A topic in the Open Knowledge Graph — a free, open map of 15,290 topics and the order to learn them in.
While compositionality is a guiding principle—a complex expression's meaning is determined by its parts and structure—it faces challenges from phenomena like idioms ("kick the bucket"), scope ambiguities, and context-shift cases. Understanding where and why compositionality succeeds and fails refines our conception of how meaning is constructed and transmitted.
Start with clear compositional cases (arithmetic expressions, simple predicates) then move to failures: "pull someone's leg" doesn't mean making leg-pulling motions. Study how the same structure can yield different meanings in different contexts ("That's a bright idea" versus "That's a bright student"). Examine whether compositionality is semantic law or pragmatic principle, and whether partial compositionality is defensible.
You've mastered the compositionality principle: the meaning of a complex expression is a function of the meanings of its parts and their mode of combination. The power of this principle is that it explains how finite minds understand an infinite number of sentences — you don't memorize meanings whole, you compute them. But the principle faces stress tests that reveal both its scope and its limits. Understanding where compositionality holds and where it strains illuminates the deeper architecture of how meaning is constructed.
The clearest case of compositional success is logical and mathematical language. "Three plus four equals seven" means what it does because "three," "four," "plus," and "equals seven" each make determinate contributions, and the syntactic structure specifies how those contributions combine. From your study of first-order logic syntax, you recognize that the same compositional story applies to natural language predicates: "the cat slept" means what it does because "the cat" picks out a referent and "slept" applies a predicate, combined by standard predication. The syntactic structure is the composition rule; the semantic values of the parts are the inputs; the output is the sentence's meaning.
Idioms are the canonical challenge. "Kick the bucket" does not mean what a compositional analysis would predict from the literal meanings of "kick" and "bucket." The meaning (to die) seems to be a stored, non-compositional whole. But careful analysis complicates this picture. Idioms may be compositional at a metaphorical level — the image of a bucket being kicked encodes a conventional metaphorical mapping that the community has fixed. Even if the idiomatic meaning cannot be computed from *literal* meanings of the parts, it may still be derived compositionally from *conventionalized metaphorical* meanings that are themselves stored. Whether this counts as "real" compositionality is a substantive question, but it shows that apparent failures often mask more subtle compositional structure.
Context-shifting poses a deeper challenge. Consider "I," "here," "now" — these pick out different individuals, places, and times depending on context. The standard response is that compositionality operates over character (a function from context to content, in Kaplan's framework) rather than content directly. The semantic value of "I" is not an individual but a rule: "in any context, pick the speaker." Composition applies to characters; context then yields content. This extension handles many apparent failures while revealing that the compositionality principle operates at a level of abstraction richer than surface form. Persistent cases that resist even this treatment — donkey anaphora ("Every farmer who owns a donkey beats it"), ellipsis, and nominal coercion — remain active research areas where the genuine limits of compositionality are tested. These cases are not failures of the principle so much as reminders that the syntax-semantics interface is more complex than any simple formulation of compositionality captures.
Topics in reflective domains aren't scored by quiz answers. Read, reflect, and mark when you've thought it through.