📄 Book Reviews: Meaning and Speech Acts Volume I: Principles of Language Use Volume II: Formal Semantics of Success and Satisfaction

Summary

This is Amichai Kronfeld's 1993 review in Computational Linguistics of Daniel Vanderveken's two-volume Meaning and Speech Acts. The book builds on Searle and Vanderveken (1985) and tries to give speech act theory a full logical treatment. It places illocutionary acts, rather than sentences, at the centre of meaning, and it joins truth-conditional semantics with theories of when an act succeeds and when it is satisfied.

Volume I sets out the conceptual foundations. Volume II presents a first-order illocutionary logic, an ideal object language, an intensional model theory and a complete axiomatization, plus translation rules from English performative verbs into that logic.

Kronfeld praises the formal achievement. He disputes the philosophical thesis that thought is fundamentally built out of illocutionary acts: thoughts lack success conditions, and beliefs need individuating representations, whereas speech acts can carry singular propositions. He sees real computational promise in the work, especially for plan-based models of speech acts and for fine-grained analysis of performatives. He names extending the English-to-logic translation rules as the main open task.

Key points

• Every illocutionary force is broken into six components:
  • illocutionary point,
  • mode of achievement,
  • propositional content conditions,
  • preparatory conditions,
  • sincerity conditions,
  • degree of strength (for example, a request versus a command).
• An illocutionary act is modelled as a force applied to a propositional content. It is evaluated along two axes: success (was the act performed?) and satisfaction (did the world come to fit?). Truth is only one case of satisfaction.
• Volume I, Chapter 6 analyses many English performatives and arranges them in derivation trees. For example, among directives, suggest leads to recommend via advice, and to alarm via warn, caution and alert. The reviewer notes the individual definitions are close to dictionary entries.
• Volume II, Chapter 7 gives rules that map the syntax of English performative sentences onto illocutionary logical forms. Its appendices, partly with David K. Johnston and François Lepage, prove completeness results.
• The reviewer's main critique is that the semantics of speech acts admits singular propositions (reference without an individuating representation), while the semantics of thought cannot (citing Kronfeld 1990).

Interesting ideas

• I1. Break each speech-act type into a fixed set of separate components instead of using a single opaque label, so that related acts differ in just one or two slots, such as strength or preparatory conditions.
• I2. Judge an utterance on two independent axes, whether the act was successfully performed and whether its content was later satisfied, rather than only true or false.
• I3. Arrange performative verbs in derivation trees, where each verb inherits from a more basic act and adds conditions, as in suggest → advise → recommend.
• I4. Treat the meaning of a word as its contribution to possible speech acts in context; this gives words like alas and hurray a formal role instead of leaving them unanalysed.
• I5. Model communicative planning as a derivation from axioms describing the current world toward theorems describing the world where the goal holds.

For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

• I1 (frames): crate prag could store each speech act as a six-slot record, so the difference between request and command becomes an explainable change in the strength slot.
• I2 (world): the world of the text could mark a promise or command as performed at one event and fulfilled or broken at a later one, which supports questions like "did the fox keep its word?"
• I3 (world): this hierarchy could sit in the compiled ontology beside VerbNet classes, letting the system generalize from known parent acts to rarer child verbs.
• I4 (frames): interjections and expressive particles could be mapped to expressive acts with sincerity conditions (speaker regret or joy), feeding irony and implicature detection.
• I5 (answer): story readers could explain a character's request or warning by reconstructing the goal it serves, giving a step-by-step justification for "why did X say that?" questions.