A blank document asks an English learner to solve several problems at once: choose an idea, order it, find the words, build the sentences and check the grammar. For a B1 writer preparing a 250-word company update, that pile-up often produces a careful opening followed by rushed, repetitive sentences. A brief content plan before drafting separates idea work from sentence work. The plan should be small enough to finish, and the writer still needs time to monitor the language while composing.
That is too much at once.
Two kinds of planning
Researchers distinguish pre-task planning from online planning. Pre-task planning happens before the first sentence. The writer decides what the text will say and how the parts will connect. Online planning happens during composition: pausing over a verb, rereading a clause, moving a sentence or checking whether a pronoun has a clear reference. The distinction matters because each kind of time appears to support a different part of performance. In Rod Ellis and Fangyuan Yuan's 2004 study, 42 Chinese learners wrote picture-based narratives under one of three conditions. Some received 10 minutes to plan before writing, some had unpressured time while writing, and some wrote without planning support. Pre-task planning was associated with faster production and a wider range of verb forms. Unpressured online planning was associated with more error-free clauses.
One study with one narrative task cannot settle how every learner should write an email, an essay or a GEPT response. A teacher who applies the result to a timed report, a complaint letter and an examination essay should treat the transfer as a working classroom hypothesis, then inspect what each writer actually produces. It does give teachers a sharper diagnosis. A learner who has little to say needs help generating and arranging content. A learner whose ideas arrive quickly while tense and agreement collapse needs room to monitor during the draft. Giving both learners the same generic instruction to "plan more" hides the difference.
What belongs in the plan
A usable plan records meaning, sequence and a small language problem. It does not contain polished sentences. If the writer spends 17 minutes composing the plan for a 25-minute task, the plan has become a first draft with less room to revise.
Consider a hypothetical project coordinator in Hsinchu writing an update about a delayed shipment. Four notes are enough:
- purpose: explain the two-day delay and protect the client relationship
- order: cause, current status, revised delivery date, next contact
- evidence: customs release received at 14:37 on Thursday
- language check: past tense for completed events; will for the next update
Those notes remove four decisions from the opening sentence. They also preserve the route through the paragraph when the writer stops to search for clearance or checks the difference between delay and postpone. The oddly precise time, 14:37, has a job: it tells the client which event has already happened. Decorative detail would only consume attention. The plan can be shorter for a familiar task. A three-line email may need only "request, reason, deadline." A source-based academic response needs more: claim, evidence, relationship between sources and the writer's position. My preference is a plan that looks slightly incomplete. Its gaps leave decisions for the draft, where full sentences make the consequences visible.
Protect time inside the draft
Planning before writing does not guarantee accurate prose. The 2004 study found the strongest accuracy result in the condition with unpressured online planning, where writers could allocate attention during composition. A later research synthesis and meta-analysis by Johnson and Abdi Tabari examined an inconsistent body of L2 writing research and treated planning as several operationally different conditions. That variation is a warning against turning "10 minutes of planning" into a universal prescription. A classroom task can protect online planning with two simple constraints. Set a drafting checkpoint after roughly one third of the available time, then reserve the final 4 minutes for a targeted read. The checkpoint asks whether the text still follows the planned route. The final read follows one feature at a time: verb time, sentence boundaries or reference words such as it, this and they. Searching for every possible error at once recreates the original pile-up.
This is also where a model answer can help.
Showing it before the learner has planned may replace the learner's decisions with imitation. Showing it after a first attempt creates a comparison. Our article on using a model answer after a first draft explains why that order can direct attention to a visible gap. For planning practice, ask the learner to compare routes through the content before comparing individual phrases.
Match the support to the bottleneck
Some writers already over-plan. They build elaborate mind maps, colour-code every branch and delay the moment when language has to appear on the page. Give them a 3-minute cap and require a rough first sentence when the timer ends. Others begin instantly and discover their position halfway through paragraph two. Give them a fixed planning frame and withhold the document until the frame has four entries. Task familiarity changes the calculation as well. An experienced sales assistant may plan a routine follow-up email in 23 seconds because its communicative sequence is known. The same writer may need 8 minutes to organise a complaint that cites three invoices and proposes a remedy. Planning time should follow the number of unresolved decisions, rather than the nominal CEFR level of the writer.
Teachers can assess the plan separately from the finished text.
Check whether each planned point appears, whether the order survived and where the writer departed from it. A useful departure shows that drafting produced a better idea. A missing final point may show that lexical searches consumed the remaining minutes. These traces tell the teacher which constraint to change in the next task. For tomorrow's writing, use one sheet. Draw a line about one third of the way down. Put four content notes above it, draft below it, and circle the exact minute reserved for the final language check. If the last note never reaches the paragraph, the next lesson has a concrete place to begin.