A new architecture for functional grammar

The question of discourse representation in Functional Discourse Grammar Focus of attention in discourse The complementarity of the process and pattern interpretations of Functional Gr

for Functional Grammar

Printed on acid-free paper which falls within the guidelines

of the ANSI to ensure permanence and durability

Library of Congress Cataloging-in-Publication Data

A new architecture for functional grammar / edited by J Lachlan

Mackenzie, Maria de los Angeles, Gomez-Gonzalez

p em, — (Functional grammar series ;'24)

Includes bibliographical references and index

ISBN 3-11-017356-5 (pbk : alk paper)

1 Functionalism (Linguistics) 2 Discourse analysis I, Macken-

zie, J, Lachlan — lí GómezGonzáles, Ma de los Angeles (Maria

de los Angeles) III Series

Bibliographic information published by Die Deutsche Bibliothek

Die Deutsche Bibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data is available in the Internet at <http://dnb.ddb.de>

© Copyright 2004 by Walter de Gruyter GmbH & Co KG, D-10785 Berlin

All rights reserved, including those of translation into foreign languages No part of this book may be reproduced or transmitted in any form or by any means, electronic or mechanical,

including photocopy, recording or any information storage and retrieval system, without

permission in writing from the publisher

Printed in Germany.

The question of discourse representation in

Functional Discourse Grammar

Focus of attention in discourse

The complementarity of the process and pattern

interpretations of Functional Grammar

Functional Discourse Grammar and language

production

J Lachlan Mackenzie

Comment clauses, Functional Discourse Gram-

mar and the grammar-discourse interface

The problem of subjective modality in the

Remarks on layering in a cognitive-functional

language production model

Discourse structure, the Generalized Parallelism

Hypothesis and the architecture of

Functional Grammar

Towards a speaker model of

This volume arises indirectly from the Ninth International Conference on Functional Grammar (ICFG9), which was held at the Universidad National

de Educacién a Distancia (UNED), Madrid, from 20 to 23 September 2000 Some of the chapters derive from papers given there, others have been spe- cially commissioned This book addresses, and offers solutions to, 4 problem that was keenly felt at that conference: Functional Grammar (FG), the model of the organization of human languages proposed by the late Simon Dik, was failing to achieve one of its central ambitions, to account for regularities of form that are attributable to the use of language in dis-

course

The first chapter of this volume, by Kees Hengeveld of the University

of Amsterdam (Netherlands), proposes that the solution to the problem lies

in turning the original model on its head He rejects the bottom-up ap- proach, which started with the individual predicate and worked upwards to

a fully specified underlying representation, in favour of a top-down ap- proach, which takes the speaker’s intention to communicate as its starting point At the same time, he develops suggestions in Dik’s posthumous work that a FG should have a modular architecture, proposing three levels

of analysis for each discourse unit to be analysed: the interpersonal, the

representational and the expression levels Within each of these levels, the

nesting of layers which came to be the trademark of FG in the nineties has been preserved Hengeveld sketches various ways in which this architec- ture, to be known as Functional Discourse Grammar (FDG), is more adequate than the previous model, by showing how it can account for phe- nomena involving interactions between levels

All but the last of the remaining chapters offer reactions by other schol- ars to Hengeveld’s bold proposals Matthew Anstey, Sessional Lecturer at the School of Theology, Charles Sturt University (Australia) and PhD can- didate at the Vrije Universiteit, Amsterdam (Netherlands), offers a thoughtful, critical and thoroughly documented history of FG from its ear- liest beginnings to the present He shows how FG, in its various guises, has struggled with the same seven problems throughout its existence and, even

in its metamorphosis as FDG, continues to face many of the same difficul- ties

FDG brings grammar closer to cognitive science in introducing a Cognition module In his chapter, Carlos Inchaurralde of the University of Zaragoza (Spain), assesses the cognitive adequacy of FDG While stressing the impenetrability of cognitive processes, he points out various ways, for example in a language like Japanese with morphosyntactic marking of speakers’ awareness of social roles, that cognition impacts linguistic form

in a manner that can be modelled within FDG

The following chapter, by John H Connolly of the University of Loughborough (United Kingdom), turns to another module of FDG, that for the Communicative Context He details the multiplicity of contextual factors that can influence the form of a discourse, and proposes a notation for representing contextual information He distinguishes carefully between information which belongs to the contextual module and that which is miore properly assigned to the interpersonal level The chapter contains a detailed application of Connolly’s proposals to an invented mini-discourse Discourse analysis is also central to the chapter by Francis Cornish, of the University of Toulouse (France), but his data are all taken from attested language use His purpose is to show how the new FDG architecture pro- vides a sound environment in which to account for phenomena of topic and focus in-discourse He achieves this-aim-by.confronting FDG-with work from the Columbia School of Linguistics (CS), which he finds to lack the tools needed to deal with the flow of discourse; on the other hand, incorpo- ration of CS’s concern with attention and accessibility would further strengthen FDG treatments of language use

Michael Fortescue, of the University of Copenhagen (Denmark), ob- serves a certain contradiction in the fact that FDG mimics language production while remaining a grammar His position is that we need to ad- dress both Process and Pattern in a complementary fashion He illustrates this thesis with data from Nootka, a language whose structures are heavily controlled by considerations of discourse pragmatics Fortescue concen- trates on Focus, showing that in languages like Nootka it belongs to both Process and Pattern; in a language where Focus is not encoded grammati- cally, it will be absent from the account of Pattern For FG, the conclusion might be to retain the traditional architecture for Pattern and adopt the new for Process

The tension between static grammar and dynamic language production also underlies the contribution by J.Lachlan Mackenzie, of the Vrije Uni- vérsiteit Amsterdam (Netherlands) He proposes a variant on Hengeveld’s proposals, dubbed Incremental Functional Grammar, in which the interper- sonal and expression levels account for real-time processes The

interpersonal component analyses the Move as a sequence of Acts, each of which is composed of temporally successive Subacts The role of the repre- sentational component, in this view, is that of a declarative grammar, providing semantic frames which organize focal and topical units into pro- positional form

Peter Harder, of the University of Copenhagen (Denmark), welcomes Hengeveld’s three-tier model, sharing Fortescue’s view that we need to recognize both the ‘cooking’ and the ‘recipe’, as Harder has it, even at the cost of a certain duplication of information He concentrates on construc- tions in which there is a mismatch between the two, with especial reference

to cases where the main clause serves to ‘hedge’ the content of the subor- dinate clause, as in I'm afraid John is ill

In the chapter by Marfa de los Angeles Gémez-Gonzalez, of the Univer- sity of Santiago de Compostela (Galicia, Spain), it is again FDG’s potential for treating the dynamics of discourse and of clause construction that is central She confronts Hengeveld’s proposals with other recent develop- ments in cognitively oriented linguistics, bringing them together in her model of Incremental Discourse Cognitive Grammar She shows how clauses with complex beginnings result from the speaker’s incremental manipulation of the addressee’s focus of attention

Modality has long been a crucial issue in FG Its treatment in FDG is the subject of the contribution by Jean-Christophe Verstraete, of the Uni- versity of Leuven (Belgium) The chapter concludes that the new architecture offers an improvement in this respect, allowing ~ as Ver- straete’s new data demand — a four-way distinction such that both epistemic and deontic modality (analysed as tensed and tenseless at the representational level) can be either subjective (located at the interpersonal level) or objective (and located at the representational level)

The ambition of a Functional Discourse Grammar is comparable in many ways to the scope of Functional Procedural Grammar as developed

by Jan Nuyts, of the University of Antwerp (Belgium) In a chapter which shares Verstraete’s concern with modality, he provides a critical review of FDG, arguing that layering is properly situated in cognition, and not in the grammar itself Nuyts asks grammarian to be content with dealing with grammatically coded distinctions, and not to dabble in the study of concep- tualization, which is non-linguistic, but hierarchical

Directly opposed to such functional minimalism, as Fortescue calls his analogous proposal, is the contribution by Ahmed Moutaouakil (University

of Rabat, Morocco), who argues for an expansion of the layering principle upwards into the structure of discourse and downwards into the inner struc-

ture of the word Moutaouakil proposes a Generalized Parallelism Hy- pothesis which entails that all products of linguistic activity, from text to word, are to be seen as richly layered discourses, with the proviso that the realization of the many layers is increasingly constrained as one descends

to ever smaller units

The final reaction to Hengeveld’s proposal takes the form of a substan- tial chapter by Dik Bakker, of the University of Amsterdam (Netherlands), and Anna Siewierska, of the University of Lancaster (United Kingdom) Unlike their colleagues, they concentrate on the expression level, imple- menting proposals for an incremental, dynamic view of the morphosyntax

of utterances They propose a formalism that works top-down, left-to-right and depth-first and permits constrained inheritance and percolation They then return to a central issue of this book, the relation between grammar and production, between pattern and process, arguing that FDG offers a model of the speaker’s actual processes of production, and backing this up with a detailed analysis of a few lines of transcribed speech

This volume closes with an Epilogue by Kees Hengeveld, in which he gives various responses to issues raised in the preceding chapters and then refines the model originally proposed This chapter can thus be read as rep- resenting the current status~questionis in FDG Hengeveld begins~by identifying FDG as a pattern model: although the structures of languages reflect the processes of communication, FDG does not model those proc- esses Yet he indicates his sympathy with dynamic implementations of the model He goes on to welcome the various contributions to clarification of the cognitive and contextual components of the model and ends by offering

an expanded version of FDG

The editors hope that this book will give a lively impression of the de- bates currently being conducted about the architecture of FDG As Anstey says, now “FG could do with more engineers and builders” The current volume can be seen as a blueprint for the work of those labourers

We hereby acknowledge the financial support provided by the Spanish Ministry of Science and Technology (Ministerio de Ciencia y Tecnologia, MCYT), the European Regional Development Fund (Fondo Europeo de Desarrollo Regional, FEDER), and the Xunta de Galicia (XUGA) under project BFF2002-02441, PGIDITO3PXIC20403PN, entitled “Discourse Analysis in English: Aspects of Cognition, Typology and Second Language Acquisition/Análisis del discurso en lengua inglesa: Aspectos cognitivos, contrastivos y de adquisicién", and by the Netherlands Organization for Scientific Research, for making it possible for Dr Gomez- Gonzalez to visit the Netherlands to work on the production of this book

Our gratitude goes to Susanna Hop MA for her invaluable assistance with word-processing Last but not least, we wish to thank the authors of the various chapters for their contributions, and for their patience as they : awaited the appearance of this book

GH GivTop GPH GPSG HDPSG IDCG IFG

IL TLL INFRM

IS Iso

IT

L Lab

Leoog

Lex LFG

Functional Procedural Grammar

formal features General Functional The- ory

Givenness Hierarchy Given Topic

Generalized Parallelism Hypothesis

Generalized Phrase Structure Grammar hearer

Head Driven Phrase Structure Grammar Incremental Discourse Cognitive Grammar Incremental Functional Grammar

Interpersonal Level Tilocution (of the expres- sion)

Informing move Information Structure International Standardi- zation Organization Interpersonal Theme actual language label

general cognitive repre- sentation language Lexeme

Lexical Functional Grammar

LIBMSEC Lancaster IBM Spoken

English Corpus

Open Systems Intercon-

nection Reference Model

x

x

speaker speaker Referential Act; relation- ship

Resumed Topic Representational Level Referential Phrase Rhetorical Structure Theory

Sentence; speaker Segmented Discourse Representation Structure Setting(s)

Systemic Functional Grammar

State of Affairs Situation Theory Subcategorization Sub Topic Ascriptive Act; discourse type

Textual Theme utterance utterance; discourse unit Underlying Clause Underlying Representa- tions

Utterance Focus West Coast Functional- ism

Individual; proposition term variable

a development First of all, there are many linguistic phenomena that can only be explained in terms of units larger than the individual sentence: dis- course particles, anaphorical chains, narrative verb forms, and many other aspects of grammar require an analysis which takes the wider linguistic context into consideration Secondly, there are many linguistic expressions which are smaller than the individual sentence, yet function as complete and independent utterances within the discourse This requires a conception

of utterances as discourse acts rather than as sentences, as has been shown

in Mackenzie (1998)

Hannay and Bolkestein (1998) argue that the proposals’ which have been developed aiming at the expansion of FG into a grammar of discourse represent two different approaches In the first, the discourse level is cov- ered by additional hierarchically superordinate layers This approach, called the upward layering approach in Hannay and Bolkestein (1998), is exemplified by Hengeveld (1997) and Moutaouakil (1998) In the second approach, the discourse level is handled by a separate component, linked to the grammatical component through an interface Hannay and Bolkestein (1998) call this the modular approach, examples of which are Van den Berg (1998) and Vet (1998)

In this chapter I want to claim that an adequate model of the grammar of discourse requires the integration of these two approaches, i.e I will argue that both the application of extended layering and the recognition of vari-

ous levels of analysis are necessary The model of a Functional Discourse Grammar (FDG) presented here is thus both hierarchical and modular A major feature of the model is that it works in a top-down fashion, that is, decisions at higher levels and layers of analysis determine and restrict the possibilities at lower levels and layers of analysis This feature of FDG will

be treated first, in Section 2 Section 3 then presents a general outline of the model, focusing on the various levels of analysis and the complex inter- faces linking them to one another The layers to be distinguished at the various levels are presented in detail in Section 4 Section 5 looks at the dynamic top-down construction of basic linguistic expressions within FDG

by analyzing a series of illustrative examples More complex examples which involve intricate interactions between the various levels of analysis are discussed in Section 6 The paper is rounded off in Section 7

2 Top down

In Levelt (1989) the speech production process is described as a top-down process, running from intention to articulation His analysis suggests that the speaker first.decides.on a communicative purpose, selects the informa- tion most suitable to achieve this purpose, then encodes this information grammatically and phonologically and finally moves on to articulation Levelt shows that there is ample support in psycholinguistic research for this conception of speech production

The speech production model used in FG (Dik 1997a: 60) has a quite different orientation It starts out with the selection of predicate frames that are gradually expanded into larger structures, which when complete are expressed through expression rules In view of Levelt’s (and many other psycholinguists’) findings, this organization of the grammar runs counter to the standard of psychological adequacy that FG should live up to (Dik

1997a: 13-14)

In the model defended here production is therefore described in terms of

a top-down rather than a bottom-up model This step, apart from having a higher degree of psychological adequacy, is crucial to the development of a grammar of discourse: in a top-down model, the generation of underlying structures, and in particular the interfaces between the various levels, can

be described in terms of the cornmunicative decisions a speaker takes when constructing an utterance, as will be illustrated in Sections 5 and 6

3 Outline of the model

Figure 1 gives the basic outline of FDG It shows that FDG distinguishes three interacting levels: the interpersonal level, the representational level, and the expression level, in that hierarchical order The presence of these three levels as separate modules within the model is the major difference from earlier upward layering proposals

Layering applies at each level separately, ic each level is organized hi- erarchically, as will be shown in the next section This hierarchical organization of the model is the major difference from earlier modular pro- posals

Mapping rules link the interpersonal to the representational level, in those cases in which semantic content is necessary for the transmission of a certain communicative intention Expression rules then link the interper- sonal and representational levels to the expression level In cases in which only pragmatic content has to be transmitted, expression rules directly link the interpersonal to the expression level The various linking mechanisms may be interpreted as interfaces which define the possible correspondences between layers at different levels

The three levels interact with a cognitive component and with a com- municative component The cognitive component represents the (long- term) knowledge of the speaker, such as his communicative competence, his knowledge of the world, and his linguistic competence The speaker draws on this component at each of the three levels

The communicative component represents the (short-term) linguistic in- formation derivable from the preceding discourse and the non-linguistic, perceptual information derivable from the speech situation As far as the linguistic information is concerned, the communicative component is fed

by the interpersonal and expression levels, and feeds the representational level in order to enable later reference to earlier acts and expressions, as will be illustrated in Section 6 Of course, short-term information may be selected for long-term storage and is in that case passed on to the cognitive component.

4, Levels and layers

4.1 Introduction

As was mentioned in the previous section, each level of analysis in Figure

1 is organized hierarchically In this section I will first of all review each of the levels separately Then I will present the full model and compare it to the earlier layered sentence model, as presented in Hengeveld (1989)

4.2 The interpersonal level

The hierarchical structure of the interpersonal level is presented in Figure

Figure 2, The interpersonal level

Note that this representation is non-exhaustive, in the sense that there are higher levels of discourse organization which are not captured here In 4.6 I will return to this point

At the interpersonal level a central unit of analysis is the move (M), de- fined in Kroon (1995: 66), following Sinclair and Coulthard (1975), as “the minimal free unit of discourse that is able to enter into an exchange struc- ture” As such, the move is the vehicle for the expression of a single communicative intention of the speaker Examples of such communicative intentions are inviting, informing, questioning, threatening, warning, rec- ommending, etc

In order to achieve his communicative intention, the speaker executes one or more discourse acts (A), defined in Kroon (1995: 65) as “the small- est identifiable units of communicative behaviour” A move consists of one central act, which may be supported by one or more subsidiary acts Every act may be characterized in terms of its illocution (ILL), by which I here mean the illocution as coded in the expression® (Wik 1997b: ch 11) Tllocu- tions are represented as abstract iflocutionary frames,’ which take the participants (Py) in the discourse act, i.e the speaker (Ps) and the addressee (P,), and the communicated content (C), i.e the information transmitted in the discourse act, as their arguments In order to build up the communi-

cated content the speaker may have to execute one or more ascriptive acts (T) and one or more referential acts ®: it is the speaker who refers to enti- ties by using referring expressions,’ and it is the speaker who ascribes properties to entities by applying predicates to these referring expressions

4,3 The representational level

The hierarchical structure of the representational level is presented in Fig- tre 3

(pị: [e;: [fu) œ0] (e2)1 (bị)

Figure 3 The representational level

Note that, again, this representation is non-exhaustive There are higher levels of semantic organization which are not captured here (see Section

4.6)

In transmitting his communicative intention the speaker will in most cases have to fill his utterance with basic’ semantic content, i.e with de- scriptions of entities as they occur in the non-linguistic world These entities are of different orders: third-order entities or propositional contents (p); second-order entities or states of affairs (e); first-order entities or indi- viduals (x); and zero-order entities or properties (f) Within the maximally hierarchical representation given in Figure 3 the propositional content (p:) contains the description of a state of affairs (e;), which contains the descrip- tion of a property (f,) and the description of an individual (x;) Note, however, that all entity types may also be expressed directly, i.e non- hierarchically, through lexical items

4.4, The expression level

The hierarchical structure of the expression level is presented in Figure 4

(Parai: [(S¡: [(CH¡: [ŒrPi: [Lexi)] (PrP) (RPx: [Œex2)] (RP))] (CY) Sp)] ŒPara,))

Figure 4 The expression level

This representation is just a simplified example of what the expression level might look like It is an example, since every language has its own

expression possibilities, which lead to different expression units in their grammars It is simplified, since the expression possibilities of a single language are generally much more refined than in the example I have given here This simplified example will suffice, however, to illustrate the points

I want to make in this chapter It is a representation of constituent structure which starts at the level of the paragraph (Para), which may consist of one

or more sentences (S), each of which may contain one or more clauses (Cl), which may contain one or more predicate phrases (PrP) and referen- tial phrases (RP), cach of which may contain one or more lexemes (Lex)

It is important to note that the expression level corresponds with what in Levelt’s production is the product of grammatical and phonological en- coding Articulation, the final step, is not a level within the grammar, but the actual output of the grammar

4.5 Integration

The levels and layers discussed so far are given in Figure 5 The elements

in boldface at the interpersonal and representational levels in this figure correspond to units that were present in the layered representation of clause structure defended in Hengeveld (1989) and its upward-layering elabor- ation in Hengeveld (1997) The correspondences may be listed as follows:

The introduction of the variable T at the interpersonal level makes it possible to distinguish systematically between ascription as an act of the

speaker, and the entity type which is described within this act of ascription Often the speaker will use the description of a zero-order entity (f) to give content to his ascriptive act, but he might also use, for instance, a first- order entity (x) in a classifying or identifying construction Similarly, the variable R allows for a systematic distinction between the act of referring

on the one hand, and the entity type referred to on the other Frequently the speaker will use the description of a first-order entity to give content to his referential act, but reference to other types of entity is equally possible The introduction of the variable C opens up a way to distinguish the in- formation communicated in a discourse act from the nature of the entity type the description of which is used to transmit that information As a re- sult, it is no longer necessary to assume that every discourse act contains a propositional content, i.e a third-order entity In many circumstances it is sufficient, for instance, to communicate information by simple reference to

a first-order entity

A further difference between Hengeveld (1989, 1997) and the current proposal concerns the presence of the Expression Level in the model The major motivating factor for the introduction of this level is the existence of meta-linguistic expressions (Sweetser 1990) or reflexive language (Lucy 1993) This phenomenon will be illustrated in Section 6

4.6 Upward layering

The previous sections have shown that each level or module in the proposed model has its own layered structure Section 4.5 has stressed the relevance of further downward layering at the interpersonal level Further upward layer- ing is necessary too, but will not be dealt with here A major point, however,

is that upward layering is not restricted to the interpersonal level, but is a fea- ture of all levels of analysis Thus, at the interpersonal level there may be linguistic reasons to distinguish, for instance, between the layers of Turn and Exchange in dialogues; at the representational level languages may give spe- cial treatment to the layers of Episode and Story in narratives; and at the expression level there may be reasons to distinguish layers, for example Sec- tion and Chapter in written communication In each case, the possible mappings of interpersonal to representational to expression categories have

to be determined partly on a language-specific basis Thus, a Move is mapped onto an Episode onto a single sentence in narratives in many lan- guages of Papua New Guinea, whereas it is commonly expressed through a paragraph in most Western European languages

5 Making choices at the interpersonal level

Within the model proposed here, the construction of linguistic expressions can be interpreted as a decision-taking process on the part of the speaker This process applies in a left-right and top-down fashion

Left-right decisions apply at the interpersonal level only For instance, only after deciding on the communicative intention Warning for a certain move may the speaker select, for instance, the appropriate discourse acts Vocative, and Imperative Only after deciding on the discourse act Voca- tive does the necessity to execute a referential act become obvious, etc The result might then be something like (1):

Top-down decisions are of a more complex nature These concern the decisions the speaker makes with respect to (i) the semantic content neces- sary to successfully execute an interpersonal act, and (ii) the expression category necessary to successfully transmit his communicative intentions

In what follows I will restrict myself to some examples of this decision- making process In order to facilitate the interpretation of the representa-

tions l use tables in which the various levels correspond with rows: the first row contains the interpersonal units, the second row the representational units, and the third row the expression units Furthermore, I concentrate on the representation of the relevant part of the example under consideration Within the example this part is printed in italics; within the representation

it is separated from the remaining part by means of vertical lines

Let us first consider a lexical, i.e ready-made C If the speaker wants to express his frustration about the way things are going he may select a lexi- cal item which serves this purpose directly An expression such as damn has pragmatic, not semantic content Therefore, the speaker may move di- rectly from the interpersonal to the expression level:

The expressive illocution takes care of the prosodic contour of this one- word expression

The next example concerns a lexical R If the speaker wants to draw the attention of someone present in the speech situation he may simply call his name Here we have a referential act which makes use of a lexical item (Lex) which does not have semantic content, but only referential content Therefore, the speaker may move directly from the interpersonal to the ex- pression level again:

This may be contrasted with a case in which a referential act (R) again refers to a propositional content (p) but is expressed by means of a referen- tial phrase:

(Au IDECL (Pu)sp (Pa)aaar (Cr: L Ry) 1(CĐ1(AU)

(Ai: [DECL (Pi)sp (Pa)acar (Cu: £ Œ) LC) (Ay)

(fy) (Lexy)

But the speaker may also decide on a first-order entity (x), expressed as

a referential phrase (RP) to transmit the same kind of information, as in the following example:

6 Complex interactions between levels

6.1 Introduction

Let me now turn to more complex interactions between the various levels

As Figure 1 already showed, the communicative context feeds into the rep- resentational level The preceding discourse is of course part of this communicative context, and units within this discourse may be used for later reference This is achieved in the model presented here by having

these units reappear within the representational level In this way we may account for constructions like the following:

b 7s that a threat or a promise?

(10) My brother-in-law, if that’s the right word for him, is a poet

In (9b) the demonstrative that refers to the preceding move In (10) it refers to the preceding lexeme brother-in-law These examples thus illus- trate that elements from both the interpersonal and the expression levels are available for later reference once they are produced, i.e they may become part of the representational level in ensuing communication Reference to elements from the interpersonal and expression levels will be studied sepa- rately below In order to account for these cases the model presented here (Figure 7 below) allows for the copying of elements from the interpersonal and the expression levels to the representational level via the communica- tive context

In order to show their different status, variables from both levels are written with lower-case letters when they are used In the next sections I will give some examples of how these variables are used in analyzing a va- riety of constructions which involve metacommunicative and metalin- guistic expressions

6.2 Referents from the interpersonal at the representational level

In this section I will discuss two construction types that can only be prop- erly understood if we allow units from the interpersonal level to enter the representational level: hedged performatives and identity statements Consider the following examples from Spanish:

‘I'm afraid that Juan is ill.’

‘I'm afraid that Juan is il’

In (11a) the speaker simply expresses his state of mind The embedded clause, in which the subjunctive is used, represents what he fears might be the case In (11b) the indicative is used in the embedded clause This sen- tence, unlike (lla), is an example of a so-called hedged performative (Fraser 1975), in which the embedded clause represents the actual informa- tion the speaker wants to transmit, but which he ‘hedges’ since he thinks the addressee might not like what he has to say In cases like these the ac- tual communicated content (c) is hidden in the embedded clause (Cl, which itself is the expression of a referential act (R), so that this sentence may be represented as follows:

A second case in which units from the interpersonal level figure at the representational level concerns so-called identity statements (Declerck

1988, Hengeveld 1992, Keizer 1992) as illustrated in (12):

Sentences like (12), with a prosodically prominent copula, serve the pur- pose of stating that the act of referring to an object by using a certain name

is equivalent to the act of referring to that same object by another name; hence they are statements about the validity of acts of reference Therefore the representation of e.g the Morning Star may be as follows:

(Au [DECL Pi)sp Pa)adar (Ci: [ (Ry) J €1)

(2)

(RP)

This representation states that in (12) the speaker executes a referential act

by making reference to a referential act (1), which is expressed as a referen- tial phrase (RP)

6.3 Referents from the expression level at the representational level

Units from the expression level may enter the representational level as well I will discuss two cases here: metalinguistic conditionals and direct speech

An example of a metalinguistic conditional is given in (13):

The word that refers to the preceding word concert, which is a case of ref- erence to the code rather than to the message Thus there is a referential act (R) in which reference is made to a lexeme (lex) which is expressed as a lexeme (Lex), as indicated in the following representation:

(lex) (Lex;)

(14) He said: “;Cémo estas?”

and that direct speech reports may contain meaningless noise, as in:

(15) He said: “gagugagugagu”

The latter example furthermore shows that the imitated code can be any part of the expression level

In (14) the speaker refers (R) in the second argument of the verb say to

a previous sentence (s) which in the actual expressionis‘reépeated through

imitation This example imay thus be represented as follows"!

(Au [DECL (Pos (Pa) agar (Ci: L- (Ry) 1 Cy Ad)

What are the restrictions on top-down decisions within the produc- tion process, i.e what do the interfaces between the three levels of grammar look like?

What is the internal structure of the cognitive component and how does it interact with the three levels of grammar?

What is the internal structure of the contextual component, particu- larly with respect to the representation of the non-linguistic context, and how does it interact with the three levels of grammar?

None of these questions is new to FG I hope that the model of FDG presented here will provide the basis for an integrated approach to these central issues in linguistic theory

Notes

ber of people The Amsterdam FG-DISCO group has met at inegular intervals over the last few years, and has been a very inspiring environment for discussion of the topics dealt with in this chapter I am indebted to the members of this group, Machtelt Bolkestein, Mike Hannay, Caroline Kroon, Lachlan Mackenzie, Rodie Risselada, and Co Vet, for the many open- minded and inspiring discussions we have had A special word of thanks goes to Mike Hannay, for a revival of the group’s activities when the time was there; and to-Lachtan Mackenzie for joining this revival Outside the FG-D SEWENG PAM OAVOGKE Bie period, Thave had countless

TRUNG TÂM HỌC LIỆt

£.09 244

discussions with Gerry Wanders about the topic of this chapter, in which she has manifested herself as a critical and generous sparring partner J am grateful to her for her help Given the extensive interaction with all of these colleagues over a long period of time, it is hard to do justice to their indi- vidual contributions to the contents of this chapter As a result, 1 do not want to claim originality for many of the ideas presented in this chapter, only for the way these are put together

Crevels (1998), Gémez Solifio (1996), Hengeveld (1997), Jadir (1998), Kroon (1997), Liedtke (1998), Mackenzie (1998, 2000), Moutaouakil (1998), Rijkhoff (1995), Steuten (1997, 1998), Vet (1998)

possibilities 1 include not only sentences types, but also prosodic encoding, morphological encoding, and conventionalized lexicalization patterns

verb

ence by the act of referring”

is red, we ascribe to it the property of redness, but we predicate of it the

References

Berg, Marinus E van den

A Machtelt Bolkestein (eds), 77~106

Clark, Herbert H and Richard J Gerrig

Connolly, John H

Hannay and A Machtelt Bolkestein (eds), 167-190

Connolly, John H., Anthony A Clarke, Steven W Garner and Hilary K Palmen

viduals linked up by a computer network In: John H Connolly, Roel

M Vismans, Christopher S Butler and Richard A Gatward (eds), 33-58

Connolly, John H., Roel M Vismans, Christopher S Butler and Richard A Gat-

ward (eds)

York: Mouton de Gruyter.

Crevels, Mily

(eds), 129-148

Declerck, Renaat

Leuven University Press and Dordrecht: Foris

Dik, Simon C

Clause (K Hengeveld, ed.) (2d rev.] Berlin and New York: Mouton

de Gruyter

Structures (K Hengeveld, ed.) Berlin and New York: Mouton de Gruyter

Fraser, Bruce

Speech Acts (Syntax and Semantics 3) New York: Academic Press Gómez Soliño, José S

cional In: M Teresa Caneda Cabrera and Javier Pérez Guerra (eds),

Os Estudios Ingleses no Contexto das Novas Tendencias Vigo: Uni- versidade de Vigo

Hannay, Mike and Machtelt Boikestein

vii~xii

Hannay, Mike and Machtelt Bolkestein (eds)

delphia: Benjamins

Hengeveld, Kees

25: 127-157

New York: Mouton de Gruyter

Vismans, Christopher S Butler and Richard A Gatward (eds), 1~16 Jadir, Mohammed

A Machtelt Bolkestein (eds), 43-58

Keizer, M Evelien

Dissertation, Free University Amsterdam.

Kroon, Caroline

at Amsterdam: Gieben

John H Connolly, Roel M Vismans, Christopher S Butler and Richard A Gatward (eds), 17-32

Levelt, Willem J M

Press

Liedtke, Frank

and A Machtelt Bolkestein (eds), 107-128

Lucy, John A (ed.)

bridge: Cambridge University Press

Lyons, John

Mackenzie, J Lachlan

Machtelt Bolkestein (eds), 267-296

Linguistica Hafniensia 32: 23-44,

Moutaovakil, Ahmed

Grammar In: Mike Hannay and A Machtelt Bolkestein (eds), 25—

42

Rijkhoff, Jan

grammar/pragmatics interface Working Papers in Functional Grammar 58 Amsterdam: University of Amsterdam

and Pupils London: Oxford University Press

Steuten, Ans A G

Funcytional Grammar perspective In: Connolly et al (eds), 59-73

and A Machtelt Bolkestein (eds), 59~75

Sweetser, Eve E

semantic structure Cambridge: Cambridge University Press.

Vet, Co

ity and discourse moves In: Mike Hannay and A Machtelt Bolkestein (eds), 1~24.

Matthew P Anstey

1 Introduction

Hengeveld’s chapter (this volume) was written with two purposes in mind.!

On the one hand, it is the latest attempt within Functional Grammar to for- mulate an explicit model for the detailing of natural language phenomena

On the other, it is a manifesto for those working within the Functional Grammar movement, a framework wherein scholars may locate their contri- bution to the general theory This new architecture, therefore, is not only a design for a theoretical ‘space’ It is also, metaphorically speaking, a design for a collective, academic space, a conceptual meeting-place for those work- ing in the many divergent subdisciplines of the linguistic sciences who wish

to identify themselves to some degree with Functional Grammar

Accordingly, this chapter is an evaluation of Hengeveld’s proposal in light of these two purposes, as seen from the perspective of the evolution of Functional Grammar Three important questions will be addressed As a framework for Functional Grammar, what is its relationship to previous models? Does it address the objections raised against previous models? As a framework for Functional Grammarians, does it create sufficient space for the Functional Grammar academic community to continue to share one roof?

2 Preliminaries

2.1 Chronological Overview

Clearly, such a historical overview needs to be highly selective I have concentrated on the core of Functional Grammar, organizing the material

into five periods Each FG, refers not only to the period but also to its ca- nonical FG publication FG,, however, refers to both the period and Hengeveld’s Functional Discourse Grammar

FG; covers 1997 to September 2000, beginning with the publication

of the two volumes of The Theory of Functional Grammar (Dik

1997a; 1997b);

FG, extends from the ninth International Conference on Functional Grammar (Madrid, Spain) onwards, where Hengeveld’s (2000) Functional Discourse Grammar was first presented

Figures 1 and 2 provide an overview of these five periods.”

Dik 1978 Functional Grammar

Dik 1980 Studies in Functional Grammar

Hoekstra et al 1980 Special edition on Functional Grammar Glot 33—4)

Bolkestein et al 1981 Predication and expression in Functional Grammar

Martin Mingorance 1981 Translation of Dik 1978 into Spanish

°82: Functional Grammar workshop during 13th Int Congress of Linguists, Tokyo, Japan Dik 1983 Advances in Functional Grammar

P84: ist Functional Grammar conference, Amsterdam, The Netherlands

'85: Functional Grammar symposium on predicate operators, Amsterdam, The Netherlands Bolkestein et al 1985 Syntax and pragmatics in Functional Grammar

Bolkestein et al 1985 Predicates and terms in Functional Grammar

['86: 2nd Functional Grammar conference, Antwerp, Belgium

| $7: Functional Grammar symposium on the computer, Amsterdam, The Netherlands fuyts and de Schutter 1987, Getting one's words into line

an der Auwera and Goossens 1987 Ins and outs of the predication

'88: 3rd Functional Grammar conference, Amsterdam, The Netherlands

IƑG›

1989-1997

Dik 1989 The theory of Functional Grammar

Moutaouakil 1989 Pragmatic functions in a Functional Grammar of Arabic

P90: 4th Functional Grammar conference, Copenhagen, Denmark

fuyts et al 1990 Layers and levels of representation in language theory

(Hannay and Vester 1990 Working with Functional Grammar: descriptive and compute- ional applications

Siewierska 1991 Functional Grammar

P92; Sth Functional Grammar conference, Antwerp, Belgium

Keizer 1992 Reference, predication and (in)definiteness in Functional Grammar

IRijkhoff 1992 The noun phrase: a typological study of its form and structure

Fortescue et al 1992 Layered structure and reference in a functional perspective

’'94: 6th Functional Grammar conference, York, England

Engberg-Pedersen et al, 1994 Function and expression in Functional Grammar

°95: Simon Dik dies after three years of illness

Kroon 1995 Discourse particles in Latin

|'96: 7th Functional Grammar conference, Cérdoba, Spain

iDevriendt et al 1996 Complex structures A functionalist perspective

ingberg-Pedersen et al 1996 Content, expression and structure: Studies in Danish Func- ional Grammar

Butler et al 1997 A fund of ideas: recent developments in Functional Grammar

(Connolly et al 1997 Discourse and pragmatics in Functional Grammar

PGs

1997-2000

Dik 1997 The theory of FG Part I: the structure of the clause

Dik 1997 The theory of FG Part II: complex and derived structures

'98: 8th Functional Grammar conference, Amsterdam, The Netherlands

bertz et al 1998 The structure of the lexicon in Functional Grammar

Hannay and Bolkestein 1998 Functional Grammar and verbal interaction

Faber and Usén 1999, Constructing a lexicon of English verbs

Martin Arista et al 1999 Nuevas perspectivas en Gramdtica Funcional

IFGs

12000

onwards

Aackenzie 2000 First things first Towards an Incremental Functional Grammar

Pérez Quintero 2001 Special edition of Revista Canaria de Estudios Ingleses Challenges and Developments in Functional Grammar

Figure 2 Historical Overview of Functional Grammar

2.2 Central Problems

In addition to the evolution of FG, we wish to pay attention to several re- current problems that emerge These provide various perspectives from which to evaluate Hengeveld’s proposal It is beneficial to articulate them

at this stage.*

structure have in the grammar?

derlying representations (URs) actually represent and how should they be interpreted?

communication as process?

defined and applied? How many are needed?

phenomena beyond the sentence level?

logical adequacy mean and does FG fulfil it?

mally and explicitly represent language structure?

3 FGo - prior to 1978

3.1 Publications prior to FGo

Two of Simon Dik’s earliest writings provide helpful background to FGp Firstly, Dik (1966: 406; cf 1967b) indicates that his first disagreement with generative-transformational grammar concerned the strict division be- tween grammar and semantics For example, the words manner, many, manifest, manly, manifold and so forth ali contain the string ‘man’ but such

a purely formal observation “can never tell us which of these [recurrences] are grammatically significant and which are not” Therefore, some seman- tic criterion is necessary to determine this Dik concludes that “no grammatically relevant element can be found in complete independence of meaning” Nevertheless, he maintains, “[w]ithout being committed to rec- ognize the exclusive rights to the title of ‘grammar’ claimed for it, one should not underrate the value of the contribution that especially the trans-

formational component of generative grammar can make to general linguistic theory” (Dik 1966: 410)

Secondly, the only published evidence of Dik’s familiarity with Prague School linguistics is in a book review wherein he first mentions “function- alism’” (Dik 1967a) Dik approves of Jakobson’s view of the ‘means-ends model’ of language, writing “(t]his means that each fact of language is evaluated not only with respect to the system as a whole, but also with re- spect to the ultimate function it fulfils in the larger setting of extra- linguistic reality” (Dik 1967a: 87) These two basic meanings of ‘func- tional’ appear in all of Dik’s writings

Of more interest is Dik’s criticism of the Prague School’s theme-rheme analysis He argues that their analysis is “arrived at impressionistically” and lacks scientific rigour “In the first place”, he writes, “it is rather mis- leading to subsume it under syntax and treat it as a part of grammar Rather, it should be regarded as a phenomenon depending on the interpre- tation of the utterance in context and situation and thus already presupposing the grammatical (and the semantic) structure of the utterance

In the second place, if the distinction is really to be workable, there should

be some principled basis for finding theme and rheme in any given sen- tence” (Dik 1967a: 86) This is the view now widely advocated in Functional Graramar (cf Bolkestein 1998),

“any version of ‘restricted’ constituent structure grammar” (1968: 199), two are central and will come to characterize all FG models These are the

“independent introduction of grammatical functions” and a “different no- tion of ‘derivation’”

Dik’s justification of grammatical functions as being basic to linguistic description consistently starts from the observation that formally equivalent constituents do not have equivalent linguistic properties Therefore, con- stituents must have additional functional properties Dik does not claim that this insight is original and is happy to assume Longacre’s (1965: 65) definition: “By function is meant the particular office or role of one distin-

guishable part of a construction type in relation to other parts of the same construction”.°

What precisely then is a ‘grammatical function’? Dik writes that “ grammatical functions are irreducible aspects of grammatical structure, which can be partly correlated to formal features ., but cannot possibly be completely reduced to these” (1968: 154) His argument is essentially one

of explanatory power: if we assume that a grammatical system has “gram- matical functions’, then such a system provides a superior account of linguistic data However, the axiom of functional primitives is- itself not

Dik’s “different notion of ‘derivation’” is the forerunner to the ‘avoid transformations’ constraint found in FG; onwards “The basic (negative) property of-a functional-grammar”, he writes; “ -is that it does-not include transformational rules” (1968: 163) To achieve this Dik posits a monostratal underlying representation of the functional, categorial, and constitutional elements of each linguistic expression

More specifically, FGo has three basic components: constituents (that is, lexical categories) such as verb, noun, adjective, article; categories such as noun, pronoun, noun phrase; and functions such as subject, object, predica- tor, modifier, indirect object, question How then does one select the necessary functions to describe a language? Dik provides the following stipulation, anticipating in many respects the central philosophy of FG, :

“At least so many functions are set up for a language as there are gram- matical (not semantic) differences between the linguistic expressions of that language which cannot be correlated to differences in constituency and/or in categorization” (1968: 176)

FGp distinguishes four sets of rules that relate constituents, categories, and functions together to form an ‘independent linguistic expression’ The application of these rules results in a single UR that combines functional (CAPITALS), categorial (normal) and constituent (italic) elements For ex- ample, the sentence (1) has the underlying structure (2) (Dik 1968: 198):’

(2) ile( (Sdea (SUBJ (np (apsy (DET (art(the) )+HEAD (n,, (man) ) ) ) )+PRED (EV £Vinee (EVinee past (EVinte past 3d ps ›g(C2/09) ) ) ) ) ) ) )

Dik considers the various possible relationships between semantics and grammar and concludes, along the lines of Weinrich (1966), “[i]n my opin- ion these various developments point to the possibility that [grammar is dependent on semantics] might finally prove to the right one” (1968: 293) This revealing quotation underlines the grammatical nature of the functions Dik posits “Perhaps we could even go further”, he continues, “and include the full semantic description of a linguistic expression in its grammatical specification” These are, however, “speculations which have little more than a programmatic value They might, however, provide interesting pos- sibilities for a further elaboration of the theory of functional grammar”

(1968: 293)

Thus, for Dik FG is an exclusively grammatical, non-semantic system, yet it includes ‘functional primitives’ The semantics of a language is a dif- ferent matter This bracketing out of semantics from the system of grammar will change in subsequent FG models, where the UR is inter- preted as conveying grammar and meaning

Matthews’s (1969: 350) review of Dik’s dissertation is noteworthy He views the dissertation positively but has grave doubts about FGp Although modelled after tagmemics,Š FGo, unlike tagmemics, has no primitive that corresponds to syntagmeme (or to Halliday’s structure) Matthews takes this oversight as problematic to the whole concept of FGo: “But until he does so, his present formalization is literally not worth a moment’s consid- eration” (1969: 358) In other words, this is the first sign of the problem of structure (PR1) Matthews’s observation boils down to asserting that the URs of FGo contain insufficient information to generate (via expression rules) the correct surface form Bakker and Siewierska (2000), over thirty years later, conclusively demonstrate that this problem remains; but see Bakker and Siewierska (this volume).°

4, Dik’s writings between 1968 and 1978

The “interesting possibilities” that a semantically based grammar could of- fer occupied Dik for the rest of his life In particular, the decade after his dissertation was spent on eclectically combining various components to as- semble FG, Three of his publications in this period are illustrative

Dik (1973b: 838) first refers to the predicate frame and operators when