The Silent Speaking Partner
In a paired Speaking assessment (B1, 12 minutes: 2 minutes long turn each, 6 minutes interaction, 2 minutes wrap-up), Learner A speaks fluently and dominates; Learner B gives one-word answers and nods. B is not rude - B is anxious and has a stutter that worsens under time pressure. The interaction criterion rewards 'initiating and sustaining turns.' A's score is rising; B's is collapsing. The examiner must score each candidate independently.
1. How should the examiner handle the interaction so that B's score reflects B's ability rather than A's dominance?
Read the worked answer
The examiner must manage turn-taking without coaching content.
Actions:
- Intervene on floor-holding, not ideas. Use neutral prompts directed at B: 'B, what's your view on that?' This is standard examiner practice and does not supply language.
- Protect processing time. Give B a few seconds of silence after a question; do not let A fill the gap. The stutter worsens under time pressure, so a slower pace is an accommodation, not a hint.
- Score the interaction criterion on B's initiations that occurred, not on A's volume. B's one substantive turn may still demonstrate turn-taking ability.
What NOT to do: do not penalise B for A's dominance, and do not reward A for speaking over B. A's fluency score is separate; A's interaction score should reflect whether A invited B in - A never did, which is itself evidence.
2. A's score is rising partly because A talks more. Is that fair? How do you separate the two candidates' scores?
Read the worked answer
It is not fair if volume alone raises A's score. The criteria must be applied to quality of interaction, not quantity.
Separation method:
- A's interaction score: A sustained turns but did not initiate exchange - no questions to B, no response to B's contributions. Under 'initiating and sustaining turns,' A sustained monologue but failed the interactive element. This should temper A's score, not inflate it.
- B's interaction score: judge on the turns B did take - did B respond relevantly, did B attempt to initiate even once? A single well-placed question from B is stronger evidence than ten words of filler.
- Independent descriptors: write brief evidence notes per candidate as you go ('A: 4 long turns, 0 questions to partner'; 'B: 2 relevant responses, 1 attempted question').
Principle: score the behaviour against the descriptor, and explicitly note dominance as a factor in A's interaction mark.