This paper reports on a Content and Language Integrated Learning (CLIL) project in a French high school, focusing on second language acquisition and learner corpus research. Specifically, the presentation focuses on the subject-specific vocabulary acquisition, its use and development by learners of a foreign language within a hard-CLIL immersive setting, along with the principal results.
A multimodal CLIL corpus of spoken and written productions was compiled capturing the characteristics of a year-long history course taught in English at B2 CECRL level (L1 French) in a French secondary school. The corpus consists of 16 hours of classroom video recordings (transcribed and annotated) plus over 230 written materials produced by teachers and pupils. Accompanying data were obtained through interviews and questionnaires. The resultant mixed multimodal corpus contains almost 120,000 words of teachers' and learners' input and output.
The paper draws on the theoretical research and analysis of the corpus exploring the pupils' interlanguage via 50 corpus-based and corpus-driven measures using CLAN, AntConc and SketchEngine. The focus is on CLIL learners' interlanguage profile and the interaction between input and output. The interlanguage features examined are : frequent and specific vocabulary, N-grams, errors, systematicity and creativity in learners' oral and written productions. Comparisons are made with similar academic and general corpora: BASE, BAWE, BNC and COCA.
Some of the main conclusions drawn are :
CLIL subject-specific terminology is highly specific, acquired mainly through repetitive multi-level interaction patterns and pre-fabricated words bundles. The interlanguage is characterised as being systematic, creative and highly interactive. Errors play the fundamental role in the foreign language literacy learning process.
Study of N-grams and collocations reveals various types of subject-specific phraseological units relevant to the context studied, and having very little in common with general corpora, while keeping some common traits with academic corpora. The terminology acquisition is organised in bundles, or building blocks (Biber, 2006) used by learners in contexts both similar and different from those given by the teacher.
Various interaction techniques are used by two co-intervening teachers (English and history) including trailing-off (MacWhinney, 2000), prompting and questions-answers sessions. A strong connection and interdependence exist between the input given (what is taught/received) and the learners' output (production). Multi-levelled cumulative input is created both by teachers and learners: those speaking engender input for those who listen.
The paper concludes with a reflexion on the perspective of an ecological learning system (Doyle, 1986): learners' output is conditioned by their capacity to handle multidimensional classroom factors (among others) and constraints at a given time period.
BIBLIOGRAPHY.
Biber, D. (2006). University language: A corpus – based study of spoken and written registers. Amsterdam: Benjamin.
Doyle, W. (1986). Paradigmes de recherche sur l'efficacité des enseignants. In M. Crahay & D. Lafontaine (Éds.). L'art et la science de l'enseignement (p. 435‑482). Éditions Labor.
MacWhinney (2000). The CHILDES Project : Tools for Analyzing Talk: Vol. Part 2: The CLAN Program (3rd Edition). Lawrence Erlbaum Associates.
