Can formal semantics really explain meaning compositionally, or do intonation, context, and discourse do more of the work than we admit?
My links: https://linktr.ee/frictionphilosophy.
1. Guest
Barbara Partee is Distinguished University Professor Emerita of Linguistics and Philosophy at the University of Massachusetts Amherst, and her work has focused on formal semantics and language.
2. Interview Summary
Barbara Partee frames the interview around how to make the intuitive slogan of ‘compositionality’ precise: yes, sentence meaning depends on the meanings of parts and how they’re combined, but every term in that slogan (“meaning,” “part,” “combined,” “structure”) is theory-laden, so different formalizations yield different predictions. She illustrates how semantic considerations can adjudicate between syntactic analyses, using relative clauses: if “boy” and “who loves Mary” are both treated as predicates, composition can be simple set intersection—yielding “boy ∩ loves-Mary”—whereas building the definite “the boy” too early wrongly hardwires uniqueness and then can’t be “undone.” Along the way she credits insights from Richard Montague and notes parallels with W. V. O. Quine (including Quine’s Word and Object) about treating common nouns as predicates rather than names.
A major through-line is that many apparent counterexamples to ‘compositionality’ are really prompts to enrich what counts as “structure.” Partee points to intonation and focus as meaning-relevant (and often truth-condition- or presupposition-relevant): shifting stress in “John didn’t invite Mary to the party” changes what’s being denied, and similar effects show up with contrastive stress and “just anyone.” She also treats some constructions as having built-in “slots” that syntax doesn’t resolve, like genitives: “John’s team” contains an implicit relation variable (team John plays for/owns/writes about, etc.) that must be fixed pragmatically by context and shared expectations. This sets up her broader picture of semantics working in tandem with pragmatics and discourse—e.g., modeling how conversation updates a common ground, referencing work associated with Hans Kamp and Irene Heim on discourse and ‘context change’ perspectives.
In the latter part of the conversation she zooms out to what formal semantics has clarified over the decades: once you allow intermediate “open” structures (roughly, lambda-abstracts) rather than only closed sentences, quantification and scope phenomena stop looking like hacks and start looking systematic. She then uses anaphora puzzles—‘pronouns of laziness,’ donkey sentences, and the ‘paycheck’ pattern (traceable in the literature to Peter Geach and his Reference and Generality)—to show how natural language can require richer machinery than simple bound-variable models. Finally, she emphasizes both the strengths and limits of formal semantics: it excels at logical/structural effects (the kind that even large-scale statistical systems can miss, as in negation), but it can’t ignore lexical meaning entirely—illustrated by mass/count noun distinctions, cross-linguistic classifier strategies, and a lattice-based way of modeling why plurals and mass nouns pattern together.
3. Interview Chapters
00:00 - Introduction
00:53 - Compositionality
08:11 - Idealization
13:50 - Relevance of idealizations
19:02 - Non-syntactic expression
24:43 - Genetives
28:41 - Pronouns of laziness
33:45 - Pragmatic pronouns
36:31 - Pragmatics
40:32 - Anaphoric and deictic
42:25 - Other sorts of pronouns
43:34 - Takeaways
47:35 - Progress in linguistics
59:14 - Time and tense
1:06:24 - Meaning
1:14:20 - Many and few
1:24:26 - Difficult puzzles
1:31:18 - Modal terms
1:33:30 - Value of linguistics
1:41:12 - Conclusion
