Pragmatic Abilities in Autism Spectrum Disorder: A Case Study in Philosophy and the Empirical

away from these: To do otherwise would be to make our proposal unduly speculative.) Next, we explain how positing two processes accounts for our data about pragmatic abilities in ASD, while leaving the deficit-based methodology safe from the threat seemingly posed by the abilities we have uncovered. As a first pass at the contrast between primary and secondary pragmatic processes, consider functional characterizations of each. Beginning with secondary processes (simply because they are more familiar), they are always postpropositional. That is to say, secondary pragmatic processes always take propositions as input. More specifically, focusing on the speech act of assertion, they take the proposition literally stated as input, and yield propositions that are nonliterally conveyed. In the context of communicative abilities in ASD, it is useful to divide these into two classes: non-ironic cases on the one hand—conversational implicatures, metaphor, and indirect speech acts—and irony on the other. (We will see why shortly.) Primary pragmatic processes, in contrast, take the standing meaning of expressions as input (this being assigned by semantics in our sense) and yield the literal content of the speech act as output. Again, in the context of ASD, it is useful to divide these into subvarieties. On the one hand, there are inputs whose content is less than fully propositional. These include expressions that contain overtly context-sensitive expressions like pronouns (“he,” “this”), tense markers, and special words like “now” and “here,” as well as expressions that have multiple senses. If primary pragmatic processing does not occur in these cases, no truth condition results. For example, the sentence type “He lived there until yesterday” patently has no truth value, even once all the nonlinguistic facts are fixed, for there is no specification of who is being spoken about (nor when, nor where). This must be settled by what we call “slot filling.” Or again, “The bank was flooded,” that type, lacks a truth condition because it is ambiguous. (In fact, it is both lexically and structurally ambiguous.) On the other hand, there are inputs to primary pragmatic processes where the standing meaning is a proposition, but where the literally stated proposition is not the semantically encoded one. Thus, in reporting upon a party, someone who says “Everyone got drunk” produces a sentence that (ignoring tense) actually does express a proposition—namely that for every person in the universe, there is some time in the past at which they got drunk—however, this is not what is claimed. In the literature, this is referred to as “free enrichment.” In sum, we have (iv) Subvarieties of Primary Pragmatic Processes (a) Input is pre-propositional (1) Assignment of referents to context-sensitive items (slot filling) (2) Selection of sense (disambiguation) (b) Input is propositional (free enrichment) Now, as this last example already hints, these subvarieties are not mutually exclusive. Very often more than one applies before the literal speech act content is arrived at. Consider, for instance, typical uses of “I’ve had breakfast” and “You’re not going to die.” Slot filling is required for the pronouns and for tense. But even once this happens, the results—namely that there is some time in the past at which the contextually fixed speaker breakfasted and there is no time in the future at which 312 Jessica de Villiers, Robert J. Stainton, and Peter Szatmari