Fluidité, complexité et morphosyntaxe dans la production écrite en FLE

Sammanfattning: The aim of this study is to examine the development of linguistic proficiency in written production of second-language French. In a longitudinal study of 30 months, we follow the written production of 5 guided learners of French as a second language. In order to apprehend proficiency we use the conventional parameters fluency, complexity and accuracy. The fluency and complexity are apprehended by conventional indicators used in the literature, but the accuracy is apprehended in particular morphosyntactic contexts: finiteness, subject-verb agreement and past tense. This study also takes an interest in the use and development of explicit and implicit knowledge, which are closely related to the concept of proficiency. Written production is particularly suitable for this kind of study as it gives more time to control the output before and after the writing than oral production does. The learners? computer-written production is recorded by the means of the program ScriptLog and a video-filmed thinking aloud protocol. This methodology allows us to follow the written production in real time. The protocols from ScriptLog and the TAP provide the material for the analysis of the development of linguistic proficiency. In addition to the results indicating a non-controversial, anticipated general gain in fluency, complexity and accuracy, we observe important inter-individual differences between the 5 learners. One group of learners produces written French L2 with more fluency and less verbalisation, which indicates a lesser use of explicit knowledge. The other group produces with less fluency and more verbalisations, which indicates a more important use of explicit knowledge. It would seem that the learners in the more fluent group control the output somewhat less than the learners in the less fluent group. As hypothesized, the written production of the more fluent group is somewhat less accurate as to the finiteness and the subject-verb agreement. When it comes to the past tense, we must add another parameter to the accuracy, variation between passé composé and imparfait. In this case, the production of the more fluent group is not more accurate, but there is more variation between the two tenses. These learners give more personalised background information to the reader. The less fluent group, seems to concentrate on one of the two tenses, mostly passé composé, and they communicate less background information to the reader. Finally, following Krashen (1981) and Pica (1985) we also made the hypothesis that simple grammatical rules are easier to learn in an explicit way than rules trying to reflect a more complex grammatical structure, complex rules. In this thesis, the simple rule is exemplified by the subject-verb agreement in 3rd person plural, and the complex rule by the distinction between the past tenses passé composé and imparfait. This hypothesis was confirmed : the treatment of the simple rule, measured by pausing, was more explicit than the complex rule. The thesis ends with some research propositions in the psycholinguistic and didactic fields.