摘要:Labov (1972a, 1972b) criticized Chomsky's (1965) notions that variation is an "error" and that speech communities are homogeneous. He maintained that variation is not random but rather systematic and rule-governed, and that in order to describe language used by members of a speech community it is necessary to study the vernacular, which provides the most "systematic data". He further maintained that "the only way to obtain sufficient good data on the speech of anyone person is through an individual, tape-recorded interview" (1972b: 181). Since then, the sociolinguistic interview has been the preferred source of data in sociolinguistic research (Milroy 1987; Coulmas 1997; Feagin 2001). Yet Labov also acknowledged that there are a number of problems associated with studying spontaneous speech, including the "rarity" of certain forms (l972a:188). Moreover, an enormous amount of work is involved in data collection, transcription and coding (Milroy 1987; Feagin 2001).In this paper I present results of a research project which employed two paper and-pencil tasks, a controlled-production task and an acceptability judgment task, to study auxiliary use and its correlates in Acadian French. These tasks made it possible to collect data from a large number of subjects while ensuring that the linguistic contexts were consistent across subjects; it was also possible to study linguistic phenomena which are rare or non-existent in oral speech.