The purpose of this Chapter is to provide some indication of the scope of linguistic pragmatics. First, the historical origin of the term pragmatics will be briefly summarized, in order to indicate some usages of the term that are divergent from the usage in this book. Secondly, we will review some definitions of the field, which, while being less than fully satisfactory, will at least serve to indicate the rough scope of linguistic pragmatics. Thirdly, some reasons for the current interest in the field will be explained, while a final section illustrates some basic kinds of pragmatic phenomena. In passing, some analytical notions that are useful background will be introduced.

The origin and historical vagaries of the term pragmatics

The modern usage of the term pragmatics is attributable to the philosopher Charles Morris (1938), who was concerned to outline (after Locke and Peirce) the general shape of a science of signs, or semiotics (or semiotic as Morris preferred). Within semiotics, Morris distinguished three distinct branches of inquiry: syntactics (or syntax), being the study of “the formal relation of signs to one another”, semantics, the study of “the relations of signs to the objects to which the signs are applicable” (their designata), and pragmatics, the study of “the relation of signs to interpreters” (1938: 6).

To squeeze all that goes under the rubric of pragmatics within the confines of a linguistics textbook would be neither possible nor desirable. Consequently this book is quite conservative in scope and approach, and considers the main topics in a particular tradition of work. This is the largely Anglo-American linguistic and philosophical tradition that builds directly, for the most part, on philosophical approaches to language of both the logical and ‘ordinary language’ variety (an exception is the set of topics treated in Chapter 6, which has a sociological origin). In contrast, the continental tradition is altogether broader, and would include much that also goes under the rubric of sociolinguistics. But even within this much narrower field, this book is in some ways restricted, since its main aim is to provide an introduction and background to those topics that, perhaps largely for historical reasons, are central to the Anglo-American tradition of work in pragmatics. The would-be pragmaticist must understand these issues in depth, if he or she is to understand the background to a great deal of current research in both linguistics and philosophy.

One major way in which this book is perhaps innovative is the inclusion in Chapter 6 of a brief review of work in conversation analysis. Apart from its demonstrable importance for theories of language usage, work in conversation analysis contributes directly to many of the same issues that have preoccupied philosophers of language, and thence linguists, while employing a startlingly different methodology.

Serialization

If (5: 30) is a reasonable means for determining the deep structure properties of all those five pretty girls, we obviously need a set of rules that will ensure that the items constituting the terminal string appear in correct surface order. Whereas conventional transformational rules may not only change the order of the categories referred to in the structural index of a given T-rule, but may also add and delete material (not) generated by the phrase structure rules, the question of sequence for our purposes seems to have top priority, although obviously some form of deletion is required to account for e.g. kind and some in (5: 31). This matter, however, is linked with the question of serialization.

The surface order of the elements in an English NP allows a certain amount of fluctuation within a stable, basic scheme, a fluctuation which to a large extent is due to the possibility of a number of adjectival modifiers occurring with the same noun (cf. e.g. Goyvaerts 1968; Sussex 1974), and also to a number of idiosyncratic properties of certain quantifiers, notably all.

However, the rules we need to transform a terminal string like that in (5: 30) into an acceptable surface string all conform to the basic principle of subjunction.

Subjunction

The term ‘subjunction’ – and its ally, ‘adjunction’ – in the sense in which it is to be taken here, is introduced by Anderson (1971c) and further developed by him in subsequent works, notably (1977: ch. 2).

Features on FC

It is assumed that FC dominates linguistic material that would be interpreted as having grammatical or functional meaning, such as the articles, (other) determiners, quantifiers, and classifiers, whereas the denotative categories (L) dominate material that would be interpreted as having lexical meaning. The present section will be devoted to an account of the – functional – features on FC. The – lexical – features on L will be dealt with in §5.3. Cf. Bierwisch (1971) for detailed arguments in favour of a similar distinction between two comparable types of semantic features.

Central to the present account are the traditional grammatical categories – case, gender, tense, mood, etc. – among which standard descriptions of the (classical) Indo-European languages usually recognize as the indispensable nominal categories the following: case, gender, number, and person. Views differ as to whether definiteness should be accorded categorial status (possibly subsumed under the label ‘article’) on a par with these four, and the status of person as a purely nominal category is sometimes disputed.

There are good reasons not to regard either definiteness or person as being of the same primitive status as case, gender, and number (see §§8.1, 10.2.2 for details), and there are further good reasons to suppose that these three primitive categories must be supplemented with yet a fourth which is not normally given independent categorial status, namely, countability (cf. §5.1.3).

‘Referential functive’ defined

In this the final part I shall apply the theoretical framework developed in Part II to selected areas of English nominal composition.

From the very first inception of the research reported on in the present work my focus of attention was on pronouns, and it was my original plan ‘simply’ to account for pronominalization in English, essentially on the basis of a substitution model à la Harweg (1968) and Crymes (1968); several – abortive – attempts were made to come to grips with what Hjelmslev calls ‘la nature du pronom’ in this way. The present work may be read as the testament of my disillusionment with such an approach. From being traditionally a comparatively clearly delimited, closed class with a rather small membership, the class soon became invaded from all sides, by modal verbs (Crymes), by definite NPs (Harweg) and by determiners (Lyons 1973; Sommerstein 1972; Sørensen 1958). The result was that ‘the pronoun’ gradually receded to the point of vanishing (Postal 1966), hiding behind all manner of phenomena which all seemed to have one thing in common: referentiality. From being a major class within – apparently – easy access for investigation, it became a minor class virtually inaccessible.

The shift in viewpoint from ‘pronouns’ to ‘referentiality’ as the basic notion thus shifted the approach in the direction of Collinson (1937) with its basic appeal to ‘indicaters’. Although this is in many respects an interesting work I think it is fair to say that it is somewhat impressionistic and too intuitive to constitute a principled basis for a general theory of linguistic reference.

Referential function and referential potential

A number of points have been raised in a cursory fashion during the first two chapters which we need to take up in more detail. Prominent among these are the notions of the referential function of language and the distinction between different forms of semantic analysis.

It is universally accepted these days, I think, that there is no causal relationship between words and things. That which we call a rose by any other name would smell as sweet. Although this point may be held to be trivial now, one of the implications of it merits some attention. Since it is an important point, and since it has not been generally appreciated, if at all, I shall state it as concisely and as forcefully as I can:

1. (1) There is no logical requirement on language, conceived of as a rule-governed system of signs, that it should correlate with things, events, states of affairs, etc. at all. Differently formulated: It does not necessarily follow from the fact that we can speak that we should be able also to speak about something.

The uninterpreted, formal languages of logicians give substance to (1). Since natural languages differ from formal languages at least in respect of being ‘interpreted’ we may say, in spite of (1), that there are correlations between signs and things, but that these correlations must be contingent rather than necessary.

Pronominalization

Among the clearly defined areas of contemporary linguistic research, that of pronominalization is the only one to subsume matters of reference in any consistent way.

There are two fundamental aspects involved in the linguistic analysis of pronouns. Although they are interrelated they call for a distinction which is not always clearly drawn. One aspect is concerned with the internal analysis of pronouns; another is concerned with the function of the pronouns in linguistic utterances. Moreover, it is not immediately obvious what the relationship between these two aspects is in inferential terms: can the internal make-up of pronouns be inferred from a consideration of the functions they (may) perform, or is the range of functions performed by a given pronoun dependent in some way upon its internal make-up? In the present section I shall enquire into these questions on the basis of a number of more-or-less recent treatments of the pronoun in the linguistic literature.

It would appear to be the case that within each aspect three different views are in contention. Thus within the functional aspect we can distinguish (I) ‘co-reference’, (II) ‘substitution’, and (III) ‘indication’; and within the aspect of internal composition, (A) their composition as NP, (B) as non-derived, and (C) as involving syncretism/segmentalization. These may not be absolutely clear-cut distinctions, particularly with respect to (A) and (C), and they may not cover the whole field of pronominalization.

The present work is intended as a contribution towards the clarification of one of the problems with which man has preoccupied himself at all times: the relationship between words and things. It is not a philosophical work although the problem is often considered to be primarily philosophical in nature. But the problem has many aspects. Some of these are philosophical, some are pragmatical, some are sociological – and some are linguistic. It is the latter aspects that will be treated here, although some notice will be taken of what philosophers have had to say on the question.

My interest in the problem was first awakened when I participated in a research seminar on the problems of ‘linguistic representation’, instigated and led by Professor Gunnar Bech, in 1968. Owing to his sad illness, which to the detriment of us all has since forced him to retire prematurely, no very clear conclusions emerged from the seminar, so when a few years later the British Council and the Danish Research Council for the Humanities made it financially possible for me to devote three years to full-time research, I began to explore the area on my own. I am grateful to these two bodies for their support. I am likewise grateful to the Department of English at the University of Copenhagen for supporting my application for such a long leave of absence from my departmental duties.

Introductory comments on ‘meaning’

It is the ultimate task of semantics to describe the meaning of the words, phrases and sentences in particular languages and in language generally. This ultimate goal can only be achieved through the achievement of various intermediate goals: (a) a definition or explanation of the term ‘meaning’; (b) the development of a notational system in which meanings may be articulated; and (c) the development of a system of general rules matching the meanings of given words, phrases and sentences with particular notational specifications. The achievement of these intermediate tasks would enable us to fill in statements of the following form: (a) ‘meaning is __’, or ‘meanings are __’; (b) ‘the notational specification x is a specification (representation) of a meaning’; and (c) ‘the notational specification x is a specification (representation) of the meaning of y, because __’. The issues relating to (a) will be considered under the general heading of different approaches to the notion of meaning, those pertaining to (b) under semantic representation, and those that come under (c) under the term semantic analysis.

Although many diverse answers have been given to the question ‘What is meaning?’ throughout the history of linguistics and philosophy (e.g. physical entities, modes of signifying, mental images, concepts, ideas, the uses of words, speaker's stimulus/hearer's response) it appears that only two answers need be taken seriously: (1) meaning is a set of conditions; and (2) meaning is a set of relations.

Form and function

The grammatical category with whose description I shall be concerned in the present work is the N(oun) P(hrase), or – to put it more precisely – I shall be concerned with an analysis of the linguistic material which is dominated by the label NP in a standard Chomskyan phrase-marker. I shall, however, disregard predicative NPs, i.e. NPs functioning as subject or object complements, which means that only NPs functioning as subject, object, indirect object or complement to a preposition will be taken into account.

I shall call every occurrence of a NP a referential expression, and I shall define a referential expression as an expression by the employment of which we may speak about objects, persons, substances, occurrences, emotions, places, etc. The italicized parts of (1) are referential expressions:

1. (1)(a) Charles felt a piercing pain in his head

2. (b) Many are of the opinion that the Concorde should never have been built

3. (c) In the recent Home Internationals Series, Scotland beat England

4. (d) Harry was a cab-driver

5. (e) Everything was quiet in the village

6. (f) Have you sold the green coat that was in that window yesterday?

7. (g) She could not explain her feelings to anyone for all the gold in the world

8. (h) There are no unicorns

9. (i) John's reckless driving caused the accident.

Indefiniteness: sortal and selective expressions

In chapter 4 we established a distinction between three referential functions, sortal, selective, and identitive, and in chapter 7 we found that four classes of referential functives can be identified on distributional grounds. In this and the following two chapters I shall attempt to synchronize the notions of referential function and referential functive. This endeavour, which might be described as an attempt to synchronize functional and formal phenomena, constitutes the synthesis of the present work.

Let me begin by reminding the reader of my use of the term ‘referential expression’ (§4.1). A referential expression is a NP considered as a distributionally defined syntagm, the meaning of which equips it as a potential linguistic correlate of non-linguistic entities. According to which kind of referential function a given NP is held to perform, we may speak of sortal, selective, or identitive expressions, and according to which type of referential functive forms (part of) a given NP, we may likewise speak of quantified, E-classified, determiner, or pronominal expressions. Instead of the latter designation we shall usually speak simply of pronouns, however.

It is clear already, for numerical reasons, that there is no one-to-one correlation between, on the one hand, sortal, selective, and identitive expressions, on the other quantified, E-classified, determiner and pronominal expressions. It is therefore not the case that, e.g. a determiner expression always performs one and only one type of referential function.

My main concern in the present work has been to argue in favour of recognizing a distinction between two equipollent types of linguistic analysis. Of these, the syntactico-semantic analysis is well established and has been throughout the history of linguistics. In contrast, the referential–semantic analysis has received only sporadic and informal treatment, more often than not by philosophers rather than by linguists. Furthermore, when linguists have concerned themselves with the analysis of data that lend themselves to RS-analysis, they have more often than not conducted their enquiries within the framework which is designed to account for SS-phenomena. This has led to a certain amount of inconsistency within the SS-analysis while at the same time complicating the issues unnecessarily.

The area of English nominal composition which has provided data for an empirical assessment of the validity of the theoretical framework established in Part II has been restricted to the area in which the referential functives play a major part. This area is at the same time one of the areas least satisfactorily dealt with by current SS-practice.

Yet it is not only with respect to the referential functives that the referential–semantic theory may contribute with new insights into linguistic structure. Another general area where this type of analysis can reasonably be expected to yield felicitous results comprises many aspects of adjectival constructions. The serialization of attributive adjectives is clearly bound by a system of rules which the SS-analysis has no obvious way of accounting for.