Loading field map
How does context shape meaning beyond the literal content of words and sentences? This question drives philosophical pragmatics, a subfield that emerged from the recognition that truth-conditional semantics alone cannot account for how speakers actually communicate. The central tension runs between two broad approaches: one that treats pragmatic meaning as a matter of rational, cooperative inference, and another that sees it as the output of a dedicated cognitive module. Six frameworks, developed from 1953 to the present, have defined the debate.
The starting point for philosophical pragmatics is Ludwig Wittgenstein's later philosophy, especially the Philosophical Investigations (1953). Wittgenstein argued that the meaning of a word is not a fixed object or truth condition but is constituted by its role in concrete social activities—what he called "language-games." This was a direct challenge to the dominant tradition, from Frege through the early Wittgenstein and the logical positivists, which treated meaning as a matter of reference and truth. By insisting that "meaning is use," Wittgenstein provided a philosophical foundation for treating context and practice as central to meaning. However, his account remained deliberately unsystematic: it offered a picture of how language works rather than a technical model of pragmatic inference. Later frameworks would take up the task of building precise theories on the ground Wittgenstein cleared.
J. L. Austin and John Searle transformed Wittgenstein's insight into a systematic theory of language as action. In How to Do Things with Words (1962), Austin distinguished between the locutionary act (saying something with a certain meaning), the illocutionary act (what one does in saying it, such as promising or ordering), and the perlocutionary act (the effect on the hearer). Speech Act Theory shared Wittgenstein's focus on use, but it introduced a crucial new element: conventional rules. For Austin and Searle, the force of an utterance—whether it counts as a promise, a warning, or a request—is determined by social conventions and felicity conditions, not by the speaker's rational calculations. This emphasis on convention set Speech Act Theory apart from the inferential model that would soon follow.
H. P. Grice's William James lectures (published in 1975 as "Logic and Conversation") introduced a radically different picture of how pragmatic meaning works. Grice argued that speakers and hearers are engaged in a cooperative enterprise governed by a Cooperative Principle and a set of maxims (quantity, quality, relation, manner). When a speaker appears to violate a maxim, the hearer infers an implicature—a meaning that goes beyond what is literally said. For example, if A asks "Is there any coffee?" and B replies "There's a Starbucks on the corner," the implicature is that B does not have coffee but is suggesting where to get it.
Grice's model contrasted sharply with Speech Act Theory. Where Austin and Searle appealed to conventional rules, Grice appealed to rational inference: hearers work out implicatures by reasoning about the speaker's intentions and the cooperative context. This made Gricean pragmatics a theory of communication rather than a theory of social acts. Grice also drew a sharp boundary between semantics (what is said, determined by truth conditions) and pragmatics (what is implicated, determined by inference). That boundary would become a major site of debate.
David Kaplan's work on demonstratives and indexicals ("Dthat," 1978; the monograph Demonstratives, 1977) introduced a formal treatment of context-sensitivity that complicated the semantics-pragmatics boundary. Kaplan distinguished between character (a rule that maps contexts to content) and content (the proposition expressed in a context). For indexicals like "I," "here," and "now," the character is a constant rule (e.g., "I" refers to the speaker), but the content varies with context. This showed that some context-dependence is built into the semantic system itself, not added by pragmatic inference.
Context and Indexicality thus occupied a middle ground. It agreed with Grice that there is a level of literal, truth-conditional content (what is said), but it argued that this content is already shaped by context in ways that Grice's framework did not fully capture. The framework did not replace Gricean pragmatics; rather, it forced a refinement of the semantics-pragmatics boundary. Later theorists would ask: if indexicality is semantic, how much more context-sensitivity might also be semantic?
Neo-Gricean pragmatics, developed by Laurence Horn, Stephen Levinson, and others, aimed to make Grice's system more precise and economical. Horn reduced the maxims to two principles: the Q-Principle (say enough) and the R-Principle (say no more than necessary). Levinson added a three-tiered system of heuristics (Q, I, M) to account for generalized conversational implicatures—inferences that arise by default, without special contextual cues. For example, the scalar implicature from "some" to "not all" is a generalized implicature: it occurs unless the context blocks it.
Neo-Gricean pragmatics preserved Grice's core commitment to rational, cooperative inference, but it narrowed the focus by treating generalized implicatures as semi-automatic defaults. This brought the framework closer to the kind of systematic, predictive model that linguists and cognitive scientists wanted. However, it also created a tension: if implicatures are defaults, are they still inferences in Grice's sense, or are they something closer to semantic rules? That question would be taken up by the next framework.
Dan Sperber and Deirdre Wilson's Relevance: Communication and Cognition (1986) proposed a radical alternative to the entire Gricean tradition. Relevance Theory replaces the Cooperative Principle and maxims with a single cognitive principle: human cognition is geared to the maximization of relevance. In communication, hearers assume that an utterance is optimally relevant—that it yields enough cognitive effects for the processing effort required. Pragmatic inference is not a matter of reasoning about cooperation but of applying a domain-specific, modular cognitive mechanism.
Relevance Theory differs from both Gricean and Neo-Gricean pragmatics in several key ways. First, it rejects the idea that implicatures are derived from maxims; instead, all pragmatic interpretation (including the recovery of explicit content, which Sperber and Wilson call "explicature") is driven by the search for relevance. Second, it denies that there is a sharp boundary between semantics and pragmatics: even the proposition expressed by an utterance is underdetermined by its linguistic meaning and must be enriched pragmatically. Third, it treats pragmatic processing as automatic and modular, not as conscious reasoning.
This framework transformed the debate by shifting the explanatory burden from rational cooperation to cognitive architecture. It also absorbed some of the insights of Context and Indexicality: the underdetermination thesis generalizes Kaplan's point that context shapes content, but it extends the phenomenon far beyond indexicals to every aspect of utterance interpretation.
Today, philosophical pragmatics is a field of live disagreement. The leading frameworks—Gricean/Neo-Gricean pragmatics and Relevance Theory—agree on several fundamental points: that linguistic meaning underdetermines speaker meaning, that context plays an essential role in interpretation, and that pragmatic inference is central to communication. They disagree, however, on the architecture of that inference.
Gricean and Neo-Gricean approaches maintain that pragmatic inference is a form of rational, cooperative reasoning, governed by principles that speakers and hearers consciously or tacitly follow. Relevance Theory argues that this picture is too intellectualist and that pragmatic processing is better explained by a dedicated cognitive module operating below the level of conscious reasoning. The debate has empirical consequences: Relevance Theory predicts that pragmatic interpretation is fast, automatic, and domain-specific, while Neo-Gricean models predict that it is sensitive to explicit reasoning about cooperation.
Speech Act Theory and Wittgensteinian Language-Games continue to play important roles, but more as philosophical backdrop and specialized tools than as active research programs in pragmatics proper. Speech Act Theory remains influential in legal philosophy, ethics, and social ontology, where the conventional force of utterances matters. Wittgenstein's language-game concept is a recurring reference point for philosophers who resist systematic theorizing about meaning.
Context and Indexicality has been largely absorbed into the broader debate about the semantics-pragmatics boundary. The question Kaplan raised—how much context-sensitivity is semantic?—is now pursued in the philosophy of language under headings like "contextualism" and "semantic minimalism," which are in direct dialogue with Relevance Theory's underdetermination thesis.
The field's central tension thus remains unresolved. Is pragmatic inference a special case of general rational cooperation, or is it a distinct cognitive module? The answer will shape not only how we understand communication but also how we model the mind.