Fluent but Arguing the Other Way
A B2 writing paper asks for a 200–250 word opinion essay on the prompt: Should schools ban smartphones? Argue for the ban and respond to one counter-argument. A candidate writes 240 words of fluent, well-linked prose with two minor slips. She argues that phones should not be banned, gives three reasons, and never addresses a counter-argument. The rater awards B2 because the language is strong. The examination requires a minimum B2 on the writing component in order to pass. A colleague challenges the decision, and you are asked to settle it.
1. Is B2 defensible on this script?
Read the worked answer
The language sample is genuinely B2, but the script does not deliver the performance the task asked for, and the component is a whole, not a language score.
The prompt fixes both the position (argue for the ban) and one required element (a counter-argument). She takes the opposite position and omits the element. Whatever the quality of her sentences, the script is not evidence that she can do what the B2 writing descriptors describe - presenting a viewpoint and weighing an opposing one.
The defensible settlement is to separate the two judgements rather than let fluency carry everything. Score language at B2 and task fulfilment below threshold, then apply the component rule; the component result is below B2. If the analytic split cannot be applied reliably with the current scale, cap the overall band at B1. What is not defensible is a B2 pass driven by grammatical accuracy when the candidate wrote a different essay from the one set.
2. Suppose the prompt had read: Discuss whether schools should ban smartphones. How does your decision change, and what does that tell you about the task itself?
Read the worked answer
It changes substantially. Under that wording the opposite position is a legitimate choice, so the only defect is the missing counter-argument - a partial rather than a wholesale failure of task fulfilment. With language comfortably at B2 and 240 coherent words, a single missing element at the top of the word range is borderline; where the ambiguity sits in the prompt rather than in the response, resolve it in the candidate's favour and pass the script with an annotation.
The larger lesson is about construction. A prompt that can be read as prescribing a position or as inviting a discussion will produce scripts that raters then punish. Before the paper goes live, someone should ask of every prompt: could a reasonable candidate answer this and still fail? If yes, rewrite it. And when a construction fault is found after the event, it is the candidate who should be protected, not the item.