Wim Gombert

CHAPTER 3. This study 51 programs in reading and listening skills was relatively easy as these tests were readily available, and scores could easily be compared with the national average scores as a benchmark. But measuring the language production skills of speaking and writing was more challenging. Because speaking and writing skills cannot really be measured by means of closed question types, reliable free response tests for the assessment of oral and written production had to be designed and validated. For reading and listening, standardized Cito tests were thus used. For speaking and writing, students’ pro ciency was measured while speaking or writing about a given topic. e choice of topics and the amount of knowledge the students have on a topic may well impact on results. erefore, to avoid bias as a result of di erences between cohorts, a special 30-hour intervention was designed and implemented in both programs. During the nal two academic years, students in both programs worked on seven di erent newsworthy topics that formed the basis of the oral, listening, vocabulary and writing activities that they did in class during these nal two years. e speaking test, which was done in pairs, was done on the basis of four topics that were preselected by the teacher; each student pair was allowed to pick two out of four subjects, one of which was subsequently chosen at random by the interviewer at the beginning of the test. For the writing test, which was administered individually, the same four topics were used. At the beginning of the test, two out of four topics were selected and presented by the teacher and each student chose one of these topics to write about at the start of the test. e general idea was to have students prepare the topics to the extent that they felt comfortable, but spontaneous language was still possible, encouraged and, indeed, needed. STUDY DESIGN AND ANALYSES e overall research question for this classroom study is as follows: Which of the two CLT programs is more e ective a er six years of French instruction in a Dutch secondary school curriculum: a weak version of CLT with a predominantly explicit, structure-based textbook program or a strong version of CLT with a predominantly implicit, dynamic usage-based program? To answer the overall research question, four sub studies were conducted to investigate the e ectiveness of the programs pertaining to the four individual language skills. E ectiveness is operationalized as students’ scores on reading, listening, writing and speaking tests administered in the nal year of secondary school. In general, variation in outcomes might be triggered by group e ects, and multilevel analyses might be needed to take these group e ects into consideration, but as students in the same cohort o en changed groups in their six years of secondary school,