Definition
Semantics is the branch of inquiry concerned with meaning — what signs, words, and sentences refer to, and how they do so. It asks the deceptively simple question: how does a mark on paper or a sound in the air come to be about something in the world? That question turns out to be one of the deepest in philosophy, linguistics, and computer science, with ramifications that reach into artificial intelligence, logic, and cognitive science.
At the most basic level, semantics distinguishes itself from syntax and pragmatics. Syntax governs the formal structure of expressions — the rules that determine whether a sentence is grammatically well-formed. Semantics governs what well-formed expressions mean — the mapping from symbols to interpretations. Pragmatics governs how meaning shifts in context — how the same sentence can mean different things depending on who says it, to whom, and when. All three layers interact, but semantics occupies the middle position: it assumes that syntax has already delivered a structured expression and asks what that structure represents.
Formal semantics, developed in logic and mathematics, specifies meaning with mathematical precision through model theory and truth conditions. Compositional semantics — the idea that the meaning of a complex expression is a function of the meanings of its parts and how they are combined — gives formal semantics much of its power, and underlies the architecture of programming languages and type systems. Natural-language semantics is more complicated, because ordinary language is context-sensitive, vague, and metaphorically flexible in ways that resist full formalization.
Why it matters
How it works
Reference, sense, and truth conditions
The most influential early framework in formal semantics distinguishes reference (what a term picks out in the world) from sense (the mode of presentation by which it picks that thing out). Two expressions can share a reference while differing in sense: "the morning star" and "the evening star" both refer to Venus, yet they carry different descriptive content. This distinction has practical consequences in contexts where the route to the referent matters — in legal definitions, scientific reductions, and database lookups, for instance.
Truth-conditional semantics takes a different route: instead of asking what words refer to, it asks under what conditions a sentence is true. To know the meaning of a sentence is to know what the world would have to be like for it to be true. This approach integrates naturally with model theory and has become the standard framework in formal logic and the foundation of most programming-language semantics.
Lexical and compositional meaning
Lexical semantics studies the meanings of individual words — how they are structured internally (prototypes, necessary-and-sufficient conditions, family resemblances), how they relate to other words (synonymy, antonymy, hyponymy), and how they shift across contexts. Compositional semantics studies how word meanings combine. The core claim is that sentences are not just bags of words; their structure — subject, predicate, modifier — determines how individual meanings contribute to the whole. This principle is powerful enough to generate the meaning of sentences never encountered before, which is why both humans and well-designed formal languages can be understood productively.
Where it goes next
Semantics sits at the intersection of language, mind, and computation. Its formal wing merges with logic and programming-language theory; its empirical wing merges with cognitive linguistics and psycholinguistics. Any system that processes symbols — from a database engine to a large language model — is implicitly implementing a theory of semantics, whether or not its designers acknowledge it.