Pragmatics | behavior.engineering

Pragmatics is the study of what language actually means in context, as opposed to what it literally says. When someone asks “can you pass the salt?” they don’t want a yes-or-no answer — they want the salt. Human communication is full of these pragmatic leaps: implied meaning, conversational implicature, shared assumptions that go unsaid. For AI models, pragmatics is one of the hardest aspects of language to get right — understanding not just what a user literally wrote, but what they actually mean, need, and expect given the conversational context. For behavior architects, pragmatics is a useful lens for investigating why models sometimes produce technically correct but contextually inappropriate responses: the model understood the words but not the intent behind them.

SemanticsThe study of meaning in language — what words, phrases, and sentences refer to and signify.
Linguistic AnalysisThe systematic study of language features in model outputs — such as vocabulary, syntax, tone, and discourse structure — to understand behavioral patterns.
Instruction AmbiguityThe quality of an instruction or prompt that allows for multiple reasonable interpretations, potentially leading to inconsistent or incorrect model responses.
Discourse StructureThe organization and coherence of language across multiple sentences or turns — how ideas connect, flow, and build on each other in extended text or conversation.
Model JudgmentA model's ability to reason through ambiguous or novel situations and arrive at contextually appropriate decisions without explicit instructions for every case.