A cognitive functional approach to nominalization in english

Trang 2

DE

Trang 4

Printed on acid-free paper which falls within

the guidelines of the ANSI

to ensure permanence and durability

ISBN 3 11 017809 5

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 2003 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

Trang 6

The present study reflects a research project to which many people, each in their own way, have contributed I would like to thank those people here

First and foremost I would like to thank Kristin Davidse As my Ph.D supervisor, she was involved in my research on nominalization patterns from its earliest stages onwards I am deeply grateful for the continual en-

couragement she has given me, for the openness with which she has shared

her own learning with me, and for the precision with which she has read my texts, time and again supplying extensive commentary This book, in which any remaining shortcomings are my own, in the first place represents the dialogue I have been able to conduct with her

I would also like to thank Hubert Cuyckens, Ad Foolen, Ronald Lan- gacker and Willy Van Langendonck for their rigorous comments on an earlier version of this book In addition, I am grateful to René Dirven and Ronald Langacker who, as editors of the Cognitive Linguistics Research Series, guided me through the proposal and writing process and whose careful comments on various chapters of the book I have greatly appreci- ated I thank the managing editor of the series, Birgit Sievert, for her pa- tience and for her professional assistance throughout the publication process I thank Jurgen Benteyn for his swift and always good-humoured help with the formatting of the manuscript, and Hubert Cuyckens, Kristin

Davidse and Lieven Vandelanotte for their general support Thanks also to Lieven Vandelanotte for his help with the drawings and to An Laffut and Lieven Vandelanotte for proof-reading parts of the manuscript

My thanks also go to a number of people who have in the past years commented on my work and helped me with their suggestions and criticisms In addition to Hubert Cuyckens, Kristin Davidse, Ad Foolen, Ronald Langacker and Willy Van Langendonck, these are: Eirian Davies, Renaat Declerck, Dieter Kastovsky, An Laffut, Odo Leys, Lachlan Mackenzie, Bill McGregor, Hans-Jérg Schmid, Lieven Vandelanotte, Frederike van der Leek, Jean-Christophe Verstraete and Emma Vorlat The University of Leuven has contributed to the publication of this book by allowing me to work on it as a postdoctoral fellow of the Research Council (postdoctoral

grant PDM/02/034)

Iam most grateful to my parents and my parents-in-law for their unfail-

Trang 7

Part I Towards a theoretical-descriptive approach to nominalization Chapter 1: Introduction Chapter 2: Theoretical assumptions 1 1.1 1 14 2 2.1 22 3 3.1 3.2 343 33.1 3.3.2 3.4, 4 3 5.1 5.2 6

Language: The relation between system and usage A usage-based approach to language

Cognitive Grammar: The interface between system and

usage

The language system as network

The natural symbolic relation between lexicogrammar and semantics

A natural description of language

The natural or non-arbitrary relation between grammar and meaning

Constructions: Composition, classification and function A functional approach to composition

Levels of functional analysis

A functional approach to classification

An internal functional analysis of constructions Classes as complex categories: Schema and prototype Composition and classification: Conclusion

Functional layers of organization

Constructions and the relations between them

Agnation and enation: A definition

Agnation and enation: Theoretical significance and heuristic value Theoretical assumptions: Conclusion Chapter 3: Nominalization 1 1.1 1.2

Trang 8

es 3.1 3.2.1 3.2.2 3.1 5.2 5.3

Nominalizations as non-arbitrary symbolic units Nominalizations as composition and reclassification Constructions as assemblies of symbolic units Nominalization as functional reclassification

Rank shift and reclassification

Nominalization as functional reclassification A multifunctional account of nominalization

The role of agnation in the analysis of nominalization Networks of agnation

Agnation as a descriptive heuristic

The schematic nature of agnation

Towards a theoretical-descriptive approach to nominalization: Conclusion Chapter 4:.The functional organization of nominal and clause 1 2 2.1 2.11 2.1.2 2.2 2.3 243.1 2.3.2 2.343 24 2.5 3

From noun type specification to nominal From process type specification to clause The clausal head

Finite vs atemporal clausal heads /

The clausal head, Objects and Complements

Instantiation and quantification of the clausal head Clausal grounding

The finite element Person deixis

Clausal grounding: Conclusion

The speech-functional role of the Subject

The functional organization of nominal and clause: Con-

clusion

Nominalization as functional reclassification

Part I Deverbal -er nominalization

Chapter 5: Deverbal -er suffixation: Towards a descriptive position

Introduction

Trang 9

-er nominalizations as natural symbolic units -er suffixation as a complex conceptual category

The verbal base of deverbal -er nominals

Non-agentives as external arguments -er nominals profiling oblique participants

-er nominals based on unaccusatives Middle-based -er nominals

-er nominalizations as Subject names Lexicalization versus ad hoc nominalization

Towards a coherent account of deverbal -er nominalizations Chapter 6: The middle construction VaR AONE Ne

TỐ Middle formation: A state of the art The representational versatility of the middle formation

Towards a semantic typology of middle constructions Middle constructions and modality

The modality of ‘letting’

The middle as ‘letting’ construction Conclusion Chapter 7: A multifunctional approach to -er nominalization WN ee ee CC 3.1 3.1.2 3.1.3, 3.2 3.2.1 322 3.2.3 3.3 Introduction Representational semantics Agentive -er nominalization Non-agentive -er nominalization

A semantic typology of non-agentive -er nominalizations

The constructional properties of -er nominalization Lexicalized -er nominalizations

Non-agentive -er nominalizations Agentive -er nominalizations

A note on instrumental -er nominalization Ad hoc nominalizations

Phoric ad hoc nominalizations Non-phoric ad hoc nominalizations Ad hoc -er nominalization: Conclusion

Trang 10

Part Wil Factive nominalization

Chapter 8: Factive nominalization: Towards a descriptive po-

4.1 4.2,

sition Introduction

Factivity as truth presupposition Factivity as embedded projection Embedding vs taxis Projection vs expansion Facts as embedded projections Conclusion Delineation of the fact category Ambiguity.and vagueness

Gerundive nominalizations: Acts or facts? Towards an internal, nominal analysis Halliday’s dependency analysis

Type vs instance nominalization

Chapter 9: A functional analysis of factive nominalizations as 1.1 1.2 241 2.2 2.3, 3.1 3.2 3.21 3.2.2 nominal constructions

that-factives: From finite clause to proper name strategy Downranking as functional reclassification

The nominal paradigm

the fact that-constructions: A case of apposition

The structural characteristics of the fact that- constructions

The fictional properties of the fact that-constructions the fact that-structures: Conclusion

Gerundive facts: From atemporal clausal head to common noun or proper name strategy

From atemporal clausal head

.- {0 common noun or proper name strategy

Trang 11

3.2.3 The nominal functioning of gerundive nominalization:

Conclusion 239

3.3 Factive gerundive nominalization: Integrating the clausal

Trang 13

Introduction

The focus of the present study is on deverbal nominalization in English The analysis of nominalization that is presented in it has mainly been prompted by two observations First, the existing descriptive analyses of nominalization fail to embed the peculiarities of specific nominalization types in a coherent and systematic theoretical account of the fundamental constructional mechanisms that underlie nominalization in general Sec-

ondly, the description of nominalization systems tends to be biased towards

the clausal categories which they realize, to the neglect of their nominal characteristics Following Lees’s (1960) influential analysis of nominalization in English, nominalizations are still basically regarded as clause-like constructions used in nominal units in an unfathomable way The absence of a coherent theoretical perspective on the process of nominalization has led to descriptive fragmentation Morphological processes of nominalization such as -er or -ee suffixation, for instance, seem hardly relatable to nominalization types which incorporate clause-like structures, as in [His answering the phone] surprised me Also, the one-sided, clause-based na-

ture of most descriptions has elucidated only part of the lexicogrammatical

and semantic properties of nominalizations

These problems reflect deficiencies of theory as well as of description A central theoretical problem is, for instance: how can we, given the pre- vailing emphasis on structural constituency, model nominalization, which seems to elude and transcend ordinary constituency relations? A central descriptive problem is: which features of clause and nominal enable the reclassification from the one into the other that is generally held to be involved in nominalization? Most attention so far seems to have gone to the reclassification of verb into noun, and the ‘reification’ of a situational ref- erent into an entitized one that seems to go with it However, as a reclassification mechanism, this is too thin There must be other factors, mainly analogies between clause and nominal, which motivate the possibility of reclassifying a clausal configuration into a nominal one

To tackle the theoretical as well as the descriptive issues, this study is composed of a general theoretical-descriptive part (Part I) and two case studies (Parts II and [) Part I presents the theoretical and descriptive con-

Trang 14

ap-proach can account for the lexicogrammatical properties and meanings of specific nominalization systems Part II zooms in on the system of deverbal -er nominalization (e.g., remind -» reminder) and in Part II, three types of

factive nominalization are considered (viz [That he killed her] surprises

me, [The fact that he killed her] surprises me and [His killing her] sur-

prises me)

The main theoretical motif in this study is that nominalization strongly

calls for a functional rather than purely structural approach Crucially, the

reclassification from verbal/clausal starting point into nominal class is func-

tional in nature: nominalization involves the functional reclassification of a

particular level of assembly in the organization of the clause into a nominal structure A nominalization always adopts external nominal functions, but, as will become clear, it may also acquire internal nominal functional organization The reason why the identification of nominalizations as con-

structions of the nominal class requires a functional view on composition is that nominalizations do not always have discernible structural components

for each function which they realize: as in ordinary nominals, the internal functions which nominalizations realize are not necessarily reflected iconically in their structure It is therefore only a functional analysis which can

shed light on what nominalized structures have in common with the iconi-

cally and non-iconically coded members of the nominal class in general Moreover, it is only by adopting a functional viewpoint that one can relate nominalizations which contain part of a clause or a full clause (e.g., her

signing the contract, that she signed the contract) to nominalizations at

word level (e.g., baker) Like nominalizations based on a simple verb stem (e.g., bake ~» baker), nominalizations containing a clausal unit reclassify

this clausal unit into a unit with nominal functional behaviour

Because the analysis of nominalization requires a radically functional approach, I will in Chapter 2 discuss two linguistic frameworks which have shown a particular interest in the functional nature of language, i.e Cogni- tive Grammar (as developed in Langacker 1987a and 1991) and Systemic- Functional Grammar (Halliday 1994) I will go more deeply into some of their basic tenets, such as, firstly, the interaction between language usage and the language system (Section 1); secondly, the natural symbolic relationship between lexicogrammar and semantics (Section 2); thirdly, the

‘intrinsic’ functional nature of linguistic structure and the need for a functional approach to the notions of ‘rank shift? and ‘class’ (Section 3);

Trang 15

related structures and options in which both Cognitive Grammar and Sys- temic-Functional Grammar situate a construction) (Section 5)

In Chapter 3, I will show how these basic theoretical assumptions enable us to identify some of the lacunae and weak points in the existing ap- proaches to nominalization Importantly, these principles also point out a possible course of action in the description of nominalized constructions Going through the main theoretical divisions of Chapter 2 again, I will argue, firstly, that nominalization patterns have to be categorized both by schematization and extension It also has to be taken into account that nominalizations may display varying degrees of entrenchment or automatization: they may be Jexicalized and function as fixed expressions in language, or they may only be established as grammatical patterns or schematic units, without themselves being included as conventional units in the language system (Section 1) Secondly, I will stress the importance of viewing nominalizations, like all constructions, as natural symbolic units of which the meaning is encoded in the lexicogrammar (Section 2)

Thirdly, I will argue that nominalizations have to be viewed as functional configurations which are situated on a particular level or rank of functional organization (Section 3) As such, they find themselves in between a ‘higher’ rank in which the nominalization itself serves a function

and a ‘lower’ level that constitutes the nominalization’s own internal or-

ganization A full description of a nominalized structure requires an analysis of its external, synthetic functioning, of its internal, analytic functional outlook, and of the way in which the external and internal functions are combined in the construction itself Put differently, some nominalized constructions may represent an at first sight inexplicable mixture of external, nominal behaviour and internal, clausal categories, but the nominalized structure itself succeeds in combining these seemingly divergent functions

into one construction and therefore has to be accounted for Moreover, for a

number of nominalizations, we need a concept of downranking, which allows the possibility of functional configurations occurring at a level of organization which is not the one on which they normally function: clause- like nominalizations such as that-clauses can then be identified as having been downranked from clausal to nominal level while preserving part of their internal clausal outlook By the same token, downranking entails its own form of reclassification and involves external as well as internal re-

classification: by being reclassified, the clausal structure takes on the

grammatical and semantic properties inherent to the nominal unit Failure to recognize the reclassification involved in the rankshifted type of nomi-

nalization has been one of the main gaps in the description of nominaliza-

Trang 16

Fourthly, nominalizations are best described from a multifunctional perspective They encode representational semantics, they are related to the ground and they serve textual functions Their import cannot therefore be reduced to their representational semantics (Section 4) Finally, I will also argue that, if it is used systematically, the paradigmatic relationship of agnation forms an important descriptive heuristic in the analysis of nominalization patterns (Section 5) It helps to shed light on the clausal categories that are present in nominalization, as well as on the similarities and differences between distinct nominalization types such as factive that- nominalizations, the fact that-constructions and gerundive factives

The theoretical principles discussed in Chapters 2 and 3 reveal that an

insightful description of the functional properties of nominal and clausal organization is needed before one can tackle those of nominalized constructions In Chapter 4, therefore, I will complete the theoretical-descriptive

framework by elaborating on the internal functional organization of nomi-

nals and clauses I will argue that nominalization or the functional reclassification from processual starting-point to nominal construction is possible due to the fundamental correspondences that exist between the functional organization of the nominal and that of the clause: clausal categories can be mapped onto what is basically a nominal structure because the functional organization of clauses and nominals is to a large extent equivalent (Lan- gacker 1991) Chapter 4 will be organized around the functional categories which Langacker (1991) identifies for the nominal and the clause, i.e those of type specification, instantiation, quantification and grounding I will first

demonstrate the relevance of these functional categories to the description

of clauses and nominals (Sections 1 and 2) In a third section, then, I will formulate a tentative analysis of the basic constructional mechanisms that underlie nominalization I will argue that the integration of clausal and nominal properties in nominalized constructions develops along the lines of the functional categories of type specification, instantiation, quantification and grounding

Parts I and Ill of this volume will then present two descriptive case studies The nominalizations which I will analyze in them broadly cover the

spectrum of clausal levels of assembly to which nominal reclassification can apply: they include nominalizations derived from a simple verb stem, i.e -er nominalizations (e.g., baker, bestseller) (Part If); nominalizations based on what I will call an atemporal clausal head (e.g., signing the contract) and nominalizations which are centred on finite clauses (e.g., that she

signed the contract, the fact that she signed the contract) (Part II) The

Trang 17

Each case study will start with a chapter in which I look at the main in-

sights thus far offered in the literature and establish my own descriptive position In Chapter 5 an overview will be given of the literature on deverbal -er nominalization; Chapter 8 will discuss the most interesting analyses

that have thus far been proposed for the system of factive nominalization

In my own descriptive analysis of -er derivation and factive nominalization (developed in Chapter 7 and Chapter 9 respectively), I will attempt to show that it is possible to move beyond existing descriptions and provide an answer to some of the long-standing moot points in the analysis of specific nominalization types if the functionally-oriented approach which I set out in Part I is systematically applied Importantly, my analyses will focus not on the external functioning of the nominalized constructions, but on their internal functional organization, i.e on the clausal and nominal categories which they integrate

Apart from the general functional perspective which I will take and the

central role which I will assign to the functional categories of the nominal

and the clause, two main motifs can be said to run through both my descriptive case studies First, my analysis will be based on the assumption that the semantics of a construction is encoded in its lexicogrammatical properties and can therefore be revealed through a careful analysis of those properties

I will, in other words, not only present a detailed analysis of the lexico-

grammar of -er nominalizations and factive nominalizations; I will also try to shed light on the basic meanings which these nominalization types encode

Secondly, in each of these studies, the description of the syntagmatic integration of clausal and nominal properties in the internal organization of the nominalized constructions is supplemented with observations about the

paradigmatic relations which the nominalized structures hold In other

words, the syntagmatic properties of nominalizations will be systematically considered against the background of the larger system network which they

form part of and which they select from Not surprisingly, the paradigmatic

relations that will turn out to be most valuable to the elucidation of nominalizations are those with clausal structures and with nominals First, due to the clausal nature of their starting point and the clausal categories which they embody, nominalizations tend to be related to clausal structures The

identification of related or ‘agnate’ clause types will be shown to play a

crucial role in the analysis of nominalized constructions In my description of -er nominalizations (Chapter 7), I will thus elaborate on the systematic

relationship that exists between -er suffixation (e.g., bestseller) and the

Trang 18

(presented in Chapter 6) sheds a new light on the system of -er nominalization

Apart from being related to clausal structures, however, nominalizations also connect with nominal constructions: by being made to function within

the nominal paradigm, nominalizations enter into relation with other nomi-

nal structures, nominalized as well as non-nominalized The gerundive type

of factive nominalization, that-factives and the fact that-constructions can

thus be considered as nominalized structures which are systematically related to each other and are linked up within the system of language In my analysis of factive nominalization, I will show that it is only by taking into

account the properties of each of them that one can come to a more accurate description of the overall system of factive nominalization and its fac-

tive reading (Chapter 9) Importantly, the various types of factive nominalizations do not only relate to each other: as nominal constructions, they also

enter into relation with specific types of non-nominalized nominal structures Once again, the elucidation of the precise paradigmatic relations which each nominalization type has with ‘ordinary’ nominal constructions will be argued to be vital to the elucidation of its properties In the Conclu-

sion (Chapter 10), finally, I will attempt to answer the question as to what

Trang 19

Theoretical assumptions

Nominalization represents a theoretical challenge and has to be situated in an overall theory of the language system It is therefore essential to first clarify the basic theoretical assumptions on which my analysis of nominalization is based This is even more important as my approach is eclectic and some of the terminology which I use requires clarification In this chapter I will therefore zero in on the linguistic system and discuss the various ‘levels’ that can be distinguished in it: in a first section, I elaborate on the language system as such and its relation to language usage; the second section

deals with the various strata that can be distinguished in language and fo-

cuses on the natural relation between lexicogrammar and semantics; in Sec- tion 3, I deal with the ranks or the levels of organization that can be distinguished within the grammar and I introduce two of the notions that are crucial for the analysis of nominalization, viz reclassification and rank shift Section 4 focuses on the multifunctional approach which I advocate for the description of nominalization patterns; and, finally, Section 5 points out the

heuristic value of considering the relations that exist between different con-

structions

1 Language: The relation between system and usage 1.1 A usage-based approach to language

The analysis of nominalization set out in this study is rooted in a view of language which is at once functional and cognitive It is functional in that it builds on the assumption that the nature of the language system is “closely related to the social and personal needs that language is required to serve”

(Halliday 1970b: 142): the language system is viewed as functional with

Trang 20

functional level of organization in language will be discussed in more de-

tail)

Apart from viewing language as functional, I also adhere to the claim that the language system — and the constructions in it ~ is determined by certain cognitive abilities of the language user: the language system is viewed as an integral part of human cognition (Langacker 1987a) Lan- guage is held to refer to “those aspects of cognitive organization in which resides a speaker’s grasp of established linguistic convention”, and the linguistic abilities of the language user are believed to “not necessarily constitute an autonomous or well-delimited psychological entity” (Langacker 1987a: 57): there are a number of basic psychological phenomena involved in language use which are essential to the language system, without being limited to it (Langacker 1999) Because the cognitive processes which they embody equally contribute to their make-up, an analysis of linguistic struc-

tures should articulate what we know of cognitive processing, i.e it must

have “psychological reality” (Langacker 1987a: 42)

The language system is, however, not only shaped by the functions

which it serves and by the cognitive abilities of the ones who use it: it also

itself sanctions language use “Usage events” (Langacker 1987a: 66) or

actual “instantiations” of language (Halliday 1992: 20; Langacker 1999:

99) are sanctioned by the restrictions and options of the system of language underlying them: they are ‘systemically motivated’ (Langacker 1987a: 426) As Hjelmslev (1961: 39) puts it, an actual instance of language (which he calls a ‘process’ or ‘text’) only “comes into existence by virtue of a system’s being present behind it, a system which governs and deter- mines it in its possible development” This is not to say that language is a

generative or constructive system which generates well-formed, novel structures as its output: it is not the language system which is responsible

for assembling novel expressions Construing language is viewed as “something that speakers do” (Langacker 1987a: 65), which ties in well with Hjelmslev’s ‘processual’ interpretation of language use (Hjelmslev 1961: 39) Actual language use is “a problem-solving activity that demands a

constructive effort and occurs when linguistic convention is put to use in

specific circumstances” (Langacker 1987a: 65) In short, the language system may provide the necessary “symbolic potential”, but it is left for the speaker to recognize this potential, ie to “exploit it in a fashion that re- sponds to all the varied constraints inherent in the situation” (Langacker

1987a: 66)

Trang 21

lin-guistic structure” and the “source of language change” (Langacker 1987a: 70) If we are to elucidate particular constructions in the language system, therefore, we must consider them from a usage-based perspective, i.e we must attach substantial importance to their instantiations in actual language use Hjelmslev (1961: 11) speaks of the “empirical principle”, which states that “a theory must be capable of yielding, in all its applications, results that agree with so-called (actual or presumed) empirical data” A grammar should be at once a grammar of the language system and a grammar of the uses which it is put to (Halliday 1994: xxii); it must work bottom-up rather than top-down (Langacker 1988: 132)

Langacker’s cognitive model of grammar is aimed at modelling the language system in a usage-based and psychologically plausible way (Lan- gacker 1987a, 1988, 1990, 1991, 1999) It describes the interface between system and usage, and the insight into the organization of the language system gained from it in detail and with reference to the various psychological processes that play a role in it Because Cognitive Grammar posits a fundamental equivalence between, on the one hand, the interface between system and usage, and, on the other hand, the organization of the language system itself, — both of which are argued to be modelled on the same, essentially cognitive principles ~ I feel that it goes a long way towards the formation of a usage-based theory of language I will therefore start with a

brief discussion of Langacker’s analysis of the interaction between lan-

guage use and the language system, and of the model which he proposes to account for the organization of the language system A more detailed analysis of particular aspects of the language system will follow in later sections, together with a discussion of some basic, functional theoretical

constructs which I consider to be compatible with Langacker’s model and

prerequisite for a truly usage-based account of language

1.2 Cognitive Grammar: The interface between system and usage

Langacker considers the following psychological abilities of the language

user to be crucial for language use: symbolization, composition, comparison/ categorization and schematization

Trang 22

which associates the semantic unit [OPENER] with the phonological unit [opener])

Composition refers to the ability to integrate two or more (component) structures to form a composite structure (Langacker 1999: 94) Composi- tion gives rise to relationships on the syntagmatic plane of language (Lan- gacker 1987a: 75) An example of a composite structure is can opener, which combines the symbolic units [[CANJ/{can]] and [[OPENER]/[opener]], which in tum consists of [[OPEN)/[open]] and

[[-ERI/I-enl

Apart from symbolization and composition, the abilities of comparison and schematization are also relevant to language use According to Lan- gacker (1987a), the language user solves the problem of coding a detailed, context-dependent conceptualization in terms of a linguistic expression by making comparisons or categorizing judgements (Langacker 1987a: 65—

73):_he/she assesses whether.a specific expression can be categorized as a _

member of the category defined by a conventional linguistic unit or a unit that is widely shared by the language users of a speech community A usage event is therefore always systemically motivated or “structured and evaluated with reference to the conventional units of a grammar” (Lan- gacker 1987a: 426)

Because structures which at first sight appear very different, may be “quite comparable in a coarse-grained view” (Langacker 1999: 93), catego-

rization necessarily involves the ability of schematization: to determine

whether a usage event satisfies the specifications of specific conventional units, one may have to abstract away from certain points of difference and portray the distinct structures with lesser precision and specificity (Lan- gacker 1999: 93) A usage event can then either be identified as elaborating/instantiating a category, or it can be said to extend it When a linguistic category is elaborated or instantiated by a usage event, the latter conforms to the specifications of the category but is, schematically speaking, “characterized in finer-grained detail” (Langacker 1999: 93) An elaborative or instantiating novel usage is conventional or — in more traditional terminology — grammatical: it is conforming to the linguistic convention embodied by the linguistic unit, i.e it is fully sanctioned by it When, on the other hand, a discrepancy exists between the category that is defined by the linguistic unit and the usage event which is related to it, the category is extended and the usage event is either i//-formed or non-conventional

In general, a novel usage, irrespective of whether it merely elaborates or

extends a conventional unit, can become conventionalized or acquire unit

Trang 23

equipment, its meaning was extended This extension has by now achieved unit status because of its frequent use and the entrenchment resulting from it (Langacker 1999: 108) Entrenchment or the automatization of certain structures is a matter of degree and the boundary between units and non- units fluctuates continually because “every use of a structure reinforces it

and entrenches it more deeply, whereas non-use has the opposite effect”

(Langacker 1999: 100)

1.3 The language system as network

Crucially, the relationships of symbolization and composition, comparison and schematization, which the language user establishes when construing language, are also claimed to determine the internal organization of the language system itself The terminology used to describe language use consequently applies to the language system as well

The language system consists of three types of units: firstly, semantic (e.g., [PRINT]), phonological (e.g., [p]) and symbolic units (the latter associate a semantic with a phonological unit, ¢.g., [[PRINT}/[print]]); secondly, schematic symbolic units or grammatical patterns (e.g., [[V/ ]- {[-ER\/[-er]}), and, thirdly, the categorizing relationships of elabora- tion/instantiation, extension and mutual similarity (Langacker 1988: 134, 147) These units have acquired unit status because they are entrenched: the language user can employ them “in largely automatic fashion” because they constitute ‘pre-packaged assemblies’ (Langacker 1987a: 57) They are conventional in that they are shared and recognized as being shared by a substantial number of language users (Langacker 1987a: 62)

The simplest kind of symbolic unit is the morpheme, “in which a se-

mantic and a phonological structure participate as unanalyzable wholes in a

symbolic relationship” (Langacker 1987a: 58; see also Halliday 1961) The combination or (syntagmatic) integration of basic symbolic units leads to the formation of a composite symbolic structure or a grammatical construction The term ‘grammatical construction’ applies to the component structures which the composite structure consists of, to the relationship of integration that exists between them and to the resulting composite structure (Langacker 1987a: 277)

Trang 24

schemata, which capture generalizations and serve as ‘templates’ for the

assembly of novel expressions (Langacker 1988: 148): the schema [[V/ ]- [{-ERV/-er]] is thus instantiated by [[OPEN/open]-[-ER/-er]] Because it is believed that a speaker’s knowledge of a construction is not restricted to a single rule or schema standing in isolation, but also embraces the speaker’s “tmowledge of how the rule is ‘implemented’ with respect to more specific structures” (Langacker 1988: 151), the more specific instances of constructional schemata are argued to be part of the system of language as well As Langacker puts it (1987a: 29), it is “gratuitous to assume that mastery of a rule like N + -s, and mastery of forms like beads that accord with this rule, ate mutually exclusive facets of a speaker’s knowledge of his language; it is perfectly plausible that the two might sometimes coexist” In addition to being composite, a grammatical construction is therefore also complex: it does not reside in a single structure, but rather in “a family of structures

connected by categorizing relationships” (Langacker 1988: 149)

The ‘family of structures’ that makes up a complex category is best viewed as a schematic network or a network in which “linguistic structures of any kind and any size are linked in pairwise fashion by categorizing relationships” (Langacker 1999: 103): the superordinate nodes in the network are then schematic and the subordinate nodes are either subschemata or more specific structures which elaborate/instantiate or extend the schema

An example of a schematic network is given in Figure 1 (based on Lan- gacker 1988: 131): SCHEMA THING PL x -§ DOG PL TREE PL dog ~§ tree -S

Figure 1, A schematic network

Trang 25

in-clude schematization among the categorizing relationships because a category’s extension from the prototype tends to be “accompanied by a certain amount of upward growth, as schemas are extracted to generalize over a more diverse array of category members”

The vision of the language system that emerges through Langacker’s

model “is one of massive networks in which structures with varying degrees of entrenchment, and representing different levels of abstraction, are linked together in relationships of categorization, composition, and symbolization” (Langacker 1999: 95) The description of a construction involves the elucidation of the relationships which the construction embodies as well as the relations which it contracts with other constructions in the network,

In the following sections, I will zoom in on specific aspects of these relationships I will start with the relationship of symbolization and briefly

discuss the natural or non-arbitrary nature of the symbolic relation be-

tween lexicogrammar and semantics (Section 2) In Section 3, I will consider the syntagmatic relations that exist between the components of a construction and lay out the main properties of a radically functional view of composition and classification Section 4 will go more deeply into the various types of functional relationships that are at work in linguistic patterning and it will discuss Halliday’s tristratal functional interpretation of linguistic structure Finally, in Section 5, I will look at the various types of paradigmatic relationships that obtain between specific constructions in the language system and I will point to their value as a descriptive ‘heuristic’ I will argue that, apart from the categorizing relationships of elabora- tion/instantiation, extension and mutual similarity which Langacker dis- cusses, the relation of agnation, introduced in Gleason (1965), should be included in the description of linguistic patterning as well

2, The natural symbolic relation between lexicogrammar and

semantics

2.1, A natural description of language

The idea that a description of language, apart from being usage-based, should also be natural is prominently present both in Langacker’s cognitive model of grammar and in Halliday’s systemic-functional approach In general, the notion of ‘naturalness’ refers to the concern to develop a description of language which “deals with data in their own terms, with full regard

Trang 26

phe-nomena” (Langacker 1987a: 13) A natural model of language is a model which respects the nature of language as a complex system In Langacker (1987a), a natural analysis of language is, for instance, argued to accom- modate not only discrete, but also non-discrete aspects of language structure; it is said to carry the simplifications and idealizations which are sometimes needed to begin with the analysis of language not too far, so as not to lose sight of the actual complexity of the subject matter; and, finally, a

natural description is claimed to operate with substantive constructs, or

constructs which are sufficiently clear to be elucidated “with a certain amount of precision and detail” and which are real in that they correspond to “some actual feature of linguistic organization” (Langacker 1987a: 22)

One of the unnatural distinctions which circulate in linguistics is that be-

tween lexis and grammar, and, related to it, the assumption that linguistic theory needs the constructs of lexicon and syntax to describe it Syntax is

thought of as the “domain of generality and regularity, of productive rules

with fully predictable outputs”, and the lexicon is viewed as “the domain of irregularity, idiosyncrasy, and lists” (Langacker 1987a: 26) The distinction between lexicon and syntax has, for instance, been argued to underlie the

differences between nominalizations such as John’s refusing the offer and

John’s refusal of the offer (Chomsky 1970: 187): the latter, so-called ‘derived’ type of nominal has been argued to form part of the lexicon, mainly

because the semantic relation which it has with the verb from which it is

derived tends to be idiosyncratic Nominals of the former, i.e gerundive, type, by contrast, are claimed to be part of the syntactic component of language, being derived from clauses by means of ‘syntactic transformations’ without fundamental changes to the clausal meaning (Lees 1960; Chomsky 1970): their meaning is argued to be related to that of the underlying clause in a regular way

Gerundive nominals of the ‘mixed’ (Chomsky 1970: 215) or ‘action’ type (Lees 1960), such as John's refusing of the offer, however, make clear

that the constructs of ‘lexicon’ and ‘syntax’ lack substance and cannot be strictly distinguished between Like gerundives of the type John’s refusing

the offer, the meaning of action nominalizations seems regular, which suggests that they are ‘syntactically’ derived; and yet, action gerundives are not fully productive (e.g., *the being of a woman, *his having of a car) and they have the internal structure of a nominal, two features which are alleged to classify them with lexically-based nominals The problematic use of the constructs of ‘lexicon’ and ‘syntax’ in the analysis of nominalizations thus

illustrates that there is no reason to assume that the distinction between

Trang 27

“most delicate grammar” (Halliday 1961: 267), its relation with grammar is argued to take the form of a cline or a continuum (an idea which is also found in Langacker’s work) and the term /exicogrammar is used instead of

grammar

Apart from the artificial distinction between lexis and grammar, the re-

lationship that has most often been misrepresented is the fundamental rela-

tion between lexicogrammar and semantics: grammar and semantics have often been conceived as autonomous entities or separate ‘components’ of language (Langacker 1987a: 12; see also Halliday 1988) A description of language which is truly natural, however, can only be arrived at when the relation between grammar and semantics is recognized to be natural or non-arbitrary

2.2 The natural or non-arbitrary relation between grammar and meaning Language is essentially symbolic in nature in that it consists of an open- ended set of linguistic symbols or signs, “each of which associates a semantic representation of some kind with a phonological representation” (Lan- gacker 1987a: 11) In the case of most of the lexical items which are not further analyzable into morphemes, the symbolic association between meaning and form is arbitrary, i.e there is no natural relation between the meaning of the lexical item and the particular phonological form it is mapped onto (Saussure 1972) Well-known exceptions to the arbitrariness of unanalyzable lexical items are instances of onomatopoeia, such as hiss and buzz

One of the basic tenets of both cognitive and functional models of lan-

guage is that not only lexical items but also grammatical patterns are signs or symbolic combinations of meaning and form As Langacker (1987a: 12) puts it, “morphological and syntactic structures themselves are inherently symbolic” and this “above and beyond the symbolic relations embodied in

the lexical items they employ” A grammatical construction is thus sym-

bolic not only in that it contains symbolic elements: the patterns along which these symbolic components are integrated is held to be symbolic as

well Grammatical patterning itself is believed to associate a particular

meaning with a particular form.' Most importantly, and unlike in most simple lexical items, the symbolic relationship established in grammatical patterns is held to be non-arbitrary or natural, with the grammar ‘encoding’ or ‘realizing’ the semantics (Halliday 1994: xvii)

However, without further qualification, the Saussurean, bipolar interpre- taton of the linguistic sign, which p

TRƯỜNG ĐẠI HỌC NGOẠI NGỮ- PHQGHN

TRUNG TÂM HỌC LIỆU

E.09444

Trang 28

nological form, seems unfit to capture the non-arbitrary nature of gram~

matical structure: “A semiology in which meaning is said to be directly associated with phonology cannot be sensibly linked to a theory of grammar” (Davidse 1991: 1) As already hinted at in Saussure’s work, it is necessary to posit a third or intermediate stratum, in between the semantic and phonological ones, to account for the non-arbitrary nature of grammatical struc-

ture Halliday (1987, 1992) has used Hjelmslev’s model of the linguistic

sign in an attempt to clarify the relation between lexicogrammar and se-

mantics.”

Hjelmslev (1961) takes over the Saussurean concepts of content and expression to talk about the linguistic sign and proposes that the content plane

of the linguistic sign has to be further analyzed into content-form and con-

tent-substance Content-form, which Hjelmslev describes as language- specific encoding (Hjelmslev 1961: 52), represents according to Halliday

(1987) the lexicogrammatical stratum of language It is by passing through

this lexicogrammatical stratum that general communicative ‘purport’ (Hjelmslev 1961: 51) or meaning which “can be ‘translated’ between languages” (Davidse 2000a: 33) is formed into content-substance, i.e into sign-specific meanings, which can then be mapped onto a particular phonological form (Hjelmslev 1961: 52)

The linguistic sign is, in short, not an expression that points to a content outside the sign Rather, it is “an entity generated by the connexion be-

tween an expression and a content” (Hjelmslev 1961: 47), with the content-

level further divided into a lexicogrammatical and a semantic plane Cru- cially, it is only by virtue of content-form or lexicogrammar that content-

substance or semantics exists: content-substance appears by “the form’s

being projected on to the purport, just as an open net casts its shadow down on an undivided surface” (Hjelmslev 1961: 57) Or, as Langacker (1987a: 98) puts it, semantic structure is conceptual structure that is “shaped for symbolic purposes according to the dictates of linguistic convention” The relationship between content-substance and content-form is therefore natural or non-arbitrary: the meaning of a linguistic sign is encoded or realized via the lexicogrammar of the particular language it belongs to Any attempt to construe semantics and lexicogrammar as discrete components of the language system is, consequently, at odds with the very nature of the linguistic sign

To establish the meaning of ‘symbolically complex’ units (Langacker

1987a: 82) ~ i.e units which integrate two or more symbolic units into a grammatical construction —, then, we have to analyze their lexicogrammar

in detail Such an analysis mainly involves an elucidation of various ‘con-

Trang 29

syntagmatic context of the linguistic sign and the relations that are established through composition In Section 5, I will discuss the systemic context of linguistic signs or their position “within the schematic networks that col- lectively constitute the grammar of a language” (Langacker 1987a: 401)

3 Constructions: Composition, classification and function®

the components of a complex expression should not be thought of as

providing the material used to construct it — their function is rather to cate- gorize and motivate facets of the composite structure (Langacker 1991:

508)

Owing to the structural diversity of nominals, a universally valid schematic characterization must be couched in terms of meaning and semantic function rather than formal properties such as constituency (Langacker 1991:

54)

In this section, a radically functional approach to the analysis of constructions is described, which views them as functional configurations, the components of which are essentially component functions and not necessarily component structures A radically functional analysis of composition thus posits that it is functions rather than, for instance, constituents, which form the overarching motivating principle in composition The main properties

of a radically functional approach to composition will be sketched out as follows: in a first part (Section 3.1), I will focus on the role of functions in

the analysis of composition and posit that the components of a construction are best viewed as component functions

Trang 30

3.1 A functional approach to composition

In Langacker’s (1987a) model of language, composition is said to involve components or component structures which are turned into a composite structure ot a grammatical construction The notion of ‘component’ captures the compositional relation by which components are integrated into a composite whole Like the terms ‘construction’ and ‘assembly’, it hints at the active nature of the constructive efforts of the language user “Compo- nent’ is preferred to the notion of ‘constituent’, which, as pointed out by McGregor (1997), is too much linked to one specific type of syntagmatic relationship, viz that of constituency (i.e “the syntagmatic relationship between a part and the whole to which it belongs”, McGregor 1997: 21) The term ‘constituent? moreover suggests that the components of a construction are building blocks out of which the composite structure is formed

(Langacker 1987a, 1999) A ‘building-block’ view on composition, how-

ever, fails to account for those constructions in which “components are only partially discernible (or even indiscernible) within the composite whole” (Langacker 1999: 152) Even though composite structures like these are non-prototypical or ‘marked’, they form an essential part of the

nature of language, and should therefore be included in an account of com-

position that aims to be natural

It is the description of constructions in which not all components are discernible that illustrates best the assets of a functional approach to composition Components which are only partially discernible or indiscernible cannot be referred to as component structures As Langacker’s discussion

of non-canonical nominals makes clear, such components are best described in functional terms, i.e as component functions which are not realized

by specific component structures A functional analysis of such non- canonically coded constructions not only sheds light on the constructions themselves, but it also succeeds in revealing what they have in common with other, canonically coded members of their class

Let me illustrate this briefly (a more elaborate account of the functional interpretation of nominals given in Langacker 1991 follows later in Chapter 4) In a canonically coded nominal construction, Langacker argues, each

semantic function that is characteristic of nominals is associated with a dis-

Trang 31

to realize functions that are similar to those encoded in those three black cats: it profiles a single instance of the schematically specified type “human female’, is inherently definite, and its designatum is identified as being distinct from the speech-act participants (Langacker 1991: 148) It is thus the

functional analysis of she which establishes it as a member of the class

‘nominals’ and links it up with other nominal constructions (From a strictly structural perspective, she seems to have hardly anything in common with nominals like those three black cats.)

The analysis of a construction thus essentially involves elucidating the functions which it expresses (e.g., Haas 1954; Halliday 1994; McGregor 1997) and constructions can be interpreted as ‘configurations of functions’ (Halliday 1994) In the following section, I will zoom in on the unique role which functions play in the description and classification of constructions

3.2 Levels of functional analysis

In Haas (1954), it is claimed that a functional analysis of composition is necessarily two-directional: functional relations are said to characterize a construction both analytically and synthetically An analytic definition of a construction involves the analysis of the functions served by its components, whereas a synthetic definition refers to the functions which the.composite whole itself plays in a larger configuration (Haas 1954: 61) A con-

struction can, in this perspective, be described in functional terms exter-

nally (synthetically) as well as internally (analytically)

Apart from the external and internal functions of a construction, the construction itself should be considered as well, as the level at which the various external and internal functions are integrated.’ In this way, the construction itself constitutes a third level of linguistic analysis It is not “algo-

rithmically deducible” from its components, but must be treated in a non-

reductive way, “as a separate entity in its own right” (Langacker 1987a:

87)

Trang 32

labels a fixed number of layers in the hierarchy of constituents, such that any constituent can be assigned to one or other of the specified layers, or

ranks” (Halliday 1966b: 111) The various levels of organization that can thus be distinguished correspond to specific, determinate stretches of struc-

ture, characterized by a distinction in ‘size’, ie the sentence, clause, phrase, word and morpheme (Halliday 1961) These levels are claimed to be related in such a way that, “going from top (largest) to bottom (small- est), each ‘consists of” one, or more than one, of the unit next below” (al- liday 1961: 251).° The analysis of a construction, then, requires that it be accounted for at all ranks This is called the ‘requirement of total account-

ability’ in Halliday (1966b: 113) A clause, for instance, is analyzed into

phrases, which themselves consist of words, which, in turn, are made up of morphemes The rank scale allows for downward ‘rank shift’, whereby a

symbolic unit is used at a rank which is either equal to or lower than its

own rank, The use of a clause as nominal, as in (hat Caesar was deaaj was obvious to all (Halliday 1994: 264), for instance, is considered to be a

case of downward rank shift

The limitations of this “strong version” of the rank hypothesis (McGregor 1991: 121) and the need for refining and restricting it have been repeatedly pointed out (e.g., Matthews 1966; Huddleston 1988; McGregor 1991) Firstly, it has to be acknowledged that certain items, such as con- junctions (e.g., and, but), interjections (e.g., hey, eh) and adjuncts (e.g., surely, probably, yet) cannot be included in the rank scale (Matthews 1966, McGregor 1991), Secondly, the requirement of total accountability at all ranks is arguably more illuminating in some cases than in others As Mat- thews (1966: 103) points out, an utterance like Yes! consists of a morpheme assigned to sentence level, and analyzing it as ‘one sentence, which is one clause which is one phrase which is one word which is one morpheme’ does not seem to add anything to its analysis Many criticisms are, moreover, provoked by the ‘structuralist’ formulation of the rank scale in Halli- day (1961), where the rank scale was presented as being a rank scale of (canonically coded) structures

Nonetheless, the idea that constructions occupy a particular position on

a rank scale of symbolic units is valuable to the description of constructions, provided it is interpreted functionally and it is not detached from Hal-

Trang 33

the rank scale to so-called minimal or functional bracketing, that is ““brack-

eting together only those sequences that have some function relative to a larger unit” (Functional bracketing differs from ‘immediate’ constituent analysis, which brackets together units that may not have any function at all and which reveals “the order in which all the pieces are put together, pair by pair”, Halliday 1994; 23.)

From a functional perspective, the rank scale can thus be viewed as a hi-

erarchy of basically functional configurations and the analysis of a construction can be said to involve a description of its functioning in higher-

level configurations, a description of the lower-level functions which it em-

bodies and the integration of these two into a coherent analysis of the rank that lies in between, i.e that of the construction itself

3.3 A functional approach to classification

Symbolic units are typically assigned to particular classes because of their

(external) behaviour in larger configurations and, in some cases, because of basic semantic features which they have In functionally-oriented ap- proaches to language, it is usually pointed out that classes of symbolic units are best derived on the basis of the larger structural configurations in which they can occur, i.e on the basis of how they function in them (e.g., Halliday 1961; Quirk et al 1985; McGregor 1997) The class of the noun phrase or

nominal, for instance, is thus characterized as a class of symbolic units

which “typically functions as subject, object, and complement of clauses

and as complement of prepositional phrases” (Quirk et al 1985: 245) In Croft (2000: 16), syntactic categories, including classes, are argued to be

“construction-specific” or derivative only “from the constructions that de- fine them”

Apart from classification on the basis of external functioning, one also

Trang 34

the noun class, then, the distinction between the subclasses of count and

mass nouns mainly depends on “whether the profiled region is construed as being bounded within the scope of predication” (Langacker 1991: 18):

count nouns (e.g., ake) include the boundaries of what they designate in their scope of predication, whereas mass nouns (such as water) do not

The external, grammatical functioning of a symbolic unit and the highly schematic semantic characterizations of some basic classes have weighed heavily on discussions about grammatical classes Interestingly, internal properties have not: internally, the various instantiations of a particular class are typically considered to be too diverse to be generalized across Nominals, which vary from having a common noun as head, to being a proper name or a pronoun, seem to be no exception to this The internal structural diversity that is characteristic of many classes has, in fact, explic-

itly been said to render classification on the basis of internal properties im- possible Halliday (1961: 261) thus states that a class is “not a grouping of

members of a given unit which are alike in their own structure”: he (1961:

261) holds that, in terms of the rank scale, classes are derived ‘from above’ and not ‘from below’,

In what follows, an alternative approach to grammatical classes will be

presented which does build in the perspective from below In a first part, I will argue that a radically functional approach to classification includes the

internal functional organization of symbolic units, irrespective of whether

the internal functions are symbolized separately or not A functional ap-

proach to grammatical class which includes the internal functions in its analysis manages to identify symbolic units which are structurally widely divergent as members of one and the same class In a second part, I will

argue that classes which are functionally derived allow for a finer-grained

description of what can be said to constitute a prototypical instantiation of a class

3.3.1, An internal functional analysis of constructions

A radically functional approach to classification identifies classes not only in terms of their behaviour in larger functional configurations, but also by

the presence of a particular set of component functions More particularly,

it views classes as combinations of certain external and internal functions

The inclusion of internal functional properties in the classification of a

symbolic unit is in accordance with Haas’s claim that “many constructions can be adequately defined only by a combination of analytic and synthetic

Trang 35

(Haas 1954: 70) Until recently, it was difficult to see how such an analysis could be envisaged of the nominal, in which, in view of the topic of this study, we are particularly interested While generalizations that deal with the nominal’s external functioning are fairly readily arrived at (as, for instance, in Quirk et al 1985), internal functional generalizations are not This is not only because of the apparent disparity of the units that can function in the nominal slots of a construction, but also, and more fundamen- tally, because the question of the internal functioning of the nominal has tended to be approached from an exclusively structure-based perspective It is only Langacker’s (1991) analysis of the internal functional organization of the nominal that makes clear that the various instantiations of the class can be generalized across also internally Langacker’s analysis is pathfind- ing, both descriptively and theoretically, and adopts a view of the nominal which is very much ‘radically functional’ in that it recognizes that “seman-

tic function (rather than constituency) is the critical factor for understand-

ing their internal organization” (Langacker 1991: 51)

The semantic functions which, according to Langacker, are characteris-

tically realized by nominals are those of type specification, instantiation,

quantification and grounding (Langacker 1991) The “universally valid schematic characterization” which Langacker proposes for the nominal is that it profiles “a thing construed as an instance of some type and further incorporates some specification of quantity and grounding” (Langacker 1991: 54) Simple nouns provide nothing more than a type specification: they specify “the basis for identifying various entities as being representa- tives of the same class”, but are “not tied to any particular instance of that class” (Langacker 1991: 53) Full nominals such as the site, an excellent site and two convention sites in the Midwest, on the other hand, presuppose instantiation of the type in question and designate one or more instances In full nominals, “information is furnished concerning both the number of instances and their status vis-a-vis the speech-act participants” (Langacker

1991: 53)

Trang 36

(such as she, which was discussed earlier), proper names probably represent one of the most extreme cases of departure from the iconic type of coding: in proper names, Langacker (1991: 59) proposes, “type, instantiation, quantity, and grounding are conflated in a single expression” In the proper name Stan, for instance, a type specification is incorporated (that of ‘male human’), and because it characterizes a specific person, it presup- poses instantiation and quantification Grounding is subsumed as well, “for the nominal is definite and portrays the profiled individual as being uniquely apparent to the speaker and hearer on the basis of this name alone” (Langacker 1991: 59) Because these functions are fulfilled, the pro-

per name qualifies as a nominal, and this in spite of its non-canonical cod-

ing (Langacker 1991: 59) As Langacker (1991: 53) concludes, the functions of type specification, instantiation, quantification and grounding thus allow us to “profitably examine the organization of nominals from the

standpoint of semantic function, abstracting away from any details of struc-

tural implementation”

Classification by internal functional properties should be distinguished

from intrinsic or notional classification Notional classification makes ref-

erence to abstract cognitive events and, to make sure that all members of the class (including the less prototypical ones) instantiate the class schema, it is necessarily highly schematic (Langacker 19876: 54).’ As Langacker (1991: 52) himself points out, the characterization ‘thing or region in some domain’ applies to nominal predications “as a broadly defined class”: it is too abstract to distinguish simple nouns like site from full nominals like an excellent convention site, “each of which profiles a thing and qualifies as a noun in this inclusive sense of the term” (Langacker 1991: 52) The intrinsic classification of nominals is especially meant to distinguish them from

relational predications The functional description in terms of type specification, instantiation, quantification and grounding, in contrast, applies to

the internal organization of nominals and helps to identify the differences between nominal constructions and simple nouns (Langacker 1991: 53) Classification in general, then, is intrinsic as well as functional, and the functional criteria involve both the external functions served by the symbolic unit and the internal functions of its components

3.3.2 Classes as complex categories: Schema and prototype

Langacker’s functional description of nominals shows that it is feasible to

Trang 37

of some type’, for instance, is applicable to all the members of the class

‘nominal’ Put differently, all nominals instantiate the same schema, which “captures the pertinent generalization” (Langacker 1988: 130) All nominals are fully compatible with its specifications, but characterize them in finer detail or “elaborate the schema in different ways along various pa- rameters, to yield more precisely articulated notions” (Langacker 1987a: 68) By combining a schema with various instantiating structures, classes form complex categories, or categories residing “in a family of structures connected by categorizing relationships” (Langacker 1988: 149)

Among the categorizing relationships in a complex category, however, Langacker distinguishes not only categorization by schema, but also cate-

gorization by prototype or extension He (1987a: 371) views the two modes of categorization as “inherently related and describable as aspects of a uni-

fied phenomenon”, and his network model is a ‘synthesis’ of categorization by prototypes and by schemas Categorization by prototype offers a perspective on classification which is not so much alternative as complemen- tary to that of categorization by schema Both categorization by schema and categorization by prototype can be said to involve comparing a usage event or farget structure with a conventionalized unit or standard Instantia-

tion implies that all the specifications of the standard are satisfied by the

usage event The standard is then viewed as a schema which only differs from its instantiations in degree of schematicity or ‘granularity’ This is true of, for instance, the schematic characterization of nominals as ‘grounded instances of some type’ Extension of the standard, on the other hand, implies that the target structure is only partially sanctioned by the standard or deviates from it The standard is then conceived as the prototype of the category

It follows that a ‘natural’ approach to classification should not only look for schematic generalizations, but should also take account of the occur- rence of prototypical and less prototypical instances of a class (Langacker 1987a: 371) Instead of considering all members of a class as being fully compatible with a schema, categorization by prototype judges class mem- bership through a “perception of similarity’ (Langacker 1987a: 69), which

is based on what is conceived to be a prototypical instance of the class Pro-

Trang 38

With regard to the general classes of language, Langacker argues that they are “centered on a prototype that reflects a basic aspect of our every- day experience” (Langacker 1991: 521) The archetypal conception of ‘physical object’ is thus generally held to constitute the semantic prototype for nouns In his functional analysis of nominals, then, Langacker takes the structural way of coding as starting-point: prototypical are held to be nominals “whose structure most directly mirrors semantic function” (Langacker 1991: 143), ie nominals which reflect their component functions through distinct and overt levels of constituency.’ It follows that, in Langacker’s view (as opposed to, for instance, that of Van Langendonck 1999), prototypical nominals incorporate both a head noun and a determiner (Langacker 1991: 143), Examples of prototypical nominals are this cup, some milk, a big house, those three black cats Non-prototypical are nominals such as three from Toledo, the poor, for Harold to resign: they occur less fre- quently or “only in special grammatical circumstances” (Langacker 1991:

143)

Interestingly, by linking prototypicality to iconicity, Langacker’s analysis of the internal organization of nominals also allows one to distinguish degrees of non-prototypicality: some nominals, for instance, deviate from

the prototype in that the semantic functions which they realize are associ-

ated with the nominal as a whole (e.g., personal pronouns, proper names, thaf-nominalizatlons; Langacker 1991: 148), others consist of two nomi-

nals, related by means of apposition (e.g., the fact that whales are mam-

mals; Langacker 1991: 149); still others lack a head noun because the grounding predication itself functions as nominal (¢.g., that, these, some, any)

3.4, Composition and classification: Conclusion

In short, a radically functional approach to composition gives priority to functions rather than to constituents and posits that an accurate description and classification of symbolic units has to be based on a functional analysis which includes the external and the internal functional organization of the unit in question These functional levels are situated on a rank scale of functional configurations

Yet another characteristic of a functional analysis of linguistic patterning is that it considers the components of a composite whole to serve sev-

eral functions at the same time This kind of ‘multifunctional’ view on lan-

Trang 39

4 Functional layers of organization

The key to a functional interpretation of grammatical structure is the prin-

ciple that, in general, linguistic items are multifunctional (Halliday 1994:

30)

Many of the functional theories of language are ‘multifunctional’ and posit that constructions are characterized by different layers of functional organization (e.g., the Prague School; the functional schools of Dik and of Givén; the systemic-functional school of Halliday) The multifunctional approach to linguistic organization which I present here is that of Halliday’s systemic-functional model (with McGregor 1997 as most interesting recent development in it), and this because of its emphasis on the semiotic princi-

ple or the correlation between symbolization and semantic value for each

layer of organization

The systemic-functional interpretation of the multifunctional hypothesis can be illustrated by means of the categories of Subject, Theme and Agent in English (based on Halliday 1970b, 1988, 1994) If we take an English

clause like

(1) The duke gave my aunt this teapot

the element considered as Subject is the duke Its role in the clause can be characterized in various ways: the duke can be analyzed as the Agent of the

process with which it is construed (i.e gave); it can be said to be “the con-

cern of the message” (Halliday 1994: 30), and it is that of which something is predicated Not all English Subjects, however, realize these three features Consider the following clause:

(2) This teapot my aunt was given by the duke

The Subject here is my aunt, but the Agent-role is realized in the by-phrase

(i.e by the duke) and the message concerns this teapot The latter two fea-

tures — the role which a unit plays in the process designated by the main verb and being the concem of the message — appear to be features which can, but need not be, associated with the Subject: they can also be realized by other components of the clause.”

What at first sight appear to be characteristic properties of a single unit

thus turn out to be basically functions on the clausal level Rather than be-

Trang 40

single component (as in 1) or are realized by distinct components (as in 2) In the case of a unit such as the duke in (1), Halliday has labelled its various functions as Theme, Subject and Actor: as a Theme, the duke is the “point of departure” of the message conveyed by the clause (Halliday 1994: 34); the duke is also Actor or the “doer of the action” (Halliday 1994: 30); and, finally, in its function of Subject, the duke is the one “on which rests the truth of the argument” (Halliday 1994: 30), A unit such as the duke in (2) is thus multifunctional, i.e it serves several functions in the clause at a time

To fully appreciate Halliday’s multifunctional view on linguistic units, three aspects of it have to be further looked into, viz the so-called metafunctional origin of the functions which it involves, its configurational na-

ture, and, thirdly, the significance that is attributed to the paradigmatic or systemic choices that lie behind each metafunctional configuration The paradigmatic or systemic aspect of constructions will constitute the focus of Section 5 The metafunctional and configurational nature of Halliday’s multifunctional approach to linguistic organization will be elaborated now

Let us start with the notion of ‘metafunction’: what characterizes a

metafunction and how does this relate to the multifunctional view on linguistic units? Language users are believed to select from among a large

number of interrelated options which together constitute as it were the

‘meaning potential’ of language (Halliday 1970b: 142) It is one of the most fundamental claims of the systemic-functional approach to language that the options which are embodied in this meaning potential and which are reflected in the grammatical options of a language “combine into a very few relatively independent ‘networks’” (Halliday 1970b: 142) Within the massive network of options that constitutes the grammar of a language, one can thus distinguish certain clusters or sub-networks in which the “selections made by the speaker at one point tend to determine, and be determined by, the selections he makes at another” (Halliday 1979: 61) These clusters of lexicogrammatical options have been interpreted by Halliday as very general functions of language and they have therefore been called ‘metafunctions’ (Halliday 1994) Halliday distinguishes the ideational, the

interpersonal and the textual metafunction

The ideational or representational metafunction covers all those options

in language which enable language users to express their “experience of the

real world, including the inner world of [their] own consciousness”

(Halliday 1970b: 143) These options either represent experience directly in terms of processes, participants and circumstances (i.e experientially), or they represent experience indirectly, in terms of the fundamental logical relations in language (i.e logically) (Halliday 1979: 59) In the clause, for

Định dạng
Số trang	293
Dung lượng	16,69 MB