Modeling “non-literal” social meaning

Abstract

Truth-conditional and socially indexical meanings have traditionally been studied in separate subfields. However, recent years have seen promising attempts to unify the semantics and pragmatics of the two (e.g. Smith et al., 2010; Acton and Potts, 2014). In particular, Burnett (2017, 2019) introduces a formalization of social meaning in terms of the Rational Speech Act (RSA) paradigm (Goodman and Frank, 2016). Building on this work, we address a central observation of contemporary sociolinguistics, that a linguistic variant may be used to index only some aspects of a speaker’s identity. For instance, an adult can use childlike language features to convey not that they are a child (first order indexicality), but that they have certain traits associated with children, like cuteness or innocence (second order indexicality). Similarly, Eckert (2008) notes that some suburban Detroit teenagers use phonetic and syntactic forms conventionally associated with urban Detroit (such as vowel backing and negative concord), and hypothesizes that this is not to signal urbanity (i.e., I am from urban Detroit) per se, but rather to affiliate with certain perceived aspects of urbanity, such as being “autonomous, tough, and street-smart”.