unified theory of meaning in language

The aim of this book is not to present a unified theory of meaning in language (I am not even sure that that would be a worthwhile project), but to survey the full range of semantic phenomena, in all their richness and variety, in such a way that the reader will feel, on completing the book, that he or she has made face-to-face contact with the undeniably messy ‘real world’ of meaning. At the same time, it aims to show that even the messy bits can, at least to some extent, be tamed by the application of disciplined thinking. As far as seman- tic theories are concerned, I have been unashamedly eclectic, adopting what- ever approach to a particular problem seems genuinely to shed light on it. If there is a theoretical bias, it is in favour of approaches which, like the cognitive linguistic approach, embrace the continuity and non-finiteness of meaning.

This is not intended to be a ‘baptismal’ text; it would probably not be suit- able for absolute beginners. The sort of readership I had in mind is second- or third-year undergraduates and beginning postgraduates who have completed at least an introductory course in linguistics, and who require an overview of meaning in language, either as preparation for a more detailed study of some particular area, or as background for other studies. I would hope it would be found useful, not only by students of linguistics, but also students of ancient and modern languages, translation, psychology, perhaps even literature.

Most of the material in the book has grown out of courses in general seman- tics, lexical semantics, and pragmatics, given to second- and third-year under- graduates and postgraduates at Manchester University over a number of years. I owe a debt to generations of students in more than one way: their undisguised puzzlement at some of my explanations of certain topics led to greater clarity and better exemplification; critical questions and comments not infrequently exposed weaknesses in the underlying arguments; and very occasionally, a genuine flash of insight emerged during a classroom discussion.

The final form of the text was significantly influenced by constructive com- ments on a draft by Jim Miller of the University of Edinburgh, an anonymous American reviewer, and John Davey of Oxford University Press, although, of course, full responsibility for remaining imperfections lies with myself.

Preface xi

The organization of the book is as follows. It is in four parts. Part I dis- cusses a range of basic notions that underlie virtually all discussions of mean- ing within linguistics; Part 2 concentrates on aspects of the meanings of words; Part 3 deals with semantic aspects of grammar; Part 4 introduces the core areas of pragmatics, and highlights the relations between meaning and context.

Within Part I, Chapter I provides a very general introduction to questions of meaning, locating the linguistic study of meaning within the wider context of the study of signs and communication in general. Chapter 2 introduces a set of fundamental conceptual tools, mostly drawn from the field of logic, which, because of their wide currency in discussions of semantic matters, constitute indispensable background knowledge for a study of meaning in language. In Chapter 3, a number of concepts are introduced for the descrip- tion of meanings and differences of meaning. A basic dichotomy (based on Lyons 1977) is introduced between descriptive and non-descriptive meaning and, under each of these headings, important types and dimensions of vari- ation are described. It is rare to encounter any extended treatment of these topics in semantics textbooks, yet a mastery of them is essential to anyone who wishes to talk in a disciplined way about meanings. Chapter 4 discusses the way(s) in which simpler meanings are combined to form more complex meanings.

In Part 2, Chapter 5 provides a general introduction to the study of word meanings, first discussing whether there are any restrictions on what sort of meanings words can bear, then distinguishing the meaning of a word from that of a sentence or discourse, and the meanings of full lexical items from the meanings of grammatical elements. In this chapter the major approaches to lexical semantics are also outlined. In Chapter 6, the focus is on the range of variation observable in a single word form in different contexts, ranging from arbitrarily juxtaposed homonymies to subtle modulations of sense. Chapter 7 introduces a conceptual approach to lexical semantics, beginning with a dis- cussion of whether and to what extent word meanings can be equated with concepts. The discussion continues with an outline of prototype theory, the currently dominant approach to natural conceptual categories, and its rele- vance for the study of word meanings. Chapters 8 and 9 deal with relations of sense between lexical items which can occupy the same syntactic position—in other words, paradigmatic sense relations, such as hyponymy, meronymy, incompatibility, synonymy, antonymy, complementarity, reversivity, and con- verseness. Chapter 10 looks at larger groupings of words—word fields— mainly structured by the sense relations examined in the previous two chap- ters. Chapter 11 describes the main types of process, such as metaphor and metonymy, which enable new meanings to be produced from old ones. In Chapter 12, meaning relations between words in the same syntactic construc- tion, that is, syntagmatic sense relations, are examined. Topics discussed include the nature of normal and abnormal collocations, reasons for a

xii Preface

tendency for certain types of words to co-occur, and the nature and con- sequences of selectional pressures of words on their partners in a string. Chapter 13 outlines the componential approach to the description of word meaning, which specifies meaning in terms of semantic primitives.

The focus in Chapter 14, which constitutes the whole of Part 3, is on the sorts of meanings associated with various grammatical entities. First there is a discussion of the problem of whether there are any constant meanings attached to categories such as noun, verb, and adjective, and functions such as subject and object. There then follows a survey of the sorts of meaning borne by grammatical elements of various sorts, such as number and gender in the noun phrase, tense, aspect, and modality in connection with the verb, degree in the adjective, and so on.

Part 4 covers topics which are usually considered to fall under pragmatics, in that either they involve aspects of meaning which cannot be satisfactorily treated unless context is taken into account, or they are not propositional in nature (or both). Chapter 15 is concerned with reference, that is, establishing connections between utterances and the extralinguistic world. Reference is portrayed as the assigning of values to variables, the variables being signalled by definite expressions and the values being items in the extralinguistic world. Various strategies for indicating (on the part of the speaker) and determining (on the part of the hearer) correct referents are discussed, including the use and interpretation of deictic elements, names, and descriptions. Chapter 16 provides an outline of speech act theory, mainly following Austin and Searle (1969). It discusses the acts that people perform when they are speaking—acts such as stating, requesting, warning, congratulating, commanding, and so on. The range of different types of speech act is surveyed and their nature exam- ined. Chapter 17 deals with conversational implicatures, that is, those aspects of the intended meaning of an utterance which are not encoded in its linguistic structure, but are, as it were, ‘read between the lines’. Different types of con- versational implicature are described and some proposed explanations of how they arise are considered.

The concluding chapter briefly surveys the areas covered in the book, sug- gests practical applications of the study of meaning, and highlights areas which are currently poorly understood, and where further research is needed. Each chapter except Chapter I and Chapter 5 contains a set of discussion questions and/or exercises, suggested answers to which will be found at the end of the book.

Part 1

In this first part of the book, a number of fundamental, but fairly general notions are introduced, which need to be grasped before the more detailed discussions in later sections can be properly appreciated. Chapter 1 has a scene-setting function, identi- fying the place of linguistic signs and linguistic communication in the broader domains of semiotics and communication in general. Chapter 2 introduces a num- ber of vital conceptual tools drawn from the field of logic. Chapter 3 surveys the range of different sorts of meaning, and dimensions of variation in meaning. Chapter 4 discusses the notion of compositionality, one of the essential properties of lan- guage, and its limits.

Fundamental Notions

This page intentionally left blank

CHAPTER 1

Introduction

1.1 Communication

1.2 Semiotics: some basic notions

1.3 Language and other communicative channels

1.4 Characteristics of linguistic channels

5 1.5 Approaches to the study of meaning

7 1.6 The linguistic study of meaning

1.7 Branches of the study of meaning 8

Suggestions for further reading 9

10

12

15

16

This page intentionally left blank

CHAPTER 1

Introduction

1.1 Communication

Meaning makes little sense except in the context of communication: the notion of communication therefore provides as good a place as any to start an exploration of meaning. Communication can be conceived very broadly, including within its scope such matters as the transfer of information between biological generations via the genetic code, the interaction between a driver and his car, and indeed any sort of stimulus-response situation. Here we shall confine ourselves to what is surely the paradigm communicative scenario, namely, the transfer of information between human beings.

1.1.1 A simple model

Let us begin with a simple model, as shown in Fig. I.I (after Lyons 1977). In the model, the process begins with a speaker who has something to

communicate, that is, the message. Since messages in their initial form cannot be transmitted directly (at least not reliably), they must be converted into a form that can be transmitted, namely, a signal. In ordinary conversation, this involves a process of linguistic encoding, that is. translating the message into a linguistic form, and translating the linguistic form into a set of instructions to the speech organs, which, when executed, result in an acoustic signal. The initial form of this signal may be termed the transmitted signal.

6 Meaning in language

Every mode of communication has a channel, through which the signal travels: for speech, we have the auditory channel, for normal writing and sign language, the visual channel, for Braille, the tactile channel, and so on. As the signal travels from sender to receiver, it alters in various ways, through distor- tion, interference from irrelevant stimuli or loss through fading. These changes are referred to collectively as noise. As a result, the signal picked up by the receiver (the received signal) is never precisely the same as the transmitted signal. If every detail of the transmitted signal was crucial for the message being transmitted, communication would be a chancy business. However, effi- cient communicating systems like language compensate for this loss of infor- mation by building a degree of redundancy into the signal. Essentially this means that the information in a signal is given more than once, or is at least partially predictable from other parts of the signal, so that the entire message can be reconstructed even if there is significant loss. It is said that language is roughly 50 per cent redundant.

Once the signal has been received by the receiver, it has to be decoded in order to retrieve the original message. In the ideal case, the message reconstructed by the receiver would be identical to the message that the speaker started out with. Almost certainly, this rarely, if ever, happens; how- ever, we may presume that in the majority of cases it is ‘close enough’. All the same, it is worth distinguishing three aspects of meaning:

(i) speaker’s meaning: (ii) hearer’s meaning: (iii) sign meaning:

In the case of an established signalling system like language, the meanings of the signs are not under the control of the users; the signs are the property of the speech community and have fixed meanings. Of course on any particular occasion, the signs used may be ad hoc or conventional, if ad hoc, they may be prearranged or spontaneous.

speaker’s intended message hearer’s inferred message this can be taken to be the sum of the properties of the signal which make it (a) more apt than other signals for conveying speaker’s intended message, and (b) more apt for conveying some messages than others.

1.1.2 Language as a sign system Any natural human language is a complex sign system, ‘designed’ to ensure infinite expressive capacity, that is to say, there is nothing that is thinkable which cannot in principle be encoded (provided no limit is placed on the complexity of utterances). Each elementary sign is a stable symbolic associ- ation between a meaning and a form (phonetic or graphic); elementary signs may combine together in a rule-governed way to form complex signs which convey correspondingly complex meanings.

Introduction 7

1.2 Semiotics: some basic notions

1.2.1 Iconicity

Signs can generally be classified as iconic or arbitrary. Iconic signs are those whose forms mirror their meanings in some respect; signs with no natural analogical correspondences between their forms and their meanings are called arbitrary. A simple example is provided by the Arabic and Roman numerals for “three”: 3 and III. The Arabic form gives no clue to its meaning; the Roman version, on the other hand, incorporates “threeness” into its shape, and is thus iconic. Iconicity is a matter of degree, and usually coexists with some degree of arbitrariness. Three horizontal lines would be just as iconic as the Roman III: the fact that in the Roman symbol the lines are vertical is arbitrary, as is the fact that its size corresponds to that of letters.

Iconicity enters language in several guises. The majority of words in a natural language are arbitrary: the form of the word dog, for instance, does not mirror its meaning in any respect. However, the so-called onomatopoeic words display a degree of iconicity, in that their sounds are suggestive (to varying degrees) of their meanings:

bang clank tinkle miaow splash cuckoo peewit curlew whoosh thud crack ring wheeze howl rumble, etc.

The predominance of arbitrariness in the vocabulary is not an accidental feature, but is a crucial ‘design feature’ of natural language. There is a limited stock of dimensions of formal variation in linguistic signs; if all signs were iconic, it is difficult to see how universal expressivity could be achieved.

Some iconicity is also apparent in grammar. For instance, words which belong together tend to occur together. In The tall boy kissed the young girl we know that tall modifies boy and not girl because tall and boy come next to each other in the sentence. In some languages this relationship might be shown by grammatical agreement, which is a kind of resemblance, and therefore also iconic. Another way in which iconicity appears in the grammar is that grammatical complexity by and large mirrors semantic complexity.

1.2.2 Conventionality Many of the signs used by humans in communication are natural in the sense that they are part of genetically inherited biological make-up and do not have to be learned, although a maturational period may be necessary before they appear in an individual, and they may be moulded in various ways to fit particular cultural styles. The sort of signs which are natural in this sense will presumably include facial expressions like smiling, frowning, indications of fear and surprise, and so on, perhaps many of the postural and proxemic signs

8 Meaning in language

that constitute the so-called ‘body language’, certain types of gesture, vocal indications of excitement, desire, etc. (whether or not linguistic), and many more. Natural signs are likely to be the most cross-culturally interpretable.

Other signs have conventionally assigned meanings; they have to be specif- ically learned, and are likely to differ in different communities. Linguistic signs are the prototypical conventional signs. Even onomatopoeic words usually have a significant conventional element; often the iconic nature of the word can only be appreciated, as it were, with hindsight. Take the Turkish word bulbul. What does it refer to? A baby’s babbling? The noise of a mountain spring? In fact, it means “nightingale”. Looking back, one can make the connection. It is not only linguistic signs that are conventional. Obscene or offensive gestures, for instance, can vary quite remarkably cross culturally: I was once reprimanded for pointing the soles of my feet at the Prime Minister of Iraq (in Arab culture this is disrespectful: my disrespect was entirely inadvertent). Even in Europe, conventional gestures can differ: Greeks are famously—and slightly inaccurately—said to shake their heads to say “Yes”, and nod to say “No”.

1.2.3 Discreteness

Some signs can vary gradually in their form, and their meanings vary in paral- lel with the change of form, like the fisherman’s indication of the size of ‘the one that got away’; these are called continuous signs. Other signs have fixed shapes, and must be chosen from a limited inventory: intermediate forms are not allowed, the range of possibilities is ‘chunked’; such signs are described as discrete. Linguistic signs are virtually all of the discrete variety. Again, this is not an accidental feature, but has a close connection with iconicity and arbitrariness: continuous signs are necessarily iconic; arbitrary signs are necessarily discrete.

1.3 Language and other communicative channels

The prototypical scenario for linguistic communication is two people engaged in face-to-face conversation. Of course, in such an encounter, lan- guage signals are exchanged; but normally so are many other types of signal, and these modify and/or supplement the linguistically encoded message. Let us, then, briefly look at the semiotic environment of language in a typical conversation.

The signs that accompany language can be divided into two major types— paralinguistic and non-linguistic. The defining characteristic of paralinguistic signs will be taken here to be an extreme dependence on the accompanying language. Either they cannot be produced except during speech (because they are carried on the voice), or they cannot be interpreted except in conjunction

Introduction 9

with accompanying language. Examples of the first variety are abnormal volume, tempo, pitch, and voice quality; to function as signs, there must be a departure from some (personal) baseline or norm. For instance, abnormally high volume, fast tempo, or high pitch typically signal a heightened emo- tional state. Examples of the second variety include pausing, emphatic ges- tures, and gestures which metaphorically depict, for instance, direction of motion.

The functions of paralinguistic signs can be conveniently classified under three headings:

(i) Punctuation: there are signs which have functions parallel to those of punctuation in written language, mainly to segment the stream of speech so as to facilitate processing.

(ii) Modulation: this involves the addition of an emotive or attitudinal colouring to the linguistically encoded message.

(iii) Illustration: some signs ‘depict’ a key element in the message, such as a direction of movement, or a shape; the depiction may be relatively literal, like the hand movements of someone describing the climbing of a spiral staircase, or metaphorical, as when vertical and parallel hands accompany the setting of limits of some kind.

Not all the signs that occur alongside language are paralinguistic in the sense defined. For instance, one may smile or frown while speaking, and this may well ‘modulate’ the message. But smiles and frowns (and many other signs) are perfectly interpretable and capable of being produced in the absence of any accompanying lan