Abstract
This paper examines the way in which, three different paradigms for the study of interlanguage handle the phenomenon of variability in interlanguage systems: a Chomskyian paradigm proposed by Adjemian (1976, 1981); the Monitor Theory proposed by Krashen (1976, 1981) and the Continuum paradigm proposed by Tarone (1979, 1982). The paper presents data from several studies showing that interlanguage speech production varies systematically with elicitation task; it compares the fundamental assumptions of each of these three paradigms with regard to their views of the nature of the system which underlies learner utterances and of the methodology appropriate to the study of this system; and it concludes that the Continuum paradigm accounts for the data better than the other two paradigms, because of its underlying assumptions.

This publication has 0 references indexed in Scilit: