ABSTRACT
Pity the poor analyst, who has to do the best he can with meanings that are as elusive as a piece of wet soap in a bathtub.
(Dwight Bolinger Aspects o f Language 1975:205)
We are in a hall, waiting for someone to speak. He goes up to the microphone, opens his mouth to say something, but before any sound comes out collapses from a heart attack. We may presume that at the moment he collapsed, the would-be speaker had formulated some language expression which he was about to communicate to his audience: but it died with him. The point of recounting this sad tale, is to demonstrate that no matter what language the speaker may have in his head, it is of no consequence to anyone else - including the linguist - until he has uttered it. Thus, the source for linguistic data is the speech act: where a speaker S makes an utterance U in language L to hearer H in context C. This is not to deny that a whole range of language expressions which could be uttered never are;1 but these are only interesting because they could potentially be uttered; and they are only recognizable to someone other than the person who thinks them up, when they a r e uttered: after all, linguists deal in language, not telepathy.
