Expressives and argument extension

ORLANDO, ELEONORA; LO GUERCIO, NICOLÁS FRANCISCO

Workshop; SALT 32 - Semantics and linguistic theory; 2022

Linguistic Society of America

According to Potts (The logic of conventional implicatures, 2005), expressive adjectives (EAs) and epithets constitute a natural semantic class: neither kind of expression make any at-issue contribution, and both possess functional, non-at-issue expressive meanings. In addition, both compose with other meanings via the rule CI application (see Potts, 2005, p. 165), that is, they take an at-issue argument and return the same argument plus an expressive meaning in the non-at-issue dimension. Despite these similarities, EAs and epithets show a noteworthy difference, to wit, only the former exhibit what Gutzmann (The grammar of expressivity, 2019) calls argument extension, that is, a mismatch between syntax and semantics whereby EAs target a syntactic constituent other than the one they directly modify. Thus, the speaker of (1) can hardly be taken to hold a negative attitude towards cakes in general; the most natural reading, instead, is that she holds a negative attitude towards a particular cake denoted by the DP ‘the cake’, or alternatively, towards the whole situation expressed by the entire sentence. By contrast, in (2) the speaker’s negative attitude clearly targets John:1)The dog has eaten the damn cake.a.#The ☹️cake/☹️The cake/☹️The dog has eaten the cake.2)That bastard John has eaten the cake.a.☹️JohnIn this article, we will advance a semantico-pragmatic view of EAs that explains this contrast. More specifically, we will argue that EAs are isolated CIs, that is, expressions that bear no at-issue content and exhibit a propositional, non-functional non-at-issue content. The present view explains the contrast between EAs and epithets in terms of a difference in expressive content. EAs bear a propositional expressive content, hence they do not combine with other expressions in terms of functional application. Moreover, EAs’ expressive content is unspecific regarding the target of the attitude attributed to the speaker. Together, these two facts leave the possibility open for the audience to make different pragmatic inferences regarding the target of the speaker’s negative attitude. Argument extension results from these inferences. By contrast, epithets have functional expressive meaning and apply directly to the constituent they modify via the rule CI application, just as in Potts’ view. Hence, epithets leave no room for the triggering of pragmatic inferences in the audience concerning the target of the speaker’s negative attitude.In addition, we will argue that the view has further advantages, to wit, besides accounting for some known facts concerning EAs, like their impossibility to appear in predicative position and to combine with degree modifiers, it allows explaining some facts that have not been discussed so far in the bibliography on EAs so far, namely that EAs cannot only target asserted contents but also conversational implicatures, presuppositions and even further non-communicated contents that may be mutually manifest in the context.