At-issueness Does Not Predict Projection

The notion of PROJECTION was introduced in Langendoen & Savin (1971) and has since been employed to describe the ability of certain implications to survive embedding under entailment-canceling operators, e.g. negation or modals. The term has traditionally been applied to presupposed inferences (Karttunen 1973; 1974; Heim 1983; Beaver 2001; a.o.) and has more recently been discussed in the context of non-presupposed projective content, including inferences triggered by appositives, expressives, and evidentials (Potts 2005; Koev 2013; 2016; Murray 2014; AnderBois et al. 2015). In turn, Potts (2005) introduced the term AT-ISSUE content to describe implications that constitute the main point of an utterance. In a series of recent papers, David Beaver, Craige Roberts, Mandy Simons, and Judith Tonhauser (henceforth BRST) made the influential proposal that projection can be explained in terms of at-issueness, hypothesizing that there is a perfect correlation between these two linguistic categories, in the sense that semantic content projects if and only if it is not at issue (see especially Simons et al. 2010 and Beaver et al. 2017). This squib critically evaluates BRST’s proposal and raises several theoretical and empirical issues. I argue that although projection and not-at-issueness appear to be strongly correlated, there is no perfect overlap in the way envisaged by BRST. In particular, semantic content may project and be at issue, and it may not project but be not at issue. 1. Defining and Diagnosing Projection Projection is the ability of semantic content to remain unaffected by the presence of entailmentcanceling operators such as negation, modals, etc. Simons et al. (2010) define this notion as follows.