This talk describes how we can use real-world corpus data and corpus linguistics tools to carry out research in pragmatics. In particular, I discuss the creation of a speech-act annotated business email corpus and its use for research into both the linguistic aspects of speech acts (lexicon, syntax, discourse), and into the development of an automated tool for speech act recognition, which currently achieves an average accuracy of 75%. Among the questions the talk addresses are:

• what problems do we encounter in trying to manually annotate a corpus for speech acts?
• what is the relationship between particular phrases and their pragmatic function?
• can computer programs perform speech act recognition reliably?
• how do we use speech acts in business communication?

Present tense verbal paradigms in British Isles dialects of English show a number of different variants, including systems which have -s for all persons — as in I loves, you loves, she loves — and the extraordinarily bizarre Standard English system which has zero marking except for the -s of the third-person singular - exactly what you would not expect. But the dialects of East Anglia — Norfolk and Suffolk, and including the urban dialects of Norwich and Ipswich — have, uniquely in the the British Isles, zero-marking for all persons.

I blame the Spanish Inquisition. If you'd like to know why, please come along and find out.