Paul Pietroski

Davidsonian analyses of action reports like ‘Alvin chased Theodore around a tree’ are often viewed as supporting the hypothesis that sentences of a human language H have truth conditions that can be specified by a Tarski-style theory of truth for H. But in my view, simple cases of adverbial modification add to the reasons for rejecting this hypothesis, even though Davidson rightly diagnosed many implications involving adverbs as cases of conjunct-reduction in the scope of an existential quantifier. I think the puzzles in this vicinity reflect “framing effects,” which reveal the implausibility of certain assumptions about how linguistic meaning is related to truth and logical form. We need to replace these assumptions with alternatives, instead of positing implausible values of event-variables or implausible relativizations of truth to linguistic descriptions of actual events

Speakers can use sentences to make assertions. Theorists who reflect on this truism often say that sentences have linguistic meanings, and that assertions have propositional contents. But how are meanings related to contents? Are meanings less dependent on the environment? Are contents more independent of language? These are large questions, which must be understood partly in terms of the phenomena that lead theorists to use words like ‘meaning’ and ‘content’, sometimes in nonstandard ways. Opportunities for terminological confusion thus abound when talking about the relations among semantics, pragmatics, and truth. As Stalnaker (2003) stresses, in Quinean fashion, it is hard to separate the task of evaluating hypotheses in these domains from the task of getting clear about what the hypotheses are. But after some stage-setting, I suggest that we combine Stalnaker’s (1970, 1978, 1984, 1999, 2003) externalist account of content with Chomsky’s (1965, 1977, 1993, 2000a) internalist conception of meaning.

Chomsky’s (1995, 2000a) Minimalist Program (MP) invites a perspective on semantics that is distinctive and attractive. In section one, I discuss a general idea that many theorists should find congenial: the spoken or signed languages that human children naturally acquire and use— henceforth, human languages—are biologically implemented procedures that generate expressions, whose meanings are recursively combinable instructions to build concepts that reflect a minimal interface between the Human Faculty of Language (HFL) and other cognitive systems. In sections two and three, I develop this picture in the spirit of MP, in part by asking how much of the standard Frege-Tarski apparatus is needed in order to provide adequate and illuminating descriptions of the “concept assembly instructions” that human languages can generate. I’ll suggest that we can make do with relatively little, by treating all phrasal meanings as instructions to assemble number-neutral concepts that are monadic and conjunctive. But the goal is not to legislate what counts as minimal in semantics. Rather, by pursuing one line of Minimalist thought, I hope to show how such thinking can be fruitful.

Some cases of implicit knowledge involve representations of (implicitly) known propositions, but this is not the only important type of implicit knowledge. Chomskian linguistics suggests another model of how humans can know more than is accessible to consciousness. Innate capacities to focus on a small range of possibilities, thereby ignoring many others, need not be grounded by inner representations of any possibilities ignored. This model may apply to many domains where human cognition “fills a gap” between stimuli and judgment.

A sentence like every circle is blue might be understood in terms of individuals and their properties or in terms of a relation between groups. Relatedly, theorists can specify the contents of universally quantified sentences in first-order or second-order terms. We offer new evidence that this logical first-order vs. second-order distinction corresponds to a psychologically robust individual vs. group distinction that has behavioral repercussions. Participants were shown displays of dots and asked to evaluate sentences with each, every, or all combined with a predicate. We find that participants are better at estimating how many things the predicate applied to after evaluating sentences in which universal quantification is indicated with every or all, as opposed to each. We argue that every and all are understood in second-order terms that encourage group representation, while each is understood in first-order terms that encourage individual representation. Since the sentences that participants evaluate are truth-conditionally equivalent, our results also bear on questions concerning how meanings are related to truth-conditions.

Davidson conjectured that suitably formulated Tarski-style theories of truth can “do duty” as theories of meaning for the spoken languages that humans naturally acquire. But this conjecture faces a pair of old objections that are, in my view, fatal when combined. Foster noted that given any theory of the sort Davidson envisioned, for a language L, there will be many equally true theories whose theorems pair endlessly many sentences of L with very different specifications of whether or not those sentences are true. And if L includes words ‘true’, then for reasons stressed by Tarski, it’s hard to see how any truth theory for L could be correct. Moreover, each of these concerns amplifies the other. Appealing to possible worlds will not help with Foster’s Problem, for reasons that Chomsky discussed in the 1950s, and appealing to trivalent models of truth will not avoid concerns illustrated with Liar Sentences.

Quantificational determiners are often said to be devices for expressing relations. For example, the meaning of _every_ is standardly described as the inclusion relation, with a sentence like _every frog is green_ meaning roughly that the green things include the frogs. Here, we consider an older, non-relational alternative: determiners are tools for creating restricted quantifiers. On this view, determiners specify how many elements of a restricted domain (e.g., the frogs) satisfy a given condition (e.g., being green). One important difference concerns how the determiner treats its two grammatical arguments. On the relational view, the arguments are on a logical par as independent terms that specify the two relata. But on the restricted view, the arguments play distinct logical roles: specifying the limited domain versus supplying an additional condition on domain entities. We present psycholinguistic evidence suggesting that the restricted view better describes what speakers know when they know the meaning of a determiner. In particular, we find that when asked to evaluate sentences of the form _every F is G_, participants mentally group the Fs but not the Gs. Moreover, participants forego representing the group defined by the intersection of F and G. This tells against the idea that speakers understand _every F is G_ as implying that the Fs bear relation (e.g., inclusion) to a second group.

In Conjoining Meanings, I argue that meanings are composable instructions for how to build concepts of a special kind. In this summary of the main line of argument, I stress that proposals about what linguistic meanings are should make room for the phenomenon of lexical polysemy. On my internalist proposal, a single lexical item can be used to access various concepts on different occasions of use. And if lexical items are often “conceptually equivocal” in this way, then some familiar arguments for externalist conceptions of linguistic meaning need to be reevaluated.