Marking an Interpretation Mark: When a Script Disagrees with the Key
You are the subject convenor for Urdu in a district pre-board examination taken by 62 second-PUC students. Question 4 of the Nasr section carries 4 marks: «Trace the change in the narrator's attitude towards his father and state what the lesson asks the reader to take away.» The provisional key awards 2 marks for the change (one per textual moment) and 2 marks for naming the lesson's stated message - respect for the dignity of labour. During moderation you find 14 scripts that trace the change accurately through two scenes but then argue the lesson's real message is different: that the father's silence is a dignity the narrator only belatedly learns to read. These answers are anchored in the text, not invented. Six other scripts name the approved message correctly but cite no scene at all. You have one afternoon to finish moderation, and a head of department who wants the key applied «as printed».
1. How do you mark the 14 scripts offering the alternative reading, and what do you record?
Read the worked answer
Decision: accept them, subject to an anchor test.
- The second bullet tests interpretation, not recall. A message mark should reward a reading the text can bear, not a phrase the candidate has memorised.
- Apply a rule rather than a hunch: 4/4 if two scenes are cited and the stated message follows from them; 3/4 if the message is coherent but only one scene is cited; 2/4 if the message is asserted with no scene.
- Do not deduct for «not matching the key». Reword bullet 2 in the moderation file as «any defensible statement of the lesson's takeaway, supported by the scenes cited in bullet 1», and attach two of the 14 answers as anchor scripts.
- Log the change with time and initials. The paper is defensible only if the revised descriptor is visible to anyone auditing the batch later.
2. A script names the approved message correctly but cites no scene. How many of the 4 marks, and on what reasoning?
Read the worked answer
Award 1 out of 4.
- The question has two requirements: tracing a change, and drawing a takeaway. This script does only the second, and does it by bare assertion.
- The tracing bullet (2 marks) is unmet: 0. On the message bullet (2 marks), the idea is correct but unexplained, so it earns half of that bullet: 1.
- Write in the margin that the lost marks are for evidence, not for the opinion - otherwise the candidate concludes that the «wrong» answer is being punished while the right phrase earns nothing.
- The temptation to give 2 comes from recognising a familiar sentence. But a guidebook supplies that sentence to anyone; the question explicitly asks the candidate to connect it to the text, and this answer does not.
3. A colleague objects that you cannot revise a descriptor after 40 scripts are already marked. How do you respond and what do you do practically?
Read the worked answer
Re-mark Question 4 in all 62 scripts before results are released.
Changing a descriptor midway is exactly how inconsistency enters a batch; the answer is to re-apply one standard to everyone, not to freeze the error.
- Stop marking the rest of the paper and finish only Question 4.
- Re-mark Question 4 blind to the earlier mark, using the anchor test above.
- Compare old and new totals; list every script that moves.
- Keep the original mark visible but treat the new column as final.
- Record the rationale and the anchor scripts in the moderation file.
To the colleague: the printed key is a guide, not statute. Since Question 4 is a single 4-mark item, re-marking 62 scripts takes well under half an hour - a small cost against a challenge from a parent or the board.