip-on microphones. All four recordings were made in a setting familiar to the speakers, who were left on their own and had control over the subjects they talked about . Because and cos will be treated as one marker here, although it has been shown that they may behave significantly differently. See Passot (2006).

TSL[v.20020404] Prn:10/05/2007; 9:59



F: PB16108.tex / p.2 (118)

Frédérique Passot

and the duration of the recording. 30,000 words were collected over 4 hours of recording, and minutely transcribed.2 Examples are also drawn from the spoken part of the British National Corpus (henceforth BNC) and from Biber et al.’s (1999) grammar (LGSWE). After describing the various configurations in which the pattern can be found, we analyze the relationship between the three segments, and in particular between A and B, and between A and A’. Semantically weakened compared to their written counterparts, because and so readily depart from the logical role ascribed to them by traditional grammar and adopt a behavior that is closer to that of discourse markers. What type of link do they establish between the segments they introduce? We eventually propose a hypothesis as to why the argumentative spiral should be viewed as an essential dynamic pattern with respect to discourse progression and the construction of the symbolic space in which views are exchanged and negotiated.

. Various spiral configurations The structures featuring both because and so that we wish to discuss – which we will tentatively call spirals – occur in various configurations. The prototypical configuration can be analyzed as a three-fold schema. However, our corpus provides us with a number of examples displaying a binary pattern. . A ternary pattern .. A because B so A’ The basic and most common configuration is found in example (1), in which James inquires whether Celia needs John’s approval before issuing him with the keys to the laboratory John runs:

. Our transcription conventions are the following: – {duration in sec.}: silent pause – : any audible perturbation in the speech signal described in the remark, e.g. (duration): an audible in-taking of breath – *segment1* *segment2*: overlapping speech segments – phoneme::: phoneme lengthening – phoneme-: interrupted utterance

TSL[v.20020404] Prn:10/05/2007; 9:59

F: PB16108.tex / p.3 (119)

A because B so A’

(1) Celia&James: John’s permission, because_020 J *do you technically* need John’s sort of permission *then* C *no* J to get hm no C no J (0.35) ok (0.55) C no because:: I mean {0.50} I know he wants you to have keys so that’s fine J does he {0.80} did he tell you that ** C *well if you’re doing it* after hours you have to

James’s question receives a three-fold answer. Each of the three points made by Celia aims at assuring James that she does not need John’s permission: the negation, uttered several times, the because clause and the so clause. It appears that the explanation conveyed by the content of the because clause is relevant to both the negation on its left and the so clause on its right. The two segments are indeed highly redundant in terms of their informational import. This example shows a circular pattern in three parts: Celia answers James’s question, then explains her answer with a causal clause in because that she then uses as a support for an anaphoric so clause that takes up again her answer to James. As we will see in Section 3, cases where the because clause precedes its matrix are extremely rare in conversational English. Interestingly enough, though, they can also feature the spiral pattern under discussion: (2) spoken BNC, text=“HYR”, n=“52” PS3KF notice that in this model because we’ve logged both dependent and the independent variables, right, the coefficients that we estimate are elasticities, right, so we can read those coefficients off directly as elasticities

Here, the spiral is flattened by the linear order of the segments. The schema is the following: – – –

because B (we’ve logged both the dependent and the independent variables) A (the coefficients that we estimate are elasticities) so A’ (we can read those coefficients off directly as elasticities).

.. A because B so The looping schema is also visible in examples where the A because B sequence is followed by so only: all the elements being given to reconstruct A’, there is no need to utter it. In example (3), Helen is told that she has missed a phone call from someone who Rebecca thinks may be a potential employer:



TSL[v.20020404] Prn:10/05/2007; 9:59

F: PB16108.tex / p.4 (120)

 Frédérique Passot

(3) Helen&Rebecca: RFI, because_024 R I think what happened was she tried to phone you on your mobile telephone but if i::t rang twice H right R and cos it is as easy as on a computer *I could find out she had left a message* H *actually probably it wasn’t her because* I didn’t *give them my f- my phone number so* R *oh ok probably she left* a message on the H yeah

So invites Rebecca to reconstruct clause A for herself from the elements provided by Helen in her because clause. Rebecca’s next cue is indicative of the steps she takes into the argumentative process: the exclamation (oh), the assertion marker (ok) and the epistemic adverb (probably) show that she is gradually embracing Helen’s standpoint and making it hers. The beginning of the loop is enough to suggest it entirely and to inscribe the segments of this example within a ternary system. . Binary variants Our corpus and the spoken BNC provide variants of the ternary system that either consist of two or four segments. .. A because A’ In the following example, this ternary system appears in a synthesized form in which the middle stage is no longer expressed. (4) Celia&James: Diphthongs2, because_018 C *I think* it’s easier actually if you’re putting the vowel at the beginning J oh it is it is an easier *thing but* C *because it’s* m- much easier to say J mm but the the experiment I’ve based this on was based in in fixes which were er vowel consonant C mm

The informational import of the subordinate is extremely low. Detached from its matrix, the clause tautologically reasserts the content expressed in the matrix: it is easier to say because it is easier to say.

TSL[v.20020404] Prn:10/05/2007; 9:59

F: PB16108.tex / p.5 (121)

A because B so A’

.. A because B C so A’ This four-fold configuration is at the fringe of the phenomenon that caught our attention. Example (5) illustrates a more complex pattern, spread out over two speakers’ turns, in which segment C is inserted in the middle of the loop. (5) spoken BNC, text=“JA9” n=“262” PS41Y Mm. Well certainly, you know, I think there there is work looking at the files that go through me, there is work for a second complaint examiner, because a lot of the files from investigators are actually fairly straightforward now PS420 Yeah. PS421 mm, I know Sarah’s not dealing with them, so you know I’m sure that there would be enough work for a complaint examiner, but whether that would then mean investigators were light is is really PS420 Well there is there is an argument that says if P As do the twenty six five work, or the bulk of it, that Sarah might be one team complaint examiner.

Not only is the spiral realized over two different speakers’ turns, but it seems to be inscribed within a broader pattern. The rephrasing of A (there is work for a second complaint examiner), A’ (so you know I’m sure that there would be enough work for a complaint examiner), is used concessively and is followed by a fifth segment, D (but whether that would then mean investigators were light is is really), introduced by but. The duplication of the content of A in a so clause that follows a because clause, though relatively rare in our corpus, is far from being anecdotal. The intuition that it is emblematic of a larger tendency in spoken English is supported by a number of examples that do not necessarily follow the rules of the prototypical configuration of the spiral configuration described above. To explore that point, we wish to analyze the relationship between the three segments of the spiral, and especially A and B, and A and A’.

. A because B: A dissolving causal link To explain the relationship between A and B in the spiral structures, let us focus on determining how causal the context in which because clauses are used is. The grammar of English allows a because clause to appear on either side of the matrix or even in medial position. In practice, the order seems to be restricted to the A because B pattern. The preferred final position is indicative of a certain type of relationship between the two clauses. Arguably, it allows for a greater autonomy of the because clause.



TSL[v.20020404] Prn:10/05/2007; 9:59

F: PB16108.tex / p.6 (122)

 Frédérique Passot

. Narrow scope: A tight causal link Narrow scope signals close ties between the clauses. The subordinate brings an explanation for the process described in the main clause. Studies on causal clauses traditionally call upon specific syntactic criteria used to determine the scope of a subordinate clause. Among them are clause order, the polemic focalization of the subordinate clause (it clefting, negation or contrastive interrogation), why questioning, the premodification of the conjunction (e.g. only because), and its occurrence within a prepositional phrase (e.g. because of ).3 An extensive manipulation test based on these syntactic criteria cannot be successfully carried out on examples from a spoken corpus, due to the heavy influence of other parameters on interpretation, among which disfluencies or intonation, and non standard syntax. However, the low number of examples for which narrow scope is already encoded in the syntax can be viewed as a frequency indicator. As far as clause order is concerned, the matrix clause almost invariably precedes the subordinate. The reversed order, indicative of a narrow scope relationship, is marginal both in our corpus (4%) and in the LGSWE corpus (5%). So are all the other syntactic forms under scrutiny. It is particularly striking to observe that causal subordination is not correlated with the occurrence of why questions. On the whole, utterances with an obvious narrow scope are rather rare in our data, making it difficult to draw any conclusions. We have to turn to the broad scope category of causal clauses to learn more about the context in which because and cos clauses appear. . Broad scope and autonomy Broad scope clauses constitute metalinguistic comments on the speech act itself. In this sense, they justify the very utterance of the main clause. Our focus here has been to examine the degree of independence broad scope clauses enjoy with regard to the matrix. They are generally said to differ from narrow scope ones prosodically, and specifically to be characterized by a so-called “comma intonation”, in which pauses play a major role.4 In our data, pauses not only signal broad scope but more importantly tend to question the very link between clauses. The clauses are all the more detached from one another as a silent pause generally gives the interlocutor the opportunity to intervene briefly between the matrix and the subordinate. It is the case in . For a detailed account of the appropriate tests, see for example Rutherford, W. (1970). . See Rutherford, W. (1970).

TSL[v.20020404] Prn:10/05/2007; 9:59

F: PB16108.tex / p.7 (123)

A because B so A’ 

example (6). At one point in the conversation, Celia needs to readjust her clip-on microphone. James asks her to make sure that her recorder is still recording. After receiving a positive response, he goes on to explain why he is so worried about that, recounting the problems he had the day before. (6) Celia&James: Recording, cos_020 C that looks as if that microphone’s {0.70} s::lightly *at a funny angle* J *w-ye* {0.50} probably then you can see your recorder there is there a C yes it is moving J C er (0.55) J {0.50} cos when I forgot to turn the tape recorder on yesterday I’d still been adjusting recording levels all the time it’s just the tape wasn’t running

More than 3 seconds elapse between James’s matrix and the metalinguistic cos clause he appends. In this gap, Celia reassures him, laughs, utters a hesitation marker, breathes in audibly, while James laughs, breathes in, and pauses for another half second. The following example brings further support to our claim. Helen and Rebecca share their impressions about the type of people who attend Westminster school, where Frank has studied. (7) Helen&Rebecca: Westminster, cos_039 H *I think I kn-* I should have asked him if he knew some people I he probably hated them {} would have hated them R I think they’re pretty quite I don’t know really H no I’m just thinking like the people I know who went to Westminster R yeah H who I know from Oxford are really {} nasty R okay H wawa *public school* R *cos I’ve met* some of his friends and they’re really:: {} H *nice* R *laid-* yeah *quite nice people*

Can this type of relationship still be considered as syntactic subordination when the matrix and the subordinate clause are not only separated by extremely long silent pauses but also by whole independent utterances as in example (7) above? These views are supported by the work of a number of researchers on spoken language. In her analysis of this phenomenon, Couper-Kuhlen’s (1996) distinction between two different configurations involving because takes into account

TSL[v.20020404] Prn:10/05/2007; 9:59

F: PB16108.tex / p.8 (124)

 Frédérique Passot

informational, pragmatic and prosodic parameters. The first configuration (CSI) corresponds to the narrow scope category, where the two clauses form one predication unit in which the matrix is the center. In the second configuration (CSII), each clause works as an information unit and can be a turn-constructional unit. On the informational and pragmatic levels, both clauses can trigger markers of acknowledgement or longer reactions produced by the interlocutor. On the prosodic level, the causal clause is marked by a reset in the declination line, a sign of functional autonomy. Debaisieux (forth.), analyzing very similar phenomena in spoken French, goes so far as to revoke the “subordination” label applied to broad scope relationships and proposes to reserve it for cases where a narrow link is attested. In many cases, telling whether the scope of the subordinate is narrow or broad is a difficult task. Following Vallée (2003), who observes that a metalinguistic clause can also have some narrower impact on the matrix, it may seem wise to acknowledge that the two categories are not necessarily mutually exclusive. Indeed, the two interpretations can be viewed as the attracting poles on a continuum.5 However, Vallée’s (2003) concluding claim that because is always causal is not supported by data from spoken English. Very often used outside of a strictly causal context and in growingly autonomous clauses, because tends to slide away from causality and take on different roles. In this context where strict causal subordination is rare, it may seem strange for because and so to appear jointly in structures that may at first sight be reminiscent of strongly marked argumentation using narrow scope subordination. Additionally, with the duplication of one of its terms, the argumentative pattern resembles a logical loop. What is the relationship between A and A’?

. A and so A’: A qualitative shift The relationship between A and because B is not always causal. Likewise, there is no necessary logical relationship between because B and so A’. Contrary to because B, so A’ necessarily occurs after A and should be analyzed in parallel to the clause that it is echoing. The relationship between A and A’ is not one of identity, thereby ensuring that the A because B so A’ pattern is not circular. Indeed, a qualitative shift is observed in our examples between A and A’. This shift takes several aspects that we would like to discuss.

. The same phenomenon is attested to in spoken French with parce que, as discussed by Simon & Grobet (2002: 649).

TSL[v.20020404] Prn:10/05/2007; 9:59

F: PB16108.tex / p.9 (125)

A because B so A’

. A semantic or referential shift Example (8) below presents a fairly circular aspect. Looking at A and A’ shows that the two ends of the pattern do not meet, and that there is enough difference between them to support the spiral hypothesis. (8) spoken BNC, text=“KP8” n=“3457” PS52U It’s, what they’ve done is they’ve closed er, er, erm <pause> branch, an office <pause> massive office in London and they’ve moved them all over to Leeds because they can’t get the people to <pause> take jobs in London <pause> so they’ve transferred it all to Leeds

In this example, the loss of specificity in the reference caused by the change from branch, office and them to it is counterbalanced by a gain in precision in the terminology used. The verbs used in the first part of the structure (close and move) are synthesized in one verb in A’: transfer is the appropriate term to describe both processes of closing and moving offices. In a lot of cases, A’ wraps up the information contained in A in a more specific and effective way. The shift brought by A’ with respect to A seems to be a natural side effect of the semantic import of so. Used as a resultive or an inferential marker, or in a purpose clause, so marks a transition between one level of symbolic representations to another. In this sense, A’ segments are characterized by a change of perspective from that expressed in the presupposition it is built on (here, the content of B), but also from the preceding segments (including A). Interestingly, this shift can be observed both in examples where A’ introduces a generalization, as above, or an illustration of the statement in A. The following example features two occurrences of so, the second one only being inscribed in the spiral pattern under scrutiny. Both markers, however, signal a move from more generic statements to more specific ones: (9) spoken BNC, text=“FM7” n=“21” PS1SF Tony’s come here and he’s gonna tape recorder everything that’s says, that is said, your voice, my voice, so we want it all nice and properly done because er they’re gonna make a special project our of it, reference to spoken word and the written word so you must ask your questions nice and clearly yeah?

Not only is the content of A’ semantically more precise (the verb ask and its object your questions disambiguate all and done), but the scope of the reference of the subject is also narrowed down from an all-inclusive we to a very specific you.



TSL[v.20020404] Prn:10/05/2007; 9:59

F: PB16108.tex / p.10 (126)

 Frédérique Passot

. A modal shift The most common type of shift between A and A’ is a modal one. A modal shift is one that betrays the symbolic stance taken by the speaker on the matter under discussion and the speaker’s anticipation of the addressee’s own posture. Grammatically, it is to be observed in the use of modal auxiliaries, reporting verbs, or intensive adverbs among others. In our examples, traces of modalization can be on either or on both sides of the marker so. In the following example, the speaker uses reported speech in A while resorting to unmediated assertion in A’, thereby suggesting a shift in perspective: (10) spoken BNC, text=“H49” n=“968” PS1XG Erm <pause> they say they would not be willing to change their valuation, because that was the valuation er come to by the District Value Office from , so they are not willing to, to come down in price.

In example (9) quoted above, the speaker first expresses a mild suggestion that is rephrased in A’ as an injunction for the addressee to watch their English while the tape recorder is on. In example (5) above, the modality of the assertion evolves from belief to certainty. I think and I’m sure, however, both prefix subordinate clauses in which the mood of the main verb counterbalances the modality expressed in the matrix: the simple present of assertion contrasts with the conditional of epistemicity. In all cases, the rephrasing provides a different perspective on what is uttered and leaves space for linguistic and metalinguistic commentary. The quick analysis of the relationship between A and A’ has shown that A’ is not to be viewed as the sterile repetition of A. Marked by a qualitative shift on the semantic, referential and modal levels, the two segments are not interchangeable. The pattern under scrutiny can therefore be considered as a complex structure with an apparent duplication of actually qualitatively different elements.

. Similarities with hypotactic and paratactic structures The spiral that caught our attention when looking at because clauses bears resemblance to phenomena that have prompted interesting hypotheses concerning syntactic and pragmatic organization. The first hypothesis we will consider relies on hypotactic structures, while the second one is based on linear ordering. To what extent can these hypotheses be applied to our spiral patterns?

TSL[v.20020404] Prn:10/05/2007; 9:59

F: PB16108.tex / p.11 (127)

A because B so A’ 

. A hypotactic hypothesis The syntactic explanation that we wish to examine follows De Cola-Sekali’s (1991) claim that, in because utterances, because q is never really fronted. Her analysis of the because q p sequence questions traditional views on this rarely found form in conversational speech. Can this model be used to account for the various configurations from Section 1? .. Because q, p revisited Based on the relative position of the target clause (p) and the reference clause (q), De Cola-Sekali analyzes the clauses as inscribed within a three-fold relation. The relationship between the target (p) and the reference (q) used to reach the target implies a fixed order in the segments, the target being necessarily located before the reference. Therefore, p in the canonical order (in initial position) and p in the reverse order do not have the same status. When because q is in initial position, p should be regarded as a rephrasing of a prior clause (p1) that can be retrieved or inferred from the preceding context. Hence the following pattern, put forward by De Cola-Sekali (1991: 339), in which the because clause stands between the target and the rephrased target: 1. cible d’un débat → 2. relation à un repère → 3. réassertion6

.. Application of this model to our examples Do the similarities of the phenomenon found in our corpus with this thoughtprovoking model allow us to view our so clause as being De Cola’s postposed clause p? If so, the rarity of because fronting in conversational English would be if not accounted for at least counterbalanced by the existence of an alternative structure in which all three parts of the targeting and referencing process are expressed. The existence of the configuration because B A so A’ suggests limitations of the application of De Cola’s hypothesis to our examples from Section 2. In the following example, two sequences appear jointly, which are normally incompatible if they belong to the same paradigm: a fronted because clause and a sequence where p is followed by a so clause. (11) Helen&Rebecca: Staying awake till 7, because_028 R how much actually I’ve got hm about fifteen francs if you want mmoney for drink or something H actually I’ve got my water that’s fine R you can use that with *pleasure* . 1. target of the debate → 2. link to a reference → 3. reassertion.

TSL[v.20020404] Prn:10/05/2007; 9:59

F: PB16108.tex / p.12 (128)

 Frédérique Passot

H R H R

*no it’s alright* I got water *hm* *hm* and then because Sarah {0.20} didn’t have to teach now she’s not gonna come in so {0.70} I’ll be all alone how come she doesn’t have to teach and you do

In this extract, Helen lists her day’s worries: her next class does not start before 7pm, she feels very tired, she forgot her purse at home and she will have to spend several hours alone in the staff room. Let us focus on her last cue. The ternary pattern is present, but it is difficult to find a target to the left of the reference constituted by the because clause. Indeed, it does not seem possible to restore a p1 clause from the preceding context which p would rephrase. The clause that rephrases p1 is located further down: the so clause expresses a generalization (I’ll be all alone) with regard to p1 (she’s not gonna come in). Should we look at the broader context, and consider that Helen’s exclamation (oh I don’t know if I can stay awake here till seven that’s awful) from a few seconds before is indeed p1, the target of the connection? We do not claim to be able to bring a firm answer to this question. From a pragmatic point of view, what is at stake in our example is expressed in the so clause. Followed by a long pause (70 centiseconds), the marker unifies the preceding discourse units and prefixes an additional relevant unit. . A paratactic explanation In our opinion, the phenomenon under scrutiny is best accounted for outside the framework of syntactic analysis, namely within that of discourse analysis. Though it uses the markers of connection in our examples, it shares a number of features with paratactic structures. Morgenstern & De Cola-Sekali’s (forth.) analysis of the acquisition of argumentation in French reveals a looping pattern in four parts. One-year-ten-month-old Léonard (L) is drawing with his mother (M):7 . Morgenstern, Aliyah & Sekali, Martine (forth.: 6–7). Emphasis is ours. M: L: M: L: M: L:

Look, we’re going to draw a red circle She draws one. He cries and he shakes his head. exclamation + negation No? You don’t want that? Look, it’s beautiful. a red one no He hits the drawing. It’s not beautiful/good/It’s ugly? So let’s draw another one. hits the sheet of paper. L: exclamation + negation + a red one is not beautiful/good/is ugly

TSL[v.20020404] Prn:10/05/2007; 9:59

F: PB16108.tex / p.13 (129)

A because B so A’ 

M: L: M: L: M: L: L:

Regarde, on va faire un rond rouge. Elle le fait. Il crie et il secoue la tête. anonanon / anonanonanon // Non, tu veux pas ? Regarde, c’est joli. arouz / non : : : // Il tape sur le dessin. C’est pas beau ? Alors on fait un autre. tape sur la feuille. aa : / nonarouzarouzepabo //

Let us consider two sequences uttered by the child: his first two utterances (sequence 1: anonanon / anonanonanon // arouz / non : : : //) and his last one (sequence 2: nonarouzarouzepabo //). In the first sequence, Léonard builds an argumentative loop that is closely related to the ones from our corpus combining because and so. In the absence of connectives, the explicative value of arouz (un rouge, a red one) is conveyed by the order of the segments and the repetition of non (no). This first loop in A B A’ is followed by a configuration in which both members of the argumentative structure are repeated. The second sequence can be represented as A B B A’, where A’ rephrases A. The two authors analyze epabo (est pas beau, is not beautiful/good, is ugly) as a conclusive reformulation of the child’s refusal to draw a red circle which is epitomized in the opening non. A and A’ frame the duplicated segment B (arouz, red) and give it an explicative value. The child’s argumentative endeavor is underlined by the duplication of at least one of the terms of the explanation and the organization of the segments in an alternating order. This strategy allows the child to signal and bridge the gap between what he thinks and what he thinks his mother thinks. According to the authors, the pattern itself, characterized by the linear ordering assertion – explanation – reassertion, points to the existence of the argumentation. The adult speakers of our corpus have undoubtedly acquired the use of the argumentative markers of English. However, their speech retains some aspect of child argumentation. The use of the appropriate markers allows our speakers not to duplicate segments in their entirety after so in a because/so sequence. The presence of so is in itself enough to suggest the spiral pattern. The possible combination of the two modes of organization shows how redundant certain aspects of spoken English can be. The similarities with other structures suggest that the spiral pattern is not anecdotal, despite the relatively restricted number of examples. Not only is it a frequent pattern, which does not necessarily involve because and so or indeed any connectives at all, but it is also essential to discourse organization.

TSL[v.20020404] Prn:10/05/2007; 9:59

F: PB16108.tex / p.14 (130)

 Frédérique Passot

. Recursive dynamism and the argumentative spiral As seen above, the echoing segments within the spiral are not redundant. The qualitative shift observed between A and A’ and their loose relationship with B point to the dynamic character of the pattern. The role of the pattern as a dynamic principle in the organization of discourse is, in our examples, at least partly dependent on the discourse function of the connectives. . Discourse function of the connectives In most of our examples, in spite of the apparent logical relation between the segments, the hypotactic link is faint, and a lot of times non-existent. The same applies to the causal link normally underlined by because. In a non-causal context, what roles do the traditional markers of cause and result or inference have? In her paper about the paratactic use of because, Schleppegrell (1991: 323) describes three distinct roles for a non-subordinating marker: [It can be used] as a discourse-reflexive textual link which introduces a reason for or explanation of a prior statement [...] as an expressive, non-causal link which introduces elaboration of a prior proposition [...] [or] as a discourse marker which indicates continuation and response in conversational interaction.

The first case covers the metalinguistic use of because, in which the marker retains a causal function since it introduces an element that justifies the preceding speech act. The other two, however, partake in the dissolution of the causal link that is a frequent phenomenon in conversational English. In this case, the marker can be used as an elaboration marker and be involved in information management. Typically, because prefixes stretches of speech that retain an informational relationship with a previous one, without necessarily expressing causality. The marker can additionally be used as a ligational discourse connective and play an important role in turn-taking management. Because can then appear in the same positions as ligational discourse connectives, namely in turn initial position or, not uncommonly, in final position, where it delays turn alternation. Because and so can no longer be viewed as logical markers restricted to a linking function within subordination. In Schiffrin’s (1987) analysis, so and because take part in the construction of an argumentation by introducing respectively a “position”, i.e. a foreground element, and supporting background elements (“support”).

TSL[v.20020404] Prn:10/05/2007; 9:59

F: PB16108.tex / p.15 (131)

A because B so A’

. Recursive dynamism as an organizational principle of discourse Interestingly, discourse analysis restores a sense of subordination with respect to the clauses introduced by because and so, on a semantic level. According to Schiffrin (1987: 191): So and because are grammatical signals of main and subordinate clauses respectively, and this grammatical difference is reflected in their discourse use: because is a marker of subordinate idea units, and so is a complementary marker of main idea units.

This semantic subordination relative to the main line of discourse is to be viewed as an ever evolving relationship, in which a foreground element supported by an element from the background in turn becomes part of the set of background elements on which speakers can rely to further their conversation and can be used as such. This phenomenon, through which speakers look back to go further, at least partly explains why the pattern under scrutiny is indeed a fertile spiral and not a sterile circle. .. Looking back Recycling linguistic material belongs to the essential principles described by Halliday & Hasan (1976) that ensure discourse cohesion by creating a “texture”. The two authors distinguish between several types of relations among the constituents of a text. Among them are substitution and ellipsis, on the one hand, which formally link the elements on the lexical and grammatical levels, and, on the other hand, reference, which is a semantic relationship. What the types of cohesion have in common is the way they look back to find the element that is elided or to retrieve a reference in the preceding context. If speech looks back, the discursive spiral described in Section 2 is a natural pattern at various levels in its economy. In informational terms, the new element in utterance n becomes given information on which utterance n+1 can find support. The leftmost part of an utterance typically hosts low quantities of information or more consensual elements. It contrasts with the end of the utterance where new segments (or treated as new) or more crucial segments are to be found, underlined by intonation. The syntactic choices performed by the speakers reflect these considerations. Likewise, the backward orientation is at work within the paratone (or oral paragraph) when segments are recycled as the conversation unfolds. A preamble + rheme sequence can serve as a preamble for the following paratone. In example (12), the first utterance allows Hannah, Mike’s addressee, to retrieve the reference of the personal pronoun used as the subject in the second one. The first utterance provides a thematic frame within which the second one can be built. It is recycled as a preamble:



TSL[v.20020404] Prn:10/05/2007; 9:59



F: PB16108.tex / p.16 (132)

Frédérique Passot

(12) Mike&Hannah: Trip M I went with a friend called Ann {0.35} and we walked down {0.4} a river in central Greece

Mike’s concern in ordering his utterances is not a chronological one. Rather, the succession marked by and is relevant to discourse organization: once the actors are introduced, Mike can start telling the story in which they are featured. Linear order can only establish binary relations between neighboring elements. The recycling process allows these elements to extend their relational domains. By looking back and rephrasing already given elements, the discursive spiral creates a system in which the various links in the linguistic chain echo each other. Discourse progresses by successive contacts and shifts inscribed in its linearity. .. Going further Beyond the question of the informational weight of linguistic elements, the spiral pattern has an essential role to play in the construction of intersubjective relations, i.e. in how speakers position themselves with respect to what they assume the addressee’s standpoint is. The pattern is indeed a means to carry the conversation from a consensual point to another consensual point. What is at stake with A because B so A’ structures is for all the participants in the conversation to reach a consensus (presented in A and rephrased in A’): the acknowledgement of a possible discordance is used as a springboard to extend the set of elements in the shared background and eventually open new perspectives for the conversation to unfold. The vital role of the spiral is due to the markers involved. Schiffrin (1987: 207) shows that because and so work hand in hand to serve this strategy, by consolidating the set of elements shared by the speakers. [...] because can be used to preface information when the status of that information as shared background knowledge is uncertain and when that information is important for understanding adjacent talk. This use marks an information state transition because it is a shift from unshared to shared knowledge. And so can be used to preface information whose understanding is supplemented by information which has just become shared background.

In other words, the content of A is submitted to the addressee as potentially consensual. B provides elements that allow the speakers to reach the proposed consensus. In so A’, the content of A is considered as part of the shared background, given what has been said, and can therefore be modalized and elaborated on. In this sense, the spiral pattern creates the conditions in which metalinguistic commentary can develop.

TSL[v.20020404] Prn:10/05/2007; 9:59

F: PB16108.tex / p.17 (133)

A because B so A’

. Conclusions Through the analysis of a specific structure occurring in natural conversation, we hope to have shown how the markers because and so contribute to discourse organization. In a context where strict causal subordination is not the rule, because and so appear jointly in structures that may at first sight be reminiscent of strongly marked argumentation using narrow scope subordination. The apparent duplication of terms however does not create a logical loop. Marked by a qualitative shift on the semantic, referential and modal levels, A and A’ are not interchangeable. The apparent circular pattern can therefore be considered as a spiral involving qualitatively different echoing segments. Used as discourse landmarks rather than logical markers, the connectives play an active role in the enlargement of the shared knowledge on which conversation progresses. The spiral pattern underlined in our examples by the use of the markers is indeed emblematic of a more general tendency in spoken English, or in language in general: new developments find support on established bases, and are in turn recycled into background, consensual matter for newer ones. Further research could therefore extend the analysis of the phenomenon to different scales of discourse organization. We wish to examine how thematic structure is affected by the spiral pattern, following the remarks we made in a previous study (see Passot (2004: 303)). At all levels, our claim is that the apparent circularity of such structures does not hinder discourse progression. Quite on the contrary, it fuels it.

References Biber, D. et al. 1999. Longman Grammar of Spoken and Written English. London: Longman. BNC Consortium (2000). BNC World Edition. December 2000 Release. Oxford: The Humanities Computing Unit of Oxford University. Couper-Kuhlen, E. 1996. “Intonation and Clause Combining in Discourse: the Case of Because.” Pragmatics 6 (3): 389–426. Debaisieux, J.-M. forth. “Quel statut syntaxique pour les propositions « subordonnées circonstancielles » ? La distinction entre dépendance grammaticale et dépendance macrosyntaxique comme moyen de résoudre les paradoxes de la subordination.” In Revue de Sémantique et Pragmatique. Orléans: Presses Universitaires d’Orléans. De Cola-Sekali, M. 1991. Connexion inter-énoncés et structuration des relations temporelles et argumentatives en anglais contemporain. Une étude énonciative des connecteurs polyvalents SINCE et FOR. PhD thesis. Université Paris III – Sorbonne Nouvelle. Halliday, M. A. K. & Hasan, R. 1976. Cohesion in English. London: Longman. Morgenstern, A. & Sekali, M. forth. “De la parataxe aux premiers connecteurs : les prémices de l’argumentation chez le jeune enfant.” In Cahiers de l’ENS-LSH.



TSL[v.20020404] Prn:10/05/2007; 9:59

F: PB16108.tex / p.18 (134)

 Frédérique Passot

Passot, F. 2004. La hiérarchisation des constituants discursifs dans un corpus d’anglais oral spontané. Unpublished PhD thesis. Université Paris 3 – Sorbonne Nouvelle. Passot, F. 2006. “Because and cos in context.” In Aux marges du texte. Texte et co-texte, F. Girard– Gillet (ed.), 103–114. CIEREC Travaux 128. Saint-Étienne: Publications de l’Université de Saint-Étienne. Rutherford, W. 1970. “Some observations concerning subordinate clauses in English.” Language 46 (1): 97–115. Schiffrin, D. 1987. Discourse Markers. Cambridge: CUP. Schleppegrell, M. 1991. “Paratactic because.” Journal of Pragmatics 16: 323–337. Simon, A. C. & Grobet, A. 2002. “Intégration ou autonomisation prosodique des connecteurs.” In Speech Prosody 2002. Proceedings of the 1st International Conference on Speech Prosody, B. Bel and I. Marlien (eds), 647–650. Vallée, M. 2003. “Because et for partagent-ils les mêmes propriétés énonciatives ?” CORELA – Cognition, Représentation, Langages 1 (2).

TSL[v.20020404] Prn:10/05/2007; 10:06

F: PB16109.tex / p.1 (135)

Not that . . . versus It’s not that . . . Ruth Huart University Paris-Diderot

Although sentences beginning with not that seem to lack a matrix clause, the missing material cannot be recovered directly from the preceding context. Introductory not that is used to refute an apparent implication of the previous utterance. Comparison with the use of the more “complete” it’s not that, also used to refute a supposed implication, shows that the discourse functions of the two structures are actually quite different: initial not that concerns the motive for the previous speech act, whereas it’s not that involves the circumstances leading to the state of affairs that has just been described. The latter structure may involve other adverbs in addition to or in place of not, all incompatible with initial not that. We attempt to show that the two constructions differ with respect to the scope of negation, and, in enunciative terms, the construction of “notional domains”.

.

Introduction

In certain styles of writing and speaking, the sequence “not that” can begin a sentence, i.e. appear after a full stop or semi-colon, or a terminal intonation pattern. Since the conjunction that, systematically reduced phonetically in this pattern, is normally considered as a “complementizer” or subordinating conjunction, one might imagine that a comma or non-terminal intonation would be more in keeping with a subordinate, or dependent, construction. However, it is almost always impossible to annex the not that sentence to the previous utterance without adding additional material. Nor is it normally felicitous to add a “dummy” main clause, in the form of IT + the copular verb BE, as this structure, also common, does not appear in the same contexts. The aim of this paper will thus be to attempt to analyse how it is that an apparently incomplete structure functions as an independent utterance, by comparing uses found in authentic discourse of not that, on the one hand, and it’s not that, on the other. This is not, strictly speaking, a corpus-based study as the examples come from sources which happened to have been scanned by other researchers, or from personal reading, rather than from a statistically regulated data

TSL[v.20020404] Prn:10/05/2007; 10:06

F: PB16109.tex / p.2 (136)

 Ruth Huart

base. However, careful observation of about sixty occurrences with their contexts shows sufficient consistency in the way the two structures are used to allow the formulation of certain hypotheses. The analysis is conducted within the enunciative framework developed in France by Culioli (“Theory of Enunciative Operations” = TEO). The essential concepts and corresponding terminology will be defined as required. Few studies seem to have been devoted to not that as a specific discourse marker. Working in the same theoretical framework, Ranger (1998) devotes several pages to not that, which he considers to have “concessive” value, and more specifically a “rectifying concessive marker” (Ranger 1998). For Haiman (1995), introductory not that is used to show that the “speaker’s interest in what follows is minimal”. Both authors ignore the difference with it is not that and suggest that the latter is a “full form” of the structure. Before providing our own observations on the different discourse values of the two structures, we shall first look briefly at the expression of “concession” and its treatment in TEO.

. Construction of “concessive” meaning Making a concession means giving in a little, but not entirely. In terms of speech, the speaker admits that something is true which does not normally follow from something else which (s)he wishes to maintain as true: I’m supposed to spend weeks in bed after being confined, though there’s nothing wrong with me. (Barbara Vine, Asta’s Book)

Spending weeks in bed normally follows from being ill. In the case at hand, there is nothing wrong but spending weeks in bed nonetheless ensues.1 This notion has been developed extensively in two relatively recent doctoral dissertations by TEO linguists (Filippi, Ranger). Basically, the concessive interpretation results from the juxtaposition of two assertions which, according to some “natural order” are not expected to be valid concomitantly: p is the case, q is the case, while q is “normally” associated with non-p or “other than p” (Ranger 1998; Filippi 1997; Paillard 1993 . . .). This “normal” implication, dependent on primitive relations or “world knowledge”, is attributed to a theoretical ideal addressee (noted S0 ’). Typical markers are although, as, while, however, may, . Terms like nonetheless, all the same, used in the gloss, explicitly mark the processes at work: the fact that proposition 1 obtains in no way detracts from the fact that proposition 2 obtains equally / to the same extent / fully.

TSL[v.20020404] Prn:10/05/2007; 10:06

F: PB16109.tex / p.3 (137)

Not that . . . versus It’s not that . . .

none of which can be substituted for not that in the type of occurrence we are considering, mainly because they normally apply to proposition p, not q: (1) ‘Well, we’ve got peace now,’ he said. ‘Life will return to normal.’ How easy it was to express the sentiment: he almost believed it himself. ‘We’re all Americans now,’ she said. ‘That will be interesting. Not that I shall live to see it.’ ‘Better this than what we were,’ he said half-heartedly. (William Boyd, The Blue Afternoon) a. *However interesting that will be, I shan’t live to see it. b. *Interesting as that will be, I shan’t live to see it. c. ??Although that will be interesting, I shan’t live to see it.

Slightly better results can be obtained with other clause-initial connectives, such as even if or although attached to proposition q: d. ?That will be interesting although I shan’t live to see it. e. That will be interesting even if I don’t live to see it.

And still better approximation using however as a sentence adverb: f.

That will be interesting. However, I shan’t live to see it.

Yet, none of the alternative forms are really equivalent to the original, for it would obviously be ridiculous to say that something being interesting follows from or leads to someone seeing it. The closest we can get to an equivalent meaning using although might be: g. ∼That will be interesting, although I shouldn’t make such a prediction since I won’t live to see it.

In other words, the implication which is denied here is that saying that something will be interesting stems from an expectation that one will feel interest, which can only happen if one is still alive when it happens. It would seem then that the idea of concession comes from what the speaker thinks the addressee might consider an inappropriate speech act, rather than from the propositional content of the not that clause. This is borne out by the following example, in which not that follows a direct question: (2) ‘Wieland says you were in Sampaloc. In a creepjoint.’ ‘Yes.’ This did not surprise him. [. . .] ‘You go there often? Not that I care,’ he added quickly. ‘I whore myself from time to time.’ (William Boyd, Blue Afternoon)

Here, an appropriate gloss might be: ∼ you mustn’t think I care about the answer, even though I asked the question.



TSL[v.20020404] Prn:10/05/2007; 10:06

F: PB16109.tex / p.4 (138)

 Ruth Huart

Further confirmation of this metalinguistic value is to be found in the following example, where the word say appears in the not that sentence: (3) ‘All right then, I did think you were good. I didn’t think you were bad, at any rate. I thought it was a rotten play, and I detest that girl, but I thought that you were quite good. You were the best thing in it, not that that is saying much. May I have my cigarette now please?’ (Margaret Drabble, The Garrick Year)

Using a superlative (best) should be a high compliment, but the speaker hastens to cancel this interpretation of his speech. This is a clear illustration of the “rectifying” value which Ranger sums up in these terms: “from p in all its consequences to merely p” (Ranger 1997: 114), where p (you were the best thing in the play) in all its consequences would include: you were excellent, and merely p goes no further than: you were better than the others.

. Conditions for appearance of initial not that The two examples considered so far occur in direct speech, in a fictitious dialogue. It is also common to find initial not that in (imaginary) diaries, which are a sort of self-directed monologue, or in mildly polemical texts which address the reader as a potential arguer. (4) I’d like to have Luke here, in this bedroom while I’m getting dressed, so I could have a fight with him. Absurd, but that’s what I want. An argument, about who should put the dishes in the dishwasher, whose turn it is to sort the laundry, clean the toilet; something daily and unimportant in the big scheme of things. We could even have a fight about that, about unimportant, important. What a luxury it would be. Not that we did it much. These days I script whole fights, in my head, and the reconciliations afterwards too. (Margaret Atwood, The Handmaid’s Tale) (5) MY LIBERAL FRIENDS were quite unhappy with me recently when I suggested in this column that this new JFK, is – in the words of Lloyd Bentsen’s put-down of Dan Quayle – “no Jack Kennedy.” Not that they necessarily thought me wrong, of course; they just thought it was mean of me to say it (these, of course, are the same people who think nothing at all of routinely dismissing Dubya as mean-spirited, arrogant, and dumb as a post; my remark, I’m to understand, was merely opinion, and hateful opinion at that; they, on the other hand, merely state fact). (WEB Corpus, “This JFK is not that JFK” by Mickey Edwards, June 12, 2004)

In (4) the detailed description of the fights the narrator dreams of could lead us to believe they echo real fights of the past. It is this mistaken conclusion that

TSL[v.20020404] Prn:10/05/2007; 10:06

F: PB16109.tex / p.5 (139)

Not that . . . versus It’s not that . . .

she hastens to deny. The technique in (5) is identical: friends being unhappy with a statement could mean they thought the author wrong, but this was not the case. The similarity with concession is that the speaker feigns to believe that the interlocutor might make the wrong inference from the first part of the utterance. However, what is being inferred is the motive for the utterance and not the propositional content. . Enunciative analysis It is quite remarkable that in factual reports or traditional narrative with an omniscient narrator, the structure is much less frequent (no occurrences in the texts of this type we examined). This indicates that the system of “coenunciation”, fundamental in Culiolian theory, is actively at work in this structure (Culioli 1985: 61–62). We postulate that all language activity must be anchored in a situation (Sit) having as its primary coordinates a subjective dimension (S) and a spatiotemporal dimension (T). The origin of a linguistic event, the “decider” who takes charge of any modal components (S0 ), addresses an idealised interlocutor (S0 ’), who in many cases has no physical counterpart (internal monologue, rhetorical questions. . .). The utterance may be constructed to show that the two theoretical instances S0 and S0 ’ are treated as “on the same footing” or “on a different footing” with respect to the propositional content (p). In English that clauses in general, the predicative relation (propositional content stripped of modal or temporal determination) is “pre-constructed”, i.e. envisaged as potentially “being the case” due to established factors in the discourse situation, but assertion is absent. Instead of assertion (0 modality), the presence of that indicates dependence on the speaker’s choice of modality to locate the predicative relation. Choices available include cognitive status (I think, I suppose, the idea, the fact. . .), value judgements (I hope, I’m afraid, it is unfortunate, surprising. . .), epistemic evaluation (it is likely, possible. . .) or identification of another enunciative source (indirect speech). (See, for example, De Mattia 1997) With introductory not, the modality chosen is negation. To deny, it is first necessary to establish potential existence. The notional content of the that clause, considered as “contained in” or “part of ” the domain associated with the previous utterance in the mind of a potential addressee, is “ejected”, because in the instance at hand it does not belong. On this analysis, that at the beginning of a clause is a sign that assertion is suspended. Thus, a sentence beginning with not that is not equivalent to a negative assertion. This would seem to be confirmed by examples in which a negative assertion follows a not that sentence, showing that the second is not fully redundant:



TSL[v.20020404] Prn:10/05/2007; 10:06

F: PB16109.tex / p.6 (140)

 Ruth Huart

(6) Johnson’s photographs of Joey have been widely reproduced, sometimes in literature about Tay-Sachs, and their visibility is always comforting. One photograph “keeps showing up places. ... And here I always thought I would be left alone with my grief.” Not that the grief itself ever goes away. According to Dr. Ronnie Sue Stangler, grief lasts [= grief does not go away]. (WEB corpus: Fred Moody, “Images become treasures to grieving parents”, The Seattle Times, March 16, 2005) (7) I, in my hard-headed way, [. . .] had simply lifted from my background what I thought would be of use: [. . .] a harsh degree of anonymity, and a respect for my father. And a certain fore-ordained distance from and attraction towards people like Mrs Scott, who had herself, it is quite easy to see, been some sort of mother-substitute for me. Not that I wanted a mother like that: I did not, even as a child, but she must have represented for me a regularity that I should have needed. (Margaret Drabble, The Garrick Year)

If the speaker feels the need to assert I did not [want a mother like that], it indicates that she has not yet done so. . Discourse function Two main uses of initial not that can be distinguished: metalinguistic rectification, as in (2) and (3) above, and change of topic, also exemplified in (3). In this section, we will examine more closely how these two subcategories function. .. Meta-linguistic rectification We have seen that initial not that is used to rectify a first impression which might result from the previous utterance, in particular from the form chosen for this previous utterance. Another type of meta-linguistic rectification concerns the lexical choices made by the speaker: (8) ‘...How’s Flora?’ ‘Oh, she’s all right. As far as the eye can see. Not that the eye can see very far.’ (Margaret Drabble, The Garrick Year) (9) Rasmus has become nearly as fond of fashion as he is of motor cars – I tell him that when these “autos” go out of favour, as they must do, he’ll be able to sell frocks–and wants to see me dressed up. I suspect he thinks it good for business to have a handsome wife about when customers come to the house. Not that I’m handsome but these days I do look smart. (Barbara Vine, Asta’s Book)

In (8), the speaker takes a critical look at the meaning of the idiomatic expression “as far as the eye can see”, which implies, taken literally, to the horizon. However, the notion of distance is only relative, and what may be “far” for some purposes is

TSL[v.20020404] Prn:10/05/2007; 10:06

F: PB16109.tex / p.7 (141)

Not that . . . versus It’s not that . . .

quite limited when measured in number of miles, for example. The initial qualification of the assertion, “she’s all right” is thus further restricted by the Not that sentence, so that in the end, the speaker leads us to understand that the assertion is not necessarily valid. (I say she’s all right, but she may not be because perception can be deceptive). In (9), it is the word handsome which requires rectification. By a series of predications revolving around her appearance in her husband’s eyes, the speaker first implies that <wife, be handsome> might be the case, before rejecting that formulation in favour of a more appropriate one. The statement he wants to see me dressed up is to be construed as a modal judgement: = the state of affairs <wife, be dressed up> has a positive value for the husband. <S, be dressed up> is then assimilated to <S, be handsome>, also positive (good for business).

However, the qualification is ultimately considered less appropriate than . Notice that the relation <wife, be handsome> is never asserted, nor is the negation in not that equivalent to asserting: “I’m not handsome”, the only true assertion being: “I do look smart”, in which the use of do is eloquent. The notional domain constructed is something like <S, be nice to look at>; the so-called “emphatic do” shows that the validity of predicating this quality to the subject has been put into question before being selected as the right value (Souesme 1989). The hesitation around the validity of the predication concerns the linguistic form with which to “incarnate” the notion: /look smart/ approximates /be handsome/, whether or not /handsome/ really applies.2 To generalize from these two examples, we can say that the notional content of the that clause refers to a property or a set of properties associated with the terms used in the previous utterance. The negation shows that, in light of these properties, the terms in question are not entirely adequate for the speaker’s purposes. Adjustments are needed before the “right” choice can be endorsed. .. Change of topic In some cases, an entirely different discourse strategy seems to be involved: the not that sentences closes the discussion, and allows the speaker to move on to something else. An example of this was (3) above, where, having dismissed the compliment he did not want to make, the speaker asks for a cigarette.

. Small capitals between slashes are conventionally used to represent the “notion” or set of properties associated with the expression, rather than the expression itself.



TSL[v.20020404] Prn:10/05/2007; 10:06

F: PB16109.tex / p.8 (142)

 Ruth Huart

(10) Blue eyes aren’t necessarily beautiful and I’m sure mine aren’t. They are too blue, too bright, the kind of colour that is better on a peacock or a kingfisher. In fact, they are exactly the colour of the butterfly’s wing in the brooch I had from Tante Frederikke for my sixteenth birthday. Not that it matters what colour they are. No one looks at an old woman’s eyes and I feel like an old woman, though I’m not quite twenty-five. (Barbara Vine, Asta’s Book) (11) Still, he had a good eye for the look of a thing, for the shape of a thing, and that’s how he had ended up in the job at Morgan Hero, twenty years and counting in a printing firm in the Euston Road, designing the way all kinds of things should be folded – envelopes, direct mail, brochures, leaflets – not much of an achievement, maybe, but you’ll find things need folds, they need to overlap, otherwise life would be like a broadsheet: flapping in the wind and down the street so you lose the important sections. Not that Archie had much time for the broadsheets. If they couldn’t be bothered to fold them properly, why should he bother to read them (that’s what he wanted to know)? (Zadie Smith, White Teeth)

Extract (10), from an imaginary diary, is another clear example. After a discussion of the precise colour of her eyes, which would seem to imply that finding the right description matters to her, the narrator denies this. Excluding this “normal” component from the situation at hand serves as a jumping off point for another topic – her age. In (11), we are plunged into the stream of consciousness of Archie, who recognizes in himself a gift for folding, which he first belittles (not much of an achievement) and then shows to be necessary (but things need folds). The image used to demonstrate this necessity is that life without folds would be like a broadsheet, i.e. a large unfolded newspaper. The metaphor concerns the physical aspect, but the negative appreciation expressed also includes the content (you lose the important sections). The sentence beginning with not that introduces a complete change of subject: Archie is not interested in current events. The linguistic function of not that remains the same: saying that parts of the newspaper are important implies taking time to read them, but this is not true of Archie. However, the further discourse function is to offer a transition to another topic: if you think what I have just said includes p, forget it; now that that is taken care of, we can get on with essentials. In other words, not that both looks back to close and ahead to open.

. Contrast with It’s not that In the contexts we have studied so far, it would not always be impossible to add it is or it was before not that, but even where possible, the effect produced is strange,

TSL[v.20020404] Prn:10/05/2007; 10:06

F: PB16109.tex / p.9 (143)

Not that . . . versus It’s not that . . .

if not incompatible with what follows. The best candidate for substitution among the examples above would be (9), where there is direct reference to what precedes (the word handsome is used) and an opposition introduced by but follows. (9’) I suspect he thinks it good for business to have a handsome wife about when customers come to the house. It’s not that I’m handsome but these days I do look smart.

In the other cases, particularly the last one, adding it would be totally incongruous: (11’)

??It was not that Archie had much time for the broadsheets.

When real occurrences of it is/was not that are considered, it becomes obvious that conditions of usage are quite different. Moreover, the adverb not assumes its usual role of negating the validity of a predicative relation, and can permute or combine with other adverbs: it is (not) just / only / simply that. . .; it is probably / certainly (not) that. . ., none of which can appear sentence initially followed by a that clause. . The value of it In this construction, the pronoun it can be considered as “referential” or “vaguely referential” to the extent that we are obliged to refer back to the previous utterance to understand what is denied. In fact, it may be said to refer to the state of affairs that triggered the previous utterance, whose content is always taken into account and reassessed under a new light. (12) When I went out this morning a woman asked me if polar bears3 in the streets of Copenhagen. She is one of our neighbours and she [. . .] thinks I must be a savage and half-witted too because I’m not English and I don’t speak English well and stumble over words Most people here feel like that about us. It is not that there are no foreigners (as they see us), they are used to people from Europe but they don’t like us, any of us. They say we live like animals and take away their jobs. (Barbara Vine, Asta’s Book) (13) The teacher’s pet is the only person Dinah has met, except for Catherine Cleary, who’s completely and utterly tone deaf. It’s not that she’s a bit off key, or that she hits too many flat notes. The sound that she makes in singing

. This syntactically anomalous sequence was found in the scanned version of the original. As it appears in a fictitious diary, we can perhaps see this as a stylistic ellipsis of the copula: “if there are polar bears. . ..”.



TSL[v.20020404] Prn:10/05/2007; 10:06

F: PB16109.tex / p.10 (144)

 Ruth Huart

lessons bears no relation to the music whatsoever. (Barbara Trapido, Frankie & Stankie)

As with initial not that, the speaker denies an inference that the supposed listener might be tempted to make, but instead of the justification for the speech act, the inference concerns the propositional content. A statement is asserted all the more strongly as the origin of the situation considered is not (or not only) what might have been expected. Suspicion of foreigners (12) could be due to not being acquainted with any; in the case at hand, no foreigners is not the right explanation. Indications of being tone deaf (13) are frequently “singing a bit off key” or “hitting too many flat notes”; in the case at hand, the indications are far worse. Schematically, it is a “place-holder” for the true fact(s) behind what has just been stated. However, the it in this construction is not simply a dummy subject of the type found in cleft sentences, for example. One observation in favour of this statement is that the affirmative counterpart it is that. . . can only appear after the negative form. An example of this can be found in the passage which follows immediately from extract (5), extended here as (14): (14) MY LIBERAL FRIENDS were quite unhappy with me recently when I suggested in this column that this new JFK, is – in the words of Lloyd Bentsen’s put-down of Dan Quayle – “no Jack Kennedy.” Not that they necessarily thought me wrong, of course; they just thought it was mean of me to say it (these, of course, are the same people who think nothing at all of routinely dismissing Dubya as mean-spirited, arrogant, and dumb as a post; my remark, I’m to understand, was merely opinion, and hateful opinion at that; they, on the other hand, merely state fact). So let me put it another way: John Kerry’s bigger problem is not that he’s not Jack Kennedy; it’s that he’s not Ted Kennedy.

In this case, the pronoun it substitutes for the NP John Kerry’s bigger problem. However, the same type of rectification is frequently found without a full NP in the preceding context: (15) There is much to be said for this collection of essays and, given the contributors, it is not unjust of the dust-jacket to call it a ‘major’ work, but the claim that this is a ‘new’ assessment hardly stands up to examination. [. . .]. It is not that some of the individual essays are not excellent; it is that an atmosphere of idolatry pervades the whole, even as a clove of garlic does a stew. (WEB corpus)

A justification is sought for the judgement expressed: this collection is not a new assessment. One possible factor is eliminated before a more accurate one is selected. The it in the affirmative version is a repetition of the previous one, but the

TSL[v.20020404] Prn:10/05/2007; 10:06

F: PB16109.tex / p.11 (145)

Not that . . . versus It’s not that . . .

order could not be reversed. As observed in Section 2, negation is normally a secondary operation, once existence has been postulated. Thus one can only conclude that the first it is also anaphoric. It would in fact be perfectly possible to replace the pronoun by a vaguely referential noun phrase: the problem is not that / the thing is that. . . Obviously, with it, no actual term in the text is replaced and the reference is quite vague. However, the link with the content of the previous utterance is clearly more direct than in the case of initial not that. Note, for example, that when introduced by it is not, the that clause may show ellipsis of a constituent of the preceding sentence, which was not at all the case with initial not that: (16) I don’t date often. It’s not that I don’t like to, but just consider myself kind of picky. (WEB corpus) = It’s not that I don’t like to (date) (8’) ‘...How’s Flora?’ ‘Oh, she’s all right. As far as the eye can see. Not that the eye can see very far.’ ??not that it can Ø very far (9”) I suspect he thinks it good for business to have a handsome wife about when customers come to the house. Not that I’m handsome but these days I do look smart. *not that I am Ø

This important syntactic difference is accompanied by a significant difference in discourse function. Example (16) provides a simple illustration of the most common pattern for the structure under study: an assertion ( is the case) is followed by denial of what might be supposed to be a cause leading to the state of affairs described (I don’t date because I don’t like to date). This cause is then replaced by another: I am picky is the real reason. The conjunction but, which marks a change in orientation, occurs very frequently to correct the putative “false impression”. However, a strict dichotomy between a “wrong” and a “right” value for it (the trigger) is by no means typical, and different causes can coexist. In the next section, we will examine how the construction is used to approximate the “right” value. . Search for the proper “trigger” It is worth noting that what we are informally calling a “trigger” is frequently an observable manifestation of a mental state, attitude or emotion: (17) We had arrived half an hour early, naturally, and had a reserved compartment. We put Pascal and Joe into the compartment, then David, Flora, and I set off to have a look round the station. I must confess that I was quite excited: I like journeys, I like stations, it is not that I am full of hope, but I am easily filled



TSL[v.20020404] Prn:10/05/2007; 10:06

F: PB16109.tex / p.12 (146)

 Ruth Huart

with some kind of anticipation which could well be mistaken for it. (Margaret Drabble, The Garrick Year)

Just as in (13) a correlation was made between signs of being tone deaf and the diagnosis of this handicap, in (17), there is an attempt to link the cause of excitement to its presence. In this particular case, we are not far from the “metalinguistic rectification” observed in (9): /anticipation/ is not exactly /hope/, but almost, as far as perceptible signs are concerned. The difference is in what is rectified: in (9), it was the choice of the phrase handsome wife that needed qualification; here, it is what the speaker wishes to put forward as an explanation for liking journeys and stations. S0 asserts <S, like journeys> is the case. S0 ’ might suppose <S, be full of hope> is the origin of <S, like journeys> This causal relation is denied.

In the segment following the denial, we can observe signs of construction of “a boundary zone” between what is “really /hope/” and what is “really not /hope/”. Since the term does not seem entirely appropriate to the speaker, she chooses another, anticipation, which however she also qualifies (some kind of ). Thus, the factors which cause the subject to like journeys and stations do not correspond fully to the properties that define the notion /hope/, nor to those which define the notion /anticipation/ – they are in the boundary zone between the two (not really p). Instead of one alternative, several possible “triggers” may be evoked: (18) A change generally came over the demeanour of Mr. Justice Edmondson when the time came for him to sum up. His lethargy was shed and those who suspected he had not been attending or that he was indifferent to the proceedings, understood that their suspicions were unfounded. It was not that he suddenly became alert. It was not that he began to speak in circumlocutory or mellifluous phrases but rather that he showed a grasp of what had gone before and a mastery of the facts, an absolute ability to distinguish between the subtle hints of the defence and the prosecution and the true hard evidence. (Barbara Vine, Asta’s Book)

Here again, the statement to be justified concerns the diagnosis of a mental state: people understood that Mr. Justice Edmondson was paying attention. Possible signs of the change are first rejected before a more satisfactory one is chosen. In this case, the affirmative statement finally chosen contains the comparative term rather, indicating that within the set of possible causes of people’s understanding, the one selected is more satisfactory than the others proposed, which implies that it might not be the absolutely right choice. In other words, the utterance again contains signs of construction of a boundary zone.

TSL[v.20020404] Prn:10/05/2007; 10:06

F: PB16109.tex / p.13 (147)

Not that . . . versus It’s not that . . .

. Negation, but not exclusion Qualification showing that the “trigger” sought may not be unique can be found in both clauses. On the one hand, we can find modifiers linked to the negation, as in (19) and (20), and on the other, as in (16) above and (21) below, the assertion which follows not that may contain a “limiting adverb”. (19) I began to miss London: it was not so much that I had many close friends there, for I have few friends, but I missed variety. My tastes are shallow; my life is shallow; and I like anonymity, change and fame. In Hereford I could have none of these things. . . (Margaret Drabble, The Garrick Year)

The phrase “not so much X (as Y)” implies that X cannot be totally eliminated: having friends, even if they are not many, is part of the reason for missing London. The but clause which follows introduces the dominating factor, without necessarily eliminating the first. In the following extract, this idea of partial contributors to an existing state of affairs is made explicit: (20) But [a TV series] does have to be full of characters and story lines that are both compelling and comprehensible, and “Tilt” is short on both counts. It’s not just that the poker scenes won’t make total sense if you aren’t a player (and they don’t); it’s not just that the premiere is long on style and short on substance (and it is); it’s certainly not that that the first hour poses a lot more questions than it answers (which it does). The fatal flaw is that, after watching the first hour, there’s no compelling reason to come back for the other eight episodes to find out the answers. “Tilt” just doesn’t seem to be worth investing the time. (WEB corpus, Scott D. Pierce, There’s not a great deal to like about ESPN’s ‘Tilt’, Deseret Morning News)

Thanks to the statements between brackets, it is obvious that the content of the that clause is not denied in spite of the preceding negative adverb. What is denied is the weight of each of these factors in the final judgement (the fatal flaw). At the same time, the “limiting” adverb just shows that each of the factors mentioned belongs on the critic’s list of “flaws”. The pronoun it may be considered as referring to what motivates the statement “Tilt falls short”: factor 1, factor 2, factor 3 . . . are all valid, but none is the exclusive factor. Exactly the same type of progression is present in the following passage, in which we find the NP the reason instead of the “vaguely referential” pronoun it: (21) The trouble with Miss de Sancha is, like so very many people, she is trying to shift the blame away from herself - to present herself as a victim and hence solicit sympathy. It will not wash. Quite apart from the issue of talent, the



TSL[v.20020404] Prn:10/05/2007; 10:06

F: PB16109.tex / p.14 (148)

 Ruth Huart

reason she cannot be taken seriously as an actress is not that she had an affair and society always takes it out on the woman. It does not any more. Nor is the reason that she had an affair with somebody famous. Unknown actors and actresses do that all the time. No. The reason is that because she had an affair with the ‘Minister for Fun’, a politician who, even before the news got out, was a bit of a joke. If only she had been a real Scarlet Woman, then who knows? (WEB corpus, The Sunday Telegraph)

Here it is clear that the negation concerns the link established by the speaker between the content of the that clause and the state of affairs observed. Why can she not be taken seriously? Not because A (which is the case), not because B (which is the case), not because C (which is the case), but because D. The fact that several possibilities can coexist before the “right” choice is made sometimes results in the need to single out the latter by means of a qualifying term. In (18) (repeated below), we observed the presence of rather after but, which can be considered to mark a “centring” operation (Gilbert 1989; Salkie this volume): (18) A change generally came over the demeanour of Mr. Justice Edmondson when the time came for him to sum up. His lethargy was shed and those who suspected he had not been attending or that he was indifferent to the proceedings, understood that their suspicions were unfounded. It was not that he suddenly became alert. It was not that he began to speak in circumlocutory or mellifluous phrases but rather that he showed a grasp of what had gone before and a mastery of the facts, an absolute ability to distinguish between the subtle hints of the defence and the prosecution and the true hard evidence. (Barbara Vine, Asta’s Book)

When we speak of centring in this case, we mean entering the notional domain from the outside, approaching the centre from the periphery. Another marker which corresponds to this operation is just, already seen in the negative sentence, but also frequently used in the rectifying sentence. (22) Well, I did something that I swore to myself that I would never do. I bought an SACD player. Don’t get me wrong, it’s not that I have anything against SACDs or anything, it’s just that I’m still carrying a grudge (sort of) from when we all got duped with the introduction of CDs. (WEB corpus)

The informal tone imitates the spoken language, and the introductory phrase “don’t get me wrong” clearly indicates that the speaker places himself in a dialogue situation with an interlocutor: S0 ’ = S0 . In his attempt to clear up the potential “misunderstanding”, he hovers between the interior and the exterior of the notional domain /have something against SACDs/. To swear never to buy an SACD player could lead one to believe this predicate applies; this is denied partially, but not entirely, as CDs are not exactly SACDs and just restricts actual

TSL[v.20020404] Prn:10/05/2007; 10:06

F: PB16109.tex / p.15 (149)

Not that . . . versus It’s not that . . .

objection to potential objection, i.e. mistrust due to a memory of an unpleasant experience with a related product. In familiar terms, we can sum up the oscillation as “I don’t but I do”. Through the use of just, the preceding not is close to not quite. It’s just that frequently appears after other partial negations with the same sort of restrictive function (Aijmer 2002). (23) This is not to say that grieving parents do not move on – it’s just that they move on to a different life. Instead of getting over their tragedies by forgetting them, as others suggest, they recover and move on by remembering. Images become treasures to grieving parents. (WEB corpus, Fred Moody, The Seattle Times) (24) John says that from casual conversations with Haldane it was clear that he’d thought about the problem that William Hamilton, at Oxford, became famous for, and had provided one of the early solutions to it. That doesn’t in any sense discredit Hamilton, it’s just that the basic idea had been anticipated by Haldane. (WEB corpus, Brian Goodwin, “Biology Is Just a Dance”)

Just limits the relevant information to a portion of what has just been denied: it is not the case that p is not the case; p is the case to a limited extent.

We call this “centring” because the boundary area constructed by the shift from an apparent implication to a denial is eliminated thanks to the “right” choice of a single value within the domain: passage from F (or B, for Boundary: neither really p nor really not p) to I (Interior: precisely p, once the irrelevant properties have been eliminated). This sort of inter-subjective adjustment is made perfectly explicit in the following dialogue: (25) “I never said she acts spacey.” “Yes, you did.” “No, I didn’t.” Privately she added “People who put words in my mouth” to her list of pet peeves. “Oh, well. So you think she doesn’t act spacey.” “It’s not exactly that. It’s just that she. . .always tries to please everybody. It’s like she changes personalities to fit in with whoever she talks to.” “Holy Saint Mary Francis.” (Web corpus, Emily Chung, Speak (short story contest winner))

The discussion centres around one character’s interpretation of the other’s opinion: S’ think/not think that p (<she, act spacey>) is the case.



TSL[v.20020404] Prn:10/05/2007; 10:06

F: PB16109.tex / p.16 (150)

 Ruth Huart

Denying having asserted p is not exactly equivalent to thinking non-p. The relevant information is restricted to what seems to be a sub-domain of, or a domain related to, “acting spacey”, i.e. changing personalities. What is particularly interesting in this case is that two clearly anaphoric terms, it and that, are used in the first sentence with referential value: IT (what I think) is not exactly THAT (she does or does not act spacey). The same two forms appear in the rectifying sentence with the values we have been examining. While it is generally admitted that demonstrative that and connective that, which have distinctly different phonetic realizations (the first is never reduced and the second always is)4 , are no longer the same grammatical entity, the semantic link between the two is brought out here by the proximity between the two occurrences and the possibility to rephrase one by incorporating the other: It’s not exactly that I think she doesn’t act spacey. Moreover, demonstrative that is among the very limited number of terms which can be found after sentence initial not: (26) There was no such thing as society in New Orleans last week. Please, not that. Not that quote again – “There is no such thing as society.” (Web corpus: September 12, 2005, Clive Crook, National Journal, “The Good Ol’ USA?”)

These observations tend to confirm that the “complementizer”, like the deictic that, is used to “point to” an entity whose existence has already been established. A possible extension of the present study could be a revival of the question of how the various uses of that relate in present-day English. Another might be a closer look at sentence initial not in other contexts (not only, not once, not again. . .) to see what they share in terms of speaker involvement.

. Conclusion Now that we have seen how the two structures are used in discourse, we are in a position to propose a representation of the basic difference in value, in particular with respect to the construction of negation. Initial not that reflects the passage from the interior to the exterior of a pre-constructed notional domain: the predicative relation represented in the that clause is treated as though it were “rejected” from its normal place as motivation for a speech act. Not that

: p is not the case (in spite of appearances). The boundary between the two zones is empty, so that going from I to E can often mean entering a new domain (changing the subject). . See Berkenfield (2001) for a statistical study of the phonetic realizations of that in its different uses in spoken English.

TSL[v.20020404] Prn:10/05/2007; 10:06

F: PB16109.tex / p.17 (151)

Not that . . . versus It’s not that . . .

In the it + be construction, it has “vaguely referential” value, depending for its meaning on the notional content of the previous assertion: a state of affairs for which a “trigger” is sought. X (the trigger) =

(whether or not

is the case is irrelevant)

Various adverbs can be used to show to what extent the content of the that clause is considered as the origin of the state of affairs evoked by the previous utterance. In other words they limit the identification marked by the copula. The negative adverb not can be used alone to deny the relation, or accompanied by just (19), so much (18) and no doubt other restrictive adverbs to show that various “triggers” can be identified. We have seen that the negative statement is sometimes followed by an affirmative one built on the same model. The configuration reflects the construction of the interior plus a boundary zone, where there is room for “not really p”. Instead of a clear-cut “exit”, we find a tentative “approach”, which can be reiterated or multiplied. The clearly different interpretations and discourse functions of the two configurations would seem to indicate that one is not simply a syntactic variant of the other. To the question “if the not that sentence is elliptical, what is elided?” we would reply, “nothing”. Placing a term in initial position is a device available to the speaker to show that the reference point for anchoring the utterance is the enunciative situation itself. Initial not that has modal value, in that it provides a direct (non-mediated) reflection of the speaker’s representation of an idealized addressee. The pronoun it, on the other hand, is fundamentally anaphoric and serves as a representative of an existing state of affairs.

References Aijmer, K. 2002. English Discourse Particles: Evidence from a corpus. [Studies in Corpus Linguistics 10]. Amsterdam/Philadelphia: John Benjamins Berkenfield, C. 2001. “The role of frequency in the realization of English that.” In Frequency and the Emergence of Linguistic Structure [Typological Studies in Language 45]. Bybee and P. Hopper (eds), 281–307. Amsterdam/Philadelphia: John Benjamins. Culioli, A. 1985. Notes du séminaire de DEA 1983–1984, Poitiers/Université Paris 7. Culioli, A. 1990. “La Négation : marqueurs et opérations”, in Pour une linguistique de l’énonciation, T1, 91–113. Gap/Paris: Ophrys. De Mattia, M. 1997. “A propos de That/Ø et des frontières du discours indirect”, Anglophonia 2, 53–82. Toulouse: Presses Universitaires du Mirail. Filippi, C. 1997. “Though et la relation de concession : opération énonciative et jeu sur le domaine notionnel.” In Actes du colloque La Notion, Université Paris 7, février 1996, M.-L. Groussier & C. Rivière (eds), 91–98. Gap/Paris: Ophrys Gilbert, E. 1989. “Quite, Rather”, Cahiers de Recherche en Grammaire anglaise, T.4, 5–61. Gap/Paris: Ophrys



TSL[v.20020404] Prn:10/05/2007; 10:06



F: PB16109.tex / p.18 (152)

Ruth Huart

Haiman, J. 1995. “Moods and Meta Messages: Alienation as a Mood.” In Modality in Grammar and Discourse, TSL32, Joan Bybee and Suzanne Fleishman (eds), 329–345. Amsterdam/ Philadelphia: John Benjamins. Paillard, M. 1993. “Les chemins de la concession : quelques contrastes entre anglais et français.” In Subordination. Travaux Linguistiques du CERLICO 6, C. Muller & D. Roulland (eds.), 207–226. Rennes: Presses Universitaires de Rennes 2. Ranger, G. 1997. “An Enunciative Study of the Rectifying Concessive Constructions not that, except and only”, Anglophonia 2, 105–125. Toulouse: Presses Universitaires du Mirail. Ranger, G. 1998. Les constructions concessives en anglais: une approche énonciative, Cahiers de recherche, numéro spécial. Gap/Paris: Ophrys. Souesme, J.-C. 1989. “DO, deux valeurs, une fonction.” In Exploration en linguistique anglaise, A. Gauthier (ed), 91–151. Bern: Peter Lang.

TSL[v.20020404] Prn:26/03/2007; 15:08

 

In search of operations

F: PB161P4.tex / p.1 (153)

TSL[v.20020404] Prn:10/05/2007; 10:31

F: PB16110.tex / p.1 (155)

‘He’s a cop but he isn’t a bastard’ An enunciative approach to some pragmatic effects of the coordinator but Martine Sekali University of Paris X, Nanterre

This paper uses the Theory of Enunciative Operations and suggests new concepts for a linguistic analysis of ‘pragmatic’ effects such as the construction, through the use of grammatical markers, and connectives in particular, of indirect meaning and argumentative force. Defining a three-term relation, the coordinator but is considered here as a counter-subordinator which both creates an argumentative domain attributed by the speaker to a previous utterance and corrects it in an inter-personal relation of discordance with the addressee. We propose a detailed contextualized analysis of the construction and retrieval of implicit utterances and argumentative power in interviews and political debates as the result of enunciative (versus extra-linguistic) operations marked by but interacting with a combination of specific markers within the utterances it links.

.

Preliminary remarks

My purpose here is to analyse the linguistic and enunciative process by which the coordinator but, interacting with the inner structures of the utterances it links, can build argumentative effects that have so far been analysed mainly with the tools of pragmatics and logic. Using the concepts developed within the Theory of Enunciative Operations (T.E.O),1 I will consider the construction of indirect meaning, or implicit utterances, as well as the positioning of the speaker with respect to mental representations which are attributed to the addressee. I will also consider the notion of ‘argumentative force’ of utterances as the result of linguistic (versus extra-linguistic) operations marked by a combination of specific markers. This paper will also raise a few questions as to the method for approaching such pragmatic values in linguistics. Indeed, argumentation being a linguistic . See Culioli 1990–1999.

TSL[v.20020404] Prn:10/05/2007; 10:31

F: PB16110.tex / p.2 (156)

 Martine Sekali

phenomenon, (unless of course one comes to blows..), how can we provide a useful description of the linguistic operations involved in argumentative phenomena, and particularly, of the subjective relations (or inter-personal positions) which are constructed through the use of specific markers? Argumentative (or qualitative) connection has been widely studied by linguists interested in the logical ordering and semantic presuppositions of connected notions. It has also been studied within the framework of pragmatics, where the extra-linguistic context is a major parameter for the analysis of speech acts and indirect meaning.2 So it is worth noting that what, originally, is a properly linguistic phenomenon, appears to be very difficult to apprehend with intra-linguistic tools. One of the reasons for this difficulty is the fact that, through argumentative subordination or coordination, an utterance is given a power it does not necessarily have on its own. When using argumentative connectives, the speaker asserts that one of the utterances has the power to validate or invalidate the other one, or more precisely, that the subjective endorsement of an utterance enables or, on the contrary, prevents the endorsement of the other one. Another question is the level of the utterance on which the connective operates. When dealing with argumentative connection, we must consider that the connective does not only link predicates, or arguments, but whole utterances, the dictum and the modus, therefore complex notional domains.3 In that respect, it is reasonable to say that argumentative connectives not only mark a logico-semantic link between notions, they construct a double set of relations. First of all, a subjective (and inter-subjective) relation is defined on the endorsement of utterances (therefore an operation of subjective anchoring, or subjective location): who accepts the utterance as true and do speaker and addressee agree or disagree on this truth? It is important to note here that I will use the terms speaker and addressee with the theoretical status of the French ‘énonciateur’ and ‘coénonciateur’, i.e. as origins of endorsement and modality in utterances. In the sense of ‘énonciateur’, the speaker is to be understood as an abstract linguistic entity, a subjective origin, and not a ‘locuteur’ actually producing speech, except of course in a situation of dialogue. Likewise, the addressee is to be considered as a linguistic construct, a product of discourse rather than a discourse producer, although it can become an interlocutor (or co-speaker) in a dialogical context. The addressee is therefore to be understood as a representation of the other in speech. As a subjective and linguistic counterpart to the original speaker (S0 ), the addressee (S1 ) will act as an anchoring point for the endorsement of representations that S0 rejects or concedes, according to the connective used. . See Anscombre & Ducrot 1983; Ducrot 1980; Anscombre 1995. . For a technical definition of the concept of ‘notional domain’, see Culioli 1999, Vol. 3.

TSL[v.20020404] Prn:10/05/2007; 10:31

F: PB16110.tex / p.3 (157)

‘He’s a cop but he isn’t a bastard’

Then a relation between the connected utterances is also marked, with the location of one utterance relative to the other. My aim here is to try and identify the enunciative operations whereby such inter-subjective and inter-propositional relations are structured by the operator but. With the coordinator but, we are dealing with qualitative (as opposed to quantitative) connection, where the connected utterance does not provide a quantitative (i.e. spatio-temporal) location of the validation of the previous one, but operates on its notional properties and subjective endorsement. I propose to examine the construction of referential values and subjective positions in compound utterances connected by but, with a particular focus on the linguistic structuring of indirect mental representations, or implicit meaning. The examples are taken from various political debates or speeches, as well as radio interviews and a few dialogues taken from novels.

. But-coordination as a three-term relation: The linguistic construction of an intermediary representation I hypothesize here that indirect meanings or implicatures are not encoded in the utterances prior to their connection, but are linguistically constructed through the association of the enunciative operations marked by but and the structuring operations within the connected utterances. (1) “What’s wrong with him?” “He’s drunk.” “But it’s only two o’clock.” “He’s been drunk for a long time.” (John Grisham, The Runaway Jury, p. 156)

In this dialogue, the first assertion (P) “he’s drunk” refers to a simple representation, but its coordination by the interlocutor to the second utterance “it’s only two o’clock” marks a second operation. The connective directs us back to a set of properties that could be associated by the addressee to this representation, i.e. to a series of secondary representations for which the first utterance “he’s drunk” could serve as an anchoring point, and therefore could help the addressee endorse and validate. The mere presence of but creates a link between a simple representation and a series of secondary ones, or between P and its notional domain. It also locates those representations relative to the enunciators in such a way that the speaker of but marks a relation of subjective discordance with the addressee as regards the endorsement of one of those secondary representations. The utterance (Q) here introduced by “but it’s only two o’clock”, will then develop the content (dictum and modus) of one particular representation on which this inter-subjective rela-



TSL[v.20020404] Prn:10/05/2007; 10:31



F: PB16110.tex / p.4 (158)

Martine Sekali

tion of discordance will bear. The domain of reference allotted to P by the use of the connective is thus narrowed down by the utterance of Q for the purpose of enunciative rectification. Through the connection, an indirect meaning (which I shall call i, as in implicit), is built, and can be interpreted so that ‘Q is an occurrence of representation which is exterior to the domain of secondary representations associable with P’. Q therefore both reveals and rejects i: ‘2 o’clock in the afternoon is no time to be drunk’, or ‘what you’re saying is impossible, abnormal, surprising for me, because Q’. The connective but marks that, according to its speaker (S0 ) (or original enunciator, which serves as the origin for all subjective, qualitative constructions in the field of person, aspect and modality), Q does not belong to the class of occurrences of the notion P, or to its set of properties. It is therefore an intermediate mental representation, or referential value (‘one can be drunk at any time of the day’) which is the target of a qualification, (or rather, with but, of a disqualification), and this intermediate implicit reference (located relative to the addressee) is made explicit and accessible through the coordinated utterance (Q). It is worth noting that the connection marked by but thus defines a threeterm relation rather than a binary one. The connective operates what could be called a form of counter-subordination (‘don’t think that since P then i, for Q) and the propositional content of i is linguistically structured as a mirror image (or reversed value) of Q. These enunciative operations are also exemplified in (2): (2) He’s a cop but he isn’t a bastard.

In this example, but Q ‘but he isn’t a bastard’ reveals an implicit relation of subordination between P ‘he’s a cop’ and i ‘a cop is a bastard’, where P has the power to validate i (‘since he is a cop, then he must be a bastard’). The speaker (S0 ) implies that, for the addressee (S1 , his subjective counterpart), ‘be a bastard’ belongs to the set of properties associated to the notion ‘be a cop’, and rejects this representation, using the subject ‘he’ as an example. This implicit subordination (located relative to the addressee) is thus both created and disqualified though the use of the connective but. But Q in (2) enables the speaker to create an implicit mental representation such as i = Q’ (or i is a reversed representation of Q). What is striking is that the implicit utterance will be exactly the opposite if one changes the modus in Q: (2’) He’s a cop, but he is a bastard

implies this time that ‘cops are usually not bastards’. Furthermore, since Q is determined by the speaker as being the case, the intermediate representation is disqualified: i is not an occurrence of the notional domain of P.

TSL[v.20020404] Prn:10/05/2007; 10:31

F: PB16110.tex / p.5 (159)

‘He’s a cop but he isn’t a bastard’ 

The construction of reference (and inference) is an important issue here, and in the case of connectives such as but, the distinction between conventional implicature and conversational implicature seems artificial. The speaker’s meaning intentions are understood and retrieved because they are linguistically marked, both by the connective and by the semantic and grammatical structure of the connected propositions. Implicit reference and implicatures are clearly built through a complex linguistic process involving back-reference processing and modal positioning. It should be stressed here that not only does the connective but operate a reconfiguration of the notional domain, it also acts as a modal. Through the use of but, the implicit utterance i is the target of an inter-subjective modality of discordance; i is ascribed to an addressee serving as an anchor point, or medium, for this modality of discordance. The addressee, as a linguistic representation of otherness, can also take on the role of an alter ego, and, as a springboard for denied implicit representations, it will help us retrieve the speaker’s own mental representations. In those cases, but Q will often reveal the speaker’s subconscious, as Freud’s concept of ‘denial’ explains (‘I dreamt of a woman, but she wasn’t my mother’): by refuting an anticipated interpretation attributed to the psychologist, the speaker reveals his own mental representations. Coordination with but is also very interesting to analyse in fiction for its role in the linguistic construction of characters through their speech. In (3) for example: (3) Dr Herder: His lordship is a paranoid schizophrenic. Sir Charles: But he’s a Gurney ! Dr Herder: Then he’s a paranoid schizophrenic Gurney who believes he’s God. Sir Charles: But we’ve always been Church of England! (Peter Barnes, The Ruling Class, I.3)

the character of Sir Charles is partly built through his use of but in his replies : ‘But he’s a Gurney’, following our former analyses, builds an implicit intermediary representation such as ‘be a Gurney’ cannot be an occurrence of the notional domain associated with ‘be a paranoid schizophrenic’ and vice versa. At this stage I would like to go back on the linguistic definitions of some of the concepts I have used so far, such as ‘secondary representations associable with P’, ‘notional domain of P’, ‘set of properties of the notion P’ etc., considering that the connective marks a secondary operation on a predication P which remains validated and endorsed by the speaker. I would say that argumentative (or qualitative) connection, whatever connective is used, defines one of the utterances as the anchoring point for the subjective endorsement of the utterance to which it is linked. Two locating operations are therefore marked:

TSL[v.20020404] Prn:10/05/2007; 10:31

F: PB16110.tex / p.6 (160)

 Martine Sekali

– the location of one predication relative to the other one – the location of the connected utterances relative to the subjective coordinates (S0 , S1 etc.) of the situation of utterance. Within argumentative connection, as soon as an utterance is connected with another one, it becomes a kind of macro-notion and is allotted a structured notional domain which I choose to call its argumentative domain. The notional properties of the argumentative domain of such an utterance are to be considered as a bundle or family of other utterances which, from the subjective viewpoint of the speaker, can share its modal endorsement. This family of utterances is made up of all the potential continuations of P, in other words, the implicit utterances which can be validated and endorsed by means of P. The utterance P thus becomes a potential constitutive anchoring point for this bundle of utterances. It is the connective which constructs the very presence of an implicit utterance, its semantic value then depending on the connected predications, as I will develop below. An utterance thus provided with an argumentative domain through its connection with another one will have the status of anchoring point for the modal endorsement of the other one, in a potential process of qualitative subordination. This location can build various semantic values such as cause, condition, justification etc. but also contradiction and opposition, depending on the connective used and the inner structure of the connected utterances. The connective but is particularly interesting because: –

– –

it endows P with an argumentative domain, therefore making it the potential anchoring point for the endorsement of a family of implicit utterances, which creates the existence of indirect meaning it opposes speaker to addressee with respect to the configuration of this argumentative domain, it introduces an utterance Q which both structures and disqualifies one of the implicit utterances within P’s argumentative domain.

The construction of P’s argumentative domain is thus re-configured, updated by the speaker, in a relation of inter-subjective discordance: the speakers are not ‘on the same wavelength’.

. The incidence of the inner structure of P and Q on the retrieval of the implicit utterance In the above section, I mentioned the fact that implicit meanings are linguistically constructed through the association of the enunciative operations marked by but and the structure of the connected utterances. The examples given illustrated the

TSL[v.20020404] Prn:10/05/2007; 10:31

F: PB16110.tex / p.7 (161)

‘He’s a cop but he isn’t a bastard’

case where the propositional content of i was linguistically structured as a mirror image (or reversed value) of Q. In fact, my observation of a large corpus of coordinate clauses connected by but revealed two main forms structuring the implicit utterance i: a. a form in which Q takes up the grammatical structure of P with a change of one of its lexical entries or with a different modality (Q = P’) b. a form in which Q does not directly take up P but structures a reversed value of i (i = Q’ or Q = i’) It is necessary now to go into more detail and analyse the structuring process of implicit utterances according to the form involved, with a close observation of more contextualized examples. . Form 1: Q = P’ (4) (about the war in Iraq) And now we’re fighting them now. And it’s hard work. I understand how hard it is. I get the casualty reports every day. I see on the TV screens how hard it is. But it’s necessary work. (G. Bush, First Presidential debate, 30/09/2004)

This rather characteristic example presents a similar grammatical structure in P and Q: it be + adj + work, but with a different instantiation of the adjective. Hard and necessary are not semantically opposed: the connective but defines them as opposed with respect to an argumentative project, a secondary meaning. The change of adjectives in Q induces a re-configuration (or re-formatting) of P’s argumentative domain, with the exclusion of one of its potential continuations or implicit utterances: here the predication of a wish to stop fighting in Iraq. The speaker G. Bush ascribes this implicit predication to the addressee (the viewer/his opponent), who serves as a medium for a modality of discordance. For G. Bush, the validated utterance ‘it is hard work’ does not have a status of locator for the subjective endorsement of such an implicit predication as ‘we must stop fighting’. Clearly, in this configuration (Q = P’), and unlike the second one (Q = i’), the exact semantic and grammatical structure of the implicit predication is not always retrievable, but the very presence of such a secondary indirect meaning is definitely made obvious by the use of the connective. It is important to add, and we will expand further on this phenomenon as we go along, that the inter-subjective relation of discordance marked by but also defines an argumentative hierarchy: the speaker G. Bush posits that difficulty is not a sufficient argument for stopping the war; necessity is defined as a stronger argument against this conclusion. G. Bush here appropriates such values as courage in



TSL[v.20020404] Prn:10/05/2007; 10:31

F: PB16110.tex / p.8 (162)

 Martine Sekali

face of adversity (cf. the pronoun ‘we’), even though he only shares the soldiers’ difficulty ‘on the TV screen’. In the process, the problem of the loss of human lives, previously mentioned in the debate by J. Kerry, is rejected by G. Bush as a secondary issue. In (5) below, the compound utterance delivered by J. Kerry presents the same form where Q takes up the structure of P: (5) Civilians get onto aircraft, and their luggage is X- rayed, but the cargo hold is not X-rayed. Does that make you feel safer in America? (J. Kerry, First Presidential debate, 30/09/2004)

This time the instantiation of the subject is modified (‘the cargo hold’ versus ‘their luggage’) and the assertive modality is reversed (negative versus positive). As a reversed formulation of P, Q re-configures P’s argumentative domain, so that P loses its status of potential locator for the subjective endorsement of an implicit predication, i.e. here the assertion of civilians’ security. This disqualified implicit utterance (i) is immediately retrievable, and confirmed by the question that follows: ‘Does that make you feel safer in America?’. The connective but posits J. Kerry in a relation of discordance relative to his addressee as regards the nature of P’s argumentative domain, then marks the exclusion of one of its potential properties through Q taking up and modifying its instantiation. The same process can be found in the following example, which illustrates the importance of the interaction of combined grammatical markers in the linguistic construction of semantic values: (6) Now we’re told by some that law and order shouldn’t be made an election issue. But it’s not we politicians who make it an election issue. It’s you the people. (M. Thatcher, Birmingham Speech 19/04/79)

Let us first note that the addressees to whom M. Thatcher attributes false implicit predications are here explicitly designated (though with indefinite reference) by the pronoun ‘some’. The utterance P takes the form of indirect speech expressing regret concerning a state of fact (‘should not be’). The connected utterance but Q introduces the refutation by M. Thatcher of an implicit accusation defining the Conservative Party as responsible for the choice of insecurity as a main campaign issue. Here again Q takes up the passive structure of P (with no agent expressed for the predicate ‘make’) but in a negative cleft sentence. This cleavage operation retrieves the agent of ‘make’ and marks a contrast on the subject pronoun (‘We’ ⇒ ‘you the people’). The implicit accusation is made explicit for rectification. The enunciative operations marked by but (subjective discordance and disqualification of implicit predications through a re-configuration of P’s argumentative domain) are best analysed in large discourse extracts, where the role of the connective in the progression of discourse appears decisive. In the following ex-

TSL[v.20020404] Prn:10/05/2007; 10:31

F: PB16110.tex / p.9 (163)

‘He’s a cop but he isn’t a bastard’ 

ample (7), taken from John Grisham’s novel The Runaway Jury, a medical expert is cross-examined during a trial against the tobacco industry. The argumentative stake for the defence is not an easy one: the idea is to exculpate the tobacco industry by demonstrating, not that tobacco has no effect on the smokers’ health, but that it is impossible to prove that tobacco is directly responsible for lung cancer. (7) “Smoking greatly increases the risk of lung cancer.” “But it doesn’t cause it every time, does it?”

The connective but instantly marks an inter-subjective relation of discordance between the defendant and the expert. Yet the discordance does not concern the actual validation and endorsement of the expert’s utterance P ‘Smoking greatly increases the risk of lung cancer’. This predication is considered as true by both speakers. But its argumentative power, that is, its capacity to validate other potential predications, is refuted by the defendant. But introduces an utterance (Q), which takes up P with a different verb and modality. The same subject ‘smoking’ is referred to by the pronoun ‘it’, but ‘increase’ becomes ‘cause’ and ‘greatly’ becomes ‘every time’. Through this modification, the issue is diverted from a comparative evaluation of risks to a strict and systematic determination of causality, i.e. whether or not the subject ‘smoking’ always validates the predicate ‘cause cancer’. Moreover, the modality in Q is a negation with a question-tag asking for confirmation. The defendant entices the expert to endorse the negative assertion of Q, which disqualifies one of the implicit utterances in P’s argumentative domain (i): ‘Smoking causes lung cancer’. Here again, the endorsement of Q as a reconfiguration of P both reveals and disqualifies this implicit representation. P can no longer be the locator for the subjective endorsement of i. The expert’s answer satisfies the defence counsel in so far as it confirms the disqualification of an implicit meaning and even makes it explicit: “No. Not every smoker gets lung cancer.” “Thank you.”

The exchange carries on with a series of but connectives, which, on the one hand, maintain the relation of discordance on the speakers’ secondary, indirect representations, and, on the other hand, create a to and fro motion on the definition of the argumentative domain of the initial utterance. A linguistic battle starts where each speaker tries to dominate the progression of discourse and have the last word: “But for those who smoke, the risk of lung cancer is much greater.” “Do you admit that air pollution causes lung cancer?” “It might. But I stand by my research. Rural smokers get lung cancer more than rural non-smokers, and urban smokers get cancer more than urban nonsmokers”.

TSL[v.20020404] Prn:10/05/2007; 10:31

F: PB16110.tex / p.10 (164)

 Martine Sekali

The expert uses the connective but so as to deviate from the implicit predication of non-systematic responsibility of tobacco. He thus re-introduces his initial predication P, reformulating his comparative evaluation through the words ‘risk’ and ‘greater’, which re-establish a potential responsibility of tobacco on the development of cancer. The defendant’s question: ‘Do you admit that air pollution causes lung cancer?’ is quite skilful, as it shifts the argumentative issue from the instantiation of the predicate (‘cause’ versus ‘increase the risk of cancer’) to the instantiation of the subject for the predicate ‘cause lung cancer’: ‘air pollution’ replaces ‘smoking’, with an interrogative modality. The question of risk is thus discarded, the issue becoming the strict identification of a culprit. The linguistic ploy is obvious: if the expert endorses the assertion ‘air pollution causes lung cancer’, then the subject ‘smoking’ is no longer alone in validating the predicate ‘cause cancer’, and the tobacco industry can no longer be proved guilty. The expert is trapped. He uses but again trying to disentangle himself from the defendant’s implicit representations, and to recover his initial utterance P, whose argumentative domain has been completely re-formatted. But the battle is lost. It can also be noted that in answering ‘it might’, the expert falls even more into the trap. The modal verb marking a potential truth, a risk that pollution (in the same way as tobacco) may cause cancer, then the prosecutor could just as well incriminate any polluting industry. This exchange illustrates the complex process by which the connective but can help structure indirect representations. Followed by an utterance Q which takes up and reorients the initial utterance P (‘smoking increases the risk of cancer’), but defines an inter-subjective relation of discordance between speakers on P’s capacity to be the anchoring point for the endorsement of an implicit predication. But Q = P’ here invalidates the potential endorsement by the addressee of an implicit predication i: ‘Tobacco is the only cause of lung cancer’. Although this predication has never been uttered directly, the combined operations described above have made it retrievable. The same analysis could be made of the rest of the cross examination, which follows the same process: (8) “I read that report”, Dr Kilvan said. “Do you have an opinion on it?” “No, heredity is not my specialty.” “So you can’t say yes or no on the issue of whether heredity might be related to smoking and lung cancer.” “I cannot”. “But you don’t contest this report, do you? (John Grisham, The Runaway Jury, p. 204)

TSL[v.20020404] Prn:10/05/2007; 10:31

F: PB16110.tex / p.11 (165)

‘He’s a cop but he isn’t a bastard’ 

By modifying the structure of P, the speaker corrects his interlocutor regarding P’s argumentative domain. In doing so, the speaker overpowers his interlocutor and controls the discourse continuity. . Form 2: Q = i’ or i = Q’: i is a reversed representation of Q This configuration has already been mentioned above through the analysis of examples (1) to (3). The construction and retrieval of implicit representations in this case is made easier by the fact that, contrary to Form 1, utterance Q does not necessarily take up the structure of P, but produces a mirror (or reversed) image of the propositional content and modal endorsement of the implicit utterance itself. Illustrations of such uses appear in examples (9) and (10) below, for which I propose a close analysis: (9) As soon as possible after the election, a Conservative Government will restore service pay to the full amount recommended by the Armed Services Pay Review Board. I know this will increase expenditure, I don’t deny it. But there’ll be no home policy at all unless we are properly defended because there’ll be no home. You can’t prosper unless you survive. (M. Thatcher, Birmingham Speech 19/04/79)

The utterance P (‘it will increase expenditure’) here contains the modal will, which defines the extra expenditure as inherent to a prospective increase of service pay. The full endorsement of the assertion and its modality by S0 is confirmed by ‘I don’t deny it’. The utterance Q following the connective but is no contradiction or opposition to this previous assertion P, but introduces a counter-argument to the conclusion M. Thatcher’s detractors could draw from P. Using but, she announces a counter-subordination and a subjective discordance on P’s argumentative domain. One of the potential occurrences of P’s domain (and therefore potential conclusions drawn by detractors) is made explicit by the utterance Q: ‘there’ll be no home policy at all unless we are properly defended because there’ll be no home’, where unless marks a conditional relation between the existence of an efficient national army and the existence of such things as ‘home’ and a home policy. The implicit utterance which is disqualified here can therefore be retrieved by reversing this conditional relation in Q: more money for defence means less for home policy. Hence in Form 2, the connective but, inter-acting with the linguistic operations within P and Q, builds an implicit utterance where i is a reversed representation of Q (Q = i’ or i = Q’). This linguistic pattern is commonly used in political campaign speeches because it enables politicians to manipulate their detractors, to whom they ascribe implicit assertions so as to better disqualify them and introduce their own claims. More than a simple coordinating device, but can therefore be considered on the level of inter-propositional and inter-subjective re-

TSL[v.20020404] Prn:10/05/2007; 10:31

F: PB16110.tex / p.12 (166)

 Martine Sekali

lations as a counter-subordinator, operating a three-term relation of qualitative locations: P is defined as the anchoring point for the qualitative endorsement of i by the addressee and at the same time this relation is disqualified by the speaker through Q. In the following address to G. Bush and the viewers, J. Kerry provides us with a real gloss of this three-term relation of qualitative locations and countersubordination: (10) We have to succeed. We can’t leave a failed Iraq [assertion of P]. But [subjective discordance on a potential P/i subordination] that doesn’t mean it wasn’t a mistake of judgment to go there and take the focus off of Osama bin Laden. It was! [explicit formulation and disqualification of i through the assertion of Q]. (J. Kerry, First Presidential debate, 30/09/2004)

The linguistic construction of subjective relations is essential here. But invalidates i for being out of P’s argumentative domain and irrelevant. In doing so, it also dismisses mental representations that are attributed to the addressee, and reasserts the speaker’s authority in the discourse progression. The co-speaker’s representations may sometimes be conceded, but they are always presented as beside the point, irrelevant to the on-going discussion, or simply invalid.

. But and the notion of ‘argumentative force’ As illustrated in example (10) above, the enunciative operations described for the connective but are closely linked with another argumentative phenomenon, which has always been analysed within the theoretical framework of pragmatics: i.e. what Ducrot and Anscombre called ‘argumentative force’.4 The connective but clearly introduces a hierarchy in the endorsement of discourse: the utterance Q, which is endorsed by the speaker S0 as pre-eminent, operates a qualitative and argumentative boost. Using but, the speaker is setting himself in command of the discursive exchange and taking control of its progression and thematic direction. . Inter-subjective relations and discourse orientation The inter-subjective relation of discordance marked by the connective but is very often used in interviews, dialogues or debates precisely because it enables the speaker to dissociate himself from his addressee, divert and re-route the conver. See Anscombre & Ducrot 1983 and also Ducrot 1980.

TSL[v.20020404] Prn:10/05/2007; 10:31

F: PB16110.tex / p.13 (167)

‘He’s a cop but he isn’t a bastard’ 

sation, cut the co-speaker short or go back to a former issue. In short, but gives the speaker the power to appropriate the orientation of discourse. The 2004 first presidential television debate between G. Bush and J. Kerry abounds in such uses of the connective but: (11) J. Lehrer: We’ll come back to Iraq in a moment. But I want to come back to where I began, on homeland security. This is a two-minute new question, Senator Kerry. As president, what would you do, specifically, in addition to or differently to increase the homeland security of the United States than what President Bush is doing? (First Presidential debate, 30/09/2004) (12) J. Lehrer: New question, two minutes. Senator Kerry, you mentioned Darfur, the Darfur region of Sudan. Fifty thousand people have already died in that area. More than a million are homeless. And it’s been labeled an act of ongoing genocide. Yet neither one of you or anyone else connected with your campaigns or your administration that I can find has discussed the possibility of sending in troops. Why not? J. Kerry: Well, I’ll tell you exactly why not, but I first want to say something about those sanctions on Iran. (First Presidential debate, 30/09/2004) (13) I’m not going to talk about a difference of character. I don’t think that’s my job or my business. But let me talk about something that the president just sort of finished up with. (J. Kerry, First Presidential debate, 30/09/2004)

In the examples above the connective but is always used to set up a subjective discordance between speakers on the continuation of discourse and the choice of topics: the speaker of but breaks with the co-speaker’s direction and imposes his own. In the following example, it is the debate moderator, Jim Lehrer, who intervenes in the debate so as to make G. Bush react to one of J. Kerry’s previous utterances. It must be noted that changing the argumentative direction of an utterance in such a way means going counter to an explicit or implicit potential continuation, so that this operation of subjective discordance and discourse disruption also contributes to the linguistic creation of indirect meaning: (14) J. Kerry: Just because the president says it can’t be done, that you’d lose China, doesn’t mean it can’t be done. I mean, this is the president who said “There were weapons of mass destruction,” said “Mission accomplished,” said we could fight the war on the cheap - none of which were true. J. Lehrer: Thirty seconds, Mr. President. G. Bush: You know my opinion on North Korea. I can’t say it any more plainly. J. Lehrer: Well, but when he used the word “truth” again... G. Bush: Pardon me? J. Lehrer: ... talking about the truth of the matter. He used the word “truth”

TSL[v.20020404] Prn:10/05/2007; 10:31

F: PB16110.tex / p.14 (168)

 Martine Sekali

again. Did that raise any hackles with you? G. Bush: Oh, I’m a pretty calm guy. I don’t take it personally. (First Presidential debate, 30/09/2004)

In his first answer to J. Kerry, G. Bush clearly refuses to discuss the issue of North Korea, and completely ignores his opponent’s accusations of untruthfulness. J. Lehrer, expecting a reaction, uses the connective but to break into the exchange and give back to J. Kerry’s utterance an argumentative power which had been denied or ignored by G. Bush. This discordant steering of the debate is introduced by but and followed with a reference to an element of J. Kerry’s utterance aimed at provoking a reaction (‘He used the word “truth” again. Did that raise any hackles with you?’). But G. Bush here maintains his course of argumentation and is not disturbed by Lehrer’s strategy. The operation of discourse switching or re-orientation marked by but is also quite clear in the following example, taken from an interview of M. Moore by the journalist A. Collins, and which I propose to analyse in its progression: (15) Michael Moore: I wouldn’t know how to organise myself to film so that B follows A and C follows B. I don’t think you should do that, especially a documentary film like this. Your feet should not be in cement. You should be open to whatever happens and go with it. It drives the budget and the producers batty because it can get costly but if you’re willing to do that you’ll end up with something that will reach a wider audience because you’ve got a much more interesting film. Andrew Collins: But documentary can be as manipulative as fiction, can’t it? If you want to make it that way, you can make the facts fit. MM: But that’s true of anything. AC: But you’re led by the subject rather than the other way around. Some documentary film-makers just make the films that they want to make, they know what they want to make before they start. (Guardian interview of M. Moore by Andrew Collins, 11/11/2002)

The first use of the connective but in M. Moore’s speech is non-dialogical and marks that the absence of pre-established framework in documentary films is more important to him than the financial problems this openness can cause. It is noticeable that starting from this assertion, the dialogue between M. Moore and A. Collins is then built on a series of subjective dissociations and disqualifications of implicit meaning. The journalist first intervenes with a connective but so as to redirect the conversation on the ideological rather than the artistic implications of Moore’s assertion. Taking up M. Moore’s predicative relation to question its validity in the particular domain of documentary films, the journalist re-formulates Moore’s utterance P in ‘But documentary can be as manipulative as fiction, can’t it? If you want to make it

TSL[v.20020404] Prn:10/05/2007; 10:31

F: PB16110.tex / p.15 (169)

‘He’s a cop but he isn’t a bastard’ 

that way, you can make the facts fit.’ (form 1: Q = P’), by modifying its modality (‘should’ becomes ‘can’) and suggesting an opposite predicate (from ‘be open’ to ‘be manipulative’). By doing so, Andrew Collins rectifies P’s domain and launches a discordant argumentative progression on the implicit qualities of ‘objective truth’ which could be associated with documentary films on account of their improvised form. The process of counter-subordination is here again exemplified: an implicit mental representation is thus structured, attributed to his co-speaker, and disqualified. The connective but, cooperating with contextual markers, constructs the presence of innuendoes as well as a subjective dissociation of the journalist from his interviewee. A. Collins’s following utterance (‘you can make the facts fit’) is a development of Q, where the pronoun ‘you’ is ambiguous and could be interpreted as generic or as referring to M. Moore specifically, thus adding to the slight provocation. Quite adequately, M. Moore also starts his answer with the connective but so as to dissociate himself from A. Collins, but uses the very same argumentative strategy. Taking up the journalist’s predicative relation ‘documentary can be as manipulative as fiction’ as valid whatever the subject (‘But that’s true of anything), M. Moore disqualifies his interviewer’s implicit (but perfectly understood) personal accusation. The journalist then tries to adjust the fire, rectifying, not what he said, but the implicit meaning he has created. In his utterance: ‘But you’re led by the subject rather than the other way around. Some documentary film-makers just make the films that they want to make’, A. Collins clearly contrasts the pronoun ‘you’ with ‘some documentary film-makers’, using but again, which could be paraphrased as ‘don’t think that my predication was aimed at you specifically, because my use of the pronoun ‘you’ was a generic one.’ The use of the connective but in interviews and debates is particularly revealing of an enunciative fight for the control of the orientation and direction of discourse. The co-speakers’ utterances are thus taken up and corrected, not in their validation itself, but in their power to validate implicit ones, and this operation of counter-subordination is the result of a double process of inter-subjective location and linguistic reformatting of argumentative domain. . Counter-subordination and argumentative force This operation of counter-subordination seems to me to be the linguistic origin of what is called in pragmatics the ‘argumentative force’ of an utterance. Through the enunciative operations marked by but, utterance Q is endowed with stronger argumentative force precisely because it introduces an element which disqualifies and overpowers a former relation, and also because it defines an inter-subjective relation of discordance on the endorsement of this former relation. I might add

TSL[v.20020404] Prn:10/05/2007; 10:31

F: PB16110.tex / p.16 (170)

 Martine Sekali

that the inter-subjective locations cannot but be taken into account as an essential parameter in the linguistic analysis of argumentative connection. The status of stronger argument which is ascribed by but to utterance Q, is quite often explicitly paraphrased in the utterance itself: (16) I shot the entire movie in continuity, but more importantly I shot the entire Omaha Beach sequence in continuity. (S. Spielberg, DGA Magazine, January 1999, interview about the making of Saving Private Ryan) (17) People are still hurting in the state of Michigan. I know that I travelled here a lot, I heard the stories. But the fundamental question is, which candidate can continue to grow this economy? And that’s George W. Bush. (President Bush’s Radio Address, 10/30/2004)

Yet these paraphrases are redundant, or emphatic, because the stronger status of the argument developed in Q is linguistically constructed through connection itself. Example (18) below is very interesting to analyse in detail in that respect: (18) Read the letter sent to me by Dr Safa Hashim, who lives here in Glasgow, and who says he is writing despite his fears of Iraqi retribution. [He says the principle of opposing war by the public is received warmly by Iraqis for it reveals the desire of people to avoid suffering. But he says it misses the point because the Iraqi people need Saddam removed as a way of ending their suffering.] (T. Blair,15/02/2003 speech on Iraq)

The interaction of such connectives as but, for and because in this utterance by an Iraqi doctor, quoted by Tony Blair in one of his pro-war speeches, again illustrates the process of counter-subordination marked by but, and also reveals the setting up of a scale of argumentative power given to the utterances. This scale is certainly not a logical one, but rather a subjective evaluation marked by specific linguistic operators. The argumentative domains structured in the utterances P and Q on either side of the connective but are developed through the use of qualitative connections introduced by for and then because. These conjunctions also mark inter-propositional as well as inter-subjective locations. The connective but will define the second location (marked by because) as superior to and disqualifying the first one (marked by for), in a complex combination of operations: P [U1 for U2]5 desire to avoid suffering (U2) locates (cf. for) the endorsement by S0 of: is received warmly (U1)

but Q [U1’ because U2’] but need to end suffering (U2’) locates (cf. because) the endorsement by S0 of: misses the point (U1’)

. U1 and U2 stand for ‘Utterance 1’ and ‘Utterance 2’.

TSL[v.20020404] Prn:10/05/2007; 10:31

F: PB16110.tex / p.17 (171)

‘He’s a cop but he isn’t a bastard’

Several remarks must be made about this compound utterance: –

–

–

–

P and Q present a parallel complex structure, Q taking up the predication expressed in P with a change of predicates (‘desire to avoid’ ⇒ ‘need to end’) but keeping the same complement ‘suffering’. We are dealing with Form 1 here (Q = P’). The qualitative connection with for within P is also interesting: as is the case in any argumentative subordination, this connective marks that the validation of U2 (‘it reveals the desire of people to avoid suffering’) is set as the anchor point, or locator, for the validation and endorsement of the predicative relation in U1 (‘the principle of opposing war by the public is received warmly by Iraqis’) U1 thus belongs to the set of properties included in U2, i.e. to its argumentative domain. But, on the level of inter-subjective relations, for marks the argumentative relation as thematic or pre-constructed, i.e. non-polemical, taken for granted and endorsed by any potential addressee. The particularity of this connection is that even though the locating argument (Q) is new and introduced by the speaker himself, its argumentative domain is at once represented as consensual. Argumentation introduced by for is thus added to U1 but not necessary to U1. In that respect, I would say that for coordinates U1 with a kind of argumentative subordination that is appositive rather than determinative, to use terms commonly applied to describe relative clauses: ‘I assert U1, and I remind you, if need be, that U2 locates U1.’ On the other hand, the argumentative subordination marked by because in Q sets the validation and subjective endorsement of U2’ as determinative for the endorsement of U1’, so that the inter-subjective relation is here completely different. The argumentative relation is defined as rhematic, the speaker introducing an argument which he endorses personally, in a potential discordance with his addressee. Contrary to for (and since, for that matter), because builds an argumentative link which is not granted in advance, but can still be discussed. In that respect, because is particularly compatible with but, which operates a preliminary discordance on P’s argumentative domain. But counters the subordination explicitly expressed by for, breaks with the consensual endorsement of U1 in P, and introduces a new argument which, as a result, gains a priority status.

. When but can build notions of explanation and condition Semantic values such as explanation or condition can sometimes be retrieved in but compound-utterances. It appears that the linguistic construction of these relational notions can also be explained by the operations described above, i.e. the



TSL[v.20020404] Prn:10/05/2007; 10:31

F: PB16110.tex / p.18 (172)

 Martine Sekali

process of counter-subordination and the organization of argumentative domains into a hierarchy. There again, the core-operation of the connective but will interact with other linguistic operations within the connected utterances in the complex process of meaning construction. In (19), for example: (19) Storywise, I stuck to written script very closely, but I had a very good written script. (S. Spielberg, DGA Magazine, January 1999, interview about the making of Saving Private Ryan)

The relation between P and Q can be paraphrased in (19) as a restrictive explanation, such as ‘I usually don’t stick to written script, but this time I did, only because I had a very good script’. The utterance Q is here set as a better anchoring point for the validation and endorsement of P, than another potential, implicit one (such as ‘I always do’). The connective but, as always, operates a counter-subordination, but in this pattern, it disqualifies i as an argument for P (and not P as an argument for i, as was the case in former examples) and posits Q as a better argument. This operation thus introduces Q as the right explanation for P, as opposed to an implicit one. The notion of restrictive explanation in such utterances connected by but originates in its double process of subjective rectification and hierarchical organization of argumentative domains. The same process explains the conditional relation between Q and P that can be retrieved in example (20): (20) Yes, let the United Nations be the way to deal with Saddam. But let the United Nations mean what it says; and do what it means. (T. Blair, 15/02/2003 speech on Iraq)

In this example, but links two pseudo-imperative predications (therefore predications which are not validated in the situation of utterance) in such a way that the second one is set as the conditional anchoring-point for the concession of the first one: ‘ok for the prospective validation of P, providing that Q is also validated’. Tony Blair here imposes Q as a prerequisite for the acceptance of the UNO as the one to deal with Saddam. This notion of ‘restrictive condition’, or ‘prerequisite’, appears clearly in another possible paraphrase of the utterance; ‘ok but first let the UN mean what it says’. This conditional relation is due to the fact that, interacting with the modality in both of the connected utterances, but defines Q as a priority locator for the endorsement of a potential acceptance of P. The connective but thus operates a shift in responsibilities: the validation and endorsement of ‘let’ no longer depend on the speakers, they are subordinate to the validation of Q.

TSL[v.20020404] Prn:10/05/2007; 10:31

F: PB16110.tex / p.19 (173)

‘He’s a cop but he isn’t a bastard’ 

It is worth noting that with but, the conditional relation is set in reverse order compared to the one built by the connective and linking an imperative predication with a back reference assertion. The famous example:6 (21) Think of baked beans and you think of Heinz

can be analysed as the fusion, through the use of the connective and, of two notional domains. The two mental representations are associated in such a way that the suspended validation of P (cf. the imperative modality) is set as the anchoring point for the validation of Q (if you think of baked beans, then you think of Heinz). Using but instead of and in (21’): (21’) Think of baked beans but think of Heinz

one realizes that it is no longer possible to take up the potential subject you in the second predication because the order of locations is reversed: this time it is utterance Q which gains the status of conditional anchoring point for the endorsement of P: ‘it’s ok for you to think of baked beans, providing they’re Heinz.’

. Conclusion As is the case with most discourse connectives, various meanings or referential values can be retrieved from the use of but. Yet considering connectives as simply polysemic does not help us understand the process by which grammatical markers can contribute to the construction of meaning, whether explicit or implicit. My aim here was to propose a contextualized analysis of but which would both help reveal its core operation and the intra-linguistic contextual parameters with which it can interact to build a final referential value. Even though the extra-linguistic context can certainly modify or influence the interpretation of an utterance, argumentative or ‘pragmatic’ phenomena can still be analysed without necessarily resorting to extra-linguistic tools, in so far as they are subjective representations built by specific linguistic operations. As regards the connective but, its argumentative power appears to be due to the enunciative operations it marks, in association with the inner-structures of the utterances it links: a. The indirect, implicit meaning that can be retrieved through its use is not common extra-linguistic knowledge or ‘topos’7 ; its presence stems from the linguistic construction of an argumentative domain which is attributed, through the connective but, to the first utterance in the compound structure (P). This . Example borrowed from Perrin 1992. . See Anscombre 1995 and Sarfati 1995

TSL[v.20020404] Prn:10/05/2007; 10:31

F: PB16110.tex / p.20 (174)

 Martine Sekali

domain can be understood as a bundle of implicit utterances defining P’s notional properties. b. But also operates a relation of inter-subjective location whereby the speaker is set in a position of discordance relative to his addressee on the endorsement of one of the implicit utterances in P’s argumentative domain. This implicit utterance is selected by the introduction of the second utterance in the compound structure (Q) and revealed by its inner structure according to two main configurations (Q = P’) and (Q = i’), but other forms can also be found, especially when but interacts with other connectives. c. The implicit representation is both revealed and denied in the process of counter-subordination and subjective discordance marked by but. The speaker of but thus appropriates the discourse continuation through a double process of subjective and propositional locations in a three-term relation, one of the terms being a linguistically constructed implicit one.

References Anscombre, J. C. 1995. Théorie des topoï, Paris: Kimé. Anscombre, J. C., Ducrot, O. 1983. L’argumentation dans la langue. Bruxelles: Pierre Mardaga. Bouscaren, J., Chuquet, J., Danon-Boileau, L. 1992. Introduction to a Linguistic Grammar of English. An Utterer-centred Approach. Gap: Ophrys. Culioli, A. 1990, 1999. Pour une linguistique de l’énonciation. Vol.1: Opérations et représentations; Vol. 2: Formalisation et opérations de repérage; Vol. 3: Domaine notionnel. Gap: Ophrys. Ducrot, O. 1980. Les échelles argumentatives. Paris: Minuit. Ducrot, O. 1980 Les mots du discours. Paris: Minuit. Gournay, L. 2002. “Nothing but but. Opération sémantique et caractérisation syntaxique”. Travaux du Cerlico 15. Presses Universitaires de Rennes. Groussier, M.L. & Rivière, C. (Eds.) 1997. La Notion, actes du colloque La Notion, HDL, Gap: Ophrys. Liddle, Michael (ed.). 1995. Antoine Culioli. Cognition and Representation in Linguistic Theory. C.I.L.T. Vol. 112. Amsterdam: John Benjamins. Perrin, I. 1992. “Think of baked beans and you think of Heinz”: les impératifs dits “conditionnels” in TELOS 1: Approches énonciatives de l’énoncé complexe, Louvain: Peeters. Sekali, M. 1991. “Connexion inter-énoncés et relations intersubjectives: because, since et for.” Langages n◦ 104. Sekali-DeCola, M. 1992. “Intersubjectivité et implicite par la connexion inter-énoncés: une étude du connecteur anglais BUT.” in TELOS 1: Approches énonciatives de l’énoncé complexe, Louvain: Peeters. Sekali, M. 1992. “Subordination temporelle et subordination subjective: l’exemple de since.” Travaux du CERLICO n◦ 5, Subordination/Subordinations. Presses Universitaires de Rennes. Sarfati, G. E. 1995. “Pragmatique et sens commun, note pour une pragmatique topique.” Mélanges David Gaatone, Amsterdam: Benjamins, Linguisticae Investigationes, 363–370.

TSL[v.20020404] Prn:10/05/2007; 10:31

F: PB16110.tex / p.21 (175)

‘He’s a cop but he isn’t a bastard’

Simonin, J. 1975. “Pour une typologie des discours.” In J. Kristeva et al. eds. Langue, discours, société: Pour Emile Benveniste, Paris: Editions du Seuil. 85–121. Simonin, J. 1984. “De la nécessité de distinguer énonciateur et locuteur dans une théorie énonciative.” In Revue du DRLAV n◦ 30, Paris VIII – Vincennes. 55–62 Yaguello, M. (ed.). 1994. Subjecthood and Subjectivity. The Status of the Subject in Linguistic Theory, HDL: Ophrys.



TSL[v.20020404] Prn:10/05/2007; 13:45

F: PB16111.tex / p.1 (177)

Continuity and discontinuity in discourse Notes on yet and still Graham Ranger ICTT, Université d’Avignon et des Pays de Vaucluse

yet and still each have three clearly differentiated uses relating to considerations of aspect, degree and argumentation. This paper provides a description of yet and still showing how these uses are linked. Within the Theory of Enunciative Operations I provide a representation of schematic forms for yet and still enabling us to derive local values as functions of the notional domains involved. yet marks the construction of an offline position in opposition with a preconstructed position in non p. still constructs an occurrence of a notion p identified to and continuous with a previous occurrence. These hypotheses are supported by close analysis of genuine utterances. Aspectual, quantifying or argumentative uses of yet and still derive from the properties of the domains concerned.

Introduction The problems involved in comparing and describing the markers yet and still are common enough in linguistic analysis and concern essentially the irregular match between forms and values. Firstly, the forms yet and still appear to overlap partially in that it is often, but not always, possible to use one of the two markers to reformulate an utterance including the other apparently without engendering any huge changes in the resulting values. And so the following pairs obviously construct similar and locally identifiable (if not identical) representations: (1) I haven’t eaten yet. / I still haven’t eaten. (2) There is yet / still more money to come in. (3) Yet / Still there’s a shroud of secrecy surrounding that report.

Secondly, each of the two forms is liable to receive a number of clearly differentiable interpretations, or values, which we might, for ease of presentation, refer to

TSL[v.20020404] Prn:10/05/2007; 13:45

F: PB16111.tex / p.2 (178)

 Graham Ranger

as aspectual, focal and argumentative. These are illustrated respectively for yet and still by the examples below: (4) I haven’t eaten yet. (5) [. . .] are you still Vice Chair Person? (6) They’re probably away on yet another holiday (7) Or easier still I’ll tell you what. (8) Yet there’s a shroud of secrecy surrounding that report (9) Still it hasn’t stopped me yet.1

The present paper aims to provide a description of the connectives yet and still which accounts for the similarities between the two markers while also explaining how it is that each marker may be used in a variety of different ways. More specifically I aim to show, within the framework of the Theory of Enunciative Operations, how it is possible to construct an abstract representation of schematic forms for yet and still which allow us to derive the local values illustrated above as functions of the notional domains involved.2

.

Theoretical preliminaries

I will be using the concept of the notional domain as elaborated by Culioli and his collaborators over the last forty years or so.3 We might consider the notional domain as a powerful topological template for the metalinguistic representation of operations of categorisation whereby phenomenological occurrences are located relative to linguistic representations. I will very briefly illustrate how things function using the following text: (10) The path from the celebrated singer to the respected maestro – a walk on a tightrope? “At the beginning it was all very complicated. Conducting is per se difficult, but practice makes it easier. Naturally, I always had the respective orchestra and audience on my side. But the critics were very sceptical to begin with. They didn’t understand why I, a singer, should suddenly want to lead an orchestra. But slowly their attitude changed and now, fortunately, it is . From this point on, examples will, unless otherwise mentioned, be taken from the BNC Sampler corpus. . The following text is, to a large extent, a development of work on concessives in Ranger (1999). . Core texts in English are Culioli (1990a), (1990b), and (1995). The introduction by Bouscaren, Chuquet and Danon-Boileau (1992) is also very useful.

TSL[v.20020404] Prn:10/05/2007; 13:45

F: PB16111.tex / p.3 (179)

Continuity and discontinuity in discourse 

quite different. The scepticism has gone. Recently, for instance, I conducted the Chicago Symphony Orchestra, not at the opera but in the concert hall, with a symphonic repertoire. The members of the orchestra were happy after the performance. The audience was happy after the performance. And so it seems that it dawned on the critics that just because I am a singer it doesn’t mean that I cannot conduct an orchestra. I am not really a singer. I am not really a conductor, either. I am basically, and foremost, a musician. That’s what counts.”4

Here the tenor Plácido Domingo elaborates a discourse in which he situates himself and his own activities singing and conducting relative to the notional domain of /singer/. This engages typical and non typical representations of what a singer is, what being a singer implies, what a conductor is etc. in a fairly complex procedure whereby a subject constructs referential values. This can be represented with some success on a notional domain potentially organised into four areas: an Interior, corresponding to notions of typicality, an Exterior, corresponding to an operation of differentiation relative to the Interior, a Centre, engaging typicality as well as the high degree, and a Boundary area, compatible with both the Interior and the Exterior. This may be represented diagrammatically as follows:5 Interior

Exterior

Boundary

Centre

Figure 1.

In addition to these four positions on the notional domain, it is often useful to construct a fifth, off-line position, situated on a different plane from the notional domain above, from which both the Interior and Exterior positions are accessible. This model, particularly useful when dealing with questions of modality, interrogatives etc., may be represented using a branching path as follows: I

E

IE

Figure 2. . http://www.tenorissimo.com/domingo/Articles/buehne7800.htm . This representation is based on, for example, that in Culioli (1995: 65).

TSL[v.20020404] Prn:10/05/2007; 13:45

F: PB16111.tex / p.4 (180)

 Graham Ranger

Hence an interrogative such as Are you (or are you not) a singer? marks an operation whereby a speaker situates himself in the off-position IE, envisaging both I and E as possibilities and calling upon a cospeaker to help him opt for one or the other. The use of figures to represent the notional domain, here and further on, serves a double purpose. On the one hand, it is hoped that the figures make the concepts intuitively easier to grasp. On the other, a graphical representation involves a number of quasi topological constraints which oblige us to maintain conceptual coherence more thoroughly than a purely discursive representation can. Both these metalinguistic representations will serve us in the following discussion of yet and still.

. Case studies . Hypotheses Given the limitations of space, I would like to present the hypotheses to account for the various uses of yet and still now, before going on to look at how these hypotheses work on genuine examples: – – –

yet marks an operation whereby a speaker constructs an offline position IE in opposition with a preconstructed Exterior position E; still constructs an occurrence identified to and continuous with a preconstructed occurrence on the same domain; These operations are associated with all uses of yet or still. Aspectual, focal or argumentative values can be shown to derive from the properties of the notional domains concerned (relating, among other things, to the class of instants, degree and subjective localisation).

The terms construction and preconstruction are used semi-technically. Informally we can say that a representation which a speaker constructs is one which is presented as new, whereas a representation which a speaker preconstructs is presented as existing prior to utterance. I would like to begin by illustrating these hypotheses on examples of aspectuomodal determination before extending the analysis to look briefly at focal uses and at more length at argumentative uses. . Aspectuo-modal uses In studying the various aspectuo-modal uses of still and yet the figures above presenting the notional domain are articulated in relation to an abstract ordered class of instants.

TSL[v.20020404] Prn:10/05/2007; 13:45

F: PB16111.tex / p.5 (181)

Continuity and discontinuity in discourse

Let us begin with the following classic, but nonetheless authentic, example of an utterance featuring yet. (4) I haven’t eaten yet.

We shall represent the process eat as a bounded interval on the class of instants, ordered from left to right conventionally to represent the passage of time. Let us consider this interval as the Interior of a composite notional domain.6 We can then consider that the Exterior corresponds to the area which follows on from the Interior and is, crucially, determined by the fact of not being the Interior. An utterance like I have (already) eaten situates a subject on the Exterior and, since this is determined relative to the Interior, we can easily derive the well known values of contiguity, consequence, resulting state etc. for the present perfect.

t

Interior eat

Exterior not/ no longer eat

Figure 3.

Hence I have eaten situates us within an area I have labelled both not eat and no longer eat determined relative to a previous occurrence of eat. The above diagram gives the impression of a certain symmetry to the left and right of the process eat, but the position to the left of the Interior, or before eat, is very different from the Exterior area on the right. This relates to a very fundamental difference between the before and the after. Before an event takes place, there is no way of knowing with certainty whether it will or will not come about. After an event has taken place, the situation is necessarily determined in respect of its existence. A static, two dimensional representation necessarily obscures this fact. Hence we consider that the area to the right of eat corresponds to the Exterior E, while the area to the left actually corresponds to the offline position IE, an area which is identifiable neither with I nor with E but from which both positions are potentially accessible. In saying I haven’t eaten yet a subject constructs a position to the left of eat, in IE, relative to a preconstructed position to the right of eat, in E. The event . Composite in that the bounded interval defines both an aspectuo-temporal and a notional area.



TSL[v.20020404] Prn:10/05/2007; 13:45

F: PB16111.tex / p.6 (182)

 Graham Ranger

eat, represented by a bounded interval, functions as a threshold discontinuity between two complementary areas:8 IE

Interior eat

7

establishing a

Exterior not/ no longer eat

t construction

preconstruction

Figure 4.

The model sketched out above links modal and aspectual fields.9 The position in IE, corresponding to the before, from an aspectuo-temporal perspective, is simultaneously a modally disconnected position allowing us to represent the opposing possibilities of validation or non validation.10 This position is often constructed in English by the use of the infinitive to v or a modal, as in the example below: (11) This positive approach to the challenges offered by the Tunnel is also evident in the activities of Projenor, a company set up by a number of organisations involved in the Channel Tunnel including Eurotunnel, SNCF, local and regional authorities in France, and major financial institutions such as the Belgian Credit Communal and the Midland Bank. Projenor was established in response to fears about the corridor effect and aims to promote the economic, cultural and environmental restructuring of the regions around the Tunnel. It is involved, as a catalyst, in major projects along the French coast and inland at Arras and Lille. However, despite its interests in all parts of the Euro-zone bounded by London, Paris and Brussels it has yet to launch its first British project.

Here, given the aims of Projenor we might expect it to be involved in British projects. In other words we might have expected it to have launched its first British . Hirtle (1977:32) also invokes the concept of the threshold. . This is why the event eat can be constructed either as a Boundary value (relative to before and after the meal) or as an Interior value (relative to eat/ not eat). . The marker already functions similarly except that, contrary to yet, already preconstructs a position in IE and constructs a position in E relative to this preconstruction. Of course already does not possess a clear argumentative use in English, unlike French déjà. . Validation and non validation denote operations whereby a representation (conventionally p) is situated on the Interior or on the Exterior of a given domain: respectively p is the case (validation of p) or p is not the case (non validation of p).

TSL[v.20020404] Prn:10/05/2007; 13:45

F: PB16111.tex / p.7 (183)

Continuity and discontinuity in discourse 

project. The launch threshold has not been crossed, however, and the representation constructed is similar to that in the previous example (cf. the reformulation it has not yet launched . . . ). (12) He is seen as a reforming Communist and was, for a while, in charge of the youth section of the party and a provincial Communist Party leader. His opposition while in office was such that he was ousted – to run a technical publishing house – but he is still in his early fifties and may yet have a political future.

Here the modal may allows us to represent the available possibilities have/ not have a political future (IE) relative to a preconstructed position E according to which the subject is politically finished (he was ousted). The possibility of have a political future is left open in virtue of the subject’s relatively young age: he is still in his early fifties. In this way the threshold of age, beyond which a political future is apparently no longer conceivable, has not yet been crossed. The interpenetration of modal and aspectual values means that we can represent the situation convincingly either using the branching path model or the class of instants as below.

have a political future

not have a political future

preconstruction

construction

Figure 5.

political future

age limit

no political future

t construction

preconstruction

Figure 6.

Unlike yet, which opposes discontinuous options, still functions in terms of continuity by identifying a given occurrence with a previous occurrence in the same

TSL[v.20020404] Prn:10/05/2007; 13:45

F: PB16111.tex / p.8 (184)

 Graham Ranger

topological space such that the continuity between occurrences is maintained. If we compare the pair I haven’t eaten yet and I still haven’t eaten we can consider that still constructs a position to the left of eat identified to and continuous with a preconstructed position on its left. eat t preconstruction

construction

Figure 7.

Given that we are on the ordered class of instants, such that each instant contains the potential for temporal differentiation with the preceding one, it is easy to see that, the further right we go, the greater the likelihood that the threshold represented by eat will have been crossed. Put simply, and given what we know about eating, and about the passing of time, the later it gets the more likely it is one will have eaten!11 In the example below, the relationship between still and yet is again easy to appreciate: (13) And, health and safety watchdogs say they’re not convinced the Channel Tunnel has sufficient safety measures in place to protect passengers. The tunnel is due to open next May, but the Health and Safety Executive say procedures still haven’t been tested.

still identifies a position in IE (before test) with a previous position in IE. If we had had procedures have not been tested yet, we would have constructed a position in IE, in opposition with a preconstructed position in E where the tests would already have been made. In the following example, yet can easily replace still. The passage of time implied by reminder letters and the mention of a date limit, allows us to construct the potential for differentiation which renders still and yet practically equivalent. (14) Reminder letters have been sent to those course organisers with courses ending by 31 October, who have still to send in completed questionnaires.

. The constructions we can operate on a predicate like eat are heavily affected by sociocultural considerations: in cultures where it is habitual to eat only one meal a day in the evening, I still haven’t eaten would make no sense at all if pronounced at three o’clock in the afternoon, say.

TSL[v.20020404] Prn:10/05/2007; 13:45

F: PB16111.tex / p.9 (185)

Continuity and discontinuity in discourse

. Focal uses The study of yet or still in focal uses, with comparatives, notably, or with series, provides a number of interesting and revealing oppositions. Again, in keeping with our hypotheses, still is more likely to appear in continuous series, while yet displays an affinity for discontinuous series. (15) I went along to Westminster Evening Institute again. I had sat the finals for Philosophy in late May. I signed for yet another year’s work, Social Psychology.

Here the fact that the subject has sat his or her finals in Philosophy allows us to construct a position to the right of a threshold representing a period of study (a year’s work, here). The use of yet in yet another year’s work takes us back to the left of this threshold point and opens a discontinuous area, a new threshold, represented by another year’s work. yet is often used in such contexts to mark that a series is not in fact concluded. Here still would be unlikely:12 (15) a. *I signed for still another year’s work.

This, I suggest, is due to the discontinuity constructed by another year. If we consider that still marks a continuity between occurrences it would be surprising to find it used in combination with such an explicit marker of discontinuity as another. The way yet works might be represented as follows:

another year’s work

t construction

preconstruction

Figure 8.

Similar arguments could be applied to the following example: (16) And now England could have a fight on their hands in their second innings on a pitch beginning to help the slow bowlers. Yet again, however, England’s spinners looked far less dangerous in such conditions. . Actually Hirtle (1977:41) lists the following counter example, featuring still another, from Webster’s dictionary: still another example of cultural misunderstanding. There is not enough context for us to discuss the example fully. I would not wish to claim that still another was inacceptable in all circumstances; a quick websearch nonetheless reveals that one is twenty-six times more likely to encounter yet another than still another!



TSL[v.20020404] Prn:10/05/2007; 13:45

F: PB16111.tex / p.10 (186)

 Graham Ranger

a. *. . . Still again, however, England’s spinners looked far less dangerous in such conditions.

Here the marker again constructs a separate occurrence of an already posited notion and it is this discontinuity which requires us to use yet, indicating that a series which should have ended (preconstruction in E) (cf. a pitch beginning to help the slow bowlers) has not in fact ended (construction in IE). With domains of degree that function continuously both yet and still appear to be possible. (17) “And she likes the house to stay just the same,” she remarked to Fru Gertlinger, as she swept back through the green-baize door for yet more toast, “so she’s not going to object to the blue room being returned to its former colours [. . .]” a. [. . . ] as she swept back through the green-baize door for still more toast [. . . ]

We could, similarly, modify (15) to give: (15) b. I went along to Westminster Evening Institute again. I had sat the finals for Philosophy in late May. I signed for yet more work, Social Psychology.

in which case still would also be possible: (15) c.

I went along to Westminster Evening Institute again. I had sat the finals for Philosophy in late May. I signed for still more work, Social Psychology.

The following example is taken from a class in which a primary school teacher is explaining how decimals function. (18) Right, now that’s what a decimal is a decimal is, that’s what a decimal is [sic], and there are some people think we should write them like this, or you couldn’t, you couldn’t do this. It’s nice just to see it once, that er really that nine should be written smaller and then that two should be even smaller still, and that three should almost be unreadable because those things are smaller and smaller aren’t they, as you go along there, these are getting tinier and tinier.13

The teacher evokes a continuous decrease in size and it is this continuity which justifies the use of still in preference to yet. Actually still is here placed after the comparative adjective and is used correlatively with even, another marker associated with establishing continuity.14 . The hesitations, repetitions etc. are characteristic of transcripts of spoken English. . Cf. Culioli (1999), (2002) and Ranger (1999) on even in an enunciative framework.

TSL[v.20020404] Prn:10/05/2007; 13:45

F: PB16111.tex / p.11 (187)

Continuity and discontinuity in discourse 

. Argumentative uses In argumentative uses we are dealing with concessive constructions in which two notional domains, associated with the propositions p and q, are typically constructed as incompatible. This incompatibility can be represented on the model of the notional domain such that an occurrence p normally implies non q while an occurrence q implies non p.15

p non q

q non p

Figure 9.

I would like to maintain that a concessive utterance of the general form still, p constructs a position identified to and continuous with a preconstructed position. Such uses typically signal that the proposition is validated or that speaker maintains his or her argumentative position, to which the proposition contributes, in the face of a potential counterargument judged ineffectual. A concessive utterance (and) yet p constructs an off-line position potentially compatible with p or q, in opposition with a preconstructed position q. This serves either to signal the paradoxical coexistence of opposing values or the refutation of a value located relative to a co-speaker. Let us begin by looking at an example featuring still: (19) Its essential elements are the first vote given to a constituency candidate, the second vote given to a party list, and that list itself. Each merits closer examination. The first vote The election of an MdB in a constituency is obviously straightforward and in numerical terms unambiguous. Even if he has fewer votes than his competitors jointly, he is still the winner by virtue of having more votes than any one of them.

In he is still the winner. . . the speaker maintains the proposition p he is the winner identified with a previous occurrence of p, thereby eliminating the potential for discontinuity present in the hypothetical clause q even if he has fewer votes than his competitors jointly. This can be represented as follows:

. I am not talking about logical implication here, of course.

TSL[v.20020404] Prn:10/05/2007; 13:45

F: PB16111.tex / p.12 (188)

 Graham Ranger

he be the winner

preconstruction

he not be the winner

construction

Figure 10.

Whereas in aspectual uses of still, continuity was maintained in the face of the discontinuities inherent in the successivity of the class of instants, in argumentative uses, the discontinuity is derived from a proposition which can function as a counter argument, generating a potential discontinuity or change in situation: he be the winner

he not be the winner

he has fewer votes than his competitors jointly

Figure 11.

The marker even prepares still of course, and this is again why yet would be very unusual in the example, and would, in any case, introduce a significant change in value. If we eliminate even we might imagine the following utterance: (19) a.

[. . . ] he has fewer votes than his competitors jointly, yet he is the winner by virtue of having more votes than any one of them.

Unlike still, the marker yet here would obviously fail to relate back to any previous occurrence of the notion associated with he be the winner and so, rather than maintaining a previously given position, yet would seem rather to highlight some irregularity or incongruity in the electoral system. (20) But under intense congressional pressure the Defence Department says that it will kill the programme next year unless the services manage to iron out serious flaws. The electronic jammer programme is at least four years behind schedule and may cost at least $1 billion more than estimated, according to one Pentagon report. Still, the Pentagon has yet to bite the bullet and recently awarded $419 million in contracts to Westinghouse and ITT Avionics Division, which have developed the system, to build the first 100 production models for installation in the aircraft [. . .]

When still is prosodically detached and in initial position it constructs slightly different values. Let us consider that this utterance opposes propositions oriented

TSL[v.20020404] Prn:10/05/2007; 13:45

F: PB16111.tex / p.13 (189)

Continuity and discontinuity in discourse 

respectively towards incompatible conclusions: supporting or “killing” a certain defence programme. The propositions can be represented as occurrences on either side of a domain. The use of still precedes a proposition telling us that support for the programme continues. This continuity is maintained in the face of a discontinuity we could derive from the preceding proposition.16 In such initial uses of still it is generally not the propositions as such which are in opposition but the conclusions one might draw from them. In common with this we often find initial still used in dialogue to signal that a speaker maintains an original argument in the face of a potential counter argument.17 Let us move on to argumentative uses of yet with the example below, taken from Stoppard’s Rosencrantz and Guildenstern are dead: (21) ROS: Mad. PLAYER: How is he mad? ROS: Ah. (To GUIL) How is he mad? GUIL: More morose than mad, perhaps. PLAYER: Melancholy. GUIL: Moody. ROS: He has moods. PLAYER: Of moroseness? GUIL: Madness. And yet. ROS: Quite. GUIL: For instance. ROS: He talks to himself, which might be madness. GUIL: If he didn’t talk sense, which he does. ROS: Which suggests the opposite. PLAYER: Of what? (Small pause.)

Here Rosencrantz and Guildenstern are discussing with the players the question of Hamlet’s curious behaviour. The problem is whether Hamlet is mad or merely melancholy. Let us consider madness and melancholy as the incompatible Interior

. The marker yet is here used aspectually and correlatively with still so that the argument, counter-argument, argument pattern marked by still is reinforced by the use of yet to position us in IE relative to the threshold event bite the bullet (that is, stop funding the project). . Hirtle (1977:42) provides the following amusing example from P.G.Wodehouse: “If you mean would I accept him if he asked me to marry him, yes I would. I’d jump into his arms.” “Well, I’m not sure I’d advise that. I don’t want to seem personal, but you’re on the solid side and he’s kind of flimsy. You might fracture something. Still, the point, the thing we’ve been trying to get at, is that your views on the subject of centre-aisleing coincide with his, so that’s all right. . .” (Wodehouse 1965: 76)

TSL[v.20020404] Prn:10/05/2007; 13:45

F: PB16111.tex / p.14 (190)

 Graham Ranger

and Exterior.18 The diagnosis swings from madness to melancholy until Guildenstern apparently opts for madness, only to reintroduce the possibility of melancholy with the use of and yet. Rosencrantz’s confirmation with quite is, of course, ironical, since the marker quite would normally be incompatible with the destabilising movement of and yet! Here is how we might represent this use of and yet on the branching path model:

melancholy

madness

preconstruction

and yet construction

Figure 12.

In (21) the pregnant And yet merely places us in IE hence potentially reopening the path towards I. This is an important feature of the analysis provided here: argumentative yet does not situate us in direct opposition with a preconstructed position but rather introduces, via the offline position, a possible access to a contradictory position. Of course the movement toward the offline position marked by yet is often followed by a proposition which takes us from IE into I as below: (22) The origins of this lay in a religion of local cults which later developed links with each other. To the ancient Egyptian this diversity was easily acceptable, being inherent to divine powers who were approached through a variety of images related to nature and animal and human life. Therefore in those terms the gods were born, lived and died and yet paradoxically were immortal.

The opposition is between being born, living and dying and being immortal, two generally irreconcilable options! The text preconstructs a position on the Exterior, corresponding to mortality, but then passes through IE to allow the validation of a position on the Interior with and yet paradoxically were immortal. The adverb paradoxically encapsulates nicely one of the effects such uses of yet can have, as . Obviously madness and melancholy are not necessarily incompatible in real life. The point is that here the subjects oppose the two in an either . . . or choice.

TSL[v.20020404] Prn:10/05/2007; 13:45

F: PB16111.tex / p.15 (191)

Continuity and discontinuity in discourse

it constructs a picture of a contrasted situation of which one might legitimately predicate two generally incompatible propositions. In other words the passage to I via IE destabilises the preconstructed position in E in such a way that the associated proposition no longer provides a sufficient or exclusive picture of the situation. Hence the referential values preconstructed for the gods were born, lived and died are dragged from E into the Boundary area, while those corresponding to the gods were immortal are also adjusted to give us non typical occurrences of E or I. We can indicate this effect on the diagram below: Immortality

the gods were immortal

BOUNDARY

Mortality

the gods were born, lived and died

Figure 13.

When the propositions are associated with marks of modal endorsement, a subject may use yet in order to undermine or invalidate a preconstructed position attributed to another speaker as below: (23) In time this led to resentment, especially among the growing educated group who felt the arrogance of the British most directly. To make matters worse the British government repeatedly announced that it would be pulling out of Egypt once the country was on a “sound” footing. Yet in practice the British seemed to dig in ever deeper. As a result, by the early 1900s, a nationalist undercurrent was developing led by a charismatic young man, Mustafa Kamil, and though there were strikes and demonstrations (which were to become a regular feature of Egyptian political life), Britain remained as unmoved, aloof and arrogant as ever.

In rough terms, the opposition is between leave Egypt and stay in Egypt. The text preconstructs a position on the Exterior with the British government [. . . ] would be pulling out of Egypt, but then passes through the offline position IE to allow the validation of a position on the Interior, the British seemed to dig in ever deeper. Since the preconstructed position is located relative to the modal endorsement of the British government repeatedly announced . . . the introduction of the opposing position has the effect of undermining the sincerity of the British government’s announcements in an argumentative movement of refutation. In dialogal contexts this destabilising movement is often used to encourage the invalidation of the preconstructed position. This is what happens in the following fairly typical courtroom example, taken from the on-line proceedings of



TSL[v.20020404] Prn:10/05/2007; 13:45

F: PB16111.tex / p.16 (192)

 Graham Ranger

dug in deeper

it would be pulling out

preconstruction

yet construction

Figure 14.

stay in Egypt

Britain dug in deeper

BOUNDARY

leave Egypt

Britain would be pulling out

Figure 15.

the Old Bailey for 1815, where the lawyer employs And yet ironically to signal a latent contradiction in the witness’s testimony by constructing two propositions as incompatibly opposed: (24) Q. Look at the prisoner, and tell me whether you are certain that he is the man – A. I am quite clear that the prisoner is the man; I have no doubt about it [. . .] Q. Did you ever see the man before – A. No; nor never after until I saw him at Bow-street; he was dressed in a dark coat, rather short for him; I think it was something of a jacket. Q. Had he on breeches or pantaloons – A. I will not swear to that. Q. And yet you will swear to the man – A. I do not swear to the lower part of his dress; I found a great deal of blood in the passage, and the watchman traced it all the way to Grosvenor-street; I only traced it down the steps. http://www.oldbaileyonline.org/html_units/1810s/t18150111-58.html

In 1815, it was, suggests the lawyer, impossible to swear to the identity of someone if one cannot also swear that he was wearing breeches or pantaloons.

TSL[v.20020404] Prn:10/05/2007; 13:45

F: PB16111.tex / p.17 (193)

Continuity and discontinuity in discourse 

. Concluding remarks I hope, with the preceding remarks, to have shown how it is possible to provide an operational description of the markers yet and still which enables us to explain their range of possible values as a function of the notional domains operated upon. still maintains a continuity between representations whereas yet enables the construction of a discontinuity between representations. When the representations involved are situated on the class of instants, we obtain aspectual values, when they are situated on a series or a gradient, we obtain focal values, and when they are situated within a class of arguments tending towards a conclusion, we obtain argumentative values. The conclusions reached are not in contradiction with less formal accounts in which yet is characterised as “significant change”, “surprise”, “paradox” and still as “continuation”, “continuance” etc.19 The destabilising effect of yet, as it takes us in the offline position, also allows us to explain its often noted affinity with negative or interrogative contexts.20 However, the description given here is made within the formal system of metalinguistic representation provided by the Theory of Enunciative Operations with the advantage that the tools used (the notional domain, the branching path model) are not limited to the description of yet and still but provide a rigorously constructed and generalisable framework for describing a vast range of heterogeneous linguistic phenomena.

References Bouscaren, J., Chuquet, J. and Danon-Boileau, L. 1992. Introduction to a linguistic grammar of English. An utterer-centered approach (translated and adapted by Flintham, R. and Bouscaren, J.), Paris: Ophrys. Crupi, C. D. 2004. But Still and Yet: The Quest For a Constant Semantic Value For English yet. Doctor of Education Dissertation, Graduate School of Education, Rutgers University. January 2004. Culioli, A. 1990. “The Concept of Notional Domain.” In Pour une linguistique de l’énonciation. Tome 1, 67–81. Paris: Ophrys. Culioli, A. 1990. “ Representation, referential processes and regulation.” In Pour une linguistique de l’énonciation. Tome 1, 177–213. Paris: Ophrys. Culioli, A. 1995. Cognition and Representation in Linguistic Theory (ed. Michel Liddle), Amsterdam/Philadelphia: John Benjamins. Culioli, A. 1990. “Even though, even if, as though, as if.” In Pour une linguistique de l’énonciation. Tome 3, 177–181. Paris: Ophrys.

. Terms we find in Hirtle (1977), or Crupi (2004) for example. . As in Traugott and Waterhouse (1969) for example.

TSL[v.20020404] Prn:10/05/2007; 13:45

F: PB16111.tex / p.18 (194)

 Graham Ranger

Culioli, A. 2002. “A propos de même” Langue française no. 133, 16–27, Paris. Hirtle, W. H. 1977. “Already, still and yet” in Archivum Linguisticum, VII N.S. No. 1: 28–45. Ranger, G. 1999. Les constructions concessives en anglais: une approche énonciative, Paris: Ophrys. Traugott, E. C. & Waterhouse, J. 1969. “Already and yet: a suppletive set of aspect markers?” In Journal of Linguistics vol. 5, 287–304. Cambridge: CUP.

TSL[v.20020404] Prn:30/05/2007; 10:15

F: PB16112.tex / p.1 (195)

Reconsidering the discourse marking hypothesis Even, even though, even if, etc. as morpheme/construction pairs François Nemo University of Orléans (France)

Discourse connectives, by definition, are landmarks of discourse. This reality should however be distinguished from the usual claim according to which they are discourse markers expliciting the nature of the relation between discourse segments. On the contrary, it can be shown that the polyfunctionality and instability of discourse connectives can only be explained in terms of integration of two connective links, one being associated with the specific contributional nature of the on-going utterance while the other is the result of the connective interpretation of the encoded meaning of the morpheme. An illustration of this approach is proposed on the connective and non-connective uses of English even, following a similar work on but, showing that the conceptual/procedural distinction may fruitfully be abandoned.

.

Life could be so simple

Because they play an important role in the interpretation of discourse and because their scope typically includes two (or more) sentences, discourse connectives (DCs) are landmarks of discourse. But as we shall see, understanding their nature and construction is far more complex than it might appear. A common sense idea about discourse connectives is that they are discourse markers, i.e. markers of discourse relations. According to this view, if S1 and S2 are two discourse segments or two discourse units, typically utterances, and if R is the relationship between S1 and S2, then the function of a DC is to mark R and/or to make R explicit. Consequently, given that different DCs can be used to mark a single discourse relationship R, it is often considered that a DC can be studied in terms of its contrastive value within a functional family.

TSL[v.20020404] Prn:30/05/2007; 10:15

F: PB16112.tex / p.2 (196)

 François Nemo

. But it’s not !! There are, however, four problems. The first problem, that I will illustrate by means of a detailed study of the connective uses of even, is that almost all DCs are polysemous (polyfunctional), i.e. they may be associated with very diverse DRs. Moreover, as we shall see in the next section, tests also show that a DC such as English but, which is usually believed to mark a contrast, may in fact be used in association with almost any kind of relationship R between S1 and S2. Furthermore, quite often DCs do not connect propositions/utterances (Ducrot & alii 1980; Rouchota 1996). Finally, they introduce attentional/contributional modifications which are not accounted for if DCs are treated as DMs. .. Testing the discourse marking hypothesis If we test the insertion of the so-called contrastive discourse marker but (Fraser 1998) in any of the standard discourse relations (DRs) identified for instance in SDRT, we are led to the observation that: – – –

– – –

but may be used to mark the absence of contrast: Peter is tall but so is John. but may be used with a DR of explanation,1 as in: Peter failed but he was sick that day in which he was sick explains why Peter failed. but may be used with a DR of consequence as in: Peter was hungry but we gave him food where we gave him food is a consequence of the fact that Peter was hungry. but may be used with a DR of background as in: Peter came in but the room was dark. but may be used with a DR of elaboration as in: Peter has bought a new house but it took him weeks to get the money. etc.

which leads us to the conclusion that in a S1/S2 sequence, the presence of but alone does not specify the S1/S2 relationship. Thus, far from “guiding the interpretation” of the sequence in which it occurs, as some linguists used to believe (Wilson & Sperber 1990, 1993; Luscher 1994), the interpretation of but seems to be embedded in the DR interpretation of the sequence. It seems hence that the interpretation of DCs is routinely guided by the nature of the context of insertion, and not viceversa. In other words, it seems that DCs are simply not Discourse Relation Markers. . Ducrot (1980), discussing counterexamples to his argumentative description of mais, provides excellent illustrations of this explanative use of mais.

TSL[v.20020404] Prn:30/05/2007; 10:15

F: PB16112.tex / p.3 (197)

Reconsidering the discourse marking hypothesis 

.. Dropping the discourse relation marking hypothesis The above tests thus falsify the thesis that DCs make explicit the nature of the discursive relationship between discourse segments, showing clearly that but does not mark a contrastive relationship. Furthermore, linguists who have adopted a functional family approach (see Nemo 2006) in the description of DCs, studying for instance reformulative discourse connectives (Rossari 1994), are led to the same conclusion: While it seems intuitively obvious that many discourse markers (DMs) are able to express discourse relations (DRs) which exist independently, the specific contributions of DMs – if any – is not clear. [Investigating the status of some consequence DMs in French] we observe that it is difficult to construct a clear and simple definition based on DRs for these DMs. Next, we show that the lexical constraints associated with such DMs extend far beyond simple compatibility with DRs. This suggests that the view of DMs as signaling general all-purpose DMs is to be seriously amended in favor of more precise descriptions of DMs in which the compatibility with DRs is derived from a lexical semantic profile. (Jayez & Rossari 1998)

From the set of test sentences presented above, it must however be concluded that the compatibility of DCs with DRs seems to go very far indeed. .. Dealing with polyfunctionality Since the actual behaviour of DCs is incompatible with the idea that discourse connectives are markers of discourse relations, we are left with only a few possibilities for the relationship between DCs and DRs. First of all, we can adopt a perspective on the diversity of uses of a DC based on Prototype Theory, according to which some uses of a DC, such as the contrastive uses of but, would be prototypical (and/or learnt initially) whereas all the others would be non-prototypical (and/or learnt later); prototypical uses would then still mark specific discourse relations. In such a perspective, it would still be possible to claim that, for instance, but is a contrastive discourse marker. In the more descriptive token-oriented perspective which I will adopt here, I start from the observation that semantic units used as discourse connectives may be used in association with very different discourse relations, and sometimes also in non-connective contexts. Central to this view is the idea that polyfunctionality and polycategoriality are the rule and not the exception. So that instead of leaving aside many uses of the semantic unit but: I am not treating other uses of but such as found in: “All but one left today”, “There was no doubt but that he won”, “it has not sooner started but it stopped”, “He was but a poor man”, “I may be wrong but I think you are beautiful”. Whether or not they could be included under my analysis is left open. (Fraser 1998)

in order to save the idea that:

TSL[v.20020404] Prn:30/05/2007; 10:15

F: PB16112.tex / p.4 (198)

 François Nemo

The core meaning of but is to signal simple contrast, nothing more, and the speaker will select it when intending to highlight a contrast. (Fraser 1998)

we need to adopt a constructional perspective according to which semantic units are inserted in discourse relating constructions; that is, I adopt an approach in which discourse connectives are morpheme/construction pairs, an approach which I shall develop here in the description of all the (connective and nonconnective) uses of the English morpheme even, as I have done recently about but (Nemo 2002, 2006), showing that these morphemes actually neither encode a (additive-contrastive) connective nor a metacommunicational nor a procedural meaning (Wilson & Sperber 1990, 1993), but some indications (see Nemo 2001, 2003) which receive a connective or non-connective interpretation, depending on their context of use. In this paper, I shall try to focus on three questions: – – –

a discussion about the nature of DCs if they are not discourse (relation) markers; a discussion about the nature of discourse relations; a non-categorial description of the English morpheme even, in all its connective (even though, even if, even so, and even then, etc.) and non-connective uses, showing thus that it can play the role of a discourse landmark.

Doing so I shall leave aside a detailed discussion of the notion of prototypicality introduced above; this notion must be considered problematic since it renders any description of DCs unfalsifiable, leading to completely arbitrary and unpredictable semantics. In such a view, a semantic unit is sometimes used because of its meaning, and in spite of its meaning in others. A prototype view furthermore forces the linguist, as a way to evade the puzzles (s)he faces, to defend the idea that some uses would have to be considered as central whereas the others would be peripheral. .. The double link hypothesis I will defend here the idea that the only way to account for the data and the tests is to acknowledge the fact that DCs, instead of merely providing information about the S1/S2 relationship, build their own link, between S1 and S2 independently of any other semantic relationships which may exist between S1 and S2. And consequently that the existence of two distinct kinds of relationship between S1 and S2, i.e. the existence of two links between them, explains the diversity of interpretations of a given DC in its different uses.

TSL[v.20020404] Prn:30/05/2007; 10:15

F: PB16112.tex / p.5 (199)

Reconsidering the discourse marking hypothesis 

. If discourse connectives are not discourse relation markers, what are they? Dropping the discourse marking hypothesis implies reconsideration of what discourse connectives are in terms of discourse structure, in terms of syntax and in terms of discourse interpretation. Hence, we need to develop new hypotheses about DCs which allow us to predict their behaviour. .. A step back: The marker hypothesis In the standard view on which the discourse marking hypothesis is based: i.

discourses (or texts) are made of discourse segments or discourse units (typically utterances); ii. interpreting the relationship between successive utterances is part of discursive competence; iii. the function of discourse connectives is to make the relationship between successive utterances explicit; iv. describing discourse connectives is a matter of describing the kind of discourse relations that are explicited/marked as well as the nature of the connected segments. .. A step further: The modifier hypothesis I shall assume, as discussed in previous work (Nemo 1999, 2006), that: 1. discourses (texts) are made of contributions C; 2. a contribution is defined in terms of “this is what you must pay attention to and what must be taken into account” (Nemo 1999; Tomasello 1995); 3. we must distinguish between intra-contributional and inter-contributional relationships. 4. a contribution may be modified by C/C modifiers; 5. the syntactic type of discourse connectives is C/C. This defines them as contribution modifiers; 6. modifying a contribution means introducing something new in the hearer’s attentional field and asking him/her to also take it into account. 7. a large part of the individual descriptions of DCs available so far in all frameworks (and possibly most of them) are consistent with the modifier hypothesis. .. Marker hypothesis vs. modifier hypothesis In a discourse formed of two utterances (or segments) U1 and U2 joined by a discourse connective (DC), the DC will be described as: – –

joining U1 and U2 in the Discourse Marker Hypothesis (DMH); joining [U1] and [U1U2] in the Contribution Modifier Hypothesis (CMH).

TSL[v.20020404] Prn:30/05/2007; 10:15

F: PB16112.tex / p.6 (200)

 François Nemo

In other words, the DMH approach considers DCs to deal with the relationship between subsequent utterances (or discourse segments of any size) whereas according to the CMH approach DCs deal with alternative discourses and the nature of this alternative. In categorial terms, the DMH describes discourse connectives as a type of word requiring two utterances (or discourse units) and forming a discourse D, whereas the CMH, in coherence with the categorial treatment of any class of modifiers (such as adverbs, adjectives, etc.), describes them as a C/C category, whose output is of the same categorial nature (contribution) as its input. Within the CMH perspective, a distinction must be drawn between U/U and C/C modifiers, i.e. between utterance modifiers (Nølke 1993) which integrate viewpoints at the level of utterances, and contribution modifiers, which integrate viewpoints at the level of contributions. .. The syntactic behaviour of DCs The classical idea that DCs connect two utterances U1 and U2 is an oversimplification even of the simplest data, since the hypothesised U2 is very frequently a simple modification of only one element of U1. Data mining shows very clearly that connecting two distinct utterances U1 and U2 constitutes only one end of the spectrum of what DCs can syntactically do, and that: – –

–

–

they sometimes introduce no linguistic elements, as in examples (1) and (2); they often introduce elements (fragments) which either complement or replace elements of the previous utterance and thus could be syntactically integrated to this utterance (I-fragments from now on), as in examples (5) to (8); they sometimes introduce elements which can be articulated syntactically (but not integrated) with the previous sentence, but which cannot stand alone syntactically (A-fragments from now on), as in examples (9) and (10); they sometimes introduce elements that cannot be integrated syntactically to the previous utterance but can syntactically stand on their own, as in “It’s beautiful but it’s expensive”.

So that we have for instance: (1) Il ira jusqu’à Dijon. Et encore! (He shall reach Dijon, if that!) (2) C’est n’importe quoi! Enfin! (This is nonsense! Well, forget it!) (3) C’est n’importe quoi! Enfin, on va faire avec. (This is nonsense! No matter, we’ll manage) (4) He arrived. Late. (5) He arrived but late.

TSL[v.20020404] Prn:30/05/2007; 10:15

F: PB16112.tex / p.7 (201)

Reconsidering the discourse marking hypothesis 

(6) He arrived but slowly. (7) He arrived but in Paris. (8) He arrived, with Paul actually. (9) It’s a friend, a very old friend in fact. (10) Ici, le port de la cravate est une tradition, et encore pas n’importe laquelle. (Here, wearing a tie is a tradition, and not just any tie at that.)

It must hence be acknowledged that if the syntactic diversity of elements that are introduced by the DC is directly predictable from the discourse modifier C/C hypothesis, it is on the contrary unpredictable from the D/UIU2 description, which seems to be an overgeneralization of one end of the spectrum of uses.

. Using even in English: Connective and non connective uses Our case study will be the English linguistic unit even, which “frequently shows up as” a component “of concessive connectives” (König 1991: 2; Culioli 1999) or as a temporal connective (even as) in fully connective uses, as well as in clearly nonconnective uses, when it is used as an adjective or a verb, together with some uses that may technically be considered as hybrid, as is apparent in the term of “additive scalar particle” itself, the connective/additive dimension being inseparable from the scalar one, as in “Nobody’s asked any nasty questions, not even the cops” (BNC, HWL 1750). Even if even in such uses has been mainly contrasted with units such as too, also, etc. and thus studied as an additive focal particle, it is technically obvious that it behaves like standard connectives because of i) its role of introducing new information to be considered; ii) giving more details about information that has just been introduced or which is available; iii) being generally inserted in a polysentential sequence. The typical usage is to introduce two chunks of information, one of which is being focused upon, the other being presupposed, and both of them being possibly argumentatively co-oriented on a scale (Anscombre & Ducrot 1983). This case has typically been considered central for its semantic description by most linguists, although it has proved in some respect difficult to generalize (see König 1991: 69–76). However, there are further uses of even playing a connective role. For instance, in an example such as “What a tremendous success! Even John agreed to sign!”, a connection is established between the two utterances in which even introduces the (possibly) strongest available backing for the first evaluation. This connection is independent of the scalar relationship between the fact that John agreed to sign and the fact that other people than John agreed to sign. The incorporation of what is added to what has just been said (e.g. the including value),

TSL[v.20020404] Prn:30/05/2007; 10:15

F: PB16112.tex / p.8 (202)

 François Nemo

which many authors seem to consider as prototypical of such uses, is indeed both typically connective and consistent with the definition of discourse connectives as contribution modifiers. Such a wide distribution is not specific to English even, since the distribution of French même, for instance, shows a similar overlap of non-connective uses (la même robe), fully connective uses (même que, quand même, tout de même) and mainly connective uses (ils sont tous venus, même Paul). . Diversity of uses of even Below is a list of corpus examples of even in different uses. These examples serve to illustrate here the various contextual instantiations of even, and they will be discussed in detail in Section 2.4. Starting with the connective uses, even is often used as the first element of an adverbial connective phrase, such as in: (11) Since then, the forces of continental drift have continued to pull the two continents apart, widening the Atlantic, but the habit of returning to the sea each year to spawn has never been broken even though it now involves such an immense journey. (BNC F9F 1637.) (12) We knew that if the mother asked for her back, we would have to give her up cheerfully, even though Joan and Susan were inseparable. (BNC AT3 1319). (13) Even if he was trying to show off, I still respected him. (BNC J0W 1740). (14) Even if you are granted non-resident status, some of your income may still be liable to British taxation. (BNC CMK 1804). (15) Even if they get me, I’ll go down fighting. (BNC FR0 4383). (16) But once you’d lied –; even if it were only by implication or simply by failing to deny something –; you were forced to go on lying. (BNC HHA 2227). (17) After my mother died, and he married Stella ; well, even then, with a new marriage, his support and guidance never faltered. (BNC HA7 2090). (18) Even as she spoke she was taking a fleecy rose-coloured towel out of the linen press, turning on the water, unscrewing a flask of bath essence which perfumed the rising steam. (BNC HA7 1321). (19) Mm, even so I mean he, he’s still got to buy it ain’t he? (BNC KDM 6804). (20) The broadcasters recognized, even so, that there were important differences, particularly in the costs of TV and in the way people might use it. (BNC CRY 1113).

Even may also be used alone as an additive scalar particle, as in: (21) All life is interdependent on the natural environment, from the smallest bacteria to the largest animal or plant ; even man. (BNC CGH 1149).

TSL[v.20020404] Prn:30/05/2007; 10:15

F: PB16112.tex / p.9 (203)

Reconsidering the discourse marking hypothesis 

(22) Spiders are known as far back as the Carboniferous, but their remains are principally known fossilized from the Tertiary ambers, where perfectly preserved specimens retain even the hairs on the legs. (BNC AMM 1052). (23) In fact, even the most basic repairs had been neglected for many years before that, so that structural problems developed where they might have been avoided. (BNC AR9 1201). (24) Even his mother had noticed it (BNC CR6 1173). (25) Hamlet is up against not just a man, but a king, he will have to strike when the king is unarmed, And he would also have to be able to explain his actions, and yet even his mother does not believe him, but only see him as mad (BNC HXH 1892).

Finally, it may be used non-connectively as an adjective, a noun or a verb: (26) Once in a while why not take advantage of the smooth, even surface to concentrate on your control and technique? (BNC G2W 1211). (27) Incidentally the odds on the hunted fox being killed are about evens. (BNC B03 1199). (28) You suggest that Detroit should join its suburban neighbours in a regional government that does away with redundancies and evens out the huge inequities in school financing and municipal services; (May 8th). (BNC CR8 315). (29) Each team had a total thirty-five handicap, so one imagined it would be an even match. (BNC ED9 710).

. Describing morphemes In spite of the polycategorial distribution of both connective and non-connective uses in so many different constructions and categorial positions, the encoded meaning of even, i.e. its signification, remains the same in all of its uses. To illustrate my claim, I shall use an Indicational-Indexical Semantics (IIS) approach. .. Indicational-indexical semantics: Encoded meaning The IIS thesis may be spelled out as follows: 1. the semantic (encoded) meaning of a morpheme provides the same indications in all uses and contexts;

TSL[v.20020404] Prn:30/05/2007; 10:15

F: PB16112.tex / p.10 (204)

 François Nemo

2. an indication is a piece of declarative information which states for instance that “There is an X and a Y (etc.) with a R(X/Y) relationship between them”2 (see Nemo 2001, 2003). 3. the possibility to use a morpheme only depends on the possibility to relate these indications to contextual elements, i.e. in each context the X and Y (etc.) must be found. This indexical process is called contextual unification; 4. the morpheme’s encoded linguistic meaning (i.e. its signification) does not specify or limit the kind of elements that may be contextually unified with the indications it provides; 5. interpretation, hence, may be described as a process of unification of semantic indications and contextual elements; 6. contextual interpretations may be memorised in the lexical memory (i.e contextual interpretations may become conventional interpretations). .. Indicational-indexical semantics: Methodology According to Linguistic Semantics and IIS: –

– – – – –

signification is accessible only by considering the diversity of uses of a semantic item (Benveniste 1954; Ducrot 1987; Cadiot 1994; Pustejovsky 1995; Bouchard 1995; Nemo 2001); signification must account for all the uses; signification serves as a tool in the interpretation process whose result is a sense (Ducrot 1987); the semantic nature of signification is different from the semantic nature of senses; the lexicalisation of an interpretation occurs whenever a use becomes a usage; its becoming a usage, and thus becoming conventionalised, does not alter the contextual nature of an interpretation;

Within such a view, interpretation may be described as a process: f(m, cstr, ctxt) = i → s

which associates a morpheme m with a construction cstr in a context ctxt, leading to a contextual interpretation i which may become lexicalised and thus become the lexical meaning s of the lexical unit which has been produced. (Gasiglia, Nemo & Cadiot 2001).

. For instance (Nemo 2006), it can be shown that English morpheme but encodes the indication that “there is a process X which has been (is/could be/could have been/should be) stopped by a stopping factor Y”.

TSL[v.20020404] Prn:30/05/2007; 10:15

F: PB16112.tex / p.11 (205)

Reconsidering the discourse marking hypothesis 

.. The morpheme/lexeme/construction distinction A consequence of the IIS description of interpretation and lexicalization in terms of: f(m, cstr, ctxt) = i → s

is that it becomes possible: –

– –

to define morphemes as form/meaning pairs, which exist independently of the context of insertion, and which consequently are purely semantic units with no syntax; to define constructions as form/meaning pairs, which exist independently of the specific semantic material they consist of (Goldberg 1995); to define lexemes as morpheme/constructions pairs.

A complete and detailed illustration of the way IIS and the morpheme/construction distinction will be provided here for the English morpheme even – exactly as it has been done for the English morpheme but (Nemo 2006), with all its nonconnectives and connective uses (e.g. with the lexemes but=almost, but=only, but=without, but=except, but=rather, etc.). It shows that the encoded signification of these morphemes in English is neither connective nor meta-communicational, but that it receives connective or discourse-related interpretations when inserted in connective positions. This allows us to formulate a “Generalized Coercion Hypothesis” according to which “the signification of a morpheme is never a prefiguration of its categorial and contextual uses”. And it also accounts for the possibility to use connectives in (almost) any kind of discourse relating contexts, as illustrated above in Section 1.1.1. . The indicational-indexical semantics of even Within the IIS approach, as far as even is concerned, there is a single morpheme even which encodes the same indication(s) in all its uses, and different lexemes of even which are associated with lexicalized meanings (senses) which are constructional and contextual interpretations of the indications encoded by the morpheme. So let’s try to find out what the indication(s) encoded by even could be and how this encoded meaning is interpreted in a connective use. First of all, let’s show that even by considering a single adjectival example, it becomes possible to account for most of the semantics of the adverbial and connective uses. Let us start for instance with the NP an even surface which we can find in example (26). To understand what an even surface is, i.e. to understand that an even surface is a flat surface, it is necessary: –

to compare the levels (L) of all the points (P1, P2, Pn) of that surface

TSL[v.20020404] Prn:30/05/2007; 10:15

F: PB16112.tex / p.12 (206)

 François Nemo

–

to get the following result L(P1) = L(P2)= L(P3), etc.,

Doing so yields as a result the truth conditions of the use of the adjective in such a context. If we move now to example (29) and consider the NP an even match, we are going through the same process of comparing the respective values of the two teams by comparing NH1, the handicap of team 1, NH2, the handicap of team 2, and observing that NH1 is no different from NH2. Of course, if we compare with example (26), the nature of what is compared is completely different, but the process and its result are identical. On the contrary, the fact that the meaning of even is flat when what is compared are levels is an illustration of the contextual nature of this interpretation, because the concept flat requires the concept of level. If we apply now to additive uses – such as (24) “Even his mother had noticed it” or (25) – what we have just found, namely that at least two things are compared and that the result of the comparison is that there is no difference between them, it becomes possible to account for most of the interpretations of our sentences: if even indicates that there is no difference between C1 and C2 and C1 is “his mother” then it implies that there are other Cs who also “noticed it” or “do not believe him” (for a standard formal description of this presupposition effect, see König 1991: 70). Thus, even if the comparison set is different (people) and even if the nature of the comparison (noticing or not, believing or not) is different, we have the same result: there are no differences between all the people involved in terms of noticing something or not, or believing somebody or not. Of course, there is something else in the interpretation of that use of even, namely that because mothers are usually inclined to believe their sons, the fact that a mother does not is the highest point of an argumentative scale (Anscombre & Ducrot 1976; Kay 1987). But this semantic content is not encoded by even itself but a mere conversational consequence of what is said: to say that there is no difference between a mother and other people as far as believing her son is concerned, implies that we would expect such a difference to exist, and it is this pragmatic inference which leads us to the common ground idea that mothers are inclined to believe their children. It can thus be shown that the scalar value of maximum is ultimately derived from the indication/assertion that there are no differences. And this is precisely why, as mentioned by Ducrot (1987), we must not mistake the tool/scaffolding (the indication) with the result/building it allows us to build. The scalar value for instance of even can be shown to be only a consequence of the fact that among all the compared Cs, it is the case that we initially think that some are more likely to be different or to behave differently. Turning now to the connective use of even if in (15), we can observe something which is now familiar: saying Even if they get me, I’ll go down fighting means that whether they get me (C1) or not (C2) it will make no difference for me as far as

TSL[v.20020404] Prn:30/05/2007; 10:15

F: PB16112.tex / p.13 (207)

Reconsidering the discourse marking hypothesis 

fighting is concerned. The only thing new here is that we are now not comparing two existing realities any more, as in the previous uses, but two alternatives. . Testing the indication As we have just seen, it is possible to postulate a semantic continuity between the connective and non-connective uses whenever even is used. Specifically whenever even is used in a new way whose interpretation has not yet been stored in the lexicon, the interpreter knows that (s)he has to find in the cotext or context two comparanda C1 and C2 such that there is no difference between them (∆C1/C2 = 0), C1 and C2 being either two realities or two alternatives. So let us see the role of this indication in each of the examples (11) to (29). First of all, in all examples (11) to (20), no matter how different they are, it can be shown that it is always the case that it is “asserted” that there is something X for which it makes no difference whether Y is the case or not.3 For instance, if we consider example (12), we knew that if the mother asked for her back, we would have to give her up cheerfully, even though Joan and Susan were inseparable), the unification pattern is “If Z then X whether Y or Not Y” with X = giving her back and Y = Joan and Susan are inseparable. Similarly, if we consider example (11) (the habit of returning to the sea each year to spawn has never been broken even though it now involves such an immense journey), the unification pattern is “X whether Y or Not Y” with X = not breaking and Y = it involves an immense journey. In examples (13) and (14), for instance even if he was trying to show off, I still respected him, the sequentially opposite pattern may be found, in which it is “asserted” that the fact that he was trying to show off (=Y) or not made no difference as far as respecting him (=X) was concerned. With example (16), the same pattern occurs, which says that no matter whether you have a good reason (=Y) or not to have lied (the explanation in the classical sense of Section 1.1.1.), you will be forced to lie (=X). In (17), what is “asserted” is that getting married did not change his support. And in (19) or (20), what is said is that admitting (=Y) something which has just been under discussion or not, does not change for instance the fact that he has to buy (=X). As for the temporal connective even as, for instance in (18), the comparison set is quite different, because what is compared in this case are not two alternatives but . I’m not claiming here that the morphemic meaning would provide a sufficient account of the interpretation of the utterances in question. Instead, I am only arguing that the same indication can be found in all examples. Accounting for the complete interpretation of each utterance would require to describe one by one all the constructions in which the morpheme is inserted.

TSL[v.20020404] Prn:30/05/2007; 10:15

F: PB16112.tex / p.14 (208)

 François Nemo

the moments at which two events are taking place. But the result is the same, and even in such a case signals that the two moments are simultaneous. The use of even as to mark the temporal concomitancy of two events is thus directly predictable from a ∆ t1/t2 = 0 comparison set. As for the additive uses, they clearly show that the “assertion” that human beings are not different from all the other living beings (as far as being interdependent is concerned) is indeed central to the interpretation of (21). Exactly as in (22) in which the absence of difference concerns the fragility of the hairs on spiders’ legs compared to other body parts. Non-connective uses are even more straightforward, since their meaning directly implies both the existence of a comparison and the zero nature of its result: saying that the odds are even or that there is a necessity to even out inequalities supposes either that the probabilities are equal or that something that is not equal must be equalized. It is thus possible to relate all the uses of even by a single semantic equation, exactly as an equation relates points. But it is also the case in connective uses that the link which is built in each use of even is doubled by an independent connective interpretation of what it may introduce: even may introduce an explanation in (16), a background in (12), an elaboration in a sentence such as “Nobody’s asked any nasty questions, not even the cops”, or a contrast in “I cannot even if I would like to”, etc. . Morphemes and constructions The study of even presented here serves as a direct illustration of the Generalized Coercion Hypothesis (GCH) introduced above, according to which the encoded meaning (signification) of a morpheme is not a prefiguration of any of its uses: a morpheme such as even instructs the hearer/interpreter to look for a set of comparanda and indicates that there is no difference between what is compared. Since the result of the process, i.e. the identification of the compared items and of the compared property, will change from context to context and from construction to construction, and is hence not encoded, we have simultaneously a single morpheme and various lexemes of even. Secondly, the study serves as a direct explanation for the falsification of the DM hypothesis by means of tests such as those presented in Section 1.1.1 above. From considering discourse relations as constructions on the one hand and from the GCH on the other hand, will follow that when it comes to interpreting a sequence of utterances, we simultaneously have to interpret the specific DRs which exist between the utterances and the use of the morpheme inserted in connective position, so that we actually are exactly in the situation described by the double link hypothesis.

TSL[v.20020404] Prn:30/05/2007; 10:15

F: PB16112.tex / p.15 (209)

Reconsidering the discourse marking hypothesis 

. Conclusion The Discourse (Relation) Marking Hypothesis is a common sense hypothesis whose only limit is that it is easily falsified whenever it is actually tested. Understanding that discourse connectives are not discourse markers, but morphemes inserted in “contribution modifying constructions”, allows us to account for apparently problematic observations. This finding allows the linguist to clearly separate two different issues, namely: – –

the identification of the encoded meaning (signification) and interpretations of the morpheme by studying all its connective and non-connective uses; the identification of contribution modifying constructions, regardless of the semantic material which is associated with them.

Using this methodology allows the investigation of the two links which are present in all the uses of all discourse connectives. The puzzling semantics, pragmatics and syntax of discourse connectives is thus not an indicator for a peripheral role in discourse; instead it is an indicator that they are both agents and patients in the process which takes place during the interpretation process and before the lexicalization of their connective uses. Therefore, discourse connectives, despite being initially morpheme/construction pairs, may simultaneously crystallize in the lexicon as landmarks of discourse and develop new uses.

References Anscombre, J.-C. & Ducrot, O. 1976. “L’argumentation dans la langue.” Langages, 42, 5–27. Anscombre, J.-C. & Ducrot, O. 1983. L’argumentation dans la langue. Bruxelles: Mardaga. Benveniste E., 1954. “Problèmes sémantiques de la reconstruction.” In Problèmes de linguistique générale, 1. Paris: Gallimard, 1966. Bouchard D. 1995. The Semantics of Syntax. Chicago: Chicago University Press. Cadiot, P. 1994. “Représentations d’objets et sémantique lexicale: Qu’est-ce qu’une boîte ?” Journal of French Language Studies 4, 1–23. Culioli, A. 1999. “Even though, even if; as though, as if.” Pour une linguistique de l’énonciation t. 3, Paris: Ophrys. 177–181. Ducrot, O. 1980. “Analyses pragmatiques.” Communications 32, p. 11–60. Ducrot O, 1987. “L’interprétation comme point de départ imaginaire de la sémantique.” In Dire et ne pas dire ([1972], 1991. Paris: Hermann. Ducrot, O. & alii. 1980. Les mots du discours, Paris: Minuit. Fraser, B. 1998. “Contrastive Discourse Markers in English.” In Andreas H. Jucker and Yael Ziv (eds.) Discourse Markers: Descriptions and Theory. [Pragmatics & Beyond, 57] Amsterdam: John Benjamins. Gasiglia, N., Nemo, F. & Cadiot, P. 2001. “Meaning and the generation of reference.” In Generative Approaches to the Lexicon (Bouillon, Pierette, ed.), Université de Genève, Genève.

TSL[v.20020404] Prn:30/05/2007; 10:15

F: PB16112.tex / p.16 (210)

 François Nemo

Goldberg, A. 1995. Constructions. Chicago: University of Chicago Press. Jayez, J. & Rossari, C. 1998. “Discourse relations vs. discourse marker relations.” ACL’98 Workshop on Discourse Relations and Discourse Markers, 72–78. Kay, P. 1987. “Even.” Linguistics and Philosophy, 13, 59–111. König, E. 1991. The Meaning of Focus Particles. A Comparative Perspective. London and New York: Routledge. Luscher, J. M. 1994. “Les marques de connexion: des guides pour l’interprétation.” In Langage et pertinence, Nancy: Presses Universitaires de Nancy. 175–226. Nemo, F. 1999. “The Pragmatics of Signs, The Semantics of Relevance, and The Semantic/ Pragmatic Interface” in The Semantics-Pragmatics Interface from Different points of View, CRiSPI Series, Amsterdam: Elsevier Science. pp. 343–417. Nemo, F. 2001. “Pour une approche indexicale (et non procédurale) des instructions sémantiques.” Revue de Sémantique et de Pragmatique, n◦ 9–10: 195–218. Nemo, F. 2002. “But (and mais) as morpheme(s).” Delta (Sao Paulo), n◦ 18–2. 87–114. Nemo, F. 2003. “Indexicalité, unification contextuelle et constitution extrinsèque du référent.” Langages, 150. 88–105. Nemo F., 2006. “Discourse Particles as Morphemes and as Constructions.” In Approaches to Discourse Particles (K. Fischer, ed.), Amsterdam: Elsevier. 415–449. Nølke, H. 1993. Le regard du locuteur. Paris: Kimé. Pustejovsky, J. 1995. The Generative Lexicon. Boston: Cambridge: MIT Press. Rouchota, W. 1996. “Discourse connectives: What do they link?.” UCL Working Papers in Linguistics 8. Rossari, C. 1994. Les opérations de reformulation. Peter Lang, Berne. Tomasello, M. 1995. “Joint attention as social cognition.” In C. Moore and P. Dunham (eds.) Joint Attention: Its Origins and Role in Development. Hillsdale, NJ: Lawrence Erlbaum, 103– 130. Wilson, D. & Sperber. D. 1990. “Forme linguistique et pertinence.” Cahiers de linguistique francaise 11: 13–35. Wilson, D. & Sperber. D. 1993. “Linguistic Form and Relevance.” Lingua 90, 1–25.

TSL[v.20020404] Prn:1/06/2007; 12:31

F: PB161IND.tex / p.1 (211)

Index

A adverb , ,  ,  afference, afferent – Aktionsart , , –,  anaphoric reference , , , , ,  and –, ,  antilinear ,  argumentation , , , , , ,  argumentative domain – argumentative force ,  aspect , , ,  assertion , , , , , , ,  B background , – (see also information structure) because –,  Biber  Blakemore  Bolinger  Boundary (see notional domain) British National Corpus (BNC) , –, , , ,  but , , , , –,  C causal relation , , –, , ,  Centre (see notional domain) centring  co-énonciateur , ,  (see also Theory of Enunciative Operations) coenunciation  coercion ,  cognitive control  coherence  cohesion , , 

collocational framework – collocations , , ,  concession ,  construction –, –, ,  continuity ,  contrast , , , –,  contribution  Coordination –, , ,  Coordinate Structure Constraint ,  Coordination of Likes ,  pseudo-coordination –, – reduplicative coordination , – counter-expectation –, ,  counter-subordination ,  Couper-Kuhlen  Culioli ,  D degree modifier , ,  deictic, deixis ,  discontinuity ,  discordance , – discourse progression , , ,  dynamicity ,  lack of dynamicity  E efference, efferent – emergent grammar , , ,  Emonds  endorsement , , , ,  subjective endorsement , ,  engagement theory 

énonciateur ,  (see also Theory of Enunciative Operations) evaluation, evaluative , ,  (see also modality) even , –, – event structure ,  Exterior (see notional domain) extraction ,  F floor-keeping manoeuvre  focalisation , , ,  foregrounding / backgrounding (see information structure) frame , – Fraser  functional split  G generic statements ,  Gilbert  grammaticalization – Grice  H heteroglossic  Hirtle  hyperbole  I implicature ,  implicit , , – indefinite ,  indexical ,  Indicational-Indexical Semantics – information structure , ,  Interior (see notional domain) interpersonal relations , ,  (see also intersubjective)

TSL[v.20020404] Prn:1/06/2007; 12:31

F: PB161IND.tex / p.2 (212)

 Index inter-propositional relations , ,  inter-subjective , , , , , ,  (see also interpersonal relations) intonation , ,  (see also prosody) involvement , ,  (see also subjectivity) it ,  J Jespersen  judgement , ,  (see also modality) just ,  justification , , , ,  K Kjellmer  Kuha  L Levinson  locating operation , , –,  M mental movement – metacommunicative ,  metalinguistic , , , , , , ,  metalinguistic rectification  modal, modality , , –, , , , , , , , , , ,  (see also evaluation, judgement, subjectivity) multifunctionality , , , ,  N negation , ,  notional domain , , , , – (see also Theory

of Enunciative Operations) Boundary , ,  Centre  Exterior , ,  Interior , , ,  offline position , ,  O ordering  order of constituents , , , , , , , ,  temporal , , , ,  Östman ,  P performance error  polyfunctionality (see multifunctionality) polysemy  posture verbs  preconstruction, pre-constructed , , , ,  presupposition ,  Prince  prosody , , ,  (see also intonation) pronoun (see also it) relative pronoun , –,  resumptive pronoun – prototype, prototypical , ,  pseudo-coordination (see coordination) quantity/quality ,  R rather –, ,  would rather – Recanati ,  reduplicative coordination (see coordination) relative clause –, – gap-filled relative clauses ,  gapless relative clauses , , 

Renouf & Sinclair  S Schiffrin , , ,  Schleppegrell ,  Schmid  semantic bleaching , ,  semantic change  shell noun  so – sooner  spiral , , , , – still , – subjective endorsement (see endorsement) subjectivity , , ,  (see also modality) subordination , , , , , , , ,  Svartvik  T telicity  that  that clause ,  Theory of Enunciative Operations , ,  Thompson ,  threshold – trigger  turn-taking , , ,  V validation ,  Vallée  Vendler  which (see also relative pronoun) – W White  Y yet , , – and yet 

Pragmatics & Beyond New Series A complete list of titles in this series can be found on the publishers’ website, www.benjamins.com 164 Englebretson, Robert (ed.): Stancetaking in Discourse. Subjectivity, Evaluation, Interaction. Expected October 2007 163 Lytra, Vally: Play Frames and Social Identities. Contact encounters in a Greek primary school. Expected September 2007 162 Fetzer, Anita (ed.): Context and Appropriateness. Micro meets macro. vi, 260 pp. + index. Expected July 2007 161 Celle, Agnès and Ruth Huart (eds.): Connectives as Discourse Landmarks. 2007. viii, 212 pp. 160 Fetzer, Anita and Gerda Eva Lauerbach (eds.): Political Discourse in the Media. Cross-cultural perspectives. viii, 376 pp. + index. Expected July 2007 159 Maynard, Senko K.: Linguistic Creativity in Japanese Discourse. Exploring the multiplicity of self, perspective, and voice. xiv, 354 pp. + index. Expected July 2007 158 Walker, Terry: Thou and You in Early Modern English Dialogues. Trials, Depositions, and Drama Comedy. 2007. xx, 339 pp. 157 Crawford Camiciottoli, Belinda: The Language of Business Studies Lectures. A corpus-assisted analysis. 2007. xvi, 236 pp. 156 Vega Moreno, Rosa E.: Creativity and Convention. The pragmatics of everyday figurative speech. xii, 243 pp. + index. Expected July 2007 155 Hedberg, Nancy and Ron Zacharski (eds.): The Grammar–Pragmatics Interface. Essays in honor of Jeanette K. Gundel. 2007. viii, 345 pp. 154 Hübler, Axel: The Nonverbal Shift in Early Modern English Conversation. 2007. x, 281 pp. 153 Arnovick, Leslie K.: Written Reliquaries. The resonance of orality in medieval English texts. 2006. xii, 292 pp. 152 Warren, Martin: Features of Naturalness in Conversation. 2006. x, 272 pp. 151 Suzuki, Satoko (ed.): Emotive Communication in Japanese. 2006. x, 234 pp. 150 Busse, Beatrix: Vocative Constructions in the Language of Shakespeare. 2006. xviii, 525 pp. 149 Locher, Miriam A.: Advice Online. Advice-giving in an American Internet health column. 2006. xvi, 277 pp. 148 Fløttum, Kjersti, Trine Dahl and Torodd Kinn: Academic Voices. Across languages and disciplines. 2006. x, 309 pp. 147 Hinrichs, Lars: Codeswitching on the Web. English and Jamaican Creole in e-mail communication. 2006. x, 302 pp. 146 Tanskanen, Sanna-Kaisa: Collaborating towards Coherence. Lexical cohesion in English discourse. 2006. ix, 192 pp. 145 Kurhila, Salla: Second Language Interaction. 2006. vii, 257 pp. 144 Bührig, Kristin and Jan D. ten Thije (eds.): Beyond Misunderstanding. Linguistic analyses of intercultural communication. 2006. vi, 339 pp. 143 Baker, Carolyn, Michael Emmison and Alan Firth (eds.): Calling for Help. Language and social interaction in telephone helplines. 2005. xviii, 352 pp. 142 Sidnell, Jack: Talk and Practical Epistemology. The social life of knowledge in a Caribbean community. 2005. xvi, 255 pp. 141 Zhu, Yunxia: Written Communication across Cultures. A sociocognitive perspective on business genres. 2005. xviii, 216 pp. 140 Butler, Christopher S., María de los Ángeles Gómez-González and Susana M. Doval-Suárez (eds.): The Dynamics of Language Use. Functional and contrastive perspectives. 2005. xvi, 413 pp. 139 Lakoff, Robin T. and Sachiko Ide (eds.): Broadening the Horizon of Linguistic Politeness. 2005. xii, 342 pp. 138 Müller, Simone: Discourse Markers in Native and Non-native English Discourse. 2005. xviii, 290 pp. 137 Morita, Emi: Negotiation of Contingent Talk. The Japanese interactional particles ne and sa. 2005. xvi, 240 pp. 136 Sassen, Claudia: Linguistic Dimensions of Crisis Talk. Formalising structures in a controlled language. 2005. ix, 230 pp.

135 Archer, Dawn: Questions and Answers in the English Courtroom (1640–1760). A sociopragmatic analysis. 2005. xiv, 374 pp. 134 Skaffari, Janne, Matti Peikola, Ruth Carroll, Risto Hiltunen and Brita Wårvik (eds.): Opening Windows on Texts and Discourses of the Past. 2005. x, 418 pp. 133 Marnette, Sophie: Speech and Thought Presentation in French. Concepts and strategies. 2005. xiv, 379 pp. 132 Onodera, Noriko O.: Japanese Discourse Markers. Synchronic and diachronic discourse analysis. 2004. xiv, 253 pp. 131 Janoschka, Anja: Web Advertising. New forms of communication on the Internet. 2004. xiv, 230 pp. 130 Halmari, Helena and Tuija Virtanen (eds.): Persuasion Across Genres. A linguistic approach. 2005. x, 257 pp. 129 Taboada, María Teresa: Building Coherence and Cohesion. Task-oriented dialogue in English and Spanish. 2004. xvii, 264 pp. 128 Cordella, Marisa: The Dynamic Consultation. A discourse analytical study of doctor–patient communication. 2004. xvi, 254 pp. 127 Brisard, Frank, Michael Meeuwis and Bart Vandenabeele (eds.): Seduction, Community, Speech. A Festschrift for Herman Parret. 2004. vi, 202 pp. 126 Wu, Yi’an: Spatial Demonstratives in English and Chinese. Text and Cognition. 2004. xviii, 236 pp. 125 Lerner, Gene H. (ed.): Conversation Analysis. Studies from the first generation. 2004. x, 302 pp. 124 Vine, Bernadette: Getting Things Done at Work. The discourse of power in workplace interaction. 2004. x, 278 pp. 123 Márquez Reiter, Rosina and María Elena Placencia (eds.): Current Trends in the Pragmatics of Spanish. 2004. xvi, 383 pp. 122 González, Montserrat: Pragmatic Markers in Oral Narrative. The case of English and Catalan. 2004. xvi, 410 pp. 121 Fetzer, Anita: Recontextualizing Context. Grammaticality meets appropriateness. 2004. x, 272 pp. 120 Aijmer, Karin and Anna-Brita Stenström (eds.): Discourse Patterns in Spoken and Written Corpora. 2004. viii, 279 pp. 119 Hiltunen, Risto and Janne Skaffari (eds.): Discourse Perspectives on English. Medieval to modern. 2003. viii, 243 pp. 118 Cheng, Winnie: Intercultural Conversation. 2003. xii, 279 pp. 117 Wu, Ruey-Jiuan Regina: Stance in Talk. A conversation analysis of Mandarin final particles. 2004. xvi, 260 pp. 116 Grant, Colin B. (ed.): Rethinking Communicative Interaction. New interdisciplinary horizons. 2003. viii, 330 pp. 115 Kärkkäinen, Elise: Epistemic Stance in English Conversation. A description of its interactional functions, with a focus on I think. 2003. xii, 213 pp. 114 Kühnlein, Peter, Hannes Rieser and Henk Zeevat (eds.): Perspectives on Dialogue in the New Millennium. 2003. xii, 400 pp. 113 Panther, Klaus-Uwe and Linda L. Thornburg (eds.): Metonymy and Pragmatic Inferencing. 2003. xii, 285 pp. 112 Lenz, Friedrich (ed.): Deictic Conceptualisation of Space, Time and Person. 2003. xiv, 279 pp. 111 Ensink, Titus and Christoph Sauer (eds.): Framing and Perspectivising in Discourse. 2003. viii, 227 pp. 110 Androutsopoulos, Jannis K. and Alexandra Georgakopoulou (eds.): Discourse Constructions of Youth Identities. 2003. viii, 343 pp. 109 Mayes, Patricia: Language, Social Structure, and Culture. A genre analysis of cooking classes in Japan and America. 2003. xiv, 228 pp. 108 Barron, Anne: Acquisition in Interlanguage Pragmatics. Learning how to do things with words in a study abroad context. 2003. xviii, 403 pp. 107 Taavitsainen, Irma and Andreas H. Jucker (eds.): Diachronic Perspectives on Address Term Systems. 2003. viii, 446 pp. 106 Busse, Ulrich: Linguistic Variation in the Shakespeare Corpus. Morpho-syntactic variability of second person pronouns. 2002. xiv, 344 pp. 105 Blackwell, Sarah E.: Implicatures in Discourse. The case of Spanish NP anaphora. 2003. xvi, 303 pp. 104 Beeching, Kate: Gender, Politeness and Pragmatic Particles in French. 2002. x, 251 pp.

103 Fetzer, Anita and Christiane Meierkord (eds.): Rethinking Sequentiality. Linguistics meets conversational interaction. 2002. vi, 300 pp. 102 Leafgren, John: Degrees of Explicitness. Information structure and the packaging of Bulgarian subjects and objects. 2002. xii, 252 pp. 101 Luke, K. K. and Theodossia-Soula Pavlidou (eds.): Telephone Calls. Unity and diversity in conversational structure across languages and cultures. 2002. x, 295 pp. 100 Jaszczolt, Katarzyna M. and Ken Turner (eds.): Meaning Through Language Contrast. Volume 2. 2003. viii, 496 pp. 99 Jaszczolt, Katarzyna M. and Ken Turner (eds.): Meaning Through Language Contrast. Volume 1. 2003. xii, 388 pp. 98 Duszak, Anna (ed.): Us and Others. Social identities across languages, discourses and cultures. 2002. viii, 522 pp. 97 Maynard, Senko K.: Linguistic Emotivity. Centrality of place, the topic-comment dynamic, and an ideology of pathos in Japanese discourse. 2002. xiv, 481 pp. 96 Haverkate, Henk: The Syntax, Semantics and Pragmatics of Spanish Mood. 2002. vi, 241 pp. 95 Fitzmaurice, Susan M.: The Familiar Letter in Early Modern English. A pragmatic approach. 2002. viii, 263 pp. 94 McIlvenny, Paul (ed.): Talking Gender and Sexuality. 2002. x, 332 pp. 93 Baron, Bettina and Helga Kotthoff (eds.): Gender in Interaction. Perspectives on femininity and masculinity in ethnography and discourse. 2002. xxiv, 357 pp. 92 Gardner, Rod: When Listeners Talk. Response tokens and listener stance. 2001. xxii, 281 pp. 91 Gross, Joan: Speaking in Other Voices. An ethnography of Walloon puppet theaters. 2001. xxviii, 341 pp. 90 Kenesei, István and Robert M. Harnish (eds.): Perspectives on Semantics, Pragmatics, and Discourse. A Festschrift for Ferenc Kiefer. 2001. xxii, 352 pp. 89 Itakura, Hiroko: Conversational Dominance and Gender. A study of Japanese speakers in first and second language contexts. 2001. xviii, 231 pp. 88 Bayraktaroğlu, Arın and Maria Sifianou (eds.): Linguistic Politeness Across Boundaries. The case of Greek and Turkish. 2001. xiv, 439 pp. 87 Mushin, Ilana: Evidentiality and Epistemological Stance. Narrative Retelling. 2001. xviii, 244 pp. 86 Ifantidou, Elly: Evidentials and Relevance. 2001. xii, 225 pp. 85 Collins, Daniel E.: Reanimated Voices. Speech reporting in a historical-pragmatic perspective. 2001. xx, 384 pp. 84 Andersen, Gisle: Pragmatic Markers and Sociolinguistic Variation. A relevance-theoretic approach to the language of adolescents. 2001. ix, 352 pp. 83 Márquez Reiter, Rosina: Linguistic Politeness in Britain and Uruguay. A contrastive study of requests and apologies. 2000. xviii, 225 pp. 82 Khalil, Esam N.: Grounding in English and Arabic News Discourse. 2000. x, 274 pp. 81 Di Luzio, Aldo, Susanne Günthner and Franca Orletti (eds.): Culture in Communication. Analyses of intercultural situations. 2001. xvi, 341 pp. 80 Ungerer, Friedrich (ed.): English Media Texts – Past and Present. Language and textual structure. 2000. xiv, 286 pp. 79 Andersen, Gisle and Thorstein Fretheim (eds.): Pragmatic Markers and Propositional Attitude. 2000. viii, 273 pp. 78 Sell, Roger D.: Literature as Communication. The foundations of mediating criticism. 2000. xiv, 348 pp. 77 Vanderveken, Daniel and Susumu Kubo (eds.): Essays in Speech Act Theory. 2002. vi, 328 pp. 76 Matsui, Tomoko: Bridging and Relevance. 2000. xii, 251 pp. 75 Pilkington, Adrian: Poetic Effects. A relevance theory perspective. 2000. xiv, 214 pp. 74 Trosborg, Anna (ed.): Analysing Professional Genres. 2000. xvi, 256 pp. 73 Hester, Stephen K. and David Francis (eds.): Local Educational Order. Ethnomethodological studies of knowledge in action. 2000. viii, 326 pp. 72 Marmaridou, Sophia S.A.: Pragmatic Meaning and Cognition. 2000. xii, 322 pp. 71 Gómez-González, María de los Ángeles: The Theme–Topic Interface. Evidence from English. 2001. xxiv, 438 pp.

70 Sorjonen, Marja-Leena: Responding in Conversation. A study of response particles in Finnish. 2001. x, 330 pp. 69 Noh, Eun-Ju: Metarepresentation. A relevance-theory approach. 2000. xii, 242 pp. 68 Arnovick, Leslie K.: Diachronic Pragmatics. Seven case studies in English illocutionary development. 2000. xii, 196 pp. 67 Taavitsainen, Irma, Gunnel Melchers and Päivi Pahta (eds.): Writing in Nonstandard English. 2000. viii, 404 pp. 66 Jucker, Andreas H., Gerd Fritz and Franz Lebsanft (eds.): Historical Dialogue Analysis. 1999. viii, 478 pp. 65 Cooren, François: The Organizing Property of Communication. 2000. xvi, 272 pp. 64 Svennevig, Jan: Getting Acquainted in Conversation. A study of initial interactions. 2000. x, 384 pp. 63 Bublitz, Wolfram, Uta Lenk and Eija Ventola (eds.): Coherence in Spoken and Written Discourse. How to create it and how to describe it. Selected papers from the International Workshop on Coherence, Augsburg, 24-27 April 1997. 1999. xiv, 300 pp. 62 Tzanne, Angeliki: Talking at Cross-Purposes. The dynamics of miscommunication. 2000. xiv, 263 pp. 61 Mills, Margaret H. (ed.): Slavic Gender Linguistics. 1999. xviii, 251 pp. 60 Jacobs, Geert: Preformulating the News. An analysis of the metapragmatics of press releases. 1999. xviii, 428 pp. 59 Kamio, Akio and Ken-ichi Takami (eds.): Function and Structure. In honor of Susumu Kuno. 1999. x, 398 pp. 58 Rouchota, Villy and Andreas H. Jucker (eds.): Current Issues in Relevance Theory. 1998. xii, 368 pp. 57 Jucker, Andreas H. and Yael Ziv (eds.): Discourse Markers. Descriptions and theory. 1998. x, 363 pp. 56 Tanaka, Hiroko: Turn-Taking in Japanese Conversation. A Study in Grammar and Interaction. 2000. xiv, 242 pp. 55 Allwood, Jens and Peter Gärdenfors (eds.): Cognitive Semantics. Meaning and cognition. 1999. x, 201 pp. 54 Hyland, Ken: Hedging in Scientific Research Articles. 1998. x, 308 pp. 53 Mosegaard Hansen, Maj-Britt: The Function of Discourse Particles. A study with special reference to spoken standard French. 1998. xii, 418 pp. 52 Gillis, Steven and Annick De Houwer (eds.): The Acquisition of Dutch. With a Preface by Catherine E. Snow. 1998. xvi, 444 pp. 51 Boulima, Jamila: Negotiated Interaction in Target Language Classroom Discourse. 1999. xiv, 338 pp. 50 Grenoble, Lenore A.: Deixis and Information Packaging in Russian Discourse. 1998. xviii, 338 pp. 49 Kurzon, Dennis: Discourse of Silence. 1998. vi, 162 pp. 48 Kamio, Akio: Territory of Information. 1997. xiv, 227 pp. 47 Chesterman, Andrew: Contrastive Functional Analysis. 1998. viii, 230 pp. 46 Georgakopoulou, Alexandra: Narrative Performances. A study of Modern Greek storytelling. 1997. xvii, 282 pp. 45 Paltridge, Brian: Genre, Frames and Writing in Research Settings. 1997. x, 192 pp. 44 Bargiela-Chiappini, Francesca and Sandra J. Harris: Managing Language. The discourse of corporate meetings. 1997. ix, 295 pp. 43 Janssen, Theo and Wim van der Wurff (eds.): Reported Speech. Forms and functions of the verb. 1996. x, 312 pp. 42 Kotthoff, Helga and Ruth Wodak (eds.): Communicating Gender in Context. 1997. xxvi, 424 pp. 41 Ventola, Eija and Anna Mauranen (eds.): Academic Writing. Intercultural and textual issues. 1996. xiv, 258 pp. 40 Diamond, Julie: Status and Power in Verbal Interaction. A study of discourse in a close-knit social network. 1996. viii, 184 pp. 39 Herring, Susan C. (ed.): Computer-Mediated Communication. Linguistic, social, and cross-cultural perspectives. 1996. viii, 326 pp. 38 Fretheim, Thorstein and Jeanette K. Gundel (eds.): Reference and Referent Accessibility. 1996. xii, 312 pp. 37 Carston, Robyn and Seiji Uchida (eds.): Relevance Theory. Applications and implications. 1998. x, 300 pp.

Connectives As Discourse Landmarks (Pragmatics and Beyond New Series)

South Slavic Discourse Particles (Pragmatics and Beyond. New Series)

Constraints in Discourse 2 (Pragmatics and Beyond. New Series)

Constraints in Discourse (Pragmatics and Beyond New Series, 172)

Discourse Constructions of Youth Identities (Pragmatics & Beyond New Series)

Discourse Constructions of Youth Identities (Pragmatics & Beyond New Series)

Perspectives on Semantics, Pragmatics and Discourse: A Festschrift for Ferenc Kiefer (Pragmatics and Beyond New Series)

Discourse Markers in Native And Non-native English Discourse (Pragmatics and Beyond. New Series)

Early Modern English News Discourse: Newspapers, pamphlets and scientific news discourse (Pragmatics and Beyond New Series)

Japanese Discourse Markers: Synchronic And Diachronic Discourse Analysis (Pragmatics and Beyond New Series)

Cognitive Semantics: Meaning and Cognition (Pragmatics and Beyond New Series)

Verbal Aspect in Discourse (Pragmatics & Beyond)

Talking Gender and Sexuality (Pragmatics and Beyond New Series)

Mental Spaces in Discourse and Interaction (Pragmatics and Beyond New Series)

Impoliteness in Interaction (Pragmatics and Beyond New Series)

Humor in Interaction (Pragmatics and Beyond New Series)

Cognitive Semantics: Meaning and Cognition (Pragmatics & Beyond New Series)

Style Shifting in Japanese (Pragmatics and Beyond New Series)

Second Language Interaction (Pragmatics and Beyond New Series)

Youngspeak in a Multilingual Perspective (Pragmatics and Beyond New Series)

Recontextualizing Context: Grammaticality Meets Appropriateness (Pragmatics and Beyond New Series)

Metapragmatics in Use (Pragmatics and Beyond New Series)

Discourse Perspectives on English: Medieval to Modern (Pragmatics and Beyond New Series)

Collaborating towards Coherence: Lexical cohesion in English discourse (Pragmatics and Beyond New Series)

Creativity and Convention: The pragmatics of everyday figurative speech (Pragmatics and Beyond New Series)

Hedging in Scientific Research Articles (Pragmatics & Beyond New Series)

Political Discourse in Transition in Europe 1989-1991 (Pragmatics & Beyond New Series)

Discourse Description: Diverse Linguistic Analyses of a Fund-raising Text (Pragmatics & Beyond New Series)

Political Discourse in the Media: Cross-cultural Perspectives (Pragmatics & Beyond New Series)

Managing Language: The Discourse of Corporate Meetings (Pragmatics & Beyond New Series)

Implicatures in Discourse: The Case of Spanish NP Anaphora (Pragmatics & Beyond New Series)

Connectives As Discourse Landmarks (Pragmatics and Beyond New Series)

South Slavic Discourse Particles (Pragmatics and Beyond. New Series)

Constraints in Discourse 2 (Pragmatics and Beyond. New Series)

Constraints in Discourse (Pragmatics and Beyond New Series, 172)

Discourse Constructions of Youth Identities (Pragmatics & Beyond New Series)

Discourse Constructions of Youth Identities (Pragmatics & Beyond New Series)

Perspectives on Semantics, Pragmatics and Discourse: A Festschrift for Ferenc Kiefer (Pragmatics and Beyond New Series)

Discourse Markers in Native And Non-native English Discourse (Pragmatics and Beyond. New Series)

Early Modern English News Discourse: Newspapers, pamphlets and scientific news discourse (Pragmatics and Beyond New Series)

Japanese Discourse Markers: Synchronic And Diachronic Discourse Analysis (Pragmatics and Beyond New Series)

Cognitive Semantics: Meaning and Cognition (Pragmatics and Beyond New Series)

Verbal Aspect in Discourse (Pragmatics & Beyond)

Talking Gender and Sexuality (Pragmatics and Beyond New Series)

Mental Spaces in Discourse and Interaction (Pragmatics and Beyond New Series)

Impoliteness in Interaction (Pragmatics and Beyond New Series)

Humor in Interaction (Pragmatics and Beyond New Series)

Cognitive Semantics: Meaning and Cognition (Pragmatics & Beyond New Series)

Style Shifting in Japanese (Pragmatics and Beyond New Series)

Second Language Interaction (Pragmatics and Beyond New Series)

Youngspeak in a Multilingual Perspective (Pragmatics and Beyond New Series)

Recontextualizing Context: Grammaticality Meets Appropriateness (Pragmatics and Beyond New Series)

Metapragmatics in Use (Pragmatics and Beyond New Series)

Discourse Perspectives on English: Medieval to Modern (Pragmatics and Beyond New Series)

Collaborating towards Coherence: Lexical cohesion in English discourse (Pragmatics and Beyond New Series)

Creativity and Convention: The pragmatics of everyday figurative speech (Pragmatics and Beyond New Series)

Hedging in Scientific Research Articles (Pragmatics & Beyond New Series)

Political Discourse in Transition in Europe 1989-1991 (Pragmatics & Beyond New Series)

Discourse Description: Diverse Linguistic Analyses of a Fund-raising Text (Pragmatics & Beyond New Series)

Political Discourse in the Media: Cross-cultural Perspectives (Pragmatics & Beyond New Series)

Managing Language: The Discourse of Corporate Meetings (Pragmatics & Beyond New Series)

Implicatures in Discourse: The Case of Spanish NP Anaphora (Pragmatics & Beyond New Series)

Recommend Documents