Context-dependent interpretation addresses the systematic gap between what a sentence literally means (sentence meaning) and what a speaker communicates by uttering it (speaker meaning). Pragmatic enrichment fills in underspecified elements — "I've eaten" is understood as "I've eaten today/recently," not "at some point in my life." Narrowing restricts a word's denotation contextually — "drink" at a party means alcoholic beverages, not all liquids. Loose use extends meaning beyond its literal boundaries — "Holland is flat" communicates an approximation that is useful despite being technically false for every square meter. These processes are rapid, automatic, and essential: virtually no utterance in natural conversation communicates exactly and only its compositional semantic content.
Collect everyday utterances and identify the gap between what was literally said and what was communicated — "The ham sandwich wants his check" in a restaurant context is a classic example. Practice distinguishing enrichment, narrowing, and loose use on the same word in different contexts (e.g., "open" in "open the door" vs. "open the wine" vs. "open person"). Compare Gricean and Relevance Theory accounts of how these processes work.
From your work in linguistic pragmatics, you know that what a sentence literally encodes and what a speaker communicates are regularly different — Grice's cooperative principle governs the inferential gap. From compositional semantics, you know that sentence meaning is built systematically from word meanings and syntactic structure. Context-dependent interpretation is the territory between these two: the mechanisms by which sentence meaning gets transformed into the richer, more specific communication that speakers actually intend and listeners actually receive.
Pragmatic enrichment is the most pervasive mechanism. Virtually every utterance contains underspecified elements that must be resolved by context to determine what was actually communicated. "She's ready" is semantically incomplete — ready for what? "I've had breakfast" communicates "today," not "at some point in my life," even though the tense is perfective, not bounded. "It's raining" is implicitly "here, now." The semantic content of these sentences is truth-conditionally complete, but the communicated proposition is more specific. Enrichment fills in the gap between the semantic skeleton and the full proposition the speaker intends. Crucially, the enriched proposition — not the semantic skeleton — is what the speaker is asserting and what the listener evaluates for truth.
Narrowing restricts a word's denotation to a contextually appropriate subset. The word "drink" denotes all liquids, but at a cocktail party "would you like a drink?" means an alcoholic beverage. "She ate the whole box" typically means the entire contents, not the cardboard packaging. "Cutting" in a kitchen context means with a knife; "cutting" in a conversation about surgery means with a scalpel. The lexicon does not store a separate entry for each contextual restriction — language would be unusable if it did. Instead, a single lexical entry is deployed with a narrowed range that context makes salient. This efficiency is not imprecision; it is the designed use of underspecification.
Loose use works in the opposite direction, extending meaning beyond its literal boundaries while communicating something approximately or schematically true. "Holland is flat" is technically false for every square meter of the Netherlands, but it communicates useful information about the region's topography relative to mountainous alternatives. "The ATM is on the corner" may be twenty meters from the actual corner. These are not lies or mistakes — they are calibrated approximations where the speaker signals that exact truth is not the goal, approximate truth is. Relevance Theory explains loose use through scales of approximation: the speaker communicates the loosest interpretation that is informative enough for the hearer's purposes.
Together, enrichment, narrowing, and loose use explain why natural language communication is so efficient and yet so reliable. A finite lexicon and a relatively sparse grammar generate infinitely varied, contextually precise communications because each utterance comes loaded with assumptions about the situation, the speaker's goals, and what the listener already knows. The core skill in pragmatics is learning to see these processes as systematic and analyzable — not as noise in the signal, but as the signal itself.