Paired Speaking Assessment in French 4
At a French 4 class of 22 students, the teacher uses a three-minute paired speaking assessment. Students receive a prompt asking them to discuss a proposed ban on single-use plastics in their town. The rubric has four categories: task completion, comprehensibility, language control, and vocabulary. In a recorded sample, Dominique, a strong student, speaks for 2 minutes 40 seconds, uses English fillers such as 'like' and 'you know,' circumlocutes around 'recycling plant' by saying 'the place where they do the things with bottles,' and avoids the subjunctive despite the prompt requiring hypothetical recommendations. Her partner, Mei, says only 'oui' and 'je suis d'accord' twice. The teacher must assign individual scores and plan the next unit. The department also wants consistent scoring across sections. The teacher has the recording and a transcript.
1. How should the teacher score Dominique and Mei separately? Identify what evidence is missing and what cannot be inferred.
Read the worked answer
Score the two students on separate evidence, not on a shared conversation average. Dominique's sample shows strong task completion and comprehensibility: she sustains the topic and is understandable. Her language control is mixed. Circumlocution, 'the place where they do the things with bottles,' is strategic competence and should be credited under communicative strategies, not penalized as a vocabulary failure, unless the rubric defines vocabulary as precision. English fillers do count against language control or fluency depending on the rubric, so the teacher must apply the descriptor literally. The subjunctive avoidance is relevant only if the prompt required hypothetical recommendations and the rubric names grammatical range; one sample may not be enough to infer stable control. Mei's evidence is insufficient. Two agreement tokens do not demonstrate low proficiency; they demonstrate that the partner dominated. Do not give her a high score by assumption or a low score by default. Reassess Mei with a different partner or a short monologue.
2. What instructional sequence would address Dominique's avoidance of the subjunctive and Mei's silence?
Read the worked answer
Instruction should separate the two problems. For Dominique, the goal is not to eliminate circumlocution but to add precision and syntactic range. Use an information-gap task in which each partner holds half of a city plan and must make recommendations with 'il faudrait que,' 'à condition que,' and 'bien que.' Give planning time and require at least three hypothetical structures; record a two-minute sample. For Mei, the goal is turn-taking and risk-taking. Assign her the expert role with information Dominique needs, provide sentence starters for disagreeing, and use a timed round-robin before the paired task. Model repair phrases and require each partner to speak for 45 seconds. This is more likely to transfer than a subjunctive worksheet because the forms are tied to a communicative function.
3. What changes to the assessment task would improve validity and inter-rater reliability?
Read the worked answer
Validity improves if the assessment separates monologic and interactive performance and if raters are trained. Use two tasks: a one-minute individual response and a three-minute paired interaction, so a dominant partner cannot erase the other's evidence. Write task-specific descriptors: what counts as task completion, what counts as a filler, and how circumlocution is scored. Train raters on anchor recordings from previous classes; double-score 20% of samples and discuss discrepancies. Use an analytic rubric rather than a single holistic score, and do not average partners. Finally, align the prompt with the unit's functions; if the unit taught recommendations, the prompt should require them, and the rubric should say so.