Perspectival expressives

Expressives, i.e. words such as “damn” or “bastard” are perspective-dependent: their content is always evaluated from a certain perspective. Because expressive content projects out of all environments, this perspective is normally the speaker's. Perspective-dependence of expressives can be investigated by focusing on exceptions to this rule. Potts (2005) presents an influential theory of expressive content as a kind of conventional implicature. It is a definitional feature of expressive content on this account that it is always speaker-oriented. This claim has quickly come under criticism, and a variety of counter-examples have been offered (cf. Amaral et al. (2007), Lasersohn (2007), Potts (2007), among others). Harris and Potts (2009) consider examples of non-speaker-oriented expressives given in the literature, as well as experimental data, and argue for an explanation based on a mechanism of pragmatic perspective shift (as opposed to an approach based on semantic binding, as in, e.g., Schlenker (2007), Sauerland (2007)). The objective of this paper is to develop a theoretical understanding of such a mechanism. The approach suggested is based on a model of discourse pragmatics which focuses on commitment attribution as an element of hearers' interpretation (based on Morency et al. (2008) and Lewis (1979)). At-issue commitments are distinguished from commitments de lingua (cf. Harris (2014, 2016)). It is a characteristic property of expressives as a lexical class that they are pragmatically “opaque” and always raise the issue of de lingua commitment. The orientation of expressive content cannot be strictly predicted, and thus a fully formal treatment is implausible, but the theory offered here accounts for all factors that influence non-speaker-oriented readings, as well as for the very strong bias towards speaker-oriented ones. A limited, testable prediction of the account is presented.