Moderating Two Raters on a 6-Mark Gadya Question
At a suburban Mumbai junior college, a Kannada teacher is moderating the prose section of a 40-mark internal test taken by 58 students. The gadya question carries 6 marks and asks students to explain the significance of a turning point in one of the prescribed lessons. The department's locally written rubric splits the marks as content 3, organisation 1, language 2, and the teacher has just discovered that the two raters are applying it differently. Rater A gives 5/6 to a 90-word retelling that ends with the sentence 'this shows the author's greatness', because it is fluent and long. Rater B gives 4/6 to a 55-word answer that names the narrator's change of attitude and ties it to the lesson's title, because it is short. A third script raises a harder problem: the student's reading is defensible from the text but was never taught in class, and Rater A has marked it wrong. Twelve scripts form the moderation sample and marks must be submitted tomorrow.
1. Decide which sample answer deserves the higher content score and justify the decision band by band using the rubric.
Read the worked answer
The 55-word interpretation should win on content, and the retelling should not be rescued by its length. The content band asks for an explanation of significance, and significance is a relation: between the incident and a character's change, the lesson's argument, its title, or its closing image. The 90-word script supplies sequence (what happened) but never supplies that relation; its final clause is an unearned evaluation that could be pasted onto any lesson in the book. Fluency and length are already rewarded by the organisation and language bands, so giving them content marks double-credits the same quality. A defensible split: retelling = content 1, organisation 1, language 2 = 4/6; interpretation = content 3, organisation 1, language 2 = 6/6. Rater A's 5 is the specific error to correct in moderation, because it is the extra point that rewards retelling. Rater B's 4 is also wrong, but in the opposite direction: short is not a fault when the demanded relation is present.
2. How should the teacher handle the third script, whose reading is textually defensible but was never taught?
Read the worked answer
Apply an evidence standard rather than a coverage standard. Ask one question of the script: does it point to words, lines or incidents in the prescribed text that support the reading? If yes, the student has done the work the content band describes and should receive content marks, even if the interpretation was not the one taught; if the script asserts a reading with no textual anchor, it earns partial content at best. Two practical consequences follow. First, consistency: having decided the standard on one script, the teacher must re-read all twelve moderation scripts for similar untaught-but-supported readings, because a standard applied once is not a standard. Second, documentation: add a line to the key recording the accepted alternative reading and the textual evidence for it, so that the second rater and any future marker reach the same result. Refusing a supported reading because 'we did not do it in class' confuses taught content with assessable skill.
3. Write the short guidance note you would add to the rubric so that the same dispute does not recur next test.
Read the worked answer
Suggested note, to sit directly under the content band: 'Content marks are awarded only for the relation the question asks about - why the incident matters, what it reveals, how it connects to the theme, title or a later scene. A retelling of the incident, however accurate, scores 0–1 on content. Length is never a criterion; a complete answer in 50 words beats an incomplete one in 100. Where a student offers a reading not taught in class, award content marks if the script cites textual support, and record the reading in the key.' The wording matters: it names what earns marks and what does not, it separates content from language so that fluency cannot migrate into the content band, and it gives the second rater a decision rule for the awkward case rather than leaving it to taste. Adding this note is also faster than re-moderating every script after the fact.