Expressive content as conventional implicature

In defining conventional implicature, Grice (1975) supplied terminology for reasoning about a group of secondary lexical entailments indicating speakers’ attitudes toward (parts of) their main assertions. The passage in question is potentially ground breaking: it draws a restrictive boundary around (Grice’s) pragmatic theory, and it homes in on a complex mode of expression that operates under the radar of most linguistic operators (e.g., verbs of saying, negation). Unfortunately, Grice’s (1975:34–35) discussion is based on an unconvincing example (therefore). Subsequent research enriched the factual basis only slightly (still, but and its synonyms). The evidence seemed not to match the potential of the original definition; at least one commentator (Bach 1999) has challenged the very existence of CIs, based largely on the apparent meagerness of the support. This bleak picture does not accurately reflect the status of CIs in natural language. In truth, Grice’s (1975) original definition picks out a broad range of expressions (Potts 2003b). The concern of the present paper is expressive (emotive, affective) modification. Many items falling under this heading match perfectly the definition of CIs that Grice laid out. Attention to their semantics validates Grice’s original definition and answers the question of how to manage expressive content, which is ubiquitous in discourse but challenging to assimilate to existing semantic theories. The class of expressive modifiers is itself huge and diverse. I limit attention mostly to expressive attributive adjectives (EAs) such as damn in (1a) and epithets like the expressions highlighted in (1b–d). Huddleston and Pullum (2002:36) identify the content of EAs as conventionally implicated; extending this insight to epithets systematizes existing observations about their interactions with commanding operators (Aoun et al. 2001).