INTRODUCTION 5 страница

Many other kinds of what would usually be called implication in the everyday sense of the term (x didn't actually say that, but it's what he/she implied], but which do not fall within the scope of 'implication' as this is defined by logicians, have been more or less plausibly accounted for by invoking Grice's notion of con- versational implicature (see 6.3). These other kinds of so-called implication include the very common tendency to interpret a conditional as a biconditional: to interpret "if p, then q" as "if and only if p), then q". For example, the utterance of

(41) If Ann passes her driving text, her parents will buy her a Porsche

288 Text and discourse; context and co -text

would normally be held to imply that Ann's parents will not buy her a Porsche if she does not pass her driving text. But this is not actually said: it is conversationally implicated and is of course defeasible.

Of particular interest in this connexion are what have come to be called scalar implicatures involving quantifiers, such as 'some' and 'many', and numerals (as well as modal, evaluative adjectives and certain other classes of expressions). To exem­plify:

(42) The Browns have two daughters

will normally implicate that the Browns have only (i.e., no more than) two daughters (and in many contexts it will also implicate that they have no sons: Have the Browns got any children?). But the implicature is readily cancelled, in an appropriate context, by adding, for example,

(43) - in fact, they have three.

Scalar implicatures of various kinds have been intensively dis­cussed in the literature. They are mentioned here because they illustrate very clearly the property of defeasibility.

The second property generally held to distinguish conversa­
tional implicature from other kinds of implication is their cal-
culability. This was mentioned above (and illustrated), in
connexion with the interpretation of metaphorical utterances
such as John is a tiger: see (32) and (36). As was emphasized at
that point, the calculability (or computability) of conversational
implicatures is heavily context-dependent and in many, if not
all, contexts non-deterministic. Whether the implicatures are in
fact calculated in actual contexts of utterance and, if so, whether
they are calculated deductively and step by step, as suggested in
(36) - these are questions which are currently the subject of
debate and research among logicians and psychologists.
Another issue, as yet unresolved, is whether the knowledge of
information which is involved in the addressee's calculation of
conversational implicatures (in so far as addressees calculate
them) is always prepositional.

 

9.5 Conversational implicatures 289

A third property of conversational implicatures, according to Grice, is what he called non-detachability. This rests, at least initially, on a straightforward application of the non-technical, everyday, distinction between meaning and form. An implica­ture is non-detachable if it is inseparably attached to the mean­ing of an utterance and does not derive merely from its form. It follows from this formulation of what is meant by non-detachability that (as Grice himself realized) implicatures based on the subprinciple of manner will not necessarily be non-detachable: the manner in which something is said (in the relevant sense of 'manner' and of 'say') affects the form of what is said. Two synonymous utterances which differ in length, in grammatical complexity or in the technicality or obscurity of the words used may well give rise, for this very reason, to differ­ent implicatures. When they do, it will obviously not be possible to substitute the one for the other in the same context without changing the implicatures associated with them. In other instances, however, it should be possible to vary the form of an utterance (i.e., an utterance-inscription) and to hold constant its meaning (what is said) without thereby affecting what is implicated.

The putative non-detachability of conversational implica­tures turns out to be highly problematical, even in the case of those which are not based on the subprinciple of manner. Apart from anything else, the question whether two expressions are synonymous (in any or all of their meaning) cannot be answered unless and until one has decided what constitutes synonymy. Many proponents of formal semantics will opt for the relatively narrow notion of descriptive synonymy, defined in terms of truth-conditional equivalence (see 2.3). Those who adopt a broader notion of synonymy might well argue that the apparent detachability of implicatures associated with two formally dis­tinct, truth-conditionally equivalent, utterances is itself evidence for the view that the alleged implicatures are part of the linguistically encoded meaning of the utterances in question and that the utterances are not in fact synonymous.

Nor is the criterion of identity of form as straightforward as it might appear to be. Much of the textbook discussion of particu-

 

290 Text and discourse; context and co -text

lar examples cites the examples in question in their written form, even when they are assumed to have been spoken. This means that potentially relevant prosodic (and paralinguistic) differ­ences of form are not generally taken into account. One such class of examples is the following set of sentences:

'It is possible that p'
in contrast with

'It is not impossible that p'.

Discussion of such examples usually proceeds on the assump­tion that (45) is grammatically unambiguous. This is, however, a questionable assumption. On certain theories of the grammati­cal structure of English, (45) splits into at least two grammati­cally and semantically distinct subclasses, one of which manifests the phenomenon of sentence-negation (or clause-negation) and the other that of phrasal (or semi-lexical) ne­gation; and this difference of grammatical structure is normally, if not always, reflected in spoken utterances by differences of stress, intonation and rhythm (see 6.5).

In conclusion, then, let me emphasize once again that many of
the attempts made by linguists and philosophers to draw a
sharp distinction between semantics and pragmatics founder on
their failure to draw, and to apply consistently, a distinction
between sentence-meaning and utterance-meaning and to say
exactly what is and what is not encoded in the structure of sen­
tences. All too often identity of orthographic form is tacitly
assumed to be a sufficient condition of sentence-identity.

9.6 WHAT IS CONTEXT?

One of the points that emerges from our discussion of Grice's notion of conversational implicature in the previous section is the double role played by context. First of all, the utterance itself is embedded in what J. R. Firth and others have called a context of situation; and, as we saw in our discussion of metaphor, in order to decide whether a metaphorical interpretation is prob- able or not, one may need to know what the context of situation

 

9.6 What is context? 291

is. Second, having decided that information is being conveyed over and above the information contained in what has been said, the addressees have to infer what this additional informa­tion is on the basis of contextual information which they share with their interlocutors.

There has been a tendency, until recently, for linguists and philosophers to neglect the context of situation in their presen­tation of Grice's maxims. It is arguable that they have, for this reason, failed to bring out as clearly as they should have done the fact that language-behaviour is a culture-dependent activity. What constitutes sincerity and politeness may differ considerably from one society to another. Nor can we assume that rationality will manifest itself, in relation to the quality of information or its relevance, in the same way in all cultures. In fact, Grice's own presentation, and that of many of his fol­lowers, may well suffer from some degree of socio-cultural bias - a bias which is now being corrected by those working in conversational analysis and in what has come to be called the ethnography of speaking.

It is arguable that Grice's work also suffers from its philo­sophical bias in favour of descriptive, or prepositional, meaning. This is revealed, not only in his acceptance of a truth-conditional theory of meaning, but also in his conception of context - in the second of its two roles referred to above. For him, and for many of those who have drawn upon his ideas, context is taken to be a set of propositions in relation to which new propositions can be evaluated for truth and added to the context (or rejected as untrue).

But much of the knowledge that is involved in the production and interpretation of utterance-inscriptions is practical, rather than prepositional: it is a matter of knowing how to do some-
thing, not of knowing that something is the case. Of course, it is always possible (in certain languages at least) to describe practical knowledge as if it were prepositional. For example, instead of saying that a speaker must be able to tell whether his or her interlocutor is of higher or lower social status, we can say that the speaker must know which, if either, of the following two propositions is true: "x is of higher status than y " and "x is of

 

292 Text and discourse; context and co -text

lower status than y " (where x and y stand for referring expres­sions which will identify the speaker and addressee respectively). However, the fact that we can formulate practical knowledge in prepositional terms, does not mean that it is in fact preposi­tional. A strong case can be made for the view (taken for granted throughout this work) that social and expressive information is non-propositional.

It would seem, therefore, that context in both of the roles identified earlier in this section is, to a considerable degree, non-propositional. One of the advantages of the theory of speech-acts that we looked at in the previous chapter is that, in Austin's formulation at least, it gives full recognition to the social basis of language. It is, as I said, a theory of social prag­matics (in the etymological sense of 'pragmatics'): a theory of a particular kind of social doing. Grice's notion of language-behaviour as co-operative interaction fits in well with this; and, as I mentioned at the end of the preceding section, it need not be coupled with the assumption that the norms, or maxims, that he formulated for one kind of discourse in one culture - one kind of language-game, as the later Wittgenstein would have put it — are universally valid.

No simple answer, then, can be given to the question "What is context?". For the limited purposes of this book, it suffices to emphasize the fact that, in the construction of a satisfactory theory of context, the linguist's account of the interpretation of utterances must of necessity draw upon, and will in turn con­tribute to, the theories and findings of the social sciences in general: notably of psychology, anthropology and sociology. For further discussion of the role of context (including co-text), as also for what has come to be called neo-Gricean prag­matics, students are referred to the 'Suggestions for further reading'. In this chapter we have concentrated on the basic concepts as they were originally developed.