Kent Bach

In _Imagination and Convention_ Lepore and Stone make two sweeping claims about language, convention, and communication. One is that linguistic communication is of what is conventionally encoded. The other, complementary, claim is that when speakers use language in nonconventional ways, their intention is not to communicate some specific thing but rather to invite the hearer into a bit of “imaginative engagement.” So understanding an utterance requires no more than disambiguating it; insofar as imaginative interpretation is required, its aim is distinct from understanding the utterance. I agree with L&S that linguistic convention is much more comprehensive than traditionally supposed and that language is often used figuratively without specific communicative intentions, but their two claims go implausibly further. Both are subject to counterexample and counterargument, and rely on reasoning that downplays some distinctions and disregards others, as abetted by casual use of such key terms as “meaning,” “interpretation,” “convention,” and “Gricean.”

In “Russell’s Theory of Definite Descriptions” (2005), Stephen Schiffer argues that referential uses of incomplete definite descriptions pose a daunting problem for Russell’s theory. Attempted solutions face a certain dilemma. Either they must require that the speaker mean a proposition that is somehow both descriptive and object-dependent or else mean two things, an object-dependent proposition as well as a descriptive one, and thereby not be speaking fully literally (the latter option would also get wrong the structure of the hearer’s inference). This chapter contends that Schiffer operates with an overly liberal conception of literality and that he saddles Russellians with certain assumptions they can do without and with too strict a Gricean model of how hearers recognize referential uses. The proposed Russellian account treats both referential and attributive uses of incomplete descriptions as special cases of the ubiquitous phenomenon of meaning something one does not make fully explicit.

Does the ‘Contextualist Platitude’, that linguistic meaning generally underdetermines speaker meaning, undermine the semantic-pragmatic distinction and support radical-sounding views that go by names like ‘contextualism’ and ‘truth-conditional pragmatics’? It does not. It does not require a radical reconstrual of semantics, at least not if we recognize that what is determined compositionally by the meanings of a sentence's constituents and their syntactic arrangement is often not a complete proposition. Contextualism trades on a number of pernicious ambiguities, relies on ‘semantic’ intuitions that are in fact sensitive to pragmatic regularities, and attributes far greater powers to conversational context than it actually has. This chapter substantiates these claims on the basis of some simple arguments for the following ten points: semantics concerns sentences, not utterances; saying something is one thing, stating or otherwise meaning it is another; semantic content always underdetermines speaker meaning; we generally don't make fully explicit what we mean, and what we don't is not part of what we say; the semantic content of a (declarative) sentence cannot be equated with what it is normally used to assert; pragmatic regularities give rise to faulty ‘semantic’ intuitions; focusing on sentences representative of those we use or might use is to commit a massive sampling error — even unusable sentences have semantic contents, however unintuitive; what is said (in the locutionary sense) matters even though understanding an utterance often does not require entertaining or representing it; context does not literally determine what is said or what is meant; and demonstratives and most indexicals do not refer as a function of context — they suffer from a character deficiency.

The traditional puzzles about belief reports puzzles rest on a certain seemingly innocuous assumption, that ‘that’‐clauses specify belief contents. The main theories of belief reports also rest on this “Specification Assumption”, that for a belief report of the form ‘A believes that p’ to be true,’ the proposition that p must be among the things A believes. I use Kripke’s Paderewski case to call the Specification Assumption into question. Giving up that assumption offers prospects for an intuitively more plausible approach to the semantics of belief reports. But this approach must confront a puzzle of its own: it turns out that every case is a Paderewski case, at least potentially.

Insensitive Semantics is mainly a protracted assault on semantic Contextualism, both moderate and radical. Cappelen and Lepore argue that Moderate Contextualism leads inevitably, like marijuana to heroin or masturbation to blindness, to Radical Contextualism and in turn that Radical Contextualism is misguided. Assuming that the only alternative to Contextualism is their Semantic Minimalism, they think they’ve given an indirect argument for it. But they overlook a third view, one that splits the difference between the other two. Like Contextualism it rejects Propositionalism, the conservative dogma that every indexicalfree declarative sentence expresses a proposition. Unlike Contextualism, it does not invoke context to fill semantic gaps and, indeed, denies that filling those gaps is a semantic matter. In rejecting Propositionalism, it is more radical, indeed, more minimalist than Cappelen and Lepore’s brand of Semantic Minimalism. It does not imagine that sentences that intuitively seem not to express propositions at least express “minimal propositions.” Radical Semantic Minimalism, or simply Radicalism, says that the sentences in question are semantically incomplete – their semantic contents are not propositions but merely “propositional radicals.”.

It is now widely recognized that there is a middle ground between being literal and fully explicit in meaning something and merely implicating it. In this case what the speaker means is, unlike implicature, an enrichment of the semantic content of the uttered sentence. The hearer needs to recognize that this semantic content includes only part of what the speaker could mean, either because it falls short of comprising a proposition or because the proposition it does comprise is not specific enough to be what the speaker could plausibly be supposed to mean. The speaker intends to communicate, depending on the case, either a completion or an expansion of the semantic content. Depraetere & Salkie propose a novel conception of this distinction, on which disambiguation counts as a case of completion and resolving semantic underspecification counts as expansion (free pragmatic enrichment). Although their approach makes sense from the standpoint of the hearer’s task of figuring out what a speaker means, I will suggest that the completion/expansion distinction as well as several important subordinate distinctions should be understood more abstractly, as different relations between what the speaker means and the semantic content of the uttered sentence.