Believing the formless?

In this note, we discuss the analyticity puzzle affecting the logicality of language hypothesis. The analyticity puzzle is the fact that only some analyticities result in ungrammaticality, which seems to conflict with the idea that an inferential device plays a role in determining the set of the possible sentences of the language. The literature includes two solutions to account for this puzzling evidence. According to one of the solutions, the deductive system can access both ungrammatical and grammatical trivialities, though only the latter can be rescued, i.e. made informative, via application of a pragmatic repair strategy, which modulates the meaning of the nonlogical material. It is then argued that syntax only excludes logically trivial (i.e. unsalvageable) structures, and that nonlogically trivial structures may even be used under their trivial readings. Our focus in this note is on a possible implication of this discussion for the analysis of belief ascriptions. In particular, we discuss that occurrences of the formula ‘Bel p’ are acceptable when p is nonlogically trivial but unacceptable when p is logically trivial. Since the ascribed propositions differ just on a logical dimension, we suggest, against classical discussion, that belief ascriptions are sensitive to logical considerations.