Formulaic Language in Computer-supported Communication: Theory Meets Reality

A recent model of language processing in normal native speakers (Wray, 2002a) proposes that speakers reap substantial benefits from storing and retrieving prefabricated utterances from memory, rather than always constructing novel ones on line. Substantial evidence from a wide range of linguistic research is consistent with the model, but independent tests of its viability would be difficult to conduct. A small number of ‘naturalexperiments’, however, provide anopportunity to gain insights into some of the parametersof themodel. One such ‘natural’ experiment is TALK, a system developed topromote conversational fluency innon-speaking individuals. The design and efficacy of TALK are explored and evaluated as a potential working model of Wray’s depiction of normal language processing. TALK is demonstrated to be a valuable tool for the pursuit of language awareness, both because it demands of its users a highly developed sensitivity about how conversation works, and because it provides researcherswith glimpses of a phenomenon that is normally inaccessible – language processing in action.