Models of second language competence: a structural equation approach

Le résumé fourni par la source

Starting with a critical discussion of the consensus reached concerning the structure of second language competence, the unitary competence hypothesis as advocated especially by Oller (1976) is then confronted with new evidence supporting a multidimensional model of foreign language ability. In order to question the plausibility of the general (unitary) factor hypothesis, its central claim, that of universal validity, is put to the test. This claim implies at least two assumptions: a) the invariance of the general factor, regardless of differences in the populations studied, and b) the invariance of this factor irrespective of the conditions under which language acquisition takes place. By showing divergent structures in subgroups of high and low first language ability, the first assumption would be falsified. By demonstrating specific influences of two teaching styles on different language components, the second assumption could be disproved. The analyses are based on data from a sample of 14 000 seventh grade students and their teachers from 427 secondary schools ('Gymnasien') being highly representative for West Germany and West Berlin. The hypotheses were tested using confirmatory factor analysis (LISREL). The results obtained show the appropriateness of the three factor model proposed (ELEMENTARY, COMPLEX, COMMUNICATIVE), which is theoretically meaningful and best fits the data. Moreover, the expected effects of first language ability grouping and of teaching strategies on the structure of second language competence clearly emerge. In spite of the limitations of the data set analysed, these results can all be interpreted as strong and further evidence against a single factor model (and its educational implications) and in favour of a multiple factor model as outlined above. Finally, didactical consequences of these findings are discussed.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.