Mainstream linguists have now begun to take an interest in the way the formal resources of language are put to use by people in communicative situations. A powerful impulse for this awakening interest came from Grice's paper ‘The logic of conversation’; (1975). Grice's work has begun to he examined, criticized, and developed (see, for example, Wilson, 1975; Kempson, 5975), and prospects are now emerging of a much more satisfying and comprehensive picture of language at work in a communicative setting. The present paper is intended as a contribution to this developing field.

Postal (1968: 300) writes as follows:

Bloomfield in effect noted the incompatibility of phenomena like … metathesis, dissimilation, etc., with the view of sound change as gradual drift in performance. Instead of admitting that these were counterexamples to the view, he simply appealed to terminology and declared that they were not sound change … But, since these types of changes are as systematic, regular, cross-lexical, etc., as other changes, this terminological attempt amounts to nothing more than the naming of a class of exceptions.

The performative analysis claims that every sentence we utter refers, self-reflexively, to our utterance of it, that every sentence comes equipped in its underlying structure with a higher performative clause of the form I (or Speaker) + Verb of Communication + You (or Hearer) (see especially Ross, 1970; Lakoff, 1971, 1972). The self-reflexive nature of the performative clause is brought out even more clearly by the addition of hereby, which functions as a deictic instrumental adverb referring to the utterance-act. In this paper I shall deal with two sets of data that have been invoked in support of this analysis and I shall try to show that on closer scrutiny they disconfirm it, at least in its generally accepted form. Both concern sentence adverbials that modify not the propositional content of the sentence to which they are attached but what I shall provisionally call the pragmatics of the speech situation; the solutions I shall suggest as an alternative to the performative analysis will, however, differ considerably for the two sets.

I. Despite much activity, the recent attention paid by generativists to diachronic syntax has led to no significant implications for a general theory of grammar and, from a purely historical viewpoint, has failed to focus on any well-defined concepts of historical change. The papers in this field stand largely in isolation from each other and show few common threads of interest or argumentation; there is little agreement on even the most fundamental concepts. Modern diachronic syntacticians, it is true, have no legacy equivalent to what was handed down to phonologists, and they must do their own pioneering work. This will preclude rapid progress, but it does not explain the current disarray of the field. The fault for that lies in ourselves and is a consequence of a theoretical approach to syntax precisely as barren as that of the neogrammarians.

Linguists referring to ‘fast speech’ and to ‘normal speech’ often do not share the same definitions of the two terms. ‘Fast speech’ is normally intended as increase in speech tempo that affects the normal/slow norm, but it is also recognized that ‘fast speech phenomena’ occur in casual speech that is not necessarily rapid, and that ‘casualness’ of speech is a complex sociolinguistic notion the components of which are hard to isolate (see, for instance, Dressler, 1974, and Vanecek et al, 1975, for a discussion of the correlation between decreased attention and phonological casualness, and other factors referred to in Dressler, 1975). It is not at all clear what ‘normal speech’ is either. It is usually identified with ‘maximally distinct/differentiated speech’, but as pointed out to me by Arnold Zwicky, this is probably not the case, since there are processes that apply to give 'careful‘ or ‘;emphatic‘ speech (as when the l in please is made syllabic). Zwicky believes that normal speech ‘is related to whatever level of representation (roughly the phonemic) figures in normal rhyming (hat and had will not count as rhymes, even if they are pronounced the same, while stepped and slept will count as rhymes, even if they are pronounced differently, say without the t in slept)’ (personal communication). This paper does not attempt to tackle the problem in its entirety; it will only deal with the tempo aspect of fast speech and with the theoretical implications of increase in rate of speech. The reason for concentrating on tempo is that it can be isolated more easily than the various factors, social and other, involved in casual speech. It will also be assumed that speech tempo is relative, i.e. that one man's normal rate of speech may be another man's fast speech tempo, and that although increase in speech tempo may sometimes result in increased attention and thus in retention of the overall level of formality (Vanecek et al, 1975), the common tendency is for increased reduction and assimilation with the increase in the rate of articulation.

The impetus for this paper is a curious case of gender disagreement in Scandinavian, whereby the rule assigning gender agreement to predicate adjectives is apparently violated. An analysis of the facts is provided, and it is shown that other semantico-syntactic phenomena of Scandinavian can be handled within the same explanatory framework.

In his (1970) article ‘Surface structure constraints in syntax’ Perimutter has argued that the restrictions on the combination of the weak atonic forms of the object pronouns in Spanish and restrictions on their order cannot be accounted for naturally either in the deep structure, by constraints on the phrase structure rules, or by constraints on the rules which introduce these pronouns.1 He concludes that a transformational generative grammar must be supplemented by positive output constraints which act as a filter accepting only sequences which match the filter and rejecting the rest as ungrammatical.