From Convention to Context
David Lewis and the Philosophy of Language
In a previous article (link below), I introduced David Lewis's theory of scorekeeping in conversation, exploring how context evolves dynamically as interlocutors exchange utterances. In this article, I go a step back to an earlier and deeper foundation of Lewis's philosophy of language: his 1969 book Convention: A Philosophical Study. Here, Lewis presents an influential account of language as a social phenomenon, governed not by explicit rules, but by shared conventions.
Lewis's work had a profound impact not only on philosophy but also on linguistics, pragmatics, and the study of meaning. However, like all powerful theories, it has limitations. In what follows, I explain Lewis's model, critically examine its assumptions, and prepare the ground for a forthcoming article which will discuss Robert Stalnaker's refinements.
Language as Convention
Lewis's central claim is simple but far-reaching: language is a system of conventions that allows communication between individuals. A convention for Lewis is a regularity in behaviour that solves a recurrent coordination problem. Crucially, it persists because all parties expect others to conform, and prefer to conform themselves on that expectation.
In the case of language, conventions link sounds (or other symbols) to meanings. For instance, English speakers conventionally use "dog" to refer to a particular category of animal. This association is not natural or necessary; it is the product of a social regularity, maintained by mutual expectations.
This idea might appear similar to Ferdinand de Saussure's earlier notion of the arbitrariness of the linguistic sign, in which a signifier (the sound or form) bears no natural connection to the signified (the concept). Saussure's insight focused on the structure of language as a system of arbitrary signs. Lewis accepts this arbitrariness, but he goes further: he asks how arbitrary signs can function reliably in communication. His answer is that they do so through socially maintained conventions—regularities of use shaped by mutual expectations. Saussure's theory is descriptive; Lewis's is explanatory.
Lewis builds on earlier ideas from game theory and decision theory. He formalises communication as a kind of coordination game: speakers and hearers coordinate their behaviour around shared linguistic signs. Success depends on a background of common knowledge and mutual expectations.
Importantly, Lewis does not view conventions as being explicitly agreed upon. They evolve organically within communities, often without participants being able to articulate the rules they are following.
Strengths of Lewis's Model
Lewis's theory provides a powerful explanation for several fundamental features of language:
Flexibility and Change: Conventions can shift over time, explaining how languages evolve.
Social Embeddedness: Meaning is not fixed by private mental representations, but by shared social practices.
Normativity: Language use carries normative force. Using words "correctly" means adhering to community conventions.
These insights have been foundational for fields such as pragmatics, sociolinguistics, and philosophy of language. By treating language as a coordination mechanism, Lewis opened new ways to think about how meaning depends on use.
Critical Perspectives
Despite its elegance, Lewis's model faces important criticisms, both internal and external.
Assumptions About Rationality
Lewis's account presumes that language users are rational agents who can coordinate based on mutual expectations. However, real speakers often act irrationally, inconsistently, or without full awareness of linguistic norms.
Children acquiring language, for instance, do not coordinate meanings through conscious expectations about others' behaviour. Nor do many everyday speakers reflect on how their utterances fit into a broader system of mutual knowledge. The psychological realism of Lewis's rational actor model is, therefore, questionable.
Theoretical Critiques
Philosophers such as Ruth Millikan have challenged the conventionalist view by proposing biological accounts of language. Millikan argues that linguistic meaning arises from evolutionary functions rather than from social conventions per se. In her view, words are tools shaped by success in communication over time, not primarily by mutual expectations.
Stephen Levinson, from a linguistic and pragmatic perspective, points out that communication often succeeds despite profound contextual ambiguity and incomplete coordination. He suggests that speakers rely heavily on inference, context, and pragmatic reasoning, rather than strict adherence to pre-existing conventions.
Thus, while conventions are part of language, they may not be as central or as stable as Lewis's theory suggests.
Linguistic Reality
Empirical studies of language variation and change reveal that linguistic conventions are rarely uniform across communities. Dialects, sociolects, idiolects, and situational variation complicate any simple picture of "shared" conventions.
For example, the pronunciation of "either" and "neither" varies widely among English speakers, yet communication typically succeeds. Conventions are thus often flexible, layered, and incomplete. Lewis's formal model struggles to capture this messiness without additional mechanisms.
Moreover, language change often occurs below the level of conscious coordination. Innovations spread through imitation, accommodation, or unconscious bias rather than through rational strategic alignment.
Towards Context and Stalnaker
David Lewis's theory of language as convention remains a cornerstone of the philosophy of language. It highlights the fundamentally social nature of meaning and provides a rigorous framework for thinking about communication as coordination.
Yet its limitations are equally instructive. Language is less stable, less conscious, and less uniform than Lewis's rationalist model assumes. Later thinkers have had to grapple with these complexities.
One such thinker is Robert Stalnaker, who built on Lewis's insights but shifted the focus towards the role of context and common ground in communication. In the next article, I shall explore how Stalnaker refines the connection between language, belief, and context, moving beyond the idea of convention alone.
Related Reading:
Further Reading:
Lewis, D. (1969). Convention: A Philosophical Study. Harvard University Press.
Millikan, R. G. (1984). Language, Thought, and Other Biological Categories. MIT Press.
Levinson, S. C. (2000). Presumptive Meanings: The Theory of Generalized Conversational Implicature. MIT Press.