One place left: the fast one or the slow one
Two eleven-year-olds are being considered for the same slot in a six-week exam group. Ana talks fast and a lot: in a mock oral she filled every silence, described the picture in nine sentences, and produced 'goed', 'buyed', 'he don't like' and 'a apple'. Asked 'What is your favourite sport?', she answered about her brother's football team without ever answering the question. Ben speaks slowly. He waits three to four seconds before answering, produces five or six short accurate sentences on the picture, self-corrects ('he go- he goes', 'the apple'), and answers the question he was asked. After about ninety seconds he ran out of things to say and simply stopped rather than filling. Both read and write at a similar level. There is one place. The oral tasks at this level ask candidates to answer simple questions, describe what is happening in a picture, and take part in a short exchange with the examiner.
1. Decide which candidate is closer to what the oral tasks actually require, and justify it by the demands of the tasks rather than by which child is easier to teach.
Read the worked answer
The tasks are not asking for accuracy; they are asking for relevant response, intelligibility, and a sustained exchange. Judged that way, Ben is ahead - narrowly.
- Ana's fluency is not communication. The exchange task requires answering the question asked; she did not, and that removes evidence from every item, not one. Her errors ('goed', 'a apple') are developmental and do not block meaning, so they are not the reason to hesitate.
- Ben answers what is asked, is comprehensible, and self-corrects, which shows he monitors form - usually a sign that the pause is processing, not blankness. His weakness is volume and speed: he stops at ninety seconds, so the examiner may not get enough evidence to reward him.
So the decision turns on which deficit is cheaper to fix in six weeks. Both are coachable, but Ana's habit costs her on every single item, while Ben's costs him on the two extended ones - and the pause is far easier to shorten than a question-ignoring reflex is to break. Take Ben, on condition the four-second latency drops; if it does not, the place is not well spent.
2. Give one intervention for each, with a two-week success marker.
Read the worked answer
Ana - answer first, then add. Twenty questions per session, five seconds to give a one-sentence direct answer, and only then one extra sentence. Record and check relevance cold, not during. Marker: 18 of 20 answers on-topic for two consecutive sessions, judged by someone who has not heard the questions before.
Ben - start and fill. Sixty-second picture talk with a four-prompt frame (who / where / what doing / one guess about feelings), plus a stock opener practised to automaticity: 'In this picture I can see…'. The target is latency, not length, so I time the gap between question and first word. Marker: starts within two seconds on 8 of 10 questions, and reaches four or more sentences in sixty seconds.
Note that I am not correcting Ana's past tense or Ben's articles in these two weeks. Both are within expectations for this level and fixing them would eat the time the real targets need.
3. Design a paired activity that uses each one's strength. State precisely what each gains.
Read the worked answer
Picture-question swap, six minutes, twice a week.
Round 1: Ana holds a picture Ben cannot see and describes it for sixty seconds; Ben must then answer three questions I supply ('Where is the boy?', 'What is he doing?', 'Is it raining?'). Round 2: they swap, with Ben given the frame and Ana given the questions.
What Ana gains: relevance under mild pressure. Ben's three questions cannot be answered by talking about her brother, so she gets immediate feedback that description has to serve a listener's needs - the exact skill the exchange task tests.
What Ben gains: speed and volume under time pressure, with a model of fluency in front of him, and a legitimate reason to speak for a full minute without stopping.
Both also rehearse the examiner role, which lowers the stakes of being questioned. I would listen to round 1 only, so that round 2 is genuinely practice rather than performance.