Ninety-Six Percent Accurate and Still Not Reading: Class V English
Mr. Kumar teaches English (Language II) to Class V in a government school in Warangal: 38 children, one 40-minute period five days a week. He ran a running record on a 60-word Class V passage about Meena, who misses her bus. Fourteen children read it at 40–60 words per minute with 95–98% accuracy; they read 'already' and 'platform' correctly. On six oral comprehension questions asked in English, the same fourteen scored 1 or 2. They answered 'Where did Meena go?' correctly, but answered 'Why was Meena worried?' with 'Yes', or 'She is running', and could not say what happened at the end. Nine other children read below 90% accuracy and stumbled on 'because' and 'school'. The remaining fifteen were in between. Mr. Kumar's plan for the week is to have the whole class read the same passage aloud four times and write the answers to the six questions on Friday; the head teacher has asked him to 'bring up the class average' before the next unit test.
1. What exactly does 95–98% accuracy with 1–2 out of 6 comprehension tell you, and what does it not tell you?
Read the worked answer
What it tells you: word recognition for this passage is adequate. Accuracy above about 95% means the text is within these children's decoding range; the bottleneck is not sounding out words, so more phonics for this group is not the answer.
What it does not tell you - two confounds before you label the group 'poor comprehenders':
- The question language. The questions were asked in English. 'Why was Meena worried?' demands a because-clause that the children may not control, even when they have understood the story. Answering 'Yes' to a wh-question is a typical second-language move: the child matches a familiar question routine to a word she recognises.
- The question type. 'Where did she go?' is literal and was answered correctly. 'Why was she worried?' and 'What happened at the end?' need inference, causal linking and end-of-story integration, plus vocabulary such as 'worried'. The failure is at text level, not at word level - and it is specific to those question types.
So the evidence shows: decoding is fine, literal retrieval is fine, inference through English question forms is weak. It does not show that the children cannot comprehend, that they have a comprehension 'disability', or anything about their comprehension in Telugu. Before concluding, re-ask two of the questions in Telugu, or offer picture options, and see how many of the fourteen answer correctly. That single check separates 'cannot infer' from 'cannot say it in English', and it decides what you teach next.
2. Rework the week for a class with three clear groups in 40 minutes a day, without abandoning the head teacher's concern about the average.
Read the worked answer
Tell the head teacher the truth about the average: drilling one passage four times will raise Friday's score and teach almost nothing - the class will have learned that passage, not reading. The average that matters is built from separated fluency and comprehension scores, which is what the new plan produces.
Structure every period as 20 minutes meaning + 15 minutes word work + 5 minutes shared close.
- Group A (the 14 fluent decoders): pre-teach four words with pictures (worried, miss, hurry, late); paired reading; sequence five pictures of the story; answer three questions using sentence frames ('Meena was worried because ______'); act the scene. Their work is inference, not more reading aloud.
- Group B (the 9 below 90%): ten minutes on the sight words 'because' and 'school' and two or three other high-frequency words; echo reading of a shorter, simpler version of the same story; then the same picture-sequencing task so they meet the same comprehension demand at a text level they can decode.
- Group C (mixed 15): paired reading with a Group A partner, then the sequencing task with sentence starters.
Whole class, daily: one teacher read-aloud of the story, with the key question first asked in Telugu and answered in Telugu, then asked again in English. Friday's test keeps the same six questions but with picture support, allows a Telugu 'because' answer translated with help, and reports fluency and comprehension as two separate scores. That is how a class average is honestly raised.
3. How would you assess comprehension for these fourteen children fairly, given how little English they can produce?
Read the worked answer
Assess comprehension separately from English production. If the answer has to be a written English sentence, you are mostly measuring writing, and the fourteen children's written English is weaker than their understanding. That produces scores that are reliable but invalid.
A practical assessment set for one short passage:
- Picture sequencing - five pictures of the story put in order. No English needed, tests narrative integration.
- True/false with a reason - 'Meena was worried because she was late.' Child says true or false, then explains in Telugu or English. The reason is the evidence; the label is not.
- Two oral questions - one literal, one inference - answered in Telugu, then re-asked in English if the child is willing.
- Retelling in the home language, scored on whether the causal link is mentioned.
- Three 'because' frames for those ready to write in English.
Scoring: report three rows - decoding/fluency, comprehension, English production. A child can be at grade level in comprehension and below in production; the report should show that, or the class will be taught the wrong thing.
Also re-test the same skill on a new passage after teaching, not on the passage used for practice. And check the pass criteria: if 12 of the 14 now answer the 'why' question correctly when allowed Telugu, the original 1–2 out of 6 was an artefact of the assessment, not a finding about the children.