Moderating a 6-mark prose inference answer: price detail versus motive
At a district trial for the Madhyamik First Language Santali paper, 38 scripts are being moderated. Item 3 carries 6 marks: 'The narrator’s grandmother refuses a buyer’s offer of forty rupees for a brass pot. What does this refusal reveal about her character? Support your answer with one detail from the passage.' The passage printed above the question says the pot came from her father’s house and was used at every family wedding. One script reads: 'She was poor and needed money, but she did not sell the pot because she was lazy. This shows she did not care about money.' The script mentions the forty-rupee offer but no other textual detail. The marking scheme awards 2 marks for accurate inference, 2 for relevant textual evidence, and 2 for coherence. One moderator proposes 3/6 for the price detail; another proposes 1/6 because the inference is wrong. You must set the score and write feedback the school can use.
1. What score out of 6 would you award, and why? Refer explicitly to the three rubric dimensions.
Read the worked answer
Score: 2/6.
- Inference (0/2): 'Lazy' and 'did not care about money' contradict the passage's emphasis on the pot's family origin and wedding use. The refusal is better read as attachment to lineage or memory, not indifference. A wrong inference earns no inference mark even if confidently stated.
- Textual evidence (1/2): The forty-rupee offer is a real detail, but it is used only to support the wrong inference. The stronger details-father's house, every family wedding-are absent. One relevant but misapplied detail is worth partial credit, not full.
- Coherence (1/2): The answer has a claim and a reason, so it is not incoherent; but the reason ('lazy') does not follow from the evidence, so it cannot earn full coherence.
Why not 3/6: A moderator may want to reward the price detail, but the rubric already gives that detail only 1 evidence mark; adding a third mark would double-count it. Why not 1/6: The script does contain a relevant detail and an explicit inference, so it is not a blank or off-topic answer. Feedback should say: 'You noticed the price-now ask what the pot meant to the grandmother. Use the father's-house and wedding details to explain her refusal.'
2. Rewrite the question so that it tests inference without inviting a character label such as 'lazy'.
Read the worked answer
Revised item: 'The grandmother keeps the brass pot even though forty rupees would be useful. What does the passage suggest the pot means to her? Use one detail from the passage and explain how it supports your answer.'
Reasoning: The original asks 'what does this reveal about her character?', which invites a trait label. The revision shifts to 'what does the pot mean?', a meaning-making task that still requires inference but anchors it in the object and its history. It also demands a warrant ('explain how it supports'), so a student cannot simply assert 'lazy'. The mark scheme should accept any plausible meaning-memory of her father, family continuity, wedding custom-provided the chosen detail genuinely supports it.
3. What single change to the marking scheme would most reduce this kind of moderator disagreement?
Read the worked answer
Split 'inference' into two visible rows: 'plausible inference about the pot's meaning' (1) and 'inference is warranted by the quoted detail' (1). Then the price-only answer can earn evidence (1) + coherence (1) + warranted-inference (0) = 2/6, while an answer that says 'the pot reminds her of her father's house, because it was used at every wedding' can earn the warrant mark even if the exact wording differs. This makes the difference between 'a detail is present' and 'the detail is used' explicit, which is where the 3/6 and 1/6 moderators were really disagreeing.