Teaching Reading Comprehension: Annotation, Question-First Reading, and Inference
For tutors: most comprehension failure is not a vocabulary problem or an intelligence problem. It is a process problem — the student reads passively, then hunts desperately. Process problems are trainable.
Ask a struggling student to describe how they tackle a comprehension passage and the answer is usually the same: read the passage once, start the questions, then re-read frantically in search of answers. Every part of that process is improvable, and the improvements are teachable in a handful of tuition sessions. Comprehension performance, at O-Level English or at General Paper, is mostly the sum of three trainable skills: reading with a pen, reading with the questions in mind, and inferring with evidence.
Annotation: reading with a pen
Passive reading leaves no trace; the eyes move, the mind wanders, and at the end the student has an impression rather than a map. Annotation forces transactions with the text. But 'annotate the passage' is useless advice on its own — students respond by underlining half the page, which is highlighting with extra effort. Teach a small, fixed system instead:
- One-line margin summaries: after each paragraph, three to six words capturing its job — 'counter-argument begins', 'example: Japan', 'writer's own view'. This is the single highest-value habit, because it produces a paragraph-by-paragraph map for the hunting phase.
- Circle connectives and pivots: however, despite, yet, on the other hand. These words mark the joints of the argument, and questions cluster around joints.
- Mark tone words: vocabulary revealing the writer's attitude — 'so-called', 'merely', 'remarkably'. Tone and attitude questions are answered from exactly these.
- A question mark for anything not understood on first pass, to return to only if a question demands it.
Cap the marking: if more than roughly a tenth of the passage is marked, the system has collapsed back into highlighting. In sessions, annotate one passage together with the tutor thinking aloud — the modelling matters more than the rules.
Question-first reading
Reading the questions before the passage transforms reading from a general activity into a search with targets. The student skims the question stems — not the full detail, just the topics and question types — and then reads the passage already knowing that something about the writer's attitude in paragraph three will matter. Two cautions keep the technique honest. First, question-first reading supplements a full read; it does not replace one, and students who only hunt for answers without ever reading the whole passage miss every question about overall argument and structure. Second, for GP-length passages, the workable version is questions first, then one complete annotated read, then targeted re-reading per question.
Alongside this, train question anatomy: circle the command word, the line references, and the mark allocation before answering. A two-mark question wants two things. 'Using your own words' is an instruction with marks attached, not decoration. A remarkable share of comprehension marks are lost by students answering a slightly different question from the one printed — usually a better-known cousin of it.
Inference: the evidence chain
Inference questions — what is implied, what the writer's attitude is, why a character responds as they do — are where weaker students guess and stronger students calculate. The difference is a discipline: every inference must be anchored to specific words in the text. Teach a three-step chain the student can say aloud: what the text states, what that suggests, and how far it can be pushed. The third step is where marks die — students routinely over-infer, leaping from 'the writer questions the policy's cost' to 'the writer opposes the policy'. A useful drill is to give an inference and ask the student to argue against it from the text; learning what the evidence cannot support is half the skill.
A drill sequence for tuition
- Weeks one and two: annotation only, on short passages, no questions. Build the habit before it must compete with answer-hunting.
- Weeks three and four: question-first reading plus annotation, untimed, with the student explaining their evidence chain aloud for every inference answer.
- From week five: timed passages, followed by review against the mark scheme — classifying each lost mark as retrieval, inference, expression, or misread question.
Parents can contribute without teaching a single technique: regular reading of editorial and opinion writing — commentary pieces, longer-form journalism — builds the background knowledge and tolerance for dense argument that GP quietly assumes. Comprehension is a skill, but it runs on knowledge, and no annotation system fully compensates for a student who has never met the ideas the passage takes for granted.
- Palincsar, A. S., & Brown, A. L. (1984). Reciprocal teaching of comprehension-fostering and comprehension-monitoring activities. Cognition and Instruction.
- National Reading Panel (2000). Teaching Children to Read: An Evidence-Based Assessment. National Institute of Child Health and Human Development.
- Duke, N. K., & Pearson, P. D. (2002). Effective practices for developing reading comprehension. In What Research Has to Say About Reading Instruction. International Reading Association.