Eight codes behind every question in the Key Stage 2 reading paper - what each one really assesses, what a question looks like, and why 2a and 2g are so easy to confuse.
6 min read
They are the eight strands of reading comprehension used to build England's Key Stage 2 English reading test: 2a vocabulary in context, 2b retrieval, 2c summarising, 2d inference, 2e prediction, 2f how content is related across a text, 2g the effect of word and phrase choice, and 2h comparison within a text. They come from the Standards and Testing Agency's test framework, and every question in a KS2 reading paper is tagged with one in the mark scheme.
England is the scope, and it is worth saying once: the content domains, the framework and the programme of study behind them are England's, and the other UK nations assess reading under their own arrangements. AgentChamp is built for children across the UK, so this is one system it fits inside, not the only one.
That tagging is why the codes turn up in gap analyses, planning grids and progress meetings: they are the shared vocabulary the test itself uses, so one school's "shaky on 2f" means what another's does.
No. The content domain is drawn from the national curriculum programme of study for Years 3 to 6, but it covers only the part a written test can reach, and it is written for the people who build the tests rather than as a scheme of work.
The gap matters. The programme of study asks children to read widely for pleasure, to discuss the books they read, to build the knowledge that makes a text about the Blitz land at all. None of that survives contact with a one-hour paper.
So use 2a-2h to check coverage, never to write the curriculum: a plan with eight objectives because there are eight domains is describing the test, not the reading. Note too that the programme of study binds maintained schools - academies, free schools and independent schools are not bound by it.
Each domain isolates one thing a reader does, and each has a recognisable question shape.
Through a single English reading paper sat in the May of Year 6 - one hour including reading time, worth 50 marks, with a reading booklet whose texts increase in difficulty, a separate answer booklet, and a raw score converted to a scaled score where 100 represents the expected standard.
Question formats vary far more than most comprehension worksheets do: multiple choice, matching, ordering, true-or-false grids, find and copy, short answers and longer explanations worth up to three marks. The mark scheme names the domain beside each question number, which is what makes question-level analysis possible.
The domains do not appear in equal measure. But a question-level analysis tells you how one cohort did on one paper, and rebuilding next year's teaching around last year's paper is how a curriculum ends up written by a test.
Key Stage 1 has its own shorter set of five domains, 1a to 1e: vocabulary (1a); key aspects of fiction and non-fiction, such as characters and events (1b); the sequence of events in a text (1c); inference (1d); and prediction (1e).
Three line up letter for letter with their KS2 counterparts - 1a and 2a on vocabulary, 1d and 2d on inference, 1e and 2e on prediction - with KS2 raising the demand rather than changing the skill: 2d is 1d plus "explain and justify those inferences with evidence". KS1 has no equivalent of 2f, 2g or 2h. Sequencing (1c) has no KS2 domain of its own; my reading is that the work reappears inside summarising and inside 2f.
The KS1 national curriculum tests became non-statutory from the 2023/24 academic year; the Year 1 phonics screening check did not, and remains statutory. The domains have not gone anywhere either. And note where Year 2 sits: a Year 2 child is in KS1, described by 1a-1e, and may be reading a text that would carry a 2d question. Key stages are administrative boundaries, not reading ones.
Treat them as lenses on one good text rather than as a menu of eight lessons. A single chapter can carry a retrieval question, an inference question worth arguing about, and one question on why the author chose that word - and it is experienced as reading, not as coverage.
Three things that paid off across nine years of doing this the hard way. Ask the question orally before anyone writes: a child who has said an answer aloud, and heard a better one, writes differently afterwards. Teach find and copy as its own discipline - copying a whole clause when the question said one word is a test-literacy problem, not a reading one. And keep the comprehension lesson and the comprehension test apart: the first should generate talk, disagreement and re-reading; the second is only a measurement, and a measurement teaches nobody anything.
The version I got wrong for years was domain-labelled worksheets attached to texts nobody cared about. I once had to sit an assessment about ducks in a pond myself. I still do not know why, and it did not do much for my love of reading. Children do the analytical work when they want to know what happens next.
Six characters, mapped onto the eight KS2 domains, with two of them carrying a pair:
Three honest caveats. This is a KS2 mapping only: we make no claim on 1a to 1e. It is a framework, not a promise that every child meets every strand in every book - a branching story's quiz draws five of the six (C, H, A, M and P), and the linear reading books add the sixth, Summary. And which letters a linear book carries follows its band, not the year group: retelling leads the earliest question-bearing books, and analysis, which needs the most fluency behind it, arrives last.
Stated by year group, because key stages cannot say it precisely: phonics missions run from Reception to Year 2; CHAMP comprehension starts in Year 1, as spoken, adult-led discussion prompts rather than a marked quiz, from the Blue book band up, with the pre-fluency bands carrying none by design; and the branching story library runs from Year 2 to Year 6.
AgentChamp keeps you signed in and remembers your reading settings using cookies and browser storage. What we store, in full.