Semantic Competence

  • M. J. Cresswell
Part of the Studies in Linguistics and Philosophy book series (SLAP, volume 36)


Chomsky’s distinction between linguistic competence and linguistic performance (Chomsky, 1965, 3–15) is well known, widely discussed and contentious. For all its difficulties it seems to me a vital distinction and I want, in this chapter, to show how an account of semantic competence can be given which links it directly with semantical theories of a truth-conditional kind.


Truth Condition Actual World Native Speaker Semantic Performance Linguistic Competence 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    A great deal of the discussion of competence and performance has been focused on the role of the ‘ideal speaker-hearer’. It may be that my construal of the distinction as one between the sentences a speaker accepts and the sentences he produces is itself a tendentious one but there nevertheless does seem to me an ability, which native speakers have, to distinguish between the sentences which are grammatically acceptable and those which are grammatically unacceptable. This ability is evidenced in part by what native speakers say they find acceptable or unacceptable although, as mentioned in Section 4, only in part.Google Scholar
  2. 2.
    Any semantics which proposes a base in formal logic (e.g. McCawley, 1971; or Lakoff, 1972) is probably, even if without knowing it, coming down in favour of a truth-conditional view of semantic competence.Google Scholar
  3. 3.
    Katz and Nagel, 1974, 313. Their main point is to show that a theory based on Carnapian meaning postulates is not adequate as a theory of semantics. This they do well, though the introduction to their paper gives the unfortunate impression that Carnap’s is a ‘most recent form of intensionalist theory’. They do not mention possible-worlds semantics.Google Scholar
  4. 4.
    I have to say ‘the English sentence’ in case it should happen that that same sentence has, in another language, a different meaning. (Or even e.g. that an eccentric English speaker uses the words ‘cat’ and ‘mat’ to mean what the rest of us mean by ‘dog’ and ‘log’.)Google Scholar
  5. 5.
    A point which will emerge later, when we discuss theories of semantic performance, is that the description of this ability, makes no reference to the purpose of uttering (2). Why someone, the child, his teacher, or anyone else should choose to utter this sentence is not, in the view of this paper, involved in the notion of semantic competence.Google Scholar
  6. 6.
    One approach is to treat each contextual feature such as e.g. time, place, utterer… etc. as an ‘index’ (Montague, 1974; Scott, 1970, 143ff; Lewis, 1972, 174–6) and say that each sentence determines a set of possible worlds only with respect to a complete assignment of contextual indices. The possible world itself can also be regarded as an index and in that case what a sentence determines when the complete context is supplied is simply a truth value. I have suggested (Cresswell, 1973, 114) that contexts of use should be construed as properties of utterances. A quite different analysis, which construes all sentences as embedded in performatives, has been advocated by Lakoff (1972) who claims (p. 569) that the need for contextual indices disappears. Still another approach, in the Davidson tradition, is advocated by Tyler Burge (1974).Google Scholar
  7. 7.
    See chapter 1 above. A ‘possible world’ simply means a way that the world might be, when this is understood as a complete alternative to the actual world. The completeness of the alternative possible worlds means that it is a mistake to think of a possible world as in some way ‘out there’, a disconnected part of the actual world (much as a theological heaven or hell). Any world like that would of course be a part of the actual world. (For similar reasons it would be logically impossible to, actually, make a journey to a non-actual world.) Possible-worlds theorists have sometimes been spoken of as if they believed that possible worlds are actual, but of course they do not. (For the use of the word ‘actual’ in possible-worlds semantics vide Note 8.)Google Scholar
  8. 8.
    The word ‘actual’ simply means the world the speaker is in. A speaker in another possible world who refers to the ‘actual’ world refers of course to his world, not ours.Google Scholar
  9. 9.
    It has been proposed (Morton, 1973) that instead of taking other possible worlds as theoretical entities on which to base semantics we take parts of the actual world which ‘represent’ these other possible worlds. My suspicion is that this will not make for as simple or general a theory, but nothing in this paper would be incompatible with such an approach.Google Scholar
  10. 10.
    Some of Jackendoff’s work on picturing (1975) seems to have links with possible-worlds semantics, at least for the syntactic category of sentence. It is a bit harder to see how to picture a functor like e.g. negation. One difficulty with the picture theory of meaning is of course to find something that a false sentence could be a picture of. Possible-worlds semantics can at least solve that problem since a false sentence is a picture of worlds other than the actual world.Google Scholar
  11. 11.
    Of course the possible-worlds theorist will argue that a truth theory of the Davidson kind will need, in order to give an adequate analysis of such intensional constructions as modals and counterfactual conditionals, to have recourse to things which will turn out to look so like possible worlds that the smaller ontological commitment will be shown to have been superficial only.Google Scholar
  12. 12.
    It is becoming best known to linguists in the form of ‘Montague Grammar’. This term can be said to have a wide use, in which it means any formal representation of a natural language which makes use of possible-worlds semantics; and a more restricted use, in which it refers either to Montague’s own work or to papers which have explicitly regarded themselves as extending it. Montague’s papers on the philosophy of language have been collected (Montague, 1974). Subsequent work by others has been collected by Barbara Partee (1976). A good short introduction to possible-worlds semantics is found in Lewis (1972), though for obvious reasons I tend to have at the back of my own mind the framework I set out in Cresswell (1973).Google Scholar
  13. 13.
    It is one of the contentions of Cresswell (1973) that very few subject terms are names of entities, even those which, like definite descriptions, have been thought to be names. It is even suggested (pp. 131ff) that cases like ‘Fido’ could also be construed as like definite descriptions. However that question does not affect the present discussion.Google Scholar
  14. 14.
    I discuss this example a little more fully in Cresswell (1975a). One problem for any discussion of semantic deviance is to recognise the difficulties produced by the existence of metaphor (Harrison, 1974, 600). Whether the semantic anomaly is explained as in Cresswell (1975a) by domains of functions or by the KF device of ‘selectional restrictions’ there will always be cases in which anomalous sentences are deliberately used for some stylistic effect. All I want to say here is that a formal analysis of metaphor is something we need as much as we lack.Google Scholar
  15. 15.
    Synonomy is a stronger relation than logical equivalence. We make some observations about this (both in the text and in footnote 20) when discussing some remarks made by Katz (1972).Google Scholar
  16. 16.
    To say that A and B are contraries is just to say that there is no world in which they are both true; to say that they are contradictories is just to say that the worlds in which A is true are precisely those in which B is not.Google Scholar
  17. 17.
    Writing ‘everyone’ as ∀, ‘someone’ as ∃ and ‘loves’ as F we get versions more familiar to logicians: (∀x)(∃y)Fxy and (∃y)(∀x)Fxy. More sophisticated ways of representing these sentences are given in Cresswell (1973, 90ff).Google Scholar
  18. 18.
    That is assuming that (Seat for one) cannot be further decomposed. One of the difficulties with the marriage of lexical decomposition and Katz’s view of the basis of semantics in terms of theoretical entities is that until we have some guarantee that lexical decomposition has gone as far as it can we have no idea of what the basic theoretical entities might be. In this respect the analogy with sub-atomic physics certainly breaks down; and so does the analogy with possible-worlds semantics. For in the latter cases we do know what the primitive entities of the theory are.Google Scholar
  19. 19.
    This is not quite accurate enough. As well as a reference to a possible world there ought to be reference to a moment of time since the same thing can undoubtedly be a chair at one time but not at another. Context-dependence of a temporal sort for the predicates which represent common names has been recognised both in possible-worlds semantics, e.g. Cresswell (1973, 180), and in KF semantics, Katz (1972, 303ff). I hope that in the light of what has already been said it should be clear that the circularity here is only apparent. In using the word ‘chair’ in my English metalanguage I have of course been trading on my reader’s knowledge of that metalanguage. Thus the manner in which the set of worlds in which x is a chair has been presented to him has used the word ‘chair’. Nevertheless it is that set of worlds, however presented, which constitutes ow(a). Google Scholar
  20. 20.
    For instance, suppose Katz’s (1972, 50) definition of analyticity is accepted and we agree to distinguish between analyticity and necessary truth (182–4). Then lexical decomposition may be necessary to the definition of analyticity. Another motivation is, I think, the belief that e.g. we are more certain that ‘x is a bachelor’ entails ‘x is male’ than we are of just which worlds are included in each proposition. My own inclination in this latter case is to link it with an explanation of vagueness in terms of a ‘communication class’ (Cresswell, 1973, 59) and say that although different sets of worlds may be assigned to ‘x is a bachelor’ and ‘x is male’ in each member of the communication class of English (i.e. in each way of making precise all the meanings of all words), yet in each of these (different) precise evaluations the set assigned to ‘x is a bachelor’ will be a subset of the set assigned to ‘x is male’. The notion of a ‘communication class’ (called by van Fraassen a ‘supervaluation’: van Fraassen, 1971, 94–6) has, I believe, a link with meaning postulates of a Carnapian kind. A Carnapian meaning postulate may be regarded as a metalinguistic statement of certain relations between symbols which hold in every member of the communication class of (say) English. It is important however to realise that meaning postulates are not theoretically necessary to the existence of the communication class, and also that they are not, in possible-worlds semantics, necessary to the definition of such notions as necessary truth or entailment.Google Scholar
  21. Another motive for lexical decomposition has been the need to give a truth-conditional semantics for propositional attitudes. Since any two logically-equivalent propositions are true in exactly the same possible worlds, then if we identify a proposition with a set of possible worlds we cannot admit distinct but logically equivalent propositions. One way of dealing with this problem is to follow Lewis’ lead and say that propositions are structured entities which reflect the way they are made up from simple meaning units. This idea is applied to propositional attitudes in Cresswell (1975b) and (1985a) and in Bigelow (1978). There is obviously a link between this view and the lexical decomposition hypothesis although its outworkings seem to be very difficult and baffling.Google Scholar
  22. 21.
    E.g. I have been told (by John Bigelow) that one of the earliest functions of language was in story-telling. Whether this is wrong or right it seems to me at least plausible to suggest that it is in the development of the imaginative faculty rather than in the necessity for day-to-day communication that we must seek the central role of language.Google Scholar
  23. 22.
    The notion of illocutionary force is based on Austin’s notion of an illocutionary act. It is used by Searle (e.g. 1969, 30).Google Scholar
  24. 23.
    With the exception of course of KF semantics which is certainly compositional. However, as I have said, I do not regard KF semantics as in competition with truth-conditional semantics; I regard the latter as a way of giving content to the former.Google Scholar
  25. 24.
    It seems to me that the idea that performatives have no truth value may have been counterproductive in their analysis. For surely what distinguishes performatives from other sentences is simply that their utterance, in felicitous circumstances, is what creates the fact that makes them true; i.e. it seems that, in human affairs, there are a large number of institutional relationships (e.g. the relationship of being under a promissory obligation) which require a convention for their creation, but a convention whose precise nature is somewhat arbitrary. In such a case what better way to create the relationship than by saying that it exists or that it has been created? But of course this analysis of performatives requires that the sentences which create the relationship can have a truth value.Google Scholar

Copyright information

© Springer Science+Business Media Dordrecht 1988

Authors and Affiliations

  • M. J. Cresswell
    • 1
  1. 1.Department of PhilosophyVictoria University of WellingtonNew Zealand

Personalised recommendations