Language and Logic advanced course:
How to do things with words: Speech Acts in Linguistics, Philosophy and Computation.

  • Regine Eckardt ()
  • Magdalena Schwager ()

External page


Speech acts have been discovered in philosophy as acts that can be done with words or moves in a language game. They still pose a challenge to semantic theory where, typically, declarative sentences are analysed as denoting propositions. The two views of sentences as containers of information, and as moves in social interaction, can not easily be reconciled.

Grammar and speech act level are not completely disconnected. Practically all languages distinguish between three sentence modes, declarative, interrogative, imperative, which are prototypically tied to the respective speech acts. But many more aspects of grammar refer to language's function in social interaction: The use of subjunctive can serve to distinguish reported speech acts from executed speech acts. Particles, discourse adverbs and NPIs can serve as speech act markers. Examples abound, consider German wohl which turns questions into self-directed questions: Wer bringt Kaffee mit? "who will bring coffee?" (real question) Wer bringt wohl Kaffee mit? "who will wohl bring coffee?" (speaker asks herself) Doch/denn distinguish verb-initial exclamatives from yes/no questions: Hat er denn einen Hund mitgebracht? "has he brought along a dog?" Hat er doch einen Hund mitgebracht! "And what he did was bring a DOG along!" Intensifiers like English so can be used to emphasize properties ("He is so tall.") but arguably (Potts) also (preconditions of) speech acts ("He is so next in line.").

Historical ties Research on grammaticalization proves that indirect directive speech acts grammaticalize into grammarical markers of imperative sentence mood. Strikingly, languages exhibit exactly those grammaticalization paths that would be predicted by Searle's theory of indirect speech acts. While this parallel is satisfactory from a theoretical perspective, its consequences are largely unexplored. Are similar grammaticalization patterns known for other indirect speech acts? Other sentence modes? Why is it that grammars limit themselves to three sentence modes while major speech act classifications rest on at least five-six major groups, which subdivide into many more specific kinds of acts?

Analysis of speech acts Formal theories of speech acts exist (e.g. SDRT, Asher/Lascarides, but also philosophical approaches ranging from Stalnaker/Lewis to Searle/Vanderveken), but are usually phrased as orthogonal to, and disconnected from truth conditional semantics of the clause. However, there are many words that have a meaning both in semantics and at speech act level; consider quantification into speech acts, cookie conditionals ("if you're hungry, there are cookies on the table") and more. Can such cross-domain phenomena be captured in a way that makes the ties between the two levels maximally transparent? Related to this is the question whether there is any way to delimit the number of speech acts, and to come up with clear criteria to identify and distinguish them.

Speech acts and automatic language processing

Finally, applications in language processing are often based on a scenario of speaker-machine interaction which involves more than a simple exchange of propositional content. Such applications often rest on simple, task-driven models of "doing things with words". Is there anything that formal semanticists can learn from such theories that have been tested against real communication? Can semanticists contribute something to this type of models?