King and Stanley on Truth-Conditional Pragmatics

Last night I read Jeffrey King and Jason Stanley's very interesting piece Semantics, Pragmatics, and the Role of Semantic Content. Jeff and Jason raise various concerns about certain forms of semantic minimalism and truth-conditional pragmatics. Let me try to re-formulate one of their concerns about one form of truth-conditional pragmatics in my own words.

Truth-conditional pragmatics distinguishes between the semantic content of a sentence relative to context and the speech act content of a sentence relative to context. The speech act content of a sentence relative to context is the content asserted by the speaker in that context. A sentence is true relative to a context c, on this account, iff the speech act content of the sentence at c is true at the index determined by c.

The worry now is this. Though the semantic content can be defined, it seems to play no significant role in the semantics. As we have just seen, it plays no role in the definition of truth-at-a-context. Admittedly, it does constrain what a sentence can be used to assert. But this feature of semantic content is pragmatic not semantic.

I think, however, that defenders of truth-conditional pragmatics may reply as follows. Consider the following analogy. One might worry that certain forms of eternalism run into similar trouble. Take, for instance:

(1) John is nice.

Given certain forms of eternalism, the semantic content of (1) relative to a context is the temporal content that John is nice (no time constituents). But the speech act content of (1) relative to a context is an eternal proposition, viz. the proposition that John is nice at t*, where t* is the time of speech. But now, one might ask, what role does the temporal content play in the semantics? Well, the eternalist will say, the temporal content does in fact play an important role in the semantics: it is the sort of content that tense operators operate on.

Perhaps defenders of truth-conditional pragmatics could reply in a similar way. Consider, for instance:

(2) France is hexagonal

Some defenders of truth-conditional pragmatics broadly defined (for instance, Soames in recent work) would say that the speech act content of (2) relative to a normal context is the proposition that France is roughly hexagonal. The semantic content of (2) is the content that France is hexagonal. Does the semantic content play a role in the semantics? Yes, it does. For 'strictly speaking' and 'loosely speaking' operate on the semantic content. As Lewis pointed out, 'Loosely speaking, France is hexagonal' is true under high standards of precision iff the semantic content of 'France is hexogonal' is true under low standards of precision.

Of course, this reply may not work in every case, but at least it is not obvious that the semantic content plays no role in the semantics just because it does not play a role in the formulation of truth-conditions for occurrences of sentences.