прагматика и медиа дискурс / Teun A van Dijk - Text and Context

ZOO PRAGMATICS

the hearer, according to further conventional criteria dependent on the status, function or position of the speaker and ad hoc situational criteria (personality of the speaker, specific circumstances, corroborating evidence, compatibility with the extant knowledge of the hearer, etc). Note that these conditions are preconditional in the sense that when they are not satisfied the illocutionary act either becomes pointless (eg superfluous) or fails by nonacceptance by the hearer.

With each specific illocutionary act, then, a set of CHARACTERISTIC or ESSENTIAL conditions must be satisfied, distinguishing it from other types of illocutionary acts. For assertion, again, this would be the condition

W s(-)K H(b)p. We see that such a condition pertains to the specific change wanted by the speaker, not to the actual change having taken place or not (as a perlocutionary effect): it is relevant for us merely to reconstruct the systematic rules and conditions determining how the hearer understands what is meant (in a broad sense) by the speaker. Similarly, in a request the essential condition is WS(.)DOH(b)p. In both cases we may embed these conditions in the schema BH(b —, because it must be generally assumed (see aboye) that the hearer believes what the speaker wants, intends and says. Theoretically, this embedding is RECURSIVE: the speaker must in turn believe that the hearer believes him, etc. The RECOGNITION clause B H(b) — is the consequence part of the illocutionary act, defining its ultimate illocutionary success.

The various sets of conditions for different (classes of) illocutionary acts need not be spelled out here since they have been extensively discussed in speech act theory. What is relevant for our discussion is the set of theoretical primitives needed to be able to characterize them. What else do we need

beyond wants, beliefs, wants and bringing about (DO)?

One further kind of condition is necessary in the characterization of illocutionary acts such as reproaching, accusing, praising, condemning, congratulating, etc, viz that the speaker thinks or finds that something (an object, event or action) is GOOD or BAD, relative to himself, relative to the hearer, or relative to some community or norm of the community. In the examples given, the presupposition is PAST(DO H(b)), embedded in Bs(.) —. Introducing two types of EVALUATION operators, viz EPO9 and E, denoting positive and negative evaluation (like and dislike), respectively, one of the essential conditions to be added is W S(a)Ky(b)E1(a) PAST(DO H(b)p) or its negative variant.

Other differences between the examples of this class of illocutionary acts are to be sought in further presuppositions with respect to the DEGREE of certainty of the speaker, the degree of like or dislike, the

of the object, event or action (dis-)liked, the degree of sincerity of the action, etc. Furthermore, there is an important condition to be formulated in terms

of STATUS, POSITION and POWER, defining the AUTHORITY of the speaker.

These are again specific functions taking participants in some specific context: only AS a judge may a participant condemn/acquit another

participant — having the function of the accused: S(a) = Y (a, c o ), where Y is some position function, defining the role of a in co . In this case, we no longer are dealing with internal structures of language users, but with their SOCIAL FUNCTIONS, which hence should be added to the list of categories of the pragmatic context as specified aboye.

Systematic research is necessary to determine which further categories are necessary in order to define the various illocutionary acts. Methodological problems will certainly arise in that respect, because some differences may not be based on clear pragmatic criteria — relating to the structure of the context — but to other social properties of the situation where the distinction between pragmatics and social theory is admittedly vague. Thus, for instance, would the important notion of POLITENESS be a pragmatic notion or a notion characterizing certain properties of social behaviour in general? Linguistically, the notion seems required in order to differentiate the use of second-person pronouns in many languages (German, French, etc) and other phrases, eg in Would you please be so kindas to give ... versus Please give .. versus Give ... In this case the delimitation between pragmatics on the one hand and 5TYLISTICS/RHETORICS on the other hand adds further confusion. The pragmatic condition would pertain to APPROPRIATENESS of an utterance, whereas the stylistic/rhetorical variations define the degree of EFFECTIVENESS of an utterance, underlying the willingness of hearers at the perlocutionary level. I may have various options to make an appropriate request, but certain requests will be more likely to be complied with than others, according to the degrees of politeness, the measure of preparation of the request (see Chapter 9), and the degree of freedom left to the hearer. At this point of the study of language USE, pragmatics, stylistics and sociology intermingle.

3.7

Whatever the precise delimitation of pragmatics and the set of categories defining pragmatic contexts, the main aim of research should be kept in mind, viz to account for certain systematic properties of language use. As an expedient strategy, then, we only are paying attention to those categories which systematically differentiate the pragmatic functions of certain linguistic expressions. In other words: a pragmatic theory should not merely give independent appropriateness conditions for utterances, but specify which properties of utterances (realized sentences and discourses) depend on these conditions.

Perhaps the most obvious relationship between semantics and pragmatics

is exemplified by PERFORMATIVE SENTENCES, such as I promise to come, I

advise you to go, etc which denote the illocutionary act executed by the very utterance of these sentences in the appropriate context. That is, in the present tense and first person they are pragmatically self-verifying: they are true by the simple fact of being uttered in some appropriate context. s

A more general relationship between pragmatics and semantics (and hence with the other grammatical properties of sentences) is constituted by the

CONTENT of the illocutionary act. Thus, in promíses and threats the proposition expressed must denote a future act of the speaker, in reproaches and accusations a past act of the hearer, in orders, requests and advices a future act of the hearer, etc. Hence personal pronouns, predicates and tenses must be such that those propositions may be satisfied.

In the case of INDIRECT SPEECH ACTS, such as Could you leed me some

money? and There is a bad tyre on that car, COUNTING As a request and a

warning, respectively, the propositional content corresponds to some necessary condition of the illocutionary act, eg pertains to abilities of the hearer or some dangerous state of affairs. 9

Finally, syntax, intonation and particles may be used as INDICATORS of certain illocutionary classes, even if this relation need neither be necessary nor sufficient: thus, the indicative syntactic structure and intonation may correspond to assertion-like or assertion-based illocutionary acts, the interrogative with questíon and request-type acts, and the imperative with commands, threats, etc. In some languages, such as German, Dutch and Greek, specific pragmatic structures may be expressed by PARTICLES. Thus one function of German doch is to express the fact that the speaker assumes that the hearer knows (or should know) already the proposition asserted by the speaker. lo

In the subsequent chapters we will be interested in particular in the systematic relations between certain properties of discourse structure, eg connectives, and properties of illocutionary acts and act sequences. In order to be able to do this, some final remarks are necessary about composite speech acts and sequences of speech acts.

3.8

Illocutionary acts, typically, do not come alone. They are part of SEQUENCES OF ACTION in general or of SEQUENCES OF SPEECH ACTS in particular. These sequences must satisfy the usual conditions for action sequences. Thus, it may be required that the final state (result) of some speech act is a necessary condition for the successfulness of a following (speech) act. In this sense, an illocutionary act may be an AUXILIARY ACT in a course of interaction. If I want to have some book in the possession of a participant in the social situation, I may expect that he will give me the book without my interference, or I may take the book, or I may give some conventional signal to the effect that the book be given to me. In the latter case an utterance of natural language will be the most prominent means to satisfy my wishes. Similarly, changes in the social situation may be brought about by illocutionary acts such that the appropriate conditions are satisfied for other actíons of the hearer: the hearer may act upon the OBLIGATIONS instituted by the illocutionary act. This reaction may again be an illocutionary act, thereby constituting a sequence of CONVERSATION. A accuses B of p, whereupon B rejects the accusatíon or makes excuses for having done p. The conversational sequence, as we said aboye, must satisfy the requirement that the final state s;

after illocutionary act F. is an appropriate initial condition for the illocutionary act F;+ 1 . I may apologize for having done p if I assume that my hearer knows that I have done p and disapproves of p, which assumption may have been (perlocutionarily) brought about by the successful accomplishment of a previous accusation by my hearer, which implies the initial conditions of my apologizing. Note, however, that such sequences of illocutionary acts need not as such be `necessary' : each illocutionary act is, in general, successful with respect to a specific structure of the context, whether this actual context is brought about by an illocutionary act or by another act or event. Therefore, we cannot simply say that an apology PRESUPPOSES an accusation or reproach, unlike the case of QUESTION and ANSWER, which however are not specific illocutionary acts but certain structural functions of illocutionary acts: I may answer a question BY all kinds of illocutionary acts. On the other hand, there are examples in which the interactional sequence is practically wholly verbal. A judge may acquit somebody only after an accusation and after assertions that are sufficient conditions for acquittal.

In the following chapters it will be investigated how an analysis of sequences of illocutionary acts relates to sequences of sentences in discourse, and thus how pragmatic coherence may codetermine the semantic coherence of a discourse. Similarly, as for semantic macro-structures, it may be the case that sequences of illocutionary acts are to be described at an additional level of global speech acts, which would be ajustification of not merely wanting to study isolated speech acts with respect to a context, but whole conversations with respect to a context.

Notes

1 After Peirce's work (see Peirce 1960), it has mainly been Morris (1946) who has formulated the task of a pragmatic component of semiotic theories. For discussion, see Lieb (1976).

2See Austin (1962) and Searle (1969) as the two basic works that have given rise to further developments in philosophical and linguistic pragmatics. In the following we take for granted that the basic results in the philosophical theory of speech acts are known.

3For recent advances in pragmatics, especially linguistic pragmatics, see Bar-Hillel, ed (1972), Kasher, ed (1976), Cole and Morgan, eds (1975), Sadock (1975),

Wunderlich, ed (1972), van Dijk, ed (1975), Wunderlich (1976).

4A similar approach is taken in Groenendijk and Stokhof (1976).

5For a discussion of the notion of 'utterance', see Kasher (1972).

6For instance it may be asked in what sense we may MEAN something without — at least tacitly — realizing some language expression. Secondly, increasingly complex cases may arise in which, ad hoc or under special agreement, we may mean q although we express p, or mean certain implications of p as given by specific pragmatic rules of conversation. For detail of these and similar problems, see current work in the philosophy of language, eg Grice (1971, 1967), Schiffer (1972).

7In addition there is a subset of specific constraints on the epistemic set of speakers. For instance for certain utterances involving modal sentences it must be the case

204 PRAGMATICS

that neither p nor -p is part of the epistemic set, eg if I assert Perhaps Peter is iii. Similarly, in composite sentence utterances it may not be the case that one part has epistemic preconditions conflicting with those of other parts of the utterance, as in

Peter is iii, but I know he isn't, or Maybe Peter is ¡II, but I know he isn't. For details and formal treatment, see Groenendijk and Stokhof (1975, 1976), who have called these kinds of conditions CORRECTNESS conditions, which are pragmatic and given parallel to the normal truth conditions of sentences, whereby sentences may be true but incorrect, or false but correct. Note that correctness conditions are a specific subset of pragmatic appropriateness conditions, because they are formulated in terms of structures of language users in contexts.

For details see Groenendijk and Stokhof (1976) and the references given there to further philosophical work on performative sentences, a notion first discussed by Austin (1961, 1962).

See Searle (1975a) and Frank (1975) for the notion of an indirect speech act.

See Franck (1977) for a discussion of the specific pragmatic function of partícles, especially in German.

SENTENCES in discourse. At the semantic level, we were primarily concerned with relations between propositions, whether these are expressed within the same composite sentence or within severa! sentences. Although sentences and sequences may be semantically equivalent they may reasonably be expected to have at least different pragmatic functions. Other systematic differences in the use of sentences and sequences are stylistic, rhetorical, cognitive and social, and will not be discussed here. It will be argued that the pragmatic distinction between the expression of information in composite sentences versus the expressión of information in a sequence of sentences depends on the intended illocutionary acts, on their internal structure, and on the ordering of such acts.

1.3

The problem of the DISTRIBUTION OF INFORMATION in discourse is not only semantic. In processes of communicative interaction this ordering depends on what we know and believe and on our beliefs about the knowledge of our conversation partners. Similarly, the information ordering is subject to our own wishes and intentions for action and our assumptions about those of the hearer. TOPICS OF CONVERSATION are initiated and changed under these constraints. Information may be more or less 'relevant' or 'important' with respect to a context thus defined. The same facts may be described from different points of view or under different 'propositional attitudes'. It is within such a framework, then, that notions like PRESUPPOSITION (eg versus ASSERTION) and TOPIC-COMMENT require further explication, viz as principies of social information processing in conversational contexts.

1.4

Besides these and other pragmatic properties of connection, coherence, information distribution, sentence and clause sequencing, perspective and relative importance in discourse, this chapter must focus on their relevance for the accomplishment of SEQUENCES OF ILLOCUTIONARY ACTS. That is, we want to know what necessary or sufficient conditions must be satisfied in order for speech acts to be combined, which acts are 'presupposed', focused upon, directly or indirectly intended, and in general how sequences of speech acts are connected and coherent.

2 Sentences and sequences

2.1

Let us start our inquiry into the pragmatics of discourse with a problem of immediate grammatical importance, viz the difference between COMPOSITE SENTENCES and In later sections the more general theoretical background for such a distinction will then be developed.

Consider, for instance, the following pairs of examples:

[1 ]a: Peter had an accident. He is in hospital. b: Peter is in hospital. He had an accident.

[2]a: Peter had an accident. So, he is in hospital.

b:Peter had an accident, so he is in hospital.

[3]Peter is in hospital, for he had an accident.

[4]a: Because he had an accident, Peter is in hospital. b: Peter is in hospital, because he had an accident.

Apparently, there are various morpho-syntactic ways to express the `same' information about an ordered sequence of facts. In all these examples, reference is made to the fact that Peter had an accident and that Peter is in hospital (now) and that the first fact caused the second fact. In other words, the different expressions are semantically equivalent at least in one sense of semantic equivalence: they have the same truth conditions.

Yet, at another leve) of analysis the equivalence does not hold. The differences appear both between sentences with distinct syntactic structure and between sentences and sequences.

Taking the last examples first, we see that subordinated causal clauses may either occur in `first' or in `second' position, viz precede or follow the main clause. Sentence [4]a however may be used in a context in which (the speaker assumes that) the hearer knows that Peter had an accident, whereas [4]b is used in a context in which the hearer knows that Peter is in hospital.' That is, the APPROPRIATENESS of the respective sentences depends on the knowledge and beliefs of speech participants at a certain point in the conversational context. On the other hand, examples [1-3] are normally used in those contexts in which the speaker has no such assumptions about the knowledge of the hearer, or rather in which he assumes that the hearer does NOT know either of the facts referred to. This means that [1]a—[3] would be inappropriate answers to any of the following previous questions of the hearer:

[5]Why is Peter in hospital?

[6]Where is Peter? They say he had an accident.

Sentence [4]b, however, is appropriate after question [5], whereas [4]a, although perhaps a bit awkward, is appropriate after [6].

The complex sentence, apparently, has properties which are similar to that of the TOPIC-COMMENT articulation : `known' elements come in first position, `new' elements in second position. Since the known element in this case is a proposition, we may say that the first clauses in [4] are PRAGMATICALLY PRESUPPOSED. 2 Hence, one of the differences between the sequences and the coordinated compound sentences is that relating to the well-known

distinction: in [1-3] each proposition expressed by the utterance of the sentence or sequence is asserted, whereas in [4] only the second position propositions are asserted and the first position sentences presupposed (in the pragmatic sense of this term, ie assumed by the speaker to be known to the hearer). Yet, there is a difficulty, because we may also maintain that both [4]a and [4]b, taken as a whole, are assertions. Below, we therefore will have to find out whether two different meanings of the term `assertion' play a role here.

208 PRAGMATICS

2.2

More crucial for our discussion, however, are the differences between [1la, [ 1 ]b, [2]a, [2]band [3] : what implications does the ordering of sentences have, if not presuppositional, and in what respect are compound sentences different from their corresponding, ie semantically equivalent, sequences?

Although [1 ]a and [1 ]b are equally appropriate in many contexts, there are also contexts in which the first seems more natural than the second, eg after a question like

[7] What happened to Peter?

eg on seeing his car badly damaged. On the other hand [1]b seems more appropriate after a question like

[8] Why doesn't Peter answer his telephone?

That is, the reason he doesn't answer his telephone is the one requested by the previous speaker, and in the answer this information is given first. The second sentence in [ 1 ]b then gives an EXPLANATION of the fact referred to by the first sentence. In [1 ]a no such explanation is given, only a representation of the facts, implying that the first fact caused the second. This relation between the ordering of facts and the ordering of clauses or sentences in a sequence will be further discussed below.

It might be argued that [2]a merely explicitly expresses the causal connection which in [1]a is only 'expressed' by syntactic ordering,' and that the same holds for [2]b. Again, however, there are different contextual conditions, hence pragmatic differences between [1 ]a, [2]a and [2]b. Sentences like [2]a are typically used when CONCLUSIONS from certain facts are to be drawn RELATIVE TO A GIVEN SITUATION. If during a board meeting several members do not show up, the president may say "Harry had to meet Pierre Balmain. So, he is in Paris", and then utter sentence [2]a, possibly by stressing Peter and he. In this respect, [2]a shares a pragmatic function with [l]b, viz it draws attention to the fact which is of primary importance or RELEVANCE FOR A CERTAIN SITUATION, but in [2]a this fact is not only referred to as a factual consequence but also as a conclusion drawn explicitly by the speaker. This is typically the case in those cases where only indirect factual evidence is present, eg in Peter's car is damaged. So, he must have had an accident. This does not hold for interclausal so in [2]b, which only expresses, coordinatingly, the causal connection between the two facts referred to by the respective clauses. ° Hence, in [2]b so is a proper whereas in [2]a sentence initial So, followed by a pause, rather relates utterances or illocutionary acts, viz those of premise and conclusion. s In that case we may speak of a

In some languages, eg in Dutch and in German, the difference between semantic and pragmatic so (dus and also respectively) may also show in the syntax. Inter-clausal (semantic) connectives are followed by Verb-Subject

ordering, whereas sentence-initial connectives followed by a pause may also have normal Subject-Verb ordering. b

Finally, we may use interclausal for in order to relate a fact which has, so to speak, `pragmatic prominence' focused on its cause or reason, much in the same way as [4]a — althoughfor-clauses cannot be presupposed. The difference with [1 ]b is that for-clauses do not have an explanatory function; they merely state a condition of another fact stated before, in the same way as interclausal so states a consequence of a fact stated before.'

2.3

Until now we have met the following differences between sentences and sequences, eg in [1-4]: different presuppositions, ie different knowledgebelief structures of the context, focus on the reason/cause or on the consequence, the relevance or importance of a certain fact for the present context, eg the interests of the hearer, indications of conclusion or explanation as specific acts. Some of these differences are rather vague and require further definition. The notion of RELEVANCE or IMPORTANCE, with respect to a certain context, should for instance be defined in terms of the action theoretical semantics used earlier in this book. In that case a fact, and hence the knowledge of such a fact, is important relative to a context or in general to a situation if it is an IMMEDIATE CONDITION for a probable event or action (or prevention of these) in that context or situation. In the board meeting situation, the proposition `Peter can't come' is more directly important for that meeting than the reason `Peter is in hospital', which is in turra more relevant than the fact that he had an accident. On the other hand, in a situation in which Peter's wife is informed of the events, the information about the accident may well be much more important than the fact that he is in hospital, which are both more important than the fact that he did not have dinner that night.

Similarly, notions like that of FOCUS and PERSPECTIVE must be made explicit to account for the differences. Thus, a sequence of facts may be described from the point of view of the time, place and involved agents of the action or event, but also from the point of view of the observen or `informant' at the time-place of the context. In the first case we may have compound sentences with semantic connectives, in the latter case, a sequence with a pragmatic connective may be more appropriate:

[9]I felt ill, so I went to bed.

[10]Peter is ill. So, he won't come tonight.

Typícal for pragmatic connectives, which may be considered as INFERENTIAL ADVERBS, is that they cannot be preceded by and, whereas semantic so can be preceded by and. Sentence initial So, used to make inferences is also typically used in dialogues, as in:

[11]A. Where is Peter?

B. He is in hospital. He had an accident. A. So, he won't come tonight. Let us start.