Literal Reading High, Inferential Reading Low
In a Class 8 English classroom of 40 students, the teacher uses a 120-word passage about a village boy who walks four kilometres to a library and hides his books under his shirt when it rains. An eight-item multiple-choice quiz has five literal items and three inferential items. Results: 82% correct on literal items and 24% on inferential items. For the item 'Why did the boy cover the books with his shirt?', 26 students chose 'because he was afraid of the rain' rather than the text-supported reason that he had no bag and wanted to protect the books. In oral retelling, most students recount the sequence accurately. When asked to find the sentence that supports an answer, 14 students can locate it; 9 can explain how it supports the answer. The teacher has been teaching comprehension by translating each sentence into Hindi and asking 'What does this mean?' Students can decode English but rarely reread to check. The next unit is a short story with a flashback. The teacher has five periods of 35 minutes.
1. Interpret the gap between literal and inferential scores, and explain why the translation routine may be contributing to it.
Read the worked answer
The data show a decoding-to-comprehension gap, not a general English deficit. Literal items at 82% and accurate oral retelling indicate students can access the gist and retrieve stated facts. The inferential items at 24%, plus 26 students choosing 'afraid of rain', suggest they answer from prior expectation rather than from the text. The boy's reason involved having no bag and protecting books; 'afraid of rain' is a plausible world inference, but it is not the textual relation the item tests. The translation routine may encourage this: if every sentence is translated, students do not have to hold the English in working memory or reread to link pronouns and causes. Only 14 students can locate supporting evidence and 9 can explain it, so the class needs explicit work on evidence, not more vocabulary lists. The next unit's flashback structure will make this harder because temporal cues matter. The judgment is to teach inference as a text-based procedure: question, locate, link, explain.
2. Design a five-period scaffolded sequence that teaches students to answer inferential questions with text evidence.
Read the worked answer
Across five 35-minute periods: Day 1, model one inference with a think-aloud. Read the target sentence, ask 'What do I know from the text? What do I know from life? Which one answers this question?' Mark the evidence. Day 2, do a teacher-led evidence hunt with three questions: one literal, one causal, one about a character's feeling. Students underline the sentence and draw an arrow to the question. Day 3, pair work with a question stem sheet: 'The text says ___. This makes me think ___.' Require one quote per answer. Day 4, small-group jigsaw: each group gets one inference item, finds evidence, and prepares a 30-second explanation; the teacher listens for the link. Day 5, independent short task with a new 80-word paragraph; students answer two inferential questions and write one sentence of evidence. Keep translation to a minimum; use bilingual glosses only for blocking words. For the flashback unit, add a timeline strip so students can track time shifts.
3. What assessment would show whether students can justify an inference, rather than just guess the expected answer?
Read the worked answer
Use a two-part assessment. Part A: a short unseen passage with three inferential questions; for each, students must select an answer and underline the evidence. Score the answer and the evidence separately. Part B: an oral or written 'explain your evidence' task: students choose one answer from the quiz and complete the frame 'I chose ___ because the text says ___.' A rubric with three levels: selects a plausible answer with no evidence; selects a text-supported answer but cannot explain the link; selects and explains the link. Also keep a running record of which question types are missed: causal, pronoun reference, character motive, or time sequence. If literal scores stay high and inference rises to 50-60% after the sequence, the intervention is working; if students still choose 'afraid of rain', reteach the specific sentence with a contrastive question: 'What would the text say if he were afraid of rain?' and have them find what it actually says.