Text and Discourse Constitution: Empirical Aspects, Theoretical Approaches 9783110862126, 9783110075663

Citation preview

Text and Discourse Constitution

Research in Text Theory Untersuchungen zur Texttheorie Editor Jänos S. Petöfi, Bielefeld Advisory Board Irena Bellert, Montreal Antonio Garcia-Berrio, Madrid Maria-Elisabeth Conte, Pavia Teun A. van Dijk, Amsterdam Wolfgang U. Dressler, Wien Nils Erik Enkvist, Abo Peter Hartmann f, Konstant^ Robert E. Longacre, Dallas Roland Posner, Berlin Hannes Rieser, Bielefeld Dieter Viehweger, BerlinjDDR Volume 4

w DE

G Walter de Gruyter · Berlin · New York 1988

Text and Discourse Constitution Empirical Aspects, Theoretical Approaches Edited by Jänos S. Petöfi

W DE

G Walter de Gruyter · Berlin · New York 1988

Printed on acid free paper (ageing-resistant — pH 7, neutral)

Library of Congress Cataloging-in-Publication Data

Text and discourse constitution : empirical aspects theoretical approaches / edited by Jinos S. Petöfi. p. cm. — (Research in text theory = Untersuchungen zur Texttheorie ; v. 4) Includes indexes. ISBN 0-89925-326-1 (U.S. : alk. paper) 1. Discourse analysis. : I. Petöfi, Jänos S. II. Series : Research in text theory ; v. 4. P302.T355 1987 87-27271 401.41-de 19 CIP

CIP-Kur^titdaufnahme der Deutschen Bibliothek

Text and discourse constitution : empirical aspects, theoretical approaches / ed. by Jänos S. Petöfi. — Berlin ; New York : de Gruyter, 1987. (Research in text theory ; Vol. 4) ISBN 3,11-007566-0 NE: Petöfi, Jänos S. [Hrsg.]; GT

ISSN 0179-4167 © 1987 by Walter de Gruyter & Co., Berlin 30 Printed in Germany All rights reserved, including those of translation into foreign languages. No part of this book may be reproduced in any form — by photoprint, microfilm or any other means — nor transmitted nor translated into a machine language without written permission from the publisher. Typesetting and Printing: Arthur Collignon GmbH, Berlin Binding: Lüderitz & Bauer, Berlin

Preface In research directed towards text constitution by means of dominantly linguistic devices (where the term 'linguistic' is to be taken in a wide sense) there are three key notions involved: textuality, connectedness (connexity, cohesion, coherence), and interpretability. However, the relationships among these key notions have not yet been specified unambiguously; some of the researchers consider them as to be mutually exchangeable, while others try to establish different kinds of dependency relations among them. In recent years the view has become increasingly dominant that as with a number of other questions in text research, this question can be answered adequately only in an interdisciplinary framework. In this interdisciplinary framework, linguistics taken in its wider sense continues to play the dominant role, even if it must be combined with methods of cognitive psychology when processual aspects of the text constitution are being analysed. The studies contained in the present volume attempt to supply a comprehensive picture of the main aspects of text constitution, and the problems, results and perspectives of linguistic research on these aspects. These studies can be divided into four groups. The studies classified into the first group treat aspects of the verbal constitution of texts: Gibbon handles in his paper the relation between intonation and discourse structure, maintaining the thesis that discourse structures and not sentence structures are the primary correlates of the structures defined by the prosodic system of a language. Harweg analyses those manifestations of the relations called by him "substitution" which guarantee the connectedness of sentence sequences consisting of one, two, or more sentences which presuppose one another. Brennenstuhl discusses on one hand aspects of the dynamics of speech act sequences, on the other hand she presents a speech-act thesaurus which is based on the idea of a threedimensional cross-classification. Hajicova and Sgall provide in their paper a survey of the linguistic questions concerning the topic-focus articulation of a sentence and discuss some of the attempts to bring this articulation into an immediate connection with the structure of the text. Rudolph analyses in her study on one hand the so-called connective relations (the relations which are assumed or postulated among states of affairs by producers/receivers), on the other hand those connective expressions with which natural languages operate to express these relations. Dorfmüller-Karpusa in her study is concerned with the question of how and to what extent temporal and aspectual relations contribute to assuring the connectedness of the text constitution. — While the topics mentioned so far belong under the heading 'linguistic research', the topic of the paper written by Garcia-Berrio and Albaladejo Mayordomo belongs both to the domain of linguistics and to rhetoric. In this paper the characteristics of the so-called macrostructures are investigated and the

VI

Preface

questions of how methods of text linguistics and of a rhetoric reinterpreted on the basis of the results of text-linguistic research can be used for their analysis. The studies in the second group deal with various aspects of research on meaning. Eikmeyer's paper provides a comprehensive picture of the research done up to now on word meaning, sentence meaning, and text meaning, taking into account both the dominantly logical research and the dominantly linguistic research, as well as the dominantly descriptive and the dominantly dynamic approach. Among the questions concerning the construction and reconstruction of meaning, the question of the relationship between literal and metaphorical meaning is of special relevance. Some main questions concerning the theory of metaphors are discussed in the study of Scholz. — In most cases, texts do not contain information on the syntactic, on the semantic, or on the pragmatic level in an explicit form which is necessary for assigning a meaning to them. In text processing the interpreter has to derive the missing information from what is given. Van de Velde investigates the different kinds of inference as well as the aspects of the processing strategies and of the knowledge sources. The two studies constituting the third group of papers discuss possibilities and means of applying the results of logical research. Ballmer is concerned in his paper with the logical aspects of such factors/phenomena as extensional reference, fiction, opaque contexts, verbal speech acts and written speech acts, dialogues, and context change. His attention is centered on so-called procedural logic and its application. — In the analysis of the coherence of texts (which is unthinkable without assigning a meaning to them), the stateof-affairs configuration which the interpreter believes to have been expressed in the text, plays a decisive role. In order that coherence and meaning interpretation can be treated adequately, it is indispensable to analyse the structure of the state-of-affairs configurations. Heydrich is dealing in his study with the aspects of setting up a logic, by means of which this structure can be analysed and represented. The fourth group contains two studies and a bibliography. Their common endeavour is to provide general information concerning the topic of the volume. Vitacolonna's paper is a survey of various attempts at explicating the notion text. The aim of Petöfi and Sözer in their paper has been to determine how the terms connexity, cohesion and coherence are to be used in the framework of a semiotic text theory. Lohmann provides a bibliographical survey of the monographs handling the topic of text-connectedness. This volume is being published in English in order to allow the widest possible access to these contributions, but since English is not the native language of most of the authors, there may be some stylistic rough edges which have not been smoothed out by the editors in order to preserve the logical precision of the arguments. Bielefeld, 1987

J.S.P.

Contents Preface

V

Part I Dafydd Gibbon Intonation and Discourse

3

Roland Harweg Sentence Sequences and Cotextual Connexity

26

Waltraud Brennenstuhl Speech Act Sequences

54

Eva Hajicova and Petr Sgall Topic and Focus of a Sentence and the Patterning of a Text

70

Elisabeth Rudolph Connective Relations — Connective Expressions — Connective Structures

97

Käthi Dorfmüller-Karpusa Temporal and Aspectual Relations as Text-Constitutive Elements . . . 134 Antonio Garcia-Berrio and Tomäs Albaladejo Mayordomo Compositional Structure: Macrostructures

170

Part II Hans-Jürgen Eikmeyer Word, Sentence, and Text Meaning

215

Oliver R. Scholz Some Issues in the Theory of Metaphor

269

Roger G. van de Velde Inferences as (De)compositional Principles

283

VIII

Contents

Part III Thomas T Ballmer Context and Context Change

317

Wolfgang Heydrich Things in Space and Time

377

Part IV Luciano Vitacolonna 'Text'/'Discourse' Definitions

421

Janos S. Petöfi and Emel Sözer Static and Dynamic Aspects of Text Constitution

440

Patricia Lohmann Connectedness of Texts: A Bibliographical Survey

478

Name Index

503

Subject Index

513

Part I

DAFYDD GIBBON Intonation and Discourse /. Introduction

In this paper I shall try to give an overview of some of the methodological problems and a few descriptive results in the area of intonation and discourse. The report is divided into four sections: 1. Methodological considerations, in which such questions as the role of discourse in general, as opposed to 'sentences', and the question of autonomy of systems will be dealt with; 2. Aspects of discourse structure, with a highly selective, relatively informal presentation of some points of relevance to intonational meaning; 3. Discussion of a specific systematic application of discourse relations to the solution of a problem in the area of Focus: so-called 'default accent'; 4. Outline of a model of prosodic phonology based on extensive perceptual and interpretative analysis of natural data, and on assumptions of concreteness in intonation which I would like to term 'Natural Prosodic Phonology'; it is intended to define procedurally interpretable prosodic structures. /. /. Is a 'sentence phonetics' possible? The commonly used phrases 'Sentence Phonology' and 'Sentence Phonetics' deserve a closer look. Everybody understands roughly what they mean — probably things such as intonation, accent, stress, rhythm, tempo, as well as word-linking phenomena at the segmental level. But I think one is justified in asking what, in fact, all this has to do with the category of sentence. Is there such a thing as 'Sentence Phonology' or 'Sentence Phonetics'? Without wishing to split hairs — or, at least, without wishing to split too many of them, I would like to point out a few things which these traditional terms imply in a modern context. First, there is the sentence/utterance distinction made (for instance) by Lyons, following a common tradition in the philosophy of language. This is not the distinction I mean; I take intonation (let me use this as a nearsynonym for the general topic) to be a set of systems which are comparable in abstractness and systematically to other systems in phonology or sentence syntax.

4

D. Gibbon

Second, there is the sentence/discourse distinction, where many specialists, including Lehiste, the Birmingham group (particularly David Brazil), the Edinburgh group, Couper-Kuhlen (l 986) and myself, have demonstrated that there is hierarchical organisation of intonation, which presumably means that not all (and perhaps not any) of these hierarchical levels relate directly to the category 'sentence'. Third, and most important, there is the question of the autonomy of levels in linguistic description: autonomy of semantics, syntax, morphology, phonology; what, then, of 'sentence phonology? It is in connection with this third aspect that I formulated the thesis of this paper: Discourse structures, not sentence structures are the primary correlates of the structures defined by the prosodic systems of language. A claim like this clearly commits me to explaining what I mean by the terms 'discourse structures', 'prosodic systems' and 'primary correlates'. The subsequent sections of the paper will be concerned with 'level autonomy', 'discourse structures' and 'prosodic systems', respectively. The notion of 'primary correlate' cannot be explained without some conception of the autonomy of levels of language description. I do not mean 'correlate' in the common phonetic sense of 'correlate in a different phonetic domain', such as acoustic or articulatory correlate: I mean semiotically functional correlation with units at other levels of language description. In the following section I shall distinguish between three different types of level autonomy and apply these to the study of intonation. 1.2. Types of level autonomy It has often been assumed that prosodic systems are autonomous in one, or more, of three ways; I shall refer to these types of autonomy as structural, procedural and functional. In a structurally autonomous system (or structurally autonomous level of analysis), no reference is made to primitive terms of other systems. Primitive terms in prosodic analysis include accents, and relative prominence relationships. In Sound Pattern of English phonology (Chomsky and Halle, 1968), stress patterning was not a structurally autonomous system, since initial stress assignments were calculated from segmental patterns and morphological categories, and hierarchical stress structures were calculated from these and syntactic structures. In metrical phonologies the situation is not so clear, since there are translation algorithms ("readjustment rules") for deriving metrical trees from syntactic trees. The situation in other theories is clearer. In the tonetic tradition (from Palmer through Kingdon, O'Connor & Arnold and Halliday to Crystal), the situation is different. Intonation patterning has an autonomously defined structure, whose relation to syntactic patterns is defined by 'unmarked' and 'marked' mappings. In these approaches, details

Intonation and Discourse

5

of mapping below the word level were often left inexplicit, since the smallest locutionary correlate of the prosodic systems at a broad phonological level was held to be the word. The structural autonomy hypothesis is held here; relatively 'flat' prosodic structures are defined (by transition network grammars or equivalent means) separately from the locutions with which they co-occur, and mapped on to semantically interpreted locutions. A procedurally autonomous system is one which is processed autonomously from other systems in a parallel processing environment. The assumption that prosodic systems are wholly or partly procedurally autonomous appears to be common; the phonetic and neurological bases for this autonomy need to be further defined. In view of the need to view a mapping between prosodic and locutionary categories as synchronisation, from a procedural point of view, processing cannot be wholly autonomous. It is plausible that the more physiologically causal aspects of paralinguistic behaviour (excited tension, etc.) are procedurally autonomous from the conventional aspects of prosody, though capable of influencing and being modelled by these. The stand taken here on this issue is that there can only be partial procedural autonomy. A functionally autonomous system is one with a semantic interpretation which does not refer to the semantic primitives of the interpretation of another system. Linguists who have held, like Martinet, that intonation is outside language proper, or those, like Schubiger, who distinguish between emotive or attitudinal, and intellectual or structural meanings of intonation, presumably hold the view that intonation is wholly or partly functionally autonomous. A linguist like Bolinger, who postulates a scale of grammaticalisation, would presumably hold that there is a scale of relative functional autonomy. The stand taken here on this issue is that most areas of prosody are not functionally autonomous. More precisely: the syntactic and discourse semantic categories on to which prosodic categories are mapped are the primary correlates, indeed 'meanings', of prosodic categories. For instance, the general meaning of a pitch accent is the focussed word with which it is synchronised. In previous studies I have referred to this variety of the functional nonautonomy hypothesis as the metalocutionary hypothesis. A different kind of functional non-autonomy hypothesis is the "shared means" hypothesis: prosodic categories have similar meanings to, and share the overall semantic load with locutions. For instance, speaker attitudes, or sentence functions (question, etc.) are said to be conveyed either by intonation, or by sentential means, or both, or the two "means" may clash in this respect. This is perhaps the most widespread view of prosodic meaning, but one which I am highly sceptical about. The extremely general, rather vague meanings of prosodic units are not suited to the attachment of heavy lexical labels in the traditional fashion: 'question, command, statement intonation', 'reserved, ironic, friendly, encouraging, insinuating tone', etc.

6

D. Gibbon

1.3. Discourse: heuristic or substantive notion?

A fourth, and final question to be asked in connection with the status of discourse is a methodological one: Does discourse have a heuristic value only, in supplying us with intuitively categorisable data types (such as formal vs. casual speech, spontaneous vs. edited speech, free production vs. reading aloud, and so on) for analysis on other levels? Or does it have the status of a substantive theoretical explicandum, like 'sentence' or 'word', or 'syllable' in our theories? So far, it has tended to have a more heuristic role. But in this regard, too, I suggest, it is desirable to perform heuristics in as systematic a fashion as possible. Simplified scale of data types Higher theoretical

Higher innovative

value

value

strict P-data

anecdotal P-data

S-data

S-data

Systematically constructed paradigms of P-data

'Invented' 'Remembered'

Q: What's going on? A: ONE pig's died one PIG's died one pig's DIED

Q: What's going on? A: One PIG's died

isolated data items used for ad hoc illustration.

Continuous recorded speech ... oh and ONE pig died because it ATE too much ... and ONE pig it was a YOUNG pig. about THAT size you know M1DDling ...

Figure 1

Figure 1 shows a simplified scale of data types. A simple distinction can be made between P-data and S-data; 'P' can be understood as 'paradigmatic' or, perhaps, 'planned' constructed data, and 'S' can be taken to mean 'syntagmatic' or maybe 'spontaneously produced', natural data. For various reasons which I shall not go into now, I prefer working with perceptual and interpretative methods on S-data. However, the majority of linguists tend to work with 'P-data' and I shall concentrate on P-data here. It should be

Intonation and Discourse

7

borne in mind, however, that native speakers (including linguists) have few descriptive terms for intonation, without special training. The main characteristics of P-data is that they are organised in systematic paradigms which are in turn defined by an explicit theory. Their role is to test predictions of this theory in a systematic way. The minimal case of a Pdata paradigm is the classical 'minimal pair'; this should indeed be the minimum requirement. This is hardly possible with S-data, which is why they are widely distrusted in linguistic circles, although lip-service is regularly paid to the ultimate desirability of describing natural data. In much actual intonation analysis, however, more informal hybrid data are used, of the 'anecdotal' type given in the centre of the figure. I would like to argue that wherever anecdotal data are used — their use is predominant in this area — they should be immediately regarded with the greatest suspicion, especially when coupled with far-reaching theoretical claims.

2. Discourse structures and discourse as context

The status of discourse as a substantive contextual explicandum for utterances can be classified in a progression from more elementary to more sophisticated under the following rough headings: (1) Functional Labelling: 'question intonation', 'normal intonation' (non-relational predicates); (2) Speech Act Explication: types of illocutionary force (statement of conditions of use); (3) Minimal Dialogue Schemata: adjacency pairs — Question-Answer, etc. (e.g. Natural Response relations); (4) Full Dialogue Model The most common procedure is to provide labels such as 'question intonation', 'normal intonation', 'focus'; these labels are simply one-place predicates which are intended to be intuitively evident and are not given further explication. Extreme cases of non-explanatory intonation labelling are to be found in the language teaching literature. If insufficient care is used, then what is called the notional fallacy in linguistics arises: in saying 'comma intonation', 'question intonation', for example, one assumes that something with phonetic relevance has been said. This is, however, not the case; there is no biunique relation between forms and such functional labels. The notional fallacy is perhaps the most common sin committed in intonation analysis.

8

D. Gibbon

The second step in this progression is the replacement of unanalysed functional labels by as explicit as possible a statement of conditions of use. The approach to language in which this kind of statement is to be found is ordinary language philosophy, as represented for instance, by Austin, Searle, Strawson and Grice. This technique of stating conditions observed by speakers with respect to their addressees' beliefs, and their own commitments to action, has been used to some extent in describing intonation and 'sentence functions' like 'question', 'statement', and the like, or the contributions of intonation to 'illocutionary force'. I have used this technique myself, and possibly the most recent attempt to use this approach is in a recent article in Language by Culicover and Rochemont (1983), in which they define different kinds of focus from this angle, for instance: Contrast!ve focus: In Foe (F(t)), the element Foci is a contrast!ve focus iff S believes that H believes that not F(t/Foci), and that H believes that F (t/ Foc2), for some Foc27^Foci (they add: '^' signifies lexical contrast). That is: if the speaker thinks the addressee believes what would ensue if the focussed element were replaced by a lexically contrasting construction, and he himself believes the opposite. A third, and rather different technique is one which is used quite frequently in the literature (e.g. by Gunter, Chomsky, Jackendoff, Ladd, Bolinger, Ronat, Lieb and many others): the 'natural response' or 'minimal dialogue schema' technique, in which a so-called 'adjacency pair' of speech acts, such as question and answer, is used, where the first item provides a context for the second. For example: Q: Has John read Slaughterhouse-Five? A: No, John doesn't READ books. This is the kind of example I used to illustrate the data types in § 1.3 above. In the final section of this paper I shall come to a detailed analysis of an explicit P-data paradigm of examples like these in order to illustrate what I mean by taking the notion of data paradigm seriously: the textual semantic properties of paradigms of this sort need to be — and can be — discussed in terms which are at least as systematic as the syntactic and phonological terms used, if the different autonomous systems of language are to be effectively linked. The ultimate aim must be, in my opinion, to establish more comprehensive dialogue models in order to create an increasingly sophisticated theoretical background for discourse description. There has been a lot of highly systematic work on discourse structure done during the past ten years or so, some of which has been explicitly related to intonation. I would like to simply mention some of the factors involved, as a kind of consensus from different approaches in text linguistics, discourse

Intonation and Discourse

analysis and sociolinguistics that I am familiar with. Relevant major levels of discourse organisation are: (1) Textual: cohesion / coherence in intonation, syntax and semantics; (2) Control: i. sequencing of 'adjacency pairs' (e.g. Question-Answer) ii. uptake monitoring and securing (e.g. 'back channel', repetition, 'WhatP'-loops). The first level, text semantic structure, is particularly important for the characterisation of coherent semantic development, contrast, anaphora and so on, which are in turn connected with questions of focus, accent placement, 'de-stressing', 'de-accenting', 'denuclearisation', and contrastive accent. The second is discourse control: sequencing constituents of discourse moves such as speech acts, implementing speaker turn control, and signalling the status of addressee uptake (understanding) as discourse progresses. It is contexts like these which are required for giving precise characterisations of question types, if old problems such as the association of high tones or rises with questions (of which there are, of course, numerous kinds) are to be seriously addressed. PARTICIPANTS

I

A

fern

I

I

B

I

C

DIALOGUE CONTRIBUTION

I

I

PRELUDE

1 PROPOSITION

fern male

+ •f

'no but you * see

+

Christmas" tea

'nobody ever 'eats their chriitmu tree

'true

+

'tea •f

'very true

•f

you 'know

+

you 'could have 'Christmas cake

+ •f

oh

for* brunch

* couldn't you

I 'don't know about 'that

Tn-m

+

'yen •f

giggle

+ +

CODA

'mince * pies cos 1 mean

"certainly [. . J

'if they're 'addicts then they'll'«! it when 'ever it is

Figure 2: Aspects of the topical structure of an English dialogue (Crystal & Davy, 1975; Extract 11, lines 27 — 38), retranscribed with main pitch accents.

10

D. Gibbon

An illustration of constituents of a discourse, here related to a schematic, ad hoc semantic topical structure, is shown in Figure 2. The body (here 'proposition') of a move is the textual core, based on semantically interpreted sentences; the initial and terminal periphery (here 'prelude' and 'coda') has discourse control function. Back channel moves have no textual body. I shall not go into further details here, but simply point out the consistency of semantic (and accentual) patterning through long stretches of S-data as an antidote to the pessimism frequently encountered in this regard.

3. A discourse-oriented model of discourse

Rather than outlining and criticising in detail various studies of discourse intonation by scholars such as Brend, Brown and co-workers, Couper-Kuhlen, Lehiste, Pike, Pilch, I shall outline the main features of a new discourseoriented model of intonation (cf. Gibbon, 1984, for more detail). The primitives are LOCAL (short-scope; roughly speaking: word-oriented) and GLOBAL (long-scope; roughly speaking: phrase/sentence/text-oriented). The LOCAL primitives are accent pulses and their modifications (some modifications being reflexes of global features determined with respect to longer intonational structures). (1) Accent pulses. Accents and perhaps other intonational categories may be thought of as laryngeal pulses. The larynx is regarded as a complex elastic body whose steady state with respect to vocal cord tension is altered by 'stretching' (POSITIVE) and 'compressing' (NEGATIVE) pulses. After a pulse, the larynx returns (other things being equal) to its pre-pulse state. The auditory reflexes are a pitch change in one direction on the leading flank of the pulse, followed by a change in the reverse direction after the pulse. The articulatory 'correlates' are conceptually simpler than the auditory patterning, and are presupposed by the latter. They are more abstract, however, from an empirical point of view. The two pulse types are (note that '«' denotes approximate correlation): i. POSITIVE pulse (« upward pitch movement), denoted by 'f' ii. NEGATIVE pulse (« downward pitch movement), denoted by 'J,'. The f pulse can be felt with a finger as a narrowing of the intercartilage ridge at the front of the larynx, the |, pulse sometimes as a lowering of the entire larynx. (2) Pulse modifications. The central modification is pulse amplitude (« degree of pitch change, ft Afo (i.e. change in fundamental frequency), ft prominence), which realises accentual gradation. A second set of accent modifications affects pulse timing relative to syllable structure, including pulse length, and

Intonation and Discourse

11

i. pulse leading flank:

a. delayed triggering (syllable-initial, post-syllable-onset, post-syllablenucleus, post-syllable), as in Kingdon's delayed rise-fall, Bolinger's variant of Accent A (characteristic of some dialect and style variation and some types of emphasis, interacting here with syllable onset timing features, which produce fortition, e.g. stronger aspiration); b. slow rise time involving syllable lengthening (accounting for some rise-fall tones, Armstrong and Ward's emphatic rise-prefix, Kingdon's 'homosyllabic prehead'); ii. pulse trailing flank:

a. fade (the neutral case — cf. remarks above); b. tension increase (A rising pitch, A pulse); c. laryngeal equilibration (A level tone, cf. 'chroma' Gibbon, 1976a, b; 'stylization', Chao, 1956; Ladd, 1980; Brazil's o-tone, 1978 and § 4 below). The GLOBAL primitives can be divided into boundary features and baseline features; they are properties of longer intonational structures). (1) Boundary features. The boundary features mark the beginning and end of laryngeal pulse sequences: i. Initial boundary: a complex function of Anacrusis/Prehead and Head/ Onset height specification, where the function of Head height (Crystal, 1971; cf. also Trim, 1964: 378; Brazil's 'key'; Lehiste's 'paragraph' boundary marking, 1975; Brown et al.'s 'initial stressed peaks', 1980: 136; cf. also Ozga, 1980) appears to be primary strategically modifiable by Prehead height (cf. Bing's initial boundary tone, 1980). In articulatory terms, Head height appears to be contextually determined to some extent by subglottal conditions. ii. Final boundary, with several different feature types: a. Lengthening of accented syllable and other timing factors involving delay, such as 'terminal pause'; b. Pulse mirroring: a sequence [a pulse] Λ [ — a pulse] (the use of ' — a' is that of generative phonology; ' A ' means concatenation), with the second pulse before the soonest segment which is next lowest in sonority in the subsequent unaccented stretch (cf. Liberman's similar explanation of tone distribution in 'call contours', 1978). This also allows a generalisation over call contour phenomena and some other contour types. If the specification is α = |, then the following inversion, — a, or |, can be explained functionally as accelerating the natural postpulse decay which already receives perceptual prominence by virtue of the syllable lengthening function.

12

D. Gibbon

c. Laryngeal tensing (« pitch rise). d. Laryngeal equilibration (« pitch sustention). (2) Baselinefeatures. Throughout the history of English intonation description, indications are to be found that two kinds of 'baseline' are involved: a 'baseline proper', a virtual pitch value toward which unaccented syllables tend (depending on their relative proximity to a pulse), and a line along which auditory pitch peaks tend to be aligned (cf. Bolinger's 'tangent', or Body/Head contours). I shall call the former baseline and the latter peakline. The most similar recent applications of this distinction are by Brown et al. (1980) and Carding (1984). Given a baseline and a peakline, a number of global relational properties of prosodic patterning can be defined over peakline-baseline pairs: i. Bandwidth ('span', 'range') is a simple peakline-baseline relation defined as pitch difference (= Afo). ii. Peakline slope, [Δ peakline] i.e. type of change in peakline: up, down, level), with downdrift as the neutral case and a rise (cf. Palmer's 'scandent head') or more complex tangents as marked cases. Whether peakline slope induces baseline slope (or vice versa) is an open question. Three main kinds of 'level' peakline slope may be distinguished: a. low, as in relaxed conversational reporting style, often taken as the standard American English pattern but also characteristic of analogous styles in other dialects of West Germanic languages; b. sustained mid, as in one speech genre of overt narration; this feature is similar to the laryngeal equilibration noted twice above; c. high pitch, as in 'uncontrolled', excited speech. iii. Peakline-baseline pitch convergence, [peakline ->p baseline], again of two types: a. an unmarked type, the 'trailing off which is presumably allied to neutral downslope and may not need to be postulated separately from this; b. sustained convergence, as in mid-discourse 'narrative tags' (cf. Gibbon, 1984: § 3, § 4). iv. Peakline-baseline 'identity', [peakline = p baseline]: this is strict 'chroma' or 'stylization', which occurs in its most extreme form in chants and song (Chao, 1956), and is entirely dependent on larynx equilibration supported by negatively phased auditory feedback (cf.

§4). v. Peakline-baseline complementation, in which relative peakline-baseline heights are inverted; this is an interesting theoretical possibility with considerable descriptive potential (cf. Gibbon, 1984: Figure 2). This feature provides a device for describing certain features of style

13

Intonation and Discourse

and dialect variation in a unified fashion, and distinguishes some global uses of |/| alternation from the local uses already noted, with cases such as the following (where α< ρ β means 'a is lower in pitch than β' and α > ρ β means 'a is higher in pitch than β'): a. Edinburgh dialect: [peakline > p baseline], Glasgow dialect: [peakline < p baseline] (Brown et al., 1980: 19); b. North German dialects: [peakline > p baseline], South German dialects: [peakline < p baseline] (an oversimplification, but cf. von Essen's "Zickzackmelodie", 1956); c. In standard English dialects, child-adult and adult-child speech, with [peakline < p baseline], often functionally alternating with [peakline > p baseline]. d. As a stylistic discourse strategy in German (cf. Gibbon and Selting, 1983). The primitives are used to construct complex intonational structures by a syntax with both iterative (e.g. for sequences of accents and accent groups) and recursive (e.g. for embedded parentheses with temporary prosodic marking) PROSODIC FRAMES:

π-frame p-line > p b-line high init. pitch _Δρ = wide, downslope ι

γ-frarne ["downslope ~] [convergence]

"γ-frame [downslope ~| [convergencej

[long syll TONAL FUNCTIONS

\ l·I

b4

i

A \

H

t

[long syll.]

T

N

A /l

high] I

ACCENT Polarity: Modifications:

/ 1 / 1

τ ·

Tct = i nc. ampl. [i ong pulse_

"

[( t =

[oum

/

/

/

/

1 I

LOCUTION:

[α = Τ]/

/

1 w D m pig Figure 3

/

daid

5ikz it et tu m

1

l 1 1 1 1

1

J

Atf]

14

D. Gibbon

INTENSITY

Ξ0Θ1

unibi-1 M i

FREQUENCY •ν

Ι

· ·- v 100-

t w Ans

* Λρ On3 t ai m d3w3

t θ r i:

* '-, bE

3

z

50 Θ

.5 ] 1.5 2 2.5 3 ONCE u p o n a T I M E t h e r e were THREE B E R R S Figure 4a: Four rising-falling pulses in two accent sequences, each with falling high peakline and terminating with a fall. INTENSITY

2Θ0Ί

ΙΘΘ-

unibi-li l i

FREQUENCY

M .·

w Λ n s 3 p On3 t a l md 3W3

Θ

θ r i:

b ε

3

.5 l 1.5 2 2.5 ONCE upon a T I M E t h e r e were THREE

z

3 BERRS

Figure 4b: Four falling-rising pulses in one sequence with falling low peakline, terminating with a (rise-)fall terminal tone; the rise part of the terminal is actually due to the rising flank of the accent pulse; the terminal is a fall.

Intonation and Discourse

15

properties. An illustration of an analysis of one utterance ("Oh and One pig 'died because it 'ate too much", also from Crystal and Davy, 1975) is given in Figure 3. A -frame is a unit with recursive properties (not realised in this example); a -frame has only iterative properties. The categories ßi and ßt are initial and terminal boundary tones, while is an accent. The initial boundary consists of Anacrusis (unstressed syllables before the first accent) and //ead (first accent); the terminal boundary consists of the last (TVuclear) accent and Tail (unstressed syllables following the last accent). If only one accent occurs, then the head-nucleus distinction is neutralised, as on ate. A further illustration of a sequence containing POSITIVE pulses and one containing NEGATIVE pulses is given in Figure 4. In the sentence "Once upon a time there were three bears", the different pulse accent reflexes can be recognised particularly clearly on time and three. 4. An application: Cybernetic iconicity in call contours

Let me illustrate an application to one specific and, I think, novel point in this area. A good many years ago I looked into the claim that intonation patterns were used as illocutionary force markers. The obvious method to use seemed to be that of speech act theory, the second in the progression listed in § 2 above. The clearest case of an easily identifiable pattern both in form and function seemed to me to be contours with 'chanted' pitches, or 'call contours', with laryngeal equilibration. I first reported on these results in Gibbon (1976 a, b). It turned out that in German, this contour is used in a broader range of contexts than in English: in calls, in greetings, in farewells, in other formulae for beginning and ending transactions such as "Bitte schön", "Danke", "Bitte", but also in discourse medial contexts in order to restore communication after a misperception: students in class would call "Lauter!" if I spoke too softly, and a number of similar cases struck my attention. In terms of Searle's conditions on the use of speech acts, therefore, it seemed that the role of this intonation was not to mark an illocutionary force as such, but to mark beginnings, ends and reinstatements of states of talk, i.e. to mark a prerequisite for illocutionary force in the technical sense defined by Searle, but not exactly illocutionary force itself. If this is also taken to be an illocutionary force, as some have done, then it is a different use of the term. That was the original investigation. Later, using a model of dialogue structure rather than statements of conditions of use, I came to the conclusion that the contour was used to draw attention to the self-monitoring and othermonitoring features of dialogue from a procedural, rather than a structural point of view and that an appropriate model for the uptake control component would be a closed system self-regulation circuit with negative feedback: as

16

D. Gibbon

INTENSITY

iinibi-1 i l i

FREQUENCY

100-

G

.5

]

1.5

2

J o h n — n y h e i l α-o narrow, POSITIVE > NEGATIVE fall > rise, in general). Note that the accent pattern and the stress pattern may conflict, as the former is assigned by semantic and pragmatic discourse factors, while the latter is an abstract formal pattern. Metre: Conventionalised euphonic principle of harmonically related occurrences of syllable strength relations, characteristically in the form of a 'metrical grid'. Metre is a speech genre or style principle which varies along scales of metrical depth and stylisation, from speech to verse. Metre may conflict with both stress and accent patterns. Rhythm: Perceived subjective regularity in the temporal patterning of prominence peaks. By analogy with prominence, rhythm is a pre-theoretical notion and may be explicated as a complex function of stress pattern, accent pattern and metre expectations, i.e. of perceptual and abstract cognitive patterning. More specific phonetic terminology (pitch, intensity, etc.) may be taken from Lehiste (1970). The final term to be defined is:

Intonation and Discourse

19

Focus: The position of an accent, relative to a semantically interpreted syntactic constituent; this definition resembles definitions used in generative and metrical phonology, from Chomsky (1971) to Selkirk (1984). The following section will be concerned with a frequently discussed problem of accent and focus.

6. Application to focus and accent: default or discourse implicature? To extend the picture, I would therefore like to examine one aspect of a rather well known, but so far insufficiently explained, problem in the semantics of focus according to the data paradigm method. The data derive from Halliday and Hasan (1976) and from Allerton (1978); they have been dealt with by Ladd (1980) and others under the name of 'default accent'. The standard explanation is in terms of anaphora: Since an individual or a species implies a genus, in the well-known pair Q: Has John read Slaughterhouse-Five? A: No, John doesn't READ books. the relation of 'Slaughterhouse-Five' to 'books' is one of hyponym to superordinate term, or of specific to generic term. The latter is implied by the former, hence anaphoric, hence 'denuclearised' (Allerton), 'deaccented' (Ladd), 'destressed' (Schmerling). The accent then falls on the first available item earlier in the 'sentence', by 'default'. But this is not the whole story, as an examination of a fuller P-data paradigm shows. Accent is not simply assigned to the verb in this case by default, in the absence of an appropriate focus; on the contrary, a distinct discourse semantic focus interpretation is associated with this accent placement. In particular, the distinction between accent on the first or finite component of the verb and accent on a later component of the verb must be distinguished: No, John doesn't READ books. No, John DOESn't read books, etc. In a recent study, Fuchs (1984) distinguishes here between focus on 'ascription', i.e. the illocutionary component, with accent on the finite form of the verb, and focus on 'connection', i.e. the syntagmatic predicative component, with accent on the first nonfmite item in the verb phrase. A related distinction between accents of 'power' and accents of 'interest' is made by Bolinger (1983). These analyses do not go far enough, however.

20

D. Gibbon

The first step in a more complete analysis is to construct a P-data paradigm Question: Has Greg Answer paradigm: No, Greg Greg Greg Greg Greg Greg

drunk my whisky? HASN'T drunk your whisky hasn't DRUNK your whisky HASN'T drunk any whisky hasn't DRUNK any whisky DOESN'T drink whisky doesn't DRINK whisky

(la) (lb) (2a) (2b) (3a) (3b)

This is only a fragment of the full response paradigm, which would have to include accents on Greg or whisky, combinations of accents on different words, and formulations with non-negatives, etc. Further, a parallel question paradigm can be constructed. This is sufficient for the present discussion, however. Discourse semantic properties of members of answer paradigm CONTRAST, or FOCUS ON DENIAL OF:

a. Validity of ASSERTION (ASCRIPTION)

b. Validity of PREDICATION (CONNECTION)

1. Specific item in Existential context range (PAST)

2. General class in Existential context range (PAST)

3. General class in Universal context range

Greg HASN'T drunk any whisky.

Greg HASN'T drunk your whisky.

Greg DOESN'T drink whisky.

0

1

Greg hasn't DRUNK your whisky. 1

Greg hasn't DRUNK any whisky.

2

Greg doesn't DRINK whisky.

2 3

Figure 6

Figure 6 shows a semantically motivated tabular arrangement of the paradigm: the columns are distinguished by minimal differences in existential and universal quantification; the rows are characterised by the assertion vs. predicative presupposition distinction.

Intonation and Discourse

21

Recall that the basic method is that of "natural response" (cf. § 2). Now which of the six members of the response paradigm is 'the natural response'? I am indebted to Margret Selting for the comment that they are all perfectly natural responses. Indeed they are. However, it is possible to quantify naturalness in terms of minimal semantic differences. By this criterion, the item at the left, Greg hasn't drunk jour whisky, is 'the' semantically minimal natural response, while the item cited in the literature, at the right, is by no means semantically minimal. Furthermore, the naturalness metric can be refined by reference to the assertion vs. predicative presupposition distinction: the assertion focus may be defined as semantically more natural than the predication focus. Let us assume that the formula for the metric is simply x + y, with o ^ x ^ 2 (from left to right) and ο ^ y ^ 1 (from top to bottom) in this particular case. Then naturalness values can be calculated; they are contained in the lower right corners of the corresponding cells of the matrix in Figure 6. A more general way of formulating the metric is: Semantic natural response metric

Let η be the number of semantic implicature dimensions involved, with d the distance from the minimal (zero implicature) position on any dimension: Response naturalness value =

e.g.:

d2:

X j = ι

dr.

0

1

2

0

0

1

2

1

1

2

3

Now it may be thought that this semantic numerology is a little pointless, and explains nothing. True. It is simply an expository crutch. There is, however, something much more significant that underlies this numerical conception of naturalness, namely that of discourse implicature. I shall not recapitulate this branch of discourse theory here but simply invoke common sense interpretations of the implication relations between members of the Pdata answer paradigm. First, it is clear that on elementary semantic grounds there is an implication relation between a general statement and a specific statement. Outside logic textbooks (where ontological problems raise their ugly heads), the same applies conversationally to the all-some relation, which obtains between whisky and my whisky (with any whisky as a variant of the universally quantified expression in some contexts in English). There are, in consequence, semantic textual implicatures, as I wish to term them, from right to left: any paradigm

22

D. Gibbon

member implies its left neighbour. This type is indicated by lexical changes, each accompanied by an accent — cf. also doesn't drink ANY whisky. Second, it is also clear that to question or deny a presupposition is to question or deny a putative assertion or set of assertions which make this presupposition. In fact it is more than this: it is to .question or deny the relevance of that assertion or set of assertions, extending the Frege-Strawson line of argument to the present conversational context. Consequently, there is a predicative textual implicature, as I wish to term it, from bottom to top: any paradigm member implies its top neighbour. This is the type indicated by accent placemerit. Note that the accent thus has a positive focus meaning, and is not just a default assignment by virtue of anaphora. Third, there is the application of these implications outside the response paradigm, with respect to the 'natural response' relation between context and answer, which can be termed speech act insinuation or dialogue implicature: to utter a semantically less natural response to a question is to imply, to insinuate, to impute to the questioner a covert intention to utter the question which is structurally cognate to the response, rather than the question actually uttered. That is, the response "No, Greg DOESN'T drink whisky" can be taken to insinuate that the questioner really wanted to ask "Does Greg drink whisky?", for instance to find out something more general about Greg's habits. This is an explication of Fuchs' observation about paraphrases for assertion focus as opposed to predicative presupposition focus: Assertion: "What you are saying implies the possibility of John's reading books. Well he doesn't." Predicative presupposition: "You are saying that John reads books — let me tell you that such isn't the case." In other words, the first two implicature types are concerned with the indirectness of the answer; the third type implies that the question is an indirect speech act. This analysis of focus has been applied in detail to a particularly simple example. Applications to other examples will be more complex, but there is no reason to doubt that they are possible. It could be said, in summary, that anaphoric destressing may be a necessary condition for explaining accent positioning here. If it were also a sufficient condition then perhaps notions like 'default accent' might be adequate. I hope to have demonstrated that it is not a sufficient condition, in fact, but that a special focus distinction is involved. In more general terms, I hope to have shown that there is no reason to be pessimistic about explaining the semantics of focus but that even if a relatively elementary discourse semantic theory is used, this is far superior to no semantic theory at all.

Intonation and Discourse

23

7. Final remarks

In this paper I have primarily argued for consistent and systematic use of theory-backed analysis at all relevant levels of linguistic description, from the phonetic domain to the domain of discourse. In particular, I have illustrated, at various points, a plurality of equally valid methods of analysis, without claiming that the results of, for instance, a P-data analysis with or without experimental methods and the results of an S-data analysis could be reduced to some common denominator. It is tempting to speculate that this may be so, of course, in some more comprehensive metatheory. Beside these methodological points, a number of innovative points of substance were made: (1) the pulse accent theory with local and global primitives, and iterative and recursive construction processes; (2) application to iconicity in call contours; (3) application to focus and discourse implicatures in natural response relations. The conclusion to be drawn is that whatever area of prosody is studied, it requires thorough treatment on as many levels of analysis as are required; no level of analysis should be treated more lightly, in terms of theoretical understanding, than any other, whether at the discourse levels of text semantics and discourse control, or at phonological and phonetic levels. Note The research reported on here was partly financed by a grant from the Deutsche Forschungsgemeinschaft (Gi 143/1 — 1). I am indebted to members of the Constance Summer School of the Deutsche Gesellschaft für Sprachwissenschaft, to Betty Couper-Kuhlen, Greg Dogil, Anna Fuchs, Bob Marek, Friederike Sader, Margret Selling and Hans-Jürgen Türling for much fruitful discussion on the topics involved.

Literature Allerton, David J. 1978 "The Notion of 'Givenness' and its Relations to Presupposition and to Theme", Lingua 44, 133-168. Bing, Janet M. 1980 Aspects of English Prosody. Bloomington: Indiana University Linguistics Club (Ph. D. Thesis, Amherst 1979). Bolinger, Dwight 1983 "Affirmation and Default", Folia Linguistica 17, 99-116. Brazil, David 1978 Discourse Intonation (Discourse Analysis Monographs No. 2). Birmingham: University of Birmingham, English Language Research.

24

D. Gibbon

Brown, Gillian; Curry, Karen L.; Kenworthy, Joanne 1980 Questions of Intonation. London: Croom Helm. Chao, Yuen Ren 1956 "Tone, Intonation, Singsong, Chanting, Recitative, Tonal Composition and Atonal Composition in Chinese" in Halle, M., Lunt, H. and McLean, H. (eds.): For Roman Jakobson. The Hague: Mouton, 52—59. Chomsky, Noam 1971 "Deep Structure, Surface Structure and Semantic Interpretation" in Steinberg, D. D. and Jakobovits, L. A. (eds.): Semantics: An Interdisciplinary Reader in Philosophy, Linguistics and Psychology. Cambridge: Cambridge University Press, 183 — 216. Chomsky, Noam and Halle, Morris 1968 The Sound Pattern of English. New York: Harper and Row. Couper-Kuhlen, Elizabeth 1986 An Introduction to English Prosody. Tübingen: Niemeyer. Crystal, David 1971 "Relative and Absolute in Intonation Analysis", Journal of the International Phonetic Association 1, 17—28. Crystal, David and Davy, Derek 1975 Advanced Conversational English. London: Longman. Culicover, Peter W. and Rochemont, Michael 1983 "Stress and Focus in English", Language 59, 123-165. von Essen, Otto 1956 Grundlage der hochdeutschen Satayntonation. Ratingen: Henn-Verlag. Fuchs, Anna 1984 "'Deaccenting' and 'Default Accent"' in Gibbon, D. and Richter, H. (eds.): Intonation, Accent and Rhythm: Studies in Discourse Phonology (Research in Text Theory 8). Berlin: de Gruyter, 134-164. Girding, Eva 1983 "A Generative Model of Intonation" in Cutler, A. and Ladd, D. J. (eds.): Prosody: Models and Measurements (Springer Series in Language Communication 14). Berlin: Springer, 11—25. Gibbon, Dafydd 1976a Perspectives of Intonation Analysis (Forum Linguisticum 9). Bern: Lang. 1976b "Performatory Categories in Contrastive Intonation Analysis" in Chi^oran, D. (ed.): Second International Conference of English Contrastive Projects. Bucharest: Bucharest University Press, 145 — 156. 1984 "Intonation as an Adaptive Process" in Gibbon, D. and Richter, H. (eds.): Intonation, Accent and Rhythm: Studies in Discourse Phonology (Research in Text Theory 8). Berlin: de Gruyter, 165-192. Gibbon, Dafydd and Selling, Margret 1983 "Intonation und die Strukturierung eines Diskurses", Zeitschrift für Literaturwissenschaft (Lili) 49, 53-73. Halliday, Michael A. K. and Hasan, Ruqaiya 1976 Cohesion in English (English Language Series 9). London: Longman. Ladd, D. Robert 1980 The Structure of Intonational Meaning: Evidence from English. Bloomington: Indiana University Press. Lehiste, Use 1970 Suprasegmentals. Cambridge, Mass.: MIT Press. 1975 "The Phonetic Structure of Paragraphs" in Nooteboom, S. G. and Cohen, A. (eds.): Structure and Process in Speech Perception. Heidelberg: Springer, 195 — 203. Liberman, Mark Y. 1978 The Intonational System of English. Bloomington: Indiana University Linguistics Club.

Intonation and Discourse

25

Ozga, Janina 1980 "A Functional Analysis of Some Pause and Pitch Step-Up Combinations" in Dechert, H. W. and Rauphach, M. (eds.): Temporal Variables in Speech: Studies in Honour of Frieda Goldman-Eisler. The Hague: Mouton. Selkirk, Elisabeth 1984 Phonology and Syntax: The Relation between Sound and Structure, Cambridge, Mass.: MIT Press. Trim, John L. M. 1964 "Tonetic Stress Marks for German" in Abercrombie, D. et al. (eds.): In Honour of Daniel Jones. London: Longman, 374—383.

ROLAND HARWEG

Sentence Sequences and Cotextual Connexity^ Pluralities of sentences — sentences defined as explicitly or implicitly complete predications1 — occur in a number of basically different kinds of arrangement, two simple and a number of gradually more complex ones. The simple ones are sequence and set, the complex ones, among others, sequence of sets, set of sequences, and sequence of sets of sequences. Of these kinds of arrangement we will concentrate in this article on sentence sequences. Sets, sequences of sets, sets of sequences and sequences of sets of sequences of sentences will be discussed only incidentally and by way of contrast. The notion of sentence sequence — but the same may be true of the other modes of arrangement as well — is not unequivocal. Like many other metalinguistic concepts, it comprises two qualitatively different conceptual variants, two variants designating linguistic phenomena of different structure and mode of existence, physical on the one hand and semiotic on the other — phenomena of which I, following Pike,2 designate the former as etic and the latter as emic. Just as, for example, the undifferentiated concept of sound is split up into the concepts of phone and phoneme, the undifferentiated concept of sentence sequence can and should be split up into the concept of etic and of emic sentence sequence. /. Etic sentence sequences By an etic sentence sequence I understand a plurality of sentences which, as a rule within the boundaries of one and the same community of sentence* I wish to thank David Beal, Bochum, for reading the manuscript and suggesting several stylistic improvements. Any remaining errors or inadequacies are, of course, my responsibility alone. 1 According to the criterion of predicational completeness (put forward and discussed in Harweg, 1971d), non-restrictive relative clauses, for example, are sentences, but restrictive ones are not. One may, however, doubt whether it is justified to assign the non-restrictive relative clauses (and, similarly, certain types of conjunctional clauses or even coordinated sentences) the status of full sentences, and in an as yet unpublished article entitled "Subordinative Semi-Satzfolgen und Textkonstitution" I thus propose to interpret them as semisentences. In this article, however, I do not make use of this concept. 2 Cf. Pike (1967).

Sentence Sequences and Cotextual Connexity

27

sequence producer(s) and recipient(s),3 follow each other "physically", i.e. either temporally or spatially. The notion is not to be confused with that of 'etic texts'.4 Thus, not every etic text manifests an etic sentence sequence — for there are also etic texts which consist of only one sentence each —, and inversely, not all etic sentence sequences manifest etic texts: segments of etic texts and sequences of etic texts may be etic sentence sequences as well. With the concept of etic texts the concept of etic sentence sequences shares only linearity, but not delimitation or more precisely, observer-independent delimitation. Unlike the delimitation of etic texts, the delimitation of etic sentence sequences is solely dependent on the focus of the observer and is therefore arbitrary. The sole criterion for etic sentence sequences is, then, the linear succession of a plurality of sentences in time or space. Yet for such linear successions to emerge, mere temporal successivity — the case of spatial successivity is somewhat different — is not enough. Thus, the sentences uttered throughout the world at successive points of time do not, even under the widest possible focus, form an etic sequence of sentences, but an etic set of sequences of sentences; for a great part of these sentences is, at each of these successive points, uttered simultaneously. But even for the formation of an etic set of sentence sequences the criterion of mere temporal successivity does not suffice. For, what is brought about by this criterion alone is, at least with regard to all sentences which are temporally coextensive, etic sequences of sets of sentences rather than etic sets of sequences of sentences, and all sentences which are, temporally, not coextensive, belong, together with the sentences temporally coextensive with them, to d i f f e r e n t sequences of sets of sentences. For a sequence of sets of sentences to become a set of sequences of sentences, a criterion is needed that determines which of the successive sentences of the successive sets of sentences actually form, in any special case, a definite sequence, and the criterion which might serve the purpose is the identity of a special producer-recipient-community. Thus, a certain sequence of sets of sentences is split into etic sequences of sentences according to the extent to which subsequent sentences can be attributed to definite producerrecipient-communities. A single etic sentence sequence is therefore — provided it is an oral one — a plurality of sentences which follow each other temporally within one and the same producer-recipient-community. Theoretically, it seems as if the analyst's phenomenological point of departure were a set of sequences of sets of sentences, a phenomenon from which he had to isolate the single etic sentence sequences by having recourse to the additional criteria just mentioned. Practically, however, it is the single etic 3

4

In what follows I use alternatively the shorter terms 'producer-recipient-community' and 'sentence-sequence community'. Cf. Harweg (1968/79: 152 ff., 285 f.).

28

R. Harweg

sentence sequence which forms his phenomenological point of departure, for he is — and we all are —, at any single moment, a member of maximally one producer-recipient-community. Sentence sequences which at the same time are produced within some other producer-recipient-community are not experienced by us, we are only allowed, and indeed are forced, to a s s u m e that they are being produced. From the point of view of direct observability and its backward prolongation, namely recallability, the sentence sequences produced and received within foreign producer-recipient-communities are, strictly speaking, only hypothetical entities, entities of whose factual existence we, despite a high, but unspecific probability of their simultaneous existence, only get notice through their being mentioned to us and of whose contents we only learn through their being cited. However important the identity of a definite producer-recipient-community for the constitution of etic sentence sequences may be, it is very difficult to grasp and determine this identity exactly. The reasons are partly subjective, partly objective in nature. As subjective I regard all those reasons which are bound up with the faculty of the members of a sentence-sequence-community to recall the sentence sequence, whereas as objective I consider the reasons connected with the composition of such communities. We cannot go into details here, but have to note that in both cases we are forced to make considerable idealizations. As far as the composition of such communities is concerned, this obtains above all for communities with anonymous recipients, and in this respect the highest degree of idealization is presumably required for communities with anonymous recipient groups as diffuse as those which are addressed by broadcast and television speakers. In this connection it seems unavoidable to conceive of the recipient-group not as a factual, but as an intended one. But even then there are still considerable problems left. To cite but one: should, for instance, a certain broadcasting station or more precisely, one of its alternative programs presuppose for all of its transmissions one and the same group of intended recipients or should it presuppose different groups for different (types of) transmissions, and if so, what groups for which transmissions? In this connection, a particularly important case are series, not only fictional ones (where this is obvious), but also (where this is less obvious) non-fictional ones, such as, for example, the permanent series of radio news broadcasts. Apart from the synchronic extension of a sentence-sequence community, the diachronic extension, the life-span, of such a community is a problem as well. It is connected with the community members' faculty of recall, and this does not go back beyond a certain point of time in the individual member's childhood or, as we might say for the sake of simplicity, beyond his birth. Yet however precisely this point may be statable in an individual's life, in a community with members of different ages it is a very fluctuating point. Here, too, idealizations cannot be dispensed with.

Sentence Sequences and Cotextual Connexity

29

Unlike the formation of etic sentence sequences in the case of temporal successions of sentences, the formation of such sequences in the case of s p a t i a l successions of sentences does not seem to require the existence of a definite sentence-sequence community. This is not to say that the opposition between successivity and simultaneity has no equivalent in the sphere of space; for besides successivity there is, in the sphere of space, something like parallelity, and this might even be the primary mode of arrangement in space, successivity coming about only by a process of linearization and directionalization of certain spatial phenomena. Such a process is writing, but linearity and directionality are inherent not only in the process, but also in the products of writing. Yet just as products of speech also occur simultaneously, products of writing also occur in parallel arrangements. In space, too, we therefore need a criterion for deciding which of the spatially contiguous sentences are those that form an etic sequence — a need which makes itself felt the more so as in writing linearity is partly disguised by apparent non-linearity. We must know, for example, in what way the actual lines on a page or a sequence of pages make up that ideal and unique line that we are forced to follow in reading. True, in most books, that is, in books without footnotes and containing only one text, the number of conventions to be known in combining the actual lines is comparatively low, but in books with footnotes and/or containing more than one text, including books with parallel translations, it is comparatively high. And the number of conventions to be known may be still higher in newspapers and magazines, where sets of sentence sequences are distributed over single and subsequent pages in the most diverse modes of arrangement and where the different etic sentence sequences are to be identified with the aid of conventions such as size and arrangement of headings, type and size of letters, kind and distribution of separation lines, arrangement of columns, and place of names of authors, and sometimes even through explicit metalinguistic references, such as continued on page 12. Most of these conventions are learned by use or even grasped intuitively, but sometimes the reader in identifying the different etic sentence sequences has to have recourse to emic criteria as well. Above all this might be the case with insertions of pictures in a text. And in this case it might even happen that the determination of etic succession within one and the same sentence sequence requires emic criteria to be consulted. The identification of etic sentence sequences in books, newspapers and magazines can make use of the conventions mentioned above only to the extent that one does not transcend the boundaries of the single publication unit, that is of a single book (which, however, may consist of several volumes) or of a single number of a newspaper or magazine. For, as soon as we do so, we leave the sphere of spatial arrangement and enter, though remaining in the medium of writing, the sphere of temporal arrangement. True, we could interpret as a kind of time even the spatial succession manifested by a written sentence sequence, but this kind of time would be a very special one,

30

R. Harweg

a sort of 'text-time', whereas the time which forms the temporal basis of those written etic sentence sequences which connect subsequent books or subsequent numbers of a newspaper is time of the same kind as the one which forms the basis of oral etic sentence sequences, namely that kind of time which we might term 'world-time' instead of 'text-time' and which, in this case, is the time of publication of subsequent books or numbers of a newspaper or magazine. Now, the segments of written etic sentence sequences which are formed by successions based on 'world-time' require for their identification and determination — just as the oral sequences do — the existence of a definite, though idealized, sentence-sequence community. In the case of subsequent numbers of newspapers or magazines this community may be, for instance, the author-reader-community of a serialized novel. The reader community of a certain newspaper as a whole, however, is, of course, not a sentence-sequence community, it is at best a community associated with a sequence of sets of sentence sequences. Written etic sentence sequences are to be found not only in books and newspapers, but also in the domain of positionally fixed "inscriptions" or labels. Here, however, the distinction between etic sequences and sets of sentences often cannot be made in a clear-cut way. Sometimes — as with the labels or inscriptions on self-service vending machines — defmiteness of order is brought about by numbering the inscriptions, and a particularly interesting form of written etic sentence sequences are sequences of signpost inscriptions, that is sequences owing their sequentiality to the sequentiality of the different stages of a specific route at which they are erected.5 True, these signpost inscriptions are for the most part highly elliptical as to their sentential manifestation, but why should the sentences of etic sentence sequences not be elliptical? What is peculiar about these sentence sequences is solely that we follow them not only with our eyes, but also with our bodies, either driving a car or walking.

2. Ernie sentence sequences

Unlike etic sentence sequences, emic sentence sequences are pluralities of sentences which are not defined by the fact that their sentences follow each other, under certain conditions, temporally or spatially, but by the fact that their sentences presuppose6 each other cotextually7 or are at least, through 5 6

Sequentiality is also found among non-verbal traffic signs. Cf. Studnicki (1970). Much more often than in connection with sentence sequences the phenomenon of presupposition has been discussed in connection with single sentences, presupposition being the favourite concept of sentence grammarians who felt the necessity to cope with relations transcending the boundaries of the single sentence without being willing to enter the domain of text grammar. K. Heger's concept of'Präsuppositionsgefüge', however, which he uses to designate

Sentence Sequences and Cotextual Connexity

31

certain linguistic means, cotextually connected.8 Yet they are, notionally, not identical with emic texts9 — just as etic sentence sequences are not identical with etic texts. Neither is every emic text an emic sentence sequence — for there are also emic texts consisting of one sentence only —, nor is every emic sentence sequence an emic text: segments of emic texts may be emic sentence sequences as well. It may, however, happen that there are emic segments of emic texts that are emic sentence sequences without being recognizable as such.10 2.1. Two-sentence sequences 2.1.1. Syntagmatic substitution

The phenomenology of emic sentence sequences, that is of the cotextual connexity between the sentences of emic sentence sequences is presumably inconceivably complex, and up to now only a tiny part of it has been investigated.11 The part whose investigation has been carried farthest is without doubt the cotextual or more specifically, the syntagmatically substitutional or briefly, substitutional connexity of only two subsequent sentences. It is therefore with the substitutional connexity of two-sentence sequences12 that I will begin, confining myself to its main forms. Substitutional connexity is brought about by s y n t a g m a t i c s u b s t i t u tion, 13 and syntagmatic substitution normally consists in the coreferential

7

8

9 10

11

12

13

several ranks (and several variants of ranks) of units above that of the sentence, seems to be intended as a text-grammatical notion. Cf. Heger (1974) and Heger (1976). For the notion 'co-textual', especially as opposed to the notion 'con-textual', see Petöfi (1971: 224). The requirement of cotextual connectedness by certain linguistic means is obviously weaker than the requirement of presuppositionality, and on the whole the latter may be indeed too strong, for there may be sentences within emic sentence sequences which one may omit (and whose order one may change) without destroying thereby the emicity of the sequence. For the notion of emic texts, see Harweg (1968/1979: 152 ff., 258 f.). That is to say that their recognizability is dependent on taking into account their cotextual environment. Among the contributions to the investigation of this phenomenon (contributions employing a number of different conceptual and terminological frameworks) are, for example, Nye (1912), Harris (1952), Karlsen (1959), Koch (1965), Greimas (1966), Heidolph (1966), Pike (1967), Nickel (1968), Harweg (1968/1979), Hartmann (1968), Pisarkowa (1969), Paduceva (1970), Palek (1970), Isenberg (1971), Brinkmann (1971), Petöfi (1971), Raible (1972), van Dijk (1972), Halliday and Hasan (1976), Viehweger (1976), Werlich (1976), Agricola (1977), Dressier (1978), Weinrich (1978), Beaugrande (1980), and van de Velde (1981). That these have to function as the starting-point in our investigations is explicitly stated by Issatschenko (1965: 80). Cf. Harweg (1968/1979). The notion of substitution is, among others, also used by Halliday and Hasan (1976). But with them substitution is, though it is characterized by them as a relation within the text (1976: 89), not a syntagmatic relation, but, in much the same way as Bloomfield's term, a paradigmatic one, its sentence-connecting force being secured only

32

R. Harweg

'resumption' or syntagmatic replacement of a certain expression by one or more than one other expression in the framework of a certain sentencesequence community.14 Syntagmatic substitution is thus not identical with mere coreference. The coreference obtaining, for instance, between the two occurrences of the expression a doll within two occurrences of the sentence I've got a doll, occurrences which, we will assume, follow each other, but have different addressees, is not a syntagmatic substitution, and the two sentences do not form an emic, that is a cotextually connected sentence sequence.15 They do not even, according to our criteria, form an etic one. In addition to coreferential syntagmatic substitutions (which I also call identity substitutions16) we have a second kind of syntagmatic substitution, one based on contiguity.17 These substitutions (which I call contiguity substitutions) may in principle always be traced back to identity substitutions,18 but in practice, that is from the point of view of surface structure, this often seems to be unacceptable. It therefore appears to be justified to regard them as a special type. Syntagmatic substitution is a procedure occurring in two directions, the anaphoric and the cataphoric.19 Of these I shall discuss, in this article, only the former, which is also the main one. Within the framework of clear-cut syntagmatic substitutions20 a form which I call 'substituendum' is anaphorically 'resumed', that is syntagmatically replaced, by one or more than one other form, which I call 'substituens'. Two-sentence sequences, within this type, may therefore be connected either through the relationship 'substituendum: substituens' or through the relationship 'substituens : substituens'. Syntactically, substituenda, in the framework of two-sentence sequences, are either phrases or sentences,21 whereas substituentia are only phrases or

14 15

16 17 18 19 20

21

indirectly, namely through the artificial proviso that its replacing force is restricted to repetitions of items (cf. 1976: 89). Cf. Harweg (1971c). In most discussions of coreference as a sentence-connecting or text-forming relation such cases are not taken into account. Cf. Harweg (1968/1979: 182 ff.). Cf. Harweg (1968/1979: 192 if.) and Brinkmann (1971: 824 ff.). Cf. Harweg (1968/1979: 194 f.). Cf. Bühler (1934: 122), Harweg (1968/1979: 54) and Raible (1972: 150 ff.). Cf. Harweg (1968/1979: 151 ff. and 178 ff.). Note that these clear-cut substitutions are, as is suggested by the different names for their initial and their non-initial terms, asymmetric relations and that such asymmetric relations cannot be brought about by mere recurrences, a kind of relation which by some seems to be erroneously thought of as having nearly the same connective qualities as (clear-cut) syntagmatic substitutions. Quite often, recurrence even lacks coreferentiality — a case in which its sentence-connecting qualities are still weaker. Mere recurrence or even weaker, mere similarity of expressions is, for instance, the basis of the concept of discourse in the framework of the discourse analysis of Harris (1952). On recurrence see, among others, Beaugrande (1980: 134ff.). Strictly speaking, they may also be sequences of sentences.

Sentence Sequences and Cotextual Connexity

33

certain kinds of clauses (forming parts of sentences). Substituenda which are phrases are expressions like somebody, something, some time or somewhere; common nouns introduced by a non-generic indefinite article, expressions like some or several or cardinal numerals; and generically used common nouns in existential sentences.22 Substituenda which are sentences may be sentences of any kind including those which contain substituentia. Substituenda, in their turn, are phrase-expressions such as the anaphorical pronouns he\she\it, this, and that; the anaphorical pronominal adverbs there, at that time or tbusfso; non-generic common nouns23 introduced by this or that or the definite article24 and sometimes expanded by determinative (restrictive) attributes; and clauseexpressions such as this happened (because ...) or this he did (because ...).25 Consider these expressions as terms of syntagmatic substitutions within anaphorically connected two-sentence sequences according to the following typology 1. 1.1. 1.1.1. 1.1.2. 1.1.3. 1.2. 1.2.1. 1.2.2. 2. 2.1. 2.1.1. 2.1.2. 2.2.

identity substitutions substituendum : substituens phrase : phrase sentence : phrase sentence : constituent clause substituens : substituens phrase : phrase constituent clause : constituent clause contiguity substitutions substituendum : substituens phrase : phrase sentence : phrase substituens : substituens

Of these types consider the most specified sub-types. Of these, in turn, — type 1.1.1. may be exemplified by the — English, French, and German — two-sentence sequences — A. There is somebody26 lying on the ground. B: He seems to be sleeping — I have ordered a new car. But I shall have to wait for it several months 22 23

24

25

26

Cf. Tschauder (1979) and Herbermann (1981: 58 ff.). For the lexematic and word-formational aspects (with respect to their Substituenda) see Harweg (1968/1979: 178 ff.) and Herbermann (1981: 284 ff.), respectively. Substitutions with substituentia of this type are called 'Renominationen' by Wawrzyniak (1979: 80 f.). The more pronoun-like subclass of these substituentia is often called pro-forms. Cf. Steinitz (1974), Braunmüller (1977) and Beaugrande (1980). With some authors, however, the category of pro-forms, not unlike the traditional category of pronouns, overrides the distinction between what I call Substituenda and substituentia and loses thereby its sentence-connecting quality. The expressions in bold italics are the Substituenda and the substituentia.

34

R. Harweg

— Le roi avait une tres belle fille. II l'aimait plus que toute autre chose au monde and — A: Ich habe heute fünf Mark gefunden. B: Und wo hast du das Geld gelassen? — type 1.1.2. may be exemplified by the Italian and German two-sentence sequences — A. Carlo viene. B. Ciö mi piace and — Ich habe ihm immer geholfen. Er aber hat mir nie dafür gedankt — type 1.1.3. may be exemplified by the German and English two-sentence sequences A: Karl ist entlassen worden. B: Das (ist} aber nur (geschehen), weil er silberne Löffel gestohlen hat and — Charles has left. He has left (or: done so), although I had forbidden him to do so — type 1.2.1. may be exemplified by the Italian and French two-sentence sequences — L'amo molto. Mi ha sempre aiutato lui and — Cet komme, je le connais tres bien. U est le frere de mon ami — type 1.2.2. may be exemplified by the German two-sentence sequence — A: Er ist nicht gekommen, weil er keine Zeit gehabt hat. B: Nein, er ist nicht gekommen, weil er keine Lust gehabt hat — type 2.1.1. may be exemplified by the French and English two-sentence sequences — Hier, fat achete une vieille auto. Mais le moteur, j'ai du le remplacer\ echanger and — On the programme there was also a violin-recital. Guess who was the soloist — type 2.1.2. may be exemplified by the Italian and German two-sentence sequences — A: S met to di lavorare. B: E quäl e la ragione? and — Er half, wo er nur konnte. Aber der Dank blieb aus — and type 2.2. may be exemplified by the German two-sentence sequence — Ich habe sofort versucht, den Chef %u sprechen. Aber die Sekretärin sagte mir, er sei verreist. 2.1.2. Substitution-like deixis and coordination The cotextual connexity of the sentences of emic two-sentence sequences is not only brought about by — anaphoric — substitutions, but also by substi-

Sentence Sequences and Cotextual Connexity

35

tution-like deixis27 or coordinations,28 copulative or non-copulative. Substitution-like deixis is deixis which has the shape of syntagmatic substitutions. The objects of this kind of deixis are sentences — more precisely, utterances —, and the deictic expressions through which this kind of deixis refers to its objects are — metalinguistic — expressions which have the form of substituentia, but differ from them by the nature of their denotata. Such expressions are, for instance, expressions like this assertion or the pronoun this in sentences like This is not the case. Two-sentence sequences which owe their emicity, that is the cotextual connexity of their sentences, to such expressions are the German sentence sequences — A: Sie haben den Wagen gestohlen. B: Diese Behauptung kann ich leicht widerlegen and — A: Sie haben den Wagen gestohlen. B: Das stimmt nicht. If we consider two-sentence sequences with respect to coordinations, copulative or non-copulative, we soon notice that many, if not most coordinations occur in connection with syntagmatic substitutions. This is true above all of coordinations without frames,29 of copulative coordinations as found in the German sentence sequences — Das Schicksal unserer Zeitschrift beschäftigt mich sehr, und ich habe schon mit verschiedenen Leuten darüber gesprochen and — Ich habe auch mit Karl über die Zeitschrift gesprochen, und er hat mir versichert, mit in die Redaktion eintreten %u wollen and of non-copulative ones as manifested by the English, Italian, and German sentence sequences — Unfortunately, Charles cannot come, for his wife is ill — Gli scriverei con piacere, ma non ho il suo indiri^o and — Karl hat im Augenblick viel Zeit; dennoch kommt er %u nichts; for, all these coordinations contain, in addition to their conjunctions, substitutional relations, relations whose substituentia are the expressions darüber, er, die Redaktion (as a contiguity substituens), his wife, il suo indiri^p, and er respectively. Some coordinative conjunctions such as the German conjunctions deshalb and trotzdem manifest substituentia by themselves. If we, in addition to these clear-cut and non-elliptical substitution relations, also take into account unclear substitutions such as 7 : 7 or today : today1*® and elliptical ones (which are to be discussed later), we may ask ourselves 27 28 29

30

Cf. Harweg (1980: 304 ff.). Cf., among others, Lang (1977). In an as yet unpublished article entitled "Rahmenhaltige und rahmenlose «W-haltige Satzkoordinationen" I propose a distinction between sentence coordinations with and without certain kinds of 'framing* sentences. Cf. Harweg (1968/1979: 260 ff.).

36

R. Harweg

whether frameless coordinations are not necessarily connected with syntagmatic substitutions. At any rate, a frameless coordination without any kind of substitution, as manifested by the German sentence sequence *Das Schicksal unserer Zeitschrift beschäftigt mich sehr, und heute schneit es, seems to be unacceptable, and a frameless coordination like Ich würde Peter gerne schreiben, aber das Lexikon ist weg seems acceptable only when the expression das Lexikon is interpreted as das Lexikon, das ich dafür brauche. On the other hand, however, substitutions do not guarantee the well-formedness of coordinations either. What is also needed is the compatibility of their predicates. Not necessary (and apparently rare) are substitutions in "framed" coordinations. Framed coordinations are coordinations by which a preceding 'framing' sentence is, as it were, antithetically unfolded, unfolded either in an adversatively antithetical manner (in which the German framing sentence Wie geht es Karl und Petra eigentlich? is unfolded by the non-copulative coordination Karl geht es gut, Petra aber geht es schlecht) or in an enumeratively antithetical manner (in which the German framing sentence Karl und Petra haben %um erstenmal getrennt Urlaub gemacht is unfolded by the copulative coordination Karl ist nach Frankreich gefahren, und Petra hat eine Kreuzfahrt durchs Mittelmeer gemacht). The sentences of a framed coordination may, however, also be connected substitutionally. An example is the German sentence sequence Karl geht es gut, aber seiner Frau geht es schlecht. 2.1.3. Citing Besides syntagmatic substitution, substitution deixis and coordination there is a fourth procedure by virtue of which subsequent (but as a rule discontinuously subsequent) sentences may be regarded as connected, namely the procedure of citing.31 This procedure can take a number of forms, not only the well-known and obvious ones of citing in direct or indirect discourse, but also such unknown and latent a variety as is hidden in expressions of the type ein Herr Meier32, and all these varieties create relationships between — as a rule discontinuously — subsequent sentences (or a sentence and a sentence sequence). However, they do not create emic sentence sequences in the sense of cotextual connexity. Unlike syntagmatic substitution and even unlike substitution deixis (with which it shares the property of metacommunicativity), citing does not combine and integrate the cited sentence(s) and the citing sentence into one and the same text. It is, even in the framework of the most extensive of all text concepts proposed so far, namely the concept of 'macrotext' proposed by myself,33 not a procedure of intratextual, but of intertextual 31 32 33

Cf. Harweg (1980). Cf. Harweg (1971c). Cf. Harweg (1970). What I here call 'macro-text', in my German writings I have called 'Großraumtext'. Similarly, what I here call 'micro-text', in my German writings I have

Sentence Sequences and Cotextual Connexity

37

connection. The sentences between which it creates a relationship not only do not constitute an emic sentence sequence, they do not constitute an etic one either. For they are not confined to one and the same sentence-sequence community, but create, via a 'double member' acting, in a Janus-headed way, first as recipient and then as producer, a certain connection or communication between different sentence-sequence communities.34 By the way, such — lateral rather than linear — connections may be brought about not only through citations, but also through what I call 'semi-citations', that is through utterances of the type Charles is, as Peter has told me, ///,35 or even through mere repetition of utterances (by other speakers).

2.1.4. Substitutional

differentiations

Of the cotextual connections of the sentences of emic two-sentence sequences which are brought about by substitution, substitution deixis and/or coordination, particularly the Substitutional connections may be further differentiated according to several parameters, including — directness — 'clear-cutness' and — explicitness. In addition, the sentence sequences themselves may be differentiated according to parameters such as — — — —

distance between sentences directionality of address kind of 'linguality' and kind of linearity.

2.1.4.1. Types of Substitutional connections

The three parameters for further differentiating the Substitutional connections are, in principle, gradual parameters. On less close inspection, however, one may reduce them to binary oppositions, namely to those between — directness and indirectness — 'clear-cutness' and 'unclearness' and — explicitness and implicitness.

34 35

called 'Kleinraumtext'. A term like 'Makro-Textologie' (for what I have called 'GroßraumTextologie') is (also) used by Kalverkämper (1978: 385 ff.). Cf. Harweg (1971c). Cf. Harweg (1983).

38

R. Harweg

In the sense of this reduction and on the simplifying assumption that the substitutional connections consist of one substitution only I mean — by direct substitutional connections substitutions one of whose terms is an element of the first sentence (or this sentence itself) and one of whose terms is an element of the second sentence of the two-sentence sequence — by indirect substitutional connections substitutions one of whose terms is an element of a sentence preceding the first sentence of the two-sentence sequence (or that sentence itself) and one of whose terms is an element of the second, but none of whose terms is an element of the first sentence of the sequence (or this sentence itself) — by clear-cut substitutional connections substitutions the second of whose two terms connecting the two sentences of the sentence sequence is a clear-cut subsequential element, that is a substituens — by unclear substitutional connections substitutions the second of whose two terms connecting the two sentences of the sentence sequence is an unclear subsequential element, that is a substituendum-substituens — by explicit substitutional connections substitutions of whose terms the one occurring in the second sentence of the two-sentence sequence occurs explicitly — and by implicit substitutional connections substitutions of whose terms the one occurring in the second sentence of the two-sentence sequence occurs only implicitly. In exemplifying these categories let us begin with the direct substitutional connections. Two-sentence sequences displaying this kind of sentential connection are the sequence A: There is somebody lying on the ground. B: He seems to be sleeping with its substitution terms somebody and he, and the Italian sequence A: Carlo viene. B: do mi piace with its substitution terms Carlo viene and do. At the same time, these two sequences are examples of clear-cut and explicit substitutional connections. Indirect substitutional connections are frequently found in two-sentence sequences connecting two subsequent paragraphs36 or chapters, for in such cases it is not rare that the substitution term of the second sentence, that is of the first sentence of the new paragraph or chapter, refers to a substitution term occurring not in the last sentence of the preceding paragraph or chapter, but in one of its more remote sentences. The last sentence of this paragraph or chapter is then connected with its immediate follower sentence only indirectly, namely via that more remotely preceding sentence with which both of them are directly connected. Such a chapter-connecting two-sentence sequence is the sequence (The world has become more like that of Machiavelli than it was,) and the modern man who hopes to refute his philosophy must think more deeply than seemed necessary 36

Cf. Paduceva (1965) and Koch (1971: 206 ff.).

Sentence Sequences and Cotextual Connexity

39

in the nineteenth century. (New chapter, entitled Erasmus and More.) In northern countries the Renaissance began later than in Italy.2'1 For, the last explicit predecessor occurrences of the subsequential substitutional terms the Renaissance and Italy occur, combined into the expression Renaissance Italy, at a distance of more than half a page beforehand in the book from which the sequence is taken, their initial predecessor occurrences, that is their substituenda, occurring in the first sentence of the chapter preceding the preceding one, namely in the sentence The modern as opposed to the medieval outlook began in Italy with the movement called the Renaissance.^ From the point of view of clear-cutness this substitutional connection is one of the comparatively clear-cut ones, for the expressions the Renaissance and Italy in the sentence In northern countries the Renaissance began later than in Italy are, as is shown by their lack of stress39 (which may be inferred from the structure of the sentence), clear-cut substituentia, that is clear-cut subsequential substitutional elements. This is not so in the chapter-connecting sentence sequence These things, too, will some day, when our knowledge is wider, lend themselves to a systematic survey and to fruitful generalisation. (New chapter, entitled The Use of Language.) The most difficult step in the study of language is the first step40 a sentence sequence whose sentences are likewise indirectly connected, for its second sentence could equally well open a new text. The expression the study of language is, substitutionally, neither a clear-cut substituens nor a clearcut substituendum, it is what I call a substituendum-substituens.41 To these belong expressions like I, unemphatically stressed proper names, and unemphatically stressed common nouns in generic use (outside existential sentences) — a usage characteristic above all of scientific texts. While unclear substitutional connections are particularly frequent in scientific texts, implicit substitutional connections are frequent in advertisements. A typical example is the German sentence sequence Reinigungskosten kann man senken. Bis ^u fünfzig Prozent. Typical of such sentence sequences is that the second sentence is an expansion of the first and that it is ellipticized42 except

37

38 39 40

41 42

This sentence sequence is taken from B. Russell: History of Western Philosophy. London: Allen & Unwin, »1962, p. 498 f. Cf. Russell, op. at., p. 483. For the textological function of stress, cf. Harweg (1971a), Nikolaeva (1979 and 1982). This sentence sequence is taken from L. Bloomfield: Language. New York: Holt & Comp., 1958 ( 933), p. 20 f. Cf. Harweg (1968/1979: 28) and the accentological modifications in Harweg (1971a). For the role of ellipsis from the point of view of cotextual connexity see Karlsen (1959), Gunter (1963), Issatschenko (1965), Dressier (1978), Halliday and Hasan (1976), and Beaugrande (1980). Frequently, sentence-connecting ellipsis is found in connection with comparative expressions.

40

R. Harweg

for the expansion.43 Unellipticized this sequence would read Reinigung*kosten kann man senken. Man kann sie senken bis %u fünfzig Prozent. 2.1.4.2. Types of sentence sequences Of the above-mentioned parameters serving the further differentiation of the sentence sequences themselves that of the distance between the sentences is a gradual one as well, but this graduality can hardly be reduced to a binary opposition. It is a graduality which as a rule is both etic and emic and whose stages are manifested, among others, by the intersentential distances of paragraph-internal two-sentence sequences, the intersentential distances of paragraph-connecting two-sentence sequences, the intersentential distances of chapter-connecting two-sentence sequences,44 and the intersentential distances of (micro-)text-connecting two-sentence sequences.45 Examples of these stages (the higher of which often show no direct substitutional connection between the two sentences) are, one after the other, the sentence sequences — Gelbs Buch ist einer der bedeutendsten Beiträge %ur Schriftforschung der neueren Zeit. Aber das darin vertretene „Stufengeset^ der Schriftentwicklung" (...) ist meiner Meinung nach (...) nicht aufrecht %u erhalten*1*' — Die Texte können, im Sinne linguistischen Sprachgebrauchs, sowohl schriftlich als auch mündlich konstituiert sein. (New paragraph.) Eine so verstandene Textologie hat %tt unterscheiden ^wischen einem textimmanent und einem texttranszendent definierten Textbegriff1 — 'Oh no, no!' he repeated, with the tears streaming down his face, Oh, no, no!' (New Chapter.) Meanwhile a remarkable problem was absorbing the attention of the Catholic Church** — and Von den Entführern fehlt weiterhin jede Spur. (Interval of several days.) Die Entführer des zehnjährigen Peter M. haben heute ein neues Lebenszeichen ihres Opfers geschickt. Unlike the parameter of intersentential distance, the sentence-sequence parameters of directionality of address, kind of 'linguality', and kind of linearity 43 44

45 46

47 48

Cf. Brinkmann (1971: 705 f. and 708 ff.). Brinkmann calls this phenomenon 'Satzkonstan2'. On this level we must possibly take into account whether or not the chapters have titles. At any rate, the structure of the second sentence of a chapter-connecting two-sentence sequence may be partly determined by a chapter title. Cf., in this respect, the sentence sequence In this respect the practice of the Greeks was to be commended. (New Chapter, entitled The Place of Love in Human Life). The prevailing attitude of most communities towards love is curiously twofold (from: B. Russell: Marriage and Morals, London: Allen & Unwin, I21958: 95 f.). Cf. Harweg (1979). This sentence sequence is taken from A. Schmitt: Entstehung und Entwicklung von Schriften (ed. by C. Haebler). Köln and Wien: Böhlau, 1980, p. 4. This sentence sequence is taken from Harweg (1970: 12). This sentence sequence is taken from L. Strachey: "Cardinal Manning", in: Strachey: Eminent Victorians. London: Chatto & Windus, 1966, p. 94f.

Sentence Sequences and Cotextual Connexity

41

are not gradual ones, but allow or even require the establishment of binary distinctions. These are those between — monologicity and dialogicity — intralinguality and interlinguality and — straight and 'angular' linearity. Of these properties those of monologicity and dialogicity are primarily, it is true, properties of speech-act sequences rather than of sentence sequences, but they nevertheless leave traces in the linguistic material of the utterances and may therefore, at least secondarily, be regarded as properties of sentence sequences as well.49 Thus, for instance, a German sentence sequence like Ich trinke jeden Abend Mb'brensaft. Das ist nämlich gesund is necessarily interpreted as monologic, whereas a German sentence sequence like A: Ich trinke jeden Abend Möhrensaft. B: Das ist ja auch gesund must, on account of its internal structure, be interpreted as dialogic.50 Furthermore, sentence sequences may be either intralingual or interlingual, that is they may consist either of sentences of one and the same language or of sentences of different languages. Of these, however, the latter are comparatively rare. They occur mostly in dialogues or more exactly, in dialogues between speakers of different languages having only a passive knowledge of the language of their interlocutor. If they are monologic, they normally contain, for instance in a letter, a metalinguistic announcement of the language change, an announcement with frequently substitution-deictic connection with the preceding sentences. The language of the announcing sentence (which, at the same time, should open or even form a new paragraph) may be either that of the preceding or that of the subsequent part. An example of the first case is the German-English sentence sequence Das hat er uns versprochen. (New paragraph.) After these lines in German I will conclude the letter with some lines in English, an example of the second case the GermanEnglish sentence sequence Nach diesen Zeilen auf Deutsch will ich noch einige auf Englisch anfügen. (New paragraph.) Tomorrow, as I already may have told you, we will leave for Japan. Finally, sentence sequences can, as has been noted, be rectilinear or angular. Rectilinearity is the normal case of linearity, that type of linearity which characterizes unilinear texts and the main line of plurilinear ones. Angularity or angular linearity, on the other hand, is the mark of those places where, within a plurilinear text, a line branches off from the main line.51 Such 'branch lines' are manifested, in written texts, by footnotes. Provided they branch off at the end of a sentence and not in the middle (which would be 49

50 51

Both from the point of view of speech-act sequences and from that of sentence sequences monologicity and dialogicity are discussed in Canisius (1986). Cf. Harweg (1971b). Cf. Harweg (1974).

42

R. Harweg

a parenthetical branching off), they form, together with the sentence from which they branch off (and the sentences preceding it), an angular line. Yet angular lines result not only when there are lines branching off from the main line of a text, but also when there are lines, that is other texts, joining the main line of a text. Such junctions occur in two main forms, microtextually and macrotextually. In the case of the first we have a kind of close junction, in the case of the latter a kind of 'distant' junction. In the case of a close junction there is only a short distance between the joining text or more exactly, its last sentence and the place where it joins the main text, in the case of a distant junction, however, there is a great distance between them, etically as well as emically. Closely joining texts or sentence sequences are as a rule summaries of nonfinal parts of a text which, as sometimes is the case with serialized novels, are placed in front of a subsequent part of the text and to which this subsequent part is connected in principally the same way as it is to the parts which precede it and which are summarized by the summary. Distantly joining texts or sentence sequences, in contrast, are texts or sentence sequences which join a continuative micro-text at a macrotextual distance and which do not do so at its beginning, but in the further course of the text. An example of such a distant junction is furnished by the German sentence sequence Von den Entführern fehlt weiterhinjede Spur. (Interval of several days or weeks or months.) (...) Gan% ähnlich sind ja wohl auch die Entführer des zehnjährigen Peter M. vorgegangen. It is a sequence whose first sentence, we will assume, is the final sentence of a certain item in a radio news broadcast, of an item about the same persons who are mentioned in the second sentence of the sequence, and whose second sentence is a noninitial sentence taken from a conversation held some time afterwards, a sentence which through the elliptical expression gan% ähnlich is connected with the immediately preceding sentences of the conversation and through the expression die Entführer des zehnjährigen Peter M. with the radio news item joining it 'laterally' and from a great distance. Just like sentences, emic sentence sequences may be well-formed or not well-formed. They may be not well-formed even when their constituent sentences are well-formed. This non-wellformedness may, for example in the case of substitutionally connected sentences, be an especially substitutional one, but it may also be based on a non-compatibility of predicates. Not wellformed sentence sequences of the first kind frequently stem from the fact that the substituens, as for instance the expression the hammer within the sentence sequence */ am looking all the time for a certain tool. But I cannot find the hammer, is semantically more specific than its substituendum, in this case the expression a certain tool, whereas not well-formed sentences of the latter

Sentence Sequences and Cotextual Connexity

43

kind are, among others, as for example the sentence sequence */ should like to buy a car. It is red, sentence sequences whose sentences contain denotata which in part belong to different worlds.52 2.2. More-than-two-sentence sequences

Many sentence-sequence-specific forms of non-wellformedness begin to appear only in sentence sequences consisting of more than two sentences, but the rules they violate are for the most part still unknown. One of the few of these rules which is sometimes mentioned in the literature is the one which specifies that longer sentence sequences must have a t h e m a t i c k e r n e l underlying them.53 Sentence sequences violating this rule are sentence sequences of some length whose principle of substitutional connection is that of alternating connection.54 A sentence sequence demonstrating the violation of this rule is the unacceptable German sentence sequence *Ich bin gestern in Berlin gewesen und habe mir dort einen Mantel gekauft. In Berlin gibt es drei Universitäten. An diesen Universitäten studieren zusammen etwa 100000 Studenten. Diese Studenten kommen aus den verschiedensten Ländern. Einige dieser Länder a/erden hin und wieder von einer Hungersnot heimgesucht. Zu diesen Ländern gehört auch Indien. Ich kenne dieses Land von einer Reise durch verschiedene Länder Asiens. Diese Reise hat etliche tausend Mark gekostet. Dieses Geld habe ich von einem Verwandten geschenkt bekommen. Dieser Verwandte wohnt in München. München hat viele interessante Museen.55

It is a sentence sequence whose unacceptability is substitutionally reflected by the fact that its substitutions employ only one substituens each, so that micro-themes change from sentence to sentence. Forms of non-wellformedness which begin to appear only in more-thantwo-sentence sequences are to be found not only in the domain of substitutional, but also in the domain of coordinative connections, both copulative and non-copulative. One of these forms is a purely numerical one. The rule it violates is simply the rule that certain sentence coordinations may not 52 52 54

55

Cf. Karttunen (1972) and Enkvist (1978: 121). Cf. van Dijk (1972). Cf. Harweg (1968/1979: 250 ff.) and Danes (1970: 75), where he calls this principle simple linear progression. In exemplifying it, however, he confines himself to two-sentence sequences and thus does not arrive at a statement about the violation of the rule under discussion. An example which the author obviously thinks to be of the same kind, but which differs from our example by the fact that its sentences are not connected by coreferential substitutions, but only by non-coreferential recurrences (cf. note 20 above) is the sentence sequence */ bought a Ford. The car in which President Wilson rode down the Champs Elysees was black. Black English has been widely discussed. The discussions between the Presidents ended last week. A week has seven days. Every day I feed my cat. Cats have four legs. The cat is on the mat. Mat has three letters presented by Nils Erik Enkvist in Enkvist (1978: 110 f.). Its cotextual connexity is, on account of these non-coferential recurrences, still weaker than that of the sentence sequence presented by me. In fact, its sentences are, unlike the sentences of my sentence sequence, not connected at all.

44

R. Harweg

combine more than two sentences, and the sentence coordinations of which this rule holds true are, among the copulative ones, the frameless (but not the 'framed'!) coordinations employing the conjunction and56 and among the non-copulative ones all those which several times consecutively make use of the same conjunction. Thus, one cannot, for instance, begin a letter with an enumerative German sentence sequence like *Das Schicksal unserer Zeitschrift beschäftigt mich sehr, ich habe mit verschiedenen Leuten, auch mit meinem Sohn, darüber gesprochen, sein Wunsch und Ehrgei^, die Redaktion %u übernehmen, ist sehr groß, und sein Selbstvertrauen macht mir Eindruck51 — a sentence sequence which manifests a frameless four-sentence coordination with the conjunction and, but one must choose instead two two-sentence coordinations and write Das Schicksal unserer Zeitschrift beschäftigt mich sehr, und ich habe mit verschiedenen Leuten, auch mit meinem Sohn, darüber gesprochen. Sein Wunsch und sein Ehrgeit^, die Redaktion %u übernehmen, ist sehr groß, und sein Selbstvertrauen macht mir Eindruck. Similarly, one cannot, employing one and the same conjunction, iterate noncopulative coordinations. Sentence sequences like — ^Unfortunately, Charles cannot come, for he is in hospital, for he has fallen downstairs — *G/i scriverei con placere. Ma non ho il suo indirisgp. Ma lo cerchero or — *Karl hat im Augenblick viel Zeit. Dennoch kommt er %u nichts. Dennoch scheint er %ufrieden %u sein are not acceptable. If sentential coordinations are to constitute longer sentence sequences, they must, if they are non-copulative coordinations, strive for a change in the series of conjunctions, and sometimes even a change between synonymous conjunctions may suffice. At any rate, a German sentence sequence like Karl hat im Augenblick viel Zeit. Dennoch kommt er %u nichts. Trotzdem aber scheint er ^ufrieden %u sein does not seem to be unacceptable. If, on the other hand, copulative sentential coordinations are to constitute longer sentence sequences, they require a f r a m e , that is, an abstract predecessor sentence which they specify enumeratively and which, to a varying degree, determines their structure. A more-than-two-sentence copulative senM 57

I have tried to show this in the as yet unpublished article cited in note 29 above. This sentence sequence is adapted from a letter of Thomas Mann's. Cf. Th. Mann: Briefe 1937-1947 (ed. by E. Mann), [Frankfurt/M.]: Fischer, 1963, p. 79.

Sentence Sequences and Cotextual Connexity

45

tential coordination which is hardly determined by its framing sentence is, together with its bracketed frame, the German sentence sequence (Die Persönlichkeit der neuen Patientin erregte ungewöhnliches Aufsehen.) Der diabetische General hörte %u murren auf, die Herren mit den entfleischten Gesichtern lächelten und versuchten, ihre Beine %u beherrschen, die Magistratsrätin Spat% schloß sich ihr als ältere Freundin an, und ein Schriftsteller, der seit einigen Wochen hier lebte, verfärbte sich

An analogous sentential coordination whose structure is, in a certain sense, clearly determined by its framing sentence is, together with its bracketed frame, the German sentence sequence (A: Was haben die vier denn eigentlich über Weihnachten und Neujahr gemacht?) B: Ach, Karl ist nach Frankreich gefahren, Petra ist nach Mallorca geflogen, Erika hat eine Kreuzfahrt gemacht, und Hein^ ist %u Hause geblieben.

If, finally, sentence sequences are to be built up through substitutional rather than coordinative sentence connections, it is, as we have seen, important to choose substitution patterns which do justice to the requirement that longer substitutionally connected sentence sequences must have a kind of thematic kernel underlying them, and since the substitutional chains alternating between substituendum and substituens plausibly do not do justice to this requirement, it is reasonable to assume that the n on -alternating substitutional chains of the type 'substituendum : substituensi : substituensa ... substituensn' will do this.59 In principle, this is indeed the case, but this principle is subject to a fundamental sentence-internal topological restriction. The members of the substitutional chain may not all, if the substitutional sentence connection is not by chance to manifest a 'framed' copulative coordination, occupy one and the same sentential position. Thus, they may, for instance, not all occupy the sentence-initial position, a position which, on the other hand, is favoured with respect to the substitutional reflection of the required underlying thematic kernel. An example of this restriction is the — unacceptable — sentence sequence * Henry Edward Manning who lived from 1808 to 1892 was an English cardinal. He was born at Totteridge, a village in Hertfordshire. He was educated at Harrow and Balliol College in Oxford. He was there president of the Union. He took, in 1830, a first in 'Greats'. He entered the colonial office. He left it to become a fellow of Merton. He took holy orders in 1832. He married in 1833. He remained childless. He lost his wife in 1837. He became archdeacon of Chichester in 1841. He was influenced by the tractarian movement and finally shaken by the 'Gorham case'. He therefore entered the Roman catholic church in April 1851. He was ordained priest in the following June. He went to Rome for some years of study. He established the congregation of the Oblates of St. Charles in Bayswaier in 1857. He became

58

59

This sentence sequence is adapted from Th. Mann's Tristan. Cf. Mann: Erzählungen (Stockholmer Gesamtausgabe). [Frankfurt/M.]: Fischer, 1958, p. 220. In the terminological framework of Danes (1970: 75 f.) this type is called "der Typus mit einem durchlaufenden Thema".

46

R. Harweg

provost of the metropolitan chapter of Westminster shortly afterwards. He succeeded Cardinal Wiseman as archbishop of Westminster in 1865. He ruled his diocese with %eal and energy. He was, towards the end of his life, especially concerned with the social evils of the time. He was (...) interested in the welfare of the poor. He was, however, described as ambitious, intolerant and bigoted in his attitude to non-Catholics by unsympathetic judges. He was an ardent advocate of the definition of papal infallibility at the Vatican council of 1870. He wrote several works afterwards to explain and defend the doctrine (...). He was made cardinal in 1875. He was expected by some to be elected pope at the conclave of 1878. He was a devout and conscientious bishop. He loved the poor and was loved by them.

This sentence sequence, a modified version of a lexicon entry about the English cardinal Manning (a version adapted for the sake of demonstration) creates, since all members of the 'thematic' substitutional chain with the substituendum(-substituens) Henry Edward Manning and the substituentia he\, he2, ... hen occupy strictly sentence-initial positions, an impression of monotony and infantility. But it is, it seems to me, not only monotonous and infantile, it is, in a certain sense, also unacceptable, and its unacceptability is not only a question of style, but also a question of grammar or more precisely, of text grammar. Its unacceptability obviously resides in the fact that it manifests a 'masked' copulative coordination, a sentence sequence which only lacks the copulative conjunction and before its last sentence, a conjunction which, at least in reading the sentence sequence aloud, is immediately expected. A copulative coordination consisting of so many sentences, however, can, as we have already seen, be only a framed one, and whether this sentence sequence may be called a framed one may fairly well be disputed; for its first sentence — and only this could be the frame — encompasses, with its predicate was an English cardinal, not Manning's whole career, but only its climax. But even if this sentence also encompassed the contents of the rest of the article, this sentence sequence (even with the conjunction and before its last sentence) does not seem to be acceptable; the reason for this lying in the fact that a framed sentence coordination of so great a length, if it is acceptable at all, cannot itself form a text. In fact, the sentence sequence just presented produces not so much the impression of a coherent text, but rather the impression of an enumerative set of answers to the question 'What do you know about Cardinal Manning?' What one has to demand of longer non-alternating substitution chains of the type 'substituendum : substituensi : substituensa ... substituensn' is therefore a variation of their sentence-internal topology, a variation which at the same time seems to imply certain changes of perspective, so that one can presumably say that the demand for certain kinds of change of perspective is also a demand vis-a-vis the type of substitution chain under discussion. These two demands are fulfilled partly by purely syntactical variations such as, for instance, the change between be and him or he and his, partly by purely topological variations (which sometimes may consist of as slight an alternation as that between a sentence beginning with He married ... and a sentence beginning with In 1841 he became ...), and partly by a combination of these two types of variation. In the original version of the lexicon entry about

Sentence Sequences and Cotextual Connexity

47

Manning we find above all the purely topological type of variation, but the other two types occur as well. The original version reads, slightly abridged, as follows:60 Manning, Henry Edward (1808—1892), English cardinal, was born at Totteridge, a village in Hertfordshire, and educated at Harro» and Balliol College, Oxford, where he was president of the Union and took a first in 'Greats' (1830); entered the colonial office, which he left to become a fellow of Merton, taking holy orders in 1832. He married in 1833, his wife dying childless in 1837. In 1841 he became archdeacon of Chichester but, influenced by the tractarian movement and finally shaken by the 'Gorham case', he entered the Roman catholic church in April 1851. Ordained priest in the following June, he went to Rome for some years of study. In 1857 he established the congregation of the Obiates of St. Charles (...) in Bayswater, becoming shortly afterwards provost of the metropolitan chapter of Westminster. He succeeded Cardinal Wiseman as archbishop of Westminster in 1865 and ruled his diocese with %eal and energy, being especially concerned towards the end of his life with the social evils of the time. His interest in the welfare of the poor was conspicuously manifested during the dockers' strike of 1889; but unsympathetic judges describe him as ambitious, intolerant and bigoted in his attitude to non-Catholics. He was an ardent advocate of the definition of papal infallibility at the Vatican council of 1870 and wrote several works afterwards to explain and defend the doctrine, (...). In 1875 he was made cardinal, and some expected him to be elected pope at the conclave of 1878. He was a devout and conscientious bishop, loving and loved by the poor.

In texts which are much longer and much more complex than this lexicon entry the substitutional chain substitutionally reflecting the main thematic kernel of the text is varied by yet other and at the same time much greater variations. One of these variation forms is the insertion of framed copulative sentence coordinations frequently mediated via contiguity substitutions,61 another one is the interruption of the main substitutional chain by substitutional chains reflecting secondary themes of the text. The first of these variation forms, the insertion of coordinations, is found particularly in the case of descriptions of persons or places in narrative texts. Yet the coordinations as such are often somewhat disguised. As the German sentence sequence Man vergegenwärtige sich einen Brünetten am Anfang der Dreißiger und von stattlicher Statur, dessen Haar an den Schläfen schon merklich ^u ergrauen beginnt, dessen rundes, weißes, ein wenig gedunsenes Gesicht aber nicht die Spur irgendeines Bartwuchses %eigt. Er war nicht rasiert — man hätte es gesehen; weich, verwischt und knabenhaft, war es nur hier und da mit einzelnen Flaumhärchen besetzt. Und das sah gan^ merkwürdig aus. Der Blick seiner rehbraunen, blanken Augen war von sanftem Ausdruck, die Nase gedrungen und ein wenig %u fleischig. Ferner besaß Herr Spinell eine gewölbte, poröse Oberlippe römischen Charakters, große, kariöse Zähne und Füße von seltenem Umfange^

makes clear, they, too, often show strong variations of perspective. To decide to what extent such variations are possible or even necessary we need, however, more detailed investigations. The same holds true of the question to what extent in such or other copulative coordinations or coordination-like 60 61 62

Cf. Chambers's Encyclopaedia, London, 1968, s.v. Manning. Cf. Danes's type 'Progression mit abgeleiteten Themen' in Danes (1970: 76 f.). The sequence is taken from Thomas Mann's Tristan. Cf. Mann: Erzählungen (Stockholmer Gesamtausgabe). [Frankfurt/M.]: Fischer, 1958, p. 223.

48

R. Harweg

sentence sequences the order of the sentences may be changed or even single sentences be omitted — changes or omissions from which, to the extent they are possible, one might even be tempted to draw the conclusion that the sentences in question fail to be connected,63 but from which, as I think, one may only infer that these sentences fail to presuppose each other or to be connected explicitly.64 The other variation form, the interruption by other substitudonal chains, raises the question of how, at the end of the interruption, one is to continue the interrupted chain, and leads us thus back to those 'distance substitutions' we discussed in connection with the chapter-connecting sentence sequences. Distant connections between sentences — connections which often lead to the formation of paragraphs and chapters and thus also to sentence-sequenceinternal hierarchizations65 — are to be found not only in the substitutional, but also in the coordinative or more precisely, in the copulatively coordinative domain; for as a kind of distance coordination (a concept unknown to traditional grammar) one could, for instance, interpret the connections of sentences more or less remote from each other by means of series of ordered numerals.66 Such series are either pure series of ordinals or series consisting of cardinals and ordinals in a certain order, the ordinals, if used correctly, being definite in the pure series of ordinals and indefinite in the mixed one. An example of a copulative distance coordination with a mixed series of numerals is (together with its bracketed frame) the sentence sequence (In many speech-communities there are certain speech-forms that are uttered only under restricted circumstances. They may be called improper speech-forms. A speaker who utters them outside the restriction is shamed or punished. The strictness of the prohibition ranges from ...) Some of these improper forms are forms which denote objects or persons that are not to be named in a casual way, or perhaps not to be named at all. In English, various terms of religion, such as ... Violation of the rule ... Another type of improper forms are the so-called obscene forms. In English ... A third type of improper forms are the ominous speech-forms. They are forms which one avoids, because they name something painful or dangerous ... A fourth (and last) type of improper forms is (, finally,) to be found among the names of game animals. ...

In this sentence sequence (which is an adaptation of a sentence sequence of Bloomfield's) the coordinated sentences (each of which opens a new paragraph) are to be conceived as interrupted by sentences connected with them not through coordinations, but through — explicit or implicit — substitutions. These interruptions — a counterpart to the insertions of coordinations in substitutionally connected sentence sequences — are, in the case 63 64 65

66

Cf. Figge (1971: lolff.). Cf. note 8 above. Cf. Koch (1968), Koch (1971: 206 ff.), Giilich and Raible (1974), and Wienold (1983) with a critique of Giilich and Raible. I have discussed this phenomenon in an as yet unpublished article entitled "Zahlwortreihengeprägte Mikro- und Makro-Koordinationen".

Sentence Sequences and Cotextual Connexity

49

of Bloomfield's original sentence sequence, comparatively short, but they could be much longer. Apart from this, the coordinated sentences of Bloomfield's original sentence sequence show certain structural variations which are not limited to the transition from cardinals to ordinals which is also to be found in my adaptation and which is regular. In Bloomfield's original version, where the coordinated sentences or its equivalents read: Some improper forms denote objects or persons that are not to be named in a casual way, or perhaps not to be named at all. ... Another direction of impropriety is the tabu on so-called obscene forms. ... A third type of improper connotation is less universal among us; ... In some communities one avoids the names of game animals, either during the hunt or more generally^1

the most striking phenomenon in this respect is the lack of the ordinal numeral in the last of the coordinated sentences or its equivalent and the lack of a final-sentence marker likey'inally. The absence of this marker, as well as that of the numeral, is by no means necessary, one may, on the contrary, even ask whether it is not to be disapproved of. 3. Relationship between etic and emic sentence sequences

Having discussed first etic and then emic sentence sequences, we will finally briefly deal with the relationship between these two categories, or more precisely, with their extensional relationship; for the intensional one (whose knowledge, by the way, is a presupposition for discussing the extensional one) has, at least implicitly, already been the object of the above discussions. From a purely logical point of view, there are the following three possibilities of extensional relations between etic and emic sentence sequences: 1) Etic and emic sentence sequence are coextensive 2) One and the same etic sentence sequence consists of several emic ones and 3) One and the same emic sentence sequence consists of several etic ones. Of these three possibilities the first is undoubtedly the one which is most frequently realized. It constitutes, as it were, the normal case. The possibility which is next most frequently realized is the second one; for etic sentence sequences consisting of several emic ones occur, among others, in radio news broadcasts, letters, diaries, and dramas. The third possibility is the most rarely realized one, and at first sight one might even ask whether it can be realized at all. This holds true at least, if we admit that etic sentence sequences, not unlike the emic ones, may be discontinuous, for in that case even many of those emic sentence sequences which show macrotextual distances between some of their sentences, namely at least the oral ones among these sentence 67

Cf. Bloomfield: Language. New York: Holt & Comp., 1958 (Ί933), p. 155.

50

R. Harweg

sequences, are, etically, not pluralities of sentence sequences. On the other hand, however, there seem to exist certain written macrotextual emic sentence sequences which are pluralities of etic sentence sequences. These include emic sentence sequences consisting of continuations of news reports about one and the same actual theme distributed over several numbers of a certain newspaper. For, considering the fact that the different numbers of a newspaper do not, etically, manifest a sentence sequence, but a sequence of sets of sentence sequences and that there are no criteria allowing us to sequentialize these sets of sentence sequences in a definite way, there is no possibility either, to interpret these emic sentence sequences as single etic ones. They therefore constitute one of those rare cases where emic sentence sequences consist of more than one etic one.

References Agricola, Erhard 1977 "Text — Textaktanten — Informationskern", in: F. Danes and D. Viehweger (eds.): Probleme der Textgrammatik II (studia grammatica XVIII). Berlin: Akademie-Verlag, 11-32. de Beaugrande, Robert 1980 Text, Discourse, and Process: Toward a Multidisciplinary Science of Texts. London: Longman. Braunmüller, Kurt 1977 Referen^ und Pronominalisierung: Zu den Deiktika und Proformen des Deutschen (Linguistische Arbeiten 46). Tübingen: Niemeyer. Brinkmann, Hennig 197l2 Die deutsche Sprache. Düsseldorf: Schwann. Bühler, Karl 1934 Sprachtheorie: Die Darstellungsfunktion der Sprache. Stuttgart: Fischer. 2. Auflage 1965. Canisius, Peter 1986 Monolog und Dialog: Untersuchungen %u strukturellen und genetischen Beziehungen ^wischen sprachlichen Solitär- und Gemeinschaftshandlungen. Bochum: Brockmeyer. DaneS, FrantiSek 1970 "Zur linguistischen Analyse der Textstruktur", Folia Linguistica 4, 72—79, van Dijk, Teun A. 1972 Some Aspects of Text Grammars. The Hague: Mouton. Dressier, Wolfgang U. 1978 "Vorstudien zur Erstellung einer textsemantischen Basisstruktur, skizziert am Beispiel indogermanischer Sprachen", in W. Dressler (ed.): Textlinguistik. Darmstadt: Wissenschaftliche Buchgesellschaft, 220-231. Enkvist, Nils Erik 1978 "Coherence, Pseudo-Coherence, and Non-Coherence", in J.-O. Östman (ed.): Cohesion and Semantics (Meddelanden fran stiftelsens for Abo Akademi Forskningsinstitut, nr. 41), Abo, 109-128. Figge, Udo L. 1971 "Syntagmatik, Distribution und Text", in: W.-D. Stempel (ed.): Beiträge %ur Textlinguistik. München: Fink, 161 — 181. Greimas, Algirdas Julien 1966 Semantique structurale: Recherche de methode. Paris: Larousse.

Sentence Sequences and Cotextual Connexity

51

Gülich, Elisabeth und Raible, Wolfgang 1974 "Überlegungen zu einer makrostrukturellen Textanalyse — J. Thurber, The Lover and his Lass", in E. Gülich & K. Heger & W. Raible (eds.): Linguistische Textanalyse: Überlegungen %ur Gliederung von Texten. Hamburg: Buske, 73 — 126. Gunter, Richard 1963 "Elliptical Sentences in American English", Lingua 12, 137 — 150. Halliday, Michael A. K. und Hasan, Ruqaiya 1976 Cohesion in English. London: Longman. Harris, Zellig 1952 "Discourse Analysis", Language 28, l -30 and 474-494. Hartmann, Peter 1968 "Zum Begriff des sprachlichen Zeichens", Zeitschrift für Phonetik, Sprachwissenschaft und Kommunikationsforschung 21, 205 — 222. Harweg, Roland 1968 Pronomina und Textkonstitution, München: Fink, 2. Auflage 1979. 1970 "Zur Textologie des Vornamens: Perspektiven einer Großraumtextologie", Linguistics 61, 12-28. 1971a "Die textologische Rolle der Betonung", in W.-D. Stempel (ed.): Beiträge %ur Textlinguistik. München: Fink, 123—159. 1971b "Quelques aspects de la constitution monologique et dialogique de textes", Semiotica 4, 127-148. 1971c "Zur Textologie des Typus ein Herr Meier: Perspektiven einer nichtsubstitutionellen Textologie", Orbis 20, 323-346. 1971d "Zum Verhältnis von Satz, Hauptsatz und Nebensatz", Zeitschrift für Dialektologie und Linguistik 38, 16—46. 1974 "Bifurcations de textes", Semiotica 12, 41-59. 1979 "Satzgrenzen und Satzabstände und das Verhältnis zwischen Satz- und Textlinguistik", in J. S. Petöfi (ed.): Text vs. Sentence I. Hamburg: Buske, 181-199. 1980 "Meta-assertorische, meta-propositionale und meta-ontologische Aussagen: Ein Beitrag zur Typo- und Textologie metakommunikativer Rede", l'olia Linguist na 14, 283-328. 1983 "Semi-metakommunikative Äußerungen", Zeitschrift für Phonetik, Sprachwissenschaft und Kommunikationsforschung 36, 287 — 302. Heger, Klaus 1974 "Signemränge und Textanalyse", in E. Gülich, K. Heger und W. Raible (eds.): Linguistische Textanalyse: Überlegungen %ur Gliederung von Texten. Hamburg: Buske, 1-71. 19762 Monem, Wort, Sat^ und Text. Tübingen: Niemeyer. Heidolph, Karl-Erich 1966 "Kontextbeziehungen zwischen Sätzen in einer generativen Grammatik", Kybernetika 2, 274-281. Herbermann, Clemens P. 1981 Wort, Basis, Lexem und die Grenze ^wischen Lexikon und Grammatik: Eine Untersuchung am Beispiel der Bildung komplexer Substantive. München: Fink. Isenberg, Horst 1971 "Überlegungen zur Texttheorie", in: J .Ihwe (ed.), Literaturwissenschaft und Linguistik: Ergebnisse und Perspektiven l, Frankfurt: Athenäum, 150—173. Issatschenko, Alexander V. 1965 "Kontextbedingte Ellipse und Pronominalisierung im Deutschen" (1965), in W. Dressler (ed.): Textlinguistik. Darmstadt: Wissenschaftliche Buchgesellschaft (1978), 79-92. Kalverkämper, Hartwig 1978 Textlinguistik der Eigennamen. Stuttgart: Klett-Cotta.

52

R. Harweg

Karlsen, Rolf 1959 Studies in the Connection of Clauses in Current English, Bergen: Eides boktrykkeri. Karttunen, Lauri 1972 "Textreferenten", in F. Kiefer (ed.): Semantik und Generative Grammatik. Frankfurt/ M.: Athenäum, vol. l, 175-197. Koch, Walter A. 1905 "Preliminary Sketch of a Semantic Type of Discourse Analysis", Linguistics 12, 5-30. 1968 "Problems in the Hierarchization of Text Structures", Orbis 17, 309—342. 1971 Taxologie des Englischen. München: Fink. Lang, Ewald 1977 Semantik der koordinativen Verknüpfung (studia grammatica XIV). Berlin: AkademieVerlag. Nickel, Gerhard 1968 "Kontextuelle Beziehungen zwischen Sätzen im Englischen", Praxis des neusprachlichen Unterrichts 15, 15—25. Nikolaeva, Tatjana M. 1979 "Akcentno-prosodiceskie sredstva vyrazenija kategorii opredelennosti-neopredelennosti", in: T. M. Nikolaeva (ed.): Kategorija opredelennosti-neopredelennosti v slavjanskich i balkanskich ja^ykach, Moskva: Izdatel'stvo Nauka, 119—174. 1982 Semantika akcentnogo vydelenija. Moskva: Izdatel'stvo Nauka. Nye, Irene 1912 Sentence Connection: Illustrated chiefly from Livy. Dissertation New Haven: Yale University. Weimar: Wagner. Paduceva, Elena 1965 "O strukture abzaca", Ufenye %apiski Tartuskogo universiteta, Tartu. 181, 284 — 292. 1970 "Anaphoric Relations and their Representation in the Deep Structure of a Text", in: M. Bierwisch and K.-E. Heidolph (eds.): Progress in Linguistics. The Hague: Mouton, 224—232. Palek, Bohumil 1970 Cross-Reference: A Study from Hyper-Syntax. Praha: Universita Karlova. Petöfi, Janos S. 1971 Transformationsgrammatiken und eine ko-textuelle Texttheorie. Frankfurt/M.: Athenäum. Pike, Kenneth L. 19672 Language in Relation to a Unified Theory of the Structure of Human Behavior. The Hague: Mouton. Pisarkowa, Krystyna 1969 Funkcje skiadnioive polskich %aimkow odmiennych. Wroclaw: Zaklad narodowy im. Ossolinskich. Raible, Wolfgang 1972 Sat% und Text: Untersuchungen %u vier romanischen Sprachen. Tübingen: Niemeyer. Steinitz, Renate 1974 "Nominale Pro-Formen", in W. Kallmeyer & W. Klein & R. Meyer-Hermann & K. Netzer & H.-J. Siebert (eds.): Lektürekolleg ^ur Textlinguistik, Bd. 2: Reader. Frankfurt/M.: Athenäum, 246 — 265. Studnicki, Franciszek 1970 "Traffic Signs", Semiotica 2, 151 -172. Tschauder, Gerhard 1979 Existen%sät%e: Eine textgrammatische Untersuchung vor dem Hintergrund bestimmter Positionen der modernen Sprachphilosophie. München: Fink, van de Velde, Roger 1981 Interpretation, Kohären^ und Inferen% (Papiere zur Textlinguistik, Bd. 33). Hamburg: Buske.

Sentence Sequences and Cotextual Connexity

53

Viehweger, Dieter 1976 "Semantische Merkmale und Textstruktur", in F. Danes' and D. Viehweger (eds.): Probleme der Textgrammatik (studia grammatica XI). Berlin: Akademie-Verlag, 195-206. Wawrzyniak, Zdzislaw 1979 Untersuchungen %ur Textbildung im Deutschen (Uniwersytet Jagiellonski, Rozprawy habilitacyjne Nr. 30). Krakow. Weinrich, Harald 1978 "Die Textpartitur als heuristische Methode" (1972), in: W. Dressler (ed.): Textlinguistik. Darmstadt: Wissenschaftliche Buchgesellschaft, 391—412. Werlich, Egon 1976 A Text Grammar of English. Heidelberg: Quelle & Meyer. Wienold, Götz 1983 "Narrative Texts and Models of Hierarchical and Sequential Structure", in M. Faust & R. Harweg & W. Lehfeldt & G. Wienold (eds.): Allgemeine Sprachwissenschaft, Sprachtypologie und Textlinguistik: Festschrift für Peter Hartmann. Tübingen: Narr, 417-430.

WALTRAUD BRENNENSTUHL Speech Act Sequences /. The primacy of speech act sequences

It was the hope of the early speech act philosophers that to study speech acts would be to study linguistic communication. They (Austin, Searle) wanted to get away from the study of structural units like sentences and words and instead concentrate on action-theoretic units like the issuance of certain linguistic entities in the performance of speech acts.1 It is beyond doubt that they took thereby a step in the right direction (although not following the programme completely), namely a step towards taking seriously the simple fact that linguistic communication is a form of human behaviour and that the study of language and meaning should accordingly be a behavioral science. On the other hand, they concentrated on the unit of the single speech act in isolation, not fully realizing the fact that this concept of a single speech act would again be a construct. They fostered the hope that speech could be imagined as a simple succession of speech acts which is again a mere structuralistic approach. In any case, the compositional principles, according to which speech acts would be composed into sequences of speech acts were totally ignored by the speech act philosophers. On the other hand, when we have a look at spoken language, we are confronted with actual discourse. Speakers talk to each other, they take turns, they interrupt each other, they address and answer to each other. There are rather lengthy and likewise extremely short utterances. The difficulty for speech act analysts which arises when they are confronted with actual speech would be how to demarcate their speech act units. Where would a certain speech act begin and end? How and why would certain speech acts follow each other? "The reason for concentrating on the study of speech acts is simply this: all linguistic communication involves linguistic acts. The unit of linguistic communication is not, as has generally been supposed, the symbol, word or sentence, —, but rather the production or issuance of the symbol or word or sentence in the performance of the speech act. — More precisely, the production or issuance of a sentence token under certain conditions is a speech act, and speech acts are the basic or minimal units of linguistic communication." (Searle 1969: 16). "The characteristic grammatical form of the illocutionary act is the complete sentence (it can be a one-word sentence)" (25).

Speech Act Sequences

55

It is obvious that a speech act analyst who confines himself to the characteristics and conditions of single speech acts would have very little to contribute to a theory of spoken discourse. What matters here is the dynamical aspect of language. Not the speech act units but the speech act sequences and their dynamical properties must be in the focus of the study of discourse. Dynamical sequences of speech acts exhibit certain structural patterns that deserve to be investigated. Let us give some more arguments for our contention that the speech act sequence is the basic phenomenon of discourse (spoken as well as written). The address-reply-principle of spoken language: 1) When speaking, it is unnormal not to address oneself to someone. 2) It is unnormal not to get a reply (after having spoken to someone). In other words, speaking is normally speaking to somebody, and because one addresses oneself to somebody it is not normal for the addressee to run away after having been spoken to, but to give a reply. The address-reply-principle results naturally in the turn-taking of spoken discourse, i.e. in speech act sequences. In written language (texts) the addressreply-principle works in a refined manner. 1) In texts, it is unnormal not to address oneself to possible readers (and thereby to provoke comments or replies). 2) It is unusual not to perform a sequence of speech acts when writing a text. An argument, a story, a letter, a speech are larger text units embodying a number of acts in a certain sequence and with a certain structure. (Short notes on a piece of paper or telegrams seem to be an exception to this point. But even they usually express more than one proposition. The information to be conveyed, the specification of the writer, possibly the date.) So, the address-reply-principle results in the text structures and organizational schemes of written language. In order to decide the question whether the speech act or the speech act sequence is the basic phenomenon of discourse there remains of course the task of specifying how one speech act can be singled out from the more or less continuous flow of speech. Speech act specification:

One speech act is an utterance chunk, delimited by pauses or change of speaker, which makes sense and has an effect. One speech act may be as short as one word, for example: "Hello", "Yes", "No", "Bye", "Come!", "Push". One speech act may as well be rather long: stating, suggesting, apologizing, explaining.

56

W. Brennenstuhl

In our speech act specification above we make use of observable and of intentional aspects of the speech act. Observable aspects are the pauses in the flow of speech and changes of speaker; intentional aspects are sense and effect (or point) of what is being said, and of course the purpose or intention with which the utterance is made. In speech act theory we find the unit of one speech act only specified by these intentional aspects: a speech act is what a speaker intends to communicate. For empirical studies of spoken language such intentional specifications are useless, however. How should the intention of the speaker be observed? The empirical linguist must decide from the observable aspects of the speech act how the flow of speech is to be divided into speech acts units. For empirical purposes the specification of one speech act amounts to the specification of one turn (German: Redebeitrag). Turn specification:

One turn is an utterance chunk between pauses or speaker changes.2 The reason why one turn usually entails one speech act is that speakers simply stop talking when they have reached her or his goal of speaking. So, depending on the complexity of the idea or content to be expressed, the flow of speech constituting the speech act will be longer or shorter. As we realize it makes some trouble to get at the unit of the speech act at all. Both on the theoretic level of speech act theory as well as on the empirical level of spoken language analysis. In speech act theory the crucial and hopelessly difficult question arises what kind of a "thing" an intention is. Since this basic notion cannot remain unexplained. As we are told by Searle in the recent years it takes the development of an entire new branch of phenomenological philosophy to explain the concept of intention. In spoken language analysis we already mentioned the difficulties of spotting the observable clues with which speakers mark the beginning and end of the expression of separable goals of speaking. All these theoretic as well as empirical difficulties with the specification of the speech-act unit support our argument of the primary of speech act sequences as the basic phenomenon of discourse.

2. The dynamics of speech-act sequences

Rieser, in his paper "Dynamik in natürlichen Diskursen"3, distinguishes three types of dynamics of natural discourses (His material are transcribed "Beratungsgespräche"): 2

3

"Ein Redebeitrag ist, grob gesagt, das, was eine Sprecherin oder ein Sprecher sagt, während sie das Wort haben; er ist die Rede einer Sprecherin, die durch Schweigen oder die Rede eines anderen Sprechers begrenzt ist." (Trömel-Plötz 1982: 191). The paper was given as a talk by H. Rieser at the annual DGfS meeting at Bielefeld in February 1984.

Speech Act Sequences

57

1. Succession of discourse-contributions It comes about by the simple fact that speakers take turns in a discourse and refer to one or several of the previous discourse contributions. 2. Dynamics of a single discourse-contribution Often one discourse contribution is shaped in several onsets and trials, with hesitations, self-corrections and break-offs. These phenomena stem from the hypotheses the speaker builds about the manufacturing of the intended thoughts by the addressee. The speaker tries to control this thought-manufacturing by his words in a way conforming to his goals of speaking. 3. Dynamics of a whole discourse The macro-structure of discourses, which is followed by the participants in a discourse. Rieser is right in his claim that all three types of discourse dynamics deserve attention by the linguist. In the recent years the author, together with co-author Ballmer, propagated the idea that all information about language is to be found in language itself, namely in the lexicon of a natural language. Information about processes (and their dynamics) is to be found in the verb lexicon, and information about linguistic processes (e.g. discourses) is to be found in the lexicon of speech activity verbs. As we showed in our book Speech Act Classification* a lexical analysis, i.e. a classification, of speech activity verbs leads to an understanding of human verbal behaviour. The structuring of the speech activity lexicon (as a relevant part of the whole lexicon) into semantical categories reflects an ontological and conceptual structuring of linguistic behaviour in several aspects. In the following I would like to show how the mentioned three types of discourse dynamics, i.e. succession of discourse contributions, dynamics of a single discourse contribution and dynamics of a whole discourse, play an important role in our holistic lexical approach. Our conception that the knowledge about speaking lies in the speech activity lexicon leads to our method of holistic lexical analysis, which is a thorough empirical method, starting from the 5,000 German or English speech activity designating verbs (or verb phrases), grouping them together into 500 categories and ordering these categories into 24 models and typifications and those into 8 model groups. Categories are built according to 4

"From the point of view of linguistic semantics the dynamical aspect of language becomes particularly relevant when sequences of linguistic actions are considered. Dynamical sequences of speech acts and more generally of actions and processes exhibit special structural patterns. We hit here a point where the dynamical and the structural aspects of language intertwine. ... Thus our Lexical Analysis based essentially on the two semantic relations meaning similarity and presupposition leads to context structures guiding the dynamics of language based on the sequences of speech acts ..." (Ballmer-Brennenstuhl 1981: 47f.).

58

W. Brennenstuhl

intensity begin starting situation

end

run off elapse go on

resulting situation

Categories of a model

a) argumentative discourse (Verbal Struggle Model)

win/ lose/ cooperate

attack/ defend

makeclaims

dissent

b) single discourse-contribution (Turn-Taking Model)

announce os. take one's turn

have one's turn speak

finish

c) single discourse-contribution (Theme Models)

think

correct os. improve repeat summarize

thematize structure reorganize

verbalize

d) a whole discourse (Discourse Model)

ask for participation

address

invite Figure 1

discuss

fi finish

Speech Act Sequences

59

semantical similarity, and the model and model group structure is built by presupposition relations holding between the categories and models. The result of the procedure of categorization is a complex hierarchical structure which reveals the ontological and conceptual structuring of linguistic behaviour in its various facettes. The idea on which the formation of models is resting is the idea that linguistic acts occur in sequences and presuppose one another. In Speech Act Classification we made very clear that the first type of dynamics, the succession of discourse contributions, lies at the heart of the formation of models, especially of those showing the Ballmer/Brennenstuhl hat structure. But as can be seen in Figure 1 not only the succession of discourse contributions, e.g. in an argumentative discourse (see a), has a hat-structure organization, but also the other types of dynamics, as exemplified in b), c) and d): It does not matter whether the linguistic processes modelled by a hat-structure occur on a larger or a smaller time scale, whether a discourse as in a) and d) takes minutes or hours or (with interruptions) even months or years or whether a turn takes only milliseconds, the process structure of a beginning, ongoing and end phase can nevertheless be observed, i.e. categories of lexical items for those phases exist in the lexicon. The models which express the first type of dynamics, the succession of discourse contributions, most clearly, are the hat-structured models Verbal Struggle (for argumentative, private discourse) and Institution (for institutional, normguided, public discourse). They stand for the model group of Interaction, the interactive function of linguistic behaviour. For the case of a minimal sequence of two speech acts like question — answer, appeal — reply, order — acceptance we must go one step back in the presuppositional hierarchy of models to the model Enaction. The discourse character of those speech act pairs is more monological than dialogical, therefore not yet interactive in the full sense. The models and model groups are arranged in presuppositional order. (Cf. table on p. 60.) EM —> EN means Enaction presupposes Emotion, as you cannot, say, ask someone to do something (an enaction) without expressing the wish to have the "thing" asked for be done. In the same way presupposes Discourse Interaction, Appeal and Expression, because the higher linguistic functions in typical cases can not be used without the lower ones. One level above the interactional models in the hierarchy of Speech Activity we find the Discourse Models. In the Discourse Model itself the external organization of a whole discourse is covered (the third type of dynamics), in the Turn-Taking Model, the internal organization (the second type of dynamics). The highest level of complexity we attributed to the Text and Theme Models, which cover the socially entrenched processes of making and manipulating texts and the internal processes (the second type of dynamics) of the thematic structuring and its results.

60

W. Brennenstuhl

The Structure of the Speech Activities: (Speech Activity Thesaurus) main models and main groups

major model groups (linguistic functions)

EM

Emotion Model

Expression .

EN

Enaction Model

Appeal rr

KA i AV

Struggle Model

1

1

NO

I

W

1

°5

if influence

ο

D

1

X 1 T

1

\ N.

, ——^χ cSpeech ^η Activities

Arguing Devices Interaction^

/

Institutional Models

/

Valuation Models Discourse Models Text Models

Discourse

Theme Models

Speech Act Dynamics

Model group

Sample verbs

Geometro dynamic symbol

EX (expressives)

shout, curse, lament, salute, bitch at, scold

O-*·

EN (enactives)

suggest, beg, advise, compliment, persuade, seduce

IA (interactives)

consent, exact, dare, insult, bewilder, oppose, appease, defend, argue, refute, abandon

DC (discoursives)

report, predict, circulate, advertise, repeat, condense

O-*O-*· Q_^

Speech Act Sequences

61

As can be seen on the background of the whole Speech Activity Thesaurus the three types of dynamics as chosen by Rieser are not ordered according to complexity of degree of influence as we call the hierarchical dimension of the model system. The four major model groups of the Speech Activity Thesaurus indicate rather four basic dynamics of the underlying speech act processes: Expression (from the inner to the outer world), Enaction (from the I to the You), Interaction (from the I to the You and back), Discourse (internalized interaction, anticipated reaction of the You).5

3. The simultaneous structure of a speech act In the Speech Activity Thesaurus (SAT) we are confronted with one kind of activity, namely speaking. Activities from high up and from deep down in the hierarchy of the SAT are in that sense the same kind of activity as they are speaking activities (and nothing else). Therefore, they must have something in common which occurs in any speech activity. These processes common to the whole SAT are of course processes of articulation, thematization, verbalization, expression which function (as has been shown also by psychoand neurolinguistic research) largely automatically and uncontrolled. In fullfledged discourse (the normal case of verbal behaviour) we can distinguish Simultaneous structure of a speech act: specific speech act (goal of speaking)

intentional

discourse organization process turn organization process partially controlled

text organization process theme organization process valuation process emotion-expressive process articulatory process

5

Ballmer (1982: 522).

_

uncontrolled, automatical

62

W. Brennenstuhl

not only sequentially ordered single speech acts but also simultaneously ordered speech processes. The simultaneous structure of a speech act (in a discourse) can be modeled as simultaneously occurring processes that presuppose each other and build on another in a way that the presupposed one builds the socle of the presupposing one. The dimension which gives the direction of presupposition is the dimension of controllability: processes at the basis of the socle are least controllable and processes on top of it most controllable. Sequential structures occur on all levels, i.e. all speech processes have their phases (hats!) and can be sequenced. The speech act sequences in the original sense are to be found on the highest level. They form only the top of the iceberg. It must be stressed that the different types of dynamics of speech activity occur (in full fledged discourse) simultaneously', although they can be named and concentrated on separately as if they were single acts (therefore the separate verbs for processes that are only aspects of the whole manylevelled act).

4. The structure of the Speech Act Thesaurus

(SAT)

When we investigate linguistic behaviour we will immediately be able to distinguish frequent and rare kinds of discourse. People normally do not write books or scientific articles, they normally do not hold speeches, declare war, baptize or engage in marriage ceremonies. Speech Act Theory in its beginnings curiously enough concentrated on such rare cases of verbal

dimension discourse character

attribute | monological (mono) { polylogical (poly) privacy (priv)

privacy public (pub) self-organized (sorg) organization

normed (no)

characterized by one speaker several speakers among acquainted persons, not socially normed among unacquainted persons, to an audience, socially normed spontaneous, not following patterns or instructions, without fixed outcome of the discourse following patterns, eventually with instructions, in the extreme ritually and rigid

Speech Act Sequences

63

behaviour. But we should give the common kind of spoken discourse its place, in the center of our linguistic investigations. The commonest type of spoken discourse we would characterize as polylogical (with two or more participants), private (between persons with some acquaintance), and selforganized (spontaneous, not following rigid patterns). (Cf. the table on p. 62.) The three-dimensional cross classification of the SAT:

model groups EN BI'W

rt cfo ο

Ρ cfo ο

ft cro

TO

13

TO

f cfo

a. 0

η (1 n

c

ί*

ft CfQ

Ot?

ο Ο

8

•5

Ο

σο ο

c_ ET

r>

g

Ο

3 rt Ο

emotionalization

central linguistic behaviour

ritualization

64

W. Brennenstuhl

The attributes polylogical, private and selforganized (and their opposites monological, public, normed) comprise the three dimensions of discourse character (monologue — dialogue), privacy (social entrenching) and organisation which are apt to cross-classify the Speech Act Thesaurus. In a linear projection of the three-dimensional cross classification of the SAT (with the three ordering dimensions discourse character, privacy, organization) we clearly have the central linguistic behaviour as it occurs in private argumentation and public discussion in the center of the SAT.

discourse frequency centrality

priv mono sorg

pub mono sorg

priv mono no

pub mono no

priv dia sorg

pub dia sorg

priv dia no

pub dia no

EX

EN BEW

TX TH

TX TH

IA

DI TU ΤΗ

KIT

DI TU NO (

emotionalization

central linguistic behaviour

model groups

ritualization

The central linguistic behaviour is also the place where linguistic development takes place, it is the most dynamical part of language with respect to linguistic change and renewal. The two extremes of the SAT are too deeply rooted biologically (EM) or socially (RIT, NO §) for allowing a high change rate.

5. Spoken language analysis: its ends and limitations

In the preceding chapters I made clear, as I hope, how our lexical holistic approach can shed light on the dynamical structures of speech act sequences. Our approach is empirical in the sense that it builds on large vocabulary corpora (5,000 English speech activity verbs and 20,000 German verbs) as linguistic material, it is theoretical in the sense that the classificatory method

Speech Act Sequences

65

builds on the competence to assess similarity and presupposition relations between verb meanings. But it must be stressed that the classification is the result of an empirical method and not of a theoretic preconception. An alternative to our lexical approach for the study of the structures of speech act sequences is Spoken Language Analysis (in German: Analyse gesprochener Sprache), also called Discourse Analysis.6 In the following I would like to point at the merits, but also the problems and limitations of Spoken Language Analysis. Spoken Language Analysis claims to be empirical in the sense of analysing real natural spontaneous discourses produced in natural communication situations. (The methods of collecting the data have a lot in common with the methods of empirical ethnomethodology.) Now, since it is impossible to analyse processes which are present only for a while and then never occur again (each discourse is an absolute singularity), not the communication process itself is the analysed material but a transcription of a record of it. It takes a number of steps of abstraction (including theoretic linguistic conceptions) to arrive at such transcripts of natural discourses. The transcript of a tape record takes the two phases of segmentation (into phones, words, sentences) and codification (into orthographically correct texts, with punctuation signs and other structural signs for intonation, pauses, overlapping, stresses). The transcript reflects already several aspects of the understanding-process, namely the understanding of words and syntactical units and a gross speech act classification (declaration or question, marked by the absense or presence of question marks). Insofar linguistic competence of the linguist enters the analysed material to a not less extent than it entered the starting material of our lexical approach, namely word lists. It must be doubted whether transcripts really are the material one would need to fulfil the claim to analyse natural spontaneous discourses. Evidently verbal communication does not only take place on the acoustic level, one only has to call to one's mind the difficulties with talking on the telephone. In recent time linguists are aware of this fact and take video recordings of discourses. But this gives rise to a considerable complexivation of the transcription procedure. The communication situation, mimic expression and body-language are important for the functioning of the discourse — , but not all nonverbal processes can be coded verbally. (Just try to represent a facial expression verbally so that one is able to visualize it, even more so mimic dynamics!) I must confess that reading discourse transcripts I always have the feeling that some information is missing, so that I, to be honest, am unable to understand transcribed discourses. I can not imagine what the speakers really mean, what they understand, what their feelings are during the discourse or 6

A good presentation of Spoken Language Analysis can be found in Schank-Schoenthal (1983).

66

W. Brennenstuhl

after it. Is for instance a person who got some advice satisfied, or disturbed, is he or she rather angry at himself that she was not able to formulate her goals and conceptions more clearly? The clues for answering such questions about inner processes occurring during or after the discourse in the speaker's minds are totally missing. It is also evident that the more acquainted and familiar the participants in a transcribed discourse are (for instance parent — child, husband — wife, friends) the less understandable are the transcripts. Transcripts of discourses with some publicity, e.g. radio interviews, telephonecounseling, are better understandable, since the participants in the discourse are equally unacquainted with each other as the analyst is with the participants. Let us sum up our observations about transcripts and the information they do not include: 1. Transcripts give only partial information about the verbal side of discourses since the voices and therefore most of the intonation information is missing. 2. Nonverbal information like gestures, mimic expression, body language and other activities simultaneously to the speech activities, is missing. 3. When information about the speech information is given, it is too meagre (Schank/Schoenthal only mention 16 situational properties). 4. Totally excluded are the speaker's emotional experiences during or after the discourse, which are considerably important for the subjective assessment of the (outcome of the) discourse. Facit: The attempts to analyse spoken language aim at a written version of discourses where most of the ongoing real processes are disregarded. An analysis of spoken language in the full sense of all really ongoing processes (mentally and physically) is a hopeless undertaking. The question arises, what would a correct analysis of spoken language look like? Is it at all possible to give a correct analysis of real natural spontaneous discourses? In which situation is a participant of a discourse himself? Can we expect from the analyzing linguist to understand a discourse better than the participants in the discourse themselves? Let us handle the last two questions first. As speakers of a language we know the phenomenon of mistaking someone. When this happens reciprocally without being noticed by the speakers it is called in German "aneinander vorbeireden" (talk at crosspurposes with). A good deal of discourses take place although the participants are not coordinated, not in tune: everybody having his or her own interests in the discourse, his or her own conceptions about what has been said, or what the partner wanted to say. One follows side paths of thinking during a discourse, by which one is occupied in the moment. One does not listen all the time. No wonder, when the discourse in total is not really understandable for the analyst, being incoherent and uncoordinated. The analyst could only make full sense of the discourse contributions if he had access to what is happening inside the heads of the

Speech Act Sequences

67

speakers during the discourse! The speaker himself knows this at least for his own conscious thoughts. In this respect he is better off than the analyst. With respect to analyzing what is going on in his partner he may be in a worse position than the analyst, being preoccupied with his or her own involvement in the discourse. The analyst has two kinds of access to the discourse: a semantical-pragmatic approach and a hermenentic approach. In the semantical-pragmatic approach the analyst draws conclusions from the meaning of what is being said and the observed speech situation to what the speaker means. In the hermeneutic approach the analyst imagines to be in the role of a speaker in the discourse: he asks himself what he would have thought, experienced, intended when he had said what has been said. Both approaches allow less as well as more understanding of a discourse than the participants have. Less: Not everything going on in other persons' heads can be concluded or imagined (with the training and knowledge we have nowadays; maybe in the near future it can). More: More and correcter conclusions can be drawn by a trained analyst than by the discourse participants themselves. It can be re-imagined what they (out of negligence or ignorance) did not take into consideration. Finally, a stepwise systematic analysis from the coarse to the fine structure of a discourse can, because of the systematicity with which it is carried through, reveal more layers of meaning than the participants are aware of. To answer the question what would be a correct analysis of spoken language we can say now that there exists certainly a whole spectrum of correct analyses of a discourse, each of them taking into account different layers of meaning, with different depth of the analysis. A full blown comprehensive analysis would include the whole spectrum of correct analyses. 6. The stepwise systematic analysis of spoken discourse

A speech act analysis of spoken discourse is not a trivial task. Whether one starts from a tape recording, a video tape recording or a transcript, there always remains a degree of uncertainty which kind of speech act an utterance would express. Only a systematic analysis helps to keep the range of uncertainty small and to arrive at a full blown comprehensive speech act analysis. A systematic analysis would have to fulfil the following desiderata: 1. In the formation of utterance chunks, which one grants speech act quality, one should keep one's grid of segmentation constant; not use once a coarse and once a fine grid for the segmentation of the same discourse in the same survey. Rather, a discourse should be surveyed several times, starting from a coarse analysis (macrostructure) and resulting step by step in a fine analysis (microstructure).

68

W. Brennenstuhl

2. Throughout a survey of a discourse one should stay on the same level of abstraction. Utterances can be understood in a direct or more or less indirect way. Throughout a survey one should keep a certain degree of directness and proceed stepwise from a direct to a more and more indirect and connotation-loaded understanding. E.g. in a discourse, where a radiocounseler says to a librarian "Sie müssen ja die Bücher immer 'rein- und 'rausstellen": surface: "Sie müssen ja die Bücher immer rein und rausstellen." direct speech act: Statement conclusion: insinuation (you know only the outside of the books) indirect speech act: characterization of the addressee's professional duties indirect speech act including by demonstrating his ignorance of evaluation of speaker: librarians' duties the speaker reveals his incompetence When different opinions arise about which speech act quality should be assessed to an utterance, the difference is mostly a difference in the level of abstraction. When a record of the level to be used is being kept this will give speech act analyses of discourse more transparency and persuasiveness.

Literature Austin, John L. 1955 How to Do Things with Words. Cambridge, Mass.: Harvard University Press. (In German translation: Zur Theorie der Sprechakte. Stuttgart: Reclam, 1972.) Bach, Kent and Harnish, Robert M, 1979 Linguistic Communication and Speech Acts. Cambridge, Mass.: MIT-Press. Ballmer, Thomas T 1979 "Probleme der Klassifikation von Sprechakten" in Grewendorf, G. (ed.): Sprechakttheorie und Semantik. Frankfurt: Suhrkamp. 1982 "The Roots of Episteme: Archetypes, Symbols, Metaphors, Models, Theories", Poetics 12, 493-539. Brennenstuhl, Waltraud 1982 Control and Ability. Amsterdam: John Benjamins. 1980 "Are Speech Acts Really Actions?" in Ballmer, Th. T and Kindt, W. (eds.): Zum Thema Sprache und Logik (Papers in Textlinguistics 24). Hamburg: Buske, 335 — 342. Brennenstuhl, Waltraud and Ballmer, Thomas T 1981 Speech Act Classification. New York: Springer. Bierwisch, Manfred 1980 "Semantic Structure and Illocutionary Force" in Searle, J. R.; Kiefer, F.; Bierwisch, M. (eds.): Speech Act Theory and Pragmatics. Dordrecht: Reidel, 1 — 35. Cohen, Jonathan 1980 "Sprechakte" in Kussmaul, P. (ed.): Sprechakttheorie: Ein Reader. Wiesbaden: Akademische Verlagsgesellschaft Atheneion, 9—25.

Speech Act Sequences

69

Coulmas, Florian 1977 Rezeptives Sprechverhalten (Papers in Textlinguistics 15). Hamburg: Buske. Grewendorf, Günther 1972 "Sprache ohne Kontext" in Wunderlich, D. (ed.): Linguistische Pragmatik. Frankfurt: Akademische Verlagsanstalt Athenaion, 144—182. l lausser, Roland R. 1980 "Surface Compositionality and the Semantics of Mood" in Searle, J. R.; Kiefer, F.; Bierwisch, M. (eds.): Speech Act Theory and Pragmatics. Dordrecht: Reidel, 71—95. Meggle, Georg 1981 Grundbegriffe der Kommunikation. Berlin: de Gruyter. Rehbein, Jochen 1977 Komplexes Handeln: Elemente einer Handlungstheorie der Sprache. Stuttgart: Metzler. Ross, John R. 1970 "On Declarative Sentences" in Jacobs, R. A. and Rosenbaum P.S. (eds.): Readings in English Transformational Grammar. Waltham, Mass.: Ginn. Sadock, Jerrold M. 1974 Toward a Linguistic Theory of Speech Acts. New York: Academic Press. Schank, Gerd and Schoenthal, Gisela 1983 Gesprochene Sprache. Tübingen: Niemeyer. Searle, John R. 1969 Speech Acts: An Essay in the Philosophy of Language. Cambridge: University Press. (In German Translation: Sprechakte. Frankfurt: Suhrkamp, 1971.) 1979 a Expression and Meaning. Cambridge: University Press. 1979b "Intentionalität und der Gebrauch der Sprache" in Grewendorf, G. (ed.): Sprechakttheorie und Semantik. Frankfurt: Suhrkamp. 1981 "Intentionality and Method", Journal of Philosophy, 720- 733. Strawson, Peter F. 1968 "Wahrheit" in Bubner, R. (ed.): Sprache und Analysis. Göttingen: Vandenhoeck & Ruprecht, 96-116. 1974 "Intention und Konvention bei Sprechhandlungen" in Schirn, M. (ed.): Sprachhandlung — Existen·^ — Wahrheit. Stuttgart: Frommann-Holzboog, 74—96. Mrube, Werner 1980 "Lokutionärer Akt / Illokutionärer Akt" in Ritter, J. and Gründer, K. (eds.): Historisches Wörterbuch der Philosophie. Basel: Schwabe & Co. Bd. V, 329-343. 1981 Sprachanalytische Ästhetik. München: Fink. Trömel-Plötz, Senta 1982 Frauensprache: Sprache der Veränderung. Frankfurt: Fischer Verlag. Urmson, James O. 1974 "Einstufen" in Grewendorf, G. and Meggle, G. (eds.): Seminar: Sprache und Ethik. Frankfurt: Suhrkamp, 140—174. Wagner, Klaus R. 1978 Sprechplanung: Empirie, Theorie und Didaktik der Sprecherstrategien. Frankfurt: Hirschgraben.

EVA HAJICOVÄ AND PETR SGALL

Topic and Focus of a Sentence and the Patterning of a Text In the present paper we give a short survey of the linguistic questions concerning the topic-focus articulation (TFA) of a sentence (Section 1), as well as a characterization of some of the attempts to bring this articulation into an immediate connection with the structure of a text (Section 2); we then point out how a not fully immediate, but perhaps more legitimate connection between the two domains might be achieved (Section 3), if the relative salience of the elements of the stock of shared knowledge and the relationship between questions and answers in a dialogue are taken into account. 1.1. The importance of the dichotomy of topic and focus for the system of language and for the process of communication has been known since Weil (1844). After Gabelentz (1868, 1891) included the 'psychological subject and predicate' into the investigations of general properties of language systems, these notions were studied especially by Wegener (1885: esp. 29, 31), who stated that stress is the main means of expression of the dichotomy, further by Marty (1897), Jespersen (1924: 145ff.), and by Ammann (1928: esp. 141), who introduced the German terms Thema and Rhema. In the Prague School, Mathesius (1929, 1939, 1942) formulated a linguistic account of the dichotomy from the viewpoint of structural comparison of Czech and English, pointing out that the functions of subject and other sentence parts are concerned. Among his followers Firbas (1957, 1971, 1975) analyzed the interplay of this 'functional sentence perspective', the syntactic structure of the sentence and word order, showing that not only a dichotomy, but a whole scale or hierarchy of'communicative dynamism' is concerned. Danes (1970, 1974) systematically explored the relationships of 'theme' and 'rheme' to word order and intonation, as well as to the structure of a text. Similar questions were studied by Benes" (1968, 1973) with respect to German and by Adamec (1966), who analyzed the functions of the word order in Russian. The features of Slavonic languages with the so-called free word order and their comparison with English and French were always a rich source of interesting data, and thus TFA often was studied in connection with Russian (see esp. Krusel'nickaja, 1956; Dahl, 1969; 1974; Kovtunova, 1974). Halliday (1967) brought the questions of 'theme' and 'focus' nearer to the centre of interest of British and American linguists, and after Chomsky (1971)

Topic and Focus of a Sentence and the Patterning of a Text

71

these questions shifted their position from a marginal and rarely discussed issue to one of the central problems. Numerous analyses of TFA were presented in connection with transformational grammar or other frameworks. However, far less numerous are the attempts at a systematic description that would take into account the recursive properties of the syntactic structure and the composition of topic and focus units corresponding to phrases or syntagms, clauses, etc. Is it a realistic task to construct a procedure assigning a sentence a set of representations indicating possible repartitions of topic and focus? It may be considered as a necessary precondition of this task first to find out how topic and focus should be adequately described in the underlying structures of sentences. One of the first attempts at such a formulation (Sgall, 1967) was connected with dependency syntax as characterizing the main structuring of the underlying or tectogrammatical representations (TR's), in which the hierarchy of communicative dynamism (CD) could be rendered by the left-to-right ordering of the nodes of the dependency tree. Other trends have been connected with transformational grammar (Jackendoff, 1972) and, more recently, with intensional semantics (Stechow, 1980; Klein and Stechow, 1982). However, the empirical questions have not yet been fully explored; it seems especially that a more thorough study of intonation patterns is necessary; in many approaches the importance of these patterns is either underestimated or viewed too straightforwardly. It is important to notice that (a) and (b) are two different sentences in (1) as well as in (2) or (3), though the semantic difference is much more important in (3) than in (1); with (1) the two sets of propositions to which the two sentences correspond assign the value 'true' to the same subset of possible worlds, which is not the case with (3). (The intonation centre is denoted by italics.) (1) (a) (b) (2) (a) (b) (3) (a) (b)

Mother is coming. Mother is coming. Napoleon ruled on Elba. Napoleon ruled on Elba. Many men read few books. Few books are read by many men.

However, it also should be respected that in many cases a specific intonation pattern is not connected with a single underlying structure; ambiguity is as broadly spread in the domain of intonation (and in that of word order) as in other parts of the system of language. Thus it is not surprising that many specialists still continue to analyze the issues of topic and focus by the methods of 'soft' syntax and semantics without using an explicit framework. This approach is certainly useful in connection with such difficult tasks as is the study of the structure of text (see Section 2), or a comparative description of typologically highly different

72

E. Hajicova, P. Sgall

languages. Kuno's (1972) observations concerning such phenomena as the functions of the Japanese particles wa and ga, contrast, exhaustive listing, etc., in comparison with English, were of great significance for a general linguistic analysis of TFA. More recently, a useful survey of empirical approaches has been presented by Lutz (1981), though the author has included only three European trends (Prague School, Halliday, and Zemb) in her discussion (her bibliography being much wider). Another recent book, Brömser (1982), aims at an analysis of the main empirical questions, but in some points it seems that the author unnecessarily dwells on details of individual formulations, which mostly are nebulous; instead of paying so much attention to such differences as that between 'given', 'spoken about', or other verbal characterizations of 'topic', 'theme', etc., it might be useful to try to penetrate to questions concerning on the one hand the theoretical issues of semantics and pragmatics, and, on the other, the main empirical point: what is the nature of TFA, what aspects and differences are structured by language? Looking for an answer to this question, we pursued Sgall's (1967) attempt to connect the results of the classical Prague School analyses, based on the use of testable empirical criteria, with the methodological requirements of Chomskyan linguistics, and we came to the following results (cf. Hajicova and Sgall, 1980; for a more detailed account see Sgall, Hajicova and Panevova, 1986), which also may have a certain significance (in connection with the 'given-new strategy') in connection with a 'procedural' approach to the meaning of the sentence. If the sentence is viewed not only statically, but is understood as an instruction given by the speaker to the hearer (i.e. as an elementary structure adapted to functioning in communication), then not only the speaker's intention "to tell someone something" by certain means (cf. Searle, 1970; 42 ff.) should be taken into account, but also the fact that the speaker mostly uses such means that help the hearer to find the interpretation rather easily, avoiding much of the effort that would be necessary if the hearer had to find the meaning and reference of the expressions contained in the sentence in the vast domain of her/his memory without any aids. The sentence, as a systemic form of an elementary communicative linguistic act, is organized in such a way as to minimize this effort on the part of the hearer. Usually a sentence distinguishes certain items of the information stored in the accessible parts of the hearer's memory from the modifications concerning these points that the hearer should carry out according to the intention of the speaker.1 In uttering a declarative sentence, the speaker specifies the items of information s/he shares with the hearer and considers to be easily accessible to the latter at the given time point of the discourse (topic), and s/he specifies, further, 1

This was stated by one of the founders of general linguistics, Gabelentz (1868: 378), who distinguished between the object to which the hearer's attention is oriented by the speaker and what the speaker attempts to make the hearer think about that object.

Topic and Focus of a Sentence and the Patterning of a Text

73

what properties should be assigned to them by the hearer, in what relationships with what other items they should be introduced, or what other modifications they should undergo (focus). In sentences with such a simple structure as (1 a) the two main articulations of the sentence (the valency-based syntactic relations and the topic-focus dichotomy) coincide with each other; the topic is identical here with the subject, and the focus with the predicate, i.e. the terms subject and predicate may be used here both in their grammatical as well as in their philosophical (Aristotelian) sense. In other cases this is not so simple; either the relationship of the two elements is inverted, as in (1 b), or the sentence includes more than two elements, cf. (2), (3), or with deeper embeddings (4): (4) Jim came to the house that he wanted to buy. In (4) there are unstressed pronominal elements within the focus, which denote items readily accessible in the hearer's memory (coreferential with nouns occurring in the preceding part of the sentence). This points to the necessity to distinguish between the dichotomy of the topic and the focus of the sentence and the contextually bound (CB) and non-bound (NB) character of the individual nodes (occurrences of lexical units). In a theoretical description of language several layers of phenomena have to be accounted for jointly: (i) the individual lexical items occurring in the sentence as CB ('given', chosen by the speaker among the items s/he assumes to be easily accessible in the hearer's memory, i.e. salient, activated over an upper threshold in the stock of shared knowledge, see Hajicova and Vrbova, 1982) or NB ('new', or at least not 'recoverable' in the sense of Halliday, 1967; cf. also Chafe's, 1976, term 'identifiable'); (ii) the division of a sentence (more exactly, of one of its meanings) into its topic and its focus; (iii) the hierarchy of CD ("deep word order"). Attempting to describe these items in a uniform manner, we assume that a meaning of a sentence (its tectogrammatical representation, TR) can have the form of a dependency tree (with complex symbols occurring as labels of the nodes, the main verb constituting the root of the tree, and the left-to-right order of the nodes representing the hierarchy of CD),2 and we work with two rather strong hypotheses: 2

It is connected with certain advantages if CD is defined as a partial ordering, where the sister nodes are always ordered with respect to each other and to their common mother, but e.g. the relationship between meet and / in Fig. 1, or between meet and Latin in Fig. 2 remains undefined. It should also be noted that the relation of coordination (conjunction) requires a more complex framework than that of dependency trees, see Plätek, Sgall and Sgall (1984).

74

E. Hajiaovi, P. Sgall

(a) The boundary between topic and focus is always placed so that there is such an item A in the TR that every item of the TR that is less (more) dynamic than A belongs to the topic (focus). In the primary case the verb meets the condition on A and it is itself included in the focus. There are at least two secondary cases, in the first of which the verb meets the condition on A and is included in the topic, while in the latter the verb together with all its modifications (its daughter nodes) belongs to the topic, the focus consisting only in one or more nodes subordinated to the most dynamic of these modifications (see point (c) in Sect. 1.6 below, and the example in Fig. 2). (b) The hierarchy of CD within focus is determined by an ordering of the kinds of modification (types of dependency relation: deep cases and adverbials), which is given by the grammar; we call this relation 'systemic ordering' (SO). In individual sentences, this scale is "changed" into their hierarchy of CD by the contextually bound elements being removed "to the left"; the hierarchy of CD of CB elements is not determined by the grammar, so that permutations of parts of the topic are possible. We assume, furthermore, that every sentence has a focus (since otherwise it would convey no information relevant for communication, it would lack any illocutionary force), but that there are sentences without any topic. 1.2. It appears as useful to start the discussion of TFA by remarks concerning the hierarchy of CD, which partly corresponds to what Chomsky (1971) calls 'the range of permissible focus'. Previous investigations have led us to the conclusion that the description of CD can be based to a certain degree on more elementary notions: (i) a word is less dynamic than its governing word if the dependent word is CB; (ii) the degrees of CD of the NB sister nodes are determined by SO (cf. hypothesis (b) formulated above); (iii) with the CB sister nodes the degrees of CD are primarily distributed in such a way that a word corresponding to a more foregrounded (salient, activated) item in the stock of shared knowledge is less dynamic. A remark concerning point (iii) should be added: With our approach it is the topic proper, i.e. the least dynamic element of the upper bundle of the TR, that is most easily connected with a certain kind of contrast; since (5) can occur as contrasting, for example, with (6) rather than with (7) or with (8), we consider that the subject in (5) is the least dynamic element. The temporal (or local) setting and the unstressed pronoun also belong to the topic there (they are CB), but we consider them to be more dynamic than the subject, though for example from the viewpoint of the 'paradigmatic CD' of Buy (1981) they would be characterized differently.

Topic and Focus of a Sentence and the Patterning of a Text

(5) (6) (7) (8)

The The The The

75

tall boy visited me at the Faculty last week. blond girl visited me at home. tall boy visited you at home. tall boy visited me at home this morning.

This understanding of CD within topic seems to be in accordance with the fact that the indexical elements (including me, you, last week, this morning) are always connected with the situation of the discourse and thus can be used as CB (i.e. primarily in the topic) since they are determined by the situation, while the NP's not containing indexical elements (or proper names) generally are CB only if they were mentioned in the preceding co-text. If an item that was mentioned is assumed regularly to be more activated than an item determined by the situation of the discourse, then such an NP as the tall boy is more activated, i.e. also less dynamic than me, etc., in the given time point of the discourse. 1.3. The notion of contextual boundness should not be understood as immediately connected with the not altogether clear difference between 'given' ('known') and 'new' information. The distinction between CB and NB elements of a TR has its specific position in the patterning of the system of language. A CB element need not be assumed by the speaker to be actually known to the hearer, cf. (9);3 it may even be considered by the speaker not to exist in the actual world (not taking place, etc.), cf. (10). (9) Who is Peter? — Peter is Ann's brother (or: Peter, I don't know.} (10) Their arrival is very improbable now. Also in such a sentence as "Few books are read by many men" the set of few books need not be directly known to the hearer; it is enough if, for example, the reading of books (or just the level of cultural background) was mentioned in the preceding co-text. In other cases, for instance, a CB NP may refer to a part, an element, an aspect, a related counterpart of an already activated object or set. If the notions of topic and focus (as parts of a TR) are characterized on the basis of contextual boundness, then it is possible to avoid such inadequate assumptions as that regarding topic or focus as a single (deep or surface) constituent;4 our standpoint is corroborated by such examples as the follow3

4

This example was presented by Jacobs (1977), who was convinced that the opposition between 'known' and 'new' information played a crucial role in our framework; however, an item may be activated enough in the stock of shared knowledge to be referred to by a CB expression, even if the participants of the discourse are able neither to identify the referent of the given expression, nor to specify more than a feature loosely characterizing it. Also some disadvantages of Firbas' approach (defining 'theme' and 'rheme' on the basis of CD) may be overcome in this way; for example, the existence of topicless sentences can be

76

E. Haji£ova, P. Sgall

ing, in which the time adverbial ("for how long") as well as the adverbial of direction belong to the focus, the Actor with the verb constituting the topic: ("What trip did you make?") — "I went for the weekend to mother" (with a possible continuation "... and not for the whole week to Cannes"}. Such a 'multiple wh-question' as "What did John buy where" (Kuno and Robinson, 1972: 474) determines the Objective and the Locative as the focus of the answer "He bought some books at Heffers".5 In other examples the focus consists in the verb and one of its participants, the other participant(s) belonging to the topic: "He is writing a new book" may occur as an answer to "What is John doing?"; "Jane ate it" may be uttered, for example, as an answer to "What happened to the cake?"; "She lost it on a trip" can answer the question "Does Helen still like her new necklace?". Here the subject as well as the object clearly belong to the topic, so that there is no hope that the sentence could be analyzed into topic and focus as two constituents of any level whatsoever. Dependency syntax seems to be better suited to account for the fact that neither topic nor focus necessarily is 'simple' from whatever point of view. It is important to distinguish between 'CB' and 'presupposed', see Hajicova (1974) and Dahl (1974), since also NB items may be connected with presuppositions. On the other hand, the coinciding results yielded by the question test and by the tests using negation (see Sect. 1.5) corroborate the view that certain persons are still foregrounded enough in the stock of knowledge of a large part of mankind to be able to occur as CB (or on-stage in the sense of Dahl, 1976) without a co-textual antecedent in such sentences as Dahl's (1975) "Hitler was Λ Jew" or Atlas' (1975) "Kepler died in misery". Our approach to contextual boundness is connected with the conviction that the difference between co-text and situation is not directly relevant in this respect (cf. esp. Enkvist, 1974: 30). We assume that the activation of an clement in the stock of shared knowledge beyond a threshold, which allows for it to be referred to by a CB expression, can be conditioned either by the preceding co-text, or by the situation of the discourse. 1.4. As we have already noted in 1.2 (ii), the degrees of CD of NB sister nodes, esp. within the focus, are basically distributed in accordance with a linear ordering of the types of dependency, called the systemic ordering (SO). accounted for, while in the other case even a sentence consisting only of an impersonal verb (e.g. Lat. "pluit" — "it rains", or Czech "nelze pokralovat" — "It is not possible to continue"} would have a topic; cf. Wollmann's (1981: esp. 155) treatment of similar sentences, which — though subscribing to Firbas' views — does not fully respect the definition based on CD: he characterizes, for example "ale nelze fici" ("but it is impossible to tell") in his (4) as topicless. 5 Dahl's (1975: 345) claim that normally the verb is "in the presupposed part of the sentence" is not in accordance with the primary function of the verb, namely to predicate something about the Actor, i.e. to belong to the focus; neither "Mary didn't insult John", nor "John wasn't insulted by Mary" entail an insult on their primary readings.

Topic and Focus of a Sentence and the Patterning of a Text

77

In other words, the kinds of participants and free modifications in the valency frames are ordered, and this ordering, given by the structure of the language, is identical with the hierarchy of CD of individual sentences in the primary case. In a secondary case a CB element A precedes another (CB or NB) element B under CD (i.e. is less dynamic), though A follows after B under SO. The hierarchy of CD in a particular TR differs then from the order of its elements under SO. A criterion serving to find out the relationship of two types of modifications under SO may be formulated as follows: if a governing word can be expanded by two different modifications, A and B, occurring in this order in the surface (with normal intonation), neither of them being CB, whereas the surface word order B A (or placement of the intonation centre on A if preceding B} is possible only if B is CB, then A precedes B under SO. This can be checked by means of such examples as (11) to (13), in which the position of the Objective before the Instrument (11), Origin (12) and Locative (13) in the scale of CD of the (a) examples (corresponding, in these sentences with normal intonation, to the surface word order) is not conditioned by contextual boundness. No higher frequency is concerned here, but the fact that the Objective can be NB, since each of the (a) examples can answer such a question as "What did he do?"; these examples are ambiguous in that the Objective is NB in some of their readings, while it is CB in others (a similar ambiguity concerns also the verb). On the other hand, the (b) examples, with marked intonation (or changed word order), may not be used as answers to the quoted question; they correspond rather to a question in which their final NP is contained (e.g. "What does he use these hoes (logs) for?". This shows that the lower degree of CD of these NP's in (b) — expressed by marked intonation or by inversion — is conditioned by the fact that they are CB.6 (11) (a) (b) (12) (a) (b) (13) (a) (b)

He He He He He He

dug a hole with a hoe. dug a hole with a hoe. made a canoe out of a log. made a canoe out of a log (or He made a log into a canoe.} dug a hole in the garden. dug a hole in the garden (or In the garden he dug a hole.}

An examination of Czech in comparison with English and several other languages has led to the conclusion that the SO of some of the main participants is identical for many languages, having the form: Actor — Addressee — Objective. As for Instrument, Origin, Locative, it seems that English differs from Czech in that these three participants follow Objective 6

As so often in natural language, also here it is necessary to acknowledge the presence of different peripheral phenomena, such as inverted verb forms (e.g. make into), which make a secondary order of the sister nodes possible even with "fixed" word order.

78

E. Hajicovä, P. Sgall

in English, though they precede it in Czech. It need not be surprising that languages differ in such semantically relevant details of their grammatical structures as are those concerning SO — similarly as they appear to differ in the semantics of verbal aspects, of the articles, of dual number, etc. The difference between Czech (or Russian, etc.) and English may be connected with the typological properties of the two languages, and with the fact that those participants that are rendered in English by prepositional phrases (or by adverbs) mostly follow the primarily prepositionless participants in surface, which might have influenced the SO in English. It can be checked that a change of the order of the relevant modifications in each of the examples, if not accompanied by a transfer of the intonation centre from the end of the sentence, leads to the expected effect, namely that the modification transported to the left is then understood as being CB.7 It should be also noted that the temporal and local adverbials, which in English stand at the beginning or at the end of the sentence, often function as settings (if CB). The final position is unambiguous (if it is preceded by the intonation centre); such sentences as (14a) or (15 a) cannot immediately answer a question after which during summer or in a hotel is not activated. On the other hand, the initial position is ambiguous with a temporal adverbial, though a Locative even here occurs only as a setting: (14b) can be also used as topicless, while (15b) probably cannot, the Locative perhaps following after Actor and Manner in SO, so that only in (15c) it can belong to the focus (cf. the examples discussed by Seuren, 1969): (14) (a) (b) (15) (a) (b) (c)

Big fires occurred in several towns during summer, During summer big fires occurred in several towns. Some people can't sleep quietly in a hotel. In a hotel some people can't sleep quietly. Some people can't sleep quietly in a hotel.

Such adverbials often occupy the position of a setting, but this use does not represent their primary positions: as setting these adverbials are CB and thus they display a reduced degree of CD, while their primary positions, corresponding to their places under SO, concern the occurrences that are NB. This view is corroborated by Hakulinen (1976; as for her English wording, see p. 144), who works with a rule of adverbial fronting for settings. Another point concerns the role of topic and 'psychological' or 'logical' subject (i.e. the item that is assigned — or predicated — some quality by the sentence, that is spoken about, assigned some characteristics, etc.). As we have seen, topic need not consist in a single constituent of the sentence. Kuroda (1972: 159) quotes this feature of the topic as a reason why topic 7

The rules of surface word order exclude such an inverted order in some cases; in other cases surface ordering rules are partly conditioned by the shape of the participant in question (this concerns e.g. often, certainly and other adverbs in English).

Topic and Focus of a Sentence and the Patterning of a Text

79

should not be identified with subject ("in a logical sense"), though he does not state why he is convinced that logical subject (of a categorical judgment) must consist in a single syntactic constituent. Dahl (1974: 8 — 13) identifies his topic with Kuroda's "subject in a logical sense", or the thetic judgments with the topicless ones; such an identification is perhaps too straightforward, at least for two reasons: first, it is rather Dahl's 'background part' of the sentence (or our topic) that could be compared with the logical subject, since a verb may constitute the logical subject (or belong to it, as in English cleft sentences), though the verb does not seem to be included in Dahl's concept of topic. Second, the possibility of temporal or local settings in thetic judgments still remains to be investigated. The TR's having no topic correspond to thetic judgments, u Inch lack Kuroda's "subject on the semantic level" or "in a logical sense" and probably can be identified, not only for Japanese, with Kuno's (1972) 'neutral description'. 1.5. Among the operational criteria for determining topic, focus and CD, the question test is often used, and in the decisive cases it leads to results identical with those of the tests based on negation (within the sentence or in a 'natural response'). By means of these tests it is possible to gain a criterion enabling us to identify and distinguish the different phenomena from the domain of TFA with a similar degree of certainty and precision as has been gained, for example, for the identification of the syntactic sentence parts. Another question concerns the identification (disambiguation) of the TFA of an utterance by the hearer; this question was discussed by Hajicova and Sgall (1980: 101 ff.). The question test, as used for the identification of the boundary between topic and focus, and also of the distinctions between the degrees of CD, was characterized by Hatcher (1956), DaneS (1970), and others. This approach is based on the assumption that the contexts in which a sentence may be appropriately used can be characterized by a set of questions that can be immediately (and fully) answered by this sentence.8 The following rules may be understood as the basis of the question test: (a) If the set of all those questions for which the given sentence can serve as an 'immediate' answer (called the set of relevant questions in the sequel) is such that (aa) for some such phrases A and B included in the given sentence, that (the referent of) A is (referred to by a phrase) included in every question from the set of relevant questions in which B By an 'immediate answer' we denote, for example, "Mary has thrown them into the waste basket" as an answer to "What has Mary done with the flowers?" rather than to "What happened with the flowers?", since in the latter case (at least on one reading) the answer includes Mary as CB, though its referent is not referred to in the question (i.e. is activated enough by the situation).

80

E. Hajicova, P. Sgall

(b) (c) (d)

(e)

is included, and also (ab) in such a question (from the set) that does not include B, then in (all TR's of) the given sentence the (source of the) phrase A is less dynamic than the (source of the) phrase B; if A (from the given sentence) occurs in no element of the set of relevant questions, it is the focus proper of the given sentence; if A (from the given sentence) occurs in every element of the set of relevant questions, then it belongs to the topic of the given sentence; if (a) is met by A and B, but either A or B breaks (b) and also (c), then the sentence is ambiguous in that the phrase breaking these two conditions belongs to its topic in some of its TR's and to its focus in some other(s); if there is a pair of phrases A, B, in the given sentence, such that A and B break (aa) and neither A nor B meets (c), then the sentence is ambiguous not only with respect to the position of the boundary between topic and focus — cf. (d) and Chomsky's (1971) 'range of permissible focus' — but also in that A is less dynamic than B in some of its TR's and more dynamic in some other(s); at most in one of these two cases both A and B belong to the focus (this case can be determined on the basis of SO).

According to these rules, for example, the phrase the astronauts belongs to the topic in all TR's of (16), while from the moon belongs to the focus in all of them, and each of the words brought and minerals belong to the topic in some of them, but to the focus in others. (16) The astronauts brought minerals from the moon. This shows that to a certain degree the question test is useful not only for drawing the boundary between topic and focus (or, more exactly, between the CB and NB parts of the nucleus of a TR of a sentence), but also for identifying the degrees of CD: the elements that belong to the topic only in some of the TR's of the sentence are more dynamic than those belonging to the topic in all the TR's, but less dynamic than the (single, as the examination of hundreds of examples from different European languages suggests) element that in all the TR's belongs to the focus (and thus constitutes the focus proper).9 One of the two approaches using a test with negation is based on the properties -of negative sentences, in which (see Hajicovä, 1973) primarily it is the focus (in its relationship to the topic) what is negated, so that, for example, the (a) sentences in (17) and (18) are more properly paraphrased by the (b) than by the (c) sentences. 9

Here, as well as in the usual definition of 'focus proper' as the most dynamic element of the focus, only those TR's are taken into account in which the uppermost subtree (the root and its daughters) includes a NB element.

Topic and Focus of a Sentence and the Patterning of a Text

(17) (a) (b) (c) (18) (a) (b) (c)

81

Many arrows didn't hit the target. About many arrows it is not true that they hit the target. It is not true that many arrows hit the target. Every Sunday one doesn't work. About every Sunday it is not true that one (then) works. It is not true that every Sunday one works.

This kind of the negation test can be used to show that — as for the primary readings — in (17 a) the subject and in (18a) the temporal adverbial belong to the topic; cf. also Allwood (1974) about negating such sentences as (19) (a) Bill came here in a car. (b) Bill came here in a car. The possibilities of this test are rather limited, since for example to show in this way that Bill in (19 a) belongs to the topic would be difficult; such a simple proper name certainly is connected with a referential presupposition, even if included in the focus, so that the effect of 'being under negation' is not conclusive. The other approach to tests using negation is best known from Chomsky (1971), who uses the possible 'natural responses' or continuations of the sentence examined to show which parts of the sentence can — in some readings — belong to its focus. Other variants of the test using a negative continuation have been elaborated rather systematically by Posner (1972a) in his Kommentartest and by Boguslawski (1977). The former of these approaches pays due attention also to the different positions of the intonation centre, and makes use of the different positions of the negative particle in German, the relevance of which for the TFA was discussed with deep insight by Zemb (1968). It thus seems important that in those situations where more than one of the mentioned tests can be applied the tests coincide in their results. This coincidence corroborates the view according to which the bipartition of the sentence into topic and focus is based on the difference between its elements that are CB (in our broader sense of the term) and NB. It is this bipartition that is relevant for the semantic function of linguistic negation in a sentence, as well as for the role of the sentence in the structure of a text — in connection with the interplay of questions and answers in a dialogue, or with the natural continuations in any coherent text. 1.6. The interplay of contextual boundness with the dependency structure of a TR determines which parts of the TR are included in its topic and which of its parts belong to its focus, though the way in which this is determined is not quite simple; a more detailed discussion and a formal definition of topic and focus is contained in Sgall (1979). Here we want to present the characterization of topic and focus in an informal shape, which can be developed in this or that direction, if further empirical research shows that

82

E. Hajicova, P. Sgall

such a development is useful. Informally the focus of a TR may be specified as follows: (a) If the main verb of the TR (the root of the dependency tree) or some of the nodes which directly depend on it are NB, then these nodes belong to the focus of the TR; (b) if a node other than the root belongs to the focus, then also all nodes subordinated to it belong to the focus; (c) if the root and also all its daughter nodes are CB, then it is necessary to specify the rightmost of the daughter nodes of the root and ask whether some of its daughter nodes is (are) NB; if so, then these NB nodes belong to the focus; if not so, we again specify the rightmost of the last set of sister nodes and ask whether some of its daughter nodes is NB, etc. The nodes of the TR that do not belong to its focus constitute its topic. We characterize the focus (or the topic) neither as an assertion (or question), nor as an assertion form (with indefinite pronouns or their counterparts replacing the missing nodes, as in Chomsky's, 1971, 'presupposition'). Such means appear to be helpful in this connection as rough analogies rather than as exact semantic explicata. It seems that more elucidation can be gained by defining topic and focus as parts of the TR and by providing then a semantic interpretation in a framework of intensional logic (cf. Materna and Sgall, 1981; Kosik and Sgall, 1982). Thus for example, the sentence (20) corresponds to the TR characterized in Fig. 1 as to its primary reading, the topic here consisting in the nodes corresponding to John (topic proper) and yesterday (temporal setting), the other parts of the TR belonging to its focus. (20) Yesterday John met my brother. meei-Pret-lnd'ic-...

brother-Sing-Oef-Ob) Joan-Act

/-Appurt Figure 1: A simplified TR of "Yesterday John met my brother", where the edges of the dependency tree going down to the left mark their incident dependent nodes as contextually bound and the left-to-right ordering of sister nodes corresponds to their degrees of CD; the grammatical parts of the complex symbols correspond to the values of morphemic categories (Tense, Modality, ..., Number, ...) and to the types of dependency (deep cases and adverbial modifications).

Topic and Focus of a Sentence and the Patterning of a Text

83

Point (c) of the definition is illustrated by Fig. 2, and Fig. 3 shows another reading of the same sentence, which (similarly as many other sentences) is ambiguous as for its TFA. meetb-PKt-Indic-...

teaeher-Sing-Oef-Ob] John-Act Latin-Ob] Figure 2: A simplified TR of "John met the teacher of Latin" (appropriate for such contexts as when answering the question "Which teacher did John meet?", or when followed by "..., not of biology"); the superscript b marks the main verb as contextually bound.

Jobn-hct Latin-Ob] Figure 3: Another TR of "John met the teacher of Latin" (appropriate e.g. as an answer to "Who did John meet?", or followed by "..., not the eye-doctor").

2.1. When the connections between the TFA of the sentence and the structuring of a text are considered, it is often argued that a thematic organization essentially similar to that of the sentence underlies the structuring of any text. However, it can be shown that the structure of a text obeys different principles than that of a sentence. It is important in this connection to recall that it is not precise to characterize a text or a discourse as a sequence (or string) of sentences. The constituent parts of a text correspond to sentence tokens (utterances) rather than to sentence types; the reference assignment, i.e. the specification of the objects referred to by NP's (and other referring expressions), is crucial tor the structuring of a text, especially for its coherence, and different tokens (occurrences) of a single sentence may differ in the reference assignment, so

84

E. Hajitova, P. Sgall

that it is an utterance rather than a sentence that constitutes a part of the text. Utterances may be of other than sentential (or even linguistic) form (e.g. the nodding answering a question is an appropriate part of a dialogue, though it is not a sentence). This point is connected with the fact that discourse or text belongs to the domain of communication (language use, a kind of human activity) rather than to the system of language. It is not surprising, under the given viewpoint, that the questions of the internal structuring of a text cannot be answered by general statements concerning the structure of every possible ("well-formed") text (in a given language). Even those authors who speak of "text grammars" have found that a "theory of coherence" may be preferable (Petofi, 1978: 366), and that different kinds of text structuring can be described as being characteristic for texts of different types (these types of texts themselves can be supposed to correspond in a certain way to the forms of human cognition). Such units of text structuring as paragraphs and chapters are limited to certain kinds of (written) texts, while e.g. a dialogue can be analyzed into other constituents. One of the most systematic attempts to formulate generally valid rules of the structuring of a text is known from Heger (1977); he distinguishes "signemes" of different ranks and presents criteria for identifying them. Giilich and Raible (1977), using van Dijk's theory of the macrostructure of a text, achieve similar results and conclude that Heger's approach is connected with the advantage of specifying the individual ranks of units also in a qualitative way, but that it meets difficulties concerning the possibilities of a (recursive) embedding of dialogues into a monologue and vice versa. The differences of the structure of different kinds or types of texts, as studied, for example, by Hausenblas (1964), Posner (1972b) and others, corroborate the view that the regularities connected with the patterning of texts are of a different character than the regularities dealt with in the description of languages. Conversation, one of the basic kinds of discourse, is well known to have "rules of its own right", quite different from rules of grammar (see Mey, 1981). Similarly, there are different kinds of patterning connected with instruction, narration, technical texts, not to speak of poetic genres. Also the beginning and the end of a text are determined in the general case by criteria which cannot be regarded as intrinsic to the structure of the text. Only in some specific cases (esp. in poetic texts) such internal criteria exist, but they are certainly (in most cases) not obligatory to such a degree as rules of grammar are. The difference between the structure of a sentence and the patterning of a text is also apparent in the domain of coreference. Within a sentence there are several points in which coreference is determined grammatically. This holds for the reflexive pronouns (though esp. with the reflexive possessives the situation is not always so clear-cut as might be assumed, cf. Buy, 1981,

Topic and Focus of a Sentence and the Patterning of a Text

85

and the writings quoted there). Also in such cases as "They allowed him to do it" the coreference of the subject of the embedded clause (infinitive) with the Objective of the governing clause is conditioned by grammar. On the other hand, if crossing the boundaries between sentences, coreference perhaps always is indistinct (cf. Halliday and Hasan, 1976: 31 In), and pragmatic inferencing is involved in the resolution of anaphoric expressions (see Reinhart, ms., § 4.2). Thus in "Henry goes to a theatre with his wife every week"; "John does so only rarely" it is not clear whether John does so with his wife or not, perhaps he may be a bachelor. This concerns also most cases of optional deletion, be they of an intersentential character, or in coordinate clauses, cf. "John met Jim. He didn't say hello". (The subject is indistinct, and so is the deleted Addressee.) Even though Buy (1981: 179—201) examines the questions of coreference in a systematic way, in some cases he also overlooks the possibility of interpreting his examples with /'/ as having different antecedents: e.g. in his (76) "Charles expects that he will get a raise, but Peter does not believe it, though even Ivan counts upon it" Ivan may be understood as counting upon Charles' wages being raised (not only upon Ivan's). It seems that even within a sentence without coordination indistinctness is typical with such pronouns that can refer to elements in a preceding sentence, as: "After John met Bill, he started to learn better" (cf. "John met Bill. After this he started ...") or "Bill behaved in such a nasty way towards John that I had to tell him". Therefore it seems preferable not to use as the underlying structure a language "in which indexed variables are used and where scopal relations are overtly indicated in structural configurations", as Bach and Partee (1980) state about Extended Standard Theory and Extended Montague Grammar: We understand the cases of unclear coreference as indistinct rather than ambiguous. Keenan's (1978) criterion is not decisive here: the speaker knows whether he meant John or Bill (or another person), but this does not say for sure that ambiguity is present; the criterion is decisive for those cases in which the speaker need not know; then it is certain that ambiguity is not present. Also the VP anaphora, discussed by Partee and Bach (1981), appears to be indistinct in this respect; only a larger co-text can decide whether Mary left or stayed in such a segment of text as "John left, though Bill stayed. Mary didn't"; if we are not mistaken, such a continuation as "She thus avoided being accompanied home by X" is possible either with John or Bill in the position of X. It should be added that it is especially the deletion of the verb that is conditioned by verbal co-text. With such pronominal elements as did the same it appears that the well known conditioning by 'consituation' is present, similarly as in so many other cases concerning intersentential context. This strongly corroborates the view that text linguistics should be developed within the theory of communication rather than in parallel to the linguistics of the sentence (or even instead of the latter).

86

E. Hajicovä, P. Sgall

2.2. There are then different possible approaches to the relationship between the relatively well known TFA of a sentence and the corresponding aspects of the patterning of a text. If one uses the term 'topic' (or 'theme', as the case may be) as a technical term from the sphere of TFA, rather than in its colloquial meaning, then DaneS' (1970; 1974) 'thematic progressions' appear to constitute the aspect of text structuring that is most closely related to TFA: the most important types of thematic progressions are those characterized in Fig. 4 to 6, the first of which consists in the focus (rheme, comment) of a Ti

R,

Figure 4: Thematic progression of type 1.

T2

Figure 5: Thematic progression of type 2.

R2

Ti

Ri

| Ta

R4

H

V, \

Figure 6: Thematic progression of type 3.

X

\

\

Ti - Ri

\ \ T2 - R2

given utterance Ui reappearing as the topic of the immediately following utterance U2, as e.g. in "Last week I visited Lund. It is a nice university town ..."; in the second type the topic of Ui is identical10 to that of U2, as for example in "Charlie came to his aunt. He wanted first of all to show his politeness"; the two utterances of the third type derive their topics (themes) from a hypertheme, such as the appartment in "Three of the four rooms are rather large. The windows lead into a garden." However, often the progres10

Not a full identity is required here; such associative links as that between a whole and its parts, or those concerning hyponyms, etc., are relevant.

Topic and Focus of a Sentence and the Patterning of a Text

87

sions contain different combinations of the elements of topic and focus (cf. e.g. Juganov, 1979), so that only relatively rarely the topic of an utterance as a whole is coreferential either with the topic or with the focus of the preceding utterance. Since topic and focus can comprise more than one element, it is necessary to reckon with the whole interplay of CB and NB items, as well as of items carrying lower and higher degrees of CD. On the other hand, if 'topic' is understood as 'topic of the discourse' or in a similar way (which corresponds to the everyday usage rather than to the technical meaning the term 'topic' has acquired in connection with TFA), then it seems typical that it is not maintained that every text or discourse has a single topic.11 Several authors who have undertaken an empirical investigation that perhaps was intended to lead to a general characterization of "the topic of the text" have concluded that smaller segments of a text regularly have topics of their own. This has been stated by Marciszewski (1976: 21 ff.); also van Dijk (1977: esp. 132ff.) prefers to speak of "the topic for this passage", "topic of (a part of) a discourse", etc., instead of presenting arguments in favour of the view that every text has a single topic. Most of the more recent approaches to "discourse topic" are of a similar character, see e.g. Bayer (1979) and the survey given by Reinhart (1981). A thorough characterization of different approaches concerning the dialogue was presented by Carlson (1982), who himself analyzes the structure of the dialogue from the viewpoint of Hintikka's game-theoretical semantics. 3.1. Since the relationship between TFA and the structuring of a text thus does not appear to be a straightforward connection, it may be of interest to consider some of the characteristic relations between question and answer, which are important for the dialogue, one of the main kinds of discourse. We can state that for a declarative sentence A to be a direct answer to a whquestion conveyed by an interrogative sentence Q it is necessary that the topic-focus articulation of A corresponds to that of Q in a certain way, which has to be specified. If we restrict our investigation to interrogative sentences having normal intonation,12 we may assume that Q has a surface structure which can be characterized, with many simplifications, by (21): (21) (Verb (Αι) ... (At) 11

12

(A* + i) ... (A,))

A terminological distinction certainly would be useful here, but both the terms 'topic' and 'theme' already have been widely used in connection with the TFA of a sentence. The main task concerning the corresponding questions of the structure of a text seems to consist in identifying a (set of) concept(s) characterized clearly enough by the properties of a relatively important class of texts. Such a question as "Who came late?" can be answered by "Nobody" in a smooth dialogue, while the corresponding question with marked intonation: "Who came late?", cannot, see Hajicova (1976); also von Stechow (1980: 87) considers "No one likes Hans-Robert" to be an appropriate answer to "Who likes Hans-Robert?"

88

E. Hajicovä, P. Sgall

where / < k < », the elements A\ to Ak are CB and belong to the topic, the others are NB and included in the focus; one of the latter elements contains the interrogative phrase. As for the answer A, we assume that: (a) the item referred to by the focus proper of Q (unless it is identical with the interrogative phrase) is referred to by the topic proper of A;13 (b) the element of A corresponding to the interrogative phrase, i.e. the element carrying new information in A, is the focus of A; (c) the elements corresponding to other elements of Q belong to the topic in A;" The validity of (a) to (c) can be checked by means of such examples as (22): (22) When did the chemical industry recently cause a tragedy? — A tragedy was recently caused by the chemical industry in October. It may be assumed that in the preferred reading of the interrogative sentence in (22) a tragedy is its focus proper, while when and cause belong to the focus, too, and the subject constitutes the topic proper, recently (temporal setting) representing the more dynamic part of the topic. According to (a), then, a tragedy is the topic proper of the answer A, which is in accordance with its leftmost position as the subject of A. According to (b), in October is the focus proper of A, which is rendered by the rightmost position of this phrase. According to (c) the remaining three elements belong to the topic of A. It is known that the topic, or some parts of it, may be either deleted or substituted by pronominal elements, as in (23); however, it is an interesting question to examine how these possibilities are restricted, i.e. how the implausibility of (23a) can be predicted, or how it can be explained that (23f) does not — if we are not mistaken — contain a deleted Actor (or, more precisely, that a general Actor is present here), so that (23f) is not an immediate answer to the question from (22).

13

14

In this point we differ from Kuno (1982), who assumes that it is "a normal state of affairs" that "the focus of the question is also the focus of the answer"; it seems more adequate to consider the information not present in the question to constitute the focus of the answer in the typical cases: in the answer to "When (Why) did he come here?" the temporal (causal) adverbial rather than the verb constitutes the focus. Also Allwood (1974: 7) finds that the more a non-interrogative word in a question is "put into focus", the more its correlate in the answer is "pushed into the background" (i.e. is less dynamic, in our terminology). A certain correspondence of the syntactic positions of the elements present both in the question and in the answer is necessary, see Carlson (1982: 302 f.) about the subject of "An idiot would trust an idiot" as not corresponding to the object of "Who would trust an idiot?". However, Carlson's 'substitution' should be specified in such a way as to characterize, for example, him in "I gave him a pen" or "I took him home" as corresponding to the Benefactive in "What did you do for the boy?".

Topic and Focus of a Sentence and the Patterning of a Text

(23) (a) (b) (c) (d) (e) (f)

89

(It was) in October. It caused one in October. It did in October. One was caused by the chemical industry in October. It did it in October. A tragedy was caused in October.

According to the general assumptions of our approach, the degrees of salience (activation) of individual sentence parts change in the following respects: (a) the more an element was dynamic in Q, the more activated it is in the time point of A\ (b) if other factors influencing the scale of CD do not intervene, a more activated item should be referred to by a less dynamic element of A. It should be noted that when Q contains a series of quantifiers or a relevant 'holistic' aspect, such as in (24) or (26), the change of activation we just spoke about would cause the meaning of A not to match completely with that of Q. As the following examples illustrate, such an answer would not even be appropriate in what regards its truth conditions. (24) (25) (26) (27)

Since when does John talk to few girls about many problems? John talks about many problems to few girls since half a year ago. Since when is English spoken in New Zealand? English is spoken in New Zealand since three or four centuries ago.

It is (28) rather than (25), and (29) rather than (27), which may serve as an answer to (24) and to (26), since the scale of CD determines the scope of quantifiers, or the 'holistic' aspect, and it is thus semantically relevant. One should also bear in mind that a secondary intonation pattern (e.g. with the intonation centre on English in (26)) changes the TFA and the contextual applicability of the sentences. (28) John talks to few girls about many problems since half a year ago. (29) In New Zealand English is spoken since three or four centuries ago. 3.2. The changes of salience of the elements of the stock of shared knowledge during a discourse were studied by Hajicova and Vrbova (1982), who analyzed segments of texts and came to the following conclusions (which should be further checked and enriched by means of analyzing a large corpus of texts of different kinds): Let xa denote an expression χ referring to an object a: a" denotes that this object is salient to the degree n in the stock of shared knowledge; since the maximum of salience can more easily than other degrees be imagined to be fixed, we denote it by « = 0, reversing the direction of "growth" of the degrees; to the left (right) of the arrow we indicate the state immediately preceding (following) the utterance of a sentence S in which χ occurs; P(xa) denotes that χ is expressed either by a weak (unstressed) anaphoric pronoun

90

E. Hajicova, P. Sgall

or is deleted in S (though present in the underlying structure of S}; NPd(xa) denotes that χ is a definite NP (including such expressions as one of the ...). It appears that the following rules concerning the degrees of salience can then be formulated, where m, n ^ 0: (1) (2) (3) (4)

If P(xa), then a" -> a". If NP (xa) is in the focus of S, then tf -> a®. If NPd(xa) is in the topic of S, then a" -> a1. If a" -> am, then bm+2 obtains for every object b that is not itself referred to in (the underlying structure of) S, but is immediately associated with an item contained there (e. g. room and ceiling, school and pupils). (5) If as for xa, concerning xa is the leftmost expression in S, then a" —* a1. (6) If xa is included neither in S, nor among the associated objects (see Rule (4) above), then 0" The difference between « = 0 and « = 1 is too small to make the reference assignment univocal if a weak pronoun is used in the utterance following S. It can be assumed that a difference larger than 1 is sufficient in this respect (though our use of numerical values is meant only as a first approximation of how to render the difference between the degrees of salience, which probably are not connected with such a simple scale). Thus, for example, after such two sentences as "The school garden was full of children. Outside parents were waiting", an utterance of "Most of them seemed to be quite satisfied" can be used with them referring clearly to the parents. On the other hand, the same utterance following after "Many parents used to send their children to our courses. The children liked it here" clearly refers to the children. Several issues are treated only in a very preliminary way by the rules in their present form; the following remarks can point out some of the aspects in which a more thorough formulation should be found: (i) Rule (3) is connected with the fact that χ retains its salience in the subsequent portion of the discourse to a higher degree than if it were never included in the topic of an utterance. (ii) Rule (5) should be so formulated as to point out that it activates an item that was not among the most activated ones; in the prototypical cases « is relatively high here. S contains a topic that was not expected, often recalling an object which was not mentioned in one of the last sentences. (iii) Rule (4) or a similar rule should reflect also a rather common situation when a mentioning of a particular object (or objects) brings into foreground only a fraction of a set of objects that already have been activated. (iv) Another aspect that should be reflected consists in the difference between (a) a situation in which a single element of a set of objects that can be denominated by a certain expression is activated to a sufficiently higher degree than other objects belonging to this set, and (b) a situation in which this is not the case, so that the simple denomination would not

Topic and Focus of a Sentence and the Patterning of a Text

91

be connected with univocal reference, and the speaker needs to use a more specific denomination: thus if only a single male human being was mentioned in the preceding two utterances, be is sufficient; if two such beings do not sufficiently differ in the corresponding degrees of activation, then the teacher might be univocal; if both of them share this qualification, the teacher of chemistry or a similar denomination will be used in the next utterance. (v) Certain other scales or hierarchies should be taken into account, e.g. that concerning the more or less immediate relationships of association, cf. Rule (4), or that concerning a more or less prominent position of the utterances described (sentences metatextually opening a narration or a piece of it, or headings, etc., are more prominent than others in that the objects introduced in them retain their activation to a higher degree than others), etc. These lines of further research should be pursued to gain the possibility to characterize a token of a sentence (with its sense) as a procedural instruction aiming at certain changes in the hearer's image of the world, such as:15 (A) Identify the referents of the contextually bound expressions in your memory! (B) Modify your image of the objects identified in (A) (and of their relations to other objects) according to what the focus of S asserts of them! (C) Change the degrees of salience according to the rules just stated! (D) Notice that this information comes from me under the conditions obtaining here and now! (E) Use the piece of knowledge gained in (D) to find out (by inferencing, applying perhaps also other pieces of the knowledge you have, as well as the conversational principles and maxims) what aims I pursue by telling you S (indirect illocution, perlocution)! (F) Use the piece of knowledge gained in (D) and in (E) to appreciate the probability and/or degree of my sincerity, of the appropriateness and the more or less urgent character of my aims from the viewpoints of your interest, and decide accordingly what to do next (continue to listen, answer me, pass the salt, close the window, or, if you like, put down the earphone ...). 4. We may conclude that an application of the results of the investigations of TFA in the study of text structure has many different aspects. It is not surprising that text as an instance of human action can be structured in many ways corresponding to different chains and hierarchies of intentions and aims. The dichotomy of topic and focus, typical for the structure of the sentence, 15

The changes are carried out as soon as possible, some of them do not wait until the hearer decodes the utterance as a whole.

92

E. Hajicovi, P. Sgall

has no direct counterpart in the patterning of a text, though this dichotomy is of crucial importance in the questions concerning the succession of individual utterances (sentence tokens) in the text. During discourse the "topics" spoken about may change at different rates and in hierarchies of various shapes. The 'thematic progressions' as characterized by Danes, the 'topics of text passages' analyzed by van Dijk, Carlson and others (from various points of view), as well as the interplay of questions and answers all constitute examples of how the changes of the hierarchy of salience, underlying the TFA of a sentence, are connected with the structuring of a text. For a procedural approach to the structure of a text this hierarchy of salience has a similar significance as TFA has for such an approach to the meaning of a sentence. References Adamec, Premysl 1966 Porjadok slov v sovremennom russkom ja^yke. Prague: Rozpravy CSAV. Allwood, Jens A. 1974 Intensity, Pitch, Duration and Focus (Logical Grammar Reports 11). Göteborg. Ammann, Hermann 1928 Die menschliche Rede: sprachphilosophische Untersuchungen. Teil 2: Lebensformen und Lebensfunktionen der Rede: Das Wesen der Sat^form. Sat% und Urteil. Lahr i. B.: Schauenburg. Atlas, Jay D. 1975 "Frege's Polymorphous Concept of Presupposition and Its Role in a Theory of Meaning", Semantikos 1, 19 — 44. Bach, Emmon W. and Partee, Barbara H. 1980 "Anaphora and Semantic Structures", in J. Kreiman and A. E. Ojeda (eds.): Papers from the Parasession on Pronouns and Anaphora. Chicago: University of Chicago, 1 —28. Bayer, Josef 1979 "Diskursthemen", presented at the 14th Linguistic Colloquium, Bochum, in E. Weigand and G. Tschauder (eds.): Perspektive: textintern: Akten des 14. Linguistischen Kolloquiums Bochum 1979. Bd. l, 213-224. BeneS, Eduard 1968 "On Two Aspects of Functional Sentence Perspective", Travaux Linguistiques de Prague 3, 267-274. 1973 "Thema-Rhema-Gliederung und Textlinguistik" in H. Sitta and K. Brinker (eds.): Studien %ur Texttheorie und %ur deutschen Grammatik (Sprache der Gegenwart, Schriften des Instituts für deutsche Sprache in Mannheim 30) Düsseldorf: Schwann, 42 — 62. Bily, Milan 1981 Intrasentential Pronominali^ation and Functional Sentence Perspective in C^ech, Russian, and English (Lund Slavonic Monographs 1). Lund: Slaviska inst., Lunds Univ. Boguslawski, Andrzej 1977 Problems of Thematic-Rhematic Structure of Sentences. Warsaw: Panstwove Wyd. Naukowe. Brömser, Bernd 1982 Funktionale Satyperspektive im Englischen (Tübinger Beiträge zur Linguistik). Tübingen: Narr. Carlson, Lauri H. 1976 Dialogue Games: An Approach to Discourse Analysis (Synthese Language Library 17). Dordrecht: Reidel.

Topic and Focus of a Sentence and the Patterning of a Text

93

Chafe, Wallace L. 1976 "Givenness, Contrastiveness, Definiteness, Subjects, Topics and Point of View", in C. N. Li (ed.): Subject and Topic. New York: Academic Press, 27-55. Chomsky, Noam 1971 "Deep Structure, Surface Structure and Semantic Interpretation", in D. D. Steinberg and L. A. Jakobovits (eds.): Semantics: An Interdisciplinary Reader in Philosophy, Linguistics and Psychology. London: Cambridge University Press, 193 — 216. Dahl, Osten 1969 Topic ana Comment: A Study in Russian and General Transformational Grammar (Slavica Goetheburgensia 4). Stockholm: Almqvist & Wiksell. 1974 "Topic-Comment Structure Revisited", in 0. Dahl (ed.): Topic and Comment, Contextual Soundness and Focus (Papers in Textlinguistics 6). Hamburg: Buske, 1—24. 1975 Review of Sgall et al. (1973), Journal of Linguistics 11, 347-354. 1976 How Do Noun Phrases Refer (Logical Grammar Reports 21). Göteborg: Research Project for Logical Grammar, Department of Linguistics, University of Göteborg. Danes, FrantiJek 1970 "Zur linguistischen Analyse der Textstruktur", Folia Linguistica 4, 72 — 78. 1974 "Functional Sentence Perspective and the Organization of the Text", in F. Danes (ed.): Papers on Functional Sentence Perspective. The Hague: Mouton, 106—128. van Dijk, Teun A. 1977 Text and Context: Explorations in the Semantics and Pragmatics of Discourse (Longman Linguistics Library 21). London: Longman. Enkvist, Nils E. 1974 "Style and Types of Context" in N. E. Enkvist (ed.): Reports on Text Linguistics: Four Papers on Text, Style, and Syntax (Meddelanden fran Stifteisens for Abo Akademi Forskningsinstitut 1). Abo: Publications of the Research Institute of the Abo Akademi Foundation, 29 — 75. Firbas, Jan 1957 "Some Thoughts on the Function of Word-Order in Old English and Modern English", Sbornik pracifiles, fakulty brninske university 6, A5, 72—100. 1971 "On the Concept of Communicative Dynamism in the Theory of Functional Sentence Perspective", Brno Studies in English 7, 12—47. 1975 "On the Thematic and the Non-Thematic Section of the Sentence" in H. Ringbom et al. (eds.): Style and Text: Studies Presented to Nils Erik Enkvist. Stockholm: Spragförlaget Skriptor, 317—334. von der Gabelentz, Georg 1868 "Ideen zu einer vergleichenden Syntax — Wort- und Satz-Stellung", Zeitschrift für Völkerpsychologie und Sprachwissenschaft 6, 376—384. 1891 Die Sprachwissenschaft, ihre Aufgaben, Methoden und bisherigen Ergebnisse. Leipzig: Tauchnitz. Gülich, Elisabeth and Raible, Wolfgang 1977 Linguistische Textmodelle: Grundlagen und Möglichkeiten. München: Fink. Hajicovä, Eva 1973 "Negation and Topic vs. Comment", Philologia Pragensia 16, 81—93. 1974 "Meaning, Presupposition and Allegation", Philologia Pragensia 17,18—25, reprinted in W. Klein and A. von Stechow (eds.): Functional Generative Grammar in Prague (Forschungen Linguistik und Kommunikationswissenschaft 2). Kronberg/Taunus: Scriptor, 160 — 172. 1976 "Question and Answer in Linguistics and in Man-Machine Communication", Statistical Methods in Linguistics, 30—46. Hajicovä, Eva and Sgall, Petr 1980 "A Dependency-Based Specification of Topic and Focus", Statistical Methods in Linguistics — Journal of Linguistic Calculus 1/2, 93—140.

94

E. Hajicova, P. Sgall

Hajicova, Eva and Vrbova, Jarka 1982 "On the Role of the Hierarchy of Activation in the Process of Natural Language Understanding" in J. Horecky (ed.): Proceedings of the Ninth International Conference on Computational Linguistics. Amsterdam: North-Holland, 107—113. Hakulinen, Auli 1976 Suomen kielen generatiivista lauseoppia 2, Reports on Text Linguistics (Meddelanden fran Stifteisens for Abo Akademi Forskningsinstitut 7). Diss. Turku. Turku. Halliday, Michael A. K. 1967 "Notes on Transitivity and Theme in English", Journal of Linguistics 3, 37—81, 199_244, 4 (1968), 179-215. Halliday, Michael A. K. and Hasan, Ruqaiya 1976 Cohesion in English (English Language Series 9). London: Longman. Hatcher, Anna G. 1956 "Syntax and the Sentence", Word 12, 234-250. Hausenblas, Karel 1964 "On Characterization and Classification of Discourses", Travaux Linguistiques de Prague 1, 67-84. Heger, Klaus 1977 "Signemränge und Textanalyse" in T. A. van Dijk and J. S. Petöfi (eds.): Grammars and Descriptions. Berlin: de Gruyter, 260—313. Jackendoff, Roy S. 1972 Semantic Interpretation in Generative Grammar (Studies in Linguistics Series 2). Cambridge/Mass.: MIT Press. Jacobs, Joachim 1977 Review of Sgall, P., E. Hajicovi and E. Beneiovi, Topic, Focus, and Generative Semantics, Lingua 43, 77 — 91. Jespersen, Otto 1924 Philosophy of Grammar. London: George Allen & Unwin. Juganov, V. J. 1979 "Tekst i tipy tematiceskich progressij", in Znalenije i smysl relevych obra^pvanij. Kalinin, 157-163. Keenan, Edward L. 1978 "Some Logical Problems in Translation", in F. Guenthner, and M. GuenthnerReutter (eds.): Meaning and Translation: Philosophical and Linguistic Approaches. London: Duckworth, 157-189. Klein, Wolfgang and von Stechow, Arnim 1982 Intonation und Bedeutung von Fokus. Konstanz: Universität. Kosik, A. and Sgall, Petr 1981 "Towards a Semantic Interpretation of Underlying Structure", Theoretical Linguistics 8, 158-171. Kovtunova, Irina I. 1974 "Aktual'noje clenenije i sistema jazyka (na materiale russkogo jazyka)", in F. DaneS (ed.) Papers on Functional Sentence Perspective. The Hague: Mouton, 142—151. Krusel'nickaja, K. G. 1956 "K voprosu o smyslovom clenenii predlozenija", Voprosy ja%yko%nanija 5, 55 — 67. Kuno, Susumu 1982 "Principles of Discourse Deletion", in Preprints of the Plenary Session Papers of the XIHth International Congress of Linguists, Tokyo, 36 — 46. Kuno, Susumu and Robinson, Jane J. 1972 "Multiple Wh-Questions", Linguistic Inquiry 3, 436-487. Kuroda, S.-Y. 1972 "The Categorical and the Thetic Judgment", Foundations of Language 9, 153 — 185.

Topic and Focus of a Sentence and the Patterning of a Text

95

Lutz, Luise 1981 Zum Thema "Thema: Einführung in die Thema-Rhema-Theorie" (Hamburger Arbeiten zur Linguistik and Texttheorie 1). Hamburg: Hamburger Buchagentur. Marciszewski, Witold 1976 "Ot ponjatija temy predlozenija do ponjatija kljucevogo slova", in Naulno-technileskaja informacija 2, 18 — 25. Marty, Anton 1897 "Über die Scheidung von grammatischem, logischem und psychologischem Subject, resp. Prädicat", in Archiv für systematische Philosophie, 3, 174-190, 294-333. Materna, Pavel and Sgall, Petr 1980 "Functional Sentence Perspective, the Question Test and Intensional Semantics", Statistical Methods in Linguistics — Journal of Linguistic Calculus 1/2, 141 — 160. Mathesius, Vilem 1929 "Zur Satzperspektive im modernen Englisch", Archiv für das Studium der neueren Sprachen und Literatur 155, 202-210. 1939 "O tak zvanem aktualnim cleneni vitnem", Slovo a slovesnost 5, 171 — 174; translated as: "On Information-Bearing Structure of the Sentence" in S. Kuno (ed.): Harvard Studies in Syntax and Semantics 1 (1975), 467-480. 1942 "Rec a sloh" in Ctent ja^yce a poesii, 11 — 102; as a separate volume Prague: Cs. spisovatel 1966. Mey, Jacob 1981 "General Editor's Preface", in F. Coulmas (ed.): Conversational Routine. The Hague: Mouton, VII-X. Partee, Barbara and Bach, Emmon W. 1981 "Quantification, Pronouns and VP Anaphora" in J. A. G. Groenendijk et al. (eds.): Formal Methods in the Study of Language. Amsterdam: Mathematisch Centrum, 445-481. Petöfi, Jänos S. 1978 "A Few Comments on the Methodology of Text-Theoretical Research", Journal of Pragmatics 2, 365-372. Platek, Martin; Sgall, J. and Sgall, Petr 1984 "A Dependency Base for a Linguistic Description", in: Sgall, P. (ed.): Contributions to Functional Syntax, Semantics and Language Comprehension. Prague: Academia, 63 — 97. Posner, Roland 1972 a Theorie des Kommentieren!: Eine Grundlagenstudie %ur Semantik und Pragmatik (Linguistische Forschungen 9). Frankfurt: Athenäum. 1972b "Dialogsorten — Die Verwendung von MikroStrukturen zur Textklassifizierung", in E. Gülich and W. Raible (eds.): Textsorten: Differen^ierungskriterien aus linguistischer Sicht (Athenäum — Skripten Anglistik 5). Frankfurt/M.: Athenäum, 183—197. Reinhart, Tanya 1981 "Pragmatics and Linguistics: an Analysis of Sentence Topics", Philosophica 27, 53-93. Searle, John R. 1970 Speech Acts: An Essay in the Philosophy of Language. London: Cambridge University Press. Seuren, Pieter A. M. 1969 Operators and Nucleus: A Contribution to the Theory of Grammar. Cambridge: University Press. Sgall, Petr 1967 "Functional Sentence Perspective in a Generative Description", Prague Studies in Mathematical Linguistics 2, 203—225.

96

E. Hajicovä, P. Sgall 1979

"Towards a Definition of Focus and Topic", Prague Bulletin of Mathematical Linguistics 31, 3—25; 32 (1980), 24 — 32; reprinted in Prague Studies in Mathematical Linguistics 1 (1981), 173-198. Sgall, Petr; Hajicova, Eva and Panevovä, Jarmila 1986 The Meaning of the Sentence in Its Semantic and Pragmatic Aspects. Prague: Academia. von Stechow, Arnim 1980 Notes on Topic and Focus of Interrogatives and Indicatives. Konstanz: NUC, Dt. Bibl. Wegener, Philipp 1885 Untersuchungen über die Grundfragen des Sprachlebens. Halle/S.: Max Niemeyer. Weil, Henri 1844 De l'ordre des mots dans /es langues anciennes comparees aux langues modernes. Paris; translated as: The Order of Words in the Ancient Languages Compared with That of the Modern Languages, Boston 1887; new edition with an introduction by Aldo Scaglione (Amsterdam Studies in the Theory and History of Linguistics I, 14). Amsterdam: Benjamins, 1978. Wollmann, Alfred 1981 "Wortstellung und kommunikative Gliederung in übersetzungsäquivalenten Sätzen" in J. Esser and A. Hübler (eds.): Forms and Functions (Tübinger Beiträge zur Linguistik 149). Tübingen: Narr, 153-161. Zemb, Jean Marie 1968 Les structures logiques de la proposition allemande: Contribution ä l'etude des rapports entre le langage et la pensee. Paris: O. C. D. L.

ELISABETH RUDOLPH

Connective Relations — Connective Expressions — Connective Structures 7. Introduction

The following considerations deal with the status and function of connection and are based on the results of several years of investigations of German and various Romance languages. These studies have confirmed my belief that the Indo-European languages are not only related by the same parentage, but also share a great number of properties, particularly in sentence connection. The description of the structures in one language can therefore become the starting point for a more general consideration of the ways in which texts or parts of texts are produced and understood with the aid of connective expressions. Scholars who have treated the theory and use of connective relations and expressions are aware of the fact that the situations of natural language are far too complicated to be explained by the application of simple models. It seems impossible to formulate rules for connection, because the different connective relations are interrelated to such an extent that paraphrases, transformations, and substitutions are often acceptable within the connective expressions. Due to space limitation and given that these problems cannot easily be solved, I shall, in the following three chapters, concentrate on general considerations. They will show that connective expressions are compositional elements of the text offered by the language without any obligation of being used as necessary components of the text. The examples are taken from English grammars and handbooks or have already been discussed in linguistic research. Most of them are short and context-free. In order to give an idea of language in use other examples are taken from an editorial of The Times which appeared in September, 1982. The article can be found in its entirety in the Appendix to this article. With regard to terminology, I have consistently tried to use traditional terms, so that a few remarks will be sufficient to prevent misunderstandings. In this paper, the term "connection" refers (1) to the process of connecting two or more entities and (2) to the result of this process, the connected entities. It is used in a wide sense and does not specify a class of phenomena as being opposite to another class.

98

E. Rudolph

My special interest is directed to a type of connection which is characterized by the fact that the connected elements are configurations or representations of states of affairs. The corresponding grammatical units are known as compound sentences, whereas the mental representations are the "connective relations". In a text, which I understand as a specific utterance or a sequence of utterances, we find manifestations of connective relations which often appear explicitly. In detailed analysis of these manifestations, the term "connective expression", sometimes abbreviated as "connective", refers to conjunctions, prepositions, and other surface means of connection. In the table of terminology (cf. 6), the concepts which relate to the terms used are represented in their interdependency so that the correspondence within the five different types is demonstrated.

2. Connective relations

An answer to the question whether explicit manifestations of connective relations are to be seen as necessary elements of text construction seems to depend on the definition of the concept of connective relation. Connective relations hold a special place among the great number of possible relations because they apply to complex propositions which have arguments that themselves are propositions and that do not merely represent persons or objects. Connective relations therefore may be described as combinations of propositions. On the other hand, the question arises whether the manifestations of connective relations reflect special functions within the text or not. From an empirical point of view this cannot be explained without first defining the term text. One remarkable aspect is to be noticed which the concept of text has in common with other fundamental concepts of linguistics: the term is very useful to linguistic theory but we do not have a generally accepted definition at our disposal. In this paper I shall consider text as a specific spoken or written utterance or a sequence of utterances being structured so that a text unity is perceived by the hearer/reader. It is noteworthy that in suggestions for definitions and descriptions, the most frequently mentioned characteristics of texts are the terms coherence and cohesion, which seem to be the labels to cover the intuitively known unity of a text. It is not my aim to discuss the problems of coherence and cohesion. For present purposes it is sufficient to understand coherence as a connection in the mental representation attributed to a text, and cohesion as a connection of the elements of a sequence of sentences corresponding to a text. Among the constituting elements of text cohesion that have been enumerated, we often find connective expressions and conjunctions.

Connective Relations — Connective Expressions — Connective Structures

99

2.1. Propositions in connective relations

Presumably the main function of connective expressions is to indicate the existence of a connective relation between two propositions, most often but not always expressed by sentences. A sequence of two propositions A and B expresses a new thought A-rel-B on a level other than that of the isolated propositions, though also representable in a proposition. This new proposition is composed of the two former propositions and can be represented as a complex proposition. Unlike the propositions A and B indicating facts or events, the complex proposition corresponding to a connective relation indicates a thought on facts or events. A thought may be characterized as a fact or an event, if these terms are taken in a wider sense. But there is no doubt that a thought is another kind of fact than an action, even if a thought becomes a speech act when expressed in words. In linguistic research attempts have been made to deal with these problems when analysing compound sentences. These are usually represented in either of the following descriptions. In both cases (la) and (lb), the complex sentence (1) is represented in a relation with a higher predicate; the abstract verb to cause corresponds to CAUS in the descriptions. (1) This is more difficult in West Germany because the constitution discourages mid-term change. (The Times) (a) CAUS (p, q): p is equivalent to the first proposition, q to the second (b)

CAUS

the constitution discourages midterm change

this is more difficult in West Germany Figure 1

I will not discuss the details of these forms of analysis, which have quite different aims of description. As far as connective relations are concerned, it is clear that the explanatory force of the abstract verb is not very strong. On the other hand, it does demonstrate that the complex sentence has an additional semantic value that is not found in either of its members. We must now ask whether this semantic value can be attributed to the connective expression realized in (1) as the conjunction because.

100

E. Rudolph

2.2. Position of speaker and hearer Although manifestations of connective relations are signals of text coherence, they are far from being necessary elements of text construction. However, if a speaker intends to produce a speech act involving a connective relation and wants to be clear and unmistakable, he is forced to use a connective expression explicitly marking that relation. This argument does not contradict the wellknown examples of parallelism between connective relations expressed with or without connectives. The following examples are taken from van Dijk (1979: 514). (2) (a) John won't come tonight because he is ill. (b) John won't come tonight. He is ill. In colloquial speech the second sentence of (2b) will always be understood as the indication of a reason. It is of high plausibility that a person being ill will stay at home and not go elsewhere. Therefore, (2a) and (2b) are semantically equal when uttered at a meeting of a group of friends, van Dijk argues that "within the 'theory', there is no distinction between a complex (coordinated) sentence on the one hand and a sequence of sentences on the other hand, ...". He requires of a grammar that it capture "within a unifying account, the relations between the clauses of (2a) on the one hand, and those between the sentences in (2b) on the other hand" (1979: 514). The difficulties of interpretation are illustrated by (1), where it seems inopportune to remove the connective expression, even if the semantic relation is clear in the context, as can be seen in (3): (3) In such circumstances it would be normal for any democratic country to let another party have a try at doing better. This is more difficult in West Germany because the constitution discourages mid-term change. Hence the disturbance is greater when change comes to seem necessary. (The Times). The role of connective relations in text production seems to be determined by two factors: — Manifestations of connective relations are signals for a text, since they usually occur within a text and not in isolated position. This is true even for colloquial examples such as (2a) or (2b), which would presumably be uttered in a longer text in which the co-text is known by speaker and hearer. — Explicit manifestations of connective relations are not to be regarded as necessary elements of text production. They are not found in many texts and discourses, especially in narrative texts full of actions. As I have pointed out, a connective relation indicates a thought on facts or events. Obviously another kind of cognitive ability is demanded, if a speaker

Connective Relations — Connective Expressions — Connective Structures

101

regards events and actions not simply as they occurred but as facts that are interrelated and interdependent. In this case the speaker's position is characterized by a certain distance from the reported actions and events. The hearer is asked to keep a similar distance which prevents him from identifying himself with the actions. Such positions are normally the case in philosophy, in scientific research, and in commentaries on whatever topic. In narrative texts where the hearer/reader is invited to plunge into the story, there is no need for a speculative position, but in highly psychological novels in which the author tries to explain the motives and historical backgrounds related to the protagonists, a greater need for manifestations of connective relations will be found. 2.3. Rules and obligations

In sentence grammar, construction is guided by rules. Because connective relations are represented in texts, we have to ask what kind of rules are found in text production. It is assumed that there are obligatory rules that operate in all texts and optional rules that are only valid under certain conditions. Manifestations of connective relations do not occur in all texts, and because of this obligatory rules cannot be formulated for them. However, we can try to find out the optional rules from which the speaker can choose. If we regard connective relations exclusively from the point of view the speaker might adopt, we will not find any obligation for him to use a manifestation of a connective relation: the speaker is entirely free in his decision to choose a connective expression or to do without it. From the hearer's point of view, however, things look quite different. Understanding an entire text demands a certain amount of cognitive ability, and attention will diminish inasmuch as the text loses attractivity. Texts are produced to be understood. In verbal interaction the partners have to cooperate. Therefore the hearer demands not only that the speaker observe Grice's maxims but also that he communicate his opinion and make his standpoint clear. The speaker is not asked for permanent confessions as to his thinking and believing, but the hearer does want to know whether the speaker is convinced that the reported events are true, how he evaluates the facts, and what sort of mutual relationship he attributes to the facts. Langleben (1979: 251) gives (4a) and (4b) as examples of coherence which is apparently not affected by different forms of connection. (4) (a) Mrs Leivers sat for some time talking to the boy, although she was needed at her work. She was too polite to leave him. (D.H. Lawrence), (b) Mrs Leivers sat for some time talking to the boy. She was needed at her work, but she was too polite to leave him. The only difference between (4a) and (4b) lies in the different connective relations; in (4a) the first two sentences are connected, in (4b) the last two

102

E. Rudolph

sentences. I agree with Langleben (1979: 251) that "both (4a) and (4b) are based on a latent line of coherence 'sat talking — was needed — too polite'", but I doubt that "in case that the conjunctions are absent, the causality relations are all the same to be revealed and represented in the semantic description" of this passage, as may be proved by (4c): (4) (c) Mrs Leivers sat for some time talking to the boy. She was needed at her work. She was too polite to leave him. Those readers who would understand that Mrs Leivers' politeness did not let her return to her work would at least claim that there is a link missing between sentence 2 and 3. Even if we insert something like "She wanted to get up but could not", the contrast between was needed and too polite is not sufficiently prepared. The effect of connective expressions like although or but cannot be easily expressed by a sentence. Other forms of connecting are illustrated in the following paraphrases: (4) (d) Mrs Leivers sat for some time talking to the boy, although she was needed at her work. But she was too polite to leave him. (e) Mrs Leivers sat for some time talking to the boy, because she was too polite to leave him, although she was needed at her work. (f) Mrs Leivers sat for some time talking to the boy, although she was needed at her work. For she was too polite to leave him. The two connectives in (4d) do not give more information than one connective does. The connection with two different relations, contrast and causality, as in (4d), requires that the order of succession be reversed. If the order is maintained as in (4f), only conjunctions of coordination are acceptable. I have discussed these examples in detail because with short sentences and a nearly context-free situation the problems and difficulties of finding out rules for the use of manifestations of connective relations in text production can be demonstrated with greater evidence. Yet another example may show that the regularities in the manifestations of connective relations are of a special kind. (5) (a) Now they [= Christian Democrats] have another chance, not because they have earned it but because the Free Democrats fear being dragged downwards into oblivion by the sinking fortunes of the Social Democrats and have therefore decided to jump ship. (The Times). If we withdraw the connectives from this text segment, we render it senseless: (5) (b) They have another chance. They have not earned it. The Free Democrats fear being dragged downwards into oblivion by the sinking fortunes of the Social Democrats and have decided to jump ship.

Connective Relations — Connective Expressions — Connective Structures

103

In (5b) no one would see a latent causal relation; (5b) is more like an enumeration of facts and evaluations. Between the first two sentences and the third, there is a gap broadening with the greater complexity of the last sentence. The most suitable interpretation of the passage would be that the author, after finishing his report on the Christian Democrats, turns to another subject and evaluates the situation of the Free Democrats. In (5a), however, the obvious interpretation is that the author links up the information and his own judgement, which he gives a foundation by considering the prospects of the Free Democrats. The same order of information on facts and opinions suggests a quite different interpretation through (5a) or (5b). In other words, (5a) and (5b) are parts of different semantic macro-structures. While (5a) belongs to a macro-structure dealing with reasonable explanations of political change, the macro-structure of (5b) is to be found within the frame of objective reports on political events. It may be concluded from theses examples that the meaning of a text is essentially altered if the existing connectives are withdrawn from it or, respectively, if new connectives are inserted. However, reevaluating (2a) and (2b) might lead to a different conclusion. In this case the connective relation seems to be independent of its manifestation as a causal relation or not. The interpretation of a sequence of two propositions denoting facts as expressed in the two sentences of (2b) seems to be predetermined. It must be a connection of causality, whether the sequence is parallel to the causal clause of (2a) or parallel to a clause of result as in (2c): (2) (c) John is ill. He won't come tonight. John is ill. Therefore he won't come tonight. The connection of causality remains the same in (2) (a —c), since the facts are related in that special manner we are acquainted with by our knowledge of the world. The relation of result and the causal relation depend on the facts denoted by the propositions. The fact that a person is ill has a series of consequences, one of them being the fact that this person will not come to a certain place. On the other hand, the fact that a person is not coming may have a series of reasons, among them the fact that he is ill. In situations of personal and social life there do not exist causal relations of the strong causality which is attributed to events under the law of nature, where Λ and Β are related in such a way that A always has the result Β and Β can be predicted if A is realized. In personal and social life A may have various consequences B, C, D, ... and Ζ may have various reasons Y, X, W, ... The speaker has to express a connective relation of causality if we are to be informed which consequence or which reason among the series of possibilities is meant in the actual case.

104

E. Rudolph

2.4. Conclusions

In the world of our daily experience, as in many other possible worlds, facts and objects can be regarded in their various relations; isolated objects do not exist. Therefore we are acquainted with the interrelationship of our surroundings and do not ask for details. Only when a special relation becomes important for a situation or an event are we interested in the precise connection of facts and objects. As speakers we have an implicit and unconscious knowledge of the point when it is necessary to give detailed information so that the hearer will continue to pay attention. It is, however, difficult to predict when this point will be reached, because it depends on the whole situation of the interaction, on the social and intellectual status of the interactors, on the co- and context and many other factors. Therefore, in linguistics we generally assume a standard situation (characteriaed by an average social and intellectual status of the participants) and disregard other factors. In such a standard situation the speaker will mention every relation he believes important for the understanding of the text, but he will keep away from irrelevant relations. The hearer in the standard situation will direct his attention especially to those relations which are relevant for his understanding and comprehension and he will not take any notice of the irrelevant relations. This aspect of relevant relations has to be distinguished from relevance logics, although there are some similarities, van Dijk (1977: 25) points out that "both in natural and in 'relevance logics' propositions are required somehow to be connected when occurring in one complex formula". He explains that the antecedents of condition and implication or entailment "must be 'relevant' to their postcedents". For relevance relations very strict conditions have been formulated, whereas connective relations have not found a consistent description. This fact seems to be a consequence of two aspects of connective relations: — They are always embedded in a network of situations, presuppositions, encyclopedic knowledge, special knowledge of discourse, personal conditions and so on, interactional factors of which the participants are generally unaware. — Explicit manifestations of connective relations are signals for the author's opinion on the facts mentioned and assist the hearer/reader in interpreting the meaning of a text/discourse. But it seems impossible to predict in which situation manifestations of connective relations might appear. 2.5. Possibilities of connection Among the different means of text production found on various levels of language, connection marks several ways of connecting elements of speech, such as one element forming part of another element, one element being

Connective Relations — Connective Expressions — Connective Structures

105

dependent on another element, two elements being equivalent members of a unity, two elements being interrelated so that none of them is superior to the other, and so on. The elements, grammatically known as word classes, have different functions in the utterances: some of them refer to persons or objects; others refer to actions, states, events, situations; others refer to properties, qualities, quantities. One type of connection stands out among these different possibilities because it combines two propositions indicating independent facts or events. This type of connection is realized on the surface by prepositions and conjunctions which reflect the speaker's opinion and point of view with regard to this connection. The two aspects of connection can be distinguished as follows: The speaker's point of view

If a speaker perceives two facts/events as being related, he has to decide which type of connection best corresponds to his perception of this relationship, each connection being bound to another dominating idea. In organizing his utterance in accordance with his intentions, the speaker gives an idea of his convictions of how the world is structured. The speaker's opinion on a connective relation

If a speaker wishes to express his conviction that the facts or events are related in a connection of contrast or a connection of causality, he has to decide among several possibilities corresponding with different points of view he may adopt; each possibility is represented in a different relation. By choosing one of the relations belonging to the respective connection he expresses his opinion. He has a free choice, but he is forced to make a choice. The speaker's point of view and opinion are situated on a level of communication superior to the level of connecting facts or events. Thus the connection gives two pieces of information in an order of hierarchy. Connective relations are found within the frame of this hierarchy connection. As far as we look back in grammatical tradition, the hierarchy connection has always found a certain interest. This is because of a small set of words which chiefly or exclusively express this connection: prepositions and conjunctions. In grammatical descriptions, suggestions have been made for classifying conjunctions, starting from various points of view such as coordination vs. subordination or adverbial clauses vs. noun clauses. The traditional class names reflect semantic values corresponding to the meaning intuitively attached to the various compound sentences in natural language. Linguists are well aware of the fact that freedom of choice in natural language offers the opportunity of using another expression for almost the same meaning. Paraphrases, repetitions, the capacity of memory, and most of the ordinary means of communication have their foundation in this basic freedom of choice. The disadvantage, however, is obviously situated in the

106

E. Rudolph

difficulty of classification, all sets of expressions being open sets and subsets with overlappings and crossings. One of the tasks in text theory is the elaboration of a theory of connection comprising general as well as special possibilities of connection. This theory would have to account for the differences between connecting objects or connecting facts. It would have to make clear that connected propositions do not necessarily correspond to connected sentences, since prepositional phrases are often genuine expressions of propositions. 2.6. The hierarchy connection

In this connection, propositions are bound together under various labels. Corresponding to wider aspects of connection, the whole set of possible connections is seen as having a number of subsets, each of which would comprise several elements (cf. the table in 2.8). There seem to be four basic ways to realize the combination of two propositions denoting facts and events. Wishing to avoid a theoretical discussion of other means of representation, I will call the two propositions A and B. In a series starting from the simplest and ending in the most complex aspect of hierarchy, the set of connections may be described as follows. For reasons of clarity I give various names used in linguistics for each type of connection. Connection of addition; concomitance of A and B; equivalence hierarchy; AND-relation; description relation; summation

I.

II.

A -B

Connection of contrast; delimitation of A from B; difference hierarchy; BUT-relation; adversative relation

III.

A , B

Connection of time; succession of A and B; temporal hierarchy; THEN-relation; time-and-space-relation

IV.

Connection of causality; A is the cause of the effect B; asymmetric causal hierarchy; BECAUSE-relation; causality relation

The four basic types of connection are presented as framed labels so that they cannot be confused with other formalisations. Although the names of classification are generally known, they cannot replace exact definitions based on thorough investigations and proved by examples taken from natural language. Connecting propositions is not possible without making a new meaning, since each connection reflects a different viewpoint of the speaker with regard to the connected facts/events. Moreover, hierarchy has a meaning of its own. The minimum semantic value of hierarchy is that the connected entities

Connective Relations — Connective Expressions — Connective Structures

107

involve a special order, usually with a top entity dominating the lower entities. When only two entities are connected, as in the above case of propositions, the concept of hierarchy is modified: the act of connection itself is the dominating entity. Each connection demonstrates another mode of connecting which is equivalent to another dominating idea. The different relations, being elements of the corresponding set of connection, are bound together by this dominating idea, which can be described as a constant in comparison with the relations acting as variables. It is not easy, however, to formulate the dominating idea of each connection. The following list, which contains classifying labels of relations taken from traditional grammars and descriptions, may demonstrate the extent of the connections possible. Λ + Β A —B A , Β A -» Β

2.7.

addition, alternation, comparison, conjunction, disjunction, enumeration, exception, facts, manner, neutrality, preference, proportion, transition adversativity, concession, contrast, contrastive assertion, degree, restriction duration, distance, direction, frequency, place, time, time-when causality, condition, circumstance, comment, consequence, direct cause, finality, inference, instrument, hypothetical meaning, means, purpose, reason, result

The two functions of hierarchy connections

From the viewpoint of functional importance for the text, the four classes of hierarchy connection have different values, depending on the meaning of the connection as such. The connection A + Β means that the two propositions are closely linked so that the second proposition offers information that has not been given by the first proposition. This additional information may be of a different kind, as can be deduced from the traditional labels, but the textual importance is the same for the whole set. This connection has a special function in organizing the text. The connection A — Β means that in the speaker's opinion the two propositions denote facts that are not well matched, the second proposition being in contrast to or at least questioning the information given by the first proposition. This connection is important for communication and interaction. The connection A , Β means that for a good understanding of the information of the text as a whole, it is important to know the temporal sequence of the two propositions. This is again of textual significance for organizing the information, but the speaker's opinion is of minor importance for it. The connection A —> Β means that the speaker's view of the kind of relation between the facts denoted by the connected propositions is of importance for the text as a means of communication. The isolated facts and events are often independent of the speaker as a person, but by linking them together

108

E. Rudolph

he gives his explanatory opinion of how the world is organized in which the facts and events are located. Thus the connection of causality has great value in argumentative discourse and psychological considerations. From these descriptions it can be concluded that the four classes of connection form two pairs of function, namely (a) the speaker organizes the information in accordance with his point of view by referring to the complexity of the world mentioned: A + B and A , B; (b) the speaker indicates his position within the frame of the world mentioned by reporting his opinions of facts and events: A — B and A -> B. Obviously the amount of meaning — if we may be allowed to measure such an unmeasurable thing as meaning — is different in the two pairs: (a) conveys less meaning than (b). But considering the pairs as far as text production is concerned, (a) gains a lead on (b) because of its importance for text coherence. In accordance with the above I shall use the term connective relation only for those two sets of hierarchy connections indicating the speaker's personal opinion. Thus I shall exclude the connections A + B (addition) and A , B (time) from further consideration. This decision is not an isolated one. Though from quite different points of view, van Dijk (1977) and Pritsche (1982) make the same distinction and reserve the term 'connective relation' only for contrast and causality connections. In natural language connective relations play an important part; connective expressions are used in everyday discourse as well as in highly specified argumentations and scientific investigations. Considered from a pragmatic point of view, connective expressions can be used for a great variety of speech acts. 2.8. Table of connections and connective relations connection

hierarchy connection

A - B

A ,B

connection of contrast

Figure 2 a

connection of causality connective relations

Connective Relations — Connective Expressions — Connective Structures

109

connection of contrast

^^

adversative Λ relation )

jW

concessive relation

connection of causality /conditional ^v V relation )

relation of purpose

relation of result

causal relation

Figure 2 b

3. Connective expressions

The main function of connective expressions is to indicate that there is a connective relation, i.e. to indicate the speaker's opinion in the special relationship of two facts or events in a possible world which is in most cases the real world we are living in. For the hearer the connective expressions are signs giving instructions for cognitive operations in the task of decoding the speaker's utterances. After pointing to the variety of linguistic means for expressing the obviously finite set of relations (cf. table 3.1), I will argue that connective expressions can be used as functional indicators by the speaker as an instrument to give his personal views together with information on facts. The speaker can use connective expressions to organize the text, to create a special textual perspective, to stress climactic points and renounce former points of view, to conceal a real opinion behind conventions, and to order states of affairs in structures that give secondary importance to states of affairs of secondary value. I will not try to systematize these various possibilities because they are closely linked to context and situations and to the development of the text as such. Connective expressions are used by the speaker and are understood by the hearer as instructions for cognitive operations. The speaker, guided by his intentions, selects one connective expression among the variety offered by the language system. Connective relations have been described on the basis of their function as a set of connections indicating the speaker's personal opinion and consisting of two subsets of hierarchy connections, namely A — Β (contrast) and A -»· Β (causality). The number of elements within these two connections is finite, but it is dependent (a) on the kind of specification made, and (b) on the distinction and assignment in case of doubt. Thus according to the explications given above the adversative and

110

E. Rudolph

the concessive relations are subtypes of the connection of contrast. The connection of causality comprises as subtypes the relations of condition, purpose, result, and cause/reason. 3.1. List of connective expressions The following list in Table 1 is a survey of the most commonly used English connective expressions as they are mentioned in grammars and handbooks. On the whole, the Indo-European languages have quite comparable structures with slight differences in details, as may be seen from contrastive studies such as Henschelmann (1977) and Dorfmüller-Karpusa (1982). Table 1 demonstrates distributional differences between the various relations. Some of the relations are expressed predominantly by means of coordinate structures, others only through subordinate, non-finite, and verbless clauses and through other means. The semantic interrelationship can be seen from the fact that several connectives appear in more than one category. This is especially the case with that and so that (purpose, result), if (condition, concession), to cause (reason, result). Other connectives may also introduce a clause of time, such as while, when, since, as, as long as.

Table 1

contrast adversative

contrast concession

coordination

subordination

but yet still however only even so nevertheless notwithstanding all the same by comparison instead by way of contrast

whereas while/whilst while on the contrary when

although though notwithstanding (that) however even if /though except that save that no matter (how)

other means

in spite of despite for + all with + all notwithstanding relative pronouns and adverbs in -ever

Connective Relations — Connective Expressions — Connective Structures

Ill

Table 1 (continued) subordination

other means

condition

if unless on condition that in case (that) provided (that) supposing (that) as if though whether ... or not as long as so long as if only

in case of but for pro-clauses: if so — if not if -f non-finite and verbless clauses subject-operator inversion imperative: (then) had better + inf: or (else)

purpose

that in order that so that lest in case for fear (that)

to + inf in order to + inf so as to + inf for + noun + to

therefore thus hence so that is why owing to this for that (very) reason consequently accordingly

so that so so ... (that) such ... (that) that so much so that to such an extent that

so ... as to as a result the result is the consequence was

for

because since as now that now seeing that in that

because of on account of from -\ out of 1 f noun for 1 J phrase through

coordination

result consequence

cause

for that reason coordination without conjunction

to cause to result from

to cause to make non-finite and verbless clauses participle clause

However, not result following tion from

this semantic cumulation in some connective expressions does in any ambiguity of the corresponding connective relations. In the passages I have chosen the method of substitution and transformaone connective relation into another in order to demonstrate that

112

E.Rudolph

the instructions for cognitive operations given by the speaker generally are clear enough for the hearer to understand. In some cases, however, the hearer has to take into consideration not only the connective expressions but other signals, as well. 3.2. Substitution in the same relation: if

Two different sorts of substitution must be recognized: the substitution of expressions within the same relation and substitution with an expression from a different relation which is made possible by a semantic interrelationship. The first sort of substitution — within the same relation — pragmatically appears as the stylistic opportunity of variants. This seems to be the reason why such substitutions are mentioned in grammar. After having introduced open and hypothetical conditions Leech and Svartvik (1975: 96) give several examples to demonstrate different possibilities of applying connectives in conditional clauses: (6) (a) If you feel seasick, take one of these pills. (b) Take these pills, in case you feel ill on the boat. The difference between (6a) and (6b) is clear according to Leech and Svartvik: "In case specifies a future condition that may or may not arise", whereas for if they give the simple explication that the speaker does not know whether the state of affairs will be true. The other examples show even more that the selection of special connective expressions depends on the context, perhaps not entirely, but to a very large extent: (7) I'll lend you the money on condition that you return it within six months. (8) Provided that / so long as they had plenty to eat and drink, the men were happy. (9) In case of difficulty call the operator. In all these examples the speaker does not know whether the condition will be realized. But in (7 — 9) this general uncertainty is not part of the content of the message. Stress is placed on other aspects. The essential point for Leech and Svartvik is the following: "On condition that stipulates or lays down a condition to which a person must agree. Provided that and so long as resemble on condition that in having the restrictive implication of 'if and only if ...'." The semantic value of either connective expression is closely related to the context. A person who lends money is interested in being given a guarantee that he won't lose it; therefore he will set down conditions in a contract which has to be agreed upon by the borrower. The shortest form of such a contract is given in (7), which implies that the speaker will only lend the money if the addressee agrees to the condition. However, money-lending is not necessarily bound to such conditions, as may be seen in (7a —b), also taken from Leech and Svartvik (1975: 96):

Connective Relations — Connective Expressions — Connective Structures

113

(7) (a) I'll lend Peter the money if he needs it. (b) I'd lend Peter the money if he needed it. In (7a) the speaker makes a promise without setting down conditions as to the date of repayment, the only condition for keeping the promise being the social circumstances of the borrower. How this condition works is made clear in (7b), where the speaker will not lend the money because he is convinced that Peter does not need it. Although provided that and so long as in (8) have the same restrictive implication of if and only if, conditions are not set down like the terms of a contract; an obvious condition of life is emphasized in which the amount of food and drink consumed is related to the happiness of the people concerned. From (8) it may be deduced that the men cease to be happy as soon as they do not have plenty to eat and drink. There is an essential difference between (8) and (7). From (7) it cannot be deduced that the speaker will not lend the money if the condition of returning the money is not fulfilled within the period of six months. Given common knowledge of the real world everyone takes into account that lending an object implies both an obligation on the part of the borrower to return it as well as the expectation of the lender to get the object back, whether or not the state of affairs he hopes for is realized. The crucial point is not the interval of time established in (7), but the social implications of the act of lending which may appear as semantic properties of the verb to lend. Nevertheless, if may be substituted for the connective expressions introducing the conditional clause in (7) and (8) without noticeable change of meaning. The same does not apply to (9), where the verbless clause cannot be kept in a paraphrase beginning with if. Regardless of the verb selected, the meaning will change. Utterances with connective expressions are generally part of a text. In grammars and linguistic investigations the surrounding text is often (if not always) omitted and the common knowledge of the real world gives a frame for interpretation. No one has difficulties understanding the meaning of (9) because we remember various occasions of having read such an instruction. For other examples, such as (10), taken from van Dijk (1977: 46), it is necessary to invent an appropriate context because such examples represent neither spoken nor written language, but merely grammar language. (10) If John is ill, we must call a doctor. Who would be the speaker of (10)? In a family context, in which John, a child, is ill and his father and mother are alarmed, a visitor, perhaps a good friend or aunt, might be the speaker. Father and mother would merely say "we must call a doctor", and the grandparents would at least use the variant "jou must call a doctor". Another context: The family on holidays together with friends; the day is bright; there are plans to go to the beach; John says he would prefer to stay

114

E.Rudolph

at home, and he tries to look sick, but everybody knows that he is quite well and just has an exciting book to read. Any of the family members could be the speaker. The situations are different; in the first context John is ill, whereas in the second context he is not ill, but tries to give that impression. The following explication (van Dijk 1977: 46) will unanimously be accepted: "The speaker knows that the antecedent expresses a true proposition, but simply repeats it, as a presupposition [...], stating that such a fact is a 'sufficient condition' for the fact denoted by the postcedent." Though the explication refers to (10) it is clearly intended to cover the general semantic background of a conditional clause where since may be substituted for if. Such a substitution points to the fact that there is at least one common element in the sets of conditional relation and relation of cause/reason. This element may be described as an underlying causal connection, or to put it in van Dijk's words: "Because a reason or a cause is involved we may paraphrase these sentences with initial since." There still remains the problem with terms from logic such as "truth, falsity, sufficient and necessary condition". When used in linguistic explications they may be easily misinterpreted as implying that natural language is obedient to the laws of logics or that speakers carefully consider their words before speaking. In reality, however, we make use of the patterns of language without reflecting. 3.3. Semantic interrelationship in the connection of contrast: but — although

The second sort of substitution is a surface manifestation of the fact that the elements of a set of connections are closely interrelated. Generally this substitution is mentioned in linguistic investigations but not in traditional, normative or descriptive grammars. Quirk and Greenbaum (1974: 254) give some examples of the semantic relationships to be found in both coordination and subordination; these are also good examples of substitution: (11) (a) (b) (c) (d)

He tried hard, but he failed Although he tried hard, he failed He tried hard, jet he failed He tried hard, but jet he failed

The connection of contrast is composed of the adversative and the concessive relations. The common element of both is that the second proposition is portrayed as unexpected with respect to the first proposition. In general, we expect that trying hard will result in succeeding, since this is the basic aim of trying. The unexpectedness of the outcome is the reason why the utterance is made. Whenever the result of an action, expected and generally achieved, is not given, there is a need to express this in a connection of contrast, the two possibilities being the adversative and the concessive relations.

Connective Relations — Connective Expressions — Connective Structures

115

11 (a—d) are synonymous to the degree that they are expressions of the same connection of propositions A — B. The second proposition (B) presents a contrast to or at least questions the information given by the first proposition (A). In this special case the contrast is built up by the contrary-to-expectation fact denoted by B. The differences are of two kinds: the first one is between although and the other expressions; the second one concerns the three alternatives in the expression of B. I will not consider the second difference, which seems to be a gradual reinforcement of the contrary-to-expectation regret that may be used with stylistic motives. The first difference concerns the distinction of adversative and concessive relations. As Table 1 shows, the syntactic status of A and B differs: in the adversative expression B is marked with a conjunction of coordination; in the concessive relation A is marked with a conjunction of subordination. The adversative relation with the characteristics of unexpectedness occurs only in coordinate structures, whereas the concessive relation can only be realized in subordinate clauses. Thus (Ha) and (lib) are substitutions of a special kind: the sequence of sentences is the same, but the position of the connective changes from B to A. The consequence of this change of the marking position is a different meaning. In (11 a) the message has two parts of equal importance. The speaker in A gives a piece of information about a fact and in B he adds a surprising result, thus giving the hearer the opportunity to be as surprised as he himself might have been. In (lib) the situation is clear from the beginning: the speaker has obviously not been surprised by the 'unexpected' fact in B, but he supposes the hearer to be surprised; therefore his selection of the concessive relation gives the cognitive instruction that a surprise is to be expected. This is also the case with sequence (lib'), because the information is different: (lib') He failed, although he tried hard. When the unexpected result B is presented first, the speaker indicates through intonation that the message is not yet finished. Passage (12), which will not be discussed in detail, provides examples of the contrast relations that cannot always be substituted for one another: (12) But although the crisis has been precipitated largely by their tactical considerations it does reflect deeper stresses which derive from threats to the fundamental assumptions which have held the West German consensus together. One is that economic growth is generally always attainable, even if temporarily interrupted by world recession. But there have now been two years of stagnation. Another is that unemployment will remain modest. But the figure has now reached two million. Another is that there will always be the funds to support a reassuringly high level of social security. But cuts have now become inevitable. Another is that loyalty to the United States is compatible with the pursuit of traditional German interests to the East. But Washington is now pressing West Germany to make an unacceptable choice between East and West. (The Times).

In this passage four fundamental assumptions are listed, and each is followed by a contrasting rectification. At the top of the passage, as a kind of

116

E.Rudolph

introduction, the combination of adversative and concessive relations reveals the author's opinion on the state of affairs. 3.4. Transformation result: cause

Another semantic interrelationship is concerned with two of the four relations of the connection of causality, the relation of consequence/result and the relation of cause/reason, which now and then are described as transformations of one another (Härtung, 1964: 193; Eggers, 1970: 89; Boettcher, 1972: 84). The following examples, taken from van Dijk (1977: 47), demonstrate the specific property of this transformation: (13) (a) It has been a very hot summer, therefore the crop has been destroyed, (b) Because it has been a very hot summer, the crop has been destroyed. The order of propositions A and B is reflected in the sequence of utterances, which is the same in both examples. The two vary only with regard to the position of the connective: in (13a) it is placed at the beginning of B, in (13b) at the beginning of A. Undoubtedly the speaker's point of view in both cases is that the extreme heat is the cause of the destruction. Regarding the propositional connection it may seem to be superfluous that the speaker has the choice between two relations. On the other hand, the relations of result and cause differ in the kind of instruction given by the connective expression. In (13a) proposition A (hot summer) mentions a fact, the truth of which is not denied by the hearer; but perhaps almost forgotten, the situation being autumn or winter, proposition B (crop destroyed) supports proposition A by mentioning a fact that to a certain degree is new for the hearer. The expression therefore invites the hearer to accept the fact denoted by proposition .ß as a plausible consequence of the fact denoted by proposition A and thus to agree with the speaker's opinion. It should be clear that utterances of the type found in (13a) are mostly used in argumentations. In examples of the type (13b) the position of the clauses is generally the reverse: B because A. Presumably van Dijk has chosen the less often encountered sequence only because of the parallel to (13a), thus demonstrating the "specific property of causal implications" which involve "the implicit presence of empirical laws or law-like regularity propositions" (van Dijk 1977: 47). The usual form (13b') suggests a different argumentation: (13b') The crop has been destroyed, because it has been a very hot summer. In (13b'), proposition B (crop destroyed), the fact to be discussed or the theme to be considered, is mentioned first; speaker and hearer are convinced that it is a true fact in the real world. Proposition A (hot summer) is mentioned in second position, with because inviting the hearer to accept it as a plausible cause of B. Utterances of the type of (13b') generally declare that

Connective Relations — Connective Expressions — Connective Structures

117

a fact (proposition E] known to be true may be understood as being caused by another fact (proposition A}, likewise well-known. In this sense the causal clause with because is the answer of the speaker to the question why the fact of the main clause is a fact, why the action has taken place, why the event has occurred. Examples illustrating empirical laws may be convincing and can demonstrate the implication rule in logics, but they do not reflect the common situation of natural discourse. Here the speaker selects one cause among the bundle of imaginable causes and declares it to be the most significant cause or the most interesting in the given situation. For (13b) and (13b') we may invent a situation where farmers are discussing present difficulties and considering methods of how to avoid a bad harvest next year, since the present year has not been very successful. One of the farmers may utter (13b) in order to remind his colleagues that sometimes a bad harvest may be caused by events that are beyond the control of our human capabilities. 3.5. The aim of message

The differences between the relation of consequence/result and that of cause/ reason cannot be entirely explained by the fact that the theme/topic of one relation may in several cases be mentioned as rheme/comment in the other relation. As I have pointed out in Rudolph (1982: 200—202), the transformations of consequence/result into cause/reason and vice versa are based on the different position that a speaker has adopted towards the causal connection; for reasons of preciseness, I have called this the "causal constant". The causal constant is a name for the formula A —» B, which says that proposition Λ is the cause/reason/condition/supposition/base for proposition B, which can be described as a consequence/result/effect. As underlying semantic pattern, the "causal constant" cannot be expressed by a surface connective expression unless it is completed by the "speaker variable", which marks the speaker's point of view and opinion towards the causal connection. The causal constant A —> B remains untouched whether the speaker is of the opinion that the states of affairs are represented in a result relation or in a causal relation. As for the causal relation, the speaker's interest is focussed on proposition B, when B reflects the main information and is presented in the main clause, generally in first position. From proposition B the speaker mentally looks back to proposition A, which denotes the fact the speaker considers to be the cause or reason and which serves as explanation, explication, apology and which gives the reason why the action or event denoted in proposition B has been done or has occurred. During the whole procedure proposition B continues to be the aim of the message, its importance being stressed by the explanation/explication given in proposition A.

118

E. Rudolph

As for the relation of consequence/result, the speaker's interest is focussed on proposition A, which is presented in the main clause, always in the first position of the sentence. From proposition A the speaker looks forward to proposition B and finds there the consequence or result of the fact denoted by proposition A. In an entire text (as opposed to isolated examples), the speaker's interest can be perceived in the continuation of the text: after a causal relation he will continue the theme of proposition B; after a consecutive relation he will continue the theme of proposition A. The text of (13a) could be continued as in (14a) We have to be aware that the coming winter will be very hard; this is almost always the case after a hot summer. whereas the text of (13b) would continue as in (14b) And now we have to buy our corn from abroad, and certainly the prices will increase. In (14a) the theme of proposition A (hot summer) is continued and involves the future conditions of the weather. In (14b) the theme of proposition B (bad harvest) continues to be discussed; after the cause has been stated, the prospects for the future are mentioned. To summarize: — The relation of consequence/result and the relation of cause/reason have the same "causal constant" A -> B. — The causal constant is an underlying semantic pattern that cannot be isolated from the "speaker variable". — The speaker variable is different for each of the four specific relations found in the set of causal connections. — The differences of meaning are condensed in the aim of the message, which is decisive in the continuation of a text. To conclude this section, (15) may give an idea of the complexity of connection when embedded in a text. (15) 1. When the Christian Democrats take over, by whatever means, they will probably find themselves trying to patch up the old consensus rather than 2. striking out in bold directions. At home they will not be able to accept the whole of Count Lambsdorffs rigorous recipe for recovery because it would alienate too many people and go against the grain of the Christian Democrats' 3. own social conscience. Abroad they will be less openly critical of Washington and more openly critical of the Soviet Union. (The Times).

The connection of causality is realized in (15) as a relation of reason with the connective expression because. Obviously the aim of the message is not what has been uttered in the because-clause., although it is composed of two pieces of information. The author wants to stress another point, which is

Connective Relations — Connective Expressions — Connective Structures

119

made clear by the context. In 1. the author gives the theme of this small passage, the probability of future acts of a new government; in 2. he introduces the first aspect of his theme ("at home they will"); and in 3. he continues with another aspect ("abroad they will"). In relation to the primary aim of the message, the because-clause presents information which is only of secondary significance: it gives a forecast of possible states of affairs. Given this interrelationship, it would not be possible to transform the relation of cause into a relation of result.

3.6. Result and purpose with the same connective

The relations of result and purpose are characterized by a semantic proximity in so far as in both the speaker first draws the hearer's attention to proposition A, after which he moves on to proposition B, which denotes a state of affairs generally succeeding the state of affairs of proposition A in time. The difference between the two relations is reflected in the viewpoint towards proposition B. The relation of result generally presents two states of affairs that are situated in the real world. The relation of purpose, however, deals with two states of affairs that do not belong to the same world, proposition B always being located in another possible world than proposition A: That is why the verb construction with would or should in the clause of purpose is called 'putative'. In grammars clauses of result and clauses of purpose are generally treated together or at least back-to-back, thus pointing to their semantic proximity. Cf. statements such as "Clauses of result overlap with those of purpose both in meaning and in form" (Quirk et al., 1972: 754) or "The intended result or 'purpose' of an action is described by an adverbial of purpose" (Leech and Svartvik, 1975: 94) and "So that ... serves to express result as well as purpose" (Quirk et al., 1972: 754). But the relationship is not explicated nor is it demonstrated by transforming the same example. Such caution on the part of English grammars probably stems from the very low frequency of purpose clauses introduced by so that, as opposed to the infinitive clauses that are more commonly used for denoting purpose. The to-infinitive-clause is used even if it does not have the same agent as the main clause. This case is demonstrated by Quirk and Greenbaum (1974: 328): (16) They left the door open in order for me to hear the baby. In other languages such a distinction is made with the infinitive clause being used in the case of one agent and the finite verb clause in the case of two agents. This is especially so in languages with parallel patterns for the two cases, such as the Romance languages, as may be shown in the following table in (17):

120

E. Rudolph

(17) Connective expressions for purpose clauses: with infinitive with finite verb French pour, afin de pour que, afin que Spanish para, a fin de para qm, a fin de que Italian affine di affinche Nevertheless, the different surface realizations in various languages do not affect the semantic value of purpose as a close neighbour of result. Thus, the babysitter example, (16), being the usual formulation, can be transformed in order to show the various possibilities: (16) (a) They left the door open so that I should hear the baby (b) They left the door open so that I heard the baby In (16a) and (16b) the purpose clause and the result clause are introduced by the same connective so that and follow the same main clause based on proposition A. Purpose and result clause refer to the same proposition B, but they differ in meaning. In (16) and (16a) the speaker says that the parents left the door open intending that the babysitter hear the baby. In (16b) the speaker merely states the two facts, the second being the result of the first.

4. Connective structures Manifestations of connective relations which represent complex configurations of states of affairs generally appear as complex utterances in the form of compound sentences. As such they are often described in those sections of handbooks and grammars dealing with sentence structure or syntax. To a certain degree their position in the system of language is ambivalent: their application and functions are part of the basic knowledge of the speech community, but they do not occur in all speech acts, nor are they necessary components of texts and discourses. Nevertheless they do not occur in isolated position, but only in texts and discourses. The difference between connective relations represented in texts and connective relations in isolated examples can be seen by comparing (12) and (18) with context-free examples such as (10) and (11). 4.1. Interdependency The use of connective expressions chosen from the list presented in 3.1. depends on many factors, especially on the speaker's intentions. The speaker can organize his utterance without any connective relation, and if he has decided to use a manifestation of a connective relation he is again free in his choice of connective expressions.

Connective Relations — Connective Expressions — Connective Structures

121

Speaker intentions vary greatly and they cannot be described with simple patterns. But it can be assumed that intentions are dominated by the wish to be understood by the hearer. The first condition for successful interaction in discourse is that the speaker organize his speech act so that it is understandable. Understanding an utterance is closely related to the knowledge of the world which the participants in the communication have in common. A hearer cannot understand the subject being discussed if it is entirely unknown to him. Nor can he understand an utterance if he does not know the exact meaning of the words used by the speaker. Normal communication is located between these two extremes: the hearer knows something of the topic, but the speaker knows more about it and wishes to inform the hearer. In cases in which the hearer has little knowledge of the subject or in which the subject is difficult, the speaker can facilitate the hearer's understanding by structuring his utterance or sequence of utterances. There are several ways of structuring a text: summary and repetition, questioning and answering, the alternation of report and consideration, narrative and reflection, and so on. Connective relations do not belong to any of these types in particular, but hold a special position within the means of structuring a text, since they convey the speaker's opinion on the reported states of affairs together with the information. Certainly, the speaker is always present when speaking. But the hearer can forget the speaker's existence when listening to reports of actions and events or descriptions of states of affairs. This is the well-known case of news announcers on radio and television. The hearer is generally not interested in the announcer, but only in the news. Whereas news texts are generally written without connectives, commentaries are. Commentaries are helpful in the understanding of political and economical developments, but also exercise an influence on the hearer. In order to demonstrate the different levels in the use of connective relations I shall analyse an entire editorial of The Times as an example of a text and then a relatively short passage of the Acts of the Apostles as an example of a discourse. 4.2. Analysis of an editorial of The Times

The entire editorial, as can be seen in the Appendix, is composed of eight paragraphs of slightly different length; only the first paragraph is shorter. The text is a commentary on the current political situation in West Germany in September, 1982. Description and evaluation are mixed; in some paragraphs descriptions prevail, in others evaluations. Each paragraph can be characterized as a portion of the text, and my analysis therefore treats each paragraph separately. It can be shown that the single parts of the text each have an internal structure, but are also linked with one another.

122

E. Rudolph

Synopsis of paragraphs, connections and interpretation I. so ... that question or question

emphatic assertion, reminding the reader of a political development alternative questions opening the deliberations

Π.

enumeration of facts, coherence being guaranteed by the anaphoric value of the nouns task and consensus with the alternation of the subthemes internally-externally

III. when with + noun-phrases but largely by + -ing but also because of + noun

interpretation of a historical fact; at the same time description of actions that have followed in time enumeration of conditions for an expected development unexpected state of affairs explanation of this state of affairs which also recalls the political situation of some years ago; the two reasons are bound together by the evaluation of the degree

IV. now not because \ but because ι therefore

nor conditional with inversion

present situation explained with rather subjective suppositions which nevertheless lead to a consequence: describing the actual facts considerations on decisive motives having a thematic coherence contrary to fact conditional revealing the author's contrary opinion

V. but

although even if one is ... but another ... but another ... but another ... but

links V. to the previous parts of text by pointing to another level of explanation introduces a larger argumentative consideration on assumptions concession at a lower level: subordination of sense expressed by subordination in text structure four assumptions are quoted and rectified with facts introduced by but

VI.

coherence achieved by the anaphorical use of the nouns strains and consensus; enumeration of informative facts bound together by repetition of there is/are VII.

because hence

coherence achieved by summing up the context (in such circumstances), which leads to a normal consequence followed by an explanative assertion (linked by this) introduces the reason for the explanation the consequence of the explanation is itself an explanation

Connective Relations — Connective Expressions — Connective Structures when however not because \ but simply because } VIII. jet when

because but if however

4.3. Different

123

combination of temporal and conditional use the author's personal opinion is presented as a new attempt at describing the actual state of affairs contrastive arguments for the explanative statement of the author's view of the political background binds VIII. to VII. giving an evaluation which is in contrast to VII. marks time and condition for prospective actions being mentioned as arguments in favour of the evaluations; the following subarguments are coordinated by at home — abroad introduces a prospective explanation which is subordinated and merely supports one of the sub-arguments signal for a new start contrasting with the preceding arguments gives the condition for the subargument supporting the contrasting statement of the preceding ^»/-sentence stresses the sceptical open-end of the paragraph in which the future tense prevails (expressing hopes and fears); it is at the same time the end of the article.

levels of connection

In the editorial, different levels of connection can be distinguished which correspond to different levels of communication. Connective relations are situated on the level of reporting as well as on the levels of interchanging opinions and of argumentations and meta-communication. At the beginning, the relation of result in its emphatic form so ... that lifts a well-known but almost forgotten state of affairs into the rank of a cause with its effect by means of the description of a new, noticeable but unpredictable situation. Each of the following seven paragraphs begins with a new thought which is bound to the already existing text by a meta-communicative sentence (§ 2), by prepositional phrases (§§ 3, 6, 7), and by connective expressions (§§ 4, 5, 8). At the highest level of textual connexity, where two complex parts of text are combined, prepositional phrases and connective expressions fulfill the same task. Like hinges, they join the two parts of text without touching their independence. The difference between phrases like under these strains (§ 6) and in such circumstances (§ 7) on the one hand and but (§ 5) and jet (§ 8) on the other hand is one of semantic preciseness. The phrases, though quite general in their sense, are rather narrow when compared with the connectives. Only the above-mentioned strains and circumstances can act as conditions for the following passage, whereas but and jet mark a contrast in whatever circumstances. Considering these eight paragraphs, it can be remarked that the coherence of the text is given together with textual information. This is also true for § 3 and § 4; however, cohesion is not directly expressed in § 3,

124

E. Rudolph

and while now (§ 4) can be considered an absolute expression of time, it must not be considered a connective expression. Another level of connection is that of argumentation, which is well represented in § 7 and § 8. A very complex procedure, argumentation cannot be described in just a few words. Causal and conditional relations are used in argumentation, the subordinate clause expressing the argument: not because — but because (§ 7); if (§ 8). The consequence of the arguments is presented as a relation of result: hence (§ 7). The narrowness of connective expressions demonstrates the interdependency in all of paragraph 7: explanations of facts ("because the constitution") are mixed with suppositions ("when change comes") and personal opinions ("however"), and the final contrastive arguments are unproved statements receiving their persuasive force from a high degree of plausibility. In § 8 jet, but and however are situated at a higher level than when, because and if, which contribute to the inner preciseness of the affirmations. The description of facts ("now they have another chance") can be easily connected with personal evaluations and opinions ("not because ... but because ..."), and finally lead to another fact ("have therefore decided to jump ship"), as is demonstrated in the first complex sentence of § 4. The connective expressions usually occur in sentences with positive truth value; they confirm the impression of plausibility. The battery of buts in § 5 proves this interpretation: each ^«/-»ow-sentence corresponds to a true, real state of affairs. The state of affairs as such is not interesting, but rather its combination with the preceding assumptions. The speaker can hide behind the facts, and nevertheless his opinion is revealed by the connective relation, as is manifested in the last sentence of § 4, where the author obviously means that the Free Democrats did not wish to keep the government going. As a final remark to the editorial it should be pointed out that two of the eight paragraphs do not have a connective relation: § 2 is a report of historical facts and § 6 reflects a then current state of affairs. This corresponds very well to the observation that connective relations do not occur as often in narrative texts or parts of text as in argumentative passages. 4.4. Analysis of a discourse

The short dialogue between the apostle Peter and the magician Simon, as reported in the Acts of the Apostles 8, 18—24, is a good example of compactness, which seems to be the result of writing down a discourse with special aims. Certainly, this passage was not written as a report of spoken language, though it appears in the form of direct speech. N ovum Testamentum Latine: Actus Apostolorum 8, 18—24 18. Cum vidisset autem Simon quia per impositionem manus Apostolorum daretur Spiritus sanctus, obtulit eis pecuniam,

Connective Relations — Connective Expressions — Connective Structures

125

19. dicens: Date et mihi hanc potestatem, ut cuicumque imposuero manus, accipiat Spiritum sanctum. 20. Petrus autem dixit ad eum: Pecunia tua tecum sit in perditionem: quoniam donum Dei existimasti pecunia possideri. 21. Non est tibi pars, neque sors in sermone isto. Cor enim tuum non est rectum coram Deo. 22. Poenitentiam itaque age ab hac nequitia tua: et roga Deum, si forte remittatur tibi haec cogitatio cordis tui. 23. In feile enim amaritudinis, et obligatione iniquitatis video te esse, 24. Respondens autem Simon, dixit: Precamini vos pro me ad Dominum, ut nihil veniat super me horum, quae dixistis.

Good News for Modern Man: Acts of the Apostles 8, 18—24 18. Simon saw that the Spirit had been given to them when the apostles placed their hands on them. So he offered money to Peter and John, 19. and said, "Give this power to me too, so that anyone I place my hands on will receive the Holy Spirit." ' 20. But Peter answered him, "May you and your money go to hell, for thinking that you can buy God's gift with money! 21. You have no part or share in our work, because your heart is not right in God's sight. 22. Repent, then, from this evil plan of yours, and pray to the Lord that he will forgive you for thinking such a thing as this. 23. For I see that you are full of bitter envy, and are a prisoner of sin." 24. Simon said to Peter and John, "Please pray to the Lord for me, so that none of these things you said will happen to me."

The quoted passage consists of seven verses and is structured in four parts. In (18) the state of affairs and context are introduced; Simon's request is found in (19); Peter's answer (20 — 23) is the main and most substantial part of the discourse and is followed by Simon's final request (answer) in (24). Simon's two short utterances have the parallel structure of a request combined with the purpose which has motivated that request. It is not unusual to formulate a request and its purpose in a final relation. Peter's answer is of greater interest: four verses, four complex utterances. He starts with a malediction (20), followed by the reason (quoniam) why the speech act of malediction has been performed. As a report of the verbal interaction, (19) and (20) could be enough: the offer is rejected. This is indirectly marked by a slight pause before the sermon of (21—23) is started. Between (20) and (21) there is no coherence expressed, (21) beginning with a statement of harsh delimitation; the apostle does not wish to have anything in common with the man to whom he addresses his sermon. The second part of (21), however, is clearly marked (enim) as explanation; it is an argument justifying the preceding delimitation. Thesis and argument of (21) are continued in (22) by the consequence (itaque), an advice with two demands that are combined by et and continued by si forte with a reconciliatory prospect. But the last verse of the sermon, (23), seems to counteract this prospect; with enim it is again presented as explanation and may be related to the entire sermon as a last argument.

126

E. Rudolph

The three autems (18, 20, 24) fulfill the task of structuring the passage; as examples of the well-known Biblical autem they indicate the turn-taking in discourse. The compactness of the apostle's contribution is intensified by quotations from the Scriptures (21 and 23); with their connotations they might have had an additional effect on the hearers. On the other hand, the compactness is alleviated by the connective expressions, because they are signals for interpretation, indicating the speaker's point of view and demonstrating his opinions. It is the speaker who relates the elements of his discourse by deciding which of them receive the rank of arguments or results. In the case of a written discourse such as this passage, this decision is not made by the speakers, but by the author of the Acts of the Apostles; and similar decisions await all translators. This is manifested in the different translations of the causal relations in this short passage: the Latin version has quoniam in (20) and enim in (21) and (23), whereas the English version has for in (20) and (23) and because in (21). Other translators may make still other decisions.

5. Concluding remarks

In this paper, connective relations and their manifestations are considered as two sides of the same coin. Connective relations as the mental representation of relations of states of affairs have their counterpart in the manifestations of connective relations which appear in the connective expressions. The separation of the two aspects results from the complex nature of this type of connection, which demands thorough explications in general as well as in detail. However, this particular goal was not persued given that this paper is part of a volume dealing with compositional principles of texts and discourses. The considerations have tried to keep in mind the importance that connective relations and expressions might have in text construction. Every manifestation of a connective relation indicates a relation between two propositions A and B expressing a new thought A-rel-B. If a speaker regards events and actions not simply as they occurred but as interrelated states of affairs, his position is characterized by a certain distance from the reported actions and events. By using a connective expression he expresses his opinion and point of view with regard to this connection. The connective expression reveals the speaker's point of view and indicates which type of connection he has chosen; at the same time it shows his opinion with regard to the different connective relations. Within a systematic survey of possible connections, connective relations comprise two of the four types of hierarchy connection. The connection of addition and the connection of time inform only about the speaker's point of view with regard to the complexity of the world, whereas the connections

Connective Relations — Connective Expressions — Connective Structures

127

of contrast and causality also report his personal opinion on how the world is organized in which the facts and events are located. The two subsets of hierarchy connection forming the connective relations — contrast and causality — have a number of subtypes: the adversative and the concessive relations are subtypes of the connection of contrast; and the relations of condition, purpose, result, and cause/reason are subtypes of the connection of causality. In the different connective relations, various connective expressions are at the speaker's disposal so that he can differentiate his opinion. The list of connective expressions (3.1) shows the variety offered by the language system. The speaker, guided by his intentions, selects one connective expression which is understood by the hearer as an instruction for a cognitive operation. Semantic interrelationships within and between types of connection can be shown through the substitution of the connective expressions available. This process demonstrates that the instructions for cognitive operations given by the speaker generally are clear enough for the hearer to understand. The hearer can learn or guess the speaker's aim of message by means of the connective expression used, his assumptions often being confirmed or reinforced in the continuation of the text. Manifestations of connective relations generally appear as complex utterances in texts and discourses, but they do not occur in all possible texts because they cannot be considered as necessary components of texts. In several sorts of texts the speaker's opinion is of little importance, especially in narrative passages. In other sorts of texts, his opinion can be helpful to the hearer in understanding interrelationships and developments in the reported states of affairs. Finally, I have analysed a text and a discourse providing examples of connective structures in order to demonstrate how and to what extent the possibilities of different manifestations of connective relations can be used by speakers and writers.

128

E. Rudolph

.§ « ο

I

.— ίφ3

C 2

Λ

«*«rt0 atnΛ -rj«3 •W

^-1

nj •M

Λ

*™^ *"

-M

^

M

11

g es υ

«

G « C Ο 'Ό C

%

0-

1=3-1 ^ g c I^

Q w

*"

(y

Ut

_

U *O

| | § >+*0 α, "α. >. g G 8 .•U"S ·.

»

»

»

»

s~\

> *

yy

yy

yy

yy

/^\

>j

>»

ω 1

ν

J

α - Γ

9J

» *

a =A Γ L a =A r

L

yy

> *

*,

r L

yy

to Α — Ω

*

J

L

A

A

α —ω

i

J Ω i J Ω = ω 1 J ω 1

A

Ο

l

Ω

Γ

Ω 1 J

A 7

= A T

1

a Γ

A

Λ

0j

A

* ω Ι

/-> —

U

5

»

Figure 4

The relations are symbolized by the letters A (anterior), P (posterior), Ο (overlapping), / (identical), C (contiguous) being specified by lower indices 3, 4, 5, 6 and upper indices + or — on the O-relation. For the analysis I will use a matrix representation for the temporal relations between all pairs of intervals in which the states of affairs are localized. The diagonal elements of the matrix are obviously /-relations and, since the matrix is symmetrical with respect to the diagonal, only the elements on the one side of the diagonal are represented. Summarizing the above discussion of temporal and aspectual relations and my proposed representation, I could say that the latter is essentially topological, attempting to localize intervals of related states of affairs on the time axis. This allows for a common representation in the form of a matrix of temporal as well as aspectual relations of a text to be analyzed.

146

K. Dorfmüller-Karpusa

On the other hand, Miller and Johnson-Laird propose a representation of verbal forms as well as temporal connections based upon the calculus of predicate logic. From a phenomenological point of view Weinrich examines another function of the tenses, i.e. of the description of the states of affairs, with respect to their roles in text constitution. According to this view, the focusing roles of the tenses as a means to produce plasticity are of primary concern to him as a literary device. Although this latter point of view seems to be at variance with the consideration of tenses as describing intervals on the time axis, the combinations of perfective and imperfective descriptions also express backgroundforeground relations. These latter relations are essentially text constitutive, since they may extend beyond the limits of the sentence. Generally speaking, the relation between the point of reference and the point of event is a background-foreground one. Thus not only states of affairs expressed in the imperfective aspect function as background information but also ones in perfective aspect if their role is to convey the background information for some others also expressed in the perfective aspect (cf. text (10)). Speaking in these terms, we transpose the focus from the aspectual information to information which sometimes cannot be detected morphologically but only semantically, i.e. to the background-foreground information. In other words, we are concerned with the relation between the point of event and the point of reference, the main criterion for their distinction being their degree of prominence. Thus some states of affairs being more prominent than some others, "run vertically down through the discourse. As such they provide primary cohesion by relating parts of the discourse which are on the central threads; they also serve as a unifying principle arround which material (supportive, elaborative, explanatory) can integrate ..." (Longacre, 1983).

I also have to mention the observation of Longacre (1976) who first remarked that in classical Hebrew the prominent events in a text are distinguished by the 'waw' predicate construction. Later, Hopper (1979) related tense and aspect with background-foreground information. He pointed out also some other means which contribute to this information, e.g. word order and voice.

7, Temporal analysis of texts

In this section I shall apply some of the ideas presented above to the analysis of the temporal structure of two texts and one discourse which have been chosen so as to illustrate the different ways by which temporality and aspect contribute to the text constitution.

Temporal and Aspectual Relations as Text-Constitutive Elements

147

7.1. A text fragment of George loannou The following text fragment is the first sentence of the short story "The gum-tree" written by the Greek author George loannou. I have chosen this fragment because its structure is convenient to illustrate much of what has been described above. As a more detailed analysis of the entire text is given elsewhere, the reader can recur to the literature in order to gain an insight into the complex relations of this fragment to the macrostructure of the text (Dorfm ller-Karpusa, 1983). (10) Μολονότι είχα δημοσιεύσει ότι δε θα γιορτάσω, κάποιος φίλος θεώρησε καλό να μου στείλει πρόπερσι τη μέρα της γιορτής μου έναν φύκο. [Although I had announced (in the newspaper) that I would not celebrate my name-day, some friend considered it to be appropriate to send me two years ago on this occasion a gum-tree.] In order to segment a text into semantic units, I will not use the traditional morphologically based concepts of 'clause' and 'sentence', but rather the semantically based concepts 'communicate' and 'composition unit'. The former designates an explicit or implicit linguistic entity to which I can assign a proposition and the latter whatever linguistic entity is expressed by the author's intention between two periods. This semantically based grammatical view of a text and of its parts offers the advantage that in this way connectives as well as punctuation marks, interpreted also as connectives, obtain the I say to you* I remember (-ed) but (although)

intended activity*

I discovered

I had announced

a friend considered it to be appropriate

I wanted (I would) not to celebrate my name-day,

to send me a gumtree. Figure 5

148

K. Dorfmüller-Karpusa

status of propositions. Within this status, connectives and punctuation marks are considered as communicates, i.e. entities to which a proposition can be assigned (Petöfi, 1981; Dorfmüller-Karpusa, 1981a, 1984). Figure 5 illustrates the connective structure of this composition unit. The implicit accessibility relations to the described states of affairs as well as the performative modality have been assigned interpretatively and are indicated by an asterisk (*). Although I have not considered the temporal connectives and adverbials in this contribution, their role as well as the temporality implicated in other connectives should not be overlooked (Steube, 1980; Co Vet, 1980). Since the connective but in this figure is considered to be a communicate, i.e. a proposition with a temporality of its own, the problem of its localization on the time axis arises (Dorfmüller-Karpusa, 1983). Thus it is possible that this adversative relation is established by the author earlier than the interval of communication S, this being represented in the representation by placing the connective but under the communicate "I remember". Another possibility is that this relation is established during the interval of communication S. In the latter case the connective would have to be placed above the communicate "I remember", which then would be placed on the top of each argument of this connective relation (Dorfmüller-Karpusa, 1977, 1978). In Figure 6 the time-matrix for this composition unit is represented according to the conventions established in section 6. 8

intended activity*

1

I had announced

2

I wanted

3

Os- Os-

not to celebrate my name-day

4

A

A

OB

I

I discovered*

5

A

A

a friend considered

6

A

A

it to be appropriate

7

A

A

to send me a gum-tree

8

A

A

ol ol o5+ ol

ol ol ol ol

7

I I

I P

I

P

I

I

P

I

I

I

P: Figure 6

On the basis of this time matrix containing all the temporal and aspectual information the following remarks must be made. The identity relation between the communicates 1 and 2 amounts to assuming that the activity of

Temporal and Aspectual Relations as Text-Constitutive Elements

149

announcing is in all its phases an intended one. However, it would be unwise to postulate that indeed this interpretation reconstructs the psychological processes which actually were the case during the activity of announcing. The label 'intended activity' has a formal function in characterizing a special class of human activities. In this text all the explicitly described states of affairs are in the perfective aspect. The point of reference of the activity described in the Pluperfect tense is the state "consider" in the achievement predication "some friend considered it to be appropriate" described in Greek in the Aorist tense3. The communicate 2 described in the perfective tense Pluperfect as well as communicates 3 and 4, which are syntactically dependent upon the former, require a later point of reference on the time axis. This point of reference is complex, consisting of three explicitly described communicates, i.e. 6, 7, 8, as well as 5, the latter being assigned interpretatively. The perfective tense Aorist "θεώρησε" [considered] supports the Λ relation, which is a dominant one in the time matrix. Even the communicate 5 would be described in Modern Greek in the Aorist, i.e. "ανακάλυψα" [I discovered] being the main point of reference of 2, since the other are syntactically dependent upon it. The Ρ relation between 5 and 6—8, is explained by means of our common knowledge that the author's discovery happened after he had received the gum-tree. The perfectivity of 5 and 6 indicates that between them we have a E-R relation. The overlapping relation O~s between columns 1, 2 and row 3 can be explained as follows: The interval in which the implicitly expressed wish of the author is the case can be assumed to extend on the time axis further than any other interval. The same reason explains the overlapping relation O+5 between column 3 and row 4 as well as between columns 3, 4 and rows 5 — 8. Actually, this intended activity must be seen as the point of reference of his permanent wish. The adverbials "πρόπερσι" [two years ago] and "τη μέρα της γιορτής μου" [on my name-day] localize the former communicative unit in a measurable distance before the communication, the latter at a definite day. In principle this localization refers to 8; however, since this set of communicates is interpreted as coherent, the assignment of the same temporal specification to 4 —8 can be justified. The three diagonal blocks are formed as a consequence of the division of the composition unit into three worlds: 1, 2, the world of the intended activity, 3, 4, the world of wish, and 5 — 8, the world of discovery containing also the friend's world. The temporal structure of each of the three worlds 3

The terminology I am using pertaining to states of affairs is the following: (a) 'states', (b) 'activities' and (c) 'processes'. Whereas activities presuppose an agent, processes do not. An activity and/or a process have as a consequence the change of a state. When I mention 'events', I do this in connection with Reichenbach's proposals. Furthermore, conforming to Mourelatos' proposal I use the term 'achievement' when the predication and/or the cotext/ context allows me to conclude that the activity's goal is reached.

150

K. Dorfmüller-Karpusa

is represented by the three diagonal submatrices. The three off-diagonal submatrices represent the temporal relations between two worlds. So we can speak of "intra-world" and/or "inter-world" relations. An inter-world relation is for instance represented between the intended activity (1, 2) and the discovery (5 — 8) by the submatrix with only A relations. This matrix representation extends the concept of temporal relations between two states of affairs to relations between temporally complex configurations of states of affairs, i.e. worlds. Considering the temporal/aspectual relations of this matrix from the background-foreground dichotomy, we recognize that the foreground information given in (5) —(8) is related to the background information given in (1) —(4) by means of A\ relations. In other words, the background information consists of states of affairs ((1), (2)) anterior to and of states of affairs overlapping with ((3), (4)) the foreground information. In a simplified form, figure 6, distinguishing only between background and foreground, is represented in figure 7. 1

2

3

4

5

6

7

intended activity*

1

I had announced

2

background

I wanted

3

vs

not to celebrate my name-day

4

I discovered*

5

a friend considered

6

background

foreground

it to be appropriate

7

vs.

vs.

foreground

foreground

to send me a gum-tree

8

8

background

Figure 7

Comparing figures 6 and 7 we recognize that the relations connecting the background to the foreground which are displayed in the submatrix (1—4) (5 — 8) are anteriority (A) and overlap (O\}, with the background being more extended than the foreground on the time axis. This is expressed by means of perfect!vity vs. imperfectivity. On the other hand, the anteriority relation is based on the tense Pluperfect in relation with the tense Aorist, i.e. perfect!vity vs. perfectivity (Hopper, 1979).

Temporal and Aspectual Relations as Text-Constitutive Elements

151

7.2. A Poem by Denise Levertov In the following section I discuss the poem "To the Reader" by Denise Levertov, in which the temporal and aspectual relations play a major role in creating this composition. This poem is introductory to the collection The Jacob's Ladder and its function is to give the reader the author's perspective of her relation to the world. This perspective is intended to be a key helping the reader to disclose the meaning of Jacob's Ladder. Another clue to this role is the title The Jacob's Ladder which indicates a bridging of two worlds. (11)

To the Reader

As you read, a white bear leisurely pees, dyeing the snow saffron, and as you read, many gods lie among lianas: eyes of obsidian are watching the generations of leaves, and as you read the sea is turning its dark pages, turning its dark pages. The almost biblical monumentality of this poem, which corresponds to the title of the collection, does not only depend on the described states of affairs, but to a large degree is based upon the way these are localized on the time axis. The poet operates with large overlapping intervals described by verbs in the Present tense. It is obvious that this Present tense lacks a deictic function as far as the relation between the point of speech and the point of event is concerned. By an act which is metaphoric in a literal sense, the poet shifts her point of speech and, for that matter herself, to the point of every potential reader's activity. By this means she localizes the reader in each of the three stanzas within a quasi-omnitemporal continuum, thus achieving a quasi-omnitemporality of the activity of reading, which thus seems to represent every intellectual activity. We are concerned here with the only activity described in this poem, the other states of affairs lacking a conscious agent; they are states and processes. Based upon the temporal structure of the text, I shall attempt to give one reading of the accessibility relations of the poet, regarding the described states of affairs. This can only be done by an interpretative effort involving a large amount of world knowledge which stems from the knowledge of the ontology of objects like "gods", "generations", "dark pages", and relations like "watch", "turn", "lie".

152

K. Dorfm ller-Karpusa

.

^ 0 0

-

^ 0 0 ^ 0 0

o

^ 0 0 L

1^

oo

1 m

^

0

«< *


o >.

0 0

"*"">

"*~ΙΛ>

^H

^^^

o

^

^

^ 0 0

^ ^

K„

TTI \

K„

i—1

C S C O ^ i O v O t—

|

m l m O O

K

^ ^

TTI N

K».

K»,

l ΙΛ

Ι ΙΛ

o

"*"">

o

0 0

^ 0 0

^ 0 0 j

^ ο ο ο ^ ^ ο ο K_

^ 0 0

>^

"*"">

^ 0 0 0

^ ^

+m ^

,_

^

1 m

1 m

ο

1 m

1m

o o ^ ^ ^-i

^

^ ^

^ 0 0 ^ 0 0

^

I m l m ^ O O

w N

ι.

1 m

1 in

GO

G^

CS

CO

"^f

lO

O

r—ι

^ 0 0

§ -s J ί ! 1 -a .1 .„ | §. S^ e

c

CL, 5 5 ? ! ? ? d - , ^ * ( Λ < Λ O H *^ s ^ H ^ ^ s ^ g - a T J a a to Do Things with Words. Oxford: University Press. Bachtin, Michail 1970a Probleme* de la poetique de Dostoievski. Lausanne: L'äge de Phomme. 1970b Uceuvre de Francois Rabelais et la culture populaire au Moyen Age et sous la Renaissance. Paris: Gallimard. 1978 Esthetique et theorie du roman. Paris: Gallimard. Baldwin, Charles S. 1928 Medieval Rhetoric and Poetic. London: Mac Millan, reprinted 1959. Ballmer, Thomas T 1976 "Macrostructures" in van Dijk, T. A. (ed.): Pragmatics of Language and Literature. Amsterdam: North-Holland, 1—22. Banreti, Zoltan 1982 "The Topic of Texts and the Interpretation of Texts" in Petöfi, J. S. (ed.), 1982, 43-57. Baquero Goyanes, Mariano 1963 Perspectivismo j contraste. Madrid: Credos. 1972 Temas,formasj tonos literarios. Madrid: Prensa Espanola. 1975 Estructuras de la novela actual. Barcelona: Planeta. Third ed. Barilli, Renato 1979 Retorica. Milano: Isedi. Barthes, Roland 1970 "L'ancienne rhetorique: Aide-memoire", Communications 16, 172—223. 1974 "Introduccion al anälisis estructural de los relates", Comunicaciones 8, 9—43. French version: "Introduction ä l'analyse des recits", Communications 8 (1966), 1—27. Becker, Alton L. 1965 "A Tagmemic Approach to Paragraph Analysis", College Composition and Communication 16, 237-242. 1966 "Contribution to the 'Symposium on the Paragraph'", College Composition and Communication 17, 67' — 72. Bierwisch, Manfred 1970 "Poetics and Linguistics', in Freeman, D. D. (ed.): Linguistics and Literary Style. New York: Holt, Rinehart and Winston, 96 — 115. Black, Edwin 1965 Rhetorical Criticism: A Study in Method. Madison: University of Wisconsin Press, reprinted 1978. Booth, Wayne C. 1965 "The Revival of Rhetoric", Publications of the Modern Language Association of America 80, 8-12.

Compositional Structure: Macrostructures

207

Bornscheuer, Lothar 1977 "Zehn Thesen zur Ambivalenz der Rhetorik und zum Spannungsgefüge des ToposBegriffs', in H. F. Plett (ed.), 204-212. Bremond, Claude 1974 "La logica de los posibles narratives", Comunicaciones 8, 87—109. French version: "La logica des possibles narratifs", Communications 8, (1966), 60—77. Breuer, Dieter 1974 Einführung in die pragmatische Texttheorie. München: Fink. 1977 "Die Bedeutung der Rhetorik für die Textinterpretation', in H. F. Plett (ed.), 23-44. Carnap, Rudolph 1942 Introduction to Semantics. Cambridge, Mass.: Harvard University Press. 1955 "Meaning and Synonymy in Natural Languages", Philosophical Studies 6, 34 — 47. Christensen, Francis 1966 "Contribution to the 'Symposium on the Paragraph'", College Composition and Communication 17, 60. Christensen, Francis and Christensen, Bonniejean 1978 Notes toward a New Rhetoric. New York: Harper and Row. Second ed. Cicero, Marcus Tullius De Oratore. Edited by E. Courband and H. Bornecque. Paris: Les Belles Lettres. 3 volumes: 19676, I9604, 1971". De Bruyne, Edgar 1958 Estudios de estetica medieval. Madrid: Credos. 3 volumes. French version: Etudes d'Esthetique medievale. Brugge, 1946. van Dijk, Teun A. 1972 Some Aspects of Text Grammars: Study in Theoretical Linguistics and Poetics. The Hague: Mouton. 1976 Per una poetica generativa. Bologna: II Mulino. German version: Beiträge ^ur generativen Poetik. München: Bayerischer Schulbuchverlag, 1972. 1977a Text and Context: Explorations in the Semantics and Pragmatics of Discourse. London: Longman. 1977b "Nota sulle macrostrutture linguistiche', in Conte, M. E. (ed.): La linguistica testuale. Milano: Feltrinelli, 181 — 194. English version: "A Note on Linguistics MacroStructures" in A, P. ten Gate and P. Jordens (eds.): Linguistische Perspektiven. Tübingen: Niemeyer, 1973, 75-87. 1980a Textwissenschaft: Eine interdisziplinäre Einführung. München: dtv. 1980b Estructurasy funciones del discurso. Mexico: Siglo XXI. 1980c "Story Comprehension: An Introduction" in T. A. van Dijk (ed.), 1980, 1—21. 1982 "Introduction" in T. A. van Dijk (ed.), 1982, 1-8. van Dijk, T. A. (ed.) 1980 Story Comprehension. Special issue of Poetics 9. Amsterdam: North Holland. 1982 New Developments in Cognitive Models of Discourse Processing. Special issue of Text 2. The Hague: Mouton, van Dijk, Teun A. and Kintsch, Walter 1978 "Cognitive Psychology and Discourse: Recalling and Summarizing Stories" in W. U. Dressler (ed.): Current Trends in Text Linguistics. Berlin: de Gruyter, 61—80. Dolezel, Lubomir 1972 "From Motifemes to Motifs", Poetics 4, 55-90. Dressier, Wolfgang U. 1974 Introdu^ione alia linguistica del testo. Roma: Officina. German version: Einführung in die Textlinguistik. Tübingen: Niemeyer, 1972. Dubois, Jean et al. 1970 Rhetorique generale. Paris: Larousse.

208

A. Garcia-Berrio, T. Albaladejo Mayordomo

Faral, Edmond 1971 Les arts poetique du XX' et du XIII' siede. Paris: Champion. Fontanier, Pierre 1830 Les figures du discours. Paris: Flammarion. Ed. 1968. Forster, Edward 1958 Aspects of the Novel. London: Arnold. Franck, Dorothea 1981 "Seven Sins of Pragmatics: Theses about Speech Act Theory, Conversational Analysis, Linguistics and Rhetoric" in H. Parret et al. (eds.), 225 — 236. Garcia-Berrio, Antonio 1973 Significado actual del formalismo ruso. Barcelona: Planeta. 1975 Introduction a la Poetica clasicista: Cascales. Barcelona: Planeta. 1977 Formation de la Teoria Literaria moderna I. Madrid: Cupsa. 1978a "Lingüistica del texto y texto Hrico: La tradicion textual como contexto", Revista Espanola de Lingüistica 8, 19 — 75. 1978b "Tipologia textual de los sonetos cläsicos espanoles sobre el carpe diem", Dispositio 3, 243-293. 1979a "Text and Sentence", in J. S. Petöfi (ed.), 1979, 24-42. 1979b "A Text-Typology of the Classical Sonnets", Poetica 8, 435-458. "Lingüistica, literaridad/poeticidad (Gramätica, pragmatica, texto", 1616: Revista de la Sociedad Espanola de Literatura Generalj Comparada 2, 125 — 170. 1979d "Situacion de la teoria textual" in J. S. Petöfi and A. Garcia-Berrio, 53 — 98. 1980a Formation de la Teoria Literaria moderna II. Murcia: Universidad de Murcia. 1980b "Construccion textual en los sonetos de Lope de Vega: Tipologia del macrocomponente sintäctico", Revista de Filologta Espanola 60, 23 — 157. 1981 a "Macrocomponente textual y sistematismo tipologico: el soneto amoroso espanol de los siglos XVI y XVI y las reglas de genero", Zeitschrift für Romanische Philologie 97, 146-171. 1981b "La Poetica lingüistica y el anälisis literario de textos", Transito, h—i, 11 — 16. 1981c Semiotica textual de un discurso plastico: Enrique Brinkmann. Montpellier: Universite Paul Valery. 1982a "Definicion macroestructural de lirica amorosa de Quevedo: Un estudio de 'forma interior' en los sonetos" in Homenaje a Quevedo: Actas de la II Academia Literaria Renacentista. Salamanca. 261—293. 1982b "Problemas de la determinacion del topico textual", Anales de Literatura Espanola de la Universidad de Alicante l, 135—205. 1982c "Poetica e ideologia del discorso classico", Interse^toni 3, 501—527. 1983 "II ruolo della retorica nell'analisi/interpretazione dei testi letterari", Versus 35/36, 99-154. Genette, Gerard 1968 "La rhetorique des figures", preface to Fontanier, P. (1830). Greimas, Algirdas 1971 Semantica estructural. Madrid: Credos. French version: Semantique structurale. Paris: Larousse, 1966. Gülich, Elisabeth and Raible, Wolfgang 1977 Linguistische Textmodelle. München: Fink. Harris, Zellig S. 1974 Structural Linguistics. Chicago: University of Chicago Press. Ninth printing. Hathaway, Baxter 1962 The Age of Criticism. Westport: Greenwood. Reprinted. Heilmann, Luigi 1978 "Retorica, Neoretorica, Lingüistica" in L. Ritter Santini and E. Raimondi (eds.): Retorica e critica letteraria. Bologna: II Mulino, 9—24.

Compositional Structure: Macrostructures

209

Hendricks, William O. 1976 Semiologta del discurso literario. Madrid: Catedra. Kennedy, George 1972 The Art of Persuasion in Greece. Princeton: University Press. Kibedi Varga, Aron 1970 Rbetorique et litterature. Paris: Didier. Kieras, David E. 1982 "A Model of Reader Strategy for Abstracting Main Ideas from Simple Technical Prose", Text 2, 47-81. Klinkenberg, Jean M. 1973 "Le concept d'isotopie en semantique et en semiotique litteraire", Lefran^ais moderne 41, 285-290. 1977 "Rhetorique et specificite poetique" in H. F. Plett (ed.), 72-92. Kopperschmidt, Josef 1976 Allgemeine Rhetorik. Stuttgart: Kohlhammer. 2. Auflage. 1977 "Von der Kritik der Rhetorik zur kritischen Rhetorik" in H. F. Plett (ed.), 213-229. Lachmann, Renate 1977 "Rhetorik und kultureller Kontext" in H. F. Plett (ed.), 167-186 Lausberg, Heinrich 1960 Manual de Retorica literaria. Madrid: Gredos. 3 volumes: 1968, 1975, 1976. German version: Handbuch der literarischen Rhetorik. München: Hueber, 1960. Leoni, Federico A. and Pigliasco, M. Rosario (eds.) 1979 Retorica e science del linguaggio. Roma: Bulzoni. Leuschner, Burkhard 1972 "Grundstrukturen des 'Paragraphs': Ein Problem der Textgrammatik", Linguistische Berichte 21, 80-95. Longacre, Robert E. 1960 "String Constituent Analysis", Language 36, 63 — 88. 1968 Grammar Discovery Procedures. The Hague: Mouton. Lotman, Jurij M. 1978 La estructura del texto artistico. Madrid: Istmo. French version: La structure du texte artistique. Paris: N.R.F., 1973. Lotman, Jurij M. and Uspenskij, Boris A. 1973 Ricerche semiotiche. Torino: Einaudi. Lotman, Jurij M. y Escuela de Tartu 1979 Semiotica de la Cultura. Madrid: Catedra. Lüking, Bernd 1977 "Rhetorik und Literaturtheorie: Überlegungen zu einer interpretativen Poetik', in H. F. Plett (ed.), 45-61. Marello, Carla 1979 "Aspetti illocutori e perlocutori della Retorica" in F. A. Leoni et al. (eds.), 25 — 36. Martin, Josef 1974 Antike Rhetorik: Technik und Methode. München: Beck. Mignolo, Walter D. 1978 Elementes para una teorta del texto literario. Barcelona: Critica. Miles, Josephine 1966 "Contribution to the 'Symposium on the Paragraph'", College Composition and Communication 17, 80—82. Morpurgo-Tagliabue, Guido 1981 "Grammar, Logic and Rhetoric in a Pragmatic Perspective', in H. Parret et al. (eds.), 493-508. Morris, Charles William 1971 "Foundations of the Theory of Signs', in Ch. W. Morris: Writings on the General Theory of Signs. The Hague: Mouton, 13—71.

210

A. Garcia-Berrio, T. Albaladejo Mayordomo

Mosconi, Giuseppe 1981 a "La dimensione retorica: DalP'Arte di persuadere' alia ricerca sul parlare-comunicare e sul parlare-pensare" in G. Mosconi et al., 18—49. Mosconi, Giuseppe et al. 1981 Discorso e retorica. Torino: Einaudi. Murphy, James J. 1974 Rhetoric in the Middle Ages. Berkeley: California University Press. Parret, Herman; Sbisä, Marina and Verschueren, Jef (eds.) 1981 Possibilities and Limitations of Pragmatics. Amsterdam: Benjamins. Perelman, Chaim 1971 "The New Rhetoric" in Y. Bar-Hillel (ed.): Pragmatics of Natural Language. Dordrecht: Reidel, 145-149. Perelman, Chaim and Olbrechts-Tyteca, Lucie 1958 La nouvelle rhetorique: Tratte de ['argumentation. Paris: Presses Universitaires de France (Collection Logos). Petöfi, Jänos Sandor 1973 "Towards an Empirically Motivated Grammatical Theory of Verbal Texts" in J. S. Petöfi and H. Rieser (eds.): Studies in Text Grammar. Dordrecht: Reidel, 205 — 275. 1975 Vers une theorie partielle du texte (Papiere zur Textlinguistik 9). Hamburg: Buske. 1979a "Una teoria textual formal y semiotica como teoria integrada del lenguaje natural" in J. S. Petöfi and A. Garcia-Berrio, 127 — 145. English version: "A Formal Semiotic Text Theory as an Integrated Theory of Natural Language" in W. U. Dressler (ed.): Current Trends in Text Linguistics. Berlin: de Gruyter, 1978, 35—46. 1979b "Estructura y funcion del componente gramatical de la teoria de la estructura del texto y de la estructura del mundo" in J. S. Petöfi and A. Garcia-Berrio, 147 — 189. English version: "Structure and Function of the Grammatical Component of the Text-Structure World-Structure Theory" in F. Günthner and S. J. Schmidt (eds.): Formal Semantics and Pragmatics for Natural Languages. Dordrecht: Reidel, 1979, 303-338. 1979c "La representacion del texto y el lexico como red semintica" in J. S. Petöfi and A. Garcia-Berrio, 215—242. German version: "Textrepräsentation und Lexikon als semantische Netzwerke" in K. Heger and J. S. Petöfi (eds.): Kasustheorie, Klassifikation, semantische Interpretation (Papiere zur Textlinguistik 11). Hamburg: Buske, 341-358. Petöfi, Janos S. (ed.) 1979 Text vs. Sentence: Basic Questions of Text Linguistics (Papiere zur Textlinguistik 20, 2 vols.). Hamburg: Buske. 1982 Text vs. Sentence Continued (Papiere zur Textlinguistik 29). Hamburg: Buske. Petöfi, Jänos S. and Garcia-Berrio, A. 1979 Lingüistica del texto y critica literaria. Madrid: Comunicacion. Pike, Kenneth L. 1967 Language in Relation to a Unified Theory of the Structure of Human Behavior. The Hague: Mouton. 2nd revised edition. Plett, Heinrich F. 1977 "Die Rhetorik der Figuren: zur Systematik, Pragmatik und Ästhetik der Elocutio", in H. F. Plett (ed.), 125-166. Plett, Heinrich F. (ed.) 1977 Rhetorik: Kritische Positionen %um Stand der Forschung. München: Fink. Propp, Vladimir 1974 Morfologia del cuento. Madrid: Fundamentes. English version: Morphology of the Folktale. Bloomington: 1958 (Publication of the Indiana University Research Center in Anthropology, Folklore, and Linguistics, 10). Ramon Trives, Estanislao 1979 Aspectos de semantica lingiiistico-textual. Madrid: Istmo-Alcalä.

Compositional Structure: Macrostructures

211

1980 "Nuestro hablar: proceso pragmäticamente no exento", Monteagudo 68, 13 — 20. Rastier, Fra^ois 1972 "Systematique des isotopies" in A. J. Greimas et al.: Essais de semiotique poetique. Paris: Larousse, 80-105. Real Academia Espanola (Comision de Gramatica) 1981 Esbo^p de una nueva gramatüa de la lengua espanola. Madrid: Espasa-Calpe. 7th repr. Reiser, Brian, J.; Black, John B. 1982 "Processing and Structural Models of Comprehension", Text 2, 225 — 252. Richards, Ivor A. 1965 The Philosophy of Rhetoric. New York: Oxford University Press. Rodgers, Paul C. 1966a "A Discourse-Centered Rhetoric of the Paragraph", College Composition and Communication 17, 2—11. 1966b "Contribution to the 'Symposium on the Paragraph"', College Composition and Communication 17, 72—80. Schiaffini, Alfredo 1962 "Rivalutazione della Retorica", Zeitschrift für Romanische Philologie 78, 503-518. Searle, John R. 1969 Speech Acts. Cambridge: Cambridge University Press. 1979 Expression and Meaning: Studies in the Theory of Speech Acts. Cambridge: Cambridge University Press. Segre, Cesare 1976 Las estructuras y el tiempo. Barcelona: Planeta. Italian version: Le strutture e il tempo. Torino: Einaudi, 1976. Spillner, Bernd 1977 "Das Interesse der Linguistik an Rhetorik", in H. F. Plett (ed.), 93-108. 1979 Lingüistica j literature. Madrid: Gredos. German version: Linguistik und Literaturwissenschaft. Stuttgart: Kohlhammer, 1974. Spit2er, Leo 1974 Lingüistha e historia literaria. Madrid: Gredos. 2nd edition. English version: Linguistics and Literary History. Princeton: Princeton University Press, 1948. Todorov, Tzvetan 1971 Literaturay significacion. Barcelona: Planeta. French version: Litterature et signification. Paris: Seuil, 1971. 1974 "Las categorias del relato literario", Comunicaciones 8, 155—192. French version: "Les categories de reck litteraire", Communications 8 (1966), 125 — 152. Tomasevskij, Boris 1965 "Thematique" in T. Todorov (ed.): Theorie de la litterature des formalistes russes. Paris: Seuil, 263-307. Uspenskij, Boris A. 1973 A Poetics of Composition. Los Angeles: University of California Press. Valesio, Paolo 1980 Novantiqua: Rhetorics as a Contemporary Theory. Bloomington: Indiana University Press. Vera Lujan, Agustin 1977 ' Analisis semiologico de "Muertes de perro". Madrid: Cupsa. Weinberg, Bernard 1961 A History of the Literary Criticism in the Italian Renaissance. Chicago: The University of Chicago Press. 2 volumes. 1970 — 73 Trattati di Poetica e Retorica del '500. Bari: Laterza. 4 volumes. Wunderli, Peter 1979 "Satz, Paragraph, Text - und die Intonation" in J. S. Petöfi (ed.), 319-341.

Part II

HANS-JÜRGEN EIKMEYER

Word, Sentence, and Text Meaning

/. Preliminary remarks

Language is a continuous phenomenon, physically speaking. This is obvious for the sound waves of spoken language, but it holds true for written and printed language as well, since in these cases the language signal is represented by a continuous distribution of some color on a background. However, from the speakers' point of view when processing (i.e. producing or understanding) language, it is reasonable to assume that some discrete entity is being processed on a corresponding processing level. Consequently, one has to accept that language understanding involves processes of discretization where the hearer/reader singles out discrete elements from a continuum, and one has to accept that language production requires a reverse process yielding some continuous signal. This view of language differs from the opinion of linguists postulating the discrete character of language or taking it for granted (cf. Harris, 1968). The difference can be traced back to a fundamental difference concerning the objects to be described in linguistics. If one considers processes of language production and understanding as legitimate objects of research for linguistics, then my position can be taken. The second view seems to be adequate if the task of linguistics is limited to the description of language signals independent of real speakers or hearers, i.e. this view is typical for linguistic theories accepting the autonomy principle (cf. chapter 4). Thus, both opinions agree that language can be described by discrete entities, however, they disagree with respect to the level where discreteness is assumed: this level is an intermediate one in my opinion, but it is a basic one for proponents of the other view. If language is described as discrete — on some level at least — two questions arise: the first concerns the discrete units, the second concerns their relations and possible classifications. I will comment upon these questions in the next chapter. It is a prerequisite for the discussion of x-meanings (where denotes some unit) and principles of compositionality for x-meanings based upon the relations between units. Consequently, I will consider different approaches to semantics in the third chapter. Since every principle of compositionality for meaning depends upon the techniques for the representation of

216

H.-J. Eikmeyer

meanings for basic units and vice versa, I will have to comment upon word semantics as well as upon semantics for complex units with respect to every approach mentioned. My own view of a semantic theory will be sketched briefly in the final section. The problems I will comment upon in the following are thus mainly semantic in nature. Consequently, the question of the relevance of semantics with respect to text composition might arise. In the first instance, one can try to imagine the consequences a purely formal or syntactic approach towards text composition might have. Such an approach would consist in reformulating the hypothesis of the autonomy of syntax accepted by some representatives of the generative transformational paradigm with respect to texts. This hypothesis has been criticized already by proponents of the generative semantic approach as well as by many others, e.g. logical grammarians. If the autonomy hypothesis cannot be upheld — even as a reasonable working hypothesis — for sentences, then it cannot be applied to texts which, on a formal level, can be regarded as sequences of sentences. This point of view can be strengthened by the examination of a list of factors contributing to textuality, i.e. the quality which distinguishes texts from other sequences of sentences. Such a list can be easily compiled from the text-linguistic literature. I arrived at the following tentative list starting from Dressier (1973) and Rieser (1976): (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13)

lexical recurrences word order coreference and preforms names and descriptions time and aspect actant roles topic-comment structure connectives lexical relations between sentences contiguity relations between sentences causal relations between sentences time and place relations between sentences presuppositions.

I do not deny that formal considerations have to be taken into account with this list, but they will be dominant at best with respect to the first two factors mentioned and their relevance will continuously decrease for the rest. Moreover, when one examines the common practice in linguistic research, it seems to be obvious that semantic considerations play an important role at least during the heuristic phase of modelling. Even approaches which try to find formal indicators for textuality, coherence, cohesion etc. are pervaded in the initial phase by intuitive reasoning with respect to the meaning of expressions.

Word, Sentence, and Text Meaning

217

These points taken together clearly demonstrate that a reasonable study of text composition has to include considerations of meaning and meaning composition. A supporting argument can be inferred from the decision to favour a special class of linguistic models which I will comment on in section 4. 2. Units, categories, parts of speech

Language is not only discrete at some level but it is also regular or patterned. If we want to talk about patterns or rules within a linguistic theory, we need some entity which can be said to carry the regularity or patternedness. This entity is obviously a theoretical entity which I will call a unit (cf. Halliday, 1961). Following structuralistic practice, nothing can be said about one unit without saying something about the other units as well because a unit is characterized by the role it plays within the system of units. Halliday (1961: 251 ff.), for example gives the set of units a structure by imposing on it a taxonomy, i.e. a special kind of hierarchy, the ordering dimension of which is called rank. Rank, too, is a theoretical notion because the manifestation of this dimension may vary between languages, i.e. if a unit of higher rank 'contains' a unit of lower rank, the smaller unit is not necessarily a substring of the larger one. Halliday (1961: 252) postulates at least two units for every theory of grammar and, consequently, only two units may have a special status in the theory, the sentence and the word: There will always be one unit which, more than any other, offers itself as an item for contextual statements because it does the language work in situations: so it might as well always have the same name: 'sentence'. There will be another unit, always lower in rank, which more than any other (but again not exclusively) enters into another type of pattern and thus offers itself as an item for lexical statement; this we may as well always call the 'word'.

With respect to specific languages more detailed hierarchies of units can be established and Halliday (1961: 252 ff.) proposes the following scheme for English. It should be noted that word is a theoretical notion for Halliday and that the manifestation of the unit word in a natural language has to be determined by linguistic description. t RANK

UNITS / sentence clause group (/phrase) word \ morpheme

With respect to the upper end of Halliday's hierarchy, text linguistics has convincingly argued for a unit higher in rank than the sentence, namely the text. I do not want to enumerate all arguments raised in favour of an

218

H.-J. Eikmeyer

organizational level above the sentence because they have been sufficiently documented in the literature (cf. Petöfi and Rieser, 1973; Dressler, 1973, 1977; de Beaugrande and Dressler, 1981). However, I want to present a very condensed summary of these arguments by assigning a special status to texts in a way reminiscent of the passage of Halliday's quoted above: There will always be a unit which, more than any other, offers itself as an item for statements about verbal behaviour because it does the language work in communication: so it might as well always have the same name: "text". The vertical classification of the set of units with respect to the dimension rank is to be accompanied by a horizontal classification on each level with respect to a dimension called class by Halliday (1961). Both classifications taken together yield a system of categories, where the categories on the word level are normally called parts of speech. With respect to alternatives for Halliday's approach to the definition of categories, I want to refer to McMahon (1976: 100 ff.) who discusses four kinds of definitions for categories most commonly applied by linguists: (1) (2) (3) (4)

notional or semantic definitions inflectional definitions formal definitions distributional definitions.

McMahon observes that notional or semantic definitions presuppose a referential theory of meaning since expressions are grouped into one class if they denote the same class of entities. Inflectional definitions make use of secondary grammatical categories as, for instance, case, gender, tense, etc. Distributional definitions are based upon the occurrence of elements in environments and they can be operationalized in order to specify discovery procedures. Among formal definitions, McMahon distinguishes syntactic definitions from reductive ones. The rewrite rules of a phrase structure grammar (PSG) are formal statements of syntactic definitions since they characterize categories in terms of the roles they play in building up larger constructions. As an example for reductive syntactic definitions, McMahon refers to the definition of the iotaoperator in terms of quantifiers, junctors and equality. A third kind of syntactic definition not mentioned by McMahon is the definitional scheme of categorial grammar (CG) which specifies a finite number of basic categories and a rule for the construction of infinitely many derived categories. This construction is completely independent of natural-language units. However, CG has a general rule for the combination of expressions of certain categories yielding an expression of another category. This rule makes CG-style definitions of categories look like syntactic PSG-style definitions, but in fact they are not equivalent since the combination rule for expressions in CG is not needed for the construction of the category system itself.

Word, Sentence, and Text Meaning

219

Traditional definitions of parts of speech or categories have often been criticized by linguists because they use more than one of the aforementioned kinds of definition, i.e. they apply to morphological, syntactic and semantic criteria at the same time (cf. Lyons, 1977: 423 ff.; McMahon, 1976: 101 f.). At first glance PSG-style or CG-style definitions seem to be better alternatives, but this is simply an illusion: Even if the definitions of the categories were precise, one problem remains, namely the consistent distribution of all natural language expressions with respect to the categories. The same criticism which has been raised against traditional definitions of parts of speech can be raised against the methods of distributing expressions into categories since normally an intermixed bundle of semantic, syntactic and morphological criteria is used (cf. e.g. Cresswell's argumentation for treating nominals as of category walks with store. This is one change of the existing data base, the second one is caused by adding the instantiated event (29a) to the data base. By instantiation is meant that script roles are replaced by the constants they are associated with. Both changes presuppose that (28a) could somehow be linked up to the activated shopping-script. Technically this link is produced by a pattern matcher which checks whether the story line (28a) and the current event (29a) are structurally equivalent, i.e. whether they contain the same predicates at corresponding places and whether the existing bindings of the script roles do not contradict the constants. Since (28a) is the first line to be processed, the second condition is automatically fulfilled in this special case. If the match between the story-line and the current event is successful, the next event of the script — (29b) — becomes the current event. The data base is updated (changed) as described above and the next story-line will be processed. In cases where a story-line does not match the current event of the activated script, the next event is tried. When a suitable event can be found by iterated applications of this procedure, then the skipped events will also be instantiated and added to the data base. Thus, the final understanding of (28) by McSAM will be (30). (30) (a) (b) (c) (d) (e)

Jack went to a store. Jack picked up a kite. The store transferred possession of the kite to Jack. Jack transferred possession of some money to the store. Jack left the store.

Rephrasing McSAM in Bosch's terminology yields the following: The CM is made up from the script binding form (containing the individuals) and the list of instantiated events. The BK is formed by McSAM's procedures and by the scripts themselves. Linking up to the current CM is performed by the pattern matcher. Changes of the current CM are new bindings to script roles and instantiations of events. It has become obvious from the discussion of McSAM that processing whole texts requires some macro-structuring mechanism, as, for example, scripts. Such a mechanism has not been taken into account by Bosch since he illustrated his proposal with single sentences only. Petöfi's text semantical model discussed in the previous section corresponds to McSAM in some respects: the text type is the same (monological texts in both cases) and the mode of processing described is in both cases interpretation or understanding. Concerning the units of processing, however, the models differ because McSAM processes one sentence after the other and in the TeSWeST a text is interpreted en bloc. If one wants to consider models with more complications built in, the next step with respect to text type would be to design models for dialogues. The complication with respect to text type entails automatically a complication with respect to the mode of processing, since production has to be incorporated as well. One such model has

254

H.-J. Eikmeyer

been proposed under the name Procedural Grammar (PG) (cf. Eikmeyer and Rieser, 1982, 1983b; Eikmeyer, 1983). The production mode in this model is not worked out in greater detail, i.e. the model specifies only general plans for the production but does not describe their translation into single utterances. The processing units in PG are assumed to have phrase size in contrast to almost all artificial intelligence systems which work with sentences. Moreover, PG is a subjectivistic approach supplied with extra explanatory mechanisms for the functioning of communication. PG has been applied to studies of smaller parts of dialogues (approximately turn-size, cf. Eikmeyer and Rieser, 1982) as well as to classes of advisory discourse (part of this material has been published in Rieser (1983) where it is related to such fields as topics and rhetorics). In contrast to PG, the HAM-ANS artificial intelligence model (cf. v. Hahn et al., 1976, 1980; Jameson et al., 1980; Nebel and Marburger, 1982) describes the production mode in great detail. A special characteristic of this model is the emphasis placed upon building a model of the partner in a conversation. This is important since the impression someone has of his via-a-vis heavily influences both the interpretation of the utterances and the respective responses. The HAM-ANS group has studied this aspect in a model of a hotel manager. A comparison of the four models mentioned last is presented in Table 2. However, in this comparison the particular models figure more as representatives for classes of models than as individuals. Table 2

TeSWeST

McSAM

HAM-ANS

PG

processing unit

text

sentence

sentence

sentence

text type

monologue

monologue

dialogue

dialogue

mode(s) of processing

interpretation

interpretation

interpretation + production

interpretation (production)

3.10. Word, sentence, and text semantics

In this section I will try to sum up systematically the descriptions and discussions of the preceding sections. The set of approaches I will concentrate upon is formed by the whole domain of logical grammar (cf. 3.1.) and the individual proposals of Lyons (cf. 3.2.), Putnam (cf. 3.3.), Ballmer and Brennenstuhl (cf. 3.5.), Wildgen (cf. 3.6.), Petöfi (cf. 3.8.) and Bosch (cf.

Word, Sentence, and Text Meaning

255

3.9.). With respect to these approaches, I will consider two questions: The first concerns an assignment to systematic classes of semantic theories made up according to the hierarchy of units — word, sentence, text — mentioned in the title. The second question refers to a classification of approaches with respect to (1) the level in the hierarchy the model is designed for primarily and (2) the direction or orientation adopted which may be downwards (text —> sentence —> word) or upwards (word —> sentence —»· text). All approaches are or contain contributions to the field of word semantics. This should not be puzzling since any semantic theory necessarily has to say something about the meaning of atomic units. Consequently, discussions of word semantic questions took up a lot of room. Interestingly (or: because of the subjective selection of approaches), all approaches but Lyons' and the one based on logical semantics emphasize the importance of some kind of typicality (without a specifying prefix or with the prefix stereo-, proto- and arche-, respectively) for a semantic theory. The exclusion of Lyons and logical semantics, however, is not completely justified. Concerning Lyons, I suspect that his notion 'denotation' has a touch of stereotype, but I cannot support this by quotations. With respect to logical grammar, the notion of stereotype is not treated by standard approaches, but the proposal of liikmeyer and Rieser (1983) mentioned in section 3.3. above is a non-standard extension of logical grammar incorporating reconstructions of stereotypes. The holistic approach towards the structure of the verb-thesaurus (Ballmer and Brennenstuhl) and the archetypal approach of Wildgen both stress ontological dynamics in connection with word semantics. With respect to Ballmer and Brennenstuhl, this may be a consequence of the limitation to verbs, but archetypal semantics is designed to be uniformly dynamic for all main classes of content words (verbs, nouns, adjectives). Only the holistic model of Ballmer and Brennenstuhl, however, can claim to apply directly to a considerable fragment of a natural language, the others have been exemplified by few concrete instances only. Thus, these latter approaches have to assume that the empirical base can be enlarged inductively. Sentence semantics is the primary domain of logical grammars and in this field they have arrived at a high degree of formal explicitness. Many linguists are in a conflicting position with respect to logical semantics since on one hand, they highly appreciate its formal rigour but, on the other hand, they regret that logical semantics lacks appropriateness with respect to intuitions. In order to overcome this conflict, one can either supply the constructs of logical semantics with non-standard interpretations or set up an extension. In Lyons' model, only one aspect of meaning applies to the level of sentences, namely 'sense'. In contrast to logical semantics, however, he neither makes his notion of sense and sense-relations precise, nor does he offer an explicit principle for sense composition. Putnam's proposal is designed for words (more restrict!vely: content words or even natural-kind terms) and no statement concerning the sentence level is made. However, Eikmeyer and

256

H.-J. Eikmeyer

Rieser (1983a) may be regarded as an extension of Putnam's approach to sentences adopting descriptive tools developed by logical semantics. Archetypal semantics as described above does not apply directly to sentences and an extension would require formally a suitable theory of non-elementary catastrophes and substantially a general discussion as to the range of the notion 'archetype'. Holistic analysis of semantic space is limited to the study of typical cases, at least for the time being. Accordingly, it can be applied to the sentence level with this limitation only, i.e. it can deal exclusively with sentence resulting from verbs supplied with typical actants. On the other hand, it is intended to be integrated into the framework of a special class of logical grammars — punctuation-based grammars — proposed by Ballmer (1975). A punctuation grammar differs significantly from other logical grammars since it has no rules and all combinatory power is assigned to punctuation signs, i.e. to theoretical reconstructions of natural punctuation signs, sentence patterns or the like. I will briefly illustrate this kind of grammars by sentence (1) which may receive the structure (2) as its description. (1)

Peter loves Mary

(2)

NP Peter

V

NP

NP/V/NP/S

The two approaches not mentioned thus far in discussing sentence semantics are both designed primarily for texts. Petöfi's model applies to sentences regarded as atomic texts, i.e. as special instances of texts in general. In Bosch's incremental model, sentences are the processing units and all computations in connection with texts thus ultimately depend upon the processing of sentences. However, Bosch does not assign meanings to sentences, his model only describes the respective changes of the current context model. This difference between Bosch's and Petöfi's approach clearly carries over to the text level. The fundamental question to be settled is the question as to what kind of entities text meanings are. Linguists, philosophers and logicians have proposed several reconstructions of the notion 'word meaning'. However diverse these reconstructions may be, they nevertheless agree on the point that it is reasonable to conceive of such entities as word meanings. This holds true for all approaches I mentioned in this contribution, even for Bosch's since he indirectly reconstructs word meanings as stereotypes (at least in cases where the respective word contributes to changes of the current context model). Also the notion 'sentence meaning' is liberally used by scientists engaged in natural and formal language semantics and they ultimately resort to the corresponding conception elabo-

Word, Sentence, and Text Meaning

257

rated by logics. In Petofi's model, this conception is taken up and transformed in order to meet the needs of a text theory. These transformations mainly consist of two things: truth values are replaced by states of affairs and the semantic models are not assumed to be pre-existent but are constructed during the interpretation process. The notion 'text meaning' receives a theoretical reconstruction in this framework as a pair of a text signification (TSiR) and a text correlate (jTCoR). People working with incremental models (cf. Bosch, PG; cf. 3.9.) have doubts with respect to the notions 'sentence meaning' and 'text meaning' and their motivation primarily stems from the text processing perspective taken. The logical conception of sentence meaning finally rests upon the assumption that sentences can be described as timefree entities having an inherent (perhaps context-dependent) meaning. The proportions of texts and discourses, however, require different description techniques concentrating upon the effects of meanings and the information structure (information linking up to the current context model vs. information changing the current context model in Bosch's terminology) responsible for the effects. Thus, incremental models accept word meanings as semantic entities, but higher organizational levels with their own conceptions of meaning for the respective units are not needed. In discussing text semantics, I should not forget to mention the processalgebra of Ballmer and Dreher (1983) which is suitable for the holistic approach to the level of typical texts. Let me come to the question of orientation within the hierarchy 'word-sentence-text' taken by the approaches to be systematized. With the exception of Ballmer and Brennenstuhl, who try to base a complete theory of language upon word semantics and thus advocate an upward orientation, all other approaches are oriented downwards. Substantially, the downward orientation entails that semantic (re-)constructions on lower levels are shaped by the aim to integrate them into higher level meaning constructs. This becomes most obvious for logical semantics, many instances of which treat word semantics as a mere appendix, necessary to keep the composition machinery for sentence meanings running. The downward orientation from sentences to words is also adopted by Lyons concerning the sense aspect of meaning; denotations are genuinely restricted to the word level. I suspect that the downward perspective is also shared by Putnam whose approach may be regarded as contributing a solution to the problem of truth, which is traditionally a problem for the sentence level. Moreover, the orientation is indirectly mentioned by Wildgen's remarks concerning archetypal meaning composition and it is adopted by Petöfi when he formulates explications in the lexica as propositions in his sense. Incremental models as, for instance, Bosch's also have a downward perspective, but the hierarchy 'text -* sentence —» word' does not correctly describe these models as I mentioned in discussing the notion 'text meaning'. Instead, one should characterize them as oriented from texts to words via the information structure of the processing units.

258

H.-J. Eikmeyer

I have tried to summarize this discussion in Table 3. The meaning of the ' + '-signs in the upper half is that the respective approach applies to the corresponding level, in some cases comments have been added. In the lower half, the ' + '-signs mark the perspectives taken. laoie j

?f δ" rο p g' on rn·

n en

?' r-»·

Cd Cd S ft ;w —·

3

n3 n2

D P

§8 C/l ί-t

| o3 O:

(W*

π

Cd o o

er

M l-»

§.§.

word semantics

+

+

+

+

+

+

sentence semantics

+

+1

+2

+3

+4

+5

υ

text semantics sentence —» word tu υ ω Ο, V} U

n.

+

1

3 4 5

+

+

+

text —> sentence —» word

+

text —> information structure —» word word —* sentence —> text

2

+

+3

+5 + + +

+

For senses only. In its reformulation by Eikmeyer & Rieser (1983a). For typical cases only, Granted a theory of non-elementary catastrophes. Included as special cases.

4. Conclusion: properties of a semantic theory

In the previous section I outlined some approaches towards semantics and added some critical remarks. I tried to limit the discussion to internal questions of the respective approaches, but I am sure that I could not avoid subjective colouring. As already mentioned above, even the selection of the approaches is not claimed as representative. However, I do not want to leave to the reader the detective task of inferring my conception of a semantic theory from footprints left in the comments and from other circumstantial evidence. Instead, I will sketch my conception by a list of theses which I will try to relate to the approaches discussed before. I will additionally take into account properties of a semantic theory which I have not discussed in detail thus far but which I regard as important. No semantic theory I know of has all the

Word, Sentence, and Text Meaning

259

properties to be mentioned in the list, thus it may be regarded as a programme for a future semantic theory. Some theses will be formulated more generally for a linguistic theory and, in these cases, some consequences for the semantic component will be added. 1) Against autonomous linguistics This is the title of an article by Derwing (1980) who convincingly argues that the basic assumptions of linguistics have not changed fundamentally since de Saussure. The 'revolutions' which took place in linguistics since that time have primarily been terminological revolutions but not substantial ones. Great parts of linguistics accept the autonomy principle, be it called the language system or the competence of the ideal speaker. According to this principle, language is viewed "as a kind of supraindividualistic 'sociological reality', which supposedly maintained a kind of 'abstract existence' of its own, independent of its speakers" (Derwing, 1980: 165; my italics). From the autonomy principle, one can easily infer an explanation for the functioning of communication because one can say that every real speaker takes part in the system of his language. I do not doubt that communication works and I do not think this is the case just by chance. However, communication is to be explained from the fact that every speaker is an individual on the one hand and a member of a speech community on the other hand. Autonomous linguistics does more than give priority to the second aspect since it reifies theoretical linguistic constructs. For me the reorientation advocated by Derwing is a subjectivistic, psychological move. Correspondingly, linguistics should carefully value the approximative character of its models and try to make them overlap with psycholinguistic evidence. With respect to semantics, the psychological move entails that semantics become an integral part of a theory of language understanding and production. Logical semantics, as an instance of autonomous linguistics, postulates, for instance, that semantic models are independent of texts and their producer(s) and/or receiver(s). Moreover, these semantic models contain subtle set-theoretic constructions (e.g. possible worlds, hierarchies of semantic types) which are difficult to justify psychologically. A non-autonomous semantics could no longer use semantic models of this kind but would instead regard semantic models as representations of a semantic memory. The semantic memory would have restricted capacity, it would be built up in the course of understanding and it would be scanned in the course of language production. The entities and relations contained in the semantic memory should also have some psychological reality and, as long as we do not know very much about them, it would be methodologically reasonable not to use too intricate constructions. 2) The second dimension of language A strictly subjective linguistic theory could say a lot about individual speakers and it could generalize its statements to a certain degree by statistical methods.

260

H.-J. Eikmeyer

However, the statistically created average speaker is no longer real, he is a theoretical artifact. Stopping at this point would entail giving up the possibility of formulating generalizations and, consequently, a conception is needed to allow for statements about groups of speakers, i.e. speech communities. Roughly speaking, such a conception would make three assumptions: (a) All speakers are equipped with the same "hardware" and "initial software" when they are born; this is genetically coded, (b) All speakers of a speech community are subjected to roughly the same stimuli during language acquisition, (c) The language behavior of a speech community is regulated by some mechanism as, for instance, conventions. These three assumptions form the minimal social base for the subjective position; both taken together allow for a non-autonomous linguistic theory. I commented upon the social dimension of meaning in connection with Putnam's proposal (cf. 3.3.) which assumes that word meanings are socially fixed and that individual speakers only possess parts of the whole meaning. However, conversational analysis has demonstrated that there is more to be observed than the distribution of meaning components within a speech community. In analysing real discourses "one gets confronted with the meanings actually constructed or realized by the participants of a conversation" (Kindt, 1981: 503). These phenomena can be subsumed under the heading meaning negotiation. According to Kindt (1981), a semantic theory should treat real meanings (which would have to be detected by methods still to be developed) in specific situations and study the processes of negotiation. 3) The context-dependence of meaning This is a phenomenon well known from discussions about semantics. Meaning negotiation is one of the mechanisms which is responsible for indexicality and which is actively applied by the speakers involved in a conversation. Logical semantics as in the instance of autonomous linguistics, in contrast, studies context-dependence in a far more passive way (cf. e.g. Cresswell, 1973) by incorporating contexts as theoretical reconstruction of situations into the semantic model. Only a few approaches stress the dynamic aspect of context-change mechanisms (cf. Eikmeyer and Rieser, 1983a; Ballmer, 1978), the subsequent application of which leads to the possibility of context-dependence or indexicality. Thus, context-dependence and context-change mechanisms form part of a necessary pragmatic component of a feasible semantic theory since otherwise the enormous flexibility of natural language cannot be explained. 4) The vagueness of natural language Strongly related to context-dependence is the phenomenon of vagueness. It is by now a commonplace among semanticists that vagueness is a characteristic property of natural language which must not be regarded as a defect and which consequently must be accounted for by a semantic theory. The formal

Word, Sentence, and Text Meaning

261

systems for the descriptions of vagueness proposed thus far fall into three classes: supervaluation approaches, many-valued logics (including fuzzy logics) and threshold logics (cf. Ballmer, and Pinkal, 1983, for a collection of approaches). However, as is the case with context dependence, approaches are lacking which treat the dynamics of vagueness, i.e. which account for mechanisms applied by speakers to remove, to maintain or even to produce vagueness. 5) Stereotypy It has become obvious from many discussions in semantics that the treatment of meanings by necessary and sufficient properties along the lines of mathematical definitions is not well suited for natural languages. Achinstein (1968) for instance, proposes instead the notions 'relevant' and 'non-relevant properties', where the relevant properties can be subclassified into semantically and non-semantically relevant properties. Additionally, semantically relevant properties may be arranged on a scale according to their centrality. This description of Achinstein cannot be mapped onto Putnam's normal form for word meanings, however, the intuitions behind both approaches seem to be very similar. Putnam's proposal has been very inspiring for many linguists who haven taken up his ideas and integrated them into different frameworks. I think that a modified notion of stereotype (cf. 3.3.) can replace such concepts as intensions or senses (cf. thesis 9). 6) Dynamic semantics In formulating the last theses, I anticipated special formulations of the present one when I emphasized the dynamics underlying negotiation, contextdependence and vagueness. As I mentioned above, a dynamic semantics has to include both the ontological dynamics touched upon in sections 3.5. and 3.6. and the procedural dynamics of text processing discussed in section 3.8. and 3.9. Among procedural models, I would prefer incremental models, more specifically, incremental models with processing units smaller than sentences, i.e. of phrase size. This is only an informal characterization of the size because I do not want it to imply a preliminary decision with respect to the syntactic level the notion 'phrase' clearly refers to. 'Phrase size' could mean that the processing unit contains, let me say, up to four words approximately. However, this second definition is still problematic since it presupposes the notion 'word'. Adopting the view that the determination of processing units is a discretization process performed upon the continuous language signal requires that the signal contain some cues with respect to which the placement of unit boundaries becomes possible. These cues depend upon the medium the signal is transferred by (visual vs. acoustic) but they will be processed in connection with anticipations stemming from higher levels as e.g. syntax or semantics.

262

H.-J. Eikmeyer

The decision upon an incremental model has several implications with respect to the semantic component. Among the most prominent are implications with respect to meaning composition (cf. (8)) and the set of objects contained in the semantic memory (cf. (9)). 7) Formal rigidity The property aimed at by this thesis is one a semantic theory should inherit from logical semantics. Existing systems of this kind have been rejected in previous theses as autonomous systems (at least in their standard interpretation), however, a semantic theory should not fall back under the standards of explicitness arrived at by them. Hereby I implicitly assume that is is possible to design a model which is both formally rigid and psychologically real. I am optimistic with respect to this assumption since formal models of vagueness have shown that one can handle vague and fuzzy objects in a precise manner nevertheless. Concerning procedural models, I will add that the semantics of programming languages has paved the way for a formally satisfactory description of procedures (cf. Eikmeyer, 1983). 8) Compositionality of meaning The strongest version of a principle for meaning composition is Frege's principle which says that the meaning of a complex expression is a function of (a) the meanings of its constituent expressions and (b) the syntactic construction. According to thesis (3), contexts should be the third parameter entered into such a function. For natural languages, this version of Frege's principle still seems to be too strict. It should be replaced by a whole set of meaning composition mechanisms (cf. Kindt, 1981) which need not rest upon total functions alone, but which may use partial functions or, in order to remove determinism, even relations. In procedural semantic models meaning composition will be performed by special procedures computing the values of functions/relations. In incremental models, the computation of complex meanings is additionally complicated because all information is processed stepwise, one processing unit after the other. It may then be the case that the information necessary to compute a more complex meaning is contained in two or more subsequent processing units. In such a situation, two strategies have to be provided for: either the computation of the complex meaning is suspended until sufficient information is available or missing information has to be supplied from, for example the semantic memory by anticipations (i.e. inferences) or as defaults. Such problems are quite frequent for incremental models with processing units of phrase size, but they may also occur in models working with sentences, e.g. when cataphoric proforms have to be handled. In incremental models, meaning composition differs in another respect from meaning composition in static models. The reason is that two kinds of information have to be distinguished, namely identification information linking

Word, Sentence, and Text Meaning

263

up to the actual state of the semantic memory, and modification information changing this state (cf. 3.9.). Meaning composition in a strict sense takes place only with respect to the modification information while the identification information is essentially used by search procedures serving the purpose of pointing out objects to which the new information will be attached. Consequently, from this perspective meaning composition denotes the (incremental) process of constructing a state of the semantic memory which is assumed to be compatible with the information contained in the text. Thus, texts do not have inherent meanings per se but interpretations will be assigned to a text. If one wants to call some entity 'text meaning', then one should apply this notion to the differences between the state of the semantic memory before the text has been received and the state arrived at after the text has been processed. The linguistic theory I am aiming at is a theory for the treatment of texts and I regard meaning composition as the central aspect of text composition. The intention of a text producer is to convey some information he has coded into a specific form according to his partner model. The task of a receiver is to decipher this form. The final aim of this task is to extract the meaning from the text which obviously requires the analysis of the form. However, this formal analysis is only a prerequisite for meaning analysis and it thus has no justification on its own apart from semantic interpretation. The subordination of form to meaning has to be modelled by a subjective procedural model for text understanding, analogous arguments can be formulated for text production and production models. A decision for such a model thus stresses again the relevance of a semantic theory for questions of that kind. 9) Sense and reference I will follow common practice in logic, philosophy of language and semantics and distinguish two aspects of meaning, sense and reference. The study of sense consists in the description of semantic relations holding between naturallanguage expressions, the study of reference examines the relation of naturallanguage expressions to the 'world'. I do not think it is necessary to distinguish a third aspect such as Lyons' denotation (cf. 3.2.). This follows from theses (1) and (2) since a structuralistic lexicon is a theoretical construct which has no place within a non-autonomous model. I assume that sense determines reference as logical semantics does (cf. 3.1.), but the determination may be mediated by quite complex process including feedback from reference (or better non-reference) to senses. The 'world' in the semantic model I advocate can be compared with the semantic memory, i.e. the set of objects or individuals known together with the network of their descriptions or properties. However, the semantic memory is subject to change in the course of interpretation, since new objects or new properties will have to be added into the memory, others will be reconsidered and still

264

H.-J. Eikmeyer

others will be forgotten. (I suspect that forgetting is the most difficult phenomenon to describe. The introduction of objects into discourse worlds has been described by several approaches (cf. e.g. Chastain, 1975; Ballmer, 1972), its inverse still lacks proposals as far as I know). In a procedurally dynamic model, establishing the reference of an expression means to find a suitable object in the semantic memory. Procedures have to be designed for this aim and some way of telling whether certain objects are suitable or not is also needed. In filtering out (preferably exactly one) object from the semantic memory, senses play an important role since they are descriptions which have to match the description an object already has in the semantic memory. Things may turn out to be complicated if no suitable object can be found or if several objects are possible referents. In the first case, the search domain has to be enlarged (it is normally not the whole semantic memory but only a prominent subdomain) or the description (= sense) has to be weakened, fuzzified, bleached, etc. In the second case, either the search domain can be reduced in size or the description can be strengthened, made more precise, refreshed etc. From this informal description of search procedures, one can infer further specifications for reconstructions of senses. Since they serve the purpose of descriptions for objects, it is a reasonable assumption to use Putnam's stereotypes, i.e. theories about typical objects. However, one is not forced to accept the complete normal form proposed by Putnam, especially not the extension component (cf. 3.3.). Since this component guarantees rigidity, its incorporation is essentially the effect of an autonomous view of language. Rigidity is not suitable in combination with an incomplete, time-dependent semantic memory, the weaker assumption of sufficient similarity sketched in thesis (2) alone is adequate. Stereotypes can be reconstructed atomistically as bundles of stereotypical properties or holistically as indivisible entities (cf. 3.6.). Both reconstructions are necessary because they may be useful for the explanation of differences in processing speed: In a case where one has to determine whether an object of the search domain matches a description one will first try to make a decision holistically since I suspect such decisions to have a higher processing speed. Later on, if a holistic decision cannot be made, one will use an atomistic description, the components of which will be tested sequentially. Moreover, it should be kept in mind, that holistic reconstructions of word meaning allow for inferences differing substantially from atomistically mediated ones. The need for weakening and strenghtening descriptions requires a systematic study of sense-relations. By a systematic study, I mean an approach along the lines of holistic meaning analysis (cf. 3.5.). However, the relations used in this concrete approach (similarity and presupposition) are only few in number. Even if they turn out to be fundamental ones, one should examine them with respect to the wide range of sense-relations as proposed by Lyons, for instance (cf. 3.2).

Word, Sentence, and Text Meaning

265

With these nine theses I have roughly sketched the main characteristics of my conception of a semantic theory, however, the sketch is far from being a detailed description. Many assumptions have simply been stated and have not been defended sufficiently by arguments. Thus, this chapter is quite programmatic in nature. I hope to have demonstrated that this programme rests upon proposals and results from various sources which certainly will not be easy to integrate.

References Achinstein, Peter 1968 Concepts of Science: A Philosophical Analysis. Baltimore: Hopkins. Ballmer, Thomas T 1974 "Einführung und Kontrolle von Diskurswelten", in Rüttenauer, M. (ed.): Textlinguistik und Pragmatik. Hamburg: Buske, 185 — 225. 1975 Sprachrekonstruktionssy steme. Kronberg: Scriptor. 1978 Logical Grammar. Amsterdam: North-Holland. 1982 Biological Foundations of Linguistic Communication. Amsterdam: Benjamins. Ballmer, Thomas T and Brennenstuhl, Waltraud 1980 Speech Act Classification. Berlin: Springer. 1981a "Lexical Analysis and Language Theory", in Eikmeyer, H.-J. and Rieser, H. (eds.), 414_461. 1981b "An Empirical Approach to Frame Theory: Verb Thesaurus Organization", in Eikmeyer, H.-J. and Rieser, H. (eds.), 297-319. Ballmer, Thomas T and Dreher, Christian 1983 "Der Semantische Raum: Linguistische Analysen zur Prozeßkombinatorik", Universität Bochum: Mimeo. Ballmer, Thomas T and Pinkai, Manfred (eds.) 1983 Approaching Vagueness. Amsterdam: North-Holland. Barri, Nimrod 1977 "Giving up Word Formation in Structural Linguistics", Folia Linguistica 11, 13 — 38. de Beaugrande, Robert- , and Dressler, Wolfgang U. 1981 Introduction to Text Linguistics. London: Longman. Boolos, George S. and Jeffrey, Richard C. 1974 Computability and Logic. Cambridge: Cambridge University Press. Bosch, Peter 1983 Agreement and Anaphora: A Study of the Role of Pronouns in Syntax and Discourse. London: Academic Press. Chastain, Charles 1975 "Reference and Context", in Gunderson, K. (ed.): Language, Mind and Knowledge. Minneapolis: University of Minnesota Press, 194—269. Cohen, Antonie and Nooteboom, Sibout G. 1975 "Tentative Conclusions", in Cohen, A. and Nooteboom, S. G. (eds.), 325 — 353. Cohen, Antonie and Nooteboom, Sibout G. (eds.) 1975 Structure and Process in Speech Perception. Berlin: Springer. Cole, Ronald and Jakimik, Jola 1980 "A Model of Speech Perception", in Cole, R. (ed.): Perception and Production of Fluent Speech. Hillsdale: Erlbaum, 133-164. Cresswell, Max 1973 logics and Languages. London: Methuen.

266

H.-J. Eikmeyer

Derwing, Bruce L. 1980 "Against Autonomous Linguistics", in Perry, T. A. (ed.): Evidence and Argumentation in Linguistics. Berlin: de Gruyter, 163 — 189. van Dijk, Teun A. 1972 Some Aspects of Text Grammars. Dordrecht: Reidel. Dressier, Wolfgang U. 1973 Einführung in die Textlinguistik. Tübingen: Niemeyer. Dressler, Wolfgang U. (ed.) 1977 Current Trends in Text Linguistics (= Research in Text Theory 2). Berlin: de Gruyter. Eikmeyer, Hans-Jürgen 1983 "Procedural Analysis of Discourse", Text 3,11-38. Eikmeyer, Hans-Jürgen and Rieser, Hannes (eds.) 1981 Words, Worlds, and Contexts: New Approaches in Word Semantics. Berlin: de Gruyter. Eikmeyer, Hans-Jürgen and Rieser, Hannes 1982 "Prozedurale Analyse", University of Bielefeld: Mimeo. 1983a "A Formal Theory of Context-Dependence and Context Change", in Ballmer, T. T and Pinkal, M. (eds.): Approaching Vagueness. Amsterdam: North-Holland, 131—188. 1983b "Procedural Grammar for a Fragment of Black English Discourse", in Ballmer, T. T (ed.): Linguistic Dynamics. Berlin: de Gruyter, 85 — 178. von Hahn, Walther; Henskes, Dieter; Hoeppner, Wolfgang and Wahlster, Wolfgang 1980 "The Anatomy of the Natural Language Dialogue System HAM-RPM", in Bole, L. (ed.): Natural Language Based Computer Systems. München: Hanser/MacMillan, 119-253. Halliday, Michael A. K. 1961 "Categories of the Theory of Grammar", Word 17, 241-292. Harris, Zellig S. 1968 Mathematical Structures of Language. New York: Interscience Publ. Heydrich, Wolfgang 1982 Gegenstand und Sachverhalt. Hamburg: Buske. 1983 "Models and Realities", Text 3, 99-130. Heydrich, Wolfgang and Petöfi, Jänos S. 1983 "A Text-Theoretical Account of Questions of Lexical Structure. I and II.", Quaderni di Semantica 4, 120-127 and 294-311. Jameson, Anthony; Hoeppner, Wolfgang and Wahlster, Wolfgang 1980 "The Natural Language System HAM-RPM as a Hotel Manager: Some Representational Prerequisites", in Wilhelm, R. (ed.): GI: 10. Jahrestagung. Berlin: Springer, 459-473. Kindt, Walther 1981 "Word Semantics and Conversational Analysis", in Eikmeyer, H.-J. and Rieser, H. (eds.), 500-509. Lyons, John 1977 Semantics. (Vols. 1 and 2). Cambridge: Cambridge University Press. Massaro, Dominic W. 1972 "Perceptual Images, Processing Time, and Perceptual Units in Auditory Perception", Psychological Review 79, 124 — 145. 1974 "Perceptual Units in Speech Recognition", Journal of Experimental Psychology 102, 199-208. Mater, Erich 1966 Deutsche Verben (Bd. 1). Leipzig: VEB Bibliographisches Institut. McConkie, George W; Hogaboam, Thomas W.; Wolverton, Gary S.; Zola, David and Lucas, Peter A. 1979 "Toward the Use of Eye Movements in the Study of Language Processing", Discourse Processes 2, 157—177.

Word, Sentence, and Text Meaning

267

McMahon, William E. 1976 Hans Reichenbach's Philosophy of Grammar. The Hague: Mouton. Miller, George A. 1962 "Decision Units in the Perception of Speech", IRE Transactions on Information Theory, 81-83. 1979 "Images and Models, Similes and Metaphors", in Ortony, A. (ed.): Metaphor and Thought. Cambridge: Cambridge University Press, 202—250. Moravcsik, Julius 1981 "How Do Words Get their Meanings?", The Journal of Philosophy 78, 5-24. Nebel, Bernhard and Marburger, Heinz 1982 "Das natürlich sprachliche System HAM-ANS: Intelligenter Zugriff auf heterogene Wissens- und Datenbasen", in Nehmer, J. (ed.): GI: 12. Jahrestagung, Heidelberg: Springer, 392-402. Neubauer, Fritz and Petöfi, Jänos S. 1981 "Word Semantics, Lexicon Systems, and Text Interpretation", in Eikmeyer, H.-J. and Rieser, H. (eds.), 343-377. Nooteboom, Sibout S. and Cohen, Antonie 1975 "Anticipation in Speech Production and its Implications for Perception", in Cohen, A. and Nooteboom, S. S. (eds.): Structure and Process in Speech Perception. Berlin: Springer, 124-145. Petöfi, Jänos S. 1983a "Text, Signification, Models and Correlates: Some Aspects of Text Comprehension and Text Interpretation", in Rickheit, G. and Bock, M. (eds.): Psycholinguistic Studies in Language Processing. Berlin: de Gruyter, 266—298. 1983b "Forschung in der Semantik: Analyse and Repräsentation von Konzeptsystemen", in Petöfi, J. S. (ed.): Texte und Sachverhalte (= Papers in Textlinguistics 42). Hamburg: Buske, 72-91. Petöfi, Jänos S. and Rieser, Hannes (eds.) 1973 Studies in Text Grammar. Dordrecht: Reidel. Putnam, Hilary 1970 "Is Semantics Possible?", Metaphilosophy 1, 187-201. 1975 "The Meaning of'Meaning'", in Putnam, H.: Mind, Language and Reality. Cambridge: Cambridge University Press, 215—271. Rieger, Burghard 1981 "Feasible Fuzzy Semantics", in Eikmeyer, H.-J. and Rieser, H. (eds.), 193-209. Rieger, Burghard (ed.) 1982 Empirical Semantics. Bochum: Brockmeyer. Rieser, Hannes (ed.) 1982 Semantics of Fiction (= Poetics 1). Rieser, Hannes 1976 "Zur empirisch motivierten semantischen Beschreibung von Texten", in Rieser, H.: textgrammatik — Schulbuchanalyse — lexikon (= Papers in Textlinguistics 14). Hamburg: Buske, 4—54. 1983 "Deskriptive Rhetorik, Interaktion und natürliche Topik", in Petöfi, J. S. (ed.): Texte und Sachverhalte. Hamburg: Buske, 160—183. Schank, Roger (ed.) 1975 Conceptual Information Processing. Amsterdam: North-Holland. Schank, Roger and Abelson, Robert P. 1977 Scripts, Plans, Goals and Understanding: An Inquiry into Human Knowledge Structures. Hillsdale: Erlbaum. Schank, Roger and Riesbeck, Christopher K. 1981 Inside Computer Understanding. Hillsdale: Erlbaum.

268

H.-J. Eikmeyer

Stegmüller, Wolfgang 1979 Hauptströmungen der Gegenwartsphilosopbie. Band . 6., erweiterte Auflage. Stuttgart: Kröner. Wildgen, Wolfgang 1981 "Archetypal Dynamics in Word Semantics: An Application of Catastrophe Theory", in Eikmeyer, H.-J. and Rieser, H. (eds.), 234-296. 1982a Catastrophe Theoretic Semantics. Amsterdam: Benjamins. 1982b "Zur Dynamik lokaler Kompositionsprozesse: Am Beispiel nominaler ad-hocKomposita im Deutschen", Folia Linguistica 16, 297—344. 1983 "Skizze einer katastrophen-theoretisch fundierten dynamischen Semantik", Linguistische Berichte 84, 33-51.

OLIVER R. SCHOLZ

Some Issues in the Theory of Metaphor* 0. Introduction 0.1. The variety of metaphors

Metaphors can be used quite successfully in communication, e.g. in everyday discourse, but also in scientific or literary discourse. Nearly all of us are able to produce and use metaphors. When we make some utterances, we are able to signal to our listeners through certain means that we are speaking metaphorically. Everyday speakers of the language can recognize metaphors whenever they occur and in most cases they will have little problems in understanding them. In many cases it will be possible for the addressee to locate the particular metaphoric expressions within the larger context. Accordingly, speaker and listener seem to share the same principles in virtue of which they are able to signal and recognize metaphors as well as to work out the meaning of metaphoric expressions. Every language seems to embody an infinite (or, at least, inexhaustible) potential for new metaphors. No dictionary can anticipate all figurative uses of language, far less list all the readings of the potential metaphors. Metaphors are not restricted to special syntactic forms. Almost any (wellformed) combination of words can be used metaphorically. There seem to be very few restrictions with regards to the syntactic category of the metaphorically used expressions. Not only common nouns, adjectives and verb phrases can have figurative interpretations. Prepositions may also be understood metaphorically as in, "out of danger", "amidst many troubles", "in difficulties". The situation with regards to conjunctions is less clear. Interestingly enough, proper names have metaphorical applications. But here we have to distinguish between two quite different forms: (a) metaphorical uses of proper names whereby there occurs a change of the syntactic category (it may be argued whether those are genuine cases of metaphor at all); the most important subclass is of course the appellative use of proper names: — Goethe ist der Spinoza der Poesie. (H. Heine) — Boileau is the French Horace. * I want to thank Sussan Ameri for correcting my English.

270

O. R. Scholz

— Cesar Birotteau, ce Napoleon de la parfumerie ... (Balzac) — M. de Talleyrand, ce Roger Bacon de la nature sociale ... (Proust) — Bonn is not Weimar. — He is an Einstein. but often lexemes of the syntactic category 'proper name' are transferred to still other ones, as in — Apprendre l'art de bien Petrarquiser (Ronsard); (b) metaphorical uses of proper names while preserving the same syntactic category: — Look, Einstein is coming (said about T. B. who is a very intelligent person). Perhaps the only restriction that could be maintained within this context would be that some very general terms (like "to do", "to make", "situation", etc.) could not normally be applied figuratively. The syntactic patterns of actual or possible metaphors are also of the greatest variety.1 Very common are the forms — S is P (i.e. simple subject predicate sentences) — composite nouns — adjective + noun — noun + genetive attribute — noun + apposition — verb -f adverb. Metaphors need not in the least be syntactically anomalous, or even peculiar: They can occur in utterances of any mood, any illocutionary force and any tense; in all types of discourse (everyday, scientific, literary, religious, etc.) and in all sorts of text. You find them in fictional as well as non-fictional discourse. They may be used sincerely or insincerely. You can by way of the metaphor intentionally convey something the falsehood of which you are aware; i.e. you can lie with metaphors. Metaphors can be and often are combined with other figures of speech, e.g. with irony or hyperbole. (Indeed many metaphors seem to be at the same time hyperbolic; cp. "I have oceans of time".) 0.2. Idioms and metaphors2

It may be useful to compare metaphors to idioms: — Metaphors and idioms are linguistic devices to extend the expressive powers of a language without introducing new lexical material. Thus they fulfil an economic function. 1 2

Cp. Brooke-Rose (1958). Cp. Coulmas (1981); Davies (1982/83).

Some Issues in the Theory of Metaphor

271

— Metaphors and idioms do not seem to conform to Frege's principle. They are conventionally used with a meaning different from its constructed literal meaning (if there is one). But we have to be careful here. The literal interpretation of a metaphorically used expression may well be a function of the literal interpretations of its parts, but the metaphoric interpretation is not a function of the literal meaning of its parts alone. — Metaphors and idioms can (normally) not be translated word by word. — Many idioms and metaphors display a kind of twofold interpretability. Some idioms have meaningful literal counterparts, as metaphors have meaningful literal counterparts. For this reason, even quite frozen idioms and metaphors can be "reliteralized". This procedure has sometimes comic effects, as in the case of puns. — Many idioms simply are "petrified" metaphors. Nevertheless, there are also differences: — Idioms can be captured in a dictionary though it is not possible to anticipate all idioms. — Metaphors may become idioms, but interestingly enough, not all metaphors fade into idiomatic expressions, and not every idiom has a metaphoric origin. — Knowledge of the literal meaning of the metaphor gives one a considerable advantage for understanding the metaphoric meaning. An analogous principle does not hold for many idioms. 0.3. Reasons to employ metaphors

Our reasons to employ metaphors are manifold. The aesthetic, decorative emotive, rhetorical, euphemistic etc. functions have often been emphasized, but other (most notably cognitive) functions have usually been neglected. Rather than once more discussing the functions mentioned above, I would like to note the "organizing" and "economic" functions of the metaphor, which are interlinked. The American philosopher Nelson Goodman has noticed them both: By so putting old words to new work we save enormously on vocabulary and take advantage of established habits in the process of transcending them." (In: Johnson (ed.), 1981: 226) "This incessant use of metaphors springs not merely from love of literary color but also from urgent need of economy. If we could not readily transfer schemata to make new sortings and orderings, we should have to burden ourselves with unmanageably many different schemata, either by adoption of a vast vocabulary of elementary terms or by prodigious elaboration of composite ones. (Goodman, 1968: 80)

Metaphors are vitally important, for they allow the language users to respond to the changing world without being forced to introduce neologisms for every phenomenon, or to make their utterances unduly long-winded. Many metaphors are useful because they have organizing power. As Bergmann (1982: 243) puts it, "a metaphor has organizing power if it influences

272

O. R. Scholz

our orientation toward a subject matter". This feature is what makes metaphors valuable in the developing of scientific theories, but of course not only in this area3. 0.4. Producer and interpreter

The responsibility for the successful application of a metaphor seems to lie on the side of the speaker. If the latter is interested in successful communication, i.e. if he intends the listener to understand what he wants to say, or do what he expects him to do, he has to employ explicit clues or permit the contextual setup to make it sufficiently clear that the utterance is meant metaphorically. In addition, the speaker has to make sure that the listener grasps only the intended reading. If this is not the case, the utterer has to be prepared to correct misinterpretations by either moving to a literal paraphrase or once more attempting to use another metaphor. This does not mean that the producer of a metaphor is in a principally different situation from that of the listener when it comes to the interpretation of the metaphoric utterance. The speaker himself, precisely, is not expected to have full grasp of the complete scope of the metaphor. In his brilliant book Beyond the Letter, Scheffler (1979: 128 f.) has emphasized this point: In creating metaphor one may surprise oneself. Much of the discussion of metaphor has been conducted with reference to contexts of communication, and a mistaken, though tacit, assumption has been prevalent, i.e. that the producer of a metaphorical utterance has some special key to its comprehension which the hearer or reader can only struggle to find. In fact, the producer of any utterance, metaphorical or otherwise, may find it difficult or puzzling to interpret what has been said, and be surprised by the result of reflection on the matter. The interpretive role with respect to any utterance is not incompatible with that of producer, even when the purpose of the utterance has been straightforward communication.

/. Signaling and identifying metaphors

One of the most crucial steps in understanding the successful use of metaphors is to provide an answer to the question how we recognize metaphors and how we as speakers manage to signal that our utterances should be taken metaphorically. /./. Special clues

The speaker may use overt signals in order to make it clear that he wants to be understood in a non-literal way. He can employ, for example, formulations like "figuratively speaking, ...", "to speak figuratively...", or such phrases as "so to speak" "practically", "virtually" and so on (cp. the Latin "ut ita dicam"). In written texts quotation marks sometimes fulfil a similar function. 3

Cp. Eberle (1970); Boyd (in: Ortony (ed.) 1979: 356-408); Scheffler (1979: 128f.); Honeck/ Hoffmann (eds.) (1980); Lakoff/Johnson (1980).

Some Issues in the Theory of Metaphor

273

It is understood that such overt signals are no guarantee; the speaker may on occasion use them insincerely. Grice (1978) has pointed out that overt signals are much more unusual in the case of irony (consider e.g. "ironically speaking, ..."). He suggests two reasons: (a) Irony normally involves a far greater amount of pretending than metaphoric speech. Overt signals would, of course, spoil this effect. (b) In the case of irony the speaker disposes of conventional means other than the overt signals mentioned above, in order to indicate his intention (above all intonation). The same point can be made about similes. One would hardly say "speaking in a simile ...", because there exist other conventional means to indicate similes such as "like", "as ... as" etc. There exist some situations in which certain features bring with them that the listener/reader will be prepared for having to deal with metaphors rather than in other situations. For example, the epoch, genre, title or subtitle of a text may influence the reader's expectations as regards to frequency of figurative discourse. In both cases either when the language has not yet developed the possibilities to speak literally about a certain topic or when a particular social context prohibits literal discourse, people resort to metaphoric speech. As speaking in a literal manner about e.g. death is taboo, people are obliged to resort to metaphorical speech. Knowledge of such situations influences the expectations concerning figurative discourse. The producer of metaphor may considerably facilitate the identification and interpretation by using well-known metaphors or metaphors that — albeit new — are closely related to well-known metaphors. The previous linguistic experience of the listener will strongly guide his comprehension of non-literal utterances. In some cases he will probably interpret the utterance metaphorically in the first place (and not start with a literal interpretation). 1.2. Semantic deviance ana/or literal falsity (not necessary)

Many authors have thought that some sort of semantic deviance (of the literal reading) is at least a necessary condition for the occurrence of metaphors. Though there seems to be some sort of semantic deviance involved in many metaphors, this will not hold for all metaphors. Therefore, all approaches in terms of semantic anomaly or deviance will not be materially adequate. 1.2.1, Category mistakesI serial incorrectness (not necessary)

There exist a long standing tradition, which maintains that all metaphorically used expressions would be sortally incorrect if understood literally4. It cannot 4

Turbayne (21970), Goodman (1968: 73), for a more detailed discussion, see Lappin (1981).

274

O. R. Scholz

be denied that category mistakes often trigger metaphoric interpretations. However, the occurrence of a sortal incorrectness is by no means necessary for a metaphor. In order to see this point, we need not enter into a detailed discussion of the concepts "category mistake" or "sortal incorrectness". It will suffice to consider some examples: (1) This computer is a clown. (2) This dog is a clown. (3) Jim is a clown. (Context: Jim is a person that is not literally a clown, but may be metaphorically one)

(4) John is a clown. (Context: John is a person that is both literally and metaphorically a clown)

It should be clear that (1) —(4) could be used metaphorically in appropriate contexts, but at the most (1) and (2) could be construed as category mistakes. Even if we consider the construction of category trees a valuable project, I think no category tree could ever be constructed in such a way that every metaphor would prove to be a category mistake. Nevertheless, we should keep in mind that sortal incorrectness (of the literal interpretation) is one clue that often causes us to seek a non-literal interpretation. 1.2.2. Violation of selection restrictions (not necessary)

Many authors, writing in the Chomskyan tradition, have claimed that a violation of the selection restrictions (in a sentence) is at least a necessary condition for the occurrence of metaphors. This position is, of course, closely related to the sortal incorrectness approach. This is no accident; Katz has viewed his conception of selection restrictions as an attempt to explicate Gilbert Kyle's main ideas concerning categories (cp. Katz, 1972: 92). Even if an acceptable theory of selection restrictions can be developed, not all cases of metaphors can be judged to be violations of selection restrictions (cp. our last two examples). — There are simply other means to trigger metaphoric interpretations than category mistakes or violations of selection restrictions. A further problem is that linguists always list (what they think are) the semantic features (or defining properties) of the objects when they try to state the selection restrictions of the lexemes which stand for these objects. Instead in many metaphors, the commonly believed properties (or not even believed properties but accepted or tolerated cliches or stereotypes) are most important5. This point constitutes a fundamental criticism of many traditional theories of metaphors.

John Searle makes this especially clear in (1979).

Some Issues in the Theory of Metaphor

275

1.2.3. Literal falsity (not necessary)

Some theoreticians, even those who are otherwise skeptical about semantic anomaly theories of metaphor, try to maintain literal falsity as a necessary condition for metaphors (e.g. Loewenberg, 1975) but there are obvious counterexamples6: — Warsaw is a cold city. — No man is an island. These examples show that sentences which contain no semantic anomaly and can even be literally true (at least in some situations) can nevertheless be used metaphorically. It is quite surprising that the so-called "twice true statements" (T. Cohen, 1976) have so long escaped the attention of linguists and philosophers, since whole literary genres seem to depend on such a possibility, most notably allegories. That literal falsity is not a necessary condition for metaphors becomes especially clear when we consider the neglected topic metaphor and negation: If the principle, that negation changes the truth-value, does not — at least not in all (or even many) cases — vanish through addition of negation operators, we have a simple method to produce hosts of metaphors that are not literally false: — Achilles is not a lion. — Juliet is not the sun. — Life is not a shadow. 1.2.4. Summary The common mistake of the discussed theories consists in generalizing in an unjustified way frequent marks of metaphors and making them, thus, necessary, or even necessary and sufficient, conditions. As my examples have made clear, semantic anomaly and/or literal falsity are not necessary. (A related, though more deep-seated mistake lies in the attempt to extract from the sentence alone whether here is a metaphor or not. Dependent on the context, however, many [if not all] sentences can sometimes be interpreted literally and sometimes metaphorically.) 1.3. Towards a more adequate theory

Grice's theory of conversational implicatures provides an interesting base for a theory of metaphors. Despite its weaknesses, it suggests the means through which to handle the instances of metaphors the other theories are unable to account for. 6

Cp. T. Cohen (1976), Binkley (1974), Reddy (1969), Künne (1983: 187).

276

O. R. Scholz

In a Gricean theory, metaphors would be conceived as a special case of exploiting one or more of the so-called "conversational maxims"7. Grice himself makes the mistake to mention metaphors only in connection with exploitations of the first maxim of quality. Typically enough, he speaks of "category mistakes". Nevertheless, his framework is broad enough to handle metaphorical utterances where what is literally stated does not include a category mistake and/or is not false. The speaker can signal that he intends to convey something different from what he literally says by flouting one or several of the conversational maxims. So he may, for example, exploit the first maxim of quality by saying something, if interpreted non-metaphorically, is obviously false, e.g., "Man is a wolf?" Or he may exploit the first maxim of quantity of saying something, if interpreted literally, is obviously trivial or uninformative, e.g., "No man is an island". Or he may exploit the maxim of relation by saying something, if interpreted non-metaphorically, would be completely irrelevant, or an obvious deviation from the foregoing topic. This latter case is especially frequent. The conversational maxims may, thus, play a role on the following levels: (1) They may help to identify a metaphor in the communication context; in this complex process they may help to: (1.1) exclude literal interpretations (1.2) exclude non-literal interpretations other than metaphorical ones, e.g. ironical interpretations. (2) They may help to narrow down the range of possible interpretations of the metaphor; the reader/listener will choose among the prima facie plausible interpretations the one, those, that are most adequate vis-a-vis the conversational maxims8. Grice has also said something about how we can differentiate between metaphor and irony. It the addition of a negation leads to an utterance that does not violate one of the conversational maxims, it would be appropriate to look whether an ironic interpretation is in order. If, on the other hand, the addition of a negation does also result in the flouting of one conversational maxim, irony cannot be meant, and a metaphorical interpretation might be appropriate. Perhaps we can apply Grice's distinction between "generalized" and "particularized" conversational implicatures to the case of metaphor. For, certain metaphors do not need a specific context in order to be recognized and understood, whereas others do. Maybe we can go so far to say that "particula-

7 8

Cp. Grice (1975) and (1978). Cp. Levinson (1983), Künne (1983: 194ff.) for some principles of metaphorical interpretation, see Searle (1979).

Some Issues in the Theory of Metaphor

277

rized metaphors" are in most cases exploitations of the maxim of relation ("be relevant"), whereas generalized ones are mostly exploitations of the first maxim of quality. 1.4. Metaphor is not a speech act sui generis

Since I have come very close to a pragmatic theory of metaphor, a certain misunderstanding is near at hand. Some authors have mistakenly offered to consider metaphor as a speech act sui generis. The following objections can be made against this position: (i) As initially pointed out, metaphors can occur in every type of speech act. (ii) The following principle seems to hold: if you perform a speech act of type t by uttering a sentence p, then you will not perform a speech act of type / (i.e. the same type) by uttering the sentence "if p, then ^". It seems obvious that metaphors do not behave according to this principle, that is, a metaphor does not cease to be a metaphor, if it appears in the antecedent of a conditional. If the principle in question is deemed reasonable, it would follow that metaphors do not form a special type of speech act9. 1.5. Forms of interpretative indecision^

(i) In some situations the interpreter may be unable to decide whether he should interpret some utterance of the speaker metaphorically or nonmetaphorically, e.g. literally. (The author of a metaphor can, of course, intend such a situation.) (ii) Although there seem to be some sentences of which all parts can be simultaneously used metaphorically, the paradigmatic cases are clearly ones in which some parts of the utterance may be understood metaphorically and some literally. Now even after the listener has chosen to interpret the utterance (as a whole) metaphorically, it may remain unclear to him in a given context which elements of the expression are to be understood literally and metaphorically, respectively. Sometimes it is only obvious that certain elements, if taken literally, are incoherent. This in itself does not indicate how this prima facie incoherence can be resolved. In the case of simple subject-predicate sentences, for example, it may not be recognizable from the sentence alone whether the subject or the predicate has to be taken metaphorically.

9 10

These arguments are used by Kiinne (1983: 193). This term is borrowed from Scheffler (1979).

278

O. R. Scholz

Generally, the overall topic of the discourse provides a clue in favour of one of the possibilities. In extreme cases, though, even the whole context may leave open more than one possibility. (The author of a metaphor may intend to create such a situation.) (iii) Occasionally, though it is clear to the listener that a metaphorical interpretation is in order and he also realizes which elements of the utterance are to be taken metaphorically, radically different readings are possible. 2. Metaphors in texts 2.1. Metaphors in fictional texts

Whether a certain expression is interpreted literally or metaphorically does also depend on what the reader takes to exist in the real or fictional "world" and which natural and logical laws he takes to hold in this world. Some expressions which would receive a metaphorical reading in a non-fictional context, may in the context of a fairy tale or other fictional context be understood correctly as literal. The opposite case may also arise. This fact creates problems for the use of metaphors in particular fictional texts the fictional "worlds" of which are quite removed from that of everyday life. The difficulty for the writer consists in successfully signalling to the reader which elements are to be taken metaphorically, provided that the author wants the reader to recognize those elements. He may, of course, intend to leave the decision to the reader and that may occasionally contribute to the effect of the text. The reader will often have to choose whether a certain recognized anomaly is due to a metaphor, or other figure of speech, or to the strangeness of the fictional "world". 2.2. Chains of metaphors In many literary texts, we do not solely come across isolated cases of metaphors, but also whole chains of metaphors that hang together. Such chains can extend over few sentences as well as whole passages or even whole texts. In interesting cases, several metaphors might be coherent on a figurative level which runs parallel to the coherent literal level, as, for example, in allegories. These levels may be connected by explicit or implicit isotopic connectors. According to their occurrence in the text, they can be further distinguished as antecedent (AIC) and subsequent isotopic connectors (SIC), respectively. Subsequent isotopic connectors make a re-interpretation necessary. We may distinguish at least the following cases (E = expression; L = interpreted literally; M = interpreted metaphorically; SIC = subsequent isotopic connector; L/M = double reading — literal/metaphorical interpretive indecision).

Some Issues in the Theory of Metaphor

(a) ... -

... — (b) ... -

... — (c) ... —

EL - EL - EL - ... - Ei

- SIC - ... [first reading]

EM

— ...

[second reading] EL - EL - EL - ... - EL - SIC - ...

... —

— EM

— ... — EM

EL/M — EL/M — EL/M — ... — EL/M — ... EM

... -^EL (d) ... —

— EM

279

EM

— EM

— EM

— ... — EM

- EL - EL - ... - EL — EM

— EM

— ... — EM

— SIC — ...

- ... — SIC —

EL/M — EL/M — EL/M — ... — EL/M — ...

2.3. Expanding metaphors We may distinguish between two elementary ways of expanding metaphors: (i) One of the strategies would be to develop the metaphor by formulating related metaphors (by which I mean metaphors from the same realm). This strategy can be used in the writing of a small passage and also in the conception of a whole text as often happens in poems. The type of metaphor that has been called "fecund metaphor" by Merrie Bergmann is especially apt to fulfil this role. She characterizes this class in the following way: "A metaphor is fecund if it suggests other, related, metaphors" (1982: 243). Developed metaphor: What is life? A thawing iceboard On a sea with sunny shore; — Gay we sail; it melts beneath us; We are sunk, and seen no more. (Thomas Carlyle, from CHI bono?)

(ii) Another way to expand metaphors is to pick metaphors from a realm other than that of the first metaphor where these serve to illustrate the same point (or a similar point). The first metaphor is so to speak paraphrased by different metaphors. Metaphors from different realms for the same topic:

280

O. R. Scholz Stop and consider! 'Life is but a day; A fragile dew-drop on its perilous way From a tree's summit; 3a poor Indian's sleep While his boat hastens to the monstrous steep Of Montmorenci. Why so sad a moan? 4 Life is the rose's hope while yet unblown; 5 The reading of an ever-changing tale; 6 The light uplifting of a maiden's veil; 7 A pigeon tumbling in clear summer air; 8 A laughing school-boy, without grief or care, Riding the springy branches of an elm. (John Keats, from Sleep and Poetry) 2

These two techniques are often combined as in the following famous passage: Life's but a walking shadow, a poor player, That struts and frets his hour upon the stage And then is heard no more; it is a tale Told by an idiot, full of sound and fury, Signifying nothing. (Macbeth, V, v)

Bibliography Alston, William P. 1964 Philosophy of Language. Englewood Cliffs, N.J.: Prentice-Hall. Apostel, Leo 1979 "Persuasive Communication as Metaphorical Discourse under the Guidance of Conversational Maxims", Logique et Analyse 22, 265 — 320. Beardsley, Monroe C. 1958 Aesthetics: Problems in the Philosophy of Criticism. New York: Harcourt, Brace & World. 1962 "The Metaphorical Twist", in: Philosophy and Phenomenological Research 22, 293—307 (repr. in: Johnson, M. (ed.) (1981): Philosophical Perspectives on Metaphor. Minneapolis: University of Minnesota Press, 105 — 122). 1976 "Metaphor and Falsity", Journal of Aesthetics and Art Criticism 35, 218-222. 1978 "Metaphorical Senses", Nous 12, 3-16. Bergmann, Merrie 1979 "Metaphor and Formal Semantics", Poetics 8, 213-230. 1982 "Metaphorical Assertions", The Philosophical Review 91, 229-245. Binkley, Timothy 1974 "On the Truth and Probity of Metaphor", Journal of Aesthetics and Art Criticism 33, 171-180. Black, Max 1962 Models and Metaphors. Ithaca/N.Y.: Cornell University Press, 1962. 1977 "More about Metaphor", Dialectica 31, 431 -457 (repr. in: Ortony, A. (ed.) (1979): Metaphor and Thought. Cambridge: Cambridge University Press, 19 — 43). 1979 "How Metaphors Work: A Reply to Donald Davidson", in: Sacks, Sh. (ed.), 181-192. Brooke-Rose, Christine 1958 A Grammar of Metaphor. London: Seeker & Warburg.

Some Issues in the Theory of Metaphor

281

Cohen, Ted 1976 "Notes on Metaphor", Journal of Aesthetics and Art Criticism 34, 249-259. Coulmas, Florian 1981 "Idiomaticity as a Problem of Pragmatics", in: Parret, H. et al. (eds.): Possibilities and Limitations of Pragmatics. Amsterdam: Benjamins, 139—151. Davidson, Donald 1978 "What Metaphors Mean", Critical Inquiry 5, 31 -47 (repr. in: Johnson (ed.) (1981), 200-220). Davies, Martin 1982/83 "Idiom and Metaphor", Proceedings of the Aristotelian Society 83, 67-85. Eberle, Rolf A. 19702 "Models, Metaphors, and Formal Interpretations", in: Turbayne, C. M.: The Myth of Metaphor. New York: Columbia University Press (Appendix). Goodman, Nelson 1968 Languages of Art: An Approach to a Theory of Symbols. Indianapolis: Bobbs-Merrill. 1979 "Metaphor as Moonlighting", in: Sacks, Sh. (ed.), 175-180 (repr. in: Johnson, M. (ed.) (1981), 221-227). Grice, H. Paul 1975 "Logic and Conversation", in: Cole, P. and Morgan, J. L. (eds.): Speech Acts (Syntax and Semantics, Vol. 3). New York: Seminar Press, 41—58. 1978 "Further Notes on Logic and Conversation", in: Cole, P. (ed.): Pragmatics (Syntax and Semantics, Vol. 9) New York: Seminar Press, 113—127. Honeck, Richard P. and Hoffman, Robert P. 1980 Cognition and Figurative Language. Hillsdale: Erlbaum. Johnson, Mark (ed.) 1981 Philosophical Perspectives on Metaphor. Minneapolis: University of Minnesota Press. Katz, Jerrold J. 1972 Semantic Theory. New York: Harper & Row. Kiinne, Wolfgang 1983 '"Im übertragenen Sinne': Zur Theorie der Metapher", Conceptus 17, 181-200. Kurz, Gerhard 1982 Metapher, Allegorie, Symbol. Göttingen: Vandenhoeck & Ruprecht. Lakoff, George and Johnson, Mark 1980 Metaphors We Live by. Chicago: University of Chicago Press. Lappin, Shalom 1981 Sorts, Ontology, and Metaphor: The Semantics of Serial Structure. Berlin: de Gruyter. Levin, Samuel R. 1977 The Semantics of Metaphor. Baltimore: John Hopkins University Press. Levinson, Stephen C. 1983 Pragmatics. Cambridge: Cambridge University Press. Loewenberg, Ina 1975 "Identifying Metaphors", Foundations of Language 12, 315 — 338 (repr. in: Johnson, M. (ed.) (1981), 154-181). Mooij, Jan J. A. 1976 A Study of Metaphor. Amsterdam: North-Holland. Nieraad, Jürgen 1977 Bildgesegnet und Bildverflucht: Forschungen %ur sprachlichen Metaphorik. Darmstadt: Wissenschaftliche Buchgesellschaft. Ortony, Andrew (ed.) 1979 Metaphor and Thought. Cambridge: Cambridge University Press. Reddy, Michael J. 1969 "A Semantic Approach to Metaphor", Papers from the Fifth Regional Meeting of the Chicago Linguistic Society, 240 — 251.

282

O. R. Scholz

Sacks, Sheldon (ed.) 1979 On Metaphor. Chicago: Chicago University Press. Scheffler, Israel 1979

Beyond the Letter: A Philosophical Inquiry into Ambiguity, Vagueness, and Metaphor in

Language. London: Routledge & Kegan Paul. Searle, John R. 1979 Expression and Meaning. Cambridge: Cambridge University Press (esp. 76 — 116). Shibles, Warren A. 1971 Metaphor: An Annotated Bibliography and History. Wisconsin: Language Press. Turbayne, Colin M. 19702 The Myth of Metaphor. New York: Columbia University Press. Weinreich, Uriel 1972 Explorations in Semantic Theory. The Hague: Mouton.

ROGER G. VAN DE VELDE

Inferences as (De)compositional Principles /. Introduction

Man is a reasoning (inter)locutor and (co)agent. The communicative and cognitive processes involved in discourse production and reception provide confirmations of this basic view. It will be shown in the present paper that some of these confirmations are concerned with the solutions we propose to the following specific problems: How are the different kinds of information organized in discourse? How does man handle the verbally expressed information of discourse? How does man invoke coverbal and nonverbal information parts for an appropriate interpretation of discourse? How does man derive information from discourse? The present paper will tackle these problems by presuming that the human capacity to make inferences and to construct coherence within/between/ beyond information parts is responsible for diversified forms of discourse organization and comprehension. Limitations of space urge us to restrict our exposition to those inferences which are made in the comprehension processes of written discourse. We refer to them by the terms 'inference-making processes', 'inferential processes', 'reasoning processes', 'inferential processing', and 'inferences'. The construction of coherence which, under normal conditions, results from the integrative and interactive functioning of inferences cannot be part of our present considerations (van de Velde; 1981 a; 1981 b; 1982 a; 1985; 1986; 1987). By the phrase 'under normal conditions' we mean that no cognitive impairments and no language disorders hamper the integrative and interactive functioning of inferences. In discourse interpretation different types of human information processing are involved. They may range from low level visual/auditive perception to high level discourse processing where generalizing, deleting and constructing macro-operations (van Dijk, 1977; Kintsch, van Dijk, 1978; van Dijk, 1980) may occur. We shall be concerned predominantly with high level discourse processing, as it is actualized in syntactic processing, semantic processing, and action processing. In the first stage, we shall present a brief general outline of the allembracing functions which we ascribe to inferences.

284

R. G. van de Velde

In the second stage, it will be shown how syntactic processing takes place on the basis of syntactic inferences. In the third stage, we shall argue that semantic processing depends on the interactive functioning of two other kinds of inference: 1) inferences which have recourse to the discourse receiver's internal lexicon (termed 'ILRRR inferences'); 2) semantic-logical inferences which, among other things, enable us to discern the underlying propositions of discourse, to identify their interpropositional relationships and to handle the total prepositional content of discourse. In the fourth stage, we shall be concerned with the role which action-oriented inferences play in action processing. In the fifth stage, we shall deal with the complex relationships between pre-existing knowledge sources, inferencing and discourse processing. Thereby, we shall distinguish socioculturally determined knowledge from private knowledge. In addition, we shall also defend the view that the knowledge sources must differ depending on the type of discourse in which they are needed for an appropriate interpretation. In the sixth stage, we shall look into non-normal conditions of discourse organization and comprehension. We shall clarify in what ways the interactive functioning of inferences can be impaired in aphasia. Let us close the introductory considerations by elucidating how the views which will be set out relate to the global theme of the present volume. We conceive of discourse interpretation as a spherical complex consisting of decompositional and compositional operations. In normal discourse processing these decompositional and compositional operations converge/coincide continuously. In the one hemisphere, the decompositional operations concern the ways in which the discourse receiver disentangles the externally presented discourse information. In the other hemisphere, the compositional operations comprise the ways in which the discourse receiver invokes/makes use of not verbally expressed information in order to arrive at a successful discourse comprehension. The principles governing both operations are inferential in nature. This paper will show what their inferential nature really means. 2. On inference (-making processes)

In recent years there has been an enormous explosion in writing on the roles which information sources and related inferences play in the fields of human thought, language, memory, cognition, argumentation, action, communication, and so on. This exuberant literature extends to the fields of artificial intelligence (cf. a. o. Rieger, 1975; Charniak, Wilks, 1976; Cullingford, 1979), logic and presuppositional semantics (cf. a. o. Blau, 1978; Oh, Dinneen, 1979;

Inferences as (De)compositional Principles

285

Seuren, 1979), linguistics (cf. a. o. Bellert, 1972; Lakoff, 1972; Gordon, Lakoff, 1975), philosophy of language (cf. a. o. Franck, Petöfi, 1973; Gunderson, 1975; Margalit, 1976), cognitive psychology (cf. a. o. Clark, 1977; Noordman, 1979; Brockway, 1981), and to their neighbouring disciplines (cf. a. o. ParkerRhodes, 1978; Omanson, et al, 1978; Hildyard, Olson, 1978; Crothers, 1979a, 1979b; Goetz, 1979; Hildyard, 1979; Warren, et.al., 1979; Graesser, 1981; Agar, Hobbs, 1982). We do not have space enough to provide a complete picture of how the different views on information and inference evolved nor of what degrees of disparity and dissension manifest themselves in the terminologies and classifications. Let us only point to some areas of agreement in (1) —(5): 1) Discourse interpretation consists in going beyond the information given in a discourse di. Normally, this interpretation depends on integrating the diverse chunks of information expressed in the earlier and later parts of di. It relates these information chunks to pre-existing knowledge sources. It results in deriving new information from dt; 2) the pre-existing knowledge is not a random mixture of informational data but instead is organized according to some inferential principles; 3) in drawing consequences from a discourse di some truth assumptions must be made on local information expressed by individual utterances of d, and on the global information of di; 4) in di some specific utterances »i, ..., »5, ...., «8 are conditions of interpretation for the respective posterior utterances #2, ...., »6, ...., »9 (Kintsch, van Dijk, 1978). Then, one can speak about forward inferencing (van de Velde, 1984); 5) in cases of inferential reshuffling (or backward inferencing), posterior utterances »2, ...., »6, , »9 become preconditions for the interpretation of the anterior utterances »i, , «5, , »s (van de Velde, 1984) etc. With regard to points (4) and (5), it is worth investigating in what sense the conditions of interpretation can be conceived of as presuppositions. We shall only touch incidentally upon this line of investigation. As is convincingly demonstrated in Bellert (1973), some relevant types of presupposition can be described as particular cases of inference. Therefore, we shall concentrate predominantly on inference and on inference-making processes. We will start with the following hypothesis: The decompositional and compositional principles of discourse interpretation cannot be accounted for efficiently and appropriately if the human capacity to make inferences is disregarded. The human capacity to make inferences is most frequently manifested in language use. For this reason, it must be the first order task of cognitively oriented linguists to study those inferences which are made (possible) on the basis of the forms and meanings actualized in language use. The present paper is an attempt to conform to this first order task by laying bare how inferences are intertwined with the use of language. In so

286

R. G. van de Velde

doing, we shall take a stand which differs in many respects from the above cited literature on information and inference. Some of these writings emphasize that the critical role of inferences lies in filling in the missing slots (slotfilling inferences). Other investigations place stress on the crucial fact that inferences function as connecting devices (text-connecting inferences). Still other research reports assume that inferences serve predominantly to provide information on spatial and temporal relations (informational inferences). There are also studies which advocate that inferences are responsible for the discourse receivers' evaluative judgments (evaluative inferences), etc. Contrary to these viewpoints, our basic presumption is that the inferences we are concerned with fulfill all-embracing functions in the different modalities of language use. Of course, we cannot sufficiently illustrate and check this basic presumption within the bounds of the present paper (but see van de Velde, 1981 a; 1981 b; 1982 a; 1984; 1986; 1987). For illustrative purposes let us begin by provisionally touching upon two extremes of language processing which may show how widely and deeply inference-making processes are involved. At the one extreme, we take low level language processing of written documents. In this regard, inference-making processes start by extracting, comparing and identifying the distinctive features of graphemes as in (1) —(8): (1) tanpicitrapmuiuqollocehtoteybdoogyasotdetnawnamowgnuoyeht (2) tanpicitrap muiuqolloc eht ot eybdoog yas ot detnaw namow gnuoy eht (3) theyoungwomanwantedtosaygoodbyetothecolloquiumparticipant (4) the young woman wanted to say goodbye to the colloquium participant (5) TANPICITRAPMUIUQOLLOCEHTOTEYBDOOGYASOTDETNAWNAMOWGNUOYEHT (6) TANPICITRAP MUIUQOLLOC EHT OT EYBDOOG YAS OT DETNAW NAMOW GNOUY EHT (7) THEYOUNGWOMANWANTEDTOSAYGOODBYETOTHECOLLOQUIUM PARTICIPANT (8) THE YOUNG WOMAN WANTED TO SAY GOODBYE TO THE COLLOQUIUM PARTICIPANT

In language processing of (1) —(8) it is clear that inferences are needed. Readers can be made aware of (their need for) these inferences when they are asked why/how they arrive at judging that (1), (2), (3), (5), (6), and (7) are awkward, why/how they reject (1), (2), (3), (5), (6), and (7) as not corresponding to English orthography, why/how they detect that/where blank spaces are missing in (1), (3), (5), and (7), why/how they find out that (1), (2) and (5), (6) include the respective anagrams of (4) and (8), and so on. More specifically, the recognition of similarities and differences in the curvature and angular properties, in the circular forms, in the line segments, etc. of the letters and their combinations is dependent on forming hypotheses, on verifying/falsifying the formed hypotheses, on refining the formed hypotheses, on discarding the falsified hypotheses, and the like. In this respect, the graphemic decoding activities carried out on (1) —(8) can be conceived of as instances of problem solving. As we have demonstrated and controlled

Inferences as (De)compositional Principles

287

elsewhere, readers must, first of all, find out the real nature of the problems at hand. Then they must rely on pertinent information sources, which among other things, concern the graphemic/syntactic/semantic/contextual data. The comparison, selection, and combination of information sources are part of decision-making operations. These decision-making operations display diverse characteristics of inference. The study of analytical, sequential and holistic processing strategies of word recognition as well as the investigation of alexia/dyslexia, visual language disorders, and related receptive disabilities provides appropriate means to confirm this view (van de Velde, 1987). At the other extreme, we take high level language processing of written documents. In this respect, a wide variety of inference-making processes may be assumed. They may serve to construct syntactic and semantic hypotheses on a discourse dt, to identify the different discourse-internal links of dit to fill in the missing slots/connections in di, to find relations between d; and the discourse-external information sources, to instigate/control the local and global interpretations otdi and to explore all further interpretative elaborations of di. In high level discourse processing there is no upper limit to the applicability of reasoning processes. Accordingly, for a discourse di intrusive in-depth-interpretations may give rise to expanded versions of di (cf. also Labov, Fanshel, 1977; Warren et. al., 1979; Bellert, 1980; Cicourel, 1980; Langleben, 1981). We shall make this point clear by considering (9): (9)

The young woman wanted to say goodbye to the colloquium participant. When she glanced at him he was completely out of his mind. All day he had observed her in the auditorium. He was not blind to her attractiveness. She looked as if she were wrapped in mystery. Now it turned out that she had to leave abruptly. He could not guess the reasons for her unforeseen departure.

Though (9) is a very short discourse fragment, it may give rise to extensive inference-based elaborations of its contents. That is to say, interpretations of (9) need not be fenced in within its own strict limits of graphemic, syntactic and semantic information sources. Any interpreter may add new hypotheses to the existing interpretation(s), may interpolate new propositions between the underlying propositions of (9), may induce new knowledge data in order to arrive at a more plausible meaning analysis of (9), etc. (van de Velde, 1984). More specifically, the discourse processor may supplement the core meaning of (9) with hypotheses which concern 1) the relations between the young woman and the colloquium participant; 2) the topic(s) of the colloquium; 3) the profession of the young woman; 4) the attitudes and emotions of the male colloquium participant; 5) the reasons for the young woman's unforeseen departure, etc. We will not go here into the potential extensions of the core meaning of (9). Instead, we shall dwell on those inferences which make the interpretation of (9) as well as its potential interpretative elaborations possible. From the wide variety of inference-making processes we shall single out four major

288

R. G. van de Velde

kinds. They are taken together, because they are intertwined and close(ly tied) to human natural language. In addition, they are the basic ingredients of discourse interpretation. One admonition is still in order: in focusing on the four major kinds of inference, we will be concerned only with so-called 'descriptive interpretation' of verbal texts (Petöfi, 1981 a, 1981 b, 1983). Descriptive interpretation of verbal texts includes in this paper the combination of analytical, sequential and holistic processing strategies. That is to say, the so-called 'explanative interpretation' which, among other things, takes into account social, historical, cultural, and other contexts must be relinquished. In the same vein, the evaluative dimensions of interpretation which rely on norms, attitudes, beliefs, etc. and often reflect the subjective bias of meaning assignment (van Dijk, 1982) will have to be left out of consideration. By the same token, we shall not dwell on the broad conceptions concerned with 'speaker's meaning' (Grice, 1975), 'intended nonliteral meaning' (Gibbs, 1982), and 'illocutionary force' (Austin, 1962; Searle, 1969; 1975; 1983). They all belong to studies on the wider contexts of verbal communication. We shall account for such conceptions only to the extent that they have repercussions on the descriptive interpretation of the propositipnal content of discourse.

3. On the kinds of inference

The four major kinds of inference which we shall distinguish can be treated from the viewpoints of different disciplines. In cases where they concern the actual(ized) inference-making processes, they belong to cognitive psychology. When their dependency on the norms and rules of rational argumentation is at issue, they reside in the realms of logic. When they rely on languagebound information, they are the subject matter of (cognitively oriented) linguistics. When they relate to the intentions/beliefs/expectations of the language producer/receiver and to the communicative functioning of discourse, they are of main interest to pragmatics, philosophy, (cognitive) psychology, ethnomethodology, and other disciplines. Our approach to inference-making processes will cover the perspectives of linguistics. Therefore, we shall single out those major kinds of inference which are fundamental to language functioning: syntactic inferences, ILRRR inferences, semantic-logical inferences, and action-oriented inferences. However, if necessary, we shall cross the boundaries of linguistics and take into consideration the instructive viewpoints of other fields of research. 3.1. On syntactic inferences

When inference-making processes rely on the discourse receiver's knowledge of the rules of syntax, we term them 'syntactic inferences'. Since the study

Inferences as (De)compositional Principles

289

of syntax encompasses all lower levels of grammatical description (such as phonology, morphology, morphosyntax) we consider syntactic inferences to include reasoning processes which (can) make use of low and high level grammatical hints, cues, and means. Syntactic inferences must provide us with solutions to the following problems: 1) How can a sequentially ordered combination of sounds/letters be identified as a syntactic constituent? 2) How can the discourse receiver identify those strings of sounds/letters which are syntactically related to each other? 3) How can the discourse receiver detect what specific syntactic relations exist between the identified strings of sounds/letters? 4) What syntactic functions (subject, indirect object, direct object, etc.) are to be assigned to the identified strings of sounds/letters? 5) What syntactic devices, cues and rules (concerned with inflection, function words, order, etc.) have to be applied in order to find out what the underlying syntactic relations of multi-utterance-combinations such as (9) are? It is undeniable that we need syntactic inferences in the discourse processing of (9). For instance, in the decompositional operations of the analytical processing strategies they contribute to the identification of the syntactic constituents the young woman and the colloquium participant. In the sequential processing strategies, these syntactic inferences serve to recognize the syntactic relations between the young woman and the subsequent proforms she, her, she, she, she, her. Similarly, syntactic inferences provide the preconditions for the detection of the syntactic relations which exist between the colloquium participant and the anaphorical pronouns him, he, his, he, he. Syntactic inferences may also help the discourse receiver to find out that the tenses wanted to say goodbye, glanced, was, had observed, was, looked, were wrapped, turned out, had to leave, and could express the temporal reference of (9) in a well-organized way. Finally, since the order of the syntactic constituents is rule-governed, syntactic inferences allow us to discern in what positions the subject, the main verb, the (in)direct object, etc. occur, what subordinate clauses precede or follow the main clauses, etc. On the basis of syntactic inferences the discourse receiver may come to the conclusion that (9) displays syntactic organization (termed 'cohesion', van de Velde, 1978; 1981 a; 1987). The discourse receiver's recognition of cohesion is fundamental to her/his syntactic processing. In order to control our hypothetical views on syntactic processing we presented 48 graduate students with (9). Their task was to interpret (9) and to justify their assigned meanings on a standard examination paper. There

290

R. G. van de Velde

was no time limit imposed. All students agreed that the syntactic relations were to be identified as demonstrated above and that cohesion could be recognized in (9). 3.2. On ILRRR inferences The next problem which must be raised in this regard is whether the syntactic inferences can work on their own. We surmise that they cannot. The reason is straightforward: syntactic inferences do not operate on vacuous elements but on stretches of discourse which, among other things, carry with them semantic information. This view was also confirmed by all 48 answers provided in the interpretation tasks related to (9): the students drew the following consequences from (9) which exceed by far the domains of syntactic inferences: (10) (11) (12) (13) (14) (15) (16)

a colloquium was held the two persons referred to by (9) participated at this colloquium the two persons referred to by (9) attended one (or more) lecture(s) of the colloquium which was (were) given in an auditorium the two persons referred to by (9) met each other at the colloquium and exchanged information the two persons referred to by (9) belonged to different sexes the male colloquium participant was probably impressed by the female participant the male colloquium participant had concerns about the female participant's cognitive/ conative, affective background activities (goals/intentions/plans/wants, etc.) and about her physical actions, etc.

The main question arising here is: How do discourse receivers derive the information data (10) —(16) from (9)? It is undeniable that their exclusive reliance on syntactic inferences is an insufficient foundation for the consequences (10) —(16). Discourse receivers must also rely on those inferences which take into account the (combined) meanings of the identified syntactic constituents. Roughly speaking, the syntactic constituents of (9) may be distinguished into so-called 'content words' and 'function words'. The function words serve to express syntactic relations and thus provide hints and cues for the syntactic inferences. The content words denote perceptual, conceptual and affective information. They provide the basis for the functioning of ILRRR inferences in semantic processing. ILRRR inferences have recourse to the knowledge domains which are stored in the discourse receiver's internal lexicon. These knowledge domains are verbal and nonverbal in nature. In multidisciplinary research many terms have been used to refer to these knowledge domains. From among them we single out: 'knowledge base', 'data base', 'lexical memory', 'subjective lexicon', 'thesaurus', 'lexical knowledge', 'encyclopaedic knowledge', 'semantic memory', 'conceptual memory', 'mental lexicon', 'internal lexicon', 'background knowledge', 'common ground', 'knowledge of the world', 'shared

Inferences as (De)compositional Principles

291

knowledge', and 'background assumptions'. Space limits prevent us from going into the many intriguing issues here (van de Velde, 1986; 1987). Let us only clarify what we mean by the term 'internal lexicon'. We conceive of the mternal /exicon as the cognitively reflected representation of reality (abbreviated ILRRR). This abbreviation includes two main aspects of reflection. First, it denotes that reality is reflected in the organized information of ILRRR. Secondly, it signifies that man as a learner and thinker reflects in cognition on (the organizational features of) reality. As a matter of fact, text-grammatical studies concerned with lexical definitions (e.g. Petofi, 1973; Petöfi, Rieser, 1973) as well as research in artificial intelligence and in the cognitive sciences confirm both reflection aspects of ILRRR. Moreover, neurolinguistic and neuropsychological findings support the view that ILRRR has its own intrinsic organization which is dominated by inferential principles (van de Velde, 1987). In order to illustrate briefly the importance of ILRRR inferences we start from the following question: How can the discourse receiver draw the consequence (10)? This question is the more puzzling, because (9) does not express in explicit terms anything about a colloquium. The syntactic inferences can identify only one syntactic constituent in (9) which may serve as a clear indication to instigate interactively operating inferences which evolve in (10). This single syntactic constituent the colloquium participant leads us to consider some logico-philosophical and logico-linguistic views on presupposition. According to these views, the colloquium participant presupposes (17): (17)

There exists at least one colloquium participant

Does (17) justify in a straightforward way the consequence (10)? We surmise that it does not. The presupposition that a colloquium participant exists does not necessarily involve her/his being present at a colloquium. The discourse receiver must rely on much more extensive information in order to arrive at the consequence (10). A great deal of this information is included in ILRRR. In knowing what the colloquium participant means, the discourse receiver has at her/his disposal a store of knowledge data which reach far beyond conventional lexicographic data. These additional knowledge data pertain to reality, to the meanings of other content words, to their so-called 'semantic fields', 'semantic networks', 'conceptual dependencies', and to their connections with other cognitive domains. We cannot give here an exhaustive and sufficiently systematic description of this ILRRR information (cf. van de Velde, 1986; 1987). Let us only present in (18)-(26) those ILRRR data on the colloquium participant which may help to justify the consequence (10) (and the related consequences (11) —(12)): (18) (19)

If a colloquium is held, then there is more than just one participant If a colloquium is held, then the participants are supposed to attend one or more lectures

292

R. G. van de Velde (20) (21) (22) (23) (24) (25) (26)

If a colloquium is held, then there is/are a participant/participants who is/are a lecturer/ lecturers If a colloquium is held, then the lecturer/lecturers has/have her/his/their turn(s) to take the floor If a colloquium is held, then there is/are a participant/participants who belong(s) to the audience If a colloquium is held, then one subject matter/more subject matters is/are discussed If a colloquium is held, then the participant/participants is/are supposed to possess specialized knowledge about the subject matter(s) to be treated If a colloquium is held, then the lectures are given in a building/auditorium/room If a colloquium is held, then a specific time schedule is imposed etc.

From the answers given in the interpretation task of (9) we may conclude that the discourse receivers assumed that the antecedents of the conditionals (18), (19), (22), (23), and (25) were true with respect to the past events/actions referred to by (9). On the basis of these truth assumptions (9) made sense to them and moreover allowed them to draw consequences (10), (11), and (12). It is important to stress the following point: the truth assumptions pertaining to the interpretation of (9) are made on the antecedents of those conditionals which are part of the discourse receiver's ILRRR. On the basis of ILRRR the discourse receiver has the capacity to assign truth or falsity to the underlying propositions of a discourse di. These truth or falsity assignments are fundamental to the inferential modes of discourse comprehension. This may become clear from (27) and from its logical consequences (28) —(36): (27)

The boy looked around in the barn. He saw the rat which was bigger than the calf. Then, he observed that the calf was bigger than the horse. All of a sudden a mouse entered. It was the biggest of all the animals.

If one assumes that THE AUTHOR OF (27) AFFIRMS THAT (27) is the macropropositional attitude of (27), then one may also come to see that the propositions underlying (27) lead the receiver to make the following inferences (28)-(36): (28) (29) (30) (31) (32) (33) (34) (35) (36)

The The The The The The The The The

calf was smaller than the rat horse was smaller than the calf calf was smaller than the mouse horse was smaller than the mouse rat was bigger than the horse mouse was bigger than the rat mouse was bigger than the calf mouse was bigger than the horse horse was the smallest animal

It is an unvarnished truth that the truth or falsity assignments to (27) can be made by relying on ILRRR information about the content words rat, calf, horse, and mouse. Normal discourse receivers do not suspend their ILRRR information on these familiar content words. Therefore, they are able to recognize that (27) includes false premises, that (28) —(36) contradict their

Inferences as (De)compositional Principles

293

ILRRR information, that (28) —(36) conform to the 'ex falso quodlibet'principle, and the like. This short excursion on the truth or falsity assignments brings us to the following important point in the discussion: ILRRR inferences are, if taken on their own, unsatisfactory means for a successful interpretation of discourse. Apart from their afore-mentioned interaction with syntactic inferences, they also need close cooperation with two other kinds of inference: the semanticlogical inferences and the action-oriented inferences. :?3 On semantic-logical inferences

Semantic-logical inferences serve to handle the semantic information of ILRRR in more specific ways. Thereby, they can make use of the tools provided by the standard systems of logic. In so doing, semantic-logical inferences help the discourse receiver to disclose: (A) What individual underlying propositions belong to a discourse di; (B) what predicates and arguments can be identified in the individual underlying propositions of da (C) what intrapropositional relations exist between the identified predicates and arguments of the individual underlying propositions of di; (D) what interpropositional relations hold between the individual underlying propositions of da (E) what extrapropositional relations may be assumed for the individual underlying propositions of di. (These extrapropositional relations depend on the local higher qualifiers which can be termed 'hypersentences', 'modal qualifiers, 'attitudinal qualifications', 'propositional qualifications', 'prepositional attitudes' and the like); (F) what global qualifier is to be surmised for the whole propositional content of the discourse di; (G) what organizational properties/relations exist in/beyond the whole propositional content of the globally qualified discourse di. etc. As far as point (G) is concerned, it is obvious that there are organizational properties/relations which pertain to the internal/external aspects of the propositional content of di. The following points (a) — (c) are concerned with the internal aspects, whereas (d) —(e) relate to the external aspects: (a) Can the total propositional content of di be interpretatively reduced to a macroproposition denoting the core meaning of di? (b) Is the total propositional content of di compatible with he superscript/ global topic/global theme/title of di? (c) Are the consequences which can be drawn from the discourse di consistent with the total propositional content of di?

294

R. G. van de Velde

(d) Can the total prepositional content of di be integrated into the wider action contexts/pragmatic contexts/social contexts of di? (e) Does the total propositional content of di correspond with the discourse receiver's pertinent prior knowledge on the issues raised by di? etc. Dwelling on the points (a) — (e) would require a detailed exposition. Because of space limitations we cannot give illustrations and further specifications of (a)-(e). We refer the interested reader to van de Velde (1986; 1987). With respect to point (F), it should be emphasized that any discourse may be globally qualified by higher predicates, such as ASSERT, KNOW, BELIEVE, DOUBT, DENY, and the like. On the one hand, these global higher predicates may conform to the discourse producer's truth assumptions, beliefs, attitudes, etc. regarding di. On the other hand, they may also correspond to the discourse receiver's globally assigned truth values, etc. regarding di. It goes without saying that the discourse receiver's truth assumptions, beliefs, attitudes, etc. can coincide as well as conflict with those of the discourse producer. Whether they harmonize or not is not our concern here. We only want to emphasize that ASSERT (9), KNOW (9) and BELIEVE (9) indicate that the total propositional content of (9) is true, whereas it is false in DENY (9) and undecidable in DOUBT (9). Similarly, with respect to point (E), local higher predicates qualify the individual underlying propositions of (9). They are of particular importance for the propositions satisfying the condition of adjacency. They will have repercussions on those semantic-logical inferences which contribute to the indentification of the interpropositional relations. With respect to (E) —(G), the spatio-temporal predicates should also be specified. The reasons are obvious: what a person as discourse producer/ receiver KNOWS, BELIEVES, etc. at a specific moment rm need not be identical with her/his previous or future KNOWLEDGE DATA, BELIEFS, etc. Similarly, the content of a discourse producer/receiver's KNOWLEDGE, BELIEF, etc. may change depending on the different locations in which (s)he is. With respect to points (A) —(D), we surmise that the semantic-logical inferences become interactive with syntactic inferences and with ILRRR inferences in the following ways: (f) The interaction with syntactic inferences amounts to finding out in (9) that the syntactic constituents the young woman, the colloquium participant, etc. serve as arguments, whereas WANTED TO SAY GOODBYE, HAD TO LEAVE ABRUPTLY, GLANCED, COULD NOT GUESS THE REASONS, etc. function as predicates. (g) In the interaction with ILRRR inferences semantic information of ILRRR which is characteristic of the separate arguments and predicates is made use of for discerning which arguments/predicates are identical, overlap or include each other.

Inferences as (De)compositional Principles

295

If semantic-logical inferences can be made in the ways suggested by (f) — (g), then inferential processing of (9) yields the information that the young woman, she, her, she, she, she, and her which belong to different underlying propositions of (9) satisfy the condition of argument identity. In similar ways, it can be derived from (9) that the colloquium participant must be a male human being and that the syntactic proforms him, his, he and he are denoting one and the same argument. Finally, along the lines of (f)~(g)> it becomes possible to identify the predicates of (9) and their respective overlaps, such as between GLANCED, HAD OBSERVED and LOOKED; between TO LEAVE ABRUPTLY and UNFORESEEN DEPARTURE; between BEING WRAPPED IN MYSTERY and COULD NOT GUESS THE REASONS; and between BEING COMPLETELY OUT OF HIS MIND and BEING WRAPPED IN MYSTERY As is suggested in point (c) of (G), semantic-logical inferences also fulfill the task of providing answers to the questions 'in general what follows from a discourse d,?' and 'which particular consequences can be drawn from a discourse di?'. In (10) —(16) we illustrated already some consequences drawn from the discourse (9). The consequences drawn from the total propositional content of di are partly dependent on the consequences drawn from its individual underlying propositions. We shall briefly illustrate this point for (37) which is an excerpt of (9): (37)

He was not blind to her attractiveness

In the semantic-logical account of (37) two main predicates, viz. BLIND and ATTRACTIVE are assigned to the respective arguments he / the colloquium participant and her / the young woman. The predicate BLIND is within the scope of negation of the predicate NOT. The spatio-temporal predicates concern the environmental properties of the auditorium/building/room where the colloquium has been held, on the one hand, and the past events/actions of the colloquium setting, as they are described and suggested by the discourse (9), on the other hand. We presume that semantic-logical inferences permit us to draw, among other things, the following consequences from (37): (38) (39) (40) (41)

The young woman is held to be attractive (at least by the writer/speaker who utters (37)) The colloquium participant is sensitive to the young woman's attractiveness Probably the colloquium participant is a person with reasonable vision Probably the colloquium participant has very good spatial percepts and concepts etc.

In comparing (38) —(39) with (10) —(16) it is clear that the consequences (15)-(16) are partly dependent on (38)-(39). The propositions (38)-(39) do, in fact, follow by deduction from (37), whereas (40) —(41) are examples of inductive inference characterized by the higher modal qualifications of probability.

296

R. G. van de Velde

Interpretative elaborations of a discourse dt are to a large extent due to inference-making processes utilizing the inductive method. If the inductive method is used with respect to the inferential processing of (37) and (9), then at least the propositions (42) —(48) may be hypothetically derived: (42) Probably the young woman is a listener at the colloquium (43) Probably the young woman is also a lecturer at the colloquium (44) Probably the young woman is giving an interesting talk (45) Probably the colloquium participant does not concentrate exclusively on the young woman's talk (46) Probably the colloquium participant concentrates on the young woman's facial expression (47) Probably the colloquium participant is not a male chauvinist (48) Probably the colloquium participant is a semanticist who does not care about nonhuman possible worlds etc.

Some propositions can be selected from (42) —(48) to serve in the interpretative elaborations of (37) and (9). As noted earlier, there is no upper limit to the applicability of reasoning processes [2.]. Very often these reasoning processes lead the discourse receiver to COMPOSE interpretative elaborations. Interpretation of poetry is a good example to confirm this view. Whatever interpretative elaborations are made in the sense that propositions are interpolated in the original version di, they are severely constrained by the fact that they must be locally consistent with the individual underlying propositions of dj and that they must be globally consistent with the total prepositional content of di. Semantic-logical inferences, as they are treated here, are a suitable means to control the local/global consistency of the interpretative elaborations. In addition, ILRRR inferences place constraints on the interpretative elaborations of di. As we have seen already in connection with (27) and (28) —(36), the consequences drawn from the propositional content of di must be semantically congruent with the inferences which have recourse to the pertinent ILRRR information. In this respect, semantic-logical inferences must interact with ILRRR inferences (and with action-oriented inferences). Their interactive functioning then provides an appropriate Instrumentarium for controlling the semantic congruence both of the underlying propositions of di and of the interpolated propositions with the pre-existing ILRRR knowledge sources of the discourse receiver. We shall return to this point in [4.1.-4.2.]. 3.4. On action-oriented inferences

At this stage of the discussion the question must be raised whether the aforementioned kinds of inference provide all-round tools and satisfactory means for a complete and successful 'descriptive interpretation' of discourse. We surmise that they do not. Let us deal with this surmise by commenting on (49)-(50):

Inferences as (De)compositional Principles (49) (50)

297

The boys threw the ball through the window of the apartment building The janitor was angry and ran after them

We submitted 80 graduate students to an interpretation task of (49) —(50). They were asked whether they could recognize different sources of information in (49) —(50) and whether in interpreting (49) —(50) they would connect these different information sources. In addition, they were requested to justify the assigned connections. No time limits were imposed. All answers had to be written on a standard examination paper. We hypothesized that the decompositional operations pertaining to syntactic and semantic processing would yield some connections between (49) — (50) and their underlying propositions. All answers confirmed this hypothesis. In syntactic processing, the 80 discourse receivers could identify the syntactic relations between the boys and them. Moreover, they were able to recognize that the syntactic order of the action verbs was in accordance with the temporal order of the actions described in (49) — (50). In semantic processing, the ILRRR inferences concerned with the content words apartment building and janitor as well as the semantic-logical inferences concerned with the argument identity of the boys and them provided hints for their connecting (49) with (50). The more specific question which must be raised now is whether syntactic processing and semantic processing can provide all the required information parts to connect (49) with (50). We conjecture that they cannot: 90% of the answers confirmed this conjecture. If (49) —(50) must be considered as connected discourse, then supplementary information sources must be taken into account. The reasons have to do with the descriptive interpretation of the discourse-internal actions as they are denoted by (49) —(50). We refer to the descriptive interpretation of discourse-internal actions by the term 'action processing'. In the following, we shall demonstrate that action processing provides additional information parts for constructing connections between (49) and (50) and that action-oriented inferences are responsible for action processing. Action processing of (49) — (50) is to a large extent concerned with solving the problems of causality and intentionality as raised in (51) —(54): (51) (52) (53) (54)

What are the causes of the janitor's being angry? Did the boys have the intention to throw the ball through the window of the apartment building? What goals/plans/grounds did the janitor have in running after the boys? What goals/plans/grounds did the boys have in running away?

We presume that solving the problems (51) —(54) is dependent on the functioning of action-oriented inferences (in interaction with syntactic, ILRRR, and semantic-logical inferences). For instance, action-oriented inferences (in interaction with syntactic, ILRRR and semantic-logical inferences) may give rise to the following consequences:

298

R. G. van de Velde (55) (56)

Probably there was an open space (e.g. a lawn) in front of the apartment building Probably the boys played with the ball in the open space in front of the apartment building (57) Probably the ball was not larger in size than the window of the apartment building (58) Probably the boys made a great deal of noise in playing with the ball (59) Probably the boys destroyed the lawn in front of the apartment building (60) Probably the boys' actions were a nuisance to the tenants of the apartment building (61) Probably the tenants of the apartment building had complaints about the boys' actions (62) Probably the tenants of the apartment building had asked the janitor to prevent the boys from playing with the ball in the open space in front of the apartment building (63) Probably the janitor had forbidden the boys to play with the ball in the open space in front of the apartment building (64) Probably the boys were very angry because of the complaints of the tenants (65) Probably the boys were very angry because of the janitor's prohibition on playing with the ball in the open space in front of the apartment building (66) Probably the boys had goals/plans in protesting against the complaints of the tenants (67) Probably the boys had goals/plans in protesting against the prohibition of the janitor (68) Probably a peaceful dialogue between the janitor and the boys had become impossible (69) Probably the janitor had threatened the boys (70) Probably the boys reacted violently against the janitor's threats by throwing the ball through the window (71) Probably the window-pane was smashed completely (72) Probably the janitor was angry because of the smashed pane (73) Probably the janitor was angry because the boys ignored his prohibition/threats (74) Probably the janitor supposed that the repairs would cause difficulties for the tenants of the apartment building (75) Probably the janitor was afraid that the repair to the window-pane would cause high costs for the tenants of the apartment building (76) Probably the janitor was afraid that the tenants of the apartment building would complain about the fact that they would have to share the repair costs of the broken window-pane (77) Probably the janitor had several plans/goals/concerns in running after the boys (78) Probably the boys ran away because they were afraid they would have to pay for their destructive actions etc.

The following points are worth consideration: a) It is possible to identify properties of coherence for the actions described in (49) —(50). We refer to these properties by the term 'action-bound coherence' (van de Velde, 1984); b) in constructing action-bound coherence for (49) —(50) the discourse receiver must select particular consequences from (55) —(78); c) in deciding what consequences will fit in most appropriately with the action cotext of (49) —(50), the discourse receiver's selections are restricted by her/his requisite pertinent prior ILRRR knowledge and by rational argument; d) if the consequences selected from (55) —(78) are inappropriate, then the consequences drawn from (49) —(50) can still be extended and numerically augmented in many ways. With regard to point (d), it is undeniable that the inferential steps leading to the consequences (55) —(78) and to their potential extensions/augmentations are responsible for the development of interpretative elaborations of (49)-(50).

Inferences as (De)compositional Principles

299

Interpretative elaborations are very dependent on the functioning of actionoriented inferences. This is mainly due to the fact that action-oriented inferences in interacting with syntactic, ILRRR and semantic-logical inferences integrate the information sources provided by these inferences. In addition, as we shall see subsequently, action-oriented inferences have a much wider range of application than other kinds of inference. In the following, we shall restrict our account of action-oriented inferences to the four facets of action summed up in (H) —(K): (H) Language expresses information about actions. Thus, we speak about Verbally' or 'cotextually expressed actions' (also termed 'action cotext'). (I) Language is used in action contexts. (J) Verbally expressed actions as well as action contexts include underlying cognitive/affective/conative background activities (intentions, wants, plans, expectations, etc.) of the agent(s) involved. (K) There are innumerable cases where actions are not verbally expressed or do not occur in the action contexts of language use. In those cases, the discourse receiver must reconstruct hypothetically the pertinent actions. These hypothetically reconstructed actions must then be incorporated in the co(n)texts of action interpretation. Action-oriented inferences are directed to (H) —(K). In the following paragraphs we shall comment very briefly on their interactive and integrative functioning with ILRRR, syntactic and semantic-logical inferences. In the first place, we shall consider the question as to HOW action-oriented inferences are directed to the verbally expressed information about actions mentioned in (H). For this purpose, let us start from example (79): (79)

Arthur ploughs. Last week he spread manure. Next week he will sow. In a few months he will harvest

Any competent discourse receiver can derive as action-oriented information from (79) that different agricultural actions took place, that these agricultural actions have a well-circumscribed temporal order, that these agricultural actions display definite causal connections, that some preparatory actions must have preceded the agricultural actions of (79), that some actions will follow the agricultural actions of (79), that the agricultural actions of (79) enable the carrying out of other future actions, etc. We will restrict our illustrations to the agricultural action of PLOUGHING denoted by the predicate in the first utterance of (79). When the discourse receiver can rely on pertinent ILRRR information concerning PLOUGHING (s)he is able to make the following action-oriented inferences: (80) (81)

Probably the action of ploughing is a partial manifestation of several agricultural action sequences Probably Arthur has cognitive/affective/conative background activities (goals, intentions, plans, wants, etc.) in turning up the soil

300

R. G. van de Velde (82) (83) (84)

Probably the work of turning up the soil has a social function in providing the population with food Probably Arthur is a farmer Probably the land will no longer lie fallow etc.

It is beyond any doubt that consequences such as (80) —(84) are of importance when action-bound coherence must be established in fragments of texts like (79) which describe agricultural actions pertaining to the production of food, and the like. In the second place, we shall devote attention to the action contexts mentioned in point (I). From these diverse/extensive domains of action contexts we shall single out these actions which cooccur with discourse production/reception. We term them 'coverbal actions'. Coverbal actions are manifested in dialogues, conversations, debates, interviews, colloquia, seminars, verbal exchange in the classroom, in the court room, and the like. The coverbal actions which language users carry out provide important cues for making action-oriented inferences. This becomes strikingly clear when we consider how we derive information from the coverbal actions which actors perform in the theatre and from social behaviour situations in which many people behave as actors. Language use is only a part of human action. Consequently, there is a wide variety of coverbal actions. They encompass not only the gestures which accompany spoken language utterances. They also include gaze direction, eye contact, posture, postural shifts, and facial expression. More closely related to language expression are voice raising, tempo, and intonation. Because the present paper is concerned only with written language documents we cannot go into coverbal actions here. Space limits also prevent us from touching upon non-verbal communicative actions, such as smiling, blinking, nodding, jerking up the eyebrows, raising the shoulders, stroking the nose with one finger, etc. They all can be taken to mean something in the social setting in which they occur. In this regard, the social context of verbal intercourse must be taken into consideration (Franck, 1980; Corsaro, 1981; Fritz, 1982). In the third place, we shall direct our attention towards the aspects of action mentioned in domain (J): verbally expressed actions of domain (H) and/or actions cooccurring with language utterances belonging to domain (I) are often made possible by or go together with underlying cognitive/affective/ conative background activities (intentions, goals, plans, expectations, etc.). In this respect, the following question should be raised: How can the discourse receiver infer from the actions described by the discourse under consideration and/or from the context in which language is used what the underlying cognitive/affective/conative background activities of the agents are? The disclosure of the underlying cognitive/affective/conative background activities is often a prerequisite for an in-depth-interpretation of action

Inferences as (De)compositional Principles

301

cotexts. So, for instance, action processing of (9) requires that the discourse receiver takes into account the conditionals (85) —(88): (85) (86) (87) (88)

If χ wants to say goodbye, then χ has the intention of leaving If y observes x, then y selects those outward characteristics of χ which fit in with y's intentions/expectations/ If y observes that χ looks as if wrapped in mystery, then y searches for the reasons which may explain the outward characteristics of x's looking mysterious If y finds out that χ will leave abruptly, then y searches for the reasons for x's unforeseen departure etc.

In reading (9) the discourse receiver will experience that the antecedents of (85) —(88) turn out to be true. Accordingly, the argumentation schema of modus ponens may apply such that the consequences of (85) —(88) become true. In the course of this, the discourse receiver arrives at conclusions which may be interpolated in the propositional content of (9). The example (89) illustrates approximately how this might be done. The interpolated propositions are written between square brackets. Their tense is adapted. The separate utterances of (9) are written in upper case letters: (89)

THE YOUNG WOMAN WANTED TO SAY GOODBYE TO THE COLLOQUIUM PARTICIPANT. [The young woman had the intention of leaving.] WHEN SHE GLANCED AT HIM HE WAS COMPLETELY OUT OF HIS MIND: ALL DAY HE HAD OBSERVED HER IN THE AUDITORIUM. HE WAS NOT BLIND TO HER ATTRACTIVENESS. [The colloquium participant selected those outward characteristics of the young woman which fitted in with his intentions/ expectations.] SHE LOOKED AS IF SHE WERE WRAPPED IN MYSTERY. [The colloquium participant searched for the reasons which might explain the outward characteristics of the young woman's looking mysterious.] NOW IT TURNED OUT THAT SHE HAD TO LEAVE ABRUPTLY. [The colloquium participant searched for the reasons for the young woman's unforeseen departure.] HE COULD NOT GUESS THE REASONS FOR HER UNFORESEEN DEPARTURE.

It might be asked whether an in-depth-interpretation of (9) should take into account the underlying cognitive/affective/conative background activities in the ways sketched before. We hypothesize that it should. We checked this hypothesis by comparing the 48 interpretations the graduate students proposed for (9) with the interpretative elaboration of (89). All answers indicated that the underlying cognitive/affective/conative background activities as suggested in (85) —(88) and in (89) were taken into consideration by the interpreters. In the fourth place, we shall pay attention to two questions concerned with the domain (K): is the hypothetical reconstruction of actions necessary? Must the hypothetically reconstructed actions be integrated in the verbal cotexts? Let us look at (90) which is excerpted from (91) in order to illustrate the need for hypothetically reconstructed actions and for their interpretative integration into wider co(n)texts: (90)

... her letter came to him like a springtime sonata ...

302

R. G. van de Velde

In the cited example (90), nothing is explicitly propounded on performed actions. Does that mean that we can dispense with action-oriented inferences in the interpretation of (90)? We surmise that we cannot. Assigning a meaning to (90) requires that this meaning be integrated into the action information as it is denoted in the prior and posterior discourse fragments of (91). Moreover it must also be incorporated in the wider action contexts, but due to space limitations we cannot treat this requirement here (cf. van de Velde, 1984). (91)

For several weeks he had expected to hear from her. If she only would have left a note for him. Every morning he had run to his letter box. He realized now how anxious he was to know more about her views on the essentials of the human psyche. He knew that she was a graduate psychologist and a writer of literary discourse. °°° Her letter came to him like a springtime sonata. °°° He had observed her at a colloquium held at the end of summer. He had dreamt so many times of having communicative contact with her. He even hoped to develop a spiritual friendship with her ...

Undeniably, action processing of (91) requires that the interpreter hypothetically reconstructs non-expressed action data. From these hypothetical reconstructions the interpretative elaborations of (90) and of (91) must be developed. The hypothetical reconstruction of actions is not a free enterprise. It is predominantly restricted by the information sources of ILRRR. For instance, the knowledge data stored in ILRRR with the lexical item letter constrain in many respects the interpretation of (90). The following propositions (92) —(99) exemplify admissible interpretative extensions: (92) (93) (94) (95) (96) (97) (98) (99)

Some female person must have written a verbal message for a male person Some female person must have sent that verbal message to a male person by mail Some male person must have received the verbal message written and sent by the female person The female person will probably have used an instrument to write on the sheet(s) of paper The female person will probably have moved her arm and hand in writing down the utterances The female sender will probably have put the sheet(s) of paper into an envelope The female sender or someone else will probably have taken the letter to the post office or to a mail box A postman will probably have put the letter into the letter box of the male receiver etc.

Similarly, ILRRR information on RECEIVING A LETTER, HAVING AESTHETIC IMPRESSIONS, ENJOYING SPRINGTIME, and the like, are responsible for the constraints on the hypothetical reconstruction of actions, as is illustrated by (100) —(109): (100) (101) (102)

The male receiver will probably have opened the envelope The male receiver will probably have taken the sheet(s) of paper from the envelope The male receiver will probably have unfolded the sheet(s) of paper enclosed in the envelope

Inferences as (De)compositional Principles (103) (104) (105) (106) (107) (108) (109)

303

The male receiver will probably have read the letter with great satisfaction The male receiver will probably have heard a springtime sonata a long or short time before he read the letter The springtime sonata will probably have evoked positive effects on the male receiver The male receiver will probably have thought of the positive effects he experienced with the springtime sonata when he read the letter sent by the female person The male receiver may probably have thought of the fact that springtime has the specific effect of bringing new life in nature The male receiver may probably have felt that the letter sent by the female person had a springtime-like effect on his own emotionality and intellectual creativity The male receiver may probably have experienced a deep emotional involvement in reading the verbal message sent to him by the female person etc.

It is clear that not all the hypothetically reconstructed information on actions, as it is exemplified in (92) —(109), must be integrated in the cotextual information of (90) and (91). It is precisely in this respect that interpretative elaborations turn out not to be a free enterprise. Regarding this issue, we may adapt Goethe's famous words as follows: "The proof of the master of interpretation lies in the limitations and it is the law only that provides us with freedom". The limitations which any master of interpretation must face do not only concern the domains of hypothetically reconstructed actions. It is selfevident that these limitations as well as the law that provides us with freedom depend also on the kinds of inference we have discussed in this paper as well as on their interaction and integration and on the knowledge sources they rely on. In the next section, we shall go into these knowledge sources and into their connections with different processing strategies. 4. On processing strategies and knowledge sources 4.1. On processing strategies

The earlier parts of this paper have given prominence to the questions of how inferences determine the decompositional operations and how these decompositional operations allow the discourse receiver to disentangle the syntactic, semantic and action-oriented information of discourse. In the cognitive sciences, the decompositional operations are subsumed under the general headings of bottom-up and on-line processing strategies. Bottom-up and on-line processing strategies concern the analytical and sequential perspectives of discourse processing. In the following sections, we will show that they must be supplemented by holistic processing strategies. In current writings of the cognitive sciences, these holistic processing strategies are usually referred to by the term 'top-down processing'. Holistic processing strategies are to a large extent responsible for the compositional operations of discourse processing. To this end, they need interaction with analytical and sequential processing strategies. The notion 'interaction' denotes the exchange of information which emerges and indeed

304

R. G. van de Velde

must emerge from the cooperating inferences and from the pre-existing knowledge sources they rely on. As far as the pre-existing knowledge sources are concerned, we cannot go into the various sorts of information which the discourse receiver may invoke. We shall limit ourselves to distinguishing 'socioculturally determined knowledge' from 'private knowledge' [4.2.—4.3.] and to pointing to their roles in the interpretation of different types of discourse [4.4]. 4.2. On socioculturally determined knowledge

Socioculturally determined knowledge is usually referred to in script-based semantics by terms such as 'frames' (Minsky, 1975), 'scripts' and 'plans' (Schank, Abelson, 1977), 'schemata' (Rumelhart, 1980), 'MOPS' (Lytinen, Schank, 1982), and the like. We can connect the views of script-based semantics with our own views as follows: if socioculturally determined knowledge is related to lexical items, then it is part of ILRRR [3.2.]. For instance, if language users know what the lexical item station means, then they possess standard semantic information on schemata of actions, states, events, etc. which particularly pertain to station and which is closely connected with the higher level (action) schema of TRAVELLING BY TRAIN. As far as the organization (in the ILRRR representation) of this standard semantic information is concerned, we speak about 'schema-based knowledge'. In our view, schema-based knowledge plays a critical role in those compositional operations which guide the interpretative elaborations of a discourse di. Let us clarify this view by discussing the following example: (110)

John ran to the station. In the hall three elderly persons were standing in front of the ticket office to get a ticket for a railtour. He was afraid he would miss the Paris —Amsterdam train which was due in at 9.51 h. It was already ten minutes to ten. It seemed to be a long time before he could order a single ticket. After that, he had to queue before getting on the platform. He was glad to observe that many people were still waiting on platform six from which the train would leave. He saw that the train had already pulled into Antwerp Central. He was the last passenger to get on the train. All seats were taken.

In reading through (110) we may select different focal points (actions, events) and ask inference questions such as WHAT HAPPENED? HOW DID IT HAPPEN? WHY DID IT HAPPEN? etc. (Warren et. al., 1979). Let us place the first focal point in (110) after John ran to the station. We surmise that providing answers to these inference questions requires that the reader makes inferences as exemplified in (111) —(117). We conjecture that (111) —(117) cannot be inferred from John ran to the station, if the discourse processor has no pertinent socioculturally determined knowledge about the conditions, requirements, goals, enablements, causes, consequences of TRAVELLING BY TRAIN:

Inferences as (De)compositional Principles (111) (112) (113) (114) (115) (116) (117)

305

John hurried up to a stopping-place on the railway Presumably John wanted to book a seat in the station Presumably John wanted to catch a train Probably the station had a ticket office Probably the ticket office could be situated in the hall of the station building Probably the employee sat behind the window of the ticket office to sell tickets Probably the employee of the ticket office took money, changed it and gave tickets etc.

If we place the focal point after each of the separate utterances of the posterior parts of (110), then an unlimited number of consequences can be drawn from them. A great many of these consequences have recourse to schema-based ILRRR information on TRAVELLING BY TRAIN. If we inserted in (110) all the semantic information provided by these consequences, then (110) would be transformed into an overcomplete interpretative elaboration. For lack of space we cannot spell out all these consequences. We shall limit ourselves to a short illustration of an interpretative elaboration of (110). In so doing, we will show how the compositional operations may lead the discourse receiver to (re)construct the expanded version (118). In this expanded version we interpolated two admissible consequences after each separate utterance of (110). The interpolated propositions are presented between square brackets. The separate utterances of (110) are written in upper case letters. (118)

JOHN RAN TO THE STATION. [Presumably John wanted to book a seat in the station.] [Presumably John wanted to catch a train.] IN THE HALL THREE ELDERLY PERSONS WERE STANDING IN FRONT OF THE TICKET OFFICE TO GET A TICKET FOR A RAILTOUR. [John could predict that he would have to wait in front of the ticket office.] [Probably John had to wait until the employee who sat behind the window had given the tickets needed for the railtour to the three elderly persons.] HE WAS AFRAID HE WOULD MISS THE PARIS -AMSTERDAM TRAIN WHICH WAS DUE IN AT 9.51 H. [Presumably it was the prognosis on the inevitable waiting at the ticket office which caused John's fears about missing the train.] [Missing the train would mean that John would not be able to achieve his goal(s).] IT WAS ALREADY TEN MINUTES TO TEN. [Only one minute was left for John to catch the Paris —Amsterdam train.] [It would be necessary for the employee to change money and to give a ticket in a few seconds in order to enable John to achieve his goal(s).] IT SEEMED TO BE A LONG TIME BEFORE HE COULD ORDER A SINGLE TICKET. [Presumably John got the impression that his having to wait so long would cause his missing the Paris —Amsterdam train.] [Presumably the impression of having to wait so long caused John to get more nervous.] AFTER THAT, HE HAD TO QUEUE BEFORE GETTING ON THE PLATFORM. [Probably John could not make things go faster.] [Probably John regretted that he had to endure these unfavourable circumstances.] HE WAS GLAD TO OBSERVE THAT MANY PEOPLE WERE STILL WAITING ON PLATFORM SIX FROM WHICH THE TRAIN WOULD LEAVE. [John could surmise that the Paris —Amsterdam train had not yet left the station.] [Probably John was able to calm down.] HE SAW THAT THE TRAIN HAD ALREADY PULLED INTO ANTWERP CENTRAL. [John could expect that the Paris-Amsterdam train was not a goods train but a passenger train.] [John could presume that he was in time to catch the Paris - Amsterdam train.] HE WAS THE LAST PASSENGER TO

306

R. G. van de Velde GET ON THE TRAIN. [The door of the train was probably closed after John's boarding.] [The train would probably leave after John's boarding.] ALL SEATS WERE TAKEN. [Presumably John did not get out because of all the taken seats.] [Probably John had to stand all the way.]

As can easily be seen, the interpretative elaboration illustrated in (118) contributes to the hypothetical reconstruction of action contexts. In this respect, the discourse receiver of (110) does not draw all the possible and admissible consequences from the individual underlying propositions of (110). On the contrary, three main constraints can be placed on the compositional operations: 1) Only those inferences may/must be made which are semantically congruent with the pre-existing pertinent ILRRR knowledge sources needed for the interpretation of (110); 2) in forward inferencing only those inferences may/must be chosen which are relevant to/consistent with the posterior parts of (110); 3) in backward inferencing only those consequences may/must be drawn which are relevant to/consistent with the earlier parts of (110): In order to control whether the interpolated propositions of (118) could be considered as an integrative part of the compositional operations directed to (110) we presented 48 graduate students with (110) and (118). They were asked to decide whether their schema-based knowledge about TRAVELLING BY TRAIN allowed them to agree that (118) was an appropriate interpretative elaboration of (110). In advance, we had checked whether they shared socioculturally determined knowledge about TRAVELLING BY TRAIN. No time limit was imposed. The answers were written on a standard examination paper. They all confirmed our hypothesis that compositional operations rely to a large extent on schema-based knowledge. 4.3. On private knowledge

Notwithstanding the intersubjective agreement which can be achieved on the role of compositional operations, there still may be a number of individual differences and variations in the interpretation of discourse. These differences and variations are dependent on personality factors. As far as these personality factors are related to the discourse receiver's erudition, her/his ability to read between the lines, her/his propensity for empathic interpretation, her/his critical talents, etc., they may invoke a second kind of knowledge which we have termed 'private knowledge'. Private knowledge pertains to the private meanings which a person has in his/her ILRRR. For instance, Einstein had private meanings about the words time and space which differed from those of the other members of the linguistic community he lived in. By 'private knowledge' we mean, among other things, familiarity with specialized fields in the world of scholarship, long-term experience in specific

Inferences as (De)compositional Principles

307

disciplines, acquaintance with fine arts, and the like. As far as this private knowledge is related to lexical items, it is part of ILRRR [3.2]. Let us illustrate the role of private knowledge by considering example (119): (119)

He sat in an auditorium listening to a lecture which was remarkably illustrated with reproductions of paintings and graphic work. Slowly but surely he became more and more involved in an impressively selfimbued aesthetic experience. A young woman sat before him and stretched her arm in an attractive way. This graceful movement came to him as an invitation to use his visual senses in a trance as some painters do when facing their model over a long period of time. His glance followed the contours of her fine fingers, her slender arm and so upwards to her neck and short-cut hair. Involuntarily he associated this superb vision with the mysterious gesture of Michelangelo's Adam on the fresco of the Sistine Chapel. Renaissance ideals on harmony and beauty came to life in the person of this auditorium nymph. The enchantment fo female grace as painted by Tiziano, Giorgione, da Vinci and Raphael was called to his mind.

If an appropriate meaning is to be assigned to (119), then the individual discourse receiver must rely on her/his private knowledge about (the history of) fine arts and about the events which take place in an auditorium. Limitations of space prevent us from spelling out the propositions which can be inferred from (119). Needless to say, these inferred propositions will display more individual differences and variations than was the case with the interpretation of (110). In this respect, we presume that the pre-existing private knowledge sources of an art historian will lead to particular conclusions which will be distinct in many details from those of a psychologist. Here, the following important point should be made: if convergences/overlaps between individual interpretations of (119) occur, then they will largely depend on the commonalities in the private and socioculturally determined knowledge about what characterizes Renaissance art and which painters are representative of it. As we have suggested above, the compositional operations which are responsible for these convergences/overlaps are critical to top-down processing. 4.4. On pre-existing knowledge and discourse types

In dwelling on the examples (110) —(119), we could lift only a very small part of the enormous veil which hides the complex interrelationships of knowledge, forward/backward inferencing, and discourse processing (van de Velde, 1984; 1986; 1987). These interrelationships are the more complex the more we take into account the impact of knowledge in interpreting highly diversified types of discourse. For clarificatory purposes we distinguish three types of discourse: 1) literary discourse, 2) scientific discourse, 3) so-called pathological discourse. In each of these three discourse types the specific role of the receiver's pre-existing knowledge is of crucial importance. As far as literary discourse is concerned, we suppose that the interpreters often take into consideration knowledge about the author, her/his cultural

308

R. G. van de Velde

setting, her/his connections with literary movements, etc. In addition, literary interpretation may be based on sensitivity to literary symbols, on special gifts of empathy, on imaginative power, on familiarity with fictionality, and the like. Even knowledge of the literary genre and mastery of the literary codes may determine the receiver's understanding of literary discourse. As far as comprehension of scientific discourse is concerned, we surmise that the reader needs highly specialized knowledge about the subject matter. Moreover, her/his interpretative endeavours will be subjected to (the knowledge about the) rigorous rules of scientific argumentation. As far as so-called pathological discourse is concerned, we must draw considerable distinctions between those disturbances which reflect personality disorders (such as schizophrenia) and those impairments which are neuropathological (such as aphasia). Space limitations restrict us to making the following point: whatever deviances occur in the discourse of schizophrenics, aphasics, etc., any serious attempt at examination, assessment, understanding and treatment cannot but take into account these deviances. In these endeavours, the discourse receiver (psychiatrist, neurologist, aphasiologist, therapist, etc.) will need expert knowledge of the disabilities at hand. In addition, longterm experience, an open mind and complete attention will be of immense help in achieving diagnostic and therapeutic aims. Continuing with this topic, we shall close our exposition by reinterpreting experimental data obtained with respect to the disordered functioning of inferential processes. 5. On disordered inferential processing

Disorders in discourse processing may be due to linguistic and cognitive deficits. We shall be concerned only with those deficits which pertain to the kinds of inference discussed in this paper. From the experimental data we have reviewed elsewhere (van de Velde, 198la; 1984; 1987) we select the scrambled story test (Huber, Gleber, 1982). The results of this test provide answers to the questions as to bow the interactive and integrative functioning of inferences can be disturbed and what we may learn from these disturbances. The Huber-Gleber-test comprises nine stones. They are each presented in three versions. The first version consists of six pictures which, for instance, show what happens to a man walking with his dog in the street. We refer to it as the pictorial version. The other two versions are verbal in nature. Each of them has six utterances which are not arranged in the right order. For the verbal versions the authors draw a distinction between so-called 'redundant' and 'non-redundant' discourse fragments. The non-redundant verbal version contains only six simple utterances. Let us call the nonredundant version exemplified in (120) —(125) the type A version: (120) (121)

A man is hit on the head by a flower pot The man scolds the woman on the balcony

Inferences as (De)compositional Principles (122) (123) (124) (125)

The The The The

309

man and his dog enter the house man knocks on the apartment door woman gives a dog a bone man kisses the woman's hand

The other verbal version is a partial interpretative elaboration. We call this expanded version illustrated in (126) —(131) the type B version: (126) (127) (128) (129) (130) (131)

While walking down the street with his dog, an elderly gentleman was suddenly hit on the head by a flower pot falling from a balcony Both master and dog scolded the culprit on the balcony at the top of their voices In order to teach him a lesson, they entered the house purposefully They stopped on the second floor and the gentleman knocked on the apartment door with his cane An elderly lady opened the door and, full of sympathy, comforted the dog with a bone This made the gentleman feel good and he gallantly kissed the lady's hand

Huber and Gleber confronted 108 subjects with the pictorial and verbal versions of the scrambled story test. The subjects were divided into six groups. Each group consisted in 18 subjects: 18 Broca's aphasics, 18 Wernicke's aphasics, 18 amnesics, 18 global aphasics, 18 normals, and 18 right-hemisphere patients. Sometimes the pictorial version was given first and sometimes the verbal versions A and B were presented in alternating order before the pictoral version. Each of the separate utterances in the A version, such as (120) —(125), and in the B version, such as (126) —(131), correspond to a single picture of the pictorial version. The subjects were required to rearrange the unordered sets of pictures and the corresponding scrambled utterances of the A and B versions correctly. Patients who did not comprehend their task on demonstration were excluded. Let us now ascertain in what ways the experimental design and its results relate to our views on inferences. Above all, they conform to our views on action-oriented inferences. The pictorial version and the A and B versions have the common characteristic that they denote actions. We shall restrict our comments here to the discourse-internal actions. These discourse-internal actions are connected by causal chains. Action-oriented inferences must serve here to reconstruct the original order of the separate causes. In addition, action-oriented inferences are required here to surmise what transition between the separate causes renders the most appropriate causal connections. To this end, information emerging from the cooperation between ILRRR inferences and semantic-logical inferences must be integrated into the information data about the causal connections provided by the action-oriented inferences. The required inference-making processes may, for instance, start from conditionals, like those exemplified in (132) —(134): (132) (133)

If a person pi is hit on the head by a flower pot, then that person pi presumably looks for a cause of his being hit If a person pi finds a cause in making another person p2 responsible for the flower pot falling down, then the person pi will presumably address his complaints to the person p2

310 (134)

R. G. van de Velde If (132) and (133) are given and if the person p2 lives in an apartment building or house and if that apartment building or house has a balcony, then the person pi will presumably scold the person p2 who is present on the balcony etc.

It is beyond doubt that the action-oriented inferences must also enable the subjects to infer from the scrambled information data what temporal order is to be assumed for the described actions. This temporal order must then fit in with the hypotheses which the subjects had constructed on the effects of the separate causes and on the causal connections. The recovery of the causal chains and the detection of the temporal order are dependent on the interactive and integrative functioning of action-oriented inferences with syntactic, ILRRR, and semantic-logical inferences. In the many interesting results of the scrambled story test it turned out that Wernicke's aphasics and global aphasics had greater difficulties than the other groups in rearranging the A and B versions. In our view, this is due to the fact that Wernicke's aphasics and global aphasics have a loose organization of their semantic information in ILRRR. Consequently, the inferences which must rely heavily on the semantic information of ILRRR, viz. the ILRRR inferences, the semantic-logical inferences and the action-oriented inferences cannot become interactive in optimal ways. Nor can the information they should provide be fully integrated into the information data on the causal and temporal relations which should emerge from the action-oriented inferences. In summary, the greater difficulties of the Wernicke's aphasics and the global aphasics confirm that disturbances in the interactive and integrative functioning of inferences have repercussions on all higher levels of discourse processing. They also suggest that inferences are indispensable for syntactic processing, semantic processing, and action processing. 6. Concluding remarks

The foregoing remarks have clarified how verbally expressed information is organized in discourse processing and how inference-making processes determine the decompositional/compositional operations involved. More specifically, it has been shown that syntactic, ILRRR, semantic-logical and actionoriented inferences cooperate closely in decomposing the discourse-internal information data. It has also been demonstrated that these four kinds of inference serve to compose interpretative elaborations on the information of the discourse (fragments) under analysis. From the viewpoint of methodology, our considerations lead us to the conclusion that inferences deserve more attention than current theories of (discourse/text) grammar have paid to them thus far. In this connection, we have been able to point out that the conventionally distinguished 'natural inferences' and 'logical inferences' overlap/converge continuously in discourse processing. Discourse/text grammars should not disregard (the overlaps/ convergences of) these inferences.

Inferences as (De)compositional Principles

311

From the empirical evidence which has been adduced in this paper we may draw the consequence that our views on inference hold good for normal/ disordered high level language processing. They thus reflect realistic vistas for textlinguistic research.

References Agar, Michael and Hobbs, Jerry R. 1982 "Interpreting Discourse: Coherence and the Analysis of Ethnographic Interviews", Discourse Processes 5 , 1 — 32. Austin, John L. 1962 Hoi» to do Things with Words: The William James Lectures Delivered at Harvard University in 1955. Oxford: Oxford University Press. de Beaugrande, Robert-Alain and Dressier, Wolfgang Ulrich 1981 Introduction to Text Linguistics (= Longman Linguistic Library 26). London: Longman. Bellert, Irena 1972 On the Logico-Semantic Structure of Utterances (= Polska akademia nauk komitet j^zykoznawstwa 66). Wroclaw: Ossolinski. 1973 "On Various Solutions of the Problem of Presuppositions", in Petöfi, J. S. and Rieser, H. (eds.): Studies in Text Grammar (Foundations of Language, Supplementary Series 19). Dordrecht: Reidel, 79-95. 1980 "Sherlock Holmes' Interpretation of Metaphorical Texts: Creativity in Language Symbols", Poetics Today 2, 25-44. Blau, Ulrich 1978 Die dreiwertige Logik der Sprache: Ihre Syntax, Semantik und Anwendung in der Sprachanalyse (Grundlagen der Kommunikation). Berlin: de Gruyter. Brockway, J. P. 1981 Inferences as Human Memory (Manuscript). Charniak, Eugene and Wilks, Yorick (eds.) 1976 Computational Semantics. Amsterdam: North-Holland. Cicourel, Aaron V. 1980 "Three Models of Discourse Analysis: The Role of Social Structure"; Discourse Processes 3, 101-132. Clark, Herbert 1977 "Inferences in Comprehension", in Laberge, D. and Samuels, S. J. (eds.): Basic Processes in Reading: Perception and Comprehension. Hillsdale: Erlbaum, 243 — 263. Cole, Peter and Morgan, Jerry L. (eds.) 1975 Speech Acts (Syntax and Semantics 3). New York: Academic Press. Corsaro, William A. 1981 "Communicative Processes in Studies of Social Organization: Sociological Approaches to Discourse Analysis", Text 1, 5 — 63. Coulthard, Malcolm 1977 An Introduction to Discourse Analysis. London: Longman. Crothers, Edward J. 1978 "Inference and Coherence", Discourse Processes 1, 51—71. 1979 Paragraph Structure Inference. Norwood: Ablex. Cullingford, Richard E. 1979 "Pattern-Matching and Inference in Story Understanding", Discourse Processes 2, 319-334.

312

R. G. van de Velde

van Dijk, Teun A. 1977 "Semantic Macro-Structures and Knowledge Frames in Discourse Comprehension", in Just, M. A. and Carpenter, P. (eds.): Cognitive Processes in Comprehension. Hillsdale: Rrlbaum, 3 — 32. 1980 Macrostructures: An Interdisciplinary Study of Global Structures in Discourse, Interaction, and Cognition. Hillsdale: Erlbaum. 1982 "Opinions and Attitudes in Discourse Comprehension", in Le Ny, J.-F. and Kintsch, W. (eds.): Language and Comprehension (Advances in Psychology 9). Amsterdam: North-Holland, 35-51. Dinsmore, John D. 1981 The Inheritance of Presupposition ( = Pragmatics and Beyond II, 1.). Amsterdam: Benjamins. Dorfmüller-Karpusa, Käthi and Petöfi, Janos S. (eds.): 1981 Text, Kontext, Interpretation: Einige Aspekte der texttheoretischen Forschung (Papers in Textlinguistics 35). Hamburg: Buske. Franck, Dorothea 1980 Grammatik und Konversation (Monographien Linguistik und Kommunikationswissenschaft 46). Kronberg: Scriptor. Fritz, Gerd 1982 Kohären^: Grundlagen der linguistischen Kommunikationsanalyse (Tübinger Beiträge zur Linguistik 164). Tübingen: Narr. Gibbs, Raymond W. 1982 "A Critical Examination of the Contribution of Literal Meaning to Understanding Nonliteral Discourse", Text 2, 9-27. Goetz, Ernest T. 1979 "Infering from Text: Some Factors Influencing which Inferences Will be Made", Discourse Processes 2, 179-195. Gordon, David and Lakoff, George 1975 "Conversational Postulates", in Cole, P. and Morgan, J. L. (eds.), 83-106. Graesser, Arthur C. 1975 Prose Comprehension beyond the Word. New York: Springer. Grice, Herbert Paul 1975 "Logic and Conversation", in Cole, P. and Morgan, J. L. (eds.), 41—58. Gunderson, Keith (ed.): 1975 Language, Mind, and Knowledge (Minnesota Studies in the Philosophy of Science 7). Minneapolis: University of Minnesota Press. Hildyard, Angela and Olson, David R. 1978 "Memory and Inference in the Comprehension of Oral and Written Discourse", Discourse Processes 1, 91 — 117. Hildyard, Angela 1979 "Children's Production of Inferences from Oral Texts", Discourse Processes 2, 33 — 56. Huber, Walter and Gleber, Jochen 1982 "Linguistic and Nonlinguistic Processing of Narratives in Aphasia", Brain and Language 16, 1—18. Jarvella, Robert J. and Klein, Wolfgang (eds.) 1982 Speech, Place, and Action. Chichester: J. Wiley. Kintsch, Walter, and van Dijk, Teun A. 1978 "Toward a Model of Text Comprehension and Production", Psychological Review 85, 363-394. Labov, William and Fanshel, David 1977 Therapeutic Discourse: Psychotherapy as Conversation. New York: Academic Press. Lakoff, George 1972 "Linguistics and Natural Logic", in Davidson, D. and Harman, G. (eds.): Semantics of Natural Language. Dordrecht: Reidel, 545 — 665.

Inferences as (De)compositional Principles

313

Langleben, Maria 1981 "Latent Coherence, Contextual Meanings, and the Interpretation of Text", Text 1, 279-313. Lindsay, Peter H. and Norman, Donald A. 1977 Human Information Processing: An Introduction to Psychology. New York: Academic Press. Lytinen, Steve L. und Schank, Roger C. 1982 "Representation and Translation", Text 2, 83-111. Margalit, Avishai (ed.) 1979 Meaning and Use: Papers Presented at the Second Jerusalem Philosophical Encounter, April 1976 (Synthese Language Library 3). Dordrecht: Reidel; The Hebrew University Jerusalem: The Magnes Press. Minsky, Marvin L. 1975 "A Framework for Representing Knowledge", in Winston, P. H. (ed.): The Psychology of Computer Vision. New York: McGraw-Hill, 211-280. Noordman, Leonard G. M. 1979 Inferring from Language (Springer Series in Language and Communication 4). Berlin: Springer. Oh, Choon-Kyu and Dinneen, David A. 1979 Presupposition (Syntax and Semantics 11). New York: Academic Press. Olsen, Svend Erik 1982 "On the Information Processing Paradigm in the Study of Human Language", Journal of Pragmatics 6, 285 — 319. Omanson, Richard C.; Warren, William H., and Trabasso, Tom 1978 "Goals, Inferential Comprehension, and Recall", Discourse Processes 1, 337 — 354. Parker-Rhodes, Arthur F. P. 1978 Inferential Semantics (Harvester Studies in Cognitive Sciences). Hassocks, Sussex: The Harvester Press. Petöfi, Jänos S. 1973 "Towards an Empirically Motivated Grammatical Theory of Verbal Texts", in Petöfi, J. S. and Rieser, H. (eds.): Studies in Text Grammar (Foundations of Language, Supplementary Series 19). Dordrecht: Reidel, 205-275. 1981 a "Texttheoretische Forschung, Aspekte der Textkohärenz, Sprachunterricht", in Dorfmüller-Karpusa, K. and Petöfi, J. S. (eds.), 235-258. 1981b "Text und Bedeutung: Einige Aspekte der texttheoretischen Interpretation untersucht am Beispiel der Apg. 20, 17 — 38", in Dorfmüller-Karpusa, K. and Petöfi, J. S. (eds.), 154-215. 1983 "Explikation und Evaluation in der Textproduktion und Textinterpretation", in Petöfi, J. S. (ed.), 1983, 1-22. Petöfi, Jänos S. (ed.): 1979 Text vs. Sentence: Basic Questions of Text Linguistics (Papers in Textlinguistics 20, 1 & 20, 2). Hamburg: Buske. 1981 Text vs. Sentence Continued (Papers in Textlinguistics 29). Hamburg: Buske. 1983 Texte und S achverhalte: Aspekte der Textbedeutung (Papers in Textlinguistics 42). Hamburg: Buske. Petöfi, Jänos S. and Franck, Dorothea (eds.) 1973 Präsuppositionen in Philosophie und Linguistik (Linguistische Forschungen 7). Frankfurt/M.: Athenäum. Rieger, Charles J. 1975 "Conceptual Memory and Inference", in Schank, R. C. (ed.): Conceptual Information Processing. Amsterdam: North-Holland, 157 — 288.

314

R. G. van de Velde

Rumelhart, David 1980 "Schemata: the Building Blocks of Cognition", in Spiro, R. J., Bruce, B. C. and Brewer, W. F. (eds.): Theoretical Issues in Reading Comprehension: Perspectives from Cognitive Psycholog/, Linguistics, Artificial Intelligence, and Education. Hillsdale: Erlbaum. Schank, Roger C. and Abelson, Robert P. 1977 Scripts, Plans, Goals, and Understanding. Hillsdale: Erlbaum. Searle, John R. 1969 Speech Acts: An Essay in the Philosophy of Language. Cambridge: Cambridge University Press. 1975 "A Taxonomy of Illocutionary Acts", in Gunderson, K. (ed.), 344-369. 1983 Intentionality: An Essay in the Philosophy of Mind. Cambridge: Cambridge University Press. Seuren, Pieter A. M. 1979 The Logic of Presuppositional Semantics (First, provisional, experimental draft). Nijmegen. van de Velde, Roger G. 1978 "Einige mikrolinguistische Indizien der Textkonstitution", Linguistica Antverpiensia 12,261-281. 1979a "Probleme der linguistischen Theoriebildung einer empirischen Text-Wissenschaft", in Bergenholtz, H. and Schaeder, B. (eds.): Empirische Textwissenschaf t: Aufbau und Auswertung von Text-Corpora (Monographien Linguistik und Kommunikationswissenschaft 39) Königstein, Ts.: Scriptor, 10 — 27. 1979b "'Text' und 'Satz' in einer erweiterungsfähigen Linguistik", in Petöfi, J. S. (ed.), 1979, 572-584. 1980 "Understanding, Comprehension, and Coherence: A Plea for Interlinguistics", Linguistica Antverpiensia 14, 235 — 250. 1981 a Interpretation, Kohären^ und Inferen^. (Papers in Textlinguistics 33). Hamburg: Buske. 1981 b "Textuality and Human Reasoning", Text 1, 385-406. 1982a "Coherence Relations in Texts and Inferential Processing", in Hattori, S. and Inoue, K. (eds.): Proceedings of the Thirteenth International Congress of Linguists. Tokyo, August—September 1982, 1040-1044. 1982b "On Inferential Discourse Processing", Linguistica Antverpiensia 16, 23—45. 1983 "On Cohesion and Coherence", in M.-E. Conte (ed.): Kontinuität und Diskontinuität in Texten und Sachverhaltsfigurationen (Papers in Textlinguistics). Hamburg: Buske, to be published. 1984 Prolegomena to Inferential Discourse Processing (Pragmatics & Beyond V, 2). Amsterdam: Benjamins. 1985 "Inferences and Coherence in Text Interpretation", in E. Sözer (ed.): Text Connectivity, Text Coherence Aspects, Methods, Results (Papers in Textlinguistics 49). Hamburg: Buske, 261-298. 1986 On the Foundation of Interpretation. Leuven: Acco. 1987 Discourse Processing as Reasoning Reception. Fortcoming monograph. Warren, William H.; Nicholas, David W., and Trabasso, Tom 1979 "Event Chains and Inferences in Understanding Narratives", in Freedle, R. O. (ed.): New Directions in Discourse Processing. Norwood: Ablex. 2, 23 — 52.

Part III

THOMAS T BALLMER Context and Context Change /. Introduction For many years it was the endeavor of the author to provide for a unified treatment of the syntax, semantics and pragmatics of natural language. At the base of these investigations was the conviction that an important feature of natural language is the instrumental character of its expressions. Linguistic expressions are used by speakers and writers as instruments. Thus the problem emerges of explicating the notion of an instrument and related notions. Instruments are used primordially to aid humans (and eventually nonhumans) to reach their goals or in other words to fulfill their intentions. Instruments are thus an aid to performing actions. This being so, we are then confronted with the explication of the notion of action and its related notions. Underlying the notion of action is the notion of world change or more generally context change. If, in a typical case, an agent acts, he changes a world / (which would have come to be realized, if no agent had acted at the time in question and at any later time) to a world j (which comes to be realized, because of the agent's intervention at the time in question). Taking into account acts on a social, mental and possibly other level, an agent changes, if he acts, a context / (of a yet to be specified kind) to a context y (of a similar kind), because of his active intervention. According to this view the outset of a linguistic research program would then be to construe problems of language description, (formal) language reconstruction and other language-theoretical tasks in terms of context change, human (and non-human) action, and especially instrument-aided action. The philosophy of action is a traditional branch of philosophy, especially a branch of the philosophy of mind. As far as I can see, however, their research is centered around other problems than those which are relevant for our program. Emerging from the Wittgenstein school of linguistic philosophy and/or philosophy of language, v. Wright's (1963, 1966) and Austin's (1962) contributions are of immediate interest, v. Wright initiated the study of the logic of action, motivated mostly by his concern with deontic logic, Austin initiated the growing interest in the theory of speech acts, most efficiently carried further by Grice (1957, 1969), Strawson (1964), Searle (1965, 1969) and Schiffer (1972). Independently of these investigations, and, in fact much earlier, the German scholar Koschmieder (1929, 1935) developed the begin-

318

T. T Ballmer

nings of a theory of performatives and more generally of speech acts. An essential notion in his framework is the effective case (Effektiv-Fall), which corresponds to performatively used verbs. We do not make direct use here of his theory, however. The logic of action has been developed further by Aqvist (1971, 1975) and Brennenstuhl (1973, 1975). Both applied their results to the study of language. Nobody, however, proposed a unified theory of the syntax, semantics and pragmatics of language, founded on a theory of instrument-aided actions. Moreover, none of the logics and theories of actions proposed so far are based fundamentally on context changes. The central notions upon which logics and theories of actions have been based till now are intentions, states of affairs, counterfactuality, causation among others. These all play, of course, an important role. As we shall see, however, the structure of (instrument-aided) action is essentially determined by the notions closely related to context change. The structure of action is determined, among other things, by the situation in which the action takes place, which also determines the range of possibilities as to what actions can effect; the actual context which characterizes the passive evolution of a physical, social, mental and so forth universe, if no active interaction with the universe occurred anymore; the environment, and more specifically the time, which determines the details of the interaction of the agent with the universe; the context change function which characterizes the dynamical behavior of the contexts relative to the interactions of agents. The historical development of formal logic went through several important steps. At first propositional truth-value logic was proposed by Boole (1847). Then the development of predicate logic, functional logic and quantified logic was achieved by Russell/Whitehead (1910), v. Neumann (1928), Gödel (1930), Tarski (1935) and others. The next further step was indexical logic, in which context dependencies were incorporated explicitly (Kripke, 1963; Montague, 1970, among others), and finally intensional logic was created, including the direct denotation of intensions in the object language (Montague, 1969). It may well be, that the logic of context change, and especially procedural (cybernetic) logics prove to be another important step in the development of formal logics. Their expressive power is such that it is possible to reconstruct the calculation processes of a computer (and even interacting computer systems), the action patterns and especially the speech act patterns of humans on a par.* This paper has been written 1975/6, during the postdoctoral studies in linguistics and philosophy of language of the author at UC Berkeley. He was then engaged quite intensively in what today is called non-monotonic logics, and especially dynamic logics. Since then, however, the author has shifted his interests to less formal and abstract linguistic research. He has directed his attention, in close cooperation with Waltraud Brennenstuhl and during many years of support by the DFG (Deutsche Forschungsgemeinschaft der Bundesrepublik Deutschland) towards the investigation of natural-language predicates and predicate systems. The aim was to elaborate the logical system of the semantics of predicates of natural language, especially the verbs.

Context and Context Change

319

In this paper I am concerned with a general and by and large informal presentation of cases which demonstrate the value of the action-theoretic approach to the analysis of language. In fact many, if not all, major syntactic, semantic and pragmatic areas are influenced by the instrumental character of language. Cases which will be discussed here are: reference; fiction; opaque contexts; reference in opaque contexts; verbal speech acts, written speech acts, especially discourse, dialogues and texts; speech production, speech analysis, especially semantic and syntactic analysis; presuppositions and others.** Let us now begin with the presentation of some material which shows that context change is at the base of many phenomena of language. Language reflects so to speak linguistic communication structurally. 2. Problems of reference 2.1. Extensional reference

In the ideal case we may distinguish referent-introducing expressions and referent-using expressions. Names and pronouns provide standard examples. This turned out to be, essentially, a lexico-semantic task. It turned out, in fact, that the verbs, considered as a comprehensive, complete, and homogeneous body of natural language predicates deserve an intensive study of their own. A major interest to proceed in something which is a holistic lexico-semantic study of the Verb Thesaurus is that the verbs designate processes, in fact (virtually) all the processes relevant for human purposes. Soon it became clear that the instrumental character of natural language, as dealt with in this article, got strongly supported by these studies (cf. Ballmer/Brennenstuhl, 1981). Context change by means of the instrument 'natural language', is a processual phenomenon, after all. The late publication of this paper seems justified on the basis that only very recently the interest in dynamic logic and in the development of context change logics (cf. Kamp, 1981; Rieser/Eikmeyer, 1981; Heim, 1982) increased enormously. Having the feeling that much is said in this article which may support and supplement the issues discussed nowadays in the field, the author dares to take the risk of publishing these ideas that are part of my habilitation thesis (University of Bochum, 1977, in part published as Logical Grammar, Ballmer, 1978). The topic of speech act classification is touched upon in this paper quite naturally. This classification results directly from the investigation of context change mechanisms. The slightly modified and amended Speech Act Classification, as presented in Ballmer (1981), goes back to the one in this paper. I did not intend to alter this original one, because it seems more directly linked with the conceptions of context change. Speech act classifications have been dealt with considerably since. For a more comprehensive presentation of the topic the reader is deferred to Ballmer/Brennenstuhl (1981), to Petöfi/Kayser (1978), or arguing for an alternative point of view to Meibauer (1982). ** I would like to thank the many people who encouraged me to investigate this field, first of all R. Bartsch, M. Bierwisch, W. Brennenstuhl, J. Ihwe, W. Kummer, J. Rehbein, H. Schnelle, T. v. Dijk, and D. Wunderlich. Then I wish to thank H. Kamp who first appreciated my formal proposals; him, D. Lewis and R. Hilpinen who jointly discussed, but not unanimously agreed to a more developed stage of a logic of context change (not only for individual speech acts but for full speech act sequences); and H.-J. Eikmeyer, K. Hölker, W. Kindt, J. Petöfi, A. Oberschelp, H. Rieser, G. Todt, and many others with whom I had the opportunity to discuss issues on that topic.

320

(1)

T. T Ballmer

Charlotte skates. She hates skiing.

In addition to referring to Charlotte in the first sentence of (1), Charlotte introduces Charlotte for future reference by a feminine personal pronoun and other cases of anaphoric expressions, in other words, names change the context, which is in this case, the set of individuals immediately referable to by pronouns. Names add the individual named to that set. Other similar pairs are names and reflexives (2a), pronouns and reflexives (2b), indefinite and definite noun phrases (2c), verbs and pro-verbs (2d), sentences and prosentences (2e), time and place adverbials, and time and place proadverbials (2f, 2g), corelative clauses (2h). (2 a)

Regula watches herself in the mirror.

(2b)

He shaves himself.

(2c)

A motorcycle scorches down the street. The motorcycle is faster than the police car.

(2d)

Who rang the bell? - Thierry di'd.

(2e)

Martha got sick. It is terrible.

(2f)

Franklin lived in the 18th century. He liked to live then.

(2g)

Franklin lived in Philadelphia. He liked to be there.

(2h)

Whichever man you saw at the beach — that man is dangerous.

Quite generally we may say that any expression or form changes the context in that it introduces a corresponding individual or set of individuals, and that pro-expressions or preforms refer to the immediate context built-up by the earlier use of corresponding forms.1 Reference by pro-expressions is highly context dependent, even in fact for nouns. This is obviously true for deictic particles (like this, that, here, there, now, then, ...), for deictically used personal pronouns (I,you, he, she, ...). It is, of course, also true for intratextually used proforms, their referents depend on the (earlier or later) use of certain corresponding forms. But it is also true for these forms themselves. The picking out of a referent for names, indefinite noun phrases, dates and so forth is highly context dependent, if possibly to a lesser degree than for the corresponding proforms. Whether a form introduces an individual for future reference depends on the surrounding cotext (linguistic context): (3a) (3b) (3c) 1

Bill has a car. It is black. Bill is looking for a;car. it is black. Bill doesn't have a car. It is black.

Another somewhat refined terminology which has been used for pairs of expressions which stand in the relation of form and proform is: antecedent — anaphora, postcedent — cataphora. In this terminology the sequential order of form and proform is reflected explicitly.

Context and Context Change

321

jfff

(3d)

Bill wishes to have a carTTt is black.

(3e)

Bill regrets, having a car. It is black.

(3 f)

Bill wishes to have a car. It should be black.

These texts show that the introduction of an individual, like the introduction of a car by means of the indefinite description a car, depends on the predicate used (3b), on the negation used (3c), on the intensional construction used (3d, e, f). Moreover, the pointing to an individual is not controlled by the proform alone. This also depends on the surrounding cotext, as the following examples show. (4a)

The butterfly sat on the tree. It had green wings.

(4b)

The butterfly sat on the tree. It had green leaves.

Surrounding expressions like green wings and green leaves have a direct influence on the individual picked out. Reference may be controlled by intonational factors as well: (5 a) (5b)

Daniela slapped Marietta. Anna was angry and slapped her too. Daniela slapped Marietta. Anna was angry and slapped her too.

(6a) (6b)

John met the man in the red trousers. George missed him. Jojhn met the man in the red trousers. George missed him.

If a proform is used without a corresponding form having been used, it often, though not necessarily, gets a deictic interpretation. This forces the close relationship between deictically and intratextually used proforms. The kind and the structure of the context controls the interpretation. Often an interpretation which is in some obvious sense optimal is chosen. Thus (1) the most salient individual is picked out, (2) a previously relevant individual is picked out, (3) a commonly known individual is picked out, or (4) the nearest individual is picked out. 2.2. Fiction The use of the sentence (7 a)

Mr. Spock is a Vulcan.

uttered in normal conversation is not taken to be false, even by people (mutually) knowing that in the physical world there is no Mr. Spock, no Vulcan (the Vulcan hypothesis has been decided in the negative by physicists), no inhabitants of Vulcan, and hence no Mr. Spock who is a Vulcan. There is no explicit indication of a "fiction" operator either. Of course, in order to get your semantics straight you may postulate that names such as Mr. Spock and nouns such as Vulcan trigger a transformation (or a translation, or what

322

T. T Ballmer

not) which introduces a fiction operator in the logical analyses of the sentence in question. But you may equally well leave it to the semantical interpretation of the logical form to account for the job of assigning the referents to fictional and other peculiar names and to settle the truth status of the sentence in question. Whether you prefer the syntactical or the semantical approach you have to account for the apparent difficulty that the sentence is literally false for the world we inhabit and that it is true in some other, equally literal sense. One way to put the solution to this problem, and essentially the only solution, is to say that the use of the sentence changes the context from the actual world to a Mr. Spock- and Vulcan-relative world (or set of worlds), or, according to our common knowledge about these matters, to a Startrekrelative (set of) world(s). This is not to say that every sentence is trivially true. For instance (7b)

Mr. Spock is not a Vulcan.

is false. We, language users (and American television watchers), know enough Mr. Spock- and Vulcan-relative worlds to determine the truth-value of this sentence as false. For sentences referring to fictional (sets of) worlds, the truth-value may be undetermined. This is the case if for some of the fictional worlds in question the sentence is true and for some others it is false. This is probably the case for (7c)

Mr. Spock hates Reagan.

Because issues of that sort are not of importance in Startrek, fictional worlds of either kind — namely where (7c) is true and where (7c) is false — are compatible with Startrek-telauve worlds. The change of the context is triggered by lexical items and their meaning in the cases considered. We turn now to a more elaborate linguistic device to perform such changes of reference to other contexts. 2.3. Opaque contexts Very often the change of reference to other contexts than our actual one is made explicit by certain syntactical constructions, involving opaque context creating verbs (intensional verbs). (8a) (8b) (8c) (8d)

Bill Bill Bill Bill

believes that there is a devil. hopes that Nessie exists. fears that the president will often be shot. expects that Reagan will lose the election.

The sentences (8) refer to the actual context: the truth or falsehood of these sentences depends on what is true about the actual context. What is true in the actual context depends in turn on what is true in other contexts, namely whether the complement sentences

Context and Context Change

(9a) (9b) (9c) (9d)

(that) (that) (that) (that)

323

there is a devil. Nessie exists. the president will often be shot. Reagan will lose the election.

are true in all contexts which are compatible with what Bill believes, hopes, fears, expects, respectively. Thus the verbs believe, hope, etc. (together with their subject Bill) trigger a change of reference to a set of other, corresponding contexts. This change of reference explains easily why certain rules of inferences valid in the normal, extensional cases are blocked. Neither existential generalization nor substitution of equals is valid anymore, without further assumptions. The underlying reason for this being so is the following: If something is true in certain contexts, say the believe contexts, the inference of this being true in other contexts, say the actual context, is not licensed without severe restrictions on the contexts in question. We shall not elaborate this type of context change, which is more adequately called world change, much further. Since Carnap (1947) and Kripke (1963, 1965), a lot of work has been done to elaborate this topic, and to apply it to many linguistic and philosophical domains. These investigations fit in rather nicely with our program of context change. A topic intimately connected with world change has been neglected somewhat, however. This may be due to its complexity. It is the problem of changing the reference to individuals by the use of opaque constructions such as sentences containing intensional verbs. We shall now turn our attention to this problem. 2.4. Reference in opaque contexts Intensional verbs not only introduce new worlds, but also new sets of individuals to which names and pronouns may refer. In certain cases not even the occurrence of intensional verbs is necessary to accomplish this task, as has been shown by (7) above. There are cases, however, where the mere occurrence of a fictional name will not do the job of changing the context: (10a)

Mr. Spock exists.

or even more specifically: (lOb)

Mr. Spock exists in our world.

The latter sentence is indisputably false, the former very probably so. The following sentences, however, may easily be true: (11 a) (lib)

Daniel believes that Mr. Spock exists. Daniel believes that Mr. Spock exists in our world.

324

T. T Ballmer

Texts as the following (12)

Mr. Spock does not exist. Daniel nevertheless believes that he exists.

tell us that intensional verbs such as believe trigger a change to fictional entities. The pronoun he refers, but certainly not to an entity existing in our world. Let us introduce Ig, //, 1 2, ... as sets of individuals existing in our world, in a first order remote world, in a second order remote world, and so forth, respectively. Let us use as a first approximation the standard notations to display the semantics of sentences such as (8 a) Bill believes that there is a devil. namely (13)

Aj(iR lU -> VxDevil (x))

where Rj is a relation between the (actual) world / and the worlds j which are compatible with what Bill hopes in world /. The pertinent question is how are the individual sets /o, I\, ... and the worlds i,j connected. Is there, for instance, a dependency of the individuals of /i on the remote worlds j or is there no such dependency? To answer this question we should take into consideration our understanding of certain texts of natural language: (14a) Bill (14b) Bill (14c) Bill (14d) Bill

seeks seeks seeks seeks

his wife. He married yesterday. a secretary. He didn't find one so far. Nessie. But Nessie does not exist. a goblin.

In (14a), the individual which is the object of Bill's search exists also in the actual world. Moreover it is the same object of search in all worlds compatible with what Bill seeks. In (14b), the situation is the same in that the object of Bill's search does exist in the actual world (the secretary he seeks is already born, in the standard case). There is a difference with respect to (14a), however, in that the object of search may vary from seek- world to seekworld. Though Bill seeks one and only one secretary, there are different possible candidates. From this point of view there is no unique secretary whom Bill seeks. (14c) and (14d) are parallel to (14a) and (14b) with respect to the determinacy of the object searched. In (14c) it is the same in all seek- worlds and in (14d) it may be a different one in different seek-worlds. A formalization making use of the notation introduced so far would display the different cases as follows: (15 a) (15b) (l 5 c) (15d)

νχνχ(χΐ = χ Λ χ ε Ι 0 Λ Λ j(iR2j->(Has(Bill, χ) Λ Wife(Bill, χ)))) VxAj(iR 2 j->Vx(xj = xA X6 IoAHas(Bill, χ) Λ Secretary(x))) ΥχΥχΐχεΙ, Λ Λ j(iR 2 j ->(xj = χ Has(Bffl,x) Λ χ = Nessie))) VxAj(iR2j->Vx(xj =χ Λ χεί! Λ Has(Bill, χ) Λ Goblinfr)))

Context and Context Change

325

r

x~> is a variable denoting individual concepts, Γχ~" is a variable denoting incarnations of individuals in specific worlds, R2 is the relation which holds between the (actual world i and the worldsy which are compatible with what Bill seeks.2 3. Some specific problems of context change 3.1. Context change versus passive evolution of the context Matters can change in different ways. Some but not all of these changes we should call changes of the context. First, there are changes without the direct interference of an agent. We call this type of changes passive evolution of the context. In order to keep track of what is going on, we say that in such a case of passive evolution, the context remains constant. What changes is the environment. The context is an invariant as long as no agent interferes, it characterizes the passive evolution of the universe. Changes of the environment account then for the fact that in a discourse (16)

This is the theater. This is the town hall. This is the market place ...

on a sightseeing tour, make sense, because the objects in the focus of pointing are appropriately changing. The discourse itself does not provide for direct clues as to how the environment should change. The speaker does not perform the changes of the environment which are necessary to make the discourse true. Second, there are changes which are based exclusively on the actions of an agent. These are standard cases of context change. In these cases there is a more or less direct active interference of the agent with the universe. There are plain actions such as hitting, killing, kissing, which change the actual world, and there are language-aided actions, which change primordially the state of beliefs (and state of obligations) of the individuals involved. Some of these linguistic actions are socially relevant, others are subsidiary to the performance of those socially relevant linguistic actions. 3.2. Socially relevant linguistic actions versus subsidiary linguistic actions There is, from a social point of view, an important distinction among actions, namely those actions which are performed to have lasting results and those 2

A somewhat simpler version of the semantics can be provided omitting the quantification over individual concepts. The transworld coherence of individuals is no longer displayed, however, once this is done: (15'a) νχ(χεΙ0 Λ Λ j(iR2j^(Has(Bill,X) Λ Wife(Bill,x)))) (15'b) Λ j(iR 2 j -+νχ(χεΙ0 Λ Has(Bill.x) Λ Secretary(x))) (15'c) νχ(χεΙ,Λ Λ j(iR 2 j-»Has(Bill >X )AX = Nessie)) (15'd) Λ j(iR 2 j -*νχ(χεΙ, Λ Has(Bill,x) Λ Goblinft))) The structure of the logical form becomes somewhat more perspicuous, in this formalization.

326

Τ. Τ Ballmer

whose results are temporary and whose purpose is merely to provide the preconditions for a smooth performance of the former type of actions. They are like Wittgenstein's ladder which can be thrown away once it has served its purpose. Getting somebody aware of oneself, in order to speak with him (17a), drawing his attention to certain objects and worlds (reference) (17b, c, d), lifting these objects in a status of prominence for future pronominal reference (17e) etc. are communicatively important subsidiary actions. These are temporary. They do not, however, have a social relevance per se, as the following discourse shows: (17 a) (17b) (17c) (17d)

(17e)

- hey. Do you know Mirabelle? Do you know Leonidas? You know the woman in the bathing suit and the man in the tuxedo yesterday evening? Remember them distinctly?

The discourse gets its meaning by adding, say an assertion (17f), a question (17f) or a command (17P') which include the socially relevant predication: (17f) She is his wife. (17f) Did she leave before him? (17P) Find out what their names are. Now a lasting result has been achieved. The result of an assertion is to have established a reliable piece of information. The result of a question is to have brought the addressee under the obligation to provide a reliable piece of information (of a specific sort), if he is able to do so. The result of a command is to have brought the addressee under the obligation to do something. Speech acts can be classified according to the socially relevant lasting results they bring about. Linguistic actions which do not have such lasting results are, from the point of view of social context changes of secondary import. They are not the socially critical actions. They are not criticized, blamed, accused, withdrawn. Their functioning for the socially relevant communication processes is essential however, and for many cases unquestionably working, i.e. presupposed. Reference, establishing common grounds for socially relevant communication is hence often presupposed, and even tacit and indirect. Another typical example which demonstrates the mechanism of subsidiary reference actions by the expressions used in a discourse is the following: (18 a) Fred is waiting. He has a beard. (18b) Fred is waiting. David is reading. He has a beard. (18c) Fred is waiting. David is reading. Bill cannot sleep. He has a beard. The use of more and more expressions, before the sentence He has a beard. changes the reference of the pronoun he systematically. Individuals newly

Context and Context Change

327

introduced by the use of their name displace the aforementioned individuals in the background. They do not disappear totally from the scene, but it takes some effort to establish a reference, once they are removed from the foreground:3 (19)

Fred is waiting. David is reading. The one who is waiting has a beard.

Besides such linguistic actions establishing reference, there is a further important class of subsidiary actions mentioned briefly above. They also precede the socially relevant actions, because their purpose is to enable the smooth performance of the latter. The actions I have in mind are a mixed bag of devices to negotiate and establish common grounds, i.e. check the other's presumptions and try to reduce disagreements to enlarge mutually acceptable positions. 3.3. Verbal speech acts and written speech acts

Examples which demonstrate the mechanism of socially relevant linguistic actions are ample. We enter here the field of speech acts. Speech acts, quite generally, change the states of belief of the individuals involved. We should distinguish two main types of speech acts, the (real time) verbal speech acts and the (time-out) written speech acts. They have somewhat different properties, and merit, because of their importance, a different treatment. One thing at the base of the difference mentioned here is the distinction between action and result. Verbal speech acts (and writing acts) are actions, (discourse and) written speech acts [texts] are the result of these actions. Another thing at the base of the discussed difference is that for verbal speech acts the result, i.e. the discourse, is immediately processed by the participants, whereas for writing acts, in the standard case, the interpretation of the result of the action, i.e. the text, takes place at another time. Moreover written speech acts are directed, in general, to a larger and less restricted circle of addressees. A type of acts closely related to speech acts though utterly different and certainly much more fundamental are thought acts. We will occasionally touch this topic, as it proves useful for a clarification of speech acts.

4. Speech acts and context change 4.1. Primary considerations

Speech acts are intrinsically context changing. An appropriate semantics (and pragmatics) of linguistic expressions, used to perform speech acts, has to account for this fact, or rather, has to be based on that fact. 3

The stage and scene metaphor used here is elaborated later.

328

Τ. Τ Ballmer

In this section we are concerned predominantly with verbal speech acts performed in certain contexts (and environments). We will try to elaborate some of the facets of context changes which are due to speech acts. Because of the complexity of the domain of speech acts, we will propose different typologies and analyses in components. It is the opinion of the author that for different types of questions different such typologies and analyses prove useful. There is neither a "best" typology, nor a "best" analysis. The appropriateness may well depend on the purposes of the investigation. To begin with, let us consider a particular speech act, say an assertion, or, more specifically, the act a that S asserts to H that p, in context c (and environment e). A question of general interest is now what is it that is changed by the performance of the act a. 4.2. Various speech act analyses 4.2.1. Three level analysis It proves useful, at this stage of the discussion, to introduce the following analysis of a speech act in three levels: (20) 1. physical level (including neuro-physiological, articulatory, acoustic, auditive phenomena) 2. morpho-syntactic level 3. socio-mental level (including beliefs, states of obligation, expectations). Making use of these levels, we can say that by performing act Λ, a basic change occurred on the physical level. An electrical pattern, representing a thought of S, is transformed stepwise into an articulatory pattern, an acoustic pattern, an auditive pattern of H (and S) and finally into an electrical pattern again. The world has physically changed, a physical event originating from S has induced certain lasting results in the brain of H (and the brain of S}. Simultaneously a morpho-syntactical form is realized. Whether this is taken to be a consequence of the physical event, an act of itself, or a theoretical construct independent of the act is a problematical question. It depends on the theoretical and ontological assumption one is willing to make. All three views are possible, and defended by different philosophers and linguists. In any case, the change occurring on the morphosyntactical level is the addition of another realization of a morphosyntactical form to the already existing such realizations. What happens on the socio-mental level depends, again, very much on what the semantic and pragmatic objects are taken to be theoretically and ontologically. According to the view I would like to advocate here, these objects are essentially beliefs, including beliefs of what is to be attempted, i.e. states of obligations, beliefs about what will happen, i.e. expectations,

Context and Context Change

329

and so forth. Beliefs are mental objects. Because beliefs are about reality (and other beliefs), it seems promising to try a (possibly approximate) reduction to real objects, as much as seems useful. Therefore those objects are reconstructed relying on objects such as truth-values, individuals, times and (eventually higher order) sets and functions among such objects. The primordially independent entities worlds, propositions, beliefs, obligations, expectations, contexts, etc. are reconstructed set- and function-theoretically on top of these partly theoretical (truth-values, times4) and partly ontological entities (individuals). The changes occurring on the socio-mental level are changes of beliefs and more generally contexts. As contexts may include worlds, beliefs, states of obligations and so forth, the changes are changes of these entities independently of any set-theoretical reconstruction. Once one accepts the program of modeling these entities set-theoretically, the changes are to be modeled in the same way. Thus changes, though perhaps autochtonous and independent entities in their own right, are reconstructed here as certain (settheoretical) functions. It seems to be clear that the changes occurring on the socio-mental level of speech acts, i.e. on the objectual level in a mental sense (as excluding phonetic, graphic objects in a first approximation), are caused by the changes on the physical level (as including phonetic, graphic entities). Thus speech acts can be viewed as causally linking changes on the physical level with changes on the socio-mental level. The analyses (20) of a speech act into three levels mediates between a wellknown analysis of linguistic entities in three levels (21 a) and an analysis of actions in three levels (21b): (21 a)

1. 2. 3. (21 b) 1. 2. 3.

phonetic-graphic (phonemic graphemic) morpho-syntactic semantic-pragmatic subsidiary actions (basic actions) action type (action form) context change (physical and mental, i.e. real and socio-mental change)

The important facts which help to clarify why speech acts have an intermediary position between language (20) and action (21) are: (a) a speech act has both a linguistic (i.e. structuralistic) and an action aspect; a speech act is especially (β) a complex action, using (i) physical events, like sound production, as subsidiary actions, which are (ii) of a special linguistic form, to perform (attempts of) context changes on the socio-mental level in accordance with semantic and pragmatic rules of language. As becomes clear by this analysis, speakers performing speech acts use sounds and letters as instruments for attempts to change the context. 4

Instead of times we could take also environments. Environments are numerous enough to index the finest differences one is trying to model.

330

Τ. Τ Ballmer

4.2.2. A seven level analysis

As this analysis stands, goals, plans, intentions do not seem to play a role for speech acts. Speech acts seem to start with the production of sounds. It is clear that such a picture is inadequate. A full speech act comprises the formation of sounds out of the beliefs of the speaker as well. Thus a more complete analysis of a speech act comprises the following seven levels: (22)

a. b. c. d.

socio-mental level morpho-syntactic level physical level physical level

e. physical level f. morpho-syntactic level g. socio-mental level

of the speaker of the channel between speaker and hearer of the hearer

This analysis makes it clear that beside the causation of a change of the hearer's mind by the creation of a physical sound event (e, f, g), there is another important process going on: the causation of the creation of a physical sound event by a change of the state of mind of the speaker (a, b, c). The process (c, d, e) of information transmission is a purely physical process without an intervention of an agent, normally. Corresponding to this analysis of a speech act into different levels, some more changes of the context become interesting: the mental changes of the speaker and the process of expressing a thought in words, i.e. externalizing the mental change by attempting an appropriate physical change at the level of sounds. A most important consequence of this way of viewing the structure of speech acts is the natural analysis that speech-act sequences in dialogues and discourse get:5 For some internal or external reason a change in the beliefs of the speaker occurs which motivates him to produce an utterance, i.e. a sound string of a specific form in the context in question. For the hearer this utterance may function as an external reason for a change of his beliefs which motivates him in turn to produce an utterance. This process goes on until the states of mind of speaker and hearer are reached where no one feels motivated to produce an additional utterance. This state of affairs characterizes a natural end of the dialogue between the (two) participants. Speaker and hearer may be the same person. The sequence of speech acts is then a monologue or a (special form of a) discourse. Discourses may be written down, in the cases where a text is produced.

5

Cf. Berliner Gruppe (1975), where a typology of speech-act sequences led to a well structured classification of speech acts.

Context and Context Change

331

4.2.3. A three phase analysis: context change ontology

A quite different analysis of the structure of speech acts is possible. It is the analysis which concentrates on the socio-psychical level and examines the preconditions, the transition and the results which are caused by the performance of a speech act. The preconditions P of Ά standard case of a performance of the speech act a that S asserts to H that p, in the context c (and environment e) are that S knows that p and that H does not know that p. In other words the context c belongs to that set of contexts which satisfy the preconditions P, in short form: P(c). The result R of the performance of the speech act a of S to H in the context c is a context c' in which S (still) knows that p and in which, moreover, H knows that p. In other words, the context c' belongs to that set of contexts which satisfy the result R, in short form: R(c'). On a more elaborated level of analysis the change which occurs by the performance of a is as follows: A context c which fulfills the preconditions P is such that S knows that p, H does not know that p, S knows that H does not know that p. A context c' which fulfills the result conditions is such that S (still) knows that p, H (now) knows that p, H knows that S knows that p, S knows that H knows that p, H knows that S knows that H knows that/), ... An even more elaborated analysis would include perhaps that H wants that H knows that p and that S knows that H wants that H knows that p, or that H wants that S makes it the case that H knows that p and S knows this. The preconditions and the result conditions for contexts may be worked out in many ways. It depends, again, on the purposes which are connected with the analysis how far one should push this issue. In any case, the performance of the speech act a of S toward H by means of a linguistic expression ΓαΊ carries, in the standard case, a context c, fulfilling the preconditions Pra^, to a context c', fulfilling the result-conditions Rra-,. We may represent the situation that the use of an expression Γα~ι transforms the context c into the context c' as follows:

(23)

V

The systematic force of the transformation which is induced by the use of a linguistic expression ""a"1 in a context c is characterized by a function ξ, the context change function, ξ may be partially determined by linguistic conventions, partially by non-linguistic conventions, partially by the situation in which Γα~" is used. For many contexts, ξ may be undetermined, especially for those

332

T. T Ballmer

contexts which do not fulfill the preconditions.6 ξ characterizes an important aspect of a speech act, it characterizes its force to change the context: (24)

ξ(ΓαΛ c) = c'.

Including the conditions imposed on the contexts, ξ must fulfill the following requirement: (25a)

Pra, (c) -> (ξ(Γ(Λ c) = c' Λ Rra, (c)).

If Pra-, (c) is false, i.e. if the use of Γα~ι is unhappy, the value ξ(Γα"", c] may be undefined, which is in a sense unnatural, because, strictly speaking, some context is always produced, if also a very odd one. Another possibility would be that in case Pffr (c) is false ΑΓαι (c') becomes automatically false, too. That is, unhappiness of the use of an expression ""a"1 in context c would imply failure automatically: (25b) Pv (c) ->

fo\ c) = c' Λ Rra, (c')).

This would exclude the possibility that a context matching the result conditions could come about by luck in a situation in which Γα"" is unhappily uttered. This seems to be somewhat artificial and rigid. First, it should well be possible for an inappropriate use of an expression to lead to success: children, foreigners, drunken people etc. often do not master every rule of use and application of linguistic expressions and nevertheless they may well be able to accomplish their aims in spite of the former defect. Second, on the assumption that ξ(Γα"1, c) = c' is true, Pra-, (c) and Rra-, (c') would be equivalent. This is to say that the happiness and success of a linguistic action are equivalent, provided the context is changed according to the force of the expression: whenever an expression is happily used, it is successful, and whenever it is successfully used, it is (was) happy. This means that the appropriateness of the use of an expression equals success. Hence the success is independent of the hearer's reaction and especially his spontaneous skills to interpret even inappropriate speech acts. Whether an assertion, a question or a command is successfully performed, depends thereby solely on the preconditions and not on other factors. Of course a decision on this question depends on one's theoretical and metaphysical assumptions. A determinist may want to postulate that every result of an action is programmed in the preconditions. But for nondeterminists also this view may be persuasive. They may say that the result of a speech act is simply the understanding of the meaning of the expression used, and they may argue that this result comes about necessarily, if the happiness

6

According to this manner of conceiving speech acts, Pra-, are the happiness conditions, Rra-, are the success conditions, and ξ(ΓαΊ, c) = c' says that Γαπ counts as c' in context c.

Context and Context Change

333

conditions are fulfilled. They may concede that there are results of a speech act which depend on the hearer's (and even the speaker's) active reactions, but they point out that these are perlocutionary results and not illocutionary results. These results are not brought about solely by the meaning of the expression used in the speech act. Thus "results" for them are illocutionary results and brought about exclusively by virtue of the meaning of the expression used. I think that this is a viable approximation to the truth and valid for many interesting cases of application of the theory of speech acts. But we should keep several things in mind. First, it is perfectly legitimate to include more than merely the so-called illocutionary effects among the intended results of a speech act: If I ask somebody what time it is, I intend more than changing his beliefs about what he should do in the near future. I do not merely intend to put him under another obligation. I do not intend merely to impose an additional stress factor on his maybe already screwed up life. I intend him to obey that order. It is even (linguistic) common knowledge among us that this is so, when I have uttered rWhat time is it?"1 in a context in question. Thus it belongs to the meaning of the expression used that I want him to obey, and not only to understand that I want him to obey. The utterance event counts, at least for me, and I think also for many others, as a communicative failure, if the answer of my addressee is the utterance of T understood what you said. You wanted to put me under the obligation to tell you the time. I accept that you want to do this. As I take it, your motivation for this is that you don't know the time and that you think that I know it. In fact you are right, I know the time. Good bye."1 This case shows, I hope convincingly, that a mere understanding by the addressee of the expressions used in speech acts will not do in general enough cases. A lack of cooperation from the side of the addressee produces the cases which show its partial inadequacy. As long as cooperation pertains, the illocutionary analysis is probably correct, because excusable omissions to realize, say, obligations do not count as failure for the speaker. Second, it is perfectly legitimate to include less than the so-called illocutionary effects among the intended results of a speech act. Often communication is intentionally based on the lack of full understanding of the expression used. Typical cases are certain types of literature including modern poems, modern prose, but also standard science fiction. Technical terms, obsolete terms, loan-words, artificial words are used to increase the awe and other emotions of the addressee. Similar techniques are used in commercials, advertisements, newspapers and elsewhere. Third, it should be recognized that even requirement (25a) is not totally sound. As it stands, it is true for all contexts c, c'. A counterexample in case would be somebody's using expressions fully appropriately but not reaching the intended result, be it the illocutionary result or some other. Such cases are investigated in a more general setting in the analyses of counterfactuals

334

Τ. Τ Ballmer

and causation.7 One essential issue is here that all sorts of eventualities could arise which are not spelled out in detail in the antecedent of a restriction such as (25a). Thus the quantification about contexts should be restricted appropriately to cases which are not too extraneous. An element of idealization has to be taken into account. To conclude the discussion of the condition (25a) and (25b), we should draw the attention to two further facts. First, ξ(Γα~ν,) = c' is presupposed, if the truth of the conditions is assumed: both Pr^(c) and -tPrai(cj imply ξ(Γα"ν,) = c'. Under the illocutionary analysis this makes sense. The illocutionary force of an expression ΓαΊ is presupposed in the use of ΓαΊ. This brings us to the second fact. The context c includes, normally, the speaker's intention to perform the speech act. Under such an analysis P(rcC, c} is only true, if the speaker intends to perform the speech act of uttering meaningfully the expression ""a"1 (or what he takes to be ""a"1). Thus the speech act is precluded to be performed without the active intervention of the speaker. (25a) allows in principle also the performance of speech act without specific intention, i.e. without the speaker's active intervention. It may remain here an open question, whether such cases are of any theoretical interest. An indication that this may be so is the context changing mechanism of signs (traffic signs, posters, etc.) and texts. These have an immense effect on large numbers of addressees but without the backing by a specific speaker intention in every case. There is, however, the overall speaker intention behind the production of these types of linguistic utterances.

4,3. Interaction graphs (and their for motivation) 4.3.1. Indroduction Instead of (23), we may use a slightly different way of representing the situation that the use of an expression Γαπ transforms the context c into the context f ' , by taking into account a time axis (or more generally an environment axis) to display the temporal (or environmental) development of the speech act process.

(26)

time, environment

Goodman (1955), Aqvist (1971), Lewis (1973).

Context and Context Change

335

This representation allows us to keep track of what goes into the linguistic interaction process (""a"1, c) and what comes out of it (the transformed context c'). The circle represents the interaction zone, the function ξ determines the interaction by relating input and output systematically. The expression Γα~" is not destroyed during the speech act, i.e. is also available after the linguistic interaction in question. Therefore it is more systematic to draw the representation (26) as follows:

(27)

The time (environment) axis is drawn only occasionally. It is normally taken to point to the right. This mode of representation of speech acts with force ξ allows the bookkeeping of many important traits of linguistic and even non-linguistic processes. We may for instance introduce speakers and hearers as well, and we may connect elementary interactions to more complex ones. Speakers, hearers, time etc. are seen, as usual, as specifications of the context c or c'. The expressions ""a"1 may be omitted occasionally. In order to demonstrate the import of such representations, let us consider some of those for the classification of some non-linguistic and linguistic interactions. (28 a)

S.

.S

Examples: perform do influence operate change

(28 b) S^

c

^S'

c

experience perceive see hear recognise notice

S stands for speaker and the dash marks that he has changed during the interaction. Similarly for c. Thus (28a) underlies processes where S affects the context, and (28b) underlies processes where the context affects the speaker. The following graphs underlie communicative interactions between speaker and hearer without and with linguistic expressions:

336

Τ. Τ Ballmer

(29a)

Η

(29 b)

Examples: tell say inform ask command request

H'

S, r

c

r

cf

A dialogue is displayed as follows:

Examples: chatter argue with each other discuss debate dispute and a report is displayed as follows (29 d)

S

H

Examples: report depict describe portray convey facts

Of course these graphs only reflect very superficially what is going on in linguistic and non-linguistic interactions. They must be refined for many further purposes. On the other hand they provide, in the form they are now

Context and Context Change

337

in, a typological grid on which to project different kinds of interactions.8 A most general linguistic interaction graph (omitting the actants speaker and hearer for convenience) is the following: (30)

'd

time, environment

Let us try to interpret it. An expression rat together with a context c, the context in which ""a"1 is to be uttered , enter the zone of interaction ξ. As a result they leave, transformed as Γβ"" and d, respectively, the interaction region ξ. To every such graph belongs an equation. Generalizing the equation (24), which corresponds to (23) and (27), appropriately, we get: (31)

ξ(·αν) = (T,d)

Now, we should try to make sense out of the fact that the expression ""a"1 is transformed into another expression, Γβλ Above we said that expressions do not change during a linguistic interaction. The argument has been that the expression is not used up by such an interaction, and belongs before and after, in the same unaltered way, to the language. I do not think that this truth can be shaken, essentially. However, instead of expression types we may consider ""a"1 and Γβ~" to be expression tokens, which would in any case be more natural considering interaction processes. To stress that point, we could make it overt notationally, by representing tokens by pairs: ""a"1 = CV,v), r ~" = (""δ^,μ). But avoiding notational clumsiness, we

Cf. the fact that very simple kinds of modes of being and becoming are displayed quite naturally, too:

exist

live

be think, reflect, meditate happen, become, grow

338

T. T Ballmer

just remember that we often denote expression tokens in the linguistic actions discussed. The indices specifying the tokens are taken to be part of the environment. Once this question is settled, it is not especially difficult to interpret graph (30) and equation (31). Tokens may be transformed, for instance in the process of expression production in a speaker (cf. (22a, b, c.)), they may be transformed on the physical level (cf. (22c, d, e.)), and they may be transformed during the understanding and interpretation process of the hearer (cf. 22e, f, g.)). Let us concentrate on the last of these processes. If, say, an imperative sentence token rcf is uttered in a context c by a speaker it is quite natural for the hearer to process this speech act by transforming the imperative sentence token to another token Γβπ, characterizing the fact that the speaker has commanded the hearer something. That is, the hearer transforms the force bearing expression into a description. This transformation, of course, must be kept track of in the context, in order to remember the intent of the original speech act. Thus a corresponding context change takes place to account, among other things, for this. 4.3.2. Discussion of examples Let us discuss a specific example. The case we consider is the utterance of the imperative sentence Give Anna the ball! in a context c where the hearer does not stand under the obligation to make it the case that Anna has the ball. For simplicity we assume even that the hearer does not have any obligations whatsoever at the time in question. The relevant aspects of the context for our example are the hearer and the state of obligation CH of the hearer. All other aspects of the context are ignored for the moment.8* The beginning of the interaction is as follows: (32a)

r

Give Anna the ball!\

/QxGive (χ,Α,Β)"1

HCA

The first step displayed here is the (reconstructed) logical analysis. The surface Give Anna the ballT is assigned the logical form r © Give (H, Anna, Ballf, where the imperative operator constant r © n , and some obvious predicate and individual names are used. The imperative operator is taken here as a variable binding operator. The variable denoting the contextually determined r

8a

Because we reconstruct the processes in question formally, we lift the used expressions on a theoretical level by applying Quine's corners. Cf. chapter C. (1) or Ballmer (1975).

Context and Context Change

339

individual who is commanded to do something is bound by the operator. In the next step, the effect of the logical form on the context is exhibited: (32 b)

r

(DxGive (x,A,B)n

r

©xGive (χ,Α,Β)"1

(r©xGive (x,A,Br,H,CH) The third component l>)(rQxGive(x,A>B)'l,H>Cfi) of the outcome of the linguistic interaction process equals Has(A,B). This signifles that Η is obliged to make it the case that A has Β (i.e. that Anna has the Ball), because the proposition Has(A,B) is part of the obligations which Η has. The whole force of the command is transferred to the context, more specifically to that aspect of the context which tells what Η is obliged to realize. The logical form Γφχ&νβ(χ,Α,Β)~ί has been transformed to a corresponding logical form which tells that the incoming logical form has been worked off and is settled now. r®xGive(x,A,B)^ is read 'it has been ordered (to me, H) that by me it is given A to B'. From r(g)xGive(x,A,B)~( in the context in question it may follow, among other things, that H has understood the command, that H knows that S wants him to do that, and perhaps even that H accepted to do it and that H will do it. The latter inferences can be drawn safely at least in certain situations where the authority relations and the difficulty of the task and so forth are suitable. A somewhat simple recursive definition of the function ξ3 determining the change of the states of obligation of H would be as follows: (33)

1. ξ^,Η,ΟΟ = CH 2. ξ3(ΓΠ(γι,...,γηΓ,Η,θΗ) = CH 3. ξ3(Γα A ^,H,CH) = ξ3(Τ,Η,ξ3(ΓB).) The process (32) could be studied further, including now the world / along which the course of events evolve. Then the execution of the command could be displayed. Denoting the new state of obligation ^(r(])xGive(x,A,B)~[,H,CH) by C'H (it equals Has(A,B}}> the following graph demonstrates ffs compliance:

(32c)

... V/" C'H (ω) ωι (C'H,o)2 (C'H,i)) [«CH] ω2 (C'H,J)

The function ω, characterizing the compliance of //, has two components ωι and a>2. o>2 assigns to each state of obligation C'H and to each world /' the world /' fulfilling the obligations C' H· In our case /'ε Has(A,B). Moreover, the state of obligation is changed back, because now Η has fulfilled what he was supposed to do. This context change is accomplished by the first component ωι. ωι takes as arguments the old state of obligations C'H and the new world brought about by ω2 and readjusts the state of obligation. In our case it will be (near to) the old one, namely the empty state of affairs CH. With the techniques presented so far we are able, in principle, to reconstruct many aspects of context changes which occur induced by the use of linguistic expressions in speech acts. But sometimes we are not so much interested in the detailed effects which are performed on the context but simply whether or not something has been affected, or also whether or not something is true. In other words sometimes we want simply to evaluate an expression relative to a context and know whether it is successfully used in that context, whether it is true, and — maybe — whether it is happily used, whether it is morally good, whether it is beautiful and so forth: in short, we may be interested in judging certain expressions relative to a context.

Context and Context Change

341

4.4. Three types of context forcing A further question which may interest us sometimes is how a context forces the utterance of an expression. This is the converse question to how an utterance affects a context. The three questions: (1) how do linguistic expressions force certain contexts, (2) how are linguistic expressions judged in contexts, and (3) how do contexts force (the occurrence of) certain linguistic expressions, can be nicely displayed by our interaction graphs. The only way to do this in a reasoned manner is to recognize values as independent full-fledged entities and to rephrase (1), (2) and (3) somewhat. There are three basic kinds of speech acts: (1) (Operatives) The speaker can force a certain context to come about by attempting to make it/// to a certain sentence ( p r o p o s i t i o n ) . (2) (Judgments) The speaker can force a certain value to come about by attempting to match a sentence ( p r o p o s i t i o n ) and a context. (3) (Expressings) The speaker can be forced to utter a certain sentence (from a certain concept, p r o p o s i t i o n ) by a certain context being the case. The three basic types of speech acts are called here operatives, judgments, and expressings. In a more politically oriented mode one could call them also executives, judiciaries, and legislatives, for more or less obvious reasons.9

9

Most classifications of speech acts have more than three basic categories, mostly five. This is probably due to the fact that Austin's (1962) classification has been taken as prototype. His five categories were verdictives, exercitives, commissives, expositives, behabitives. Searle (1976), Vendler (1973). McCawley (1981) have astoundingly similar categories. For instance Searle has representatives, directives, commissives, expressives, and declarations. Searle criticizes Austin for not having based his classification on a clear and consistent principle. In contradistinction to Austin, Searle puts forward such principles. Vendler and McCawley seem to have mainly elaborated Austin's classification backing them with such principles. By using different principles, they arrived at approximately the same classification. The reason for this is, I believe, that Austin's proposal has been trusted somewhat more than the principles accepted. So, for instance, Searle's taxonomy of five classes boils down to three (or four) classes, if one adheres strictly to his principle of (direction of) fit [if one includes more of his classificatory principles a lot more than three or four classes result]: (1) the words have to match the world (representatives, expressives), (2) the world has to match the word (directives and commissives), (3) world and word need not match (fictives). The latter class does not occur in Searle's classification, however. Instead, there is a class (4) characterized by the conceptually more complex condition: world and word have to match each other (declaratives). In my reanalysis of Searle's proposal "world" has been taken in a very broad sense as including beliefs and emotions (thus "world" equals "context" in our sense of the word). If this is not done, the number of classes increases considerably, because now the directions of fit between at least words, worlds, and mental states have to be taken into account, and, more comprehensively, values. Expressives can be taken to be representatives of mental and emotional states. Commissives can be taken as a special type of directives. A systematic differentiation between the two classes would include an ego-alter distinction, which multiplies the speech act classes by about two. Distinctions between self-representatives and alter

342

T. T Ballmer

The graphs corresponding in an obvious manner to the three basic speech act types are the following:

(33a)

(1)

(2) V,

Τ

(3)

judge test assess, estimate depreciate

command order request plea ask, decide

c

express lament curse rejoice cheer

Instead of sentences Γαη we could also have displayed the conceptual (or mental) propositions α designated by Γα"1. Τ is a variable for values, such as truth, success, happiness, goodness, beauty values. When standing on the left of the interaction the entities such as sentences, contexts, and values are fixed. When standing on the right they are open. They become determinate from the left. A way to represent (33a) (1), (2) and (3) which is a little closer to the customary way of drawing our graphs is the following:

a 1 (2) Γα\ /

V

(3) CV

—\JL—c /

\

/\

(T)

Let us have a closer look at the judgments — or evaluations, or interpretation, as we may also say — of expressions used in a context. We consider here briefly the truth-functional, and a little more explicitly the success-functional evaluation of an expression in a context. First let us consider truth-functional evaluation. Truth-functional evaluation is done recursively. The truth-value of a complex expression in a context is calculated by reducing it to the truth-value of its parts. The principle underlying the reduction of meaning and truth-values to the meaning and truthvalues of the parts is called Frege's principle. It says that the meaning (truthrepresentatives would have to be included, which is of course not a very telling one. Declaratives (like definitions) are directives in the sense that the interpretation of words, a contextual factor, has to match the word. Social declaratives are directives in the sense that a social state has to fit the word, social changes of the context are performed.

Context and Context Change

343

value) of a composed expression is a function of the meaning (truth-value) of the parts. The meanings (truth-values) of the atomic expressions have to be assigned on other grounds. A simple example should demonstrate the meaning (truth-value) assignment to sentences of modal propositional logic. Let h be the function reducing the meaning (truth-value) of complex expressions to those of simpler expressions: (34)

hCV,c) = (c) (ΓπΊ: atomic propositional constant) r n Γ Ί h( —ia ,c) = -Ίη( α ,ς) (Γα"ι,Γβ"1: propositional formulas) h( r aA n,c) = h(ran,c) Λ h(r n,c) h(rD of the speech act of uttering rGive Anna the ballP in a context (H,CH), CH = 0: (36)

£(rGIVE ANNA THE BALLH,H,0) Ξ h^(t(rGIVE ANNA THE BALLP,H,0) = h (r(DxGive(x,A,Br,H,0) = h(r®xGive(x,A,By,H,Has(A,B)) = \

In accordance with Austin's and others' view the value of a speech act is trivially true or successful. Some (e.g. J. Sadock, personal communication) say that the truth of a speech act is presupposed. One might generalize this to say that the success is presupposed. An argument for this view is that it is impossible to oppose or refuse a speech act by replying NO! Another argument for this is that acts cannot be undone.10

5. A typology of speech acts based on context types

If context changing is as basic as it is claimed to be here, it should be possible to found an illuminating typology of speech acts on the types of contexts which are changed by speech acts and in which speech acts are embedded. On the basis of scanning through several thousands of speech act designating verbs,11 the following seems to be an important distinction of contexts: (37)

10

11

1. 2. 3. 4.

physical contexts mental contexts social contexts linguistic contexts

Both arguments are, taken at their face value, unsound. Replies to the speech act can be made, though by other means than just by saying no!, acts can be annulled and can be withdrawn. Berliner Gruppe (1975).

Context and Context Change

345

But, of course, a typology of speech acts of four classes is not very telling. Thus the following more detailed classes of contexts will be considered (this list is not totally exhaustive, but it is to a greater or lesser degree so): (38)

1. Physical contexts a. speech b. writing c. states of affairs 2. Mental contexts a. belief (knowledge) b. expectation c. attention d. emotion (ego) e. emotion (alter) f. obligation g. uncertainty h. memory i. conceptual k. thought 3. Social contexts a. institutional aa. law

ab. ac. ad. ae. af. b. ba. bb. be. bd. be. c.

state economy bureaucracy church moral public individual social values social structure social position social contracts social publicity existence of institution

4. Linguistic contexts a. expressions b. meaning c. reference d. discourse e. turn f. theme

In order to give an idea how speech acts can be classified according to these approximately thirty context classes, consider the following fragmentary list of speech act designating verbs. The speech act designated by such a verb alters contexts of the given type or // imbedded in contexts (situations) of the given type. (39)

1. Physical context changes a. Speech: b. Writing: c. States of affairs: 2. Mental context changes a. Belief: (Knowledge) b. Expectation; c. Attention:

repeat, reproduce, say, utter; note down, record, register, write down; desire, plan; inform, notify, report, say, tell; convert, demonstrate, persuade; instruct, lecture, preach, teach; foretell, plan, predict; accentuate, emphasise, stress; focus, point to, present; scrutinize, investigate; advise, recommend, suggest;

346

T. T Ballmer

d. Emotion (ego): e. Emotion (alter):

f. Obligation:

g. Uncertainty:

h. Memory i. Conceptual: 3. Social state changes a. Institutional:

aa. Law:

ab. State: ac. Economy: ad. Bureaucracy: ae. Church:

admonish, remind, spur on; alarm, warn; menace, threaten; curse, lament, moan, scream; caricature) deride, jeer, jest, joke, make a pun; insinuate, spit at, tease; cajole, coax, flatter, seduce; beg, implore, pray; surprise, shock, astound; commission, send for; delegate, depute, mandate; command, direct, enact, order; ask, consult, question; agree to, commit, guarantee, warrant; check, guess, test; clarify, elucidate; call into question, doubt, suspect; document, store; recall, remember; arrange, classify, group, mark, list, order, parallelize, rubrify, segment, structure;

advance, promote; expel, degrade, subordinate, suspend; charge, commission; swear; accuse, denounce, inculpate; arrest, apprehend, imprison; defend, plead for; condemn, convict, sentence; acquit, excuse, exculpate; outlaw, punish; appoint to an office, crown, delegate, depute, elect; exile, impeach; devaluate; sell, buy, contract; countersign, sign; /actuate; baptise, confirm, marry; bless, condemn, maledict; banish, excommunicate;

Context and Context Change

af. Moral: b. Public individual ba. Social values:

bb. Social structure: be. Social position: bd. Social contracts: be. Social publicity: c. Existence of institution:

347

censor, disallow, filter, prohibit, screen; appreciate, depreciate; attest, certify; acknowledge, thank; celebrate, glorify, praise; devalue, discriminate, ignore; call together, convoke, organise; argue, attack, fight, oppose; guarantee, insure, warrant; antedate, book, date, misdate, limit; advertise, cry out, declare, maintain, pronounce, publish; annul, suspend; constitute, found, establish;

4. Linguistic context changes a. Expressions: verbalise; encipher, decipher, paraphrase, summarise, translate; b. Meaning: define, interpret, redefine; c. Reference: call, dub, name, point to; d. Discourse: begin, preface, start; converse; discontinue, end, finish; e. Turn: break in, call on, intervene, howl down; begin to speak; conclude; f. Theme: elaborate, mention, raise, touch upon. These approximately two hundred speech-act-designating verbs give a hint as to how a speech act classification on the basis of a context typology can be set up. In view of the many more speech act designating (or aspects of speech act designating verbs), namely several thousands, the full structure of a classification could not be demonstrated. Here there is not space enough to work out the details.12 Remark 1: There is much to be said about what a speech act is, which expressions designate which speech acts, and what is changed by such a speech act designated by a linguistic expression. For instance take the word 12

The Berliner Gruppe (1975) has done extensive work of classification, subclassification, and structuring of the domain of speech acts and aspects thereof. Their ordering principle has been different from the one used here, however.

348

Τ. Τ Ballmer

plan under lc., which may be of a more controversial sort. Plan often designates a thought act, rather than a speech act. In this list we mean it however to designate a speech act, in accordance with one of its uses in natural language. This speech act could also more explicitly but more clumsily be described by tell one's plans. Uttering what one is going to do amounts in the normal case to a change of a physical state. This is the reason why plan has been listed under 1 (physical context changes). Of course, quite often the plans one utters are not fulfilled, for many reasons. But in a more or less strict sense the intent behind uttering plans is to alter the world according to them. Another intent behind uttering plans is to change the expectations of the addressees. That is the reason why plan has been listed under 2b also. Remark 2: There is a difference between a performative expression (say, a performative verb) and a speech-act-designating expression. Some of the speech-act-designating expressions may be used (with some small obvious modifications) to perform a speech act: thank, I thank you hereby. Many cannot be used such as: Curse! I curse hereby. The adequate way to curse is rather the use of expressions like damn, fuck you etc. We are interested here in speech acts, independent of the way they are performed, whether with explicit performatives (like / thank you hereby) or with illocutionary force indicators of some other sort (like damn). We therefore use as a convenient but very incomplete means, speech act designating verbs to get to grips with the problem of our interest. A fuller account would have to consider idiomatic expressions, too.

6. A typology of speech acts based on physical states, beliefs, and context change

One extreme stand one could take is to reduce mental, social and linguistic states all to mental states. An argument sustaining this view would have to show that social and linguistic states are in fact certain mental states. This is, according to our modern psycho-scientific paradigm, a plausible position to hold. Social relations and states are based in any case on individual beliefs, in that framework. Individuals learn about society by changing their beliefs in accordance to what they see and hear, physically and linguistically. Of course their innate predispositions may play a role too. Thus it could be argued that social states are nothing but certain beliefs pertaining to what others call social states. Social states are not taken to be part of the ontology, in that view, but only beliefs about social states. A similar argument could be run through, linguistic states, and hence the languages individuals speak, are merely mental. They are a special kind of belief. Now after having pleaded for taking social and linguistic states as mental states, namely certain beliefs, one may go further and claim that the mental states can all be reduced to certain beliefs. For the mental states representing

Context and Context Change

349

social and linguistic states, this has just been accomplished. For other mental states this reduction is done also quite easily. Epistemic states present no difficulty, because they by definition meet the requirements. Thus memory (recall, thinking of), plain present beliefs (knowledge, belief, perception) and expectations (expectation, hope, prediction) are states of belief or involve only states of belief. Moreover fictional states (dreams, thought hallucinations) are states of belief or involve only states of belief. States of abstract (theoretical, rational) existence, conditional semifactual, or counterfactual states and so on can be taken analogously. Obligational (deontic) states present more problems in being cast into the required form. The problems disappear if allowance is made for the following basic ontic things. (40)

1. physical states b. belief states e. state changes (context changes); more specifically changes on the basis of certain beliefs, and changes on the basis of certain physical states.

The state changes are taken to come about by active intervention by an agent. They do not correspond to the natural development of the context. They are, in correspondence to what has been said over and over again, context changes. The reduction of obligational states (deontic states) to certain states of beliefs is now as follows: Obligational states are beliefs. But they are not states about what has been the case, what is the case now, what is the case fictionally or abstractly, what could be the case, or what could have been the case: (40*)

Obligational states are beliefs about how to change a physical state (or more generally about how to change the context).

There are a whole lot of different kinds of obligational beliefs, if the analysis is pushed to more detail, depending on who the originator of the belief is (and who sanctions the violation, if the obligation is not fulfilled), who carries the obligation out, and what the obligation is about. Distinguishing the three important roles authority 51, subordinate S, recipient 9Ϊ, there are et least the following different notions of obligation: (40 a) (40b) (40 c) (40d)

Intention: Commitment: Advice: Command:

φ2'1: = Γφ1=>φ2"1. This is motivated by the interpretation of these formulas. For this reason it is not necessary to introduce the classical connectives independently. What has been said so far is caught in the following (translational) interpretation by h and ξ: (69 a) h(V>i) = p ,1; _ (70a) (70b) ξ(ΓωΛΐ) = i + 1 (70c) ξ('--ΐαΛΐ) Ξ ξ(ΓαΛι) (70d)

,,

fa-·, i ) r ^ l(i) n

if h( -ia ,i) =

(70e)

if if h(r-iaV) = As is always the case for such translational interpretations, the aim is to eliminate the nonstandard constants like ""Q"1, r A "*, r=>"1, and to do this in an optimal way, that is as directly and shortly as possible. In order to see the interpretation at work we must specify also how the definitional formulas are interpreted. A very simple case is where there is exactly one term the value of which can be defined. Let us consider this case first and proceed to more general ones afterwards. 8.3. Procedural logic with one definable term LPROCI

Let Γγπ be an arbitrary term, let Γδπ be a constant (i.e. undefined) term, and let Ύ1 be the definable term. Let rffi be a constant (i.e. undefined) function term. The interpretation of the definitional formula ra: = γ"1 is as follows: (69g) h(ra: = YV) = hfa^i + l) = h(rYn,i)

(69h) hCW) = γι (69i)

(= l(i) according to (69b))

364 (69k) (70g) (70h) (70 i)

(70k)

Τ. Τ Ballmer

h ξ(Γα: = γ\ϊ) Ξ i + 1 ξ(γ·,ί) = i ξ( Γ γι

The concrete instrument designated by ""ψ"1 is determined contextually, the effect on the context is determined in turn by the specific instrument used in that context. In order to determine action patterns more transparently, an 21

These complexities will not be elaborated here. It must be stressed however that they have an enormous relevance for human action and communication. Most of human and nonhuman actions are approximately but not exactly reversible.

366

Τ. Τ Ballmer

operator measuring the state of unsatisfactoriness relative to a specific type of action is helpful. rUnsat(x)"" is constative. As such it does not change the context. It is simply true or false. '"Unsat(x)"1 is true in context i iff the agent in question is unsatisfied relative to the goals of action χ. Thus (74)

r

Q(eat-a-bit A Unsat(eat-a-bit)"1

describes the action sequence of an agent in question to eat a bit, as long as he is no longer unsatisfied relative to the result of eating a bit, in other words to eat a bit until he is satisfied. TJnsat^)"1 serves as an interrupt mechanism of the iterative process programmed by rQ(x A Unsat(x)"1. We may generalize the use of the ""Unsat"1 operator to contextual instrumental formulas. An index at Γψ~ι may delimit the kinds of process units designated by ""ψ"1. rysAn may designate one contextually determined speech act. rUnsat(\|/sA)~" may test the contextually determined unsatisfactoriness relative to the set S A of speech acts. In order to represent action sequences, dialogues and texts with a procedural logic, there should be a way to specify the reason why such an action sequence and so forth is undertaken. Thus an operator must be available in the language which creates an unsatisfactoriness which can then in turn be recursively eliminated. Let Tlnsat"1 be that operator. Syntactically it can be applied to rigid and to contextual instrumental formulas just as can TJnsat"1. Semantically it creates unsatisfactoriness "out of the blue," relative to the prevailing context, of course. A more transparent way to represent an action sequence, dialogue or text formally is then: (75)

r

flnsat(v|/A) A Q(\)/A A

According to this formula, an unsatisfactoriness of a specific kind A is (contextually) created first, then the elimination of the unsatisfactoriness is undertaken recursively by contextually determined actions ψ A until an unsatisfactoriness relative to actions ΨΑ no longer obtains. With this, we could leave off this study in procedural logic. A most important notion of human and even of nonhuman society has found its way into formalization: action patterns together with their initiation. However it seems to me that a few further remarks and illustrations may prove useful for a better understanding of what has been achieved so far.

9. Applications of procedural and cybernetic logic 9.1.

Unsatisfactoriness

Before we discuss some seven regions of applications for procedural logic let us make a remark concerning the semantics of the notion of unsatisfactoriness (flnsat, Unsat). This is not a terribly difficult matter, in fact. The set of all possible contexts is divided ideally in two subsets, the satisfactory one, and

Context and Context Change

367

the unsatisfactory ones. Thus a full context is specified by (at least) a pair (i,I) the context properly spoken and the set of satisfactory contexts. Satisfactoriness is characterized by: (76)

A (proper) context / is satisfactory relative to the context class / iff / ε/.

Unsatisfactoriness prevails iff i έ I. There are two principle ways to introduce an unsatisfactoriness, either by changing / such that / έ I, or by changing 7 such that / έ I. The first case occurs in case the (proper) context develops to become dangerous, unpleasant, and so forth, the second case occurs if one makes decisions, is put under obligations and so forth. If we call / the Istvalue and / the Soil-value according to a common usage in cybernetics and (electronic) engineering, unsatisfactoriness obtains, if the Ist-value does not coincide with one of the Soil-values. This interpretation makes the logic LPROC enhanced with ""Unsat"1 and Tlnsat"1 operators a cybernetic logic. Regulations and controls may be represented most conveniently in Lern.22 We now define elementary events as processes starting with a fission of an Ist-state and a Soil-state at a time (in an environment) and the subsequent iterative fusion of the Ist-state and the Soil-state by a sequence of actions, operations, utterances, and so forth.

(77)

·

'

fusion)

... (fusion

I

An elementary event The formula (75) represents just such elementary events. The pictorial connection of (77) with the graphs used earlier in the paper is immediate. 9.2. Patterns of behavior A few typical instantiations of (75) will be given subsequently, first two formulas characterizing two important patterns of behavior: (78a) (78b)

r r

Create(Fear(\|/A)) A Q(\)/A A Create(Agression(\|/A)) A Q(\|/A A Aggression(\|/A))n

These formulas tell that at first a certain emotion, i.e. an unsatisfactoriness of a more specific kind, is created; that then actions ΨΑ are undertaken as long as this emotion prevails. In other words LPROC provides a specialization for

22

Cf. also Brennenstuhl (1976).

368

Τ. Τ Ballmer

a logic LAE of actions and emotions. Emotions function here as conditions for initiation and termination of action sequences. As such LPROC may prove to be useful for social sciences such as biology, sociology and psychology. 9.3. Computer programs

Another specialization of (75) characterizes computer programs. The task of the program is to remove a lack of knowledge. The lack of knowledge of the result of the calculation process is the specific unsatisfactoriness for the calculation. Thus (79)

r

Create (Ignorance(v|/A)) A Q(V|/A A Ignorance(yA))"1

or more abstractly (80)

""Create (Problem(\]/A)) A Q(yA A

This latter formula can be read as creating a problem and operate and test recursively until the problem has disappeared. The recursive kernel r Q(v|/A A Problem(yA)"1 can hence be rewritten in a way that brings it in close relationship to the TOTE (Test-Operate-Test-Exit) units of Miller-GalanterPribram (1970):

(81) or even more so, using an identity operator on actions ""Operate"1: (82)

r

Q(Operate(VA) A Test(VA)"1

Miller et al. did not formulate their TOTE units in logical language, which is recursively interpreted. An essential deficiency is moreover, that they do not pay attention to problem creation. 9.4. Mathematical proofs

Mathematical proofs may be cast, at least under certain decidability conditions, in a logical form similar to (75). Abstractly a mathematical proof has the form: (83)

""Create (Open-Claim(\|/A)) A Q(V|/A A OpenClaim(yA))"1

A determines, essentially, the range of admissible proof techniques, moves in the proof game etc. It is a well-known fact that proofs are meaningful only relative to such a restricted arsenal of proof instruments (ruler and compass, algorithms, model-theoretic means, ...). 9.5. Evolution strategies

Biological evolution strategies (cf. Rechenberg, 1973) can be cast in a logical form similar to (75). The basic idea behind evolution strategies is that they

Context and Context Change

369

work all the time generating mutations on populations and selecting them according to the survival value they have. Only the advantageous mutations are changing the population however in such a manner that future mutations operate on them. What counts as advantageous may be context dependent, as well as what counts as a mutation. Because evolution need not be started explicitly, no creating of an unsatisfactoriness component need be present. Moreover evolution never terminates. If a huge biosocial structure has been built up by evolution, it may well be that it breaks down totally one day — but evolution survives: it simply continues at the level which is left. Thus an evolution strategy has the following logical form: (84)

Γ

Ρ(ψΑ A

ψ A is a context dependent mutation of a set A of possible mutations. The second component of the hereditary conjunction Γ ΑΊ accounts for neutralizing the mutation if the value (survivor value, moral value, or whatever) is worse than that value before. This second component is true in all contexts, which guarantees the eternity of evolution. Evolution has no start and no end. It is a non-terminating and yet interesting algorithm, because it leaves a trace in the universe. That nonterminating algorithms are of interest has been overlooked by any mathematicians, logicians and specialists in computation. Evolution is a case in point. The reason why the second component of (84) is always true should be made explicit, because it touches an important point of evolution strategies. Either ΨΑ results in a better situation, such that rworse(val, before(val))"1 is false (the value "val" is not worse than the value "before(val)" before) or ΨΑ results in a worse situation. But then the second component is true because the inverse mutation ψΑ"1, being an operative formula is true, hence the second component is true. Moreover, the first component of (84), because it is an operative formula is always true (there are no unsuccessful mutation attempts). Thus the whole (84) is always true. This fact expresses the tautological character of the evolution principle of mutation and selection, over and above the nonempiricity which is based on the arbitrariness of what the survival values are. 9.6.

Texts, dialogues, speech act sequences

A fifth application of (75) is linguistics. Texts and dialogues, speech act sequences in general, are easily recognized as being of the form (75). Every theory of literature is based more or less directly on the fact that speech act sequences begin with an instigation of a problem and the subsequent transformation, that is recursively changing the affairs until the resolution comes about.23 23

Cf. Plato (Kratylos), Propp (1970), Greimas (1966).

370

Τ. Τ Ballmer

Often, if not regularly, the transformation involves a complication. Such a complication is a problem within a problem. Therefore more adequate logical forms of (say) texts involve an iterated application of (75). Accordingly we have the following logical forms for speech act sequences: (85) (86)

r r

Create(Problem(\|/A)) A Q(YI/A A Problem(v)/A)) Create (Problem(\jfA))) A Q(V|/A A (Create(Problem (ψΑ)) A Q(\|/A A Problem(\|/A)) Α Ρτο^6Γη(ΨΑ)

These forms are similar to simple and iterated programs (72). An appropriate context specification may render (75) as follows: (87)

r

Has(John,Key) A Looses(John,Key) A Q(Look-for(John,Key) A

According to this specification of (75) the problem consists in John's no longer having his key, he had it and lost it afterwards. The transformation consists in John's making search movements for the key as long as he does not have it (has not found it). Often a text ends with stating that the problem is solved, that is, instead of (75) we would sometimes have the more complete: (88)

r

Create(Problem(\|/A)) A Q(V|/A A Problem(yA)) A —ι Problem^)"1

and instead of (87): (89)

r

Has(John,Key) A Looses(John,Key) A Q(Look-for(John,Key) A ~iHas(John,Key))A HasQohn.Key)··

Three of the texts which may be generated by these programs, that is formulas of LPROC are the following: (90a)

John has a key. John loses the key. John looks for the key. John has the key. (90b) John has a key. John loses the key. John looks a long time for the key. John finds the key. John has the key again. (90c) John has a key. John loses his key. John looks for the key in his bedroom. He looks for it in the kitchen. He also looks for it in the bathroom. He finally finds it in the cellar. According to the context, different texts may be developed out of such procedural formulas. More interesting texts are generated by (86), because of arbitrarily many complications which may arise during the solving process for the initial problem. One may ask whether the one time embedded generation of new problems is sufficient, or whether even in the dimension of generating deeper and deeper problems there should be some recursiveness. We leave the question and its possible solution open. But a little acquaintance with actually occurring texts should show that there is no upper bound to the degree of embedding of created problems.

Context and Context Change

371

Dialogues deserve special attention. They can be generated similarly to texts. But in addition to a logical form like (75) (75)

r

Create(Problem(v)/A)) A Q(\|/A A Problem^))"1

the response behavior of addressees must be programmed also. For instance the question of the speaker should initiate a problem for the addressee, namely the problem of answering his question or passing. Thus a dialogue is characterized by a simultaneous set of two formulas, of which one is designated to characterize the starter, the other the responder (responders): (91 a) (91 b)

r

Create(Problem(\|/A)) A Q(\|/A A Problem(yA)) A ~i Problem^)"1 Q(\|/A A Problem(v|/A)n

r

The creation of problems for (91b) is done by (91a). A contextual specification of this set of logical forms could be the following: (92a)

r

—l Know(John,p) A Want(John,Know(John,p)) A Q(Ask(John,p) A Q(Wait(John) A~l VxHear(John,x)) A ~~iKnow(John,p)) A Know(John,p) n (92b) rQ(VxAnswer(Addressee,x) A Asked(Addressee) A ~~i Answered (Addressee))"1

Thus question-answering systems find a natural way of representation in or more specifically in LCYB by simultaneous formula systems. 9.7. Physical systems, physical events As a final concretization of (65) let us consider the case of physical systems and physical events. Let us specify (65) accordingly: (93a)

r

!nitial-Condition(v|/A) A Q(\|/A A nFinal-Condition (ψΑ))~"

or

(93b)

r

Q(yA A l Boundary-Condition (ψΑ))π

For physical systems and physical events, the initial condition determines to a greater or lesser degree (deterministically or stochastically) the future evolving of the system. Thus the initial conditions characterize the drive deficit, i.e. the creation of unsatisfactoriness of the system. The final condition characterizes the termination of a motion. As such its negation represents the unsatisfactoriness itself. Similarly the negation of the boundary condition represents an unsatisfactory state of the physical system, the boundary conditions are not fulfilled.

372

Τ. Τ Ballmer

The formulas (93a) and (93b) stand in close formal analogy to integral equations and differential equations characterizing the kinematics (and even the dynamics) of physical systems: (94a)

|ψ> = f |φ> αφ

(94b) Α. ψ = ΗΨ at (94c)

χ =

vdt XO

Given the obvious identifications, the analogy is striking.

10. Conclusion We have now exhibited a series of specializations of the formula (75)

r

fl nsat(\j/A) A Q(\|fA A

and some of its variants. Many if not all of these specializations described kinematical systems: elementary events, patterns of behavior, computer procedures, mathematical proofs, (biologically) evolving systems, speech act sequences, dialogues, physical systems. Important linguistic systems like speech act systems, texts, discourses and dialogues have been set on a par with other kinematical systems. We should consider this as the beginning of a new area in linguistics. Up to now linguistics has been largely taxonomic. Starting from the collection and ordering of data, linguistics has more and more developed taxonomical techniques on many levels: on the phonological, morphological, syntactical, semantical level. Taxonomic studies now begin to embrace universal grammar, i.e. they are no longer restricted to specific languages but include cross-linguistic classifications. The typology of speech acts is another recent area of taxonomical investigation in linguistics. But never until now, has the kinematics of linguistics been recognized24 and worked out accordingly. There are some taxonomical studies of global change,25 but there is no kinematical theory of such changes, mainly because of the lack of such a theory on a more local level, namely on the level of linguistic action, that is of specific language use by speaker and hearers. 24 25

There must always, of course, be some important exceptions to such categorical statements. Lieb (1970), Vennemann (1973), Givon (1975).

Context and Context Change

373

A formula like (75) characterizes the kinematics of (linguistic) actions. It does this in straightforward analogy to kinematical equations in physics (equations of motions) and biology (evolution by mutation and selection). We may thus safely state that with the establishing of (75), it has been possible to develop the intuitive idea of the instrumental character of linguistic expressions in language use with the help of the notion of context change to a point at which taxonomical linguistics is left behind. The next step is to go beyond kinematics to the dynamics of action sequences. In addition to passive and active context changes, the dynamics will provide us with laws of these changes. It may well be that the formula (75) already displays the essentials of a dynamical law of action sequences and hence of all the special cases we touched upon in our previous discussion. The formulation of a dynamics of actions must include a precise and explicit comprehension of the driving forces, in the linguistic case this means a precise and explicit comprehension of speech act forces (illocutionary forces). This more specific work for a dynamics of actions requires, as far as I can see, a fair amount of work in the taxonomy of actions and action sequences.

Literature Aqvist, Lennart 1971 a Modal Logic with Subjunctive Conditionals (= Filosofiska Studier 12). Uppsala. 1971 b "Revised Foundations for Imperative Epistemic and Interrogative Logic", Theoria 37, 33-73. 1975 "Formal Semantics for Verb Tenses as Analyzed by Reichenbach", in T. A. van Dijk (ed.): Pragmatics of Language and Literature. Amsterdam: North-Holland, 229 — 236. Austin, John L. 1962 How to Do Things with Words. Oxford: Clarendon. Ballmer, Thomas T 1971 "Gründe für eine formale Pragmatik" [Reasons for a Formal Pragmatics], in K. Hyldgaard-Jensen (ed.): Linguistik 1971. Frankfurt: Athenäum, 266 — 284. 1972a "Einführung und Kontrolle von Diskurswelten" [Introduction and Control of Discourse Worlds], in D. Wunderlich (ed.): Linguistische Pragmatik. Frankfurt: Athenäum, 183-206. 1972b Eine Semantik für Artikel. Technical University, Berlin: Mimeo. 1973 Pilot Study in Text Grammar. Technical University, Berlin: Mimeo. 1974 "Macrostructures", in T. A. van Dijk (ed.): Pragmatics of language and Literature. Amsterdam: North-Holland 1975. 1975 Sprachrekonstruktionssysteme [Language Reconstruction Systems]. Kronberg/Ts.: Scriptor. 1978 Logical Grammar: With Special Consideration of Topics in Context Change. Amsterdam: North-Holland. 1981 "Context Change and its Consequences for a Theory of Natural Language", in Parret, H.; Sbisa, M.; and Verschueren, J. (eds.): Possibilities and Limitations of Pragmatics. Amsterdam: Benjamins, 17 — 55. 1982 Biological Foundations of Linguistic Communication: Towards a Biocybernetics of Language. Amsterdam: Benjamins.

374

T. T Ballmer

Ballmer, Thomas T and Brennenstuhl, Waltraud 1972 "Versuche einer Fassung handlungstheoretischer Begriffe unter Bezugnahme auf eine Analyse von Handlungsabläufen während eines Restaurantbesuchs". Mimeo. 1981 Speech Act Classification: A Study in the Lexical Analysis of English Speech Activity Verbs. Berlin: Springer. Berliner Gruppe 1972 — 1975 Sprachliches Handeln (Kategorien, Listen, Modelle). Forthcoming. Bierwisch, Manfred 1965 "Zum Textbegriff", in J. Ihwe (ed.): Literaturwissenschaft und Linguistik I. Frankfurt: Athenäum 1971, 141-154. Boole, George 1847 The Mathematical Analysis of Logic, being an Essay toward a Calculus of Deductive Reasoning. Reprinted 1948. Oxford: Blackwell. Brennenstuhl, Waltraud 1973 Eine Theorie für erfolgreiche und erfolglose Handlungen. Mimeo. 1975 Handlungstheorie und Handlungslogik [Theory of Action and Logic of Action]. Kronberg/Ts.: Scriptor. 1976 "What We Can't Do" in Proceedings of the Twelvth Regional Meeting of the Chicago linguistic Society. 1982 Ability and Control. Amsterdam: Benjamins. Carnap, Rudolf 1947 Meaning and Necessity. Chicago: University of Chicago Press, van Dijk, Teun A. 1976 Text and Context. London: Longman. Eikmeyer, Hans-Jürgen and Rieser, Hannes 1981 "A Formal Theory of Context Dependance and Context Change", in: Th. T Ballmer and M. Pinkal (eds.): Approaching Vagueness. Amsterdam: North-Holland, 131 — 188. 1985 "Procedural Grammar for a Fragment of Black English Discourse", in Th. T Ballmer (ed.): Linguistic Dynamics (= Research in Text Theory 9). Berlin: de Gruyter, 85-178. Eikmeyer, Hans-Jürgen and Rieser, Hannes (eds.) 1981 Words, Worlds, and Contexts (= Research in Text Theory 6). Berlin: de Gruyter. Givon, Talmy 1975 Promotion, Accessibility and the Typology of Case Marking. UCLA: Department of Linguistics. Gödel, Kurt 1930 "Die Vollständigkeit der Axiome des logischen Funktionenkalküls", Monatshefte Math. Phys. 37, 349-360. Goodman, Nelson 1955 Fact, Fiction and Forecast. New York: Bobbs-Merrill 1973. Greimas, Algirdas 1966 Semantique structurale. Paris: Larousse. Grice, Herbert P. 1957 "Meaning", Philosophical Review LXVI, 377-388 and Semantics, 53-59. 1969 "Utterer's Meaning, Sentence Meaning, and Word Meaning", Philosophical Review 78, 147-177 and 1971, in J. R. Searle (ed.): The Philosophy of Language. Oxford: Oxford University Press, 54—70. Heim, Irene 1982 The Semantics of Definite and Indefinite Noun Phrases. Unpublished Ph.D. Thesis, University of Massachusetts, Amherst. Kamp, Hans 1981 "Theory of Truth and Semantic Representation", in: J. A. G. Groenendijk et al. (eds.): Formal Methods in the Study of Language, I. Amsterdam: Mathematisch Centrum, 277-322.

Context and Context Change

375

Koschmieder, Erwin 1929 Zeitbe^ug und Sprache: Ein Beitrag %ur Aspekt- und Tempusfrage. Leipzig: Teubner, 62 ff. 1935 "Zur Bestimmung der Funktionen grammatischer Kategorien", in Beitrage ^ur allgemeinen Syntax. Heidelberg: Winter. 1969, 9-69. Kripke, Saul A. 1963 "Semantical Analysis of Modal Logic I", Zeitschrift für Math. Log. und Grundlagen der Mathematik 9, 67 — 96. 1965 "Semantical Analysis of Modal Logic II", in J. W. Addison et al. (eds.): The Theory of Models. Amsterdam: North-Holland, 265-300. Lewis, David K. 1973 Counter/actuals. Oxford: Blackwell. Lieb, Hans-Heinrich 1970 Sprachstadium und Sprachsystem. Stuttgart: Kohlhammer. McCawley, James D. 1981 Everything That Linguists Have Always Wanted to Know about Logic but Were Ashamed to Ask. Oxford: Blackwell. Meibauer, Jörg 1982 "Akte oder Verben oder beides? Rezension von Ballmer-Brennenstuhl 1981", Zeitschrift für Sprachwissenschaft l, 137 — 148. Montague, Richard 1970 Formal Philosophy. R. Thomason (ed.), 1974. London: Yale University Press, von Neumann, John 1928 "Die Axiomatisierung der Mengenlehre", Math. Z. 27,669 — 752, corrected in 155,128. von Neumann, John and Morgenstern, Oskar 1943 Spieltheorie und wirtschaftliches Verhalten. Würzburg: Physica Verlag, 1967. Petöfi, Jänos S. and Heydrich, Wolfgang 1981 "Pragmatic Considerations within a Text Theoretical Framework", in H. Parret et al. (eds.): Possibilities and Limitations of Pragmatics. Amsterdam: Benjamins, 313 — 330. Petöfi, Janos S. and Kayser, Hermann 1978 "Sprechhandlungen und semantische Interpretation", in R. Meyer-Hermann (ed.): Sprechen — Handeln — Interaktion. Tübingen: Niemeyer, 1—48. Propp, Vladimir J. 1970 Morphology of the Folktale. Austin: University of Texas Press. Rechenberg, Ingo 1973 Evolutionsstrategie. Stuttgart: Frommann-Holzboog. Rehbein, Jochen 1972 "Entschuldigungen und Rechtfertigungen", in D. Wunderlich, 288 — 317. Rehbein, Jochen and Ehlich, Konrad 1972 "Zur Konstitution pragmatischer Einheiten in einer Institution: Das Speiserestaurant", in D. Wunderlich, 209-254. Russell, Bertrand and Whitehead, Alfred N. 1910 Principia Mathematica. 3 vols. Cambridge: Cambridge University Press. 1st ed. 1910-1913, 2nd ed. 1925-1929, new ed. 1950-1960. Schiffer, Stephen R. 1972 Meaning. Oxford: Clarendon. Searle, John R. 1965 "What is a Speech Act", in J. R. Searle (ed.): The Philosophy of Language. Oxford: Oxford University Press (1971), 39-53. 1969 Speech Acts: An Essay in the Philosophy of Language. Cambridge: Cambridge University Press. 1976 "A Taxonomy of Illocutionary Acts", in Gunderson, K. (ed.): Language, Mind and Knowledge (= Minnesota Studies in the Philosophy of Language VII). Minneapolis: University of Minnesota Press, 344—369.

376

Τ. Τ Ballmer

Strawson, Peter F. 1964 "Intention and Convention in Speech Acts", Philosophical Review 73, 439—460. Tarski, Alfred 1935 "Der Wahrheitsbegriff in den formalisierten Sprachen", in Stud. Philos. Comm. Soc. Phil. Pol. 1, Lemberg, and in K. Berka and L. Kreiser (eds.): Logik-Texte. Berlin: Akademie, 1971, 447-559. Vendler, Zeno 1967 linguistics in Philosophy. Ithaca: Cornell University Press. 1972 Res Cogitans: An Essay in Rational Psychology. Ithaca: Cornell University Press. Vennemann, Theo 1973 "Explanation in Syntax", in J. P. Kimball (ed.): Syntax and Semantics, Vol. II. New York: Seminar Press, 1 — 50. von Wright, Georg H. 1963 Norm and Action. London: Routledge and Kegan. 1966 "The Logic of Action: A Sketch", in N. Rescher (ed.): The Logic of Decision and Action. Pittsburgh: University of Pittsburgh Press, 121 — 136. Wunderlich, Dieter (ed.) 1972 Linguistische Pragmatik. Frankfurt am Main: Athen um.

WOLFGANG HEYDRICH Things in Space and Time* /. Introduction

1.1. It has been proposed1 that understanding natural-language texts — and perhaps understanding texts within arbitrary symbolic systems — can be conceived of as an interactive process of making comparisons between content and context, between what is being told about something and what something is being told about, between a situation which is supposed to be presented and a situation which is presupposed to be present, between a world picture which derives from the structure of a text and a structure which gives a picture of the world. According to this view, the core of text theory is a concept of interpretation which provides — roughly — for a link between texts and the world, where neither texts nor the world (or worlds) are naively assumed to be given directly, but are construed interpretatively as conceptualizations, meaningful structures, entities with import. Our rejection of both text and world as immediately given echoes the old anti-empiricist critique of the given in favour of conceptualization, construction and world-making. In Goodman's (1978: 3) words: We are confined to ways of describing whatever is described. Our universe, so to speak, consists of these ways rather than of a world or of worlds.

If we try to develop a theory of text-interpretation in terms of the comparison of meaningful structures (situations, world-pictures or whatever label seems to be appropriate), some fundamental questions are itching to be answered. What is the ontological status of the entities to be compared? And — provided these entities are made instead of given, i.e. result from constructions — we must say how they are construed and from which supposedly more elementary entities. The question concerning the ontological status of meaningful entities involves us in the notoriously difficult theory of meaning. Following the nowadays well-worn paths within the field of philosophical logic immediately leads us into the marshy regions of essentialism, where alligators like possible, impossible, and fictional entities lurk side by side with propositions and * My thanks to J. S. Petöfi, H.-J. Eikmeyer and G. Tan for discussions and valuable advice. 1 Cf. Heydrich and Petöfi (1979) and Petöfi (1982).

378

W. Heydrich

properties. Fortunately, the topic of this paper allows us to avoid these dangerous zones. Let me merely say in passing that I am sympathetic with some extensionalist (or even nominalist) account of meaning and related topics which is free of ontological commitments to intensional or hyperintensional entities (or even to abstract entities like sets)2. The preference for such an approach combines anti-empiricist scepticism towards the given with anti-essentialistic reservations towards the distinction between synthetic and analytic truths. Additionally it is sympathetic to actualistic constructivism, according to which meaningful structures have to be constructed exclusively by reference to actual entities. According to actualistic constructivism, talk about the fictive (possible or impossible) can be analysed as talk about the actual in disguise. In order to characterise the central point let me quote a passage from Rescher (1975: 8): The key point is that actuality, and the descriptive apparatus we frame — rightly or wrongly — to describe it, is for us the measure of all things: of that which is, what it is, and of that which is possible, what manner of things is at issue. The only specific possibilities that can exist for us are those that are conceived by us — those that lie within the range of the cognitive grasp of our reality-descriptive mechanisms. When this perspective is objectivized, we are led to the stance that the only possibilities that can exist per se are those articulated in terms of reality-descriptive mechanisms that are available in principle (even though they have not yet been devised, and perhaps never shall be in practice).

Anti-empiricism, anti-essentialism and constructive actualism set the scene for an ontologically neutral semantics of natural-language texts. It deals with representations of "ways of describing" as well as with an indication of "whatever is described". The interactive reconstruction of content and context (both construed as pairs of representation and indication) will (hopefully) allow us to model the process of text understanding empirically. It is not the intention of this paper to give a report on the work currently being done to spell out these ideas about text interpretation and the understanding of natural language texts in detail3. I am concerned here with things in space and time. This is a very general topic if looked at in isolation and it is a rather special problem, although an immensely important one, with respect to the task of providing for a general theory of text interpretation. Let me explain. The features of temporality and locality are omnipresent within naturallanguage texts. There are words and phrases within virtually all natural languages which name or describe certain moments or periods of time and 2

3

Extensionalists and nominalists are normally sceptical with regard to the theory of meaning. Consequently, there is not much constructive work done in this field. There are, however, some exceptions. I want to mention Martins's (1969, 1976) theory of nominal virtual classes and Eberle's (1978a, 1978b) elaboration of Goodman's (1949, 1953) proposals concerning the notion of likeness of meaning. My own ideas are developed provisionally in Heydrich (1983). Cf. Heydrich, Neubauer, and Petöfi (forthcoming).

Things in Space and Time

379

places, reference to space and time is additionally made by local and temporal deixis, e.g. by tensed verbs or special indicator words. Accordingly, our project of providing construction of content and context for natural-language texts must account in one way or another for these temporal and local features. Whatever course we take, some concept of the general nature of space and time is necessary. It might be maintained that the generality of such a concept can be restricted, because adequacy is required only within text theory and must be seen relative to the specific aims of text-theoretical interpretation. But, obviously, it is at least desirable that our conception of space and time within text theory be compatible with the nature of space and time as analysed quite independently of specific text-linguistic questions. Note that, for example, physical theories which deal with or presuppose notions of space and time are texts such that their interpretation also falls into the domain of text-theoretical research. The same is true, of course, of commonsense talk, science-fiction stories, mythical tales and other works of fiction which seem to entail concepts or stereotypes of space and time of a different nature. Hence, ideally, text theory requires a concept of space and time which is compatible with — although not committed to — what natural science says about what space and time really are. As is often the case, serious semantics for natural languages involves us here in ontological, metaphilosophical and metaphysical questions. I said above that constructive actualism tries to construe meaningful structures like content and context out of actual entities alone. Assuming that such an account is not doomed to failure from the start, we have to face the question of what makes entities actual, i.e. we have to somehow characterise the domain of actual entities. It is tempting to suppose that at least objects and events can be considered as actual entities, and, as a matter of fact, objects and events play important roles within the construction of meaningful structures that we have in mind. Supposedly, certain configurations of objects and events may be introduced as states of affairs, and complexes of states of affairs seem to be convenient constructions to capture the import of both content and context. However, in order to explicate the difference between objects and events and in order to provide for clear principles of individuation, i.e. conditions of identity for these kinds of entities, it seems to me that we have to dig deeper. Thus we are led to a more formal characterization of the domain of actual entities. We seek a group of principles which guarantees unambiguous conditions of identity for whatever is accepted as an actual entity and which is in accordance with the principle of ontological neutrality, in the sense that it does not conflict with the frugal ontology of nominalism. Such a group of principles is provided for by mereology or (under a different name) a calculus of individuals, which explicates the part-whole relation among entities. The starting point for actualistic constructions is, according to this proposal, the domain of the part-whole relation under some

380

W. Heydrich

of its different explications. Such an explication introduces what has been called a universe of individuals, and the weakest of such different possible explications which provides for the most general characterization of universes of individuals might legitimately be considered as definitional for modern nominalism4. Adopting a nominalist version of actuality, constructive actualism turns out to be the approach of construing everything needed within the interpretation of natural-language texts (or texts in general) exclusively by means of reference to individuals. Talk about objects, states of affairs, situations, content and context has to be analysed such that it is committed ontologically to individuals alone. The present paper is not concerned with the justification of such a program and it does not argue against alternative platonistic paradigms. Perhaps, it will be interesting only for those who feel some attraction for a nominalist orientation within the fields of text-theoretical semantics, linguistics, and philosophical logic, the more so as I will restrict myself to some rather elementary, although fundamental, actualistic constructions. I aim at the introduction of objects and events which are assumed to be of great importance for what is currently called the theory of reference. The construction of other likewise important semantical entities (e.g. states of affairs and situations) lies beyond the scope of this paper5. Definitions for the notions of objects, and events will be formulated in section 3.2. and 3.3. My proposals presuppose the demarcation of some special universes of individuals, and this demarcation will be given in section 2. by means of a list of twenty-one postulates formulated within the usual language of first-order logic with two added primitive predicates. I hesitate to call this list of postulates an axiomatic system since I am not at all sure that the postulates are independent of each other and whether the list gives an exhaustive characterization of the intended notion of individuals. Adequacy and independence could be secured by means of well-known model-theoretic techniques. Although I do not yet have the respective proofs, their accomplishment will however be prepared by the characterization of the notion of a model for the system of postulates (cf. section 2.9.). Let us suppose that proofs of consistency, completeness and independence with respect to the models specified can be carried out. (Possibly, amendments concerning the system of postulates as well as modifications of the notion of models would be necessary to accomplish this.) Then, clearly, these proofs, making use of set-theory, will not fit into the restricted nominalistic ontology 4

5

The classical versions of mereology are due to Lesniewski (1927 — 31) (cf. Luschei, 1962) and Leonard and Goodman (1940). A systematic study of different notions of a universe of individuals is Eberle (1970). The weakest explication of the part-whole relation referred to in the text is given by his calculus CI IV (cf. 80f.). More is said with respect to these entities in Heydrich (1982, 1983).

Things in Space and Time

381

of the system itself. Their role is to assure the success of nominalistic constructions. They are best considered as Wittgensteinian ladders to be kicked away once the task is accomplished. 1.2. What about things, space, and time with respect to the project of actualistic constructions? There are two possibilities. They might be identified with one sort or another of actual entities (i.e. individuals) or they might result from constructions which take individuals as fundamental. They might, for instance, be construed as virtual classes of individuals in the sense of Quine's and Martin's term 'virtual class'6. (This is the course I will take in my construction of objects and events in section 3.) Here, I opt for the first possibility. To be more precise, I will identify things with certain individuals which are composed out of so-called atoms (i.e. individuals which are identical with each of their parts and, hence, cannot be divided into several sub-parts) and so-called conglomerates (i.e. individuals which do not contain atomic parts and can thus als way s be divided into several subparts). The intuitive reasoning behind this decision is that I consider substance to be composed of mereological atoms (e.g. particles of matter) and both space and time as individuals without substance. Since I do not differentiate space and time in regard to their mereological structure, conglomerates are best considered as spatio-temporal individuals or areas of space-time. Things are thus conceived of as mereological sums of substance and areas of space-time, where we require that the areas are not arbitrarily chosen. Rather, the area within a thing should be interpreted as the space-time which is occupied by the thing's substance. The extension of the area depends on what space-time is occupied by the atomic parts of the thing; and I stipulate that for each atom there is a unique maximal area of space-time it occupies. — The details of this sketch will be elaborated more thoroughly in section 2. Let me point out here two peculiarities of my account. Firstly, I do not conceive of space and time or space-time as structures to be defined independently and correlated with things by a second step. This two-stage procedure is common within interpreted systems of temporal logic7 as well as within the formal semantics for fragments of natural languages, where value assignment is given relative to so-called indices, which comprise a temporal coordinate in order to provide for the interpretation of tense and temporal indicators words. (We note in passing that whereas there is a vast literature dealing with temporal logic, studies in the logic of locality are rare8.) Very often a structured set of temporal entities (mostly points of time) is assumed semantically in addition to a universe of discourse. This seems to be useful with respect to the interpretation of temporal indicators which Cf. Quine (1963) and Martin (1943). Cf. Rescher and Urquhart (1971). Cf., however, Rescher and Garson (1968) and v. Wright (1979).

382

W. Heydrich

function adverbially, but it may turn out to be insufficient if direct reference to temporal entities (say periods of time) is under discussion. Temporal entities must be within the universe of discourse as well and the structure which is assumed for the temporal coordinates within semantical indices must be reflected in the temporal entities within the universe of discourse. Generally, any semantic account which assumes a universe of discourse simply as a set without internal structure seems to be somewhat defective, at least as soon as the treatment of space and time is taken seriously. My account not only treats areas of space-time as values of object-language variables and provides for a special topological structure among areas of space-time, but also takes things as individuals with a specific internal spatiotemporal structure. Within things, substance and space-time are intimately linked and, in a sense, the whole topology of space-time is derivative. As will become clear subsequently in section 2., the spread of space-time in general depends on what things there are. Additionally, it is the combination of substance and space-time within things which allows to differentiate objects and events in our approach. A second peculiarity of my account lies in its divergence from pointontologies. Point-ontologies are quite common nowadays within treatments of space and time. However, there are some approaches which advocate an alternative orientation. Especially with respect to time, some philosophers and linguists have argued for interval semantics. There is some doubt whether extensionless instants, moments or points should be adopted as basic entities within the analysis of natural language. Needham (following Hamblin) has pointed out that e.g. activities can only occur during periods of time sufficiently long to encompass a minimum of the necessary movements involved. Saying that an activity takes place at an instant must involve some convention according to which the instant is a member of an interval during which the activity is accomplished. In the case of states we can say that they obtain at any subinterval of any interval during which they obtain. Thus, at least in simple and straightforward cases, it seems to be more natural to say that predicates of natural language hold relative to periods of time rather than relative to moments. Similar arguments have been given by Bennett and Partee, Cresswell, Taylor, and Dowty, who shows the usefulness of interval semantics for the interpretation of the progressive tense. Recently, van Benthem has given a parallel construction and comparison of interval-based and point-based temporal structures, and he has shown how temporal discourse can be interpreted with respect to both kinds of ontologies. Both Needham and van Benthem take mereology as the natural framework for their development of interval semantics, and van Benthem deals with the succession of time by means of a special order relation. Intervals may thus be compared with each other in two respects: they may be part of (discrete from, identical with or overlapping) and they may succeed each other. It should be noted, however, that although the interval perspective suggests

Things in Space and Time

383

the general framework of mereology, mereology is not committed to a treatment of time in terms of periods rather than points. It is possible to deal with points as individuals. Even if it is true that points are abstract entities and mereology is definitional for modern nominalism, it does not follow that points cannot be within the domain of a mereological theory of time. Nominalism is not at quarrel with abstract entities in general, but only with those entities which are no individuals. Points, however, may be conceived of as abstract individuals9. If, for instance, a special sort of atoms (in the mereological sense of the word) are introduced as moments of times, then certain sums of moments can be treated as periods, such that moments turn out to be periods of minimal extensions and both kinds of entities are treated as individuals on a par. Note that there might be sums of moments which are not periods according to this account, and since we think of periods as stretches of time without gaps, this seems correct. Time may be considered discrete or dense. If succession among moments is introduced as a dense ordering, each period will be composed of an infinity of moments. A conception roughly along these lines has been proposed by Martin10. My own proposals, however, are divergent. Firstly, I do not advocate an isolated treatment of time. Rather, my intention is to develop an integrated account of space and time. Secondly, I do not treat areas of space-time atomistically, but identify them with individuals without atomic parts. Individuals which are composed of atoms exclusively (they will be called 'heaps') are interpreted as substance pre-theoretically. Hence, I do not take the possible course of developing a mereological version of point-ontology. The advantage is conceptual economy. Note that the identification of some atoms as moments of time (and analogously the identification of certain atoms as points within space-time) necessitates the introduction of a special primitive predicate, whereas the difference between heaps (substance) and conglomerates (areas of space-time) can be drawn exclusively on mereological grounds. Within the literature the study of space is even more dominated by the point-perspective than the study of time. Obviously, thinking of space in terms of extensionless points is deeply rooted within the traditions of Western philosophy since the early days of the Greeks, especially since Euclid's Elements. The dominance of the point-view is also operative within the more recent study of structures of space-time such as Minkowski's four-dimensional geometry. Even the so-called causal theory of space-time (originating from the work of Robb and advocated recently by Winnie) which deals with the complex structure of space-time in terms of a primitive causal relationship among basic events conceives of events as point-like entities. The only 9

10

The possibility of abstract, concrete, universal, and particular individuals within nominalistic systems is discussed in Goodman (1951, cf. 19773: 177ff.). Cf. Martin (1965).

384

W. Heydrich

exceptions to this pervasive tendency known to me are Nicod's treatment of geometry in terms of volumes, Tarski's geometry of solids, and Whitehead's account of space and time in terms of extended events as well as the approach of proto-physics within the Erlanger Schule11. Within the approach to be presented in the following section, I propose to derive the topological structure of space-time (or to be more precise: a variety of different structures of space-time) from the relation of precedence which obtains among areas of space-time and things. Precedence is understood roughly as possible causal influence; areas and things are defined in terms of mereological relations. Thus, there are no extensionless instants, spatial points or points of space-time within the ontology of the system to be developed. Nevertheless, it can be shown that the mentioning of points is not excluded by the system. In section 3.4. I will introduce points as virtual classes of areas of space-time.

2. The System 2.1. Let me begin by characterizing a first-order language with two primitive predicate-constants: \ and «. The first is used to introduce the concept of mereological discreteness, the second stands for a certain spatio-temporal relation of precedence. I will use 'x', 4x0', 'xi', 'xa', ... as variables ranging over a universe of individuals, 'xi^xa' and 'xi « xa' are atomic formulae of my language. They may be read "xi is discrete from xz" and "xi precedes xa" respectively. Complex formulae are built up in the usual way by means of the logical constants ' ~ ' (negation), Ά' (conjunction), ' v ' (adjunction), '=>' (subjunction), ' =' (bi-subjunction), ' Λ ' (universal quantification), ' V ' (existential quantification), 'λ' (abstraction) will be used to construct n-place abstracts out of formulae (e.g. 'λχιχ2 (χ « χι Λ x0£x2)' is a two-place abstract), and abstracts are, in turn, used to construct formulae: we stipulate schematically that 'v xi ... xn' is a formula, if V is a η-place abstract. If 'v xi ... xn' is true, we say that the abstract 'v' applies to the individuals xi, ..., xn (in that order), Ύ can be used to build up definite descriptions, which are assumed to be defined contextually in Russell's way12. Discreteness is sufficient to introduce some central concepts of mereology; xi is a part of X2 (xi in bi such that xa > xt and y2>yi.

Figure 223: (B) Sbib2 precedes 20

21

22

23

If the straight line is identified with the class of real numbers together with its standard ordering, conglomerates can be conceived of as unions of classes of open intervals in model A. (Cf. note 14.) Broken lines are used to illustrate possibly interrupted periods, i.e. stretches of (space-)time with gaps. A plane might be identified with the Carthesian product of the class of real numbers with itself (IR2) and areas with regular open sets of elements of R2. (Cf. the metatheoretical definition of areas Μ 3 in section 2.9.) Dotting is used to indicate possibly perforated areas (areas with holes) whereas hatching will indicate areas without holes.

390

W. Heydrich

The interesting difference between model A and model B is that the former is linear whereas the latter is not. In a linear model, it holds that for each two conglomerates either one precedes the other or one both partially precedes and is partially preceded by the other (i.e. a part of the first precedes the second and the second precedes a part of the first). As will become clear below, it is non-linearity which allows us to differentiate between time-like and space-like regions of space-time. In linear models there are no space-like regions. Thus, conglomerates can be interpreted as areas of space-time only in some degenerate sense of the phrase in linear models: they should rather be considered as (possibly interrupted) periods of time. Despite the deficiencies of linear models, we do not exclude them from the start. Instead, the characterization of our primitive concepts of precedence and discreteness will be general enough to cover both kinds of models. Figure 3 illustrates the first two definitions of future and past within the models A and B.

Figure 3 a: (A) 3.2= Fut a ai = Past a

Figure 3b: (B) bi=Fut Sbb„ b2= Past Sbb

Note that £xi < xiOX2.

Broken lines are understood in a way such that they do not imply the exact position of gaps.

Things in Space and Time

393

2.4. With the concepts of future and past at our disposal, I can define my two relations of betweenness. The first relation is called p-betweenness, because it will be used to define what I shall call spatio-temporal positions: a conglomerate lies p-between two other conglomerates, if and only if it completely precedes their common future and is completely preceded by their common past. If we do not restrict the relation of p-betweenness to areas of space-time, but take into account things as well, my definition looks like this: D3

p-bet χ xiX2 =D Cx σι and ... anda n '> ση. A pair < Sn,P > is called a topology of space-time. It becomes clear that the space-time topologies of my models A and Β simply are > and > respectively. The models I am going to define presently are generalizations of A and Β in three respects. Firstly, I want to take into account space-time with arbitrary dimensionality; secondly, I use generalized relations of precedence among points instead of P>; thirdly, I will introduce atoms into my models which were not accounted for in the illustrative models A and B. The latter will be done by introducing a function which maps A into the set of those subsets of Sn which are identified with regular sets of points within my models. Before this function can be specified, some definitions must be given. I say that a point lies between two points if it precedes each point that is preceded by both of them and if it is preceded by every point that precedes each of the two: Ml 2 9

SBS1S2 = D AS3(S1PS3 Λ S2PS3 => sPss) Λ Λ S3(S3PS1 Λ S3PS2 ^SsPs).

The set of points between two points is called a position: M2

pOS X = D VsiS2As(sex = SBS1S2).

0 (the empty set) is a position, whereas unit sets of points are not positions. Unions of arbitrary positions are called areas. M3

area χ =D Λ s(sex => V y(pos y Λ sey Λ y^ x )).

0 is an area. An area is called connected if it cannot be separated into disjoint sub-areas: M4 con χ =D Λ yz(area y Λ area ζ Λ x = yuz r> ynz^0). 0 is not connected. These four definitions collect everything we need for a definition of regular sets of points. Such sets are connected areas which do not begin or end at several points within space-time: M5 29

regu χ =D con χ Λ ΛsiS2(si,S2ex Λ ~siPs2 Λ ~ {s I sBsiS2}^x).

The "M" indicates metatheoretical definitions.

406

W. Heydrich

The next and most important ingredient I need for the construction of my models is a function from Λ into {x | regu x}. I will use "F" to indicate such a function. Since the arguments of the function are the elements of A and its range depends on a topology of space-time < S",P > , a more explicit notation would be FA, < Sn,P > , but, for simplicity, I drop the subscripts. The task of F is to assign to each atom the maximal regular area of space-time it occupies. In order to be adequate within the construction of models, F has to respect the proviso — expressed in P 1 8 — that localities of maximal regular areas are discrete if they are assigned to different atoms. This amounts to saying that the following condition must hold: (iii)

x=/ 7 (ai) Λ y = F(&2) Λ xny^0 Λ VsiS2(si^S2 Λ Λ ~si/5S2 Λ ~

In linear models, which have < Sl,P> > as their topology of space-time, it holds that for each two elements of Sl one of them precedes the other. Hence the fourth conjunct of the antecedent is false and (iii) is vacuously true. In other models, where the topology of space-time is such that the first three conjuncts imply the fourth (like >), /''becomes one-one and there are no intersecting elements within its range. With the function F at my disposal, I define the basis of a model with respect to F. Such a basis will comprise just those sets as elements that correspond to units: Μ 6

BF = D {x| V yzw(yeA Λ regu ζ Λ regu w Λ F(y) = z Λ w ^ z Λ x = {y}uw)}.

Within each model there will be a range of things, a range of heaps and a range of conglomerates each defined with respect to F. The range of things is the set of unions of non-empty subsets of Bp: M7

TF = D { x | V y ( y ^ B F Λ y ^ 0 Λ x=IJy)}.

The range of heaps is the set of non-empty subsets of A.MS

HF = D ^A-0.

Within the next definition I use Tos x' to indicate the smallest position comprising x as a subset. The range of conglomerates is defined to be the set of all non-empty areas, which are subsets of positions of areas which are subsets of things. M9

CF =D {xjarea χ Λ χ^0 Λ Vyz(areay Λ zeTp A y i = z

y))35·

The universe of a model with respect to F is the union of the ranges of things, heaps and conglomerates: MIO

UF =D

Things in Space and Time

407

The remaining components of my definitions of models concern the interpretation of the two primitive constants. The relation of discreteness is interpreted as the set of pairs of elements of Up with an empty intersection:

Mil

R? =D {| x,yeUp Λ xny = 0}.

The relation of precedence is interpreted as the set of pairs of elements of TF or CF, such that (a) for each element s^sieP1) of the first member of such a pair there is an element 52 of its second member and (the point) si precedes (the point) S2 and (b) for each element 82(526^") of the second member of such a pair there is an element si of its first member and (the point) 52 is preceded by (the point) si:

Μ 12

R« =D {|x,yeTpUCF Λ Asi(siex Λ Λ SlPs2)) Λ

As2(S2ey Λ S2eJ'n =3 Vsi(si6x Λ S1/)S2))}.

Models are simply defined as special triples:

M13

M =D

The notion of an interpretation (with respect to a model M) of the language specified in section 2.1. arises straightforwardly. 3. Application and amplification In this section I will deal with some aspects of application and amplifications of the system sketched in section 2. My remarks mainly concern the introduction of possible things, objects and events as well as talk about points within space-time. 3.1. A characteristic feature of my system of heaps, conglomerates, and heterogeneous individuals lies in the fact that summation of atoms and conglomerates is restricted. This restriction has been introduced to provide for the intuition that each atom occupies some specific maximal area of spacetime, i.e. the greatest area where the atom actually exists. Possibly the atom might have existed somewhere else. Although my systems admits only those sums of an atom and a conglomerate which are part of a unit, we may consider arbitrary classes of atoms and conglomerates, each representing the atom at possible alternative areas of space-time. Where 'a' and 'c' indicate an atom and a conglomerate respectively, 'Sac' will either be an empty description or it will indicate a proto-unit. The abstract 'Xx(x = a v x = c)' (or '{a,c}' for short) however indicates a class (or virtual class) independently of whether Sac is a thing or not. It is tempting to accept such classes as a certain kind of possible things: possible proto-units and possible units (in the special case that the conglomerate c is regular). But in order to present a unified account of possible proto-units, units, and things, I prefer to introduce the term 'possible proto-unit' for unit-classes of classes with two elements (an atom

408

W. Heydrich

and a conglomerate). Note that my talk of possible proto-units does not presuppose possible individuals: the ultimate elements of possible proto-units (atoms and conglomerates) are actual. My account of the possible is actualistic. It does not postulate a new realm of individuals, but simply consists of a rearrangement of familiar ones, thus being in accordance with Goodman's (1955: 56f.) dictum: [...] if we restrict ourselves to predicates of actual things, we shall have enough equipment to say about the actual everything we need to say that ordinarily passes for talk about the possible. [...] discourse, even about possibles, need not transgress the boundaries of the actual world. [...] All possible worlds lie within the actual one.

As a matter of fact, the introduction of possible proto-units I have just been presenting resembles Goodman's treatment of certain possible things within phenomenalistic systems30. Once possible proto-units are available, the introduction of possible things does not present any fundamental difficulties: a possible thing is the union of a class of possible proto-units, where atoms within different elements of one and the same possible thing must be discrete and the conglomerates within different elements of the possible thing do not have localities which overlap. These conditions reflect the intuition that each atom within a possible thing occupies a certain maximal portion of the space-time which is occupied by the possible thing as a whole and that no spatial region can be shared by more than one atom at the same time. Possible units are minimal possible things (i.e. possible things with a single element). Their conglomerates have to be regular. In order to give formal definitions for this account of possible things, the language sketched in paragraph 2.1. has to be extended, since I identify possible things with certain classes of classes of individuals and these entities cannot be accounted for within a first-order language whose variables range over individuals exclusively. One possibility consists of adopting variables for classes and extending the universe of discourse, which would comprise individuals as well as arbitrary sets. Clearly, this procedure involves me with the platonistic ontology of set-theory. Alternatively, I can introduce new variables which may occur in place of η-place abstracts but are not considered to range over a new realm of abstract entities. Since η-place abstracts may be eliminated by means of the usual rules of conversion, I say that my new variables indicate virtual relations. Variables occupying the place of one-place abstracts are said to indicate virtual classes. I use V, 'v0', 'νΓ ... as variables for virtual relations. I thus incorporate the schematic letters, which were used to formulate some postulates in section 2., into the object-language. By means of variables for virtual relations we can construct second-order abstracts which indicate virtual relations of virtual relations of individuals and if we allow that new variables can also occupy 30

Cf. Goodman (1955: 49-57).

Things in Space and Time

409

the place of second-order abstracts, we may construct third-order abstracts, and so forth. In order to be completely explicit some indexing of variables seems to be convenient which allows us to identify the order as well as the adicity of variables. Such indices might be called type-indices. Variables with type-indices arrange the virtual relations within a hierarchy which corresponds to the simple theory of types. Since, however, adicity and the order of variables are very often recoverable from their context within a formula, I drop type-indices for simplicity31. Note that the mere notational device of variables for virtual relations does not involve us with abstract entities over and above individuals. The crucial point is, of course, quantification. But quantification bears ontological commitment only if interpreted referentially. Substitutional quantification is ontologically neutral. Thus I can allow for the quantification of variables in my extended language generally without the presupposition of abstract entities: quantification over individuals is interpreted referentially, whereas quantification over virtual relations is understood substitutionally32. A last point concerns identity among virtual relations. Up to now ' =' has been introduced exclusively as a two-place predicate of individuals. But an extended usage of the identity sign can easily be established: For all (substitutional reading) virtual relations vi and vz: vi is identical with V2 ('vi = V2*) iffo for all (substitutional or referential reading) virtual relations (of the appropriate type) or individuals V3...v n :vi V3...v n ^v2 V3...v n . Within the extended language33 we can define possible things and possible units. (Since the concepts to be defined will be used to introduce the notion of objects, I will not speak of possible things and possible units simpliciter but introduce the terms possible objectual thing and possible objectual unit. The reasons for this terminological manoeuver will become clear in the next section, where the distinction between objects and events will be discussed. For simplicity, however, I keep on talking about possible things and possible units where, more strictly speaking, I should use the terms 'possible objectual thing' and 'possible objectual unit'.) My formal definitions are these: D15

p-o-thing v =D Λν0(ν v 0 ^ VxiX2(atomxi Λ cong X2 Λ Vo = {xiX2})A Λ V1V2X1X2(W1 Λ V V2 ~ VI = V2 Λ V1X1 Λ V2 X2

Λ atom χι Λ atom X2=^ χι£χ2) Λ Λ V1V2X1X2(V VI Λ V V2 Λ

~ VI = V2 Λ VI XI Λ V2

X2 Λ cong χι Λ cong X2 => ~Loc xioLoc X2. 31

32

33

I keep on using "χ", "χ0", "χι", ... as variables for individuals, whereas "v", "v", "vi" are used as variables of arbitrary type. For discussion of the recent literature on substitutional quantification cf. Kripke (1975), Gottlieb (1980) and Heydrich (1982 a). A more detailled discussion of the extended language and what might be called the theory of virtual types can be found in Heydrich (1982a).

410

D16

W. Heydrich

p-o-unit v = D p-o-thing v Λ Λνιν2((ννι Λ v V2 => νι=ν2) Λ V χ (regu χ Λ vi χ))).

3.2. Strictly speaking, no possible thing is actual. Being a virtual class, it does not exist literally but is a fictitious entity34. In a derived sense, however, we may claim that the actual is among the possible: I say that a possible proto-unit is actual if the fusion of its element exists, and more generally, that a possible thing is actual if the fusions of all its elements exist. In this case the fusion of the union of its elements exists as well. (This follows from the principle of the unrestricted summation of units.) Notice, however, that the existence of the fusion of the union (of all elements) of a possible thing does not entail the existence of any fusion of one of its elements. Accordingly, I define actual possible things as follows: D1735

actual ν =D p-o-thing ν Λ Λν0(ν v0 ^ Fv0oFv0)

(Because of the reflexivity of the relation of overlapping 'Fv0oFv0' is true only in cases where there is an individual which is identical with Fv0). Let me introduce a relation of correspondence between possible things and things. A possible thing is said to correspond to a thing if the latter is identical with the fusion of the union of the elements of the former: D18

corr vix =D χ = Ρλχ0(νι ν0 Λ ν0 x0).

To each thing corresponds at least one possible thing, but some possible things do not correspond to anything. Some possible things correspond to a thing without being actual. But for each actual possible thing there is exactly one thing to which it corresponds: i.e. the fusion of the union of its elements. The converse, however, does not hold generally. To see this, consider an arbitrary thing. Each thing is identical with a fusion of proto-units with different atoms. It follows from Ρ19 that it is not the case that the localities of these proto-units overlap. Furthermore, if conglomerate-components of units are identical with their locality, it even follows that the proto-units are discrete and the thing is identical with the fusion of a unique virtual class of discrete proto-units. To each of these proto-units (say: 8x1x2), a unique possible proto-unit ({{xiX2}}) corresponds and the union of the virtual class of possible proto-units corresponding to the discrete proto-units is an actual possible thing. Hence there is a unique actual possible thing which corresponds to our arbitrary thing. This argument, however, depends on the assumption that the conglomerate-component of a unit is identical with its locality and we have seen that there are models for which this assumption is wrong. E.g. in linear models, there are no localities, different atoms may

34 35

See my discussion of fictitious entities in Heydrich (1982b). This definition will be modified below.

Things in Space and Time

411

occupy areas which overlap, and things cannot always be portioned uniquely into discrete proto-units with different atoms. Consequently, there may be things to which no unique possible thing corresponds. Whenever different units overlap, there will be more than one possible thing which corresponds to their fusion. Thus, actual possible things might be said to have more structure than things. For convenience, I will call them objects. D19

object v = D actual ν Λ p-o-thing ν

Since objects and the other possible things are virtual classes and not individuals, the mereological relations (discreteness, part and overlapping) are undefined with respect to them. Derived notions, however, may easily be introduced. E.g. I will say generally that a virtual class vi is part of a virtual class V2 (vi < va) iffo each element of vi is part of some element of va. It follows that a possible thing vi is part of a possible thing va if and only if for each element {xixa} of vi there is an element {x3X4} of vs, such that xi is part of X3 and X2 is part of X4. If an object is part of another object, the thing corresponding to the former will be part of the thing corresponding to the latter. 3.3. Let us see whether events can be introduced into my system as an additional class of entities. What is the difference between objects and events? A first idea is that events are composed out of those objects which are involved in the event. But this feature does not make them different from objects because the latter may comprise objects as parts and thus may be composed of several objects as well. A second idea is that the objects which are involved in an event may occupy scattered areas of space-time and that events tie together scattered objects. But, again, this feature does not provide for a difference between objects and events, since arbitrary scattered objects add up to nothing but objects. Perhaps events and objects are not that different and event-talk is not really about things which are different from those in object-talk, but merely provides for an altered perspective on things. Davidson once proposed as an identity criterion for events that these entities are identical if and only if they have exactly the same causes and effects36. This gives us the hint that it might be relevant for the definition of events to take into account positions of things within space-time. Remember that the position of a thing is that area of space-time which includes the position of every thing that both causally influences and is causally influenced by the thing. The area of space-time where all the effects of a thing are situated is its future, the area where all it causes reside is its past. Having identical future and past entails having identical position. Thus, positions of

36

Cf. Davidson (1969).

412

W. Heydrich

things can be considered as the feature which determines the potential positions of their causes and effects. Different things with identical position have what may be called the same causal potentiality. If the sum of a unit and its position exists, this might be said to be a unitconsidered-as-an-event. But since the existence of such sums is not guaranteed in general37, I prefer to introduce event-units in analogy to objectual units as virtual unit-classes of virtual classes with two elements: an atom and the position of a unit which has the atom as its heap-component. Possible eventunits are virtual unit-classes of virtual classes with an atom and an arbitrary position as elements. Whereas a possible objectual unit represents an atom together with some regular area of space-time which might be occupied by the atom, possible event-units represent atoms together with some area of space-time which might determine its causal potentiality. A possible eventunit {{xix2}} is actual, if it is an event-unit, i.e. if there is some regular conglomerate x, such that PosSxix = X2. The construction of possible eventthings out of possible event-units is analogous to the construction of possible things out of possible proto-units. Possible event-things are unions of virtual classes of possible event-units, where the atoms within different elements of the possible event-thing must be discrete. The localities of the conglomerates within different elements of a possible event-thing may overlap however. These conditions reflect the intuition, that each atom within a possible eventthing has its specific causal potentiality, whereas it is not excluded that one and the same causal potentiality can be shared by several atoms involved in possible event- things. The formal definitions are as follows: D 20

p-e-thing v =D Λν0(ν v0 => V x xiX2(atom χι Λ x2 = Pos χ Λ Λ VlV2XlX2(v VI Λ V V2 ~ V 1 = V 2 Λ VI Xl Λ V2 X2

D21

p-e-unit ν

Λ atom χι Λ atom X2 => xi?x2). = op-e-unitv Λ Λνιν2(ννι Λ ν V2 =3 vi=V2).

In order to apply our concept of actuality to possible objectual things as well as to possible event-things my former definition needs extension: D1 7

actual v = D (p-o-thing ν Λ Λ ν0(ν ν0 => Fv0oFv0)) v (p-e-thing ν Λ Λ χΐΧ2(ν {χΐΧ2} Λ cong X2 => V χ x2 = PosSxxi).

Using the extended concept of actuality, events can be defined on analogy with objects. Whereas the latter are actual possible objectual things, the former are introduced as actual possible event-things: D 18

37

event v =D actual ν Λ p-e-thing ν

This is because I allow for empty space-time, cf. section 2.8.

Things in Space and Time

413

Objects and events are very similar in structure. They have been construed as virtual classes of virtual classes with two individuals as members: an atom and a conglomerate. Although it is not the case generally that every object is an event or that every event is an object, each object or event comprises objects as well as events as parts. In some models there is even no difference between objects and events at all, for instance, in linear models, where no localities exist, the causal potentiality of a thing is determined by its temporality, since the regular conglomerate, which is the conglomerate-component of a unit is identical with the unit's position. With respect to these models it does not make much sense to speak about causal potentialities because the position of a thing is its temporality here and several atoms may occupy identical areas of space-time. Having introduced objects and events, I have accomplished the task outlined in section 1. of explicating the internal structure of some of the basic entities for a theory of reference. The mereological system of space and time (or space-time), outlined in section 2, has played a fundamental role within this explication. 3.4. The mereological treatment of space-time takes areas as its fundamental entities. Areas are construed as individuals without atomic parts. This account diverges from a more common line of thinking which considers points as fundamental. Undoubtedly, dealing with space-time in terms of points has turned out to be an extremely fruitful approach. By joining a very limited number of additional primitives, we may build up axiomatic system which lay the foundation of physical theories. One example is Carnap's S-system which gets along with a single primitive predicate standing for a signal relation among so-called world-points and which axiomatizes the concepts of space and time within Einstein's general theory of relativity38. A central defined notion within this system introduces the concept of signal lines which pass through the entire space-time. By means of signal lines it is possible to define certain sets of world-points as spaces upon which some special topological structure (a so-called Hausdorff system of neighbourhoods) is imposed axiomatically. An additional axiom stipulates that each space has the homogeneous dimension number three. At first sight it might be doubted whether structures like those underlying Carnap's S-system can be defined within area-based mereological space-time topologies at all, since the fundamental entitities of world-points and signal lines are not directly available here. But, of course, this does not mean that I cannot introduce world-points and signal lines definitionally into my mereological space-time topologies. Compare the analogue situation within systems as based on point-ontologies. Here we have no direct access to areas 38

Cf. Carnap (1958: 209f.). He refers to Robb's work on the causal theory of space and time.

414

W. Heydrich

which are fundamental in the mereological system. Nevertheless, areas can be defined as certain classes of world-points. In order to develop mereological analogues to systems of world-points this definitional order has to be reversed: we have direct access to areas and may try to define world-points as certain (virtual) classes of areas. If, additionally, we provide for an independent definition of signal lines, it seems to be possible to formulate a mereological version of Carnap's S-system. The task of the signal relation among worldpoints will be taken over by the relation of precedence among areas in such a system, such that no special primitive relation among world-points is needed. I shall not try to formulate a mereological version of Carnap's S-system in this paper. Instead I will restrict myself to the first steps of the development of such a version, viz. the introduction of world-points and signal lines into the system thus far developed. My constructions make use of the method of abstractive classes which has been developed and applied by Russell, Whitehead, Nicod and Tarski39. I will use the term 'abstractive class' in such a way that it indicates a virtual class of individuals of some specifiable sort. Abstractive classes are construed as virtual classes with respect to other virtual classes of individuals. A virtual class of individuals vi is an abstractive class with respect to V2 or an abstractive class of V2-s (a-class viva) iffb the conditions a) — c) obtain: a) Each element of vi is an b) Each two elements of vi c) No individual is part of not have a common part

element of V2. are such that one is part of the other. every element of vi, i.e. the elements of vi do or — equivalently — there is no nucleus of vi.

Abstractive classes are series of smaller and smaller individuals without a smallest element. (Hence, there are not abstractive classes of heaps.) The limit to which such a series converges is not an individual, it will be called an abstractive entity. Abstractive entities cannot be defined as abstractive classes since several such classes may converge to the same abstractive entity. Additionally, I note that abstractive classes of individuals of one and the same sort may converge to abstractive entities of different kinds. Consider as an example abstractive classes of positions. In linear models which are based on a topology of space-time > abstractive classes of positions converge to points. In models, which have > as their topology of space-time, they converge to points or stretches (horizontal or vertical). In models, which have >, they converge to points, certain stretches, or certain surfaces and so forth. Thus, points, stretches, surfaces, etc. may be abstracted from positions. Since stretches can be considered as

39

Cf. Russell (1914), Whitehead (1919, 1920), Nicod (1930) and Tarski (1927).

Things in Space and Time

415

being composed of points, surfaces of stretches, etc., we may think of points as minimal abstractive entities with respect to positions. In order to isolate abstractive classes which converge to minimal abstractive entities in all models, I define the relation of covering: A virtual class vi is said to cover another virtual class va (cov viva) iffn each element of vi has an element of va as a part. (Note that a virtual class can be covered by another one without being a part of it, and that a virtual class can be a part of without being covered by another virtual class.) Abstractive classes which cover each other converge to the same abstractive entity. Abstractive classes, which are covered by each abstractive class they cover, converge to minimal abstractive entities. I will call them nestings: D19

nest viva =D a-class viva Λ Av(a-class vva Λ cov viv => cov vvi).

Nestings of positions converge to points in all models. However, several nestings of positions may converge to the same point. There might even be nestings of positions which converge to the same point although no element of one of these nestings overlaps an element of any of the other ones. Figure 12 provides an example within the previously given illustrative models A and B.

Figure 12a: (A)

Figure 12b: (B)

This situation can be excluded if we turn our attention to comprehensive nestings. Generally, a virtual class vi is defined to be comprehensive with respect to a virtual class V2 (comp viva), iffo the fusion of va is an element of vi. Obviously, if vi and va are comprehensive nestings with respect to v, they will have a common element. Comprehensive nestings of positions have the maximal position (i.e. the maximal conglomerate) among their elements. If vi is a comprehensive nesting of v-s and va is an abstractive class of v-s, V2 will be part of vi.

416

W. Heydrich

Let us now consider, for some given comprehensive nesting vi of V2-s, the virtual class of all nestings v0 of V2-s such that v0 covers vi. Since v0 is a nesting, it will be covered by vi. Accordingly, all elements of our virtual class converge to the same abstractive entity. Moreover, the abstractive entity to which all elements converge will be minimal with respect to vz and if some abstractive class converges to this abstractive entity, it will be a nesting which covers vi. Consequently, it will be one of the elements of our virtual class. The union of a virtual class of this kind may thus conveniently be identified with an abstractive entity relative to V2. D 20

a-entity viva =D Vv(nestW2 Λ comp W2 Λ Λ χ(νι χ = V v0(nest v0V2 Λ cov v0v Λ ν0χ)).

The introduction of world-points and signal lines is easy after these preliminaries. World-points are simply abstractive entities with respect to positions. D21

point vi = D a-entity νιλχ χ = Pos x.

Signal lines are abstracted from another sort of individuals, viz. regular conglomerates which neither precede nor are preceded by anything which is discrete from them. Such conglomerates pass though the entirety of spacetime, they are characterised by the condition that their temporality is identical with the maximal conglomerate. D 22

s-line vi =D a-entity viXx(cong χ Λ regu χ Λ Tem χ =Ρλχ cong x).

These definitions accomplish the discussion of how to provide for mereological versions of axiomatized theories which are based on point-ontologies (like Carnap's S-system). In the context of this paper, my discussion shows that the system presented in section 2. does not prevent talk of points within space and time, although it is not committed to these entities ontologically.

Literature Bennett, Michael and Partee, Barbara H. 1972 Towards the Logic of Tense and Aspect in English. Santa Monica, Calif.: System Development Corporation, van Benthem, Johan F. A. K. 1983 The Logic of Time: A Model-Theoretic Investigation into the Varieties of Temporal Ontology and Temporal Discourse. Dordrecht: Reidel. Carnap, Rudolf 1958 Introduction to Symbolic Logic and its Applications. New York: Dover Publ. 1972 Bedeutung und Notwendigkeit: Eine Studie %ur Semantik und modalen Logik. Wien: Springer. Cress well, Max J. 1977 "Interval Semantics and Logical Words", in Ch. Rohrer (ed.): On the Logical Analysis of Tense and Aspect. T bingen: Narr, 7—29.

Things in Space and Time

417

Davidson, Donald 1969 "The Individuation of Events", in N. Rescher (ed.): Essays in Honor of Carl G. Hempel. Dordrecht: Reidel, 216—234. Dowty, David R. 1979 Word Meaning and Montague Grammar: The Semantics of Verbs and Times in Generative Semantics and in Montague's PTQ. Dordrecht: Reidel. Eberle, Rolf A. 1970 Nominalistic Systems. Dordrecht: Reidel. 1978a "Goodman on Likeness and Differences of Meaning", Erkenntnis 12, 3 — 16. 1978b "Semantic Analysis without Reference to Abstract Entities", The Monist 61, 363 — 38. Goodman, Nelson 1949 "On Likeness of Meaning", Analysis 10, 1-7. 1951 The Structure of Appearance. 19773. Dordrecht: Reidel. 1953 "On some Differences about Meaning", Analysis 13, 90 — 96. 1955 Fact, Fiction, and Forecast. 19774. Indianapolis: Bobbs-Merrill. 1978 Ways of Worldmaking. Brighton, Sussex: Harvester. Gottlieb, Dale 1980 Ontological Economy: SubstitutionalQuantification and Mathematics. Oxford: Clarendon. Hamblin, Charles L. 1969 "Starting and Stopping", The Monist 53, 410-25. 1971 "Instants and Intervals", Studium Generale 24, 127-34. Heydrich, Wolfgang 19 82 a Gegenstand und Sachverhalt: Bausteine %tt einer nominalistisch orientierten Semantik für Texte, Hamburg: Buske. 1982b "Syntactic Fiction", Poetics 11, 285-309. 1983 "Models and Realities", Text 3 (1), 99-130. Heydrich, Wolfgang; Neubauer, Fritz and Petöfi, Jänos S. (forthcoming) Systemic, Text Specific, and Interpretive Knowledge: Aspects of an Operational Knowledge Representation. Heydrich, Wolfgang and Petöfi, Jänos S. 1979 "Pragmatic Considerations within a Text-Theoretical Framework", in H. Parrett, M. Sbisa and J. Verschueren (eds.) (1981): Possibilities and Limitations of Pragmatics. Amsterdam: Benjamins, 313 — 330. Kripke, Saul A. 1976 "Is there a Problem about Substitutional Quantification?", in G. Evans and J. McDowell (eds.): Truth and Meaning: Essays in Semantics. Oxford: Clarendon Press, 325-419. Leonard, Henry S. and Goodman, Nelson 1940 "The Calculus of Individuals and its Uses", Journal of Symbolic Logic 5, 45 — 55. Lesniewski, Stanislaw 1927-1930 " podstawach matematyki, Pnyglad Filo^ofic^ny 30, 164-206, 261 -291; 32, 60-101; 33, 77-105; 34, 142-170. Lewis, David 1983 Philosophical Papers. Vol. I. Oxford: Oxford University Press. Lewis, David and Lewis, Stephanie 1970 "Holes", Australasian Journal of Philosophy 48, 206-212. Lorenzen, Paul 1961 "Das Begründungsproblem der Geometrie als Wissenschaft der räumlichen Ordnung", Philosophia Naturalis 6, 415 — 431. Lorenzen, Paul and Schwemmer, Oswald 1973 Konstruktive Logik, Ethik und Wissenschaftstheorie. Mannheim: Bibliographisches Institut. Luschei, Eugene C. 1962 The Logical Systems of Lesniewski. Amsterdam: North-Holland.

418

W. Heydrich

Martin, Richard M. 1943 "A Homogeneous System for Formal Logic", Journal of Symbolic Logic 14, 1—23. 1965 "On Time and the Null Individual", Journal of Philosophy 62, 377-87. 1969 Belief, Existence, and Meaning, New York: University Press. 1979 Pragmatics, Truth, and Language. Dordrecht: Reidel. Needham, Paul 1981 "Temporal Intervals and Temporal Order", Logique et Analyse 24, 49 — 64. Nicod, Jean 1930 Foundations of Geometry and Induction. London. New Translation 1969: Geometry and Induction. London: Routledge & Kegan Paul. Petöfi, Janos S. 1982 "Text, Signification, Models, and Correlates", in G. Rickheit and M. Bock (eds.): Psycholinguistic Studies in Language Processing. Berlin: de Gruyter, 266 — 298. Quine, Willard V. O. 1963 Set Theory and its Logic. Cambridge, Mass.: Harvard University Press. Rescher, Nicholas 1975 A Theory of Possibility: A Constructivistic and Conceptualistic Account of Possible Individuals and Possible Worlds. Oxford: Blackwell. Rescher, Nicholas and Garson, James 1968 "Topological Logic", Journal of Symbolic Logic 33, 537 - 548. Rescher, Nicholas and Urquhart, Alasdair 1971 Temporal Logic. Wien: Springer. Robb, Alfred A. 1914 A Theory of Time and Space. Cambridge: CUP. Russell, Bertrand 1905 "On Denoting", Mind 14, 479-93. 1914 Our Knowledge of the External World. London. 19695. Our Knowledge of the External World as a Field for Scientific Method in Philosophy. London: Allen & Unwin. Tarski, Alfred 1927 "Foundations of the Geometry of Solids", in A. Tarski (1956) Logik, Semantics, Metamathematics. Oxford: Clarendon, 24—29. Taylor, Barry 1977 "Tense and Continuity", Linguistics and Philosophy 1, 199—220. Whitehead, Alfred N. and Russell, Bertrand 1927 Principia Mathematica. Vol. 1. Cambridge: University Press. Winnie, John A. 1977 "The Causal Theory of Space-Time", in J. S. Earman, C. N. Glymour and J. J. Stachel (eds.): Foundations of Space-Time Theories. Minneapolis: University of Minnesota Press, 134-205. v. Wright, Georg H. 1979 "A Modal Logic of Place", in E. Sosa (ed.): The Philosophy of Nicholas Rescher. Dordrecht: Reidel, 65-73.

Part IV

LUCIANO VITACOLONNA 'Text'/'Discourse' Definitions For M. M.

0. This paper is just part of larger survey of the terms 'text' and 'discourse'; therefore I have had to make some choices and restrictions inside the mare magnum of definitions. The first restriction has been concerned with the chronological limits, as well as the use of 'text' in philology and of 'discourse' in rhetoric1. Accordingly, except for some cases, I have confined myself to an investigation taking into account only (some of) those definitions which text linguistics and text theory have put forth in the last years. (For a historical and thematic overview on textological research in Europe, see Petöfi [1985], where a bibliographical survey can be found). All that means is that I have chosen a linguistic perspective which leaves out almost entirely other 'cultural acts' (ballets, music, rituals, etc.) to which the term 'text' has been applied as well (see below § 10, and cf. Harweg, 1974: 90-92). Several ways of proceeding would be open: I might follow a chronological order or examine the terms in question with reference to groups of scholars or to research trends by simply (and taxonomically) classifying these groups, for instance as follows: (1) those who use the only term 'text'; (2) those who use only the term 'discourse'; (3) those who use both 'text' and 'discourse' either as synonyms or as different theoretical notions. It would also be possible to catalogue the definitions on the basis of particular theoretical frameworks or methodological approaches. What follows is, in a sense, a combination of these and other kinds of arrangements. I just hope the treatment will be, if not complete, at least clear. A short history of the term 'text' (textus) is in Segre (1981), who also accounts for its use in philology; for 'discourse' see s.v. discursus in Thesaurus Linguae Latinae, vol. V, Leipzig, Teubner, 1909-1934, and Segre (1979). Cf. also Brinker (1973), Plett (1975), Abad Nebot (1980); Langleben (1981: 281 ff.). A survey like this, moreover, should take into account: (1) the textual conceptions that can be found in the narrative analysis of both the Russian formalists and the French structuralists (see e.g. Todorov, ed., 1965; Chabrol, ed., 1973); (2) the textual characterizations with regard to the use of tenses (cf. Benveniste, 1959; Weinrich, 1971); and (3) the relations between stylistics and text linguistics (see Enkvist, 1978). I do hope to treat these topics adequately in the recreation of this article.

422

L. Vitacolonna

l. In the middle of the fifties, Coseriu already acknowledged the importance, rather the existence, of text linguistics, "una lingüistica del texto" (Coseriu, 1967: 289). He also employed the expression 'universe of discourse', but not in the same way logic uses it: in Coseriu's terminology universes of discourse are, for example, literature, mythology, mathematics, etc. (Coseriu, 1967: 318). If, however, we want to look for more "concrete" reference points, we have to turn to Hjelmslev and Harris. Even though these two linguists cannot be considered to be the founders of text linguistics, yet their definitions of 'text' and 'discourse' deserve attention, since it is from the criticism of these definitions that some of the early text-linguistic papers started (e.g. Bierwisch, 1965; Harweg, 1968). Hjelmslev (1961: 16) claims that the "objects of interest to linguistic theory are texts". However, he does not exactly define what a text is. In fact, after introducing the notions of 'process' and 'system', he confines himself to stating that "the process can here be called a text, and the system a language" (1961: 39). But since a process "is unimaginable [...] without a system lying behind it" and "a system is not unimaginable without a process", it is impossible "to have a text without a language lying behind it" (ibid.). Every natural language is, then, both process and system (cf. Conte 1977: 24): "In a certain sense it is said to be the same entities that enter into the linguistic process (text) and into the linguistic system" (Hjelmslev, 1961: 37). In other words: "A language may be defined as a paradigmatic whose paradigms are manifested by all purports, and a text, correspondingly, as a syntagmatic whose chains, if expanded indefinitely, are manifested by all purports" (1961: 109). What interests Hjelmslev is not to explain what a text is, but to describe the system lying behind a text (process). Therefore, to use Conte's (1977: 25) words, Hjelmslev's theory is not a theory of the text, but a theory through the text. As regards the other reference point, for example in Harris (1964), it is worth noting that he uses both 'text' and 'discourse', even if a clear-cut distinction is not drawn. According to Harris, the first problem from which it is possible to approach discourse analysis "is the problem of continuing descriptive linguistics beyond the limits of a single sentence at a time" (1964: 356). This depends on the fact that language "does not occur in stray words or sentences, but in connected discourse — from a one-word utterance to a ten-volume work, from a monolog to a Union Square argument" (1964: 357). Hence the necessity of discourse analysis, which — among other things — "yields information about stretches of speech longer than one sentence; thus it turns out that while there are relations among successive sentences, these are not visible in sentence structure [...], but in the pattern of occurrence of equivalence classes through successive sentences" (1964: 383). It is not expedient to stop here to criticize Harris's views once again (see Bierwisch, 1965; Conte, 1977: 25—27). Suffice it to say that Harris does not give us a clear

'Text'/'Discourse' Definitions

423

definition of text or discourse; anyway, given that, according to Harris (1964: 359), a discourse can be "a ten-volume work" and that a book is a "succession of sentences" or "a closed string of sentences", it follows that discourse may be seen as a closed string or succession of sentences. That this definition is not sufficient to characterize or to constitute a discourse or text has been shown by several text theoreticians; so it is not necessary to come back to this matter. 2. It was only in the middle of the sixties that the first studies in text linguistics were undertaken. In 1964 Hartmann wrote his famous paper "Text, Texte, Klassen von Texten" (see Hartmann, 1972), which is considered to be the very beginning of text linguistics. Hartmann started from the following definition: "Text [...] ist dann Sprache in dem Zustand und in der Form, wo sie Äußerung, Mitteilung und damit etwas Verstehbares ist bzw. ist geworden" (1972: 3). After introducing the concept of 'Sprache in Textform' ("eine Zustandsform von Sprache als wahrnehmbar gemachter oder gewordener Sprache", ibid.), he came to the following definition: "Mit 'Text' kann man alles bezeichnen, was an Sprache so vorkommt, daß es Sprache in kommunikativer oder wie immer sozialer, d. h. partnerbezogener Form ist; oder kürzer: Sprache kommt beobachtbar vor in Textform; und: Noch niemals ist Sprache in anderer Form als in Textform vorgekommen, d.h. in Sprachfunktion geäußert worden" (1972: 5)· Hartmann's paper, by constituting the transition between sentence linguistics and text linguistics, gave rise, of course, to a flowering of definitions of both text and discourse. One of the first definitions was provided by Koch (1965: 16), who drew the following distinction: "Any sequence of sentences temporally or spacially arranged in a way to suggest a whole will be considered to be a text"; on the other hand: "Any text (or parts of a text) having manifestations of a particular theme in common will be considered to be a discourse". (Cf. Koch, 1966). In his turn, Grimes, in regretting that the relationships "between sentences of a discourse have received only sketchy treatment" (1966: 465), writes: "Sentences in texts are obviously interrelated, not only in selection of lexical items and lexical strings, but also in the constraints imposed on pronominal usage, tense and aspect sequences, and other features of grammar" (ibid.). As can be seen, Grimes on one hand gives a glimpse of what will be called the 'connexity property', and, on the other, anticipates Harweg's (1968) views. To understand Harweg's (1968) concept of text, it is necessary to recall, first of all, his notion of 'pro-nomina' and 'pro-form(s)': for Harweg 'proform' stands for "expressions of identical or different morphological form designating the same object and replacing each other in texts under certain conditions" (Rieser, 1978: 7). Then Harweg, starting from a critical view of Hjelmslev's (1961) dichotomy 'system'/'text', considers the members of this

424

L. Vitacolonna

dichotomy to be "verschiedene Existenzformen einer Sprache" (Harweg 1968: 139), and claims: "Die kürzesten und eindeutigsten Bestimmungen der beiden genannten Existenzformen sind die, die den Text mit Hilfe des Begriffes des Nacheinander, das System mit Hilfe des Begriffes des Statteinander definieren: Der Text wird, nach diesen Bestimmungen, durch Ausdrücke gebildet, die nacheinander stehen, das System durch Ausdrücke, die, unter bestimmten Bedingungen, Statteinander stehen können" (ibid.). A/the text is, thus, a 'class' "deren Elemente in bestimmter Weise nacheinander angeordnet sind" (1968: 141). On the basis of these remarks and of his concept of'pro-form', Harweg (1968: 148) comes to define the text as "ein durch ununterbrochene pronominale Verkettung konstituiertes Nacheinander sprachlicher Einheiten", and then he adds: "Äquivalent mit dem soeben gegebenen Definiens von 'Text' ist das Definiens 'ein durch eine ununterbrochene Kette zweidimensionaler Substitutionen gebildetes Nacheinander sprachlicher Einheiten'". Undoubtedly Harweg's contribution to text analysis is substantial. However, his ideas "do not allow an intuitively adequate decision on whether a particular utterance can be regarded as a text or not, nor do they explain what properties a piece of natural language must have to constitute a text" (Rieser 1978: 8). 3. If till the end of the sixties the studies in text linguistics and discourse analysis — and, accordingly, the definitions of text or discourse — had been pursued in a linguistic framework 2 , at the beginning of the seventies Austin's and Searle's theories of speech acts began to have an influence upon text linguistic research and to turn attention towards the communicative context^). As instances of this new trend one can look at Isenberg (1970) and Schmidt (1971). To tell the truth, Schmidt (1971) does not quote either Austin or Searle, but refers to many philosophers, and especially to L. Wittgenstein. Schmidt (1971: 33) expressly writes: "Sprache wird [...] nicht mehr (wie früher häufig und auch heute noch mancherorts) primär gesehen als denotierendes Zeichensystem, sondern als Handlungs- bzw. Verfahrenssystem". On this basis he argues that "Sprache als parole oder speech, phänomenal gegeben in Texten, wird vollzogen: (a) als systematische Komplexion von Elementen nach Regeln; (b) als sozial erfolgreiches Sprachspiel in Geschichten" (1971: 38). This leads Schmidt to the following definition of 'text': "Der Text [...] ist aufzufassen als die phänomenologisch primär gegebene Vorkommensweise von Sprache. Selegierbare Textelemente vom Wort bis zum Phonem kommen vor und sind in ihrem Funktionswert fundiert im mehrheitlichen Verhalt des Textes" (1971: 47). 2

This is not completely true, since Oomen (1969: 19) had already argued that (1) "Ohne kommunikative Funktion ergibt sich kein Text", and (2) "Die jeweilige Kommunikationsfunktion steuert den spezifischen Ablauf des Textprozesses".

'Text'/'Discourse' Definitions

425

These ideas have been developed in Schmidt (1973a, b). The basic concept of Schmidt's text theory is 'das kommunikative Handlungsspiel' (Wittgenstein's notion of 'Sprachspiel' is very evident and had already been recalled in Schmidt 1971: 38). Schmidt starts from this statement: "Sprechen ist stets partnerbezogenes, intentionales und informatives (= potentiell bedeutungsvolles) Handeln, das in Form von 'Texten' vollzogen wird" (1973a: 234). So, text is "ein Phänomen, das zwei Komponenten integriert: eine geordnete Menge sprachlicher Konstituenten und ein sozial erwartbares und erwartetes Funktions- und Wirkungspotential (Illokutionspotential und Perlokutionspotential)" (1973a: 236): In other words: 'text' "benennt kein nur sprachliches Objekt, sondern eine kommunikative Rahmen-Funktion für sprachliche Elemente und Strategien, die — wie jede Funktion — zu ihrem Zustandekommen ein Code-System [...] braucht, dessen Existenz aber erst von der Funktion her gerechtfertigt werden kann. Textualität als Funktion aber kann nur im Kontext sozialer Interaktionszusammenhänge beschrieben werden" (1973b: 146). In order to understand a text, "muß ein Kommunikationspartner also nicht nur die Textzeichenmenge kennen, sondern er muß auch die im Text aktivierte Handlungsgrammatik/Handlungssemantik kennen, die die Relevanz des Textes als Realisat eines strukturell präformierten Kommunikationstypes definiert" (Schmidt 1973b: 149). Thus, Schmidt comes to the following definition of 'text': (1) Ein Text ist jeder geäußerte sprachliche Bestandteil eines Kommunikationsaktes in einem kommunikativen Handlungsspiel, der thematisch orientiert ist und eine erkennbare kommunikative Funktion erfüllt, d. h. ein erkennbares Illokutionspotential realisiert. Nur durch die von einem Sprecher beabsichtigte und von Kommunikationspartnern erkennbare, in einer Kommunikationssituation realisierte illokutive (sozio-kommunikative) Funktion wird eine Menge sprachlicher Äußerungen zu einem kohärenten sozio-kommunikativ erfolgreich funktionierenden, durch konstitutive Regeln geregelten Textprozeß (= einer Manifestation von Textualität). (2) Werden in einem Kommunikationsakt mittels verschiedener Äußerungsmengen verschiedene unterscheidbare Illokutionsakte realisiert, und lassen sich diese Illokutionsakte hierarchisch in ein kohärentes System einordnen, dann gilt die gesamte Äußerungsmenge, die die Illokutionshierarchie vollzieht, als Text; die Äußerungsmengen, die unterscheidbare integrierte Illokutionsakte vollziehen, heissen Intexte. Dabei gilt, daß Texte Sprechern zugeordnet werden. Das hat zur Folge, daß auch solche Äußerungsmengen, die von Äußerungen der Kommunikationspartner unterbrochen werden, aber vom Sprecher als zu einem Illokutionsakt gehörig angesehen werden, als ein einheitlicher Text gelten (Schmidt, 1973b: 151; cf. 1973a: 237).

4. In the meantime, but from a different theoretical viewpoint (i.e. functionalism), Halliday (1973) provided a definition of text (linked to the use of discourse) on the basis of the distinction of three functions of language: 'ideational', 'interpersonal', and 'textual' functions. The 'textual' function is concerned with the creation of text. [...] It is through this function that language makes links with itself and with the situation; and discourse becomes possible, because the speaker or writer can produce a text and the listener or reader can recognize one. A text is an

426

L. Vitacolonna

operational unit of language, as a sentence is a syntactic unit; it may be spoken or written, long or short; and it includes as a special instance a literary text, whether haiku or Homeric epic. It is the text and not some super-sentence that is the relevant unit for stylistic studies; this is a functional-semantic concept and ist not definable by size. And therefore the 'textual' function is not limited to the establishment of relations between sentences; it is concerned just as much with the internal organization of the sentence, with its meaning as a message both in itself and in relation to the context (Halliday, 1973: 107).

Hasan (1978) also starts from a functional approach. After introducing the term 'texture' and explaining what a 'structure' is, Hasan (1978: 229) states: The elements of text structure cannot be defined by reference to the rank status or sequential ordering of the lexicogrammatical units which have the function of realizing these elements. To be all viable, the definition will have to be functional [...], the functions themselves being determined by the semiotics of the text genre [...]. It seems quite irrefutable that the controls upon the structural make-up of a text are not linguistic in origin, in that language as a formal system does not enable one to predict what generalized structural formula could be associated with which genre. Instead, the control is contextual: the nearest non-linguistic analogue of a text is not a logico-mathematical formula, but a non-verbal social event. A text is a social event whose primary mode of unfolding is linguistic. If text can be seen as a bridge between the verbal symbolic system and the culture, this is because of the relationship between text and social context: text is 'in language' as well as 'in culture'. It follows from this relationship between text and context, that the specification of structural formulae for distinct genres requires a model of language in which context is a well-defined category, not just an ad hoc stand-by to which appeal can be made in analysing sentences which might otherwise prove recalcitrant.

5. In 1974 Dascal and Margalit launched their famous attack upon text linguistics (Dascal and Margalit, 1974). In substance it was a critical review of van Dijk (1972), who had argued that "discourses are the only justifiable 'natural domain' of an empirically adequate theory of language" (van Dijk, 1972: 7), and that it is useful to distinguish between 'texts' (abstract linguistic entities) and 'discourses' (observational entities). Dascal and Margalit's (1974: 213) drastic conclusion that the text-grammar research program "has not produced a convincing 'refutation' of S [en tence]-grammar, nor presented a viable alternative to it", would have been belied soon. In fact, in the last years a great deal of text-linguistic and text-theoretical studies have been produced (and are still being produced) — which not only has brought about thorough investigations of specific problems and topics, but has also enlarged the field of research by both involving different disciplines (cognitive psychology, logic, poetics, sociology, ethnomethodology, etc.) and giving rise to new theories and methods. Furthermore, text linguistics and text theory have now crossed the European borders and are becoming successful in the USA as well (cf. Beaugrande, 1980). A series of volumes witnesses all that, for example van Dijk (1977), Coulthard (1977), Dressler, ed. (1978), Petöfi, ed. (1979; 1981), Schmidt (1980a), Beaugrande and Dressler (1981)3. Inevitably, a lot of new definitions 3

Beaugrande and Dressier (1981) is a revised version of Dressier (1972). Whereas Dressier (1972: 1) gives a very concise definition of 'text' ("eine abgeschlossene sprachliche Äuße-

'Text'/'Discourse' Definitions

427

of 'text' and 'discourse' have been provided. Let's survey the most important ones. 6. van Dijk (1977) may be connected — in some respects, at least —with van Dijk (1972) and with a pragmatic point of view. In fact, after stating that "there are systematic differences between compound sentences and sequences of sentences, especially at a pragmatic level of description, and the meaning of sentences may depend on the meaning of other sentences of the same utterance although not always in the same way as the meanings of clauses in compound or complex sentences", he assumes that "utterances should be reconstructed in terms of a larger unit, viz that of text. This term will here be used to denote the abstract theoretical construct underlying what is usually called a discourse. The utterances which can be assigned textual structure are thus acceptable discourses of the language — at this level of the account of acceptability, i.e. are well-formed and interpretable" (van Dijk, 1977: 3). van Dijk, then, attaches particular attention to the 'pragmatic component', which "should not merely specify appropriateness conditions for sentences, but also for discourses" (1977: 3). These views are reaffirmed in van Dijk (1979: 512): "By the term 'text' I will [...] only understand a theoretical construct, both of a theory of language or grammar and of other theories of discourses, whereas the term 'discourse' is used as an observational term and as an intuitive, everyday term"4. The pair 'text' and 'discourse' is also traceable in Coulthard (1977). Coulthard identifies three levels in human communication: meaning or discourse, form or syntax, and substance or phonology (1977: 1). Therefore, he cannot but emphasize the property of coherence: "Sentences combine to form texts and the relations between sentences are aspects of grammatical cohesion; utterances combine to form discourses and the relations between them are aspects of discourse coherence" (1977: 6). 7. In 1977, Petöfi conducted an inquiry concerning the basic problems in the study of text and discourse (Petöfi, ed., 1979; 1981). Three of the questions which were posed concerned the relation 'text' vs 'sentence' directly. It is, then, really a summa of the research in text linguistics and discourse analysis, rung"), in Beaugrande and Dressier (1981: 3) 'text' is defined as "a communicative occurrence which meets seven standards of textua/ity", i.e.: cohesion, coherence, intentionality, acceptability, informativity, situationality, and intertextuality. Major factors about these standards are revealed by the "mechanisms which combine texts as single contributions into discourses as sets of mutually relevant texts directed to each other" (Beaugrande and Dressier, 1981: 19). A different conception is expressed by Kukharenko: "The basic property of a text (which is not observed in a sentence as an abstract language pattern) is its pragmatic relevance. A text is generated and functions only as a (process of) reflection of some concrete piece of reality, it is always communicatively-oriented and situationally-bound. As such, text should be viewed not as a theoretical construct but as an object-language element" (1979: 235 — 236; my italics).

428

L. Vitacolonna

which — as such — provides us with a great many definitions of 'text' and 'discourse'. Several contributors have tried to explain the notions of text and discourse by appealing to that of 'sentence'. For example, Wirrer (1979: 126) argues: "A definition of the terms 'text' or 'discourse' requires the definition of the term 'sentence'", since text and discourse "cover any coherent set of sentences". And Albaladejo Mayordomo (1981: 1) writes: "A text is a set formed by a number n (n > 1) of sentences with coherence and sense", which, however, seems to be in opposition to what is stated just before: "The length of the text, as a first order linguistic unit grouping smaller linguistic units, is not determinate. A single word that can communicate an autonomous or self-sufficient content separately [...] is a text, as well as a single sentence can be a text" (ibid.; my italics). Beaugrande (1979) also appeals to 'sentence', but his treatment is much more elaborated. According to him, the "most basic difference between sentence and text is that the sentence is definable within a single system, usually called grammar in linguistics and syntax in semiotics. Texts must be defined by referring to multiple systems, and to the correlations of the latter. Hence, the sentence is an essentially monosystemic entity, and the text a trans-systemic entity" (1979: 468). If so, "whether a set of elements constitutes a text can be decided only if we also 1) match its intersystemic structures with the correlation procedures of language systems and 2) its formation with its discoverable motivation" (1979: 469). Furthermore, Beaugrande (1979: 472) also tells 'text' from 'discourse': "The place of the actual occurrence of the transsystemic enity text is in discourse. A discourse may include only one text, as in the case of the novel; or many texts, as in a conversation. The text is delimited by the intention of its producer to supply a set of mutually relevant elements, and not simply by its grammatical completeness". Heger's (1979) treatment, too, is sophisticated, in that "die Frage nach dem, was ein Text sei, in mehrerer Hinsicht verschieden verstanden und daher auch entsprechend verschieden beantwortet werden kann" (1979: 49). If so, the necessity of a differentiation "müßte streng genommen dem Versuch einer Beantwortung der Frage nach dem, was ein Text sei, eine ausführliche Abhandlung über jenen metatheoretischen Fragenkomplex vorangehen lassen, für dessen Andeutung ein Stichwort wie 'Realdefinition vs. Nominaldefinition' genügen mag" (ibid.). Thus, it is expedient to pay attention at least to two dimensions, in denen Unterscheidungen anzubringen sind, die für die Verständlichkeit von Antworten auf Fragen zum Typ 'Was ist ein N?' relevant sind: a) Zum einen kann mit 'Was ist ein N?' entweder danach, für welchen unbekannten Gegenstand ein bekannter Name steht, oder aber danach gefragt werden, welcher unbekannte Name für einen bekannten Gegenstand steht. b) Zum zweiten kann unter dem, was in a) als 'Gegenstand' umschrieben wurde, entweder eine extensionale (Frege: Bedeutung; hier: Denotat) oder eine intensionale (Frege: Sinn; hier: — in Übereinstimmung mit der terminologischen Tradition der Sprachwissenschaft und im

'Text'/'Discourse' Definitions

429

Gegensatz zu derjenigen der Logik —: Bedeutung als das, was auf Grund semantischer Analysen in Form von Signifikaten, Sememen, Semen und Noemen beschrieben werden kann) Einheit verstanden werden (1979: 49 — 50).

The connection between these two research-lines can be represented by the following matrix: Name -> Gegenstand

Gegenstand —»· Name

extensional

Realdefinition

Taufakt

intensional

Semantische Analyse

Nominaldefinition

This scheme leads to the differentiation of the following kinds of possible answers to the question, "What is a text?": — man kann entweder von dem deutschen Lexem Text als der bekannten Größe ausgehen und — entweder erstens nach seinen Denotata fragen und diese Frage im Sinne einer Realdefinition durch eine Aufzählung solcher Denotata und eine Darstellung ihrer im Sinne irgendeiner Wirklichkeitsvorstellung realen Eigenschaften beantworten, — oder zweitens nach seinem Signifikat und/oder seinen Sememen fragen und diese Frage im Sinne einer semantischen Analyse durch die Angabe von deren Semen oder Noemen beantworten; — oder man kann drittens von dem im Sinne irgendeiner Wirklichkeitsvorstellung als — beispielsweise durch Ostension — bekannt unterstellten Denotat (oder von einer Klasse solcher Denotata) sowie dem lediglich einer provisorischen Verständigung dienenden deutschen Lexem Text als eigentlicher Unbekannten ausgehen und nach ihrer per Taufakt erfolgenden wechselseitigen Zuordnung sowie nach den Einzelheiten dieses Taufakts und der Rechtfertigung für die durch ihn erfolgende Zuordnung fragen; — oder man kann viertens von dem als Konjunktion bekannter Seme oder Noeme dargestellten und damit auch seinerseits als bekannt zu unterstellenden Semen sowie dem lediglich einer provisorischen Verständigung dienenden deutschen Lexem Text als eigentlicher Unbekannten ausgehen und nach ihrer durch Nominaldefinition erfolgenden wechselseitigen Zuordnung sowie nach der Rechtfertigung für diese Zuordnung fragen (Heger, 1979: 50-51).

It is evident that "allein schon auf Grund dieser rudimentären Vierteilung die möglichen Antworten auf die Frage nach dem, was ein Text sei, so grundsätzlich verschieden sein können, daß es abwegig wäre, aus der Adäquatheit einer Antwort gegenüber einer der vier Verstehensweisen dieser Frage auf ihre eigene Adäquatheit oder auf die Inadäquatheit anderer Antworten gegenüber einer der drei anderen Verstehensweisen zu schließen" (Heger, 1979: 51). Heger, then, introduces the term 'Signemrang' (whose definition is of course linked to the concept of 'Signem', that is, 'Language-Einheit' [1979: 51 ff.; cf. 1976: 30 ff.]) and points out that "ein jeweiliges SignemVorkommen auf Parole-Ebene somit hinsichtlich der Frage nach seiner hierarchischen Rangzugehörigkeit — wie auch hinsichtlich mancher anderen Frage — nur indirekt definiert ist. Diese scheinbare Trivialität eigens zu betonen,

430

L. Vitacolonna

besteht aus den folgenden Gründen Anlaß. Üblicherweise wird der Ausdruck Text eher auf ein jeweiliges an sein historisches hie et nunc gebundenes Parole-Vorkommen als auf die ihm entsprechende Langue-Einheit angewandt" (1979: 52). Thus, texts come to be "Vorkommen von Signemen" of particular degrees: "textstrukturelle Analysen haben Vorkommen des als Signem des Ranges R9 definierten Präsuppositionsgefüges, Interpretationen der Intentionen eines Textes haben Vorkommen des als Signem des Ranges RIO [...] definierten Assertierten Präsuppositionsgefüges, und den Autor oder das Publikum thematisierende Untersuchungen haben Vorkommen des als Signem des Ranges Rll definierten Assertierten Präsuppositionsgefüges mit Text'Hypersatz' zum Gegenstand" (1979: 53; cf. 1976). 8. In her contribution to Petöfi's inquiry, Langleben (1979) rejects the dichotomy 'sentence'/'text'. To her, "in order to estimate the difference between a text and a sentence, one cannot compare the two levels directly, but rather has to consider a set of four levels as a domain of three oppositions" (1979: 249). The levels are as follows:

Text Sentence cluster Non-simple sentence Simple sentence The oppositions are the following: (1) "According to the first opposition, a text is brought together with a sentence cluster and a non-simple sentence" (1979: 250); (2) The second opposition regards "the difference between a sentence cluster and a compound sentence with the same constituents, arranged in the same linear order. The crucial difference between the two types of structures lies in their communicative patterning. The informational value of constituents ranges from a topic to a focus. [...] while the focus of a simple sentence is usually provided by a word, the focus of a long, compound sentence tends to be a clause" (1979: 251); (3) "The new aspects brought in the foreground by the third opposition (by which the proper constituents of a texts are contrasted with the text as a whole) are the fixation of meaning and the global structure (1979: 253).

Very different from these positions are Weinrich's (1981) and Rickheit's (1981). Indeed, Weinrich's thesis is — at first sight — paradoxical, in that he sententiously declares: "There is no need for the text to be defined" (1981: 228), because among "all imaginable linguistic units, the oral or written unit called 'text' (more exactly speaking: 'text-in-a-situation') has a most prominent status as the only given unit. Confined in communicative boundary-markers

'Text'/'Discourse' Definitions

431

(e.g. the reunion and the break-up of a communicative group), it is given by communicative evidence" (Weinrich 1981: 228). Rickheit, in turn, after stating that we communicate "nicht in Form von Wörtern oder Sätzen, sondern in Texten",5 considers the text "als originäres sprachliches Zeichen" (1981: 181), and thus maintains "daß ein Text mehr als die Summe seiner Sätze ist und eine eigene Ganzheit bildet — ein alter Grundsatz, den wir bereits aus der Hermeneutik und Gestaltpsychologie kennen, wobei letztere eine solche Textstruktur als 'gute Gestalt' interpretieren würde, die durch eine davon abweichende Abfolge der Sätze gestört, bei Zusammenstellung eines Textes nach dem Zufallsprinzip völlig zerstört werden kann, da auch mit dem Verlust der geordneten Satzfolge die semantische Kohäsion des Textes verloren geht und somit das Verstehen für den Rezipienten unmöglich erscheint" (1981: 183). (cf. Harweg, 1976; 1979). Rickheit's words are similar to Oomen's (1979: 274): "The text forms a super-sign, to be interpreted with reference to its overall function", that is, a text is a whole that "cannot be equated with the sum of the meanings of its elements", for it "transcends the sum of the properties of its elements". "Defining a text as being more than the sum of its parts means that a text has a wholistic character" (Oomen 1979: 274). Therefore the distinctive characteristics of texts are the following: (a) "Where texts are processes of communication, sentences are units of grammar"; (b) "While texts are dynamic processes, sentences are 'static' units in a taxonomy"; (c) "While texts are wholistic in character, sentences can be analyzed through elementaristic procedures" (ibid.). 9. I cannot end this very scanty survey without referring to Lotman et al. (1975), Schmidt (1980a), and Petöfi (1980a, b; 1982). To analyse the definitions of text and discourse in these authors is very problematic, since in the first place I should expound their respective theories, which would take too much space. 10. Lotman et al. (1975), holding fast to the semiotic formulations of the Tartu School, give a definition of 'text' in terms of 'culture': "The text has integral meaning and integral function [...]. In this sense it may be regarded as the primary element (basic unit) of culture. The relationship of the text with the whole of culture and with its system of codes is shown by the fact that on different levels the same message may appear as a text, part of a text, or an entire set of texts" (1975: 8). Therefore 'text' is used by Lotman et al. (1975: 8) "in a specifically semiotic sense and, on the other hand, is applied not only to messages in a natural language but also to any carrier of integral ('textual') meaning — to a ceremony, a work of the fine arts, or a piece of 5

Cf. Harweg (1974).

432

L. Vitacolonna

music. On the other hand, not every message in a natural language is a text from the point of view of culture". The concept of 'culture text' is thus introduced: a 'culture text' is a text in that 'secondary language' which is culture (Lotman et al., 1975: 13): Three kinds of relations are possible between the text in the natural language and the verbal text of culture: (1) "The text in the natural language is not a text of the given culture. [...] All utterances to which the given culture does not ascribe value and meaning [...], from its point of view, are not texts"; (2) "The text in the given secondary language is simultaneously a text in the natural language"; (3) "The verbal text of the culture is not a text in the given natural language" (Lotman et al., 1975: 13). Nevertheless, a general concept of text is possible in connection with that of 'function': a text may be regarded as "a message which performs within the given culture a textual function" (Lotman et al., 1975: 14). 11. S. J. Schmidt has been developing for many years an empirical theory of literature (Schmidt 1979; 1980a,b; 1981). The starting points are Maturana and Varela's 'theory of cognition' and the concept of 'communicative interaction'. Schmidt, after defining 'communication' as a sort of 'instructional interaction' between two systems SI and S2 (1980b: 532), introduces the terms TEXT and KOMMUNIKAT, which are written with capital letters in order to indicate that "they are theoretical terms which ought not to be confused with the many different text-concepts used in linguistics" (ibid.). More exactly: TEXTS are defined as "physical objects (acoustic or graphematic) which communicators have learnt to produce and to receive as instruments or means of communication", while a KOMMUNIKAT is a 'cognitive representation': a communicator SI (a 'living system', in Maturana and Varela's terminology) transforms "the physical stimuli of a TEXT into neuronal signals adapted to his system-conditions and — by internal operations — assigns an emotionally charged cognitive structure to his set of stimuli. This cognitive representation (which is always emotionally loaded) whose value for Si's biographical situation is automatically checked by SI, I shall call the KOMMUNIKAT which SI assigns to a given TEXT" (ibid.). Beaugrande, in his translation of Schmidt (1980a), uses the terms 'surface text' for TEXT and 'communicative text' for KOMMUNIKAT: "An apperceptual presentation used by participants as means of communication [...] will be called a surface text. The surface text thus constitutes a means of communication produced and recognized by participants, via the constitution of their apperceptual abilities, as objects to which meanings, sense relations, and relevance can be assigned and which may be followed by further actions. If the uttering of a surface text by one participant actually brings others to carry out the above operations, we can designate the surface text as a communicative text" (Schmidt 1982: § 11.11).

'Text'/'Discourse' Definitions

433

Three kinds of 'texts' can thus be differentiated: (1) a COMMUNICATIVE TEXT: "An entity is a communicative text for a given participant in a communicative situation if and only if the participant perform a communicative interaction via an apperceptible and decodable surface text; or if the participant produces a surface text in order to perform a joint communicative interaction with other participants" (1982: § 11.37); (2) a LINGUISTIC SURFACE TEXT: "A linguistic surface text must be a material means of communication for participants in society and must meet the requirement of phoneticity/ graphematicity, lexicalicity, and syntacticity" (1982: § 11.77); (3) a LINGUISTIC COMMUNICATIVE TEXT: "A linguistic communicative text is the entity which a participant intends to present via a surface text, and thus to perform a communicative interaction in a situation, either individually or in co-operation with the other participants" (1982: § 11.80).

It is evident that these new formulations can only partially be linked to those of Schmidt (1973a, b). 12. In the light of his 'Textstruktur-Weltstruktur-Theorie' ('TeSWeST'), Petöfi considers the texts to be 'semiotic objects': "I am using the term text to designate a semiotic object, preserved in writing, on tape, or videotape, for which at least two native speakers of the given language agree that the given object is a text" (1980b: 74), while the term natural-language text "designates a subset of texts, namely that subset which contains semiotic objects that by nature belong to natural language or that have natural language in one component" (ibid.). Texts are, then, objects that can be interpreted as 'signs': "Objects with properties characteristic for texts must be understood as signs if they are to be used as texts" (Heydrich and Petöfi, 1980: 19). As regards the length of a text, "no restrictions are specified" (Petöfi, 1980a: 47): "Whether a verbal object qualifies as a text or not, can solely be decided by the interpreter of this verbal object in a given context, and his decision solely holds for himself. If an interpreter considers a given verbal object in a given context as an object which appears to him as a 'complete whole with a beginning and an end', it is expedient to say that he considers this object to be a text. (I wish to emphasize that length is no decisive factor of textuality. A verbal object having the length of a 'simple sentence' may qualify as a text, while a sequence of sentences, despite being connected in some way, may not)" (Petöfi, 1982)6. These remarks on the text length are very important, in that "the essential organizational principles of natural-language objects which are of more extended dimensions than the natural-language objects called 'simple sen6

Garcia-Berrio (1978) completely shares Petofi's viewpoint; he writes: "The text, as a theoretical unit, does not have an established length [extension prefijada]" (1978: 55); therefore, the text delimitation depends on the speaker's communicative intention (1978: 56). Cf Bertinetto (1979) and Langleben (1981: 286-287).

434

L. Vitacolonna

fences' in traditional terminology are not grammatical but coherence relations" (Petöfi, 1980a: 47)7. But it must be stressed that 'textuality' and 'coherence' do not explicate each other: "One can consider a verbal object a 'complete whole with a beginning and an end', i.e. a 'text', without considering it coherent, i.e. without being able to interpret it in a for him satisfying way. On the other hand, one may consider a sequence of sentences coherent, i.e. interpret it without regarding it at the same time as a 'complete whole with a beginning and an end', i.e. a 'text'" (Petöfi, 1982)8. Yet these statements do not imply that for Petöfi 'text' and 'coherence' are "fully subjective categories": "the task of a text theory is to investigate which are the conditions for an interpreter disposing of given properties (of a given set of knowledge concerning the communication and the world) in order in a given context he can consider a verbal object of intersubjectively determinable structure to be a text (which does not mean that it is 'coherent'), a 'coherent sequence of sentences' (which does not mean that it is a 'text'), or a 'coherent text'" (Petöfi, 1982). 13. This survey has certainly been too short and, perhaps, "tendentious". Several authors have been ignored, many theories and methods have been only touched, and — what is more — a lot of definitions have been left out. Last but not least, the arrangement itself could appear muddled. In any case, I at least trust I have shown that the scholars' disagreement over the definitions of 'text' and 'discourse' rests upon their respective (meta)theoretical and methodological approaches, and is not due to the simple sake of defining. In other words: every definition necessarily presupposes a theory and/or a method — however partial they may be. It is just this theoretical/methodological background that allows to put forth definitions from different points of view: linguistic, pragmatic, philosophical, logical, sociological, psychological, etc. Thus, in accordance with individual positions, text and/or discourse have been or may be regarded as: (a) (b) (c) (d) (e) ( f)

linguistic units (Figge, 1979; Longacre, 1979); semiotic objects (Petöfi, 1980a, b); culture-specific (Lotman et al., 1975); primary elements (Schmidt, 1971; Rickheit, 1981); context-bound (Hasan, 1978); dynamic objects (Oomen, 1979); etc.

Of great importance is also the relationship between text and discourse. Here it is also possible to distinguish between: 7

8

Cf. Garcia-Berrio (1978: 57), who considers coherence to be the key-word for the definition/ delimitation of text. Cf. Charolles (1981) and Langleben (1981: 286-287).

'Text'/'Discourse' Definitions

435

(1) those who directly oppose text to discourse, so that 'text' is very often considered to be an abstract construct, while 'discourse' is viewed as the concrete or observational datum (see e.g. van Dijk, 1977) or vice versa; (2) those who look at 'text' as a unit of 'discourse' (Bertinetto, 1979); (3) those who consider 'discourse' to be "the place of the actual occurrence" of 'text' (Beaugrande, 1979); (4) those who see 'discourse' as a set of 'texts' (Beaugrande, 1981); etc. On the other hand, there are theoreticians who oppose text/discourse to a sentence/sentence cluster (see e.g. Langleben, 1979). Hence the following schemes are possible (where: T = text; D = discourse; S(C) = sentence (cluster)): (I)

( )

T =D

vs

S(C)

(III)

T vsO

vs

S(C)

This situation is somewhat complex, but does show — I believe — that text linguistics is now a fact and that everybody concerned with the study of language, semiotics, communicative processes, and the like, has to work on the basis of a text theory, if he really wants to achieve profitable results, since linguistics can be accounted for only in a text-theoretical frame. 14. Finally, I only want to add some further studies which seem to me particularly interesting: Dressier (1978); Enkvist (1974); Eroms (1978); Fossestol (1975); Fries (1971); Hartmann (1971); Hlavsa (1976); Isenberg (1971; 1977a, b); Kalverkämper (1981); Koch (1979); Kristeva (1968); Pike (1964); Schmidt (1972); Tutescu (1980); Wunderlich (1974).

References Abad Nebot, Francisco 1980 "El texto como unidad de analisis de las ciencias filologicas", Letras de Deusto 19, 99-119. Albaladejo Mayordomo, Tomas 1981 "On Text Linguistic Theory", in J. S. Petöfi (ed.), 1-15. de Beaugrande, Robert-Alain 1979 "Text and Sentence in Discourse Planning", in J. S. Petöfi (ed.), 467—494.

436 1980

L. Vitacolonna

"Text and Discourse in European Research" in European Approaches to the Study of Text and Discourse (= Discourse Processes 3/4). Norwood, N.J.: Ablex, 287 — 300. 1981 "Linguistic Theory and Metatheory for a Science of Texts", Text 1, 113-161. de Beaugrande, Robert-Alain and Dressier, Wolfgang U. 1981 Introduction to Text Linguistics. London: Longman. Benveniste, Emil 1959 "Les relations de temps dans le verbe frangais", Bulletin de la Societe de Linguistique 54,1. Bertinetto, Pier M. 1979 "Can we Give a Unique Definition of the Concept 'Text'? Reflexions on the Status of Textlinguistics", in J. S. Petöfi (ed.), 143-159. Bierwisch, Manfred 1965 Review of Harris (1964), Linguistics 13, 61-73. Brinker, Klaus 1973 "Zum Textbegriff in der heutigen Linguistik", in H. Sitta and K. Brinker (ed.): Studien %ur Texttheorie und %ur deutschen Grammatik: Festgabe für Hans Glin^ %um 60. Geburtstag. Düsseldorf, Schwann, 9 — 41. Chabrol, Claude (ed.) 1973 Semiotique narrative et textuelle. Paris: Larousse. Charolles, Michel 1981 "Strategies for Judging Acceptability", in J. S. Petöfi (ed.): 58-75. Conte, Maria-Elisabeth 1977 "Introduzione", in Conte (ed). 11-50. Conte, Maria-Elisabeth (ed.) 1977 La linguistica testuale. Milano: Feltrinelli. Coseriu, Eugenio 1967 "Determinacion y entorno: Dos problemas de una linguistica del hablar" in E. Coseriu: Teorta del lenguaje v linguistica general. 2nd ed. Madrid: Credos: 282—323 (1st ed. in Romanistisches Jahrbuch 7, 1955-56: 29-54). Coulthard, Malcolm 1977 An Introduction to Discourse Analysis. London: Longman. Dascal, Marcelo and Margalit, Avishat 1974 "A New 'Revolution' in Linguistics? — 'Text-grammars' vs. 'Sentence-grammars'", Theoretical Linguistics, 1, 195 — 213. Dijk, Teun A. 1972 Some Aspects of Text Grammars. The Hague: Mouton. 1977 Text and Context: Explorations in the Semantics and Pragmatics of Discourse. London: Longman. 1979 "New Developments and Problems in Text-Linguistics", in J. S. Petöfi (ed.): 509-523. Dressler, Wolfgang U. 1972 Einführung in die Textlinguistik. Tübingen: Niemeyer. 1978 "Aspekte der Textlinguistik", in Germanistik II. Wien: Österreichischer Bundesverlag, 23-37. Dressler, Wolfgang U. (ed.) 1978 Current Trends in Textlinguistics (= Research in Text Theory, 2). Berlin: Gruyter. Eikmeyer, Hans-Jürgen; Heydrich, Wolfgang; Petöfi, Jänos S. 1980 Some Aspects of Formal Foundations in Text Semantics (Materialien des Universitätsschwerpunktes Mathematisierung der Einzelwissenschaften, Heft XXVI). Universität Bielefeld. Enkvist, Nils E. 1974 "Nägra textlingvistiska grundfragor", in U. Telemann and T. Hultmann (ed.): Spräket i bruk. Lund: Liber läromedel, 172—206. 1978 "Stylistics and Text Linguistics" in W. U. Dressler (ed.): 174-190.

'Text'/'Discourse' Definitions

437

Eroms, Hans-Werner 1978 "Die Arbeit am Text mit sprachwissenschaftlichen Methoden", Linguistik und Didaktik 9, 129-144. Figge, Udo L. 1979 "Zur Konstitution einer eigentlichen Textlinguistik", in J. S. Petöfi (ed.): 13 — 23. Fossest01, Bernt 1975 "Tekstlingvistikk" in R. Guldal et al. (eds.): Rapport fra faglig konferanse innen nordisk spragvitenskap, As 17-18 april 1975. Oslo: 70-85. Fries, Udo 1971 "Textlinguistik", Linguistik und Didaktik 2, 219-234. Garcia-Berrio, Antonio 1978 "Situacion de la teoria textual" in J. S. Petöfi and A. Garcia-Berrio, 53-98. Grimes, Joseph E. 1966 "Some Inter-Sentence Relationships in Huichol" in Summa Antropologica en homenaje a Roberto J. Weitlaner. Mexico: D. F. Institute Nacional de Antropologia e Historia, 465-470. Halliday, Michael A. K. 1973 Explorations in the Functions of Language. London: Arnold. Harris, Zellig S. 1964 "Discourse Analysis" in J. A. Fodor and J. J. Katz (eds.): The Structure of Language: Readings in the Philosophy of Language. Englewood Cliffs, N. J.: Prentice-Hall, 355-383 (1st printing in Language 28, 1952: 1 -30). Hartmann, Peter 1971 "Texte als linguistisches Objekt" in Beiträge %ur Textlinguistik. München: Fink: 9—29. 1972 "Text, Texte, Klassen von Texten" in W. A. Koch (ed.): Strukturelle Textanalyse — Analyse du redt — Discourse analysis. Hildesheim: Olms: 3 — 22 (1st ed. in Bogawus 2, 1964: 15-25). Harweg, Roland 1968 Pronomina und Textkonstitution. München: Fink. 1974 "Textlinguistik" in W. A. Koch (ed.): Perspektiven der Linguistik . Stuttgart: Kröner, 90-116. Hasan, Ruqaiya 1978 "Text in the Systemic-Functional Model" in W. U. Dressler (ed.): 228-246. Heger, Klaus 1976 Monem, Wort, Sat^ und Text. Tübingen: Niemeyer. 1979 "Text und Textlinguistik", in J. S. Petöfi (ed.): 49-62. Heydrich, Wolfgang and Petöfi, Janos S. 1980 "Pragmatic Considerations within a Text-Theoretical Framework" in: H. Parret; M. Sbisä; J. Verschueren (eds.): Possibilities and Limitations of Pragmatics: Proceedings of the Conference on Pragmatics, Urbino,July 8— 14, 1979 (Studies in Language Companion Series 7). Amsterdam: John Benjamins B.V., 1981: 313 — 330; also in H. J. Eikmeyer, W. Heydrich, and J. S. Petöfi, 19-42). Hjelmslev, Louis 1961 Prolegomena to a Theory of Language. Madison: The University of Wisconsin Press (1st ed. Omkring sprogteoriens grundlagelse. Kobenhavn: Ejnar Munksgaard, 1943). Hlavsa, Zdenek 1976 "Towards a Definition of a Text", in F. Danes' and D. Viehweger (eds.): Probleme der Textgrammatik (= Studia grammatica XI). Berlin: Akademie-Verlag, 41—45. Isenberg, Horst 1970 Der Begriff "Text" in der Sprachtheorie (Deutsche Akademie der Wissenschaften, Berlin, Arbeitsstelle Strukturelle Grammatik, ASG-Bericht No. 8), 1—21. 1971 "Überlegungen zur Texttheorie" in J. Ihwe (ed.): Literaturwissenschaft und Linguistik: Ergebnisse und Perspektiven. I: Grundlagen und Voraussetzungen. Frankfurt am Main; Athenäum, 155 — 172.

438 1977a

L. Vitacolonna

"Der 'Text' als kommunikative Einheit" in F. Danes and D. Viehweger (eds.). Probleme der semantischen Analyse (= Studia grammatica XV). Berlin: AkademieVerlag, 358-377. 1977b "'Text' vs. 'Satz'" in F. Danes and D. Viehweger (eds.): Probleme der Textgrammatik II (= Studia grammatica XVIII). Berlin: Akademie-Verlag, 119 — 146. Kalverkämper, Hartwig 1981 "Der Bestand der Textlinguistik I, II", Deutsche Sprache 9, 224-270, 329-379. Koch, Walter A. 1965 "Preliminary Sketch of a Semiotic Type of Discourse Analysis", Linguistics 12, 5-30. 1966 "Einige Probleme der Textanalyse", Lingua 16, 383-398. 1979 "Zur linguistischen Analyse von Texten", in B. Fabian (ed.): Ein anglistischer Grundkurs %ur Einführung in das Studium der Literaturwissenschaft. Frankfurt am Main: Athenäum, 175-194. Kristeva, Julia 1968 "Problemes de la structuration du texte", Linguistique et Litterature 12, 55 — 64. Kukharenko, Valeria 1979 "Some Considerations about the Properties of the Text" in J. S. Petöfi (ed.), 235-245. Langleben, Maria M. 1979 "On the Triple Opposition of a Text to a Sentence", in J. S. Petöfi (ed.), 246-257. 1981 "Latent Coherence, Contextual Meanings, and the Interpretation of Text", Text 1, 279-313. Longacre, Robert E. 1979 "Texts and Text Linguistics", in J. S. Petöfi (ed.), 258-271. Lotman, Jurij et al. 1975 Theses on the Semiotic Study of Culture (as Applied to Slavic Texts). Lisse: de Ridder. Oomen, Ursula 1969 "Systemtheorie der Texte", Folia Linguistica 5, 12-34. 1979 "Text and Sentences", in J. S. Petöfi (ed.), 272-280. Petöfi, Jänos S. 1980a "Written, Spoken, and Face-to-Face Verbal Communication: Some Philosophical Aspects of the Investigation of Natural Language" in H.-J. Eikmeyer, W. Heydrich, and J. S. Petöfi (1980: 43-72); also in R. Haller and W. Grassl (eds.): Language, Logic and Philosophy. Proceedings of the 4th International Wittgenstein Symposium, 28th August to 2nd September 1979, Kircbberg\Wechsel (Austria). Wien: Hölder-PichlerTempsky, 1980: 144-159. 1980b "Representation Languages and their Function in Text Interpretation" in H. Eikmeyer, W. Heydrich and J. S. Petöfi (1980: 73—182); in an abbreviated version in S. Allen (ed.): Text Processing: Text Analysis and Generation, Text Typology and Attribution. Proceedings of the Nobel Symposium 51 (= Data Linguistica 16). Stockholm: Almqvist & Wiksell International, 1982: 85-122. 1982 "Intervista con Jänos S. Petöfi" (made by L. Vitacolonna), Studi italiani di linguistica teorica ed applicata, XI: 367 — 380. (Here I have used the original text.) 1985 "European Research in Semiotic Textology: A historical, thematic, and bibliographical guide", mimeo. Petöfi, Jänos S. (ed.) 1979 Text vs Sentence: Basic Questions of Textlinguistics (= Papers in Textlinguistics, 20). Hamburg, Buske. 1981 Text vs Sentence Continued (= Papers in Textlinguistics, 29). Hamburg, Buske. Petöfi, Jänos S. and Garcia Berrio, Antonio 1978 Linguistica del textoy critica literaria. Madrid: Comunicacion. Pike, Kenneth L. 1964 "Beyond the Sentence", College Composition and Communication 15, 129 — 135.

'Text'/'Discourse' Definitions

439

Plett, Heinrich F. 1975 Textwissenschaft und Textanalyse. Heidelberg: Quelle und Meyer. Rickheit, Gert 1981 "Textprobleme aus sprachpsychologischer Sicht", in J. S. Petöfi (ed.): 181 — 190. Rieser, Hannes 1978 "On the Development of Text Grammar" in W. U. Dressler (ed.): 6-20. Schmidt, Siegfried J. 1971 "'Text' und 'Geschichte' als Fundierungskategorien", in W.-D. Stempel (ed.): Beiträge %ur Textlinguistik. München: Fink, 31—52. 1972 "Text als Forschungsobjekt der Texttheorie" in Der Deutschunterricht 24, 7 — 28. 1973a "Texttheorie/Pragmalinguistik", in H. P. Althaus et al. (eds.): Lexikon der Germanistischen Linguistik. Tübingen: Niemeyer, 233—244. 1973b Texttheorie: Probleme einer Linguistik der sprachlichen Kommunikation. München: Fink. 1979 "Empirische Literaturwissenschaft as Perspective", Poetics 8, 557 — 568. 1980a Grundriß der Empirischen Literaturwissenschaft. Part 1: "Der gesellschaftliche Handlungsbereich LITERATUR", Braunschweig: Vieweg. 1980b "Fictionality in Literary and Non-literary Discourse", Poetics 9, 525 — 546. 1981 "Empirical Studies in Literature: Introductory Remarks", Poetics 10, 317 — 336. 1982 Empirical Foundations for the Study of Literature. The Components of a Basic Theory [Translation of Schmidt (1980a) by R. de Beaugrande] (= Papers in Textlinguistics, 36). Hamburg, Buske. Segre, Cesare 1979 "Discorso", in Enciclopedia, vol. IV. Torino: Einaudi; 1056-1084. 1981 "Testo", in Enciclopedia, vol. XIV. Torino; Einaudi: 269-291. Todorov, Tzvetan (ed.) 1965 Theorie de la /literature. Paris: Editions du Seuil. Tutescu, Mariana 1980 "Proposition vs. Phrase vs. Text", Revue Roumaine de Linguistique 25, 413 — 421. Weinrich, Harald 1971 Tempus: Besprochene und erzählte Welt. 2nd ed. Stuttgart: Kohlhammer. 1981 "Text äs Primum Datum of Linguistics", in J. S. Petöfi (ed.), 228-230. Wirrer, Jan 1979 "Five Questions on Text Linguistics", in J. S. Petöfi (ed.), 123—142. Wunderlich, Dieter 1974 "Textlinguistik" in H. L. Arnold and V. Sinemus (eds.): Grundlage der Literaturund Sprachwissenschaft. II: Sprachwissenschaft (= dtv Wissenschaftliche Reihe, 4227). München: dtv, 386-397.

JÄNOS S. PETÖFI AND EMEL SÖZER

Static and Dynamic Aspects of Text Constitution1 0. Introductory remarks

In the history of scientific research sometimes the aspects of specialization, sometimes the aspects of integration dominate. In the first case, development leads to special branches of knowledge with well-defined methodologies, however, generally with a small object domain; in the latter case interdisciplinary fields of knowledge result which permanently struggle to clarify the questions of their object domains and of adequate methodologies. At present, in the history of research into verbal texts the aspects of integration seem to dominate with all the advantages and disadvantages. Linguistics, rhetorics, poetics, stylistics, cognitive psychology, ethnomethodology, artificial intelligence research — to mention only the most important ones — by continually influencing each other contribute more and more to the establishment of a textological framework which serves to investigate the apparently inseparable aspects of texts. In our present study we want to discuss some aspects of a central textological topic, namely text constitution (the organization of the verbal material of texts). First we will treat the constitution of signs in general, searching for the most general properties which can be extended to any kind of signs of any complexity. We will also analyse the possible objects, types and goals of the interpretation of signs. Then, on one hand, we will treat those basic notions which, to our mind, are of central importance in the analysis of text construction (the organization of the verbal and the represented material of texts). On the other hand, we will treat the question of which representation languages are needed to carry out text interpretation in an inter subjective way and to describe the results of text interpretation explicitly. After having explicated these basic notions, we will analyse some aspects of the constitution of a given text, where the analysis will be confined to the 1

Our paper is based on Hatakeyama-Petöfi-Sözer (1984a), on the papers discussing Hatakeyama-Petöfi-Sözer (1984a), collected in Conte (ed.), (1987), and on Hatakeyama-Petöfi-Sözer (1984b) and (1984c).

Static and Dynamic Aspects of Text Constitution

441

levels on which the analysis can be performed without having to apply more complex representation languages. In the concluding remarks we want to point out some contemporary questions concerning the strategy of textological research. /. Signification, sign, interpretation

In order that the basic questions of text constitution can be treated adequately, it is first of all necessary to discuss the most important aspects of signification, sign and interpretation. 1.1. The relation signification exists in its most general sense between an object declared/accepted as a significans and the object signified / to be signified by this significans. It is usual to demonstrate this relation by means of the socalled semiotic triangle. 1.1.1. The semiotic triangle is represented, for instance, by Lyons in the following way (Lyons, 1968: 404, here Figure 1): Meaning (Concept)

Word

Form

Referent Figure 1: The semiotic triangle

This representation, with a slight alteration in terminology, mirrors the views of the medieval grammarians which Lyons (1968: 403 f.) characterizes as follows: As the distinction was formulated by the medieval grammarians: the form of a word (the fox-part of a dictid) signified 'things' by virtue of the 'concept' associated with the form of the word in the minds of the speakers of the language; and the 'concept', looked at from this point of view, was the meaning of the word (its signiftcatio).

The terminological alteration involves the more general term referent used by Lyons to replace the original term thing in the semiotic triangle.

442

J. S. Petöfi, E. Sözer

Although in the semiotic triangle only the four elements are usually given which can also be found in Lyons' representation (i. e. word, form, meaning [concept], referent) it is also obvious from the above quotation from Lyons that there is always a fifth element involved too, namely the user/interpreter of the word, since the concept is to be understood as "'concept' associated with the form of the word in the minds of the speakers of the language". Discussions about the nature of word meaning have been and continue to be centered around the following three questions: (1) the question most often discussed is whether or not signifying things actually takes place by virtue of the concept; in other words, whether or not the concept (in another terminology: the sense, the intension) determines the referent (in another terminology: the denotatum, the extension, the extralinguistic correlate), (2) the second question concerns the character and organization of the concept assigned or assignable to a word; finally, (3) the third question raises the problem of whether it is the concept, the referent or both together which should be considered as the meaning of a word. 1.1.2. We replace the semiotic triangle by a semiotic pyramid (cf. Figure 2) in which we take the results achieved so far by research on the above questions into consideration.2 The analogen of the semiotic triangle in the semiotic pyramid is the triangle defined by the vertices indicated by the symbols "Ss", "s(Oj)" and "Oj". Before dealing with the components of this analogous triangle, let us have a look at the explication of the symbols of the semiotic pyramid. The symbol "Oj" stands for the object / state of affairs (= correlate), to which the signifier refers. The symbol "Oi" stands for the object which is used as the signifier object (= the vehicle of the significans). The symbols "s(Oj)" and "s(Oi)" stand for those stereotype knowledge-systems (of laymen and/or experts) which are assigned by the users of the signifier object Oi to the object Oj (object / state of affairs) and the signifier object respectively. The symbol "Ss" stands for the Jignificanj· conceived as a pair consisting of the object Oi and the knowledge system s(Oi) assigned to it; similarly, the symbol "Sm" stands for the Signifi.catu.rn conceived as a pair consisting of the object / state of affairs Oj and the knowledge system s(Oj) assigned to it. The symbol "Ir" stands for the /nterpreter, and the symbol " , " stands for the sign (= signum) conceived as the manifestation of the relation Ss-Sm. 2

The different figures in Petöfi (1982a, 1982b and 1984) represent the semiotic relations from different points of view. The semiotic pyramid in the present paper involves an important new angle as compared to the figures mentioned above in that it parallels the views concerning the significans component and the significatum component. This presentation is mainly a result of discussions with W. Heydrich. As to the results of research into the questions discussed here we first of all think of Putnam's respective works (cf. Putnam, 1975, 1978, and Heydrich-Petöfi, 1983). As to the semiotic relations cf. also Raible (1984), and Hatakeyama-Petöfi-Sözer (1984b).

Static and Dynamic Aspects of Text Constitution

443

lr

s(Oj) stereotype knowledge =

Oj objects/ states of affairs Figure 2: The semiotic pyramid

The arrows symbolize that both the significans and the significatum and the signum are (complex) objects depending on the interpreter. 1.1.3. The triangle Ss-s(Oj)-Oj and the traditional semiotic triangle diverge from one another in the following ways: a) In the semiotic pyramid both the significans and the significatum are a pair; the object Oi can fulfil the role of the significans only together with the knowledge /s(Oi)/ assigned to it by the users of this object as a semiotic object (= a vehicle of the significans); since — as the results of research in language philosophy show — the concept and its analogon in the semiotic pyramid /s(Oj)/ do not clearly determine the referent and its analogon in the semiotic pyramid /Oj(= the correlate)/, respectively, the meaning (in our terminology the significatum] is necessarily a pair consisting of concept and referent;

444

J. S. Pet fi, E. S zer

b) the signum Σκ,ο in the semiotic pyramid, corresponding to the element 'word' in the semiotic triangle, is in direct connection with both the Ss and the Sm, more precisely, it manifests the connection established on the basis of the knowledge system and/or the communicative activity of the interpreter(s); c) the meaning relation in the semiotic pyramid exists between the pairs Ss and Sm (we call this relation signification relation), i.e. the signification relation is neither identical with the Ss-s(Oj) (more exactly the Oi-s(Oj)) relation generally declared to be the meaning relation by linguists, nor the Ss-Oj (more exactly Οί-Oj) relation generally declared to be the meaning relation by the logicians. On analogy with the traditional terminology we call the Ss-s(Oj) relation the designation relation, while the relation Ss-Oj will be called the denotation relation. For naming the s(Oj)-Oj relation we have introduced the term correspondence, with respect s(Oj) we speak about explication relations. For simplicity's sake and in order to express the unity of the individual sign components we can use the following notation in connection with a sign: π

Σκ, or instead of O'j •"""Σκ, orL instead of s(Oi) instead "Εκ, ο"1 instead nL ~"~Σκ, o instead pair of any extent and any complexity as a sign (to express it with respect to verbal signs and not in terms of a precise terminology: not only words but also texts, further, not only words/texts to be understood in their direct sense but also those with a symbolic meaning). In the lexicon of a language ( = a system of signum) there can, of course, be represented only the conventionalized signification-relations concerning elementary signs — the so-called elementary systemic pairs. The pairs produced in the course of the communication (involved in communication situations) — and called, consequently, "communicative pairs" — can be different from the systemic ones in the case of elementary signs as well. Finally it should also be mentioned that not only objects, physical states or events can be imagined as Oj-s (i.e. as correlates) but also emotional states, either on their own or as accompanying objects, physical states or events. Both with the construction of a lexicon and in the course of interpretation it is necessary to treat the question, of how to handle these emotional states: what is to be considered as systemic in connection with them — if there is anything that can be considered as systemic at all, and how to reveal the solely communicative manifestation of them. 1.3. The basic operation of natural-language communication is interpretation. Thus, in the present subsection we want to deal with some aspects of interpretation, among them the possible objects of interpretation, the possible goals of interpretation, the different types of interpretation and, finally, the main factors of meaning-constituting interpretation. 1.3.1. A global summary of the possible objects of interpretation is given in Figure 3. The two main distinctions presented in this Figure are the following: (1) The interpretation may be directed towards the relationship between the significans and the significatum i.e. the construction of a signum in general, or it may be directed towards the functional embeddedness into various contexts of a given signum; (2) the interpretation may also be directed towards the static nature of the construction and/or of the functional embeddedness (the description of what kind of relations exist (can exist) between such and such elements), or it can be directed towards the dynamic nature of the construction and/or of the functional embeddedness (the description of how the relations between such and such elements arise (can arise)). Concerning the interpretation directed towards the static nature of the relations it is usual to speak about structural interpretation, while with respect to the dynamic nature of the relations the interpretation is usually called procedural interpretation.'' 4

As to distinguishing structural and procedural linguistics, and the aspects of dynamics in linguistics cf. Ballmer (ed.) (1985), Eikmeyer (1983), Petöfi (1983a), and Rieger (ed.) (1984).

Λ L· r - s "* S - CL

'ΪΗ

-o

O -D £ 4-> ^·£ u D C W 6C in U" TD ^C ul S « Λ 0 in ? X 4-j « 3 υ U S S 4-1 "S -g^ >, Λ »H J3 Ό >< •-t- t; δ o in -O _^T « C u J3 u < 3 H c S S (j in " »S u -C ~ ^ IS '^ " rt

a ω §

ε ε

ρP

in

a

l|

ε ε

owar

= S|

3

2

in rt SG. CL *>.O Λ

4_*

4^

U

ι™

w ^iy

E" 4- in 'S 3 ii in o1 3 QJ O r.

£

U

l l „ S

*—l

-o· 4-* § S 1 ,c 3 u σ-

U

C /-\ \-^

1

rt

.-H

p

^3

4—1 ^\

S 3 » C

'E "Ι 01 ANSWER

C203

Q-A PAIR C2Q4

("GREETING

C'Ol

LGREETING 002 [-DESCRIPTION] C'03 LDESCRIPTIONz • QUESTION ANSWER, -ANSWER 2 -ANSWER 3

C'04 C'05 C'06 O07 O08

[-QUESTION LANSWER

O09 010

- COMMENTARY Oil Figure 11

As to the aspect depending on the verbal material, we can speak of the cohesion of the linear composition of the verbal units manifesting the state of affairs / state-of-affairs configurations. The main questions concerning the cohesion of the linear composition are the following: a) which elements belong to the complete sets of possible linear arrangements of the constituents within the individual nominal, verbal, etc. phrases, and what is the relation between these sets and the linear arrangements realized in the phrases to be interpreted; b) which elements belong to the complete sets of possible linear arrangements of the immediate constituents within the elementary first-grade composition-units, and what is the relation between these sets and the linear arrangements realized in the elementary first-grade composition-units to be interpreted; c) which elements belong to the complete sets of possible linear arrangements of the immediate constituents within the complex first-grade compositionunits, and what is the relation between these sets and the linear arrangements realized in the complex first-grade composition-units to be interpreted; d) which elements belong to the complete sets of possible linear arrangements of the constituents within the second-grade composition-units, and what is the relation between these sets and the linear arrangements realized in the second-grade composition-units to be interpreted; etc. We do not find any example in (T) to discuss a); for b) we can consider the immediate constituents of every first-grade composition-units; the first-grade composition-unit of Oil can be an example for c), in which we should analyse the set of the possible linear arrangements of the immediate constituents "moi", "se dit le petit prince", "si j'avais cinquante-trois minutes ä depenser", and "je marcherais tout doucement vers une fontaine ..."; the second-grade composition-unit C203 can be an example for d), in which we

Static and Dynamic Aspects of Text Constitution

465

should analyse the set of the possible linear arrangements of O06, O07, COS. An adequate description of the linear compositional cohesion of (T) requires to perform the types of analyses enumerated above. However, we will have to abstain from performing these analyses here. 3.2.3. It is possible to apply the notion completeness also to cohesion (especially to compositional cohesion). The question to be investigated here is, why we believe (T) to be complete if we believe it to be complete, and whether one could leave out one (or some) of its composition units without causing to believe that (T) is not complete any longer. An interpreter believes (T) to be complete with respect to compositional cohesion, if the description of the state-of-affairs configuration manifest in it meets, according to his experience, the usual expectation concerning complete descriptions. We want to emphasize that not the individual states-of-affairs and the expectations concerning them are meant here. To operate with these is a task of model building and of constructing interpretamenta: What is meant here is the expectation concerning the description itself. The expectation pattern which is met by (T) can be formulated in a general way as follows: A narrator describes the meeting and the short dialogue of two persons: - A and B greet each other / = GREETING PAIR/; — Since the narrator assumes that B is unknown to the reader/listener, he briefly presents B /DESCRIPTION/; — A asks B, why he is doing what he is doing; B answers the question / = QUESTION-ANSWER PAIR/; — A is not fully satisfied by the answer/explanation of B, thus he asks another question; B answers this question, too / = QUESTION-ANSWER PAIR/; — For A the answer of B (the fact involved in the answer) appears strange/ ununderstandable, thus he gives own view about it / = COMMENTARY/. Since this expectation pattern can be considered as a stereotype, and since (T) meets this pattern, we are convinced that most interpreters will regard (T) as being complete with respect to compositional cohesion. We will leave it to the reader to investigate the question, whether one (or some) of the composition units of (T) could be left out, and if so, which one(s). 3.2.4. It is expedient also with respect to cohesion, to speak about intratextual and intertextual cohesion. I would now like to add some remarks concerning intertextual cohesion: The elements belonging to the thematic group 'thirst' play a relevant role in the whole text of Le petit prince, cf., for example, the chapters II, VII, XXII, XXIV, XXV, and XXVI. In chapter XXVI the last constituent of Oil reoccurs almost word by word: "..., et je serais heureux,

466

J. S. Petöfi, E. SÖzer

mois aussi, si je pouvais marcher tout doucement vers une Fontaine!". (It would also be interesting to analyze the symbolic meaning of the elements belonging to this thematic group.) Also the elements belonging to the thematic group 'count' are relevant, cf., for example, the chapters I, II, IV, VII, XIII, XVI, XVII, XVIII, and XXII. The fetishism of numbers appears to be one of the most characteristic properties of men both for the narrator and for the Little Prince. However, we also find intertextual relations concerning the compositional cohesion: consider, for example, the fact that the compositional set-up of many other chapters is similar to the one analysed above, and the fact that the basic compositional principle of the whole text is to present the views of the Little Prince in form of dialogues. 3.3. Though in the present paper we intend to discuss primarily the aspects of text constitution, we also want to briefly touch upon the last three factors of meaning-constituting interpretation (cf. 1.3.4 (5) to (7)) with reference to the text (T) analysed here. The relatum universe assignable to the text (T) consists of the world of the narrator, the merchant, the experts, men (in general), and the Little Prince. The only common event in these worlds is thirst and the attitude to how to quench it; the relation between these worlds is determined by the way the persons constituting the individual worlds approach the way the persons constituting the other worlds approach thirst and how to quench it. On the basis of (T) it appears that it is not good to waste time on drinking, the experts compute how much time could be saved if one did not need to drink, the merchant finds that saving time is important and sells pills which enable saving time, the Little Prince, however, does not understand this, he would spend the saved time for walking towards a spring. On the basis of this relatum universe the interpreter constructs his own interpretation model(s). Without going into details let us assume that the interpreter can easily construct a fictitious world in which everything happens (because it can happen) in the way as described in the text (T). Since the interpreting model and the relatum universe which can be assigned to the text to be interpreted are compatible with each other, the model can be accepted without any modification as an interpretamentum assignable to the text (T).15 As a consequence of compatibility, we can regard text (T) — with reference to the above outlined model and interpretamentum — as being explicitly coherent. Two further remarks to conclude: (i) It seldom happens that the interpretation process is as free of problems concerning the factors (5) to (7) as it is 15

As to questions of the interpretation process discussed here, cf. Petöfi (1983b).

Static and Dynamic Aspects of Text Constitution

467

the case here. In most cases it is necessary to explicitly complement the text to be interpreted by inferences that can be drawn from the text in order that the interpretation model can be constructed and can be compared with the relatum universe at all. (ii) We have dealt only with the so-called direct meaning of (T). The real meaning of (T) is, however, a symbolic meaning, the treatment of which would exceed both our goal and the frames of our study.

4. Concluding remarks

In our present paper we have discussed questions of the static and dynamic interpretation of text constitution. Our primary aim was to present an outline of the spectrum of those factors which (can) contribute to connexity and cohesion of texts. We are convinced that these factors can only be investigated adequately in the framework of an all-embracing theory, and we are also convinced that textological research leads sooner or later to the establishment of such a theoretical framework. To conclude our study we would like briefly to outline four groups of tasks which we consider to be central for textological research at present: (1) Relying upon the experiences of textological research until now, it seems to be necessary to examine again, which are the basic questions of establishing a flexible text typology (flexible with respect to a number of aspects to be considered). We also have to investigate the question, whether conversation analysis and the analysis of written texts require different methods, and if they do, in what respect these methods differ from each other; then we have to investigate, with which categories an optimal transcription language has to operate when transcripts are to be supplied for different types of conversations and written texts. (2) Both within the individual disciplines dealing with texts and in the various contexts of interdisciplinary cooperation it is necessary to investigate which are the basic questions of establishing an integrative textological framework. In this investigation, on one hand, the questions primarily concerning the verbal material of texts have to be critically scrutinized again, on the other hand, the questions of the relationship between verbal constitution and text meaning. In the course of striving for integration one has to reinterpret the results of Russian Formalism and those of French Structuralism as well as the results of grammatical and of hermeneutic research — just to mention four directions which cannot be said to be convergent! (3) Also the questions of possibilities and limits of structural and procedural analysis and of their interrelation have to be investigated. In connection with procedural analysis, special stress has to be laid on the differences

468

J. S. Pctöfi, R. Sözer

between the methodological problems of real time procedural analysis (e.g. the one performed with a conversation simultaneously) and those of procedural analysis with no time restriction (e.g. the procedural analysis of a written text which does not simulate a continuous real-time reading). (4) A considerable number of pilot studies ought to be done in order to investigate the interpretation types (and to explicitely simulate them as far as possible) which we have enumerated in point 1.3.1. A research field of the complexity of textology requires more extensive cooperation between the different schools in the individual disciplines and also between different disciplines than has been the trend up until now.

Bibliography Allen, Sture (ed.) 1982 Text Processing: Text Analysis and Generation, Text Typology and Attribution. Proceedings of Nobel Symposium 51. Stockholm: Almqvist & Wiksell International. Ballmer, Thomas T (ed.) 1985 Linguistic Dynamics: Discourses, Procedures and Evolution (= Research in Text Theory 9). Berlin: de Gruyter. Brown, Gillian and Yule, George 1983 Discourse Analysis (= Cambridge Textbooks in Linguistics). Cambridge: Cambridge University Press. Conte, Maria-Elisabeth (ed.) 1987 Kontinuität und Diskontinuität in Texten und Sachverbaltskonfigurationen: Diskussion über Konnexität, Kohäsion und Kohären^ (= Papers in Textlinguistics 50). Hamburg: Buske. Eikmeyer, Hans-Jürgen 1983 "Procedural Analysis of Discourse", in: Petöfi (ed.): (1983a), 11-37. Gülich, Elisabeth and Raible, Wolfgang (eds.) 1977 Linguistische Textmodelle. München: Fink. Hatakeyama, Katsuhiko; Petöfi, Jinos S.; Sözer, Emel 1984a "Text, Connexity, Cohesion, Coherence", in: Sözer (ed.) (1985), 36-106. (German version in: Conte, ed. (1987); French version as Documents de Travail et prapublications A/132—134, Universita di Urbino, Italia). 1984b "Nachtrag zu 'Text, Konnexität, Kohäsion, Kohärenz'", in: Conte (ed.): (1987). 1984c "Konnexität, Kohäsion und Sprachstruktur. Textanalyse.", in: Conte (ed.): (1987). Heydrich, Wolfgang and Petöfi, Jänos S. 1983 "A Text-Theoretical Account of Questions of Lexical Structure", Quaderni di Semantica IV: 120-127, 294-311. Heydrich, Wolfgang and Petöfi, Janos S. (eds.) 1985 Aspekte der Konnexität und Kohären^ von Texten (= Papers in Textlinguistics 51). Hamburg: Buske. Neubauer, Fritz (ed.) 1983 Coherence in Natural-Language Texts (= Papers in Textlinguistics 38). Hamburg: Buske.

Static and Dynamic Aspects of Text Constitution

469

Petöfi, Jänos S. 1982a "Representation Languages and Their Function in Text Interpretation", in: S. Allen (ed.), 85-122. 1982b "Meaning, Text Interpretation, Pragmatic-Semantic Text Classes", in: H. Rieser (ed.), 453-491. 1983 a "Aufbau und Prozeß, Struktur und Prozedur: Einige Grundfragen der prozeduralen Modelle des Sprachsystems und der natürlich-sprachlichen Kommunikation", in Petöfi (ed.), (1983b), 310-321. 1983b "Text, Signification, Models, and Correlates: Some Aspects of Text Comprehension and Text Interpretation", in: G. Rickheit und M. Bock (ed.), 266—298. 1984 "Ausdrucksfunktionen, Sätze, kommunikative Akte, Texte (Aspekte der Bedeutung und ihre Thematisierung im Rahmen einer Texttheorie)", in: A. Rothkegel and B. Sandig (eds.), 26-50. Petöfi, Jänos S. (ed.) 1983a Methodological Aspects of Discourse Processing (= Text 3, 1). Berlin: Mouton. 1983 b Texte und Sachverhalte: Aspekte der Wort- und Textbedeutung (= Papers in Textlinguistics 42). Hamburg: Buske. Petöfi, Jänos S. and Sözer, Emel (eds.) 1983 Micro and Macro Connexity of Texts (= Papers in Textlinguistics 45). Hamburg: Buske. Putnam, Hilary 1975 "The Meaning of 'Meaning'", in Mind, Language and Reality: Philosophical Papers. Vol. II., Cambridge: Cambridge University Press, 215 — 271. 1978 "Meaning, Reference and Stereotypes", in: F. Guenthner, and M. Guenthner-Reutter (eds.): Meaning and Translation: Philosophical and Linguistic Approaches. London: Duckworth, 61—81. Raible, Wolfgang 1984 "Phänomenologische Textwissenschaft. Zum Beitrag von K. Hatakeyama, J. S. Petöfi und E. Sözer (Text, Konnexität, Kohäsion, Kohärenz)", in: Conte (ed.), 1987. Rickheit, Gert and Bock, Michael (eds.) 1983 Psycholinguistic Studies in Language Processing (= Research in Text Theory 7). Berlin: de Gruyter. Rieger, Burghard (ed.) 1984 Dynamik in der Bedeutungskonstitution (— Papers in Textlinguistics 46). Hamburg: Buske. Rieser, Hannes (ed.) 1982 Semantics of Fiction (= Poetics 11). Amsterdam: North-Holland. Rothkegel, Annely and Sandig, Barbara (eds.) 1984 Text — Textsorten — Semantik: Linguistische Modelle und maschinelle Verfahren (= Papers in Textlinguistics 52). Hamburg: Buske. Sözer, Emel (ed.) 1985 Text Connexity, Text Coherence: Aspects, Methods, Results (= Papers in Textlinguistics 49). Hamburg: Buske.

470

J. S. Petöfi, E. Sözer

Appendix

(TA) OOl

(1)

C1') (3b)

Bonjour, Good morning N/m. sing.

(3c) (3d) (4a) (4b) (4c) (4d)

OBJ SE [0, P, S] •""Bonjour,

0)

prince. prince. N/m. sing.

(!') (3b)

(3c) (3d) (4a) (4b) (4c) (4d)

C'02

-

petit

the

little

V/tr. ind. pres. sing. 3° PRED

Art/def. m. sing.

SUBJ

dit

le

Adj/m. sing.

petit

103

prince."11·

Bonjour,

0')

Good morning N/m. sing.

(3c) (3d) (4a) (4b) (4c) (4d)

le

GREETING

(1)

(3b)

dit /dire/ says

OBJ SE [0, P, S] •""Bonjour,

dit /dire/ says

le

marchand.

the

merchant.

V/tr. ind. pres. sing. 3° PRED

Art/def. m. sing.

N/m. sing.

dit

le TH/sell

SUBJ

GREETING 104

marchand"11·.

Static and Dynamic Aspects of Text Constitution O03

(1)

C

(!') (3b)

This Pron/dem. m. sing.

(3c)

SUBJ1

un

marchand

de

pilules

a Art/indef. m. sing-

merchant N/m. sing-

of Prep

pills N/f. plur.

(3d) (4 a) (4b) (4c) (4d)

PRED1.COMP NP SE [S, P, P.C [NP, NP.M [NP SE.R [S, P, O]]]] Jr C' etait un marchand TH/sell DESCRIPTION 104

(1)

perfectionnees

qui

(!') (3b)

invented Ad}/f. plur.

that Pron/rel -

(3c) (3d) (4a) (4b) (4c) (4d)

O04

etait /etre/ was V/intr. ind. imp. sing. 3° PRED1

perfectionnees

SUBJ2

apaisent /apaiser/ quench V/tr. ind. pres. plur. 3° PRED2

OBJ 2

qui

apaisent

la

(1)

On

en

(!')

one

[of these]

(3b)

Pron/indef.

Pron/adv.

(3c) (3d) (4 a) (4b) (4c) (4d)

471

la

soif.

the Art/def. f. sing.

thirst. N/f. sing.

avale /avaler/ swallows

NP.MOD de

pilules

105

soifL. TH/thirst

une

par

one a [piece] Adj/num. Prep f. sing-

semaine week

V/N/ — tr. f. sing. ind. sing. 3° pres. sing. 3° SUBJ1 OBJ.ADV PRED1 OBJ1 ADV.TEMP SE [SE [S, O.A, P, O, A.T] CONJ SE [S, -|P, O]] J "On en avale une par semaine TH/count TH/time DESCRIPTION 106 105 105

472

J. S. Pet fi, E. S zer Γόη

and

one

not

(3b)

Conj

Pron/indef. — sing. 3°

Adv/neg.

(3c) (3d) (4a) (4b) (4c) (4d)

CONJ

SUBJ2

NEG

et

Γόη

O) (O

(3b) (3c) (3d) (4a) (4b) (4c) (4d)

O05

eprouve /eprouver/ feel

et

0)

(1)

V/tr. ind. pres. sing. 3° PRED2 eprouve

le

any longer Adv

the

Art/def. m. sing.

OBJ 2

plus

le

106

besoin need N/m. sing.

de to Prep

boire. drink. V/inf.

besoin

de

boire.""· TH/thirst

—

Pourquoi

vends/vendre/ sell V/— tr. ind. pres. sing. 2°

tu

ga?

you Pron/pers. — sing. 2°

that? Pron/dem. —

(!') (3b)

Why Adv/interrog.

(3c)

OBJ PRED SUBJ INTERROG SE [O [SE [I, P , S, 0]], P, S] vendstu •""Pourquoi TH/sell QUESTION 104

(3d) (4a) (4b) (4c) (4d)

plus

OBJ ja?

105

Static and Dynamic Aspects of Text Constitution

0) (!') (3b)

(3c) (3d) (4a) (4b) (4c) (4d)

C'06

(1)

dit /dire/ says V/tr. ind. pres. sing.

prince.

the Art/def. m. sing

little Adj/m. sing.

prince. N/m. sing.

petit

prince.""·

SUBJ

dit

le

103

-

0')

(3d) (4a) (4b) (4c) (4d)

petit

3° PRED

C

This Pron/dem. —

(3b)

(3c)

le

une

grosse

economic

a Art/indef. f. sing.

tremendous Adj/f. sing.

saving N/f. sing.

grosse

economic TH/saving

OBJ PRED.COMP SUB} SE [O [SE [S, I5, P-C]], P, S] Jr est une C

ANSWER I05(?)

(1)

de

temps,

(!') (3b)

of Prep

time, N/m.

(3c) (3d) (4a) (4b) (4c) (4d)

est /etre/ is V/intr. ind. pres. sing. 3° PRED

de

temps, TH/time

dit /dire/ says V/tr. ind.

le

marchand.

the Art/def. m. sing.

merchant. N/m. sing.

pres. 3° PRED

SUBJ

dit

le

104

marchand."11· TH/sell

473

474 007

J. S. Petöfi, E. Sözer experts

(1)

Les

(!') (3b)

The Art/def.

plur.

(3c) (3d) (4a) (4b) (4c) (4d)

SUBJ SE [S, P, 0] Jr Les

(1) (!') (3b)

calcules. computations. N/m. plur.

(3d) (4a) (4b) (4c) (4d) C'08

experts

fait

ont

OBJ des

108

calcules.""· TH/count

On

( ) (3b)

One Pron/indef. — sing. 3°

(3c) (3d) (4a) (4b) (4c) (4d)

SUBJ SE [S, P, 0, A.T] Jr On

0)

par a Prep

(3c) (3d) (4a) (4b) (4c) (4d)

fait des /faire/ made VIArt/indef. — tr. part. plur. pass.

ANSWER.CONT 107

(1)

(3b)

ont /avoir/ experts have N/V/m. tr. plur. ind. pres. plur. 3° PRED

ANSWER.CONT 106 semaine. week. N/f. sing. ADV. TEMP par

semaine.""· TH/time

epargne /epargner/ saves V/tr. ind. pres. sing. 3° PRED

cinquante-trois

minutes

fifty-three Adj/num.

minutes N/f. plur.

epargne TH/saving

cinquante-trois TH/count

OBJ

109

minutes TH/time

Static and Dynamic Aspects of Text Constitution

C'09

(1)

- Et

fait/faire/ does what And Pron/interrog. V/Conj tr. ind. pres. sing. 3° INTERROG PRED CONJ SE [CONJ, I, P, S, O.P] Jr Et que fait-

(3b)

(3c) (3d) (4 a) (4b) (4c) (4d)

(1) (Γ) (3b)

que

010

(1)

(!')

(3b)

(3c)

(3d) (4 a) (4b) (4c) (4d)

de

one Pron/indef. — sing. 3°

with Prep

SUBJ

OBJ.

on

de

QUESTION 106

ces cinquante-trois these fifty-three Pron/dem. Adj/num. plur.

(3c) (3d) (4a) (4b) (4c) (4d)

on

ces

cinquante-trois TH/count

minutes? minutes? N/ — f. plur. minutes?"11· TH/time

109

—

ce

que

One

fait /faire/ does

it

what

Pron/dem.

Pron/rel.

SUBJ

V/tr. ind. pres. sing. 3° PRED

OBJ NP

OBJ

On

en

with those Pron/indef. Pron/adv. — sing. 3°

OBJ.ADV

SE [S, O.A, P, O [NP, SE.R [O, S, P]]] Jr On en fait ce

ANSWER 106

109

que

475

J. S. Pet fi, E. S zer

476

Oll

(υ

1 on

(!') (3b)

one Pron/indef. — sing. 3°

(3c) (3d) (4a) (4b) (4c) (4d)

SUBJ

veut . . . /vouloir/ will ... V/tr. ind. pres. sing. 3° PRED

Γόη

veut ...""·

(1)

0') (3b)

106

se

As for me Pron/pers.

to himself Pron/pers. VIrefl. tr. ind. pres. sing. 3° PRED

sing. 1°

(3c) (3d) (4a) (4b) (4c) (4d)

the

little

Art/def. m. sing.

103

103

j'

0')

prince, N/— m. sing.

if Conj

I Pron/pers,. — sing. 1°

avais /avoir/ had VItr. ind. imp. sing. 1°

Adj/m. sing.

SUBJ

COMMENTARY 103 si

(3d) (4a) (4b) (4c) (4d)

petit

le

prince,

(3c)

le

OBJSUBJ SE [O [SE- [S]] - P, S, - 0 [-SE[ Jr dit Moi, se

(1)

(3b)

dit /dire/ says

Moi,

petit

cinquante-trois fifty-three Adj/num.

-OBJ CONJ SUBJ 1 PRED1 OBJ1 [SE [CONJ, SE [S, P, O], SE [S, P, A.M . A.L]]]]] prince, si j' avais cinquante-trois TH/count 103

109

Static and Dynamic Aspects of Text Constitution (1)

minutes

a

depenser,

je

') (3b)

minutes N/f. plur.

to Prep

spend

V/-

I Pron/pers. — sing. 1°

(3c) (3d) (4a) (4b) (4c) (4d)

minutes ä TH/time

(1) 0') (3b)

tout very Adv/-

(3c) (3d) (4a) (4b) (4c) (4d)

ADV.MANNER

tr. inf.

depenser,

marcherais /marcher/ should walk

VI-

SUBJ2

intr. cond. pres. sing. 1° PRED2

)e

marcherais

103

tout

doucement gently Adv/-

vers toward Prep

une a Art/indef. f. sing.

fontaine spring N/f. sing.

une

fontaine"11· TH/thirst

ADV.LOC doucement

vers

110

477

PATRICIA LOHMANN

Connectedness of Texts: A Bibliographical Survey The present bibliographical survey lists about 350 articles and monographs in English, French, and German on the topic 'connectedness of texts'. It has been compiled along the following lines: First the literature on connectedness has been globally surveyed and a list of the most frequently used terms concerning connectedness has been set up. This list consists of the following terms: 'coherence', 'cohesion', 'composition', 'connectedness'/'connexity', 'constitution', 'construction', 'isotopy', 'macrostructure', 'paragraph'. The next step was to consult introductions to text linguistics, the Bibliographie de Linguistique de l'Annee (1976 ff.), the Language and Language Behavior Abstracts (1970ff.), and the Bibliographies of Tannacito and Thorndyke (as to their bibliographical data cf. the respective entries in this Survey), and to select from them those items which contain the above terms (or terms which are more or less paraphrases of them) in their titles. Finally, the reference lists of the works bibliographically registered in this way were consulted as far as they were available. As a consequence of the procedure outlined above, introductions to text linguistics have not been included in this Survey, though all of them treat the topic 'connectedness'. Nor have articles or monographs been included which deal with, for example, anaphora as a text-constitutive means, unless one of the above terms occurred in their title, although these works are also essential with respect to text connectedness. Despite this deficiency, we hope that this Survey provides a useful first orientation for those who intend to deal with this topic more thoroughly. (For text-linguistic/textological research in general, cf. Petöfi, Janos S. "Text, Discourse", to appear in Encyclopedic Dictionary of Semiotics, ed. by Th. A. Sebeok, Berlin: W. de Gruyter, and "European Research in Semiotic Textology: A Historical, Thematic, and Bibliographical Guide", to appear in Folia Linguistica; as to the topic 'text connectedness', cf. the detailed survey: Charolles, M.; Petöfi, J. S.; Sözer, E. (eds.): Research in Text Connexity and Text Coherence: A Survey (= Papers in Textlinguistics 53). Hamburg: Buske, 1986). To facilitate the use of this Survey, a Terminological Index has been added in which the titles included in the Survey have been grouped according to the terms mentioned above, however, only the name of the authors/editors

Connectedness of Texts: A Bibliographical Survey

479

and the year of publication are indicated. (In the case of more than two authors/editors, the name of the first author/editor and the addition et al. has been used. It has to be emphasized that this Index is not thematic, but an entirely terminological index, since one and the same topic is often indicated by different terms and one and the same term may refer to different topics. Nevertheless, this index also supplies partial thematic information (by classing, for example, works on the topic 'macrostructure' and 'paragraph' in different groups). Adam, Jean-Michel 1978 "La cohesion des sequences de propositions dans la macrostructure narrative", Langue franfaise 38, 101 — 117. 1982 "The Macro-Structure of the Conventional Narrative", Poetics Today 3, 135-169. Adriaens, Mark 1979 "Isotopic Organization and Narrative Grammar", PTL: A Journal for Descriptive Poetics and Theory of Literature 4, 501 —544. 1975 An Isotopic Approach to Texts. Leuven: K.U.L., PhD Diss., mimeo. Agar, Michael and Hobbs, Jerry R. 1982 "Interpreting Discourse: Coherence and the Analysis of Ethnographic Interviews", Discourse Processes 5, 1—32. Agricola, Erhard 1975 Semantische Relationen im Text und im System (3rd revised edition). Halle/Saale: Niemeyer. 1st edition: Linguistische Studien 1969. 2nd edition: Janua linguarum, series minor 113. The Hague: Mouton. 1981 "Erkundungen zur Makrostruktur narrativer Texte", in Dane§, F. and Viehweger, D. (eds.), 21-39. Ajdukiewicz, Kazimierz 1935 "Die syntaktische Konnexität", Studio philosophica 1, 1—27. Andersson, Erik 1978 "An Analysis of Textual Cohesion in a Passage from Maria Gripe's 'Hugo och Josefin'", in Östman, J.-O. (ed.), 131-143. Arrive, Michel 1973 "Pour une theorie des textes poly-isotopiques", Langages 31, 53 — 63. Ballard, D. Lee; Conrad, Robert J. and Longacre, Robert E. 1971 a "The Deep and Surface Grammar of Interclausal Relations", Foundations of Language 7, 70 — 118, reprinted in Brend, R. M. (ed.): Advances in Tagmemics. Amsterdam: North-Holland, 307-355. 1971b More on the Deep and Surface Grammar of Interclausal Relations (= Language Data, Asian Pacific Series, 1). Ukarumpa, Papua, N.-G.: Summer Institute of Linguistic Publications. Ballmer, Thomas T 1976 "Macrostructures", in van Dijk, T. A. (ed.): Pragmatics of language and Literature (= North-Holland Studies in Theoretical Poetics 2). Amsterdam: North-Holland, 1-22. Banfield, Ann 1978 "The Formal Coherence of Represented Speech and Thought", PTL: A Journal for Descriptive Poetics and Theory of Literature 3, 289 — 314. Baron, Dennis E. 1976 Review of: Uspenskij, Boris (1973): A Poetics of Composition: the Structure of the Artistic Text and Typology of a Compositional Form. Berkeley: University of California Press, in: Style 10, 274-277.

480

P. Lohmann

de Beaugrande, Robert A. 1978 "Linguistic Theory and Composition", College Composition and Communication 19, 134-140. 1982a Text Production: Toward a Science of Composition. Norwood: Ablex. 1982b "Psychology and Composition: Past, Present, and Future", in Nystrand, M. (ed.), 211-268. Becker, Alton L. 1965 "A Tagmemic Approach to Paragraph Analysis", College Composition and Communication 16, 237-242. 1966 "Symposium on the Paragraph", College Composition and Communication 17, 67 — 72. 1980 "Text-Building, Epistemology and Aesthetics in Javanese Shadow Theatre", Dispositio: Revista Hispanica de Semiötica Literaria 5, 137 — 168. Beekman, John 1970 "Propositions and their Relations within a Discourse", Notes on Translation 37, 6-23. Beene, LynnDianne 1981 The Pragmatics of Cohesion. PhD Thesis, University of Kansas. Bellert, Irena 1970 "On a Condition of the Coherence of Texts", Semiotica 2, 335-363. Berrendonner, Alain 1976 "De quelques aspects logiques de l'isotopie", L'isotopie: Linguistique et Semiologie 1. Lyon: Universite Lyon II, 117 — 135. Besmertnaja, Ninel 1976 "Bericht, Beschreibung, Erörterung als grundlegende Formen der Komposition von Texten", Sprachpflege 25, 7 — 10. Biedermann, Johann 1976 "Ausdrucksformen und Funktionen von Satzkonnexionen in kohärenten Texteinheiten", in H. Jelitte (ed.): Sowjetrussische Textlinguistik Teil 1: Themen und Methoden ( = Beiträge zur Slavistik 1). Frankfurt am Main: Lang, 57 — 80. Binkley, Marilyn Rothman 1983 A Descriptive and Comparative Study of Cohesive Structure in Text Materials from Differing Academic Disciplines. Ed. Dissertation, George Washington University. Black, John B. and Bern, Hyman 1981 "Causal Coherence and Memory for Events in Narratives", Journal of Verbal Learning and Verbal Behavior 20, 267-275. Blei, Dagmar and Pfütze, Max 1975 "Text - Komposition - Emotionalität", Sprachpflege 10, 197-200. Bonnafous, Hildegard 1981 "Zur Frage der Textkohärenz in Texten verschiedener sprachlicher Tätigkeitsbereiche (Wissenschaft, Presse — Publizistik und Belletristik, Teil II), Wissenschaftliche Zeitschrift der Martin-Luther-Universität Halle-Wittenberg, Gesellschafts-\Sprachwissenschaffliehe Reihe 30, 67 — 76. Borel, Marie-Jeanne and Mieville, Denis 1983 Review of Enseignement du recit et coherence du texte, eds. Michel Charolles and Jean Peytard, in Neubauer, F. (ed.), 145-150. Brazil, David 1983 "Intonation and Connectedness in Discourse", in Ehlich, K. and van Riemsdijk, H. (eds.), 179-198. Bremond, Claude 1972 "Le 'modele constitutionnel' de A. J. Greimas", Semiotica 5, 362—382. Brinkmann, Hennig 1965 "Die Konstituierung der Rede", Wirkendes Wort 15, 157-172.

Connectedness of Texts: A Bibliographical Survey

481

Britton, James 1978 "The Composing Processes and the Functions of Writing", in Cooper, C. R. and Odell, L. (eds.), 13-28. Bunge, Mario and Garcia Sucre, Maximo 1976 "Differentiation, Participation and Cohesion"', Quality and Quantity 10, 171 — 178. Callow, Kathleen 1970 "More on Propositions and Their Relations Within a Discourse", Notes on Translation 37, 23-27. Chaplen, Frank 1970 Paragraph Writing. London: Oxford University Press. Charolles, Michel 1978 "Introduction aux problemes de la coherence des textes: Approche theorique et etude des pratiques pedagogiques", Langue frattfaise 38, 7—41. 1982 Etudes sur la coherence et l'interpretation des discours. These de doctoral d'etat, Besan5on: Universite de Franche-Comte. 1983a "Towards a Heuristic Approach to Text-Coherence Problems", in Neubauer, F. (ed.), 1-16. 1983b "Coherence as a Principle in the Interpretation of Discourse", Text 3, 1, 71—97. Charolles, Michel and Peytard, Jean (eds.): 1978 "Enseignement du recit et coherence du texte", Langue franfaise 38. Chelala, Silvia Ines 1981 The Composing-Process of two Spanish-Speakers and the Coherence of Their Texts: a Case Study. PhD Thesis, New York University. Cherchi, Lucien 1978 "L'ellipse comme facteur de coherence", Langue franfaise 38, 118—128. Christensen, Francis 1965 "A Generative Rhetoric of the Paragraph", College Composition and Communication 16, 144—156. Revised 1966, reprinted in Netv Rhetorics, ed. by M. Steinmann. New York: Scribner, 1967. 1966 "Symposium on the Paragraph", College Composition and Communication 17, 60—66. 1967 "A Generative Rhetoric of the Paragraph", in Towards a New Rhetoric, ed. Francis Christensen. New York: Harper & Row, 52 — 81. Christie, Daniel J. and Schumacher, Gary M. 1975 "Development Trends in the Abstraction and Recall of Relevant Versus Irrelevant Thematic Information from Connected Verbal Materials", Child Development 46, 598-602. Cinque, Guglielmo 1983 "'Topic' Constructions in some European Languages and 'Connectedness'", in Ehlich, K. and van Riemsdijk, H. (eds.), 7-42. Clark, Cheryl 1983 Cohesion in Spoken and Written English. PhD Thesis, Harvard University. Cohan, Carol 1976 "Writing Effective Paragraphs", College Composition and Communication 4, 363 — 365. Cohen, Ronald L. and Johansson, Bo S. 1967 "Note on the Retention of Connected Discourse", Scandinavian Journal of Psychology 8, 11-16. Cooper, Charles R. and Odell, Lee 1978 Research in Composition: Points of Departure. Urbana, 111.: National Council of Teachers of English. Correira, Alfred 1980 "Computing Story Trees", American Journal of Computational Linguistics 6, 135 — 149. Crothers, Edward J. 1978 "Inference and Coherence", Discourse Processes 1, 51 — 71. 1979 Paragraph Structure Inference. Norwood, N. J.: Ablex.

482

P. Lohmann

Crowell, Thomas H. 1973 "Cohesion in Bororo Discourse", Linguistics 104, 15-27. Cunningham, Donald J. 1972 "The Retention of Connected Discourse: a Review", Review of Educational Research 41, 47-72. Daneä, Frantisek 1974 "Functional Sentence Perspective and the Organization of Text, in Danes, F. (ed.): Papers in Functional Sentence Perspective (= Janua Linguarum, Series Minor, 147). The Hague: Mouton, 106-128. 1981 "Eine Bemerkung zur Intonation im Textaufbau", in Daneä, F. and Viehweger, D. (eds.), 88-96. DaneS, Frantiäek and Viehweger, Dieter (eds.) 1981 Sat^semantiscbe Komponenten und Relationen im Text (— Linguistica I). Praha: Ceskoslovenska akademie ved. Dascal, Marcelo and Katriel, Tamar 1979 "Disgressions: A Study in Conversational Coherence", PTL: A Journal for Descriptive Poetics and Theory of Literature 4, 203 — 232 and in Petöfi, J. S. (ed.): Text vs. Sentence Continued ( = Papiere zur Textlinguistik 29). Hamburg: Buske (1982), 76 — 95. Davis, Donald R. 1973 "Wantoat Paragraph Structure", Linguistics 110, 5 — 16. Deese, James 1961 "From the Isolated Verbal Unit to Connected Discourse", in Cofer, C. (ed.): Verbal Learning and Verbal Behavior. New York: McGraw-Hill, 11—31. Delahaye, Ives 1983 "L'ombre et le reflet: la double symetrie dans l'organisation du texte", in Borbe, T. (ed.): Semiotics Unfolding: Proceedings of the 2. Congress of the International Association of Semiotic Studies, Vienna, July 1979. Vol. II. Berlin: Mouton, 791—794. DeVilliers, Peter A. 1974 "Imagery and Theme in Recall of Connected Discourse", Journal of Experimental Psychology 103, 263-268. van Dijk, Teun A. 1973 "A Note on Linguistic Macro-Structures", in Linguistische Perspektiven: Referate des VII. Linguistischen Kolloquiums, Nijmegen 26. — 30. September 1972. Tübingen: Niemeyer, 75-87. 1976a "Narrative Macrostructures: Logical and Cognitive Foundations", PTL: A Journal for Descriptive Poetics and Theory of Literature 1, 547 — 568. 1976b "Pragmatic Macrostructures and Cognition', in: The Cognitive Viewpoint. Gent: University of Amsterdam. 1977a "Macro-Structures and Cognition", in Just, M. A. and Carpenter, P. A. (eds.): Cognitive Processes in Comprehension. Hillsdale: Erlbaum. 1977b "Semantic Macrostructures and Knowledge Frames in Discourse Comprehension", in Just, M. A. and Carpenter, P. A. (eds.): Cognitive Processes in Comprehension. Hillsdale: Erlbaum, 3-22. 1980 Macrostructures. Hillsdale, N. J.: Erlbaum. Dillon, George L. 1983 Constructing Texts: Elements of a Theory of Composition and Style. Bloomington: Indiana University Press. DiVesta, Francis J.; Schultz, Charles B. and Dangel, Timothy R. 1973 "Passage Organization and Imposed Learning Strategies in Comprehension and Recall of Connected Discourse", Memory and Cognition 1, 471—476. Dolezel, Lubomir 1973 "Narrative Composition — a Link between German and Russian Poetics", in Bann, S. and Bowlt, J, E. (eds.): Russian Formalism. Edinburgh: Edinburgh University Press, 73-84.

Connectedness of Texts: A Bibliographical Survey

483

Dorfmüller-Karpusa, Käthi 1981 "Kompositionseinheit — Kommunikat — kanonische Repräsentation" in Dorfmüller-Karpusa, K. and Petöfi, J. S. (eds.): Text, Kontext, Interpretation: Einige Aspekte der texttheoretischen Forschung (= Papiere zur Textlinguistik 35). Hamburg: Buske, 73-109. Dressler, Wolfgang U. 1983 "Textuelle Kohäsionsverfahren in der Wissenschaftssprache — eine funktioneile Ableitung", Fachsprache 5, 51—58. Ehlich, Konrad and van Riemsdijk, Henk (eds.): 1983 Connectedness in Sentence, Discourse and Text: Proceedings of the Tilburg Conference held on 25 and 26 January 1982 (= Tilburg Studies in Language and Literature 4). Tilburg. Emig, Janet 1978 Review of The Philosophy of Composition by E. D. Hirsch, Jr., Rhetoric Society Quarterly 8, 145-148. Engel-Ortlieb, Dorothea 1983 "Textkohärenz: Erzählung, Beschreibung, Bericht in der Textreproduktion", in Faust, M., Harweg, R., Lehfeldt, W. and Wienold, G. (eds.): Allgemeine Sprachwissenschaft, Sprachtypologie und Textlinguistik: Festschrift für Peter Hartmann %um 60. Geburtstag. Tübingen: Narr. Enkvist, Nils Erik 1978 "Coherence, Pseudo-Coherence, and Non-Coherence", in Östman, J.-O. (ed.), 109-128. Ezard, Bryan 1978 "Insights in Cohesion from Tawala", Oceanic Linguistics 17, 107—132. Fairley, Irene R. 1973 "Syntactic Deviation and Cohesion", Language and Style 6, 216 — 229. Ferry, Clifford Lloyd 1975 Discourse Coherence and Readability: A Study of the Effect of Coherence Marker Density on Reading Comprehension. PhD Thesis, University of Oregon. Fillol, Francois and Mouchon, Jean 1978 "Approche des notions de coherence et de cohesion sur un corpus oral", Langue fraacaise 38, 87-100. Fine, Jonathan H. 1978 "Conversation, Cohesive and Thematic Patterning in Children's Dialogues", Discourse Processes 1, 247—266. Fine, Jonathan H. and Bartolucci, Giampiero 1981 "Cohesion and Retrieval Categories in Normal and Disturbed Communication: a Methodological Note", Discourse Processes 4, 276—270. Finn, Patrick J. 1978 "Can Rules be Devised to Make Implicit Inter-Sentence Case Relationships Explicit?", Southwest Regional Laboratory Professional Papers 40, 35 — 46. Fishman, Anne Stevens 1977 Elements of Cohesion: the Effect of Noun Phrase Organisers and Anaphoric References on Paragraph Comprehension. PhD Thesis, Northwestern University (Illinois). Fosberg, Mary Dee Harris 1979 "Dylan Thomas, the Craftsman: Computer Analysis of the Composition of a Poem", Association for Literary and Linguistic Computing Bulletin 7, 295 — 300. Fowler, Roger 1977 "Cohesive, Progressive, and Localizing Aspects of Text Structure", in van Dijk, T. A. and Petöfi, J. S. (eds.): Grammars and Descriptions: Studies in Text Theory and Text Analysis. Berlin: de Gruyter, 64—84.

484

P. Lohmann

Fräse, Lawrence T. 1968 "Some Unpredicted Effects of Different Questions upon Learning from Connected Discourse", Journal of Educational Psychology 59, 197—201. 1969 "Cybernetic Control of Memory while Reading Connected Discourse", Journal of Educational Psychology 60, 49 — 55. 1969 "Paragraph Organization of Written Materials: The Influence of Conceptual Clustering upon Level of Organization", Journal of Educational Psychology 60, 394—401. Fritz, Gerd 1982 Kohären^: Grundfragen der linguistischen Kommunikationsanalyse. Tübingen: Narr. Gerdel, Florence E. and Slocum, Marianna 1976 "Paez Paragraph and Sentence Structure", in Longacre, R. E. and Woods, F. (eds.): Discourse Grammar: Studies in Indigenous Languages of Colombia, Panama, and Ecuador. Part 1 (= Summer Institute of Linguistics Publications 52). Dallas: SIL, 259—443. Giora, Rachel 1983 "Segment Cohesion: the Thematic Organization of the Text", Text 3,155—181. Gordon, Barbara Lynn 1982 An Investigation of Relations among Grammatical Subjects, Initial Sentence Elements, and Paragraph Structure in Written Expository Discourse. PhDThesis, State University of New York. Grimes, Joseph E. 1966 "Some Inter-Sentence Relationships in Huichol", in Summa Antropolögica en homenaje a Roberto J. Weitlaner. Mexico, D. F.: Instituto Nacional de Antropologia e Historia, 465-470. 1975 The Thread of Discourse. The Hague: Mouton. "Group " 1974 "Lecture du poeme et isotopies multiples", Lefranfais moderne 42, 217 — 236. Grzyb, Georg 1974 Rezension zu O spojnosci texstu [Über Textkohärenz]. Sammelband unter der Redaktion von Maria R. Mayenowa, Wroclaw, 1971", in Projektgruppe Textlinguistik Konstanz (eds.): Probleme und Perspektiven der neueren textgrammatischen Forschung I (= Papiere zur Textlinguistik 5). Hamburg: Buske, 165 — 177. Gülich, Elisabeth 1970 Makrosyntax der Gliederungssignale im gesprochenen Französisch. München: Fink. Gülich, Elisabeth; Heger, Klaus and Raible, Wolfgang 1974 Linguistische Textanalyse: Überlegungen ^ur Gliederung von Texten (= Papiere zur Textlinguistik 8). Hamburg: Buske. Gülich, Elisabeth and Raible, Wolfgang 1974 "Überlegungen zu einer makrostrukturellen Textanalyse: J. Thurber, 'The Lover and his Lass'", in Gülich, E., Heger, K. and Raible, W. (eds.): Textanalyse. Hamburg: Buske, 73 — 126 and in van Dijk, T. A. and Petöfi, J. S. (eds.): Grammars and Descriptions: Studies in Text Theory and Text Analysis. Berlin: de Gruyter, 1977, 132-175. Gutwinski, Waldemar 1976 Cohesion in Literary Texts: a Study of Some Grammatical and Lexical Features of English Discourse (= Janua Linguarum Series Minor 204). The Hague: Mouton. van Haaften, Ton; Smits, Rik and Vat, Jan 1983 "Left Dislocation, Connectedness and Reconstruction", in: Ehlich, K. and van Riemsdijk, H. (eds.), 43-70. Haberlandt, Karl and Bingham, Geoffrey 1978 "Verbs Contribute to the Coherence of Brief Narratives: Reading Related and Unrelated Sentence Triples", Journal of Verbal Learning and Verbal Behavior 17, 419-425.

Connectedness of Texts: A Bibliographical Survey

485

Hagerup-Neilsen, Allan Roy 1977 The Role of Macrestructure s and Linguistic Connectives in Comprehending Familiar and Unfamiliar Written Discourse. PhD Thesis, University of Minnesota. Halliday, Michael A. K. and Hasan, Ruqaiya 1976 Cohesion in English (= English Language Series 9). London: Longman. Hartnett, Carolyn Green 1980 Cohesion as a Teachable Measure of Writing Competence. PhD Thesis, Indiana University of Pennsylvania. Harweg, Roland 1968 Pronomina und Textkonstitution. München: Fink. 1971a "Grammatik und Textkonstitution als sprachsoziologische Probleme", in Sprache und Gesellschaft (Jahrbuch 1970 des Instituts für deutsche Sprache). Düsseldorf: Schwann, 63 — 79. 1971b "Quelques aspects de la constitution monologique et dialogique de textes", Semiotica 4, 127-148. 1977 "James Thurbers 'The Lover and His Lass' — textgrammatische Bemerkungen zur Konstitution eines literarischen Textes", in van Dijk, Teun A. and Petöfi, J. S, (eds.): Grammars and Descriptions: Studies in Text Theory and Text Analysis (= Research in Text Theory 1). Berlin: de Gruyter, 226-259. Hasan, Ruqaiya 1968 Grammatical Cohesion in Spoken and Written English, Part I (= Nuffield Programme in Linguistics and English Teaching Series 1, 7). London: Longman. Heidolph, Karl-Erich 1966 "Kontextbeziehungen zwischen Sätzen in einer generativen Grammatik", Kybernetika 2, 274-281. Hellwig, Peter 1984 "Grundzüge einer Theorie des Textzusammenhangs", in Rothkegel, A. and Sandig, B. (eds.): Text — Textsorten — Semantik: Linguistische Modelle und maschinelle Anwendung. Akten des Internationalen Kolloquiums, Saarbrücken, 18. —20. 11. 1982. Hamburg: Buske, 51-79. 1983 "Thesen zum Zusammenhang von Text- und Satzstruktur", in: Kühlwein, W. (ed.): Texte in Sprachwissenschaft, Sprachunterricht und Sprachtherapie: Kongreßberichte der 13. Jahrestagung der Gesellschaft für Angewandte Linguistik. Tübingen: Narr, 154—155. Hinds, John 1977 "Paragraph Structure and Pronominalization", Papers in Linguistics 10, 77—99. 1978 "Levels of Structure Within the Paragraph", in Jaeger, J. J., Woodbury, A. C., Ackermann, F. et al. (eds.): Proceedings of the Fourth Annual Meeting of the Berkeley Linguistics Society, 1978. Berkeley: Berkeley Linguistics Society, 598—609. 1979 "Organizational Patterns in Discourse", in: Givon, T. (ed.): Discourse and Syntax (= Syntax and Semantics 12). New York: Academic Press, 135 — 158. Hirsch, Eric Donald 1977 The Philosophy of Composition. Chicago: University of Chicago Press. Hobbs, Jerry R. 1977 "Coherence and Interpretation in English Texts", in Proceedings of the IJCAI, 1977, Cambridge, Mass., 110 — 116. 1979 "Coherence and Coreference", Cognitive Science 3, 67 — 90. 1983 "Why is Discourse Coherent?", in Neubauer, F. (ed.), 29-70. Hoffmannova, Jana 1981 "Zur Beziehung der Begriffe Textthema — Rahmen-Vertextungsmuster — Texthandlung/kommunikative Handlung/...", in DaneJ, F. and Viehweger, D. (eds.), 40-51.

486

P. Lohmann

Hohmann, Charles Frederick 1978 Assessing the Episodic Structure, Information Structure and Cohesion of Children's Written Narratives as Indices of Their Sophistication in Writing. PhD Thesis, University of Michigan. Hohulin, Richard 1971 "Cohesive Organization in Keley-i Kallahan", Pacific Linguistics, 1 — 17. Hooker, Betty 1972 "Cohesion in Ivatan", Asian Studies 10, 33-43. Howard, Linda 1977 "Certain Features of Verb Inflection as Related to Paragraph Types", in Longacre, R. E. and Woods, F. (eds.): Discourse Grammar: Studies in Indigenous Languages of Colombia, Panama, and Ecuador (= Summer Institute of Linguistics Publications 52), Dallas: SIL, 273-296. Huddleston, Rodney 1978 "On Classifying Anaphoric Relations: a Review of M. A. K. Halliday and R. Hasan, "Cohesion in English", Lingua 45, 333-354. Irmscher, William 1979 Teaching English Composition, New York: Holt, Rinehart and Winston. Jacobson, S. N. 1966 "A Modifiable Routine for Connecting Related Sentences of English Text", in Garvin, P. L. and Spolsky, B. (eds.): Computation in Linguistics. Bloomington: Indiana University Press, 284-311. Jakobsen, Lisbeth Falster and Olsen, J0rgen 1975 "Zum Problem der Textkohärenz", Text und Kontext 3, 3 — 48. 1978 "Textkohärenz und Involvierungen", Deutsche Sprache 6, 1—20. Jarvella, Robert J. 1970 "Effects of Syntax on Running Memory Span for Connected Discourse", Psychonomic Science 19, 235-236. 1971 "Syntactic Processing of Connected Speech", Journal of Verbal Learning and Verbal Behavior 10, 409-416. Jelitte, Herbert 1976a "Reduktion und Redundanz als Textkonstituenten", in Jelitte, H. (ed.), 173 — 204. 1976b "Die Substitution als Textkonstituente", in Jelitte, H. (ed.), 137 — 172. 1976c "Die Wiederholung als Textkonstituente", in Jelitte, H. (ed.), 115-136. 1978 Formen der Textkohären^ im Russischen: Eine Einführung in die Textstruktur. GrossenLinden: Hoffmann. Jelitte, Herbert (ed.) 1976 Soivjetrussische Textlinguistik ( = Beiträge zur Slavistik 1). Teil 1: Themen und Methoden, Teil 2: Übersetzte Originalbeiträge. Frankfurt am Main: Lang. Johns, Ann McClelland 1979 A Comparison of Cohesive Elements in American Business and Non-Native Speakers Written Discourse. PhD Thesis, University of Southern California. 1980 "Cohesion in Written Business Discourse: Some Contrasts", ESP Journal 1, 35—44. Jordan, Michael P. 1978 The Principal Semantics of the Nominals this and that in Contemporary English Writing. PhD Thesis, CNAA, Hatfield Polytechnic. Jussen, Heribert (ed.): 1976 Textkonstitution in einfacher Sprache und ihre Probleme. Heidelberg: Groos. Källgren, Gunnel 1979 "Some Types of Textual Cohesion and Their Effects on Texts" in Pettersson, T. (ed.): Papers from the Fifth Scandinavian Conference of Linguistics, Frostavallen, April 27-29, 1979. Stockholm: Almqvist & Wiksell, 135-146.

Connectedness of Texts: A Bibliographical Survey

487

Kallmeyer, Werner 1978 "Fokuswechsel und Fokussierungen als Aktivitäten der Gesprächskonstitution", in Meyer-Hermann, R. (ed.): Sprechen — Handeln — Interaktion: Ergebnisse aus Bielefelder Forschungsprojekten %ur Texttheorie, Sprechakttheorie und Konversationsanalyse (= Konzepte der Sprach- und Literaturwissenschaft 26). Tübingen: Niemeyer, 191—241. Kallmeyer, Werner and Schütze, Fritz 1977 "Zur Konstitution von Kommunikationsschemata der Sachverhaltsdarstellung", in Wegner, D. (ed.): Gesprächsanalysen. Hamburg: Buske, 159 — 274. Kamprad, Walter 1975 "Über die Struktur von Absätzen gesellschaftswissenschaftlicher Texte und zu Modellen für die Textarbeit", Deutsch als Fremdsprache 12, 223-234. Kantor, Robert Neal 1977 The Management and Comprehension of Discourse Connection by Pronoun in English. PhD Thesis, Ohio State University. Karlsen, Rolf 1959 Studies in the Connection of Clauses in Current English: Zero, Ellipsis and Explicit Form. Bergen: Eides Boktrykkeri. Kayne, Richard S. 1983 "Connectedness", Linguistic Inquiry 14, 223-250. Keenan, Elinor O. and Klein, Ewan 1975 "Coherency in Children's Discourse", journal of Psycholinguistic Research 4, 365 — 380. Kerbrat-Orecchioni, Catherine 1976 "Problematique de l'isotopie", Linguistique et semiologie 1, 11—33. Khatchadourian, Haig 1961 The Coherence Theory of Truth: A Critical Evaluation. Beirut: American University. Kieras, David E. 1978 "Good and Bad Structure in Simple Paragraphs: Effects on Apparent Theme, Reading Time, and Recall", Journal of Verbal Learning and Verbal Behavior 17, 13—28. 1981 "The Role of Major Referents and Sentence Topics in the Construction of Passage Macrostructure", Discourse Processes 4, 1 — 15. Kincaid, James R. 1977 "Coherent Readers, Incoherent Texts", Critical Inquiry 3, 781-802. King, David J. and Russell, Gordon W. 1966 "A Comparison of Rote and Meaningful Learning of Connected Meaningful Material", Journal of Verbal Learning and Verbal Behavior 5, 478 — 483. King, David J. 1968 "Retention of Connected Meaningful Material as a Function of Modes of Presentation and Recall", Journal of Experimental Psychology 77, 676—683. 1971 "Influence of Interitem Interval in the Learning of Connected Discourse", Journal of Experimental Psychology 87, 132—134. Kinneavy, James L. 1980 "The Relation of the Whole to the Part in Interpretation Theory and in the Composing Process", in McQuade, D. (ed.), 1—23. Kittredge, Richard 1981 "Cohesive Text Structure in Sublanguages", in Rieger, B. (ed.): Empirical Semantics: A Collection of New Approaches in the Field (= Quantitative Linguistics 12). Bochum: Brockmeyer, 446—466. Klammer, Thomas P. 1971 The Structure of Dialogue Paragraphs in Written English Dramatic and Narrative Discourse. Unpublished doctoral dissertation, University of Michigan. Klein, Klaus-Peter 1981 "Zur Konstitution von Erzählungen", in Hindelang, G. and Zillig, W. (eds.): Sprache: Verstehen und Handeln: Akten des 15. Linguistischen Kolloquiums, Münster 1980. Band 2. Tübingen: Niemeyer, 333—344.

488

P. Lohmann

Klein, Sheldon and Simmons, Robert F. 1963 "Syntactic Dependence and the Computer Generation of Coherent Discourse", Mechanical Translation 7, 50 — 61. Klein, Wolfgang 1978 Linguistik und Didaktik der Kindersprache im Grundschulalter: Untersuchungen %u Konstitution und Kohäsion von Schülertexten (= Informationen zur Sprache- und Literaturdidaktik 18). Paderborn: Schöningh. Kleine, William Michael 1983 Syntactic Choice and a Theory of Discourse: Rethinking Sentence Combining. PhD Thesis, University of Minnesota. Knapp, Donald S. 1967 Formal Factors Affecting Paragraph Division in Expository Writing. PhD Thesis, Columbia University. Koen, Frank; Becker, Alton L. and Young, Richard E. 1969 "The Psychological Reality of the Paragraph", Journal of Verbal Learning and Verbal Behavior 8, 49-53. Korpimies, Liisa 1978 "Some Aspects of Discourse and Cohesion", in Östman, J.-O. (ed.), 145 — 163. 1979 "Discourse Analysis and Cohesion in the Study of Drama", in Pettersson, T. (ed.): Papers from the Fifth Scandinavian Conference of Linguistics, Frostavallen, April 27—29, 1979. Stockholm: Almqvist & Wiksell, 147-158. Kraft, Eberhard; Nikolaus, Kurt and Quasthoff, Uta 1977 "Die Konstitution der konversationellen Erzählung", Folia Linguistica 11, 287 — 337. Krulee, Gilbert K.; Fairweather, Peter G. and Bergquist, Sidney R. 1979 "Organizing Factors in the Comprehension and Recall of Connected Discourse", Journal of Psycholinguistic Research 8, 141 — 163. Kulhavy, Raymond W.; Schmid, Richard F. and Walker, Carol H. 1977 "Temporal Organization in Prose", American Educational Research Journal 14, 115-123. Lachman, Roy and Dooling, D. James 1968 "Connected Discourse and Random Strings: Effects of the Number of Inputs on Recognition and Recall", Journal of Experimental Psychology 77, 517 — 522. Landman, Fred and Moerdijk, leke 1983 "Compositionality and the Analysis of Anaphora", Linguistics and Philosophy 6, 89-114. Langleben, Maria 1978 "On Verification of Text Coherence", Slavica Hierosolymitana 2, 155—168. 1981 "Latent Coherence, Contextual Meanings and the Interpretation of a Text", Text 1, 279-313. 1983 "An Approach to the Microcoherence of a Text", in Neubauer, F. (ed.), 71—98. Lautamatti, Liisa 1978 "Some Observations on Cohesion and Coherence in Simplified Texts", in Östman, J.-A. (ed.), 165-181. 1982 "Coherence in Spoken und Written Discourse", Nordic Journal of Linguistics 5, 117-128. Lehrer, Keith 1984 "Coherence, Consensus and Language", Linguistics and Philosophy 7, 43—55. Leuschner, Burkhard 1972 "Grundstrukturen des "Paragraphs" — ein Problem der Textgrammatik", Linguistische Berichte 21, 80-95. Levy, David Myron 1980 The Architecture of the Text. PhD Thesis, Stanford University California.

Connectedness of Texts: A Bibliographical Survey

489

Lewis, Ronald K. 1972 "Sanio-Hiowe Paragraph Structure", Papers in New Guinea Linguistics 15 (Pacific Linguistics Series A, 31), 1—9. Liebsch, Helmut 1973 "Zur kommunikativen Funktion sprachlicher Varianten in einem Vergleichstext: Eine sprachliche Detailanalyse als Beitrag zu textkonstitutiven Problemen des Ausdrucksunterrichts", Textlinguistik 3, 83 — 101. 1977 "Der Absatz als Element der Textkomposition", in Danes', F. and Viehweger, D. (eds.): Probleme der Textgrammatik II (= Studia grammatica 18). Berlin: AkademieVerlag, 147-212. Liebsch, Helmut and Pfriem, Renate 1975 "Zur Absatzgliederung von Texten", Textlinguistik 4, 88-111. Linnarud, Moira 1978 "Covert Errors in Cohesion", in Gregersen, K. (ed.): Papers from the Fourth Scandinavian Conference of Linguistics. Odense: Odense University Press, 125 — 129. Longacre, Robert E. (ed.) 1968 Discourse, Paragraph and Sentence Structure in Selected Philippine Languages (= Summer Institute of Linguistics Publications in Linguistics and Related Fields 21, 1 & 2). Dallas, Texas: SIL. Longacre, Robert E. 1970 "Paragraph and Sentence Structure in New Guinea Highlands Languages", Kivung 3, 150-163. Longacre, Robert E. (ed.) 1971 Philippine Discourse and Paragraph Studies in Memory of Betty McLachlin ( = Pacific Linguistics Series C, 22). Australian National University. 1972 Hierarchy and Universality of Discourse Constituents in New Guinea Languages (Discussion and Texts). Washington, D.C.: Georgetown University Press. 1979 "The Paragraph as a Grammatical Unit", in Givon, T. (ed.): Discourse and Syntax (= Syntax and Semantics 12). New York: Academic Press, 115 — 134. 1983 "Vertical Threads of Cohesion in Discourse", in Neubauer, F. (ed.), 99-114. Lonzi, Lidia 1983 "Types de cohesion du texte narratif, in: Borbe, T. (ed.): Semiotics Unfolding: Proceedings of the 2. Congress of the International Association for Sem to tic Studies, Vienna, July 1979. Berlin: Mouton, 915-923. Loriot, James and Hollenbach, Barbara 1970 "Shipibo Paragraph Structure", Foundations of Language 6, 43 — 66. Lundquist, Lita 1979 "From Text Coherence to Text Typology", in Pettersson, T. (ed.): Papers from the Fifth Scandinavian Conference of Linguistics, Frostavallen, April 27—29, 1979. Stockholm: Almqvist & Wiksell, 171-184. 1980 La coherence textuelle: syntaxe, semantique, pragmatique (= Erhvervssproglige skrifter 4). Kobenhavn: Nyt Nordisk Forlag. Maciejewski, Witold 1975 "On a Possibility of the Syntactic Description of the Intersentential Reference in a Literary Work", Biuletyn Fonografic^ny / Bulletin phonographique 16, 79 — 89. Mann, Renate 1976 "Textsorten: Aspekte der Textkonstitution", Die Neueren Sprachen 75, N.F. 25, 577-591. Mannhaupt, Hans-Rainer and Wirrer, Jan 1977 "Semantische Textstrukturen und Empirie: Bemerkungen zu B. F. Meyers Buch 'The Organisation of Prose and its Effect on Memory'", in Wirrer, J. (ed.): Textgrammatische Konzepte und Empirie: Probleme und Perspektiven der neueren textgrammatischen Forschung II ( = Papers in Textlinguistics 12). Hamburg: Buske, 50 — 108.

490

P. Lohmann

Marciszewski, Witold 1977 "Syntaktische Konnexität und Textkonnexität", in Daneä, F. and Viehweger, D. (eds.): Probleme der Textgrammatik II (= Studia grammatica 18). Berlin: Akademie Verlag, 173-180. Marcus, Solomon 1980 "Textual Cohesion and Textual Coherence", Revue Roumaine de Linguistique 25, 101-112. Marello, Carla 1979 "Text, Coherence and Lexicon", in Petöfi, J. S. (ed.), 618-633. Margolis, Joseph 1984 "The Locus of Coherence", Linguistics and Philosophy 7, 3 — 30. Marshall, Nancy and Clock, Marvin D. 1978 "Comprehension of Connected Discourse: a Study into the Relationships between the Structure of Text and Information Recalled". Reading Research Quarterly 14, 10-56. Mayenowa, Maria R. 1974 "Textkohärenz und Rezipientenhaltung", in Kroll, W. and Flaker, A. (eds.): Literaturtheoretische Modelle und kommunikatives System: Zur aktuellen Diskussion in der polnischen Literaturwissenschaft (= Skripten Literaturwissenschaft 4). Kronberg: Scriptor, 297-327. McCreedy, Lynn Ann 1983 Aspects of Reference, Cohesion, and Style in Three Genres of Navajo Texts (Southwest United States). PhD Thesis, Georgetown University. McCulley, George Allen 1983 Writing Quality, Coherence and Cohesion. Ed. Dissertation, Utah State University McCutchen, Deborah and Perfetti, Charles A. 1982 "Coherence and Connectedness in the Development of Discourse Production", Text 2, 113-140. McLeod, Ruth 1974 "Paragraph, Aspect, and Participant in Xavante", Linguistics 132, 51—74. McQuade, Donald (ed.) 1980 Linguistics, Stylistics, and the Teaching of Composition (= Language and Style 13, 3). Meade, Richard A. and Ellis, W. Geiger 1971 "The Use in Writing of Textbook Methods of Paragraph Development", Journal of Educational Research 65, 74—76. Mellon, John 1969 Transformational Sentence-Combining: A Method of Enhancing the Development of Syntactic Fluency in English Composition. Urbana: National Council of Teachers of English. Metzing, Dieter 1977 "Makrostrukturen als Element von T. A. van Dijks Textgrammatik-Programm", in Wirrer, J. (ed.): Textgrammatische Konzepte und Empirie (= Papiere zur Textlinguistik 12). Hamburg: Buske, 109-128. Meyer, Bonnie J. F. 1975 The Organisation of Prose and Its Effect on Memory. Amsterdam: North-Holland. Meyer, Bonnie J. F. and Rice, G. Elizabeth 1982 "The Interaction of Reader Strategies and the Organization of Text", Text 2, 155-192. Meyer, Paul Georg 1975 Sat^yerknüpfungsrelationen: Ein Interpretationsmodell für situationsabhängige Texte. Tübingen: Narr. Meyer, Robert K. 1971 "On Coherence in Modal Logics", Logique et analyse 55, 658 — 668.

Connectedness of Texts: A Bibliographical Survey

491

Meyers, Lawrence S. and Boldrick, Deborah 1975 "Memory for Meaningful Connected Discourse", Journal of Experimental Psychology, Human Learning and Memory 1, 584—591. Michel, Georg 1980 "Linguistische Aspekte der Komposition im künstlerischen Text", Zeitschrift für Germanistik 4, 430 — 446. Moles, Abraham A. 1979 "Le commentaire comme methode de composition litteraire et philosophique", Semiosis 4, 5 — 12. Monaghan, James 1979 "Some Proposals for a Co-Textual Approach to Coherence in English Texts", in Vandeweghe, W. and Van de Velde, M. (eds.): Bedeutung, Sprechakte und Texte: Akten des 13. Linguistischen Kolloquiums, Gent 1978 (= Linguistische Arbeiten 77). Tübingen: Niemeyer, 371—380. Mosenthal, Peter 1979 "Three Types of Schemata in Children's Recall of Cohesive and Non-Cohesive Text", Journal of Experimental Child Psychology 27, 129-142. Neubauer, Fritz (ed.): 1983 Coherence in Natural-Language Texts (= Papiere zur Textlinguistik 38). Hamburg: Buske. Neuner, Jerome Lawrence 1983 A Study of Cohesion in the Good and Poor Essays of College Freshmen. PhD Thesis, State University of New York at Buffalo. Nickel, Gerhard 1968 "Kontextuelle Beziehungen zwischen Sätzen im Englischen", Praxis des neusprachlichen Unterrichts 15, 15 — 25. 1970 "Some Contextual Relations between Sentences in English", in Graur, A. et al. (eds.): Proceedings of the Tenth International Congress of Linguists. Bucharest: Rumanian Academy of Sciences, 877 — 884. 1970 "Some Suprasegmental Criteria for Contextual Relations between Sentences", in Häla, B.; Romporti, M. and Janota, P. (eds.): Proceedings of the Sixth International Congress of Phonetic Sciences, Prague, 7.— 13. Sept. 1967. Prague: Academia, 671 —674. Nöth, Winfried 1977 "Textkohärenz und Schizophrenie", Zeitschrift für Literaturwissenschaft und Linguistik 23/24, 175-194. Nye, Irene 1978 "Satzverbindung, besonders bei Livius (1912)", in Dressler, W. U. (ed.): Textlinguistik (= Wege der Forschung 427). Darmstadt: Wissenschaftliche Buchgesellschaft, 15-23. Nystrand, Martin 1982 What Writers Know: The Language, Process, and Structure of Written Discourse. New York: Academic Press. Odmark, John 1981 "Textual Cohesion and Coherence and the Intrinsic Nature of the Signatum", in Kohrt, M. and Lenerz, J. (eds.): Sprache: Formen und Strukturen. Akten des 15. Linguistischen Kolloquiums, Münster 1980. Band 1. Tübingen: Niemeyer, 293—302. Östman, Jan-Ola 1978 "Introduction: Text, Cohesion, and Coherence", in Östman. J.-O. (ed.), 101 — 106. Östman, Jan-Ola (ed.): 1978 Cohesion and Semantics (= Meddelanden fran stiftelsens för Abo Akademi Forskningsinstitut 41). Abo: Publications of the Research Institute of the Abo Akademi Foundation. Paduceva, Elena V. 1974 "On the Structure of the Paragraph", Linguistics 131, 49-58.

492

P. Lohmann

Patten, Stanley Ray 1983 The Circle of Discourse: a Theory and Philosophy for Composition. PhD Thesis, Purdue University. Perkins, Kyle 1979 "Structure and Cohesion in Discourse", in Cooley, R. E., Barnes, M. R. and Dunn, J. A. (eds.): Papers of the 1978 Mid-America Linguistics Conference at Oklahoma. Norman: Dept. of Communication, University of Oklahoma, 303—306. Perlmutter, Jane, and Royer, James M. 1973 "Organization of Prose Materials: Stimulus, Storage, and Retrieval", Canadian Journal of Psychology 27, 200-209. Perry, Menakhem 1968 "The Inverted Poem: On a Principle of Semantic Composition in Bialik's Poems", ha-sifruf[, 607-631. Petöfi, Jänos S. 1981 "Texttheoretische Forschung, Aspekte der Textkohärenz, Sprachunterricht", in Dorfmüller-Karpusa, K. and Petöfi, J. S. (eds.): Text, Kontext, Interpretation: Einige Aspekte der texttheoretischen Forschung (= Papiere zur Textlinguistik 35). Hamburg: Buske, 235-285. Petöfi, Jänos S. (ed.) 1979 Text vs. Sentence: Basic Questions of Text Linguistics. Two volumes (Papers in Textlinguistics 20, 1 and 2). Hamburg: Buske. Petöfi, Jänos S. and Sözer, Emel (eds.) 1983 Micro and Macro Connexity of Texts ( = Papiere zur Textlinguistik 45). Hamburg: Buske. Peytard, Jean 1977 Lautreamont et la coherence de I'ecriture. Paris: Didier. Phillips, Brian 1975 "Judging the Coherence of Discourse", American Journal of Computational Linguistics 4, 36-49. 1978 "A Calculus of Cohesion", in Paradis, M. (ed.): The Fourth LACUS Forum, 1977. Columbia, S.C.: Hornbeam, 627—637. Pieper, Ursula 1976 "Zur Stellung einer phonetischen Komponente in der Textkonstitution", Folia Linguistica 9, 335 — 341. Polanyi, Livia 1976 "The Role of Redundancy in Cohesion and Evaluative Functioning in Narrative: a Grab for the Referential Hierarchy", Rackham Literary Studies 1 (Ann Arbor), 19-38. 1978 "On Meaning and Coherence Relations in Text", Semiotica 23, 343 — 358. Pompi, Kenneth F. and Lachman, Roy 1967 "Surrogate Processes in the Short-Term Retention of Connected Discourse", Journal of Experimental Psychology 75, 143—150. Pottier, Bernard 1977 "Micro-, meso- et macrosemantique dans l'analyse textuelle", Actes du XIV Congres international de linguistique et Philologie Romanes 1977, 4, 363 — 366. Prince, Gerald 1973 A Grammar of Stories (= De proprietatibus Litterarum, Series Minor, 13). The Hague: Mouton. Quasthoff, Uta 1976 "Makrostruktur und Gliederungsmerkmale in konversationellen Erzählungen: Gedanken zur Strukturbeschreibung von Texten", in Weber, H. and Weydt, H. (eds.): Sprachtheorie und Pragmatik: Akten des 10. Linguistischen Kolloquiums, Tübingen 1975. Tübingen: Niemeyer, 291—304.

Connectedness of Texts: A Bibliographical Survey

493

Rastier, Frangois 1972 "Systematique des isotopies", in Greimas, A. J. (ed.): Essais de semiotique poetique. Paris: Larousse, 80-105. Rath, Christel 1981 "Einige Überlegungen zur Analyse und Beschreibung von Textkohärenz", in Dorfmüller-Karpusa, K. and Petöfi, J. S. (eds.): Text, Kontext, Interpretation: Einige Aspekte der texttheoretischen Forschung (= Papiere zur Textlinguistik 35). Hamburg: Buske, 216-234. Reichman, Rachel 1978 "Conversational Coherency", Cognitive Science 2, 283 — 327. Reid, Lawrence A. 1970 Central Bontoc: Sentence, Paragraph and Discourse. Norman, Oklahoma: Summer Institute of Linguistics. Reinhart, Tanya 1980 "Conditions for Text Coherence", Poetics Today 1/4, 161-180. Rescher, Nicholas 1973 The Coherence Theory of Truth. Oxford: Clarendon. Rieger, Burghard 1977 "Bedeutungskonstitution: Bemerkungen zur semiotischen Problematik eines linguistischen Problems", Zeitschrift für Literaturwissenschaft und Linguistik (LiLi) 7, 55 — 68. Rodgers, Paul C, Jr. 1966a "A Discourse-Centered Rhetoric of the Paragraph", College Composition and Communication 17, 2 — 11. 1966b "Symposium on the Paragraph", College Composition and Communication 17, 72 — 80. Rosa, Alfred F. and Eschholz, Paul A. 1982 Models for Writers: Short Essays for Composition. New York: St. Martin's Press. Rosenberg, Sheldon 1966 "Associative Factors in the Recall of Connected Discourse", Psychonomic Science 4, 53-54. 1968a "Language Habits and the Recall of Connected Discourse", in Zale, E. (ed.): Proceedings of the Conference on Language and Language Behavior. New York: AppletonCentury-Crofts, 63 — 74. 1968b "Associative Factors in the Recall and Recognition of Nouns Embedded in Connected Discourse", Journal of Experimental Psychology 78, 254 — 260. Roy er, James M. and Cable, Glenn W. 1975 "Facilitated Learning in Connected Discourse", Journal of Educational Psychology 67, 116-123. Rudner, Richard S. 1978 "Show or Tell: Incoherence among Symbol Systems", Erkenntnis 12, 129 — 151. Ruhl, Charles E. 1973 "Prerequisites for a Linguistic Description of Coherence", Language Sciences 25, 15-18. Russell, Gordon W. and Sewell, Elizabeth C. 1972 "Serial Position Effect and Organization of Recall of Connected Meaningful Material", Psychological Reports 30, 443—446. Sachs, Jacqueline 1967 "Recognition Memory for Syntactic and Semantic Aspects of Connected Discourse", Perception and Psychophysics 2, 437—442. Saloni, Zygmunt and Trybulec, Andrzej 1974 "Coherence of a Text and Its Topology", Semiotica 11, 101 -108. Samet, Jerry and Schank, Roger 1984 "Coherence and Connectivity", Linguistics and Philosophy 7, 57 — 82.

49.4*

P. Lohmann

Sasson, Ralph 1971 "Semantic Organizations and Memory for Related Sentences", The American Journal of Psychology 84, 253-267. Schecker, Michael 1982 "Theorie des Vertextens", in Petöfi, J. S. (ed.) Text vs. Sentence Continued (= Papiere zur Textlinguistik 29). Hamburg: Buske, 191-227. Schenkein, Jim 1978 Studies in the Organisation of Conversational Interaction. New York: Academic Press. Schemer, Maximilian 1973 "Textkonstitution und -Rezeption: Zum Aufbau eines Textmodells für den Deutschunterricht", Der Deutschunterricht 25, 60 — 86. Schiffrin, Deborah 1982 Discourse Markers: Semantic Resource for the Construction of Conversation. PhD Thesis, University of Pennsylvania. Schmidt, Siegfried J. 1975 "'Negation' und 'Konstitution' als Kategorien Konkreter Dichtung", in Weinrich, H. (ed.): Positionen der Negativität. München: Fink, 393—433. Schultz, Charles B. and DiVesta, Francis J. 1972 "Effects of Passage Organization and Note Taking on the Selection of Clustering Strategies and on Recall of Textual Materials", Journal of Educational Psychology 63, 244-252. Schulze, Marlene and Bieri, Dora 1973 "Chaining and Spotlighting: Two Types of Paragraph Boundaries in Sunwar", in Hale, A. (ed.): Clause, Sentence and Discourse Patterns in Selected Languages of Nepal. Part I. Norman: The Summer Institute of Linguistics of the University of Oklahoma, 389-399. Schumacher, Gary M.; Liebert, Dale and Pass, Warren 1975 "Textual Organization, Advance Organizers and the Retention of Prose Material", Journal of Reading Behavior 7, 173 — 180. Scinto, Leonard F. M. 1983 "Functional Connectivity and the Communicative Structure of Text", in Petöfi, J. S. and Sözer, E. (eds.), 73-115. Sevbo, Irina P. 1976 "Die Erforschung der Struktur des kohärenten Textes", in Jelitte, H. (ed.)., Teil 2, 277-308. Sgall, Petr 1983 "One One Aspect of Textual Coherence", in Neubauer, F. (ed.), 135-144. Shearer, Ned A. 1972 "Alexander Bain and the Genesis of Paragraph Theory", The Quarterly Journal of Speech 58, 408-417. Shepper, Myra S. 1983 Cohesion in Student Narratives: Grades Four, Six, and Eight. Ed. D., University of Georgia. Shimmerlik, Susan M. 1978 "Organization Theory and Memory for Prose: a Review of the Literature", Review of Educational Research 48, 103-120. Shockey, Linda 1974 "Phonetic and Phonological Properties of Connected Speech", Ohio State Working Papers in Linguistics 17, 1—65. Siff, David 1975 "Teaching Freshman Composition to New York Cops", in Ohmann, R. and Coley, W. B. (eds.): Ideas for English 101. Urbana: National Council of Teachers of English, 64-71.

Connectedness of Texts: A Bibliographical Survey

495

Silberstein, Sandra V. 1982 Textbuilding Conventions and Personal Style in Oral Courtship Narrative. PhD Thesis, University of Michigan. Silman, Tamara 1968 "Die Absatzstrukturen in Goethes theoretischer und erzählender Prosa", Weimarer Beiträge 14, 854-867. Simmons, Cynthia F. 1979 Cohesion in Russian: The Major Resource of Textual Unity. PhD Thesis, Brown University, Rhode Island. Singer, Murray 1980 "The Role of Case-Filling Inferences in the Coherence of Brief Passages", Discourse Processes!, 185-203. Slamecka, Norman J. 1959 "Studies of Retention of Connected Discourse", American Journal of Psychology 72, 409-416. 1960 "Retroactive Inhibition of Connected Discourse as a Function of Practice Level", Journal of Experimental Psychology 59, 104 — 108. 1964 "Acquisition and Retention of Connected Discourse", Journal of Experimental Psychology 68, 330-333. Slobin, Dan I. 1968 "Recall of Full and Truncated Passive Sentences in Connected Discourse", Journal of Verbal Learning and Verbal Behavior 7, 876—888. Smith, Mark A. 1983 "Der kausale Vertextungstyp: Satzfolgen mit denn und weil", in Kühlwein, W. (ed.): Kongreßberichte der 13. Jahrestagung der Gesellschaft für angewandte Linguistik, GAL, Köln 1982: Texte in Sprachwissenschaft, Sprachunterricht und Sprachtherapie. Tübingen: Narr, 98-99. Smith, May J. 1983 The Effects of Text Cohesion on Reading Comprehension (Coherence, Clo^e, Anaphora, Discourse Analysis). PhD Thesis, University of Georgia. Smith, Raoul N. and Frawley, William J. 1983 "Conjunctive Cohesion in Four English Genres", Text 3/4, 347 — 373. Spivey, Nancy N. 1983 Discourse Synthesis: Constructing Texts in Reading and Writing. PhD Thesis, University of Texas at Austin. Staal, Johan F. 1967 "Some Semantic Relations between Sentoids", Foundations of Language 3, 66—88. Stalker, Douglas F. (ed.): 1984 Coherence (= Special volume of Linguistics and Philosophy 7, 1). Stammerjohann, Harro 1976 "Kohärenz und Delimitation: Zur Struktur des Absatzes", Folia Linguistica 9, 367-392. Stemmer, Brigitte 1981 Kohäsion im gesprochenen Diskurs deutscher Lerner des Englischen (= Manuskripte zur Sprachlehrforschung 18). Bochum: Seminar für Sprachlehrforschung. Stern, Arthur A. 1976 "When is a Paragraph?", College Composition and Communication 27, 253 — 257. Stone, V. Frank 1979 The Effect of Textual Cohesion on the Comprehension of Connected Discourse. Dissertation, University of Virginia. Streeter, Victor J. 1973 "A Look at Sentence-Type Cohesion", Language and Style 6, 109 — 116.

496

P. Lohmann

Sunday, Betty Louise Rhiner 1983 Cohesion and Coherence in the Language of a Selected Group of Secondary Bilingual Students. PhD Thesis, University of Illinois at Urbana-Champaign. Szanser, Adam J. 1973 "A Study of the Paragraph Structure", Statistical Methods in Linguistics 73, 79 — 90. Szwedek, Aleksander 1980 "Lexical Cohesion in Text Analysis", Papers and Studies in Contrastive Linguistics 11, 95-100. Tannen, Deborah 1983 Coherence in Spoken and Written Discourse (= Advances in Discourse Processes, 12). Norwood, N.J.: Ablex. Thomas, Gary Scott 1979 Topic Location and Paragraph Coherence: Effects on Paragraph Recall and Topic Identification. Dissertation, University of Arizona. Thorndyke, Perry W. 1978 Research on Connected Discourse: Structure, Comprehension, and Memory. A General Bibliography: 1900— 1977. Santa Monica, California: Rand Corporation. Uspenskij, Boris A. 1974 Poetics of Composition. Berkeley: California University Press. 1975 Poetik der Komposition: Struktur des künstlerischen Textes and Typologie der Kompositionsform. Frankfurt: Suhrkamp. den Uyl, Martijn 1983 "A Cognitive Perspective on Text Coherence", in: Ehlich, K. and H. van Riemsdijk (eds.), 259-283. Vaina-Pu§ca, Lucia 1974 "Pour une hierarchic des isotopies du contenu d'un texte poetique", Cahiers de linguistique theorique et appliquee 11, 317 — 326. Vasiliu, Emanuel 1979 "On some Meanings of 'Coherence'", in Petöfi, J. S. (ed.), 450-466. van de Velde, Roger G. 1978 "Einige mikrolinguistische Indizien der Textkonstitution", Linguistica Antverpiensia 12,261-281. 1980 "Understanding, Comprehension, and Coherence", Linguistica Antverpiensia 14, 235-250. 1981 Interpretation, Kohären% und Inferen% (= Papers in Textlinguistics 33). Hamburg Buske. 1983 "Coherence Relations in Texts and Inferential Processing", in Hattori, S. and Inoue, K. (eds.): Proceedings of the 19th International Congress of Linguists, Tokyo, AugustSeptember 1982, 1040-1044. Vuchinich, Samuel 1977 "Elements of Cohesion between Turns in Ordinary Conversations", Semiotica 20, 229-256. Wacha, Imre 1975 "System und Zusammenhänge der textphonetischen Ausdrucksmittel", Acta Linguistica Academiae Scientiarum Hungaricae 25, 39—75. Waltz, Nathan E. 1976 "Discourse Functions of Guanano Sentence and Paragraph", in Longacre, R. E. and Woods, F. (eds.): Discourse Grammar: Studies in Indigenous Languages of Colombia, Panama, and Ecuador. Part I (= Summer Institute of Linguistics Publications 52). Dallas: Summer Institute of Linguistics, 21 — 145. Warner, Richard George 1979 Discourse Connectives in English. PhD Thesis, Ohio State University.

Connectedness of Texts: A Bibliographical Survey

497

Wawrzyniak, Zdzislaw 1980 Hinführung in die Textwissenschaft: Probleme der Textbildung im Deutschen. Warszawa: Panstwowe Wydawnyctwo Naukowe. Weber, Heinz J. 1981 "Repräsentation von Alltagswissen und die Auffindung von Sinnzusammenhängen im Text", in Kohrt, M. and Lenerz (eds.): Sprache: Formen und Strukturen. Akten des 15. Linguistischen Kolloquiums, Münster 1980. Band 1. Tübingen: Niemeyer, 327 — 336. White, Alan R. 1967 "Coherence Theory of Truth", in Edwards, P. (ed.): The Encyclopedia of Philosophy. vol. 2 (New York: The Macmillan Company and The Free Press, London: CollierMacmillan), 130-133. Reprinted in Truth, ed. by A. R. White (1970). Garden City, New York: Anchor Books, Doubleday, 109-122. Wienold, Götz 1976 "Text Processing: Semantic Relations between Sentences and between Texts", Folia Linguistica 9, 37—44. Williams, Barbara 1970 The Well-Structured Paragraph. Columbus, Ohio: Charles E. Merrill. Williams, Joseph M. 1966 Some Grammatical Characteristics of Continuous Discourse. PhD Thesis, University of Wisconsin. Williams, Maurice and Stevens, Virginia M. R. 1972 "Understanding Paragraph Structure", Journal of Reading 15, 513 — 516. Winter, Eugene O. 1968 "Some Aspects of Cohesion", in: Sentence and Clause in Scientific English, Report of the Research Project "The Linguistic Properties of Scientific English". London: University College London, Department of General Linguistics. 1971 "Connection in Science Material: a Proposition about the Semantics of Clause Relations", Centre for Information on Language Teaching Papers and Reports 1. London: Centre for Information on Language Teaching and Research for British Association for Applied Linguistics, 14—25. 1974 Replacement as a Function of Repetition: a Study of Some of Its Principal Features in the Clause Relations of Contemporary English. PhD Thesis, University of London. 1977 "A Clause Relational Approach to English Texts: a Study of Some Predictive Lexical Items in Written Discourse", Instructional Science 6, 1—92. Winterowd, W. Ross 1970 "The Grammar of Coherence", College English 31, 823-835. Wittmers, Edith 1977 "Zu einigen Aspekten der Textkonstitution/Textkomposition als Beitrag zur Methodologie der Erfassung des Zusammenhanges sprachlicher Darstellungen", Studio grammatica 18, 213 — 235. van Wolkenten, Raymond V. 1975 A Formal Theory of Cohesive Discourse. PhD Thesis, Georgia Institute of Technology. Wrather, Nancy J. D. 1979 Side Sequences and Coherence in Children's Discourse. PhD Thesis, University of Texas, Austin. Wunderli, Peter 1979 "Satz, Paragraph, Text - und die Intonation", in Petöfi, J. S. (ed.), 319-341. Yekovich, Frank R. and Kulhavy, Raymond W. 1976 "Structural and Contextual Effects in the Organization of Prose", Journal of Educational Psychology 68, 616-635. Young, Jean Skoronsky 1979 Cohesion in Spoken Discourse: A Preliminary Analysis of the Speech of Lower-Income Black Drug Abusers. PhD Thesis, Illinois Institute of Technology.

498

P. Lohmann

Yuille, John C. and Paivio, Allan 1969 "Abstractness and Recall of Connected Discourse", Journal of Experimental Psychology 82, 467-471. Ziff, Paul 1984 "Coherence", Linguistics and Philosophy 1, 31 -42. Zolkiewski, Stefan 1972 "Poetique de la composition", Semiotica 5, 205 — 225. Zwicky, Arnold M. 1978 "Arguing for Constituents", in Farkas, D.; Jacobsen, W. M. and Todrys, K. W. (eds.): Papers from the Fourteenth Regional Meeting of the Chicago Linguistics Society. Chicago, 503-512.

Terminological Index "Coherence" Agar/Hobbs 1982 Banfield 1978 Bellert 1970 Biedermann 1976 Black/Bern 1981 Bonnafous 1981 Borel 1983 Charolles 1978, 1982, 1983a, b Charolles/Peytard (eds.) 1978 Chelala 1981 Cherchi 1978 Crothers 1978 Dascal/Katriel 1979 Engcl-Ortlieb 1983 Enkvist 1978 Ferry 1975 Fillol/Mouchon 1978 Fritz 1982 Grzyb 1974 Haberlandt/Bingham 1978 Hobbs 1977, 1979, 1983 Jakobson 1978 Jelitte 1978 Keenan/Klein 1975 Khatchadourian 1961 Kincaid 1977 Klein, S./Simmons 1963 Langleben 1978, 1981, 1983 Lautamatti 1978, 1982 Lehrer 1984 Lundquist 1979, 1980 Marcus 1980 Marello 1978 Margolis 1984 Mayenowa 1974

McCulley 1983 McCutchen/Perfetti 1982 Meyer 1971 Monaghan 1979 Neubauer (ed.) 1983 Nöth 1977 Odmark 1981 Östman 1978 Petöfi 1981 Peytard 1977 Philipps 1975 Polanyi 1978 Rath 1981 Reichman 1978 Reinhart 1980 Rescher 1973 Rudner1978 Ruhl 1973 Saloni/Trybulec 1974 Samet/Schank 1984 Sevbo 1976 Sgall 1983 Singer 1980 Smith, May 1983 Stalker (ed.) 1984 Stammerjohann 1976 Sunday 1983 Tannen 1983 Thomas 1979 den Uyl 1983 Vasiliu 1979 van de Velde 1980, 1981, 1983 White 1967 Winterowd 1970 Ziff 1984

Connectedness of Texts: A Bibliographical Survey "Cohesion" Adam 1978 Andersson 1978 Beene 1981 Binkley 1983 Bunge 1976 Clark 1983 Crowell 1973 Dressier 1983 Ezard 1978 Fairley 1973 Fillol/Mouchon 1978 Fine 1978 Fine/Bartolucci 1981 Fishman 1977 Fowler 1977 Giora 1983 Gutwinski 1976 Halliday/Hasan 1976 Hartnett 1980 Hasan 1968 Hohmann 1978 Hohulin 1971 Hooker 1972 Huddleston 1978 Johns 1979, 1980 Jordan 1978 Kallgren 1979 Kittredge 1981 Klein, W. 1978 Korpimies 1978, 1979 Lautamatti 1978 Linnarud 1978 Longacre 1983 Lonzi 1983 Marcus 1980 McCreedy 1983 McCulley 1983 Mosenthal 1979 Neuner 1983 Odmark 1981 Östman (ed.) 1978 Östman 1978 Perkins 1979 Phillips 1978 Polanyi 1976 Shepper 1983 Simmons 1979 Smith, M. J. 1983 Smith, R. N./Frawley 1983 Stemmer 1981 Stone 1979 Streeter 1973

Sunday 1983 Szwedek 1980 Vuchinich 1977 Winter 1968 Wrather 1979 Young 1984 "Composition"/"Organisation"/"Aufbau' 1 = Architecture 2= Structure Adriaens 1979 Baron 1976 de Beaugrande 1978, 1982a, b Becker 1980 Besmertnaja 1976 Blei/Pfütze 1975 Britton 1978 Chelala 1981 Cooper/Odell 1978 brothers 1979 Danes 1974, 1981 Delahaye 1983 Dillon 1983 DiVesta et al. 1973 Dolezel 1973 Dorfmüller-Karpusa 1981 Emig 1978 Fishman 1977 Fosberg 1979 2 Fowler 1977 Frase 1969 Gerdel/Slocum 1976 Giora 1983 2 Hellwig 1983 2 Hinds 1977, 1978, 1979 Hirsch 1977 2 Hohmann 1978 Hohulin 1971 Irmscher 1979 2 Kamprad 1975 Kinneavy 1980 2 Kittredge 1981 Krulee et al. 1979 Kulhavy et al. 1977 Landman/Moerdijk 1983 »Levy 1980 Liebsch 1977 2 Longacre 1968, 1970 Loriot/Hollenbach 1970 Mannhaupt/Wirrer 1977 2 Marshall/Glock 1978 McQuade 1980

499

500

P. Lohmann

Mellon 1969 Meyer 1975 Meyer/Rice 1982 Michel 1980 Moles 1979 2 Nystrand 1982 Patten 1983 Perlmutter/Royer 1973 Perry 1968 Prince 1973 2 Quasthoff 1976 Rosa/Eschholz 1982 Sasson 1971 Schenkein 1978 Schultz/DiVesta 1972 Schumacher et al. 1975 Shimmerlik 1978 Siff 1975 Silberstein 1982 2 Thorndyke 1978 Uspenskij 1974 Wawrzyniak 1980 'Williams, B. 1970 Williams, M./Stevens 1972 Wittmers 1977 Yekovich/Kulhavy 1976 Zolkiewski 1972 "Connectedness" / "Connexity" / "Connectivity" 1= 2= 3= 4=

interclausal c. intersentential c. Verkettung Zusammenhang

Agricola 1975 Ajdukiewicz 1935 'Ballard et al. 1971 a, b Beekman 1970 2 Biedermann 1976 Brazil 1983 Callow 1970 Christie/Schumacher 1975 Cinque 1983 Cohen/Johansson 1967 Cunningham 1972 DaneS/Viehweger (eds.) 1981 Deese 1961 DeVilliers 1974 DiVesta et al. 1973 Ehlich/van Riemsdijk (eds.) 1983 Finn 1978 Prase 1968, 1969

2

Grimes 1966, 1975 van Haaften et al. 1983 Heidolph 1966 4 Hellwig 1984 3 Hoffmannova 1981 Jacobson 1966 Jarvella 1970, 1971 Kantor 1977 Karlsen 1959 Kayne 1983 King/Russell 1966 King 1968, 1971 Kleine 1983 Krulee et al. 1979 Lachman/Dooling 1968 2 Maciejewski 1975 Marciszewski 1977 Marshall/Clock 1978 McCutchen/Perfetti 1982 Meyer 1975 Meyers/Boldrick 1975 Nickel 1968 Nye 1978 Petöfi/Sözer (eds.) 1983 Pompi/Lachman 1976 Rosenberg 1966, 1968a, b Royer/Cable 1975 Russel/Sewell 1972 Sachs 1967 Samet/Schank 1984 3 Schecker 1982 Scinto 1983 Shockey 1974 Slamecka 1959, I960, 1964 Slobin 1968 3 Smith 1983 Staal 1967 Stone 1979 Thorndyke 1978