Three Approaches to Pragmatics

In is most general sense, pragmatics studies the relation between lin­guistic expressions and their users. The use of the term generally implies a dichotomy between language per se - the language competence in the abstract - and the use that is made of that competence by speakers and hearers. The distinction between semantics and pragmatics, therefore. (ends to go with the distinction between meaning and use, or more generally, that between competence and performance.

This is to anticipate, however, the debates surveyed in this chapter. Such debates on the relation of semantics (0 pragmatics have been prominent in the recent history of semantics. To a considerable extent;

(his focus of interest on pragmatics has been due to the influence of three philosophers - J, L, Austin, J. R. Searle and H. P. Grice - ail of whom have in some way championed a pragmatic approach to meaning. In linguistics, too, there have been various challenges to the assumption that competence can be studied in separation from performance, and purely format theories of languages, such as transformational grammar, have suffered from a backlash. Twenty years ago pragmatics, if it was mentioned at all, was regarded as a convenient waste-bin to which to consign annoying facts which did not fit theories. Now it is one of the more vigorous areas of linguistic research.

Semantics is the level of linguistics which has been most affected by pragmatics, but the relation between semantics (in the sense of conceptual semantics) and pragmatics has remained a matter for fundamental dis­agreement. The central issue is: is it valid to separate pragmatics from semantics at all? Three logically distinct positions in this debate can be distinguished:

(1) Pragmatics should be subsumed under semantics.

(2) Semantics should be subsumed muter pragmatics.

(3) Semantics and pragmatics are distinct and complementary fields of study.

For case of reference. I shall distinguish these three positions by the use of the following terms: (1) semanticism, (2) pragmaticism, (3), complementarism. In Chapter 14 I have already considered the implications of these three positions in one field - that of presupposition. My conclusion was that the explanation of presupposition required a "division of labour* between semantics and pragmatics; in other words, I adopted a complementarist position. In this chapter, loo, I shall argue for the complcmentarist position, which is probably the one most widely espoused in linguistics today.

At a very simple level, contention between the three positions above can be traced in an ambivalence in the everyday use of the verb mean. Of two major usages of this verb, one is bivalent {'X means }") and one is trivalent ('j? means Y by A"), For example:

(1) Danker means 'ass'.

(2) When Miss Trolwood said Janet! Donkeys! she meant by this remark that Janet was 10 drive the donkeys off the lawn.

The second example is clearly concerned with meaning not just as a pro­perty of language, but as a particular speaker's use of language in a particular context. It is this latter use of meaning which is pragmatic. The question is: is meaning (1) to be assimilated to nwaninn (2), or is meaning (2) lo be assimilated to mewiing (1), or is each menmnfc distinct from the other? We may note about meaning (2) that

(i) it involves the speaker's intention to convey a certain meaning

which may, or may not, be evident from the message itself. (ii) Consequently, interpretation by the hearer of this meaning is likely

to depend on context; and

(in) meaning, in this sense, is something which is performed, rather than something that exists in a static way. It involves action (the speaker producing an effect on the hearer) and interaction (the meaning being 'negotiated* between speaker and hearer on the basis of their mutual knowledge).

The following then are outward criteria for judging whether a particular discussion of meaning takes us into the realm of pragmatics:

(a) Is reference made to addressers or addressees, or (as I shall prefer to call them, ignoring the speech/writing distinction)

SPEAKERS Or HEARERS?

(b) Is reference made to the intention of the speaker or the interpretation of the hearer?

(c) Is reference made to context?

(d) Is reference made to the kind of act or action performed by means of or by virtue of using language?

If the answer to one or more of these questions is yes. there is reason to suppose (hat we are dealing with pragmatics.

Lllocutionary Force

We begin with (he last criterion, since it involves the other three. An exploration of the approach to meaning in terms of action can fittingly begin with the philosopher J. L. Austin {How to D<f Things with Words, 1962). Austin was dissatisfied with the traditional concentration, in linguistic philosophy, on referential meaning and the truth and falsehood of statements. This led him away from the question of'what do sentences mean' towards the question "what sort of act do we perform in uttering a sentence'. This he called the illocutionary force of an utterance, distinguishing it from the locutionary meaning (roughly, the referential or cognitive meaning which had been the philosopher's traditional concern) and from the perlocutionary effect (what sort of function or fulfilment ' of intention is accomplished by the sentence). The illocutionary purport of an utterance is to be expressed in terms of what Austin called 'happiness' or 'felicity' conditions, rather than in terms of truth and falsehood. One linguist influenced by this approach, Charles J. Fitlmore, has given, by way of illustration, (he following conditions of 'happiness' or appropriaicncss for (he simple imperative utterance Please shut the doorr.

(i) The speaker and the addressee of this sentence arc in some kind of

relationship which allows the speaker to make requests of the

addressee. (ii) The addressee is in a position where he is capable of shutting the

door. (Hi) There is some particular door which the speaker has in mind and

which he has reason to assume the addressee can identify without

any further descriptive aid on the speaker's part. (iv) The door in question, is, at the time of utterance, open. (v) The speaker warns that door to become closed. We can sec that (he violation of any of these conditions would cause the utterance to be in some sense "unhappy* or inappropriate.

One way of looking at presuppositions is to regard them as such "happi­ness conditions'. In fad, conditions (iii) and (iv) above (and possibly also condition (ii)) are identifiable as presuppositions in the treatment of presupposition I have given. But conditions (i) and (v) are illocutionary in a narrower sense: they are circumstances which enter into the definition of what it is to perform a speech act of a particular sort, in this ease a request. We may call such conditions speech-act conditions. One can imagine laying down similar conditions for other kinds of speech act, such as statements, questions, promises, warnings, apologies, etc. For a question to be 'felicitous' for instance, at least the following speech-act conditions must obtain:

(a) There is a piece of information (X) of which the questioner is ignorant.

(b) The questioner wants to know (X).

(c) The questioner believes that the addressee knows (X).

(d) The questioner is in a position to elicit (X) from the addressee. Generally the illocutionary force of an utterance is not made explicit by the utterance itself; but a notable class of exceptions to this rule includes such sentences as:

I do. (uttered at a marriage ceremony) I name this ship Queen Elizabeth, (said when smashing a bottle against

the bows)

I give and bequeath my watch lo my brother, (in a wilt) t bet you sixpence it will rain tomorrow.

These sentences (Austin's own examples) illustrate the class of utterances Austin calls performaHves\ that is, they are utterances which themselves describe the speech act which they perform. Performatives took like state­ments syntactically, but as Austin points out, they differ from most gtatements in that they cannot (easily) be declared false. Thus if speaker A says 'I declare that King Charles II was a coward* and speaker B replies 'That's false', he seems to deny not the performative utterance, but the proposition that it contains, namely, 'That King Charies II was a

coward.'

The characteristic syntactic markers of a perfonnative sentence are the following:

0) The subject is in the first person. (/ or we) (u) The verb is in the simple present tense. (sfa», ask, pardon, etc.) (111} The indirect object, if one is present, is you. (iv) It is possible to insert the adverb hereby. (v) The sentence is not negative.

All these characteristics are realized in:

1 hereby declare to you my innocence.

But not all verbs referring to speech events can function as performative verbs, as we gather from the "infelicity' of these sentences:

• I hereby remark that the weather is cloudy.

• I hereby persuade you to eat fish in Lent.

• 1 hereby denigrate your parents.

Performatives are problematic semanticaily because for every non-performalive sentence it is possible to find one or more performative equivalents. Thusonecan maintain, with Austin, that the only difference between / imkr y<w in so! and Cm! is that the former »s explicitly perfor-roativc, while the latter is implicitly so. The problem is, how do we give an account of the quasi-cquivalcncc of these utterances - a matter to which I shall return shortly.

Austin's study of speech acts and performatives was taken further and systematized by his pupil J, R. Sedrle {Speech Acts, 1969), who went so far as to claim that 'a theory of language is part of a theory of action. That is, Searle in a sense regarded the whole of linguistics as pragmatics. It is not surprising, then, that he conceived of meaning in pragmatic terms: 'the study of the meanings of sentences and the study of speech acts are not two independent studies but one study from two points of view' (ihiit., p. 18). Searlc's speech-act theory may be regarded as an example of the trend t earlier called 'pragmaticism': the assimilation of semantics to pragmatics. A more extreme pragmaticist approach is that of W, P. Alston (1964).

The Performative Analysis

Almost at the same time as the publication of Searle's Speech Ads, a parallel development in the study of performatives took place within transformational grammar. Although clearly influenced by Austin and Searle, the per formative analysis (or performalive hypothesis), as it may be called, was actually an example of the opposing tendency of semanticism; the incorporation of pragmatics into semantics. This was in the heyday of generative semantics (see p. 346), when it was optimistically assumed that all factors relevant to the meaning of sentences could be included in their deep structure.

The gist of the performative analysis is that in its 'deepest structure' (which we may consider either its syntactic "deep structure' or its semantic representation), every sentence is a performative; that is, every sentence contains as its main subject a first-person pronoun, and as its main verb a performative verb in the simple present tense. For example, the declarative sentence Tomorrow will be rainy has, in this view, a deep structure of a form such as / stale that [tomorrow will be rainy] or /pre­dict that [tomorrow will be rainy}, or / warn you that [tomorrow will be rainy]. Questions and commands are given a similar deep structure analysis:

Open the door. — I command you [to open the door]. How much are those bananas? — I request of you that (you tell me [how much those bananas are]].

The advantages of the performative analysis were argued persuasively by J. R. Ross in his article 'On Declarative Sentences' (1970). He pointed out that main clauses have many things in common with clauses which are indirect statements, indirect questions, etc. For example, the emphatic reflexive pronoun is acceptable in sentences (3) and (4) below, but not in sentence (5):

(3) Tom believed that the paper had been written by Ann and himself.

(4) The paper was written by Ann and myself.

(5) * The paper was written by Ann and himself.

If we accept the performative analysis, then all direct statements come to be seen as indirect statements, and therefore a single syntactic condition can explain two sets of circumstances which would otherwise require independent explanation for direct speech and indirect speech. In the case of (3)-(4) above, the main circumstances observed are that the co-ordinated emphatic reflexive pronoun either (a) must be in the first person if it occurs in the main clause; or (b) must agree with the noun plirase of the higher clause if it occurs in a subordinate clause. But by the perfor­mative analysis, (a) becomes a special case of (b), and so need not be stated as a separate condition. If we insert the performative clause I state that... in (4) and (5), condition (b) is fulfilled in (4) but not in (5); hence (5) is ungrammatical.

Ross adduces many other arguments of the same kind, and I shall only mention one more of them here. The phrase As for... self obeys the same sort of general rule as was illustrated above for coordinated emphatic reflexives: namely, that myself (or ourselves) occurs in main clauses, while in indirect speech, there occurs a pronoun in agreement with a noun phrase in the higher clause:

(6) Tom declared that as for himself, he was ravenous.

(7) As for myself, I am ravenous,

(8) * As for himself, Tom is ravenous.

Again, if we adopt the performative analysis and analyse (7) in depth as /slate thai as for myself, 1 am ravenous, the same rule that accounts for the acceptability of (6) accounts for the acceptability of (7) in contrast to (8).

Ross's analysis also explains the semantic equivalence, or virtual equivalence, between an ordinary statement like Tomorrow will be rainy and a corresponding performative like / slate to you that tomorrow will he rainy. The difference between these, in the analysis, is simply that a transformational rule of performative deletion has applied to the former sentence, pruning away from the front of it the subject, performative verb, and indirect object.

Although it did not escape criticism (see especially Matthews, 1972), the performative analysis found widespread support among transformationalists. Nowadays, this support has all but disappeared. Per-formative sentences arc so rare, that it seems highly unnatural to argue that every single direct statement is fundamentally an indirect statement. that every direct question is fundamentally an indirect question, etc. The unnaturalness grows distinctly suspicious when we consider that a dis­course will generally have (lie same deep structure performative for each sentence it contains. So a newspaper report consisting of a hundred sentences will have 'I report that ...' or some such performative clause repeated a hundred limes. The objection becomes more substantial (as Ross himself points out) when we apply performative analysis to texts belonging to impersonal styles of discourse in which first and second pronouns arc taboo (for example, in legal documents, regulations, bureaucratic instructions). Here the performative analysis forces us to allow first-person and second-person pronouns in the underlying struc­ture of a sentence, but not to allow them in its surface structure.

There is an additional case, noted by Ross (1970, p. 255), where the performative analysis gels into difficulties:

(9) As for myself, I promise you that I'll be there.

This sentence already contains an overt performative clause / promise you..., and yet the phrase As for myself, according to an argument men­tioned in connection with examples (6) to (8), points to a higher performa­tive clause within which / promise you is embedded. But this violates another rule which Ross finds it necessary to establish: that no performa­tive can be embedded in another performative. In any case, to suppose that a double performative I slate to you l/ml I promise you underlies lllis sentence is to open llic door to potential infinite regression of performatives, one within the other:

1 state that X.

1 slate that 1 state that X.

I state that I state that I state that X ...

(etc.)

If this sort of embedding is allowed, then every simple sentence can be derived from infinitely many deep structures.

Situation of Utterance

Ross himself proposed an alternative to the performative analysis, which he called the 'pragmatic' analysis. The outline of the 'pragmatic analysis' is that the subject and performativc verb and indirect object arc (in Ross's phrase) 'in (lie air' - that is, they belong to the extra-linguistic context of the utterance rather than to its actual structure. Ross saw some advantage in the pragmatic analysis (notably that it resolves (lie difficulty of sentence (9) above), but saw no way of giving it formal status within transforma­tional grammar.

Since for me the pragmatic analysis offers hope of an attractive common-sense alternative to the performativc analysis, I shall now attempt to do what Ross declined to do - to give a reasonably precise characterization of what it means for elements of a speech act to be 'in the air' instead of being part of the underlying structural representation of a sentence. For this purpose, I shall propose that each speech act takes place in a situation of utterance, and that a situation of utterance includes:

(a) the utterance itself U

(b) the speaker/writer of the utterance s

(c) the hearer/reader of the utterance h

(d) the speech act A

(One might wish to extend the specification of a situation of utterance to include two further factors: (e) the place of utterance; (f) the time of utterance.) The utterance itself can refer to aspects of the situation of utterance by means of deictic items (p. 67) such as this, now, here. In addition, first- and second-person pronouns are defined as referring to participants in the speech act. Other linguistic elements referring to the situation of utterance are performative verbs, which, when used in the present tense (referring to the time of utterance) refer to the speech act itself. Hence we always have the ability, if we so wish, to describe overtly the situation of utterance in which we have the role of speaker, by using a performative such as / declare to you that X, in which speaker (/), speech act (declare), hearer (you) and utterance (X) are named in that order.

However, it is a general principle of conversation (sec Grice's Maxim of Quantity - p. 296) that a speaker does not bother to describe aspects of the extra-linguistic situation which are obvious to himself and to the hearer. (This principle is manifest in example (2) Janet! Donkeys! - for further examples, see p. 66.) The same principle of least effort explains why, in general, we do not specify the implicitly known features of an utterance situation by the use of a performative clause. In fact, when we do use a performative (for example, I order you to leave; I promise to pay you) often it is precisely in order to make explicit the illocutionary force which might otherwise not be appreciated. For example, if I say / promise to pay you tomorrow I make it clear that I am committing myself to a future course of action, whereas this reassurance might or might not be evident in the non-performative I'll pay you tomorrow.

We are now in a position to see how the 'pragmatic analysis' matches the performative analysis, and also has some advantages of its own. First of all, let us consider how the pragmatic analysis explains the asymmetry in examples like (3) -(5) and (6)-(8). The point to notice is that examples like

(6) Tom' declared [that as for himself, he' was ravenous].

are indirect speech sentences, in which (as we saw on p. 315) the speech-act predicate governs a metalinguistic predication:

(6a)(lom'. declare, [he': ravcnous])

So to make the generalization about (6) and (7) in pragmatic terms, we simply say: for any predication containing 'As for xself, the '.xself must refer to the speaker of the speech situation in which this occurs. In (6) tlie speaker is explicitly named as 'Tom', so this is a case of coreference to the subject of the speech-act verb; in (7) the speaker is the T implicit in the situation. To make the matter clearer, we may distinguish between the primary situation of utterance (the one which is implicit) and the secondary situation of utterance (the one referred to in the primary situation), and say that in (6) the '.xself occurs within a secondary situation of utterance, and therefore refers to the speaker of the secondary situation; whereas in (7) the 'xself refers to the speaker of the primary situation.

This generalization works as far as it goes, but does not apply to all cases. For example, in:

(3) Tom believed [that the paper had been written by Ann and himself].

there is no secondary situation of utterance. On p. 315, however, we noticed (lie parallel between indirect speech quotation, and direct thought quotation: it is the latter that is illustrated in (3). So to make the generalization work for all relevant cases, we should extend the rule so that it applies not only to utterance situations, but to 'thought situations':

Utterances containing the emphatic reflexive '-xself are grammatical in cases where '^-self refers to the speaker/thinker of the utterance/ thought situation in which it occurs.

In this way we show that the pragmatic analysis can capture the generalizations made by the performative analysis, by formulating them in terms of situation of utterance/thought, rather than in terms of syntactic or semantic constituent structure.

The parallel between (3) and (4), (6) and (7) is obvious so long as we remember that the elements of the primary situation arc implicit ('in (lie air', in Ross's phrase), wereas those of the secondary situation of utterance are explicitly referred to in the main predication ('Tom declared that ...' etc.). What is special about performatives is that, in them, it is the primary situation that is explicitly described; or to put it another way, the secondary situation and the primary situation are identical. This means, in turn, that 'I declare that' in (7a) simply makes explicit elements of the situation of utterance implicit in (7):

(7) As for myself, I am ravenous. (7a) I declare that, as for myself, I am ravenous.

By this common-sense type of analysis, then, we arrive at the conclusion that (7) and (7a) are in pragmatically equivalent (except that the nature of the speech act is covert in (7) and overt in (7a)), and that this equivalence applies to all similar cases of performativts and their non-performative analogues. Thus the pragmatic analysis can dispense with the performative-deletion transformation, which is a necessary part of the performative analysis.

Another advantage of the pragmatic analysis in terms of situation of utterance is that it docs not come to grief over utterances like:

(9) As for myself, I promise I'll be there.

Whereas the performative analysis has to postulate a double performa­tive here, the pragmatic analysis simply allows myself in (9) to refer to the (implicit) speaker of the primary situation. No additional performa­tive such as 'I state that as for myself...' need be posited.