34 Words in Four Minutes: Scoring Interaction in a Paired Oral
The oral session lasts 10 minutes: a 2-minute warm-up, a 4-minute information-gap role-play and a 4-minute discussion. In the role-play, Candidate A (Yusuf, 16, Turkish, about B1) holds a bus timetable and must ask Candidate B (Marta, 15, Polish, about A2+) about accommodation; B holds the accommodation card. The examiner's notes record that Yusuf took 78% of the turns, Marta produced 34 words in four minutes, and the examiner asked Marta six closed questions ('Tmur il-Belt? Iva?'), used English twice ('ok, so, your turn') and did not use the picture prompts. Marta's 34 words contain one error. Yusuf translated the task into English for her twice. The rating form gives separate 0–5 marks for interaction and spoken production, with descriptors requiring initiation, turn-taking and repair at each band. The session was audio-recorded; gestures were not captured.
1. Decide what Marta's interaction mark should rest on. Explain why this session is a weak measure of her interaction and state what you would record.
Read the worked answer
Record a provisional mark based on response quality only, and flag the session for a second sample. Do not award a middle band on this evidence.
- What the descriptors require. Initiation, turn-taking and repair. We have 34 words, no initiations, no repair move and no evidence of turn-taking beyond answering.
- Why the evidence is contaminated. Three separate factors suppressed output: the partner took 78% of turns and translated the task into English; the examiner asked six closed questions, which require one-word answers; and the picture prompts, which were designed to elicit, were never used. We cannot tell whether Marta cannot interact or was never given the chance.
- What can be credited. The 34 words are accurate apart from one error, so spoken production has some support. Interaction does not: accuracy in short answers is not interaction.
- What to record. A provisional interaction mark justified only by response appropriateness, explicitly marked 'insufficient evidence', plus a request for a second oral with a different partner and an examiner using open prompts.
- What not to do. Do not inflate to a middle band to look fair, and do not award zero, which would punish her for the examiner's conduct.
2. Redesign the role-play so that both candidates must speak. Give the card content and the required outcome.
Read the worked answer
Design principle: no candidate can complete the task from their own card.
Candidate A's card (bus timetable, partial)
- Departures: 16:10, 16:40, 17:20, 18:00.
- Journey time: 25 minutes.
- You do not have the stop names or the ticket price - you must ask B.
Candidate B's card (accommodation, with a constraint)
- Check-in closes at 18:00.
- The hostel is at stop 'Il-Kastell'; a ticket costs €1.50.
- You must ask A which bus arrives before 18:00 and how long the journey takes.
Required outcome
- Agree on one departure and one meeting point, and state the reason for the choice.
- Each candidate must ask at least two questions and give at least two pieces of information; cards may not be shown to the partner.
Why this works
- The check-in deadline creates a genuine information gap: B has the constraint, A has the times, and neither can solve it alone.
- The 'agree and justify' step forces negotiation, which is where turn-taking and repair become observable.
- Removing the full timetable from A prevents the dominant partner from answering both sides, as Yusuf did.
3. Classify the examiner's use of English and of closed questions: breach, minor deviation, or acceptable scaffolding? Write the note that goes on file.
Read the worked answer
Classification: English use is a breach; closed questioning is a serious deviation. Neither is acceptable scaffolding.
- English. The oral is a Maltese-language assessment. English may be used only for procedural matters before the clock starts. Using it twice during the role-play changes the construct and gives an unfair advantage to candidates who share that language.
- Closed questions. 'Tmur il-Belt? Iva?' invites a one-word answer and lowers the demand of the task. Scaffolding should open a response ('Għidli x'tagħmel nhar it-Tnejn'), not close it.
- Unused prompts. The picture prompts are part of the standardised procedure; omitting them means this session was not delivered as designed.
File note. 'Session 14: examiner used English twice during the role-play and asked six closed questions; picture prompts not used. Candidate B's low output cannot be attributed to her proficiency on this evidence. Recommend a second oral with a different partner and examiner, and a standardisation briefing on open elicitation before the next session.' The note should describe behaviour and its effect, not characterise the examiner.