Repair and Scoring in a Simulated Interpersonal Speaking Task
During a simulated conversation in Chinese, 20 students each complete six turns with the teacher as interlocutor. The prompt asks students to respond to a friend's invitation, decline politely, suggest an alternative, and confirm a time. One student, Ana, answers turns 1 and 2 in short phrases. On turn 3 she is silent for eight seconds, then says, “I don't know.” The teacher has a script with follow-up prompts but is unsure whether repeating the prompt or moving on changes the score. Later, on turn 4, Ana produces a memorized 25-second monologue about her weekend that does not answer the question. The class period is 45 minutes, and the teacher is the only rater.
1. Should the teacher repeat or rephrase the prompt on turn 3?
Read the worked answer
A limited repair is acceptable if it does not supply the answer. Repeating the same prompt once, or rephrasing it more simply, can preserve assessability without coaching. The teacher should not, however, answer for Ana, give vocabulary, or turn the turn into a guided production.
The key is consistency: if the teacher repeats for Ana, the same option must be available to all students, and the repetition should be noted. A single repetition is a procedural accommodation; repeated rescuing changes the task and inflates the score. Moving on after a reasonable wait is also defensible if the rubric scores interaction and task completion, because silence is evidence of difficulty.
2. Ana produces a memorized 25-second monologue on turn 4 that does not answer the question. How should it be scored?
Read the worked answer
Score it low on task completion and interaction, even if pronunciation and fluency are strong. The monologue may show memorized language, but it does not respond to the interlocutor's prompt, so it is not interpersonal communication.
A useful scoring note: “Fluent but off-task; does not address the invitation or propose an alternative.” If the rubric has a comprehensibility row, the student may earn some credit there, but fluency alone should not lift the overall score. This is exactly the distinction the interpersonal speaking construct requires.
3. How can the teacher maintain fairness as the only rater across 20 students?
Read the worked answer
Use anchoring and documentation:
- Follow the same script, wait time, and repair rule for every student.
- Record the conversations (with permission) so scores can be reviewed.
- Score one dimension at a time rather than holistically in one pass.
- Pre-score two or three anchor samples to calibrate.
- Write a brief evidence note for each score.
This does not eliminate rater bias, but it makes the decisions consistent and auditable. It also protects students if a score is challenged.