The three-second pause and the helpful examiner
The L2 oral examination consists of three papers marked independently by two examiners: (1) reading aloud and summarising a short unseen text; (2) a commentary and discussion on its topic; (3) a general interaction. Sofia reads fluently, but in her summary reports the author as praising the new transport policy when the text is clearly ironic. In paper 2, Examiner 1 asks open questions; Examiner 2 twice interrupts a three-second pause to supply vocabulary ('you mean congestion?') and rephrases questions into yes/no form. The three papers last 4, 6 and 5 minutes instead of the allotted 5, 8 and 7. Examiner 1's marks are 13, 12 and 14; Examiner 2's are 10, 9 and 10. The examiners must now agree one mark per paper and a single oral mark.
1. How far has Examiner 2's support contaminated the assessment, and how should papers 2 and 3 be scored?
Read the worked answer
The oral construct is 'what can this candidate do, unaided, in comprehension, production and interaction'. Examiner 2's interventions damage that construct in two distinct ways.
Supplying lexis. When the examiner offers 'congestion', the candidate no longer has to retrieve the word, so the range and lexical-resource evidence for that exchange is the examiner's, not hers. Any criterion that rewards range must be scored on what she produced before the prompt.
Closing the questions. Converting open prompts into yes/no questions removes the demand to construct an extended answer. That affects fluency, coherence and interactive competence - the very criteria paper 2 is designed to sample. The result is an inflated performance in exactly the dimension Examiner 2 is trying to help with.
But support is not automatically contamination. What the candidate does after support is legitimate evidence: does she take the offered word and build on it, repair, reformulate, ask for clarification appropriately? That belongs under interactive competence, provided it is recorded as prompted.
Scoring. Paper 2 should be marked on the unaided stretches, with the prompted exchanges noted and used only for the interaction criterion. Paper 3 (general interaction, 5 instead of 7 minutes) is largely unaffected by the specific prompts but has less evidence than intended; mark what exists and note the shortened duration. Paper 1 is a separate matter: fluent delivery does not offset a misread irony, which is a comprehension failure and should pull the summary criterion down to the middle or lower band, whatever the reading-aloud quality.
2. How should the two examiners reconcile marks of 13/12/14 and 10/9/10?
Read the worked answer
Not by averaging to 11.5, 10.5 and 12. That produces a number without a reason and conceals the fact that the two examiners are, in different degrees, marking different things.
Step 1: write before speaking. Each examiner independently writes the criterion-level evidence from their contemporaneous notes for each paper - comprehension of the text, range, accuracy, fluency, interaction, response to the task - before seeing the other's sheet.
Step 2: locate the divergence. It is largest on paper 2 (three points), which is where the prompting occurred. That is not a coincidence and it is the first thing to discuss: did Examiner 2 mark down because she knew she had intervened, or did she mark up because the interaction felt smooth? Both are distortions, and the notes should settle it.
Step 3: test the impressions against evidence. On paper 1 the three-point gap cannot rest on the reading aloud, on which both agree. It must rest on the summary, where the observable fact - the irony misread - points one way.
Step 4: agree per paper, then aggregate according to the scheme in force. If a paper remains outside tolerance after discussion, the moderation route in force applies; it should not be replaced by a private compromise.
Step 5: record the prompting. A short factual note that support was given, when and of what kind, protects the mark if it is appealed and protects the school if the practice recurs.
3. The oral ran for 15 minutes instead of the allotted 20. Is that a procedural problem?
Read the worked answer
Only if it deprived the candidate of the opportunity to demonstrate the construct. The relevant question is not the clock but whether all three papers were covered and whether the candidate was given a fair chance within each.
On the facts, all three papers took place, but each was short: 4 of 5, 6 of 8, 5 of 7. The important loss is on paper 2, the discussion, which is the paper with the most scope for extended production and the one where prompting also occurred. Fewer minutes plus more examiner talk is a compounding disadvantage: less candidate language to mark, and a higher proportion of it prompted.
What to do. Establish why the time was short. If the candidate spoke at length and simply had little more to say, that is a performance feature, not an irregularity. If the examiners started late, cut a paper short or dominated the exchange, that is a fairness issue: record it factually on the mark sheet and report it to the coordinator, who can decide whether the procedure in force requires anything further. Do not silently add marks as compensation, and do not silently ignore it.
What to avoid. Neither examiner should now try to reconstruct a longer performance in memory. The mark must rest on the evidence actually generated, with the time and the prompting recorded so that anyone reviewing the case can see the conditions under which it was produced.