This book started life as a second edition of Language, Meaning and Context (1981) and, regrettably, in several places has been announced as forthcoming under that title. It now appears with a completely different title because, in the event, it has turned out to be a very different book. It is much longer; it deals with several topics that were not dealt with at all in the earlier book; and, above all, it is written at a different level and in a different style.

Many of these differences derive from the fact that Linguistic Semantics (LS), unlike its predecessor (LMC), is intended to be used as a textbook for courses in semantics given in departments of linguistics (and related disciplines) in colleges and universities. Although LMC was not conceived as a textbook, it was quite widely used as such, until it went out of print some years ago. I hope that LS, being written especially for students of linguistics, will prove to be much more satisfactory for this purpose.

In revising the original text, apart from taking account of such recent developments as seemed to me to be relevant to what is presented as an introduction to the subject, I have found myself obliged to add several new sections and to rewrite or expand others.

INTRODUCTION

In this chapter, which constitutes the whole of Part 1, we deal with a number of concepts which are fundamental to the whole enterprise of putting linguistic semantics on a sound theoretical footing. Although it is one of the longest chapters in the book and includes several sections containing material which, at times, is quite demanding for those who are new to the subject, I have deliberately not divided it into two (or more) chapters, because I wish to emphasize the fact that everything that is dealt with here hangs together and is equally relevant throughout.

Readers who find some of the material difficult on a first reading should not be too concerned about this. They can come back to it as they proceed through the following three parts of the book and see how the various technical distinctions that are drawn here are actually used. Indeed, this is the only way of being sure that one has understood them. The fact that I have brought together, at the beginning of the book, some of the more fundamental terminological and notational distinctions which are relevant throughout should make it easier for readers to refer back to them. It should also make it easier for them to see how the conceptual and terminological framework that I am adopting compares with that adopted in other works that are referred to in ‘Suggestions for further reading’.

The fields referred to in the title of this chapter have become part of the institutional structure of linguistics. Each has scholars who specialise in it, and advertisements for teaching posts will often specify expertise in one or more of them. The divisions between them are also reflected in the organisation of lecture courses and exam papers. But are they just an institutional convenience? Or do these terms refer to real distinctions?

Let us begin by reminding ourselves of their history. The term ‘syntax’ is far older than the others, and has been used by most linguists throughout this century to refer to a branch of grammar concerned with the construction of sentences. Since the 1940s that has often been taken to exclude the study of meaning; but it does not have to be so interpreted. For Bloomfield, for example, semantics was a ‘phase’ of linguistics whose task was to relate formal features to features of ‘distinctive’ or ‘linguistic meaning’ (1935: 74, 138,141). But just as the units of Bloomfield's lexicon associated forms with meanings, so too did his units of grammar. Therefore their study was not separate from semantics, but part of it. Semantics as Bloomfield defined it was ‘ordinarily divided’ into grammar and lexicon (138), or ‘equivalent to’ the investigation of these fields (513, notes to section 5.1).

INTRODUCTION

We have been operating with the assumption that utterance-meaning is crucially dependent on context. So far, however, I have made no attempt to say what context is or how it determines the meaning of utterances and controls our understanding of them. Nor have I said anything in detail about spoken and written text: I have, however, made it clear in previous chapters that speech must be distinguished from writing (and the products of speech from the products of writing), even though, in the technical metalanguage of semantics that we have been building up throughout the book, ‘utterance’ and ‘text’ are being applied to the products of both speech and writing.

In this chapter, we shall be dealing with both text (and discourse) and context (and co-text). As we shall see, text and context are complementary: each presupposes the other. Texts are constituents of the contexts in which they are produced; and contexts are created, and continually transformed and refashioned, by the texts that speakers and writers produce in particular situations. It is clear that even sentence-sized utterances, of the kind we considered in the preceding chapter, are interpreted on the basis of a good deal of contextual information, most of which is implicit.

We shall begin by recognizing explicitly that the term ‘sentence’ is commonly used by linguists (and also by non-linguists) in two senses, one of which is, to put it loosely, more abstract than the other.

Grammar exists to code meaning. Every language has a similar set of semantic tasks to fulfil. There is a universal pool of grammatical construction types, and each language draws its own selection from the pool. According to the selection that is made, a similar type of meaning may be expressed by different grammatical means in different languages.

But the variation is not random. Each construction type in a language has a semantic effect, and although a given meaning may be expressed in different languages by constructions that are grammatically diverse, they will have similar semantics. This will be illustrated for complementation strategies in the Australian language Dyirbal, which are compared with the familiar complement clause constructions in English, and in Fijian.

Section 1 discusses the appropriate part of the semantic task which all languages have to perform; section 2 sketches the varied grammatical means for achieving it, paying attention to the different kinds of interclausal relation found in human languages; section 3 deals with complement clauses in English and Fijian; section 4 then discusses complementation strategies (which do not involve complement clauses per se) in Dyirbal; section 5 summarises the results and shows that verbs with similar meaning take complement clauses or complementation strategies, with similar meanings, across widely diverse languages.

The semantic task

The words of any language can be grouped into a number of lexical classes called SEMANTIC TYPES, which have a common meaning component and some shared grammatical properties.

INTRODUCTION

How does one set about defining the meaning of words? In this chapter, we shall see that different answers can be given to this question. We shall also see that different answers can be given for different kinds of words.

For some words, especially nouns such as ‘table’ or ‘chair’ in English, one might think that a version of the so-called referential theory of meaning, mentioned in Chapter 1, is perfectly satisfactory: one might think that they can be readily defined by identifying what they stand for. Some theorists have taken this view; and it is well represented in the literature of both linguistic and philosophical semantics. It is undoubtedly a reasonable view to take, at least for words that stand for such things as dogs and cats, or tables and chairs; and it is commonly such words that are used to exemplify, not only the referential theory, but also complementary or alternative theories of lexical meaning.

But how does one define or identify what a word stands for? Is it possible to say what one word stands for without employing other semantically related words in doing so and without saying in what respect these semantically related words are similar to one another in meaning and in what respects they differ? And what exactly does the traditional expression ‘stand for’ mean in this context? As we shall see in the following section, we have to distinguish what expressions denote from what they can be used to refer to: we have to distinguish denotation from reference.

INTRODUCTION

This chapter is pivotal in the structure of the book. It is also one of the longest, and there is a distinct change of gear. We shall be making full use of logical notions and discussing in greater detail than we have done so far the basic concepts of modern formal, truth-conditional, semantics, which, as we saw in the preceding chapter, were first developed within logic and the philosophy of language and were subsequently extended to linguistics.

There is nothing new or revolutionary about the influence of logic on linguistics (and vice versa). Grammatical theory and logic have been closely associated for centuries. Indeed, much of the terminology of traditional grammar – ‘subject’, ‘predicate’, ‘mood’, etc. – is also part of the logician's stock in trade. But does this use of the same terminology reflect any more than a purely historical, and accidental, association between the two disciplines? Does the grammatical structure of a sentence correspond directly to the logical form of the proposition it expresses? More generally, is there nothing more to the meaning of a sentence than its propositional content? These are the principal questions that we shall be addressing in the present chapter.

Our general conclusion will be that there are certain aspects of sentence-meaning that cannot be adequately represented by standard propositional logic. In coming to this conclusion, however, we shall also see that our understanding of the way meaning is encoded in sentences has been greatly increased in recent years by the attempt to describe precisely the interaction between the logical form of propositions and the grammatical structure of sentences (and clauses).

INTRODUCTION

In the last three chapters we have been concerned with lexical semantics: i.e., with the meaning of lexemes. We now move on, in Part 3, to a consideration of the meaning of sentences.

The distinction between sentences and utterances was introduced in Chapter 1 (see 1.6). The need for drawing this distinction is reinforced by the discussion of grammaticality, acceptability and meaningfulness in the following section (5.1). But our main concern in this short, and relatively non-technical, chapter is the meaningfulness of sentences. Granted that some sentences are meaningful and others meaningless, what grounds do we have for drawing a theoretical distinction between these two classes of sentences? Is it a sharp distinction? Is there only one kind of meaningfulness?

What may be described as truth-based theories of the meaning of sentences have been particularly influential in modern times, initially in philosophical semantics, later in linguistic semantics. Two of these were mentioned in Chapter 1: the verificationist theory and the truth-conditional theory (1.7). According to the former, sentences are meaningful if (and only if) they have a determinate truth-value. In formulating the verificationist theory of meaning (or meaningfulness) in this way, I am temporarily neglecting to draw a distinction (as many verificationists did) not only between sentences and utterances, but also between propositions and propositional content, on the one hand, and between truth-values and truth-conditions, on the other.

It is difficult to know quite how to respond to the request to contribute a chapter to a volume that is being published in one's own honour. Should one try to comment in detail on the other contributions? Should one attempt, instead, to provide a more general statement on how one stands now on the issues about which one has written in the past and which are still, or have recently become, prominent in the literature? I have opted for the second of these two possible responses. I will, however, comment selectively, and in some cases very briefly, on relevant points made by the other contributors.

First I must thank those of my colleagues - several of them former students of mine, all of them by now friends of long standing - who have done me the honour of contributing to this volume. I am especially indebted to Frank Palmer for agreeing to act as editor: he was already well established as a leading member of the so-called London School, based at the School of Oriental and African Studies, when I went there as Lecturer in Comparative Linguistics in 1957, and he has continued to be active in research and publication ever since. I will refer to two of his more general, introductory, works presently, when I set the scene for the discussion of grammar and meaning which follows.

Introduction

Many a student must have sighed when faced with what might seem the almost medieval casuistry of many of the distinctions in John Lyons' (1977) two-volume handbook, Semantics. Ambiguities and unclarities of every kind in our frail metalanguage for semantic analysis are there laid out for all to see; a formidable reef of difficult distinctions - types and tokens, acts and products, uses and mentions, originals and replicas, ambiguities of level, etc. - upon which we are all guaranteed sooner or later to founder. Introducing the type/token distinction in a straightforward manner, he goes on to tease us by showing how identifying different tokens of the same type can require a complex measure of similarity or identity of type, and then, having raised our anxieties, announces that it would be ‘unnecessarily pedantic’ to identify each such distinction (1977: 13–16).

One such distinction Lyons alludes to throughout the volumes may look particularly pedantic, the distinction between utterance-types and utterance-tokens, coming on top, as it does, of the distinctions between system-sentences and text-sentences, sentence-types and sentence-tokens, utterance-acts and -signals and so on. He himself seems to hint (1977: 570ff.) that the distinction may not be of any great utility (since utterancetokens are rarely constrained to type, and such types could in any case be given formal definition, for example, in terms of sentence-types or forms).

In this chapter I want to suggest that this distinction between utterance-type meaning and utterance-token meaning, or something rather like it, may indeed prove to be an important division in levels of meaning.

Concord, rection and dependency

Concord and rection both involve the assignment of a value (term) for a morphological category or categories to a victim word by a trigger or controller. Traditionally, they are differentiated, roughly, in terms of the substantive role of the trigger (see e.g. Matthews, 1981: 246). Thus, the Latin sentence in (1):

1. (1) Hostis habet muros.

2. ‘The enemy holds the walls.’

can be said to show concord between subject (hostis, the trigger) and verb (habet, the victim), which latter is said to agree with the subject - that is, is assigned (some of) the same category values, singular number and third person; while assignment of the value for the case category (accusative) to the post-verbal argument (muros) is simply determined by presence of the verb (in Latin, of a particular class) - it is not a copy of a category also manifested in the structure of the verb and thus is rectional.

This brief formulation of the traditional characterisation of concord and rection is in need of some refinement. It is not obvious that all concord, for instance, involves simply the copying of a particular member of a category. For example, a controller involving co-ordinate items of discrepant gender may trigger yet another gender selection in the victim (see e.g. Corbett, 1983, 1989, on this and other complications). But it is at least appropriate to attribute the same particular morphological category to both trigger and victim, whatever the mechanisms of matching the members (apart from simple copying).

Introduction

Many linguists working on spoken language abandon the sentence as an analytic unit as a result of data like the text in (1), which is the transcription of a conversation. The ‘ + ’ signs mark brief pauses.

1. (1) I used to light up a cigarette + you see because that was a very quiet way to go + now when I lit up my cigarette I used to find myself at Churchill + and the quickest way to get back from Churchill was to walk along long down Clinton Road + along + Blackford something or other it's actually an extension of Dick Place but it's called Blackford something or other it shouldn't be it's miles away from Blackford Hill + but it's called Blackford Road I think + uhm then along to Lauder Road and down Lauder Road.

As Crystal (1987:94) observes about another, but similar, text, it is not easy to decide whether the pauses mark sentence boundaries or whether the whole text is one loosely constructed sentence. An additional problem in (1) is that one stretch uninterrupted by pauses appears to consist of what would be several sentences in writing: It's actually an extension of Dick Place, but it's called Blackford something or other, It shouldn't be, it's miles from Blackford Hill. Two of the clauses are conjoined by but, but two are simply adjacent to the preceding one.

In this chapter I focus on a particular form of grammaticalisation process, the development of non-standard conjunctions in the rural traditional dialects (see Trudgill, 1990) of East Anglia. The factual grammatical information provided here is derived from my own personal knowledge of the dialects in question. The examples I supply for illustrative purposes are taken from dialect texts (presented here in normalised orthography) written by East Anglian writers who have made self-conscious attempts to write in the local dialect. The advantages of using this sort of material are obvious. There are now relatively few speakers of the traditional dialects left, and obtaining texts from native speakers that are of sufficient length to permit the study of grammatical features would be an extremely time-consuming process. There are also, of course, very obvious dangers with using this type of material as data. Dialect writing may be unreliable as a representation of actual dialect speech, employing inaccurate stereotypes and hyperdialectisms (Trudgill, 1986). In this particular case, however, I have selected only texts which I am confident are accurate representations of the dialect and have been written by genuine native speakers of the dialects in question. My confidence comes from a lifetime of living in East Anglia, from having had many older relatives and acquaintances who spoke the traditional dialect and from twenty-five years of academic study of these dialects. The examples, that is, should not be regarded as data as such, but simply as illustrations which support and confirm my own understanding of how these dialects work.