Marking Forty Odia Compositions

The errors are not one kind of thing, and treating them as one inflates the language score's weight.

• Orthographic error (ଣ/ନ, ଡ଼/ଢ଼, the ଯ/ୟ conjunct) - genuine, teachable, and the proper business of the 10 marks.
• Register variation (consistent Sambalpuri verb endings and pronouns) - the writing is systematic, not careless; the writer controls a variety, just not the examination variety.
• Mechanical (no punctuation) - a different skill from spelling and should be marked under organisation/presentation, not double-penalised.
• Dishonesty (the two copied scripts) - an assessment-integrity matter, not a language matter; it should not be handled by an accuracy deduction at all.

For the six Sambalpuri scripts: do not fail them, and do not pretend the difference is invisible. Award content and organisation on merit - they are described as clear and well organised. In the 10 language marks, note the dialect features once each rather than deducting per occurrence, and leave the majority of the sub-score for orthography and sentence control. The defensible principle: mark against standard written Odia for the formal register, but treat dialect as a code-switching skill to be taught - a letter to a Sarpanch and a story with dialogue require different varieties, and the student who can shift between them is more competent, not less.

Fifteen minutes daily, same structure, high frequency:

• Minutes 1–4, minimal-pair sort: eight words on cards, sorted into ଣ and ନ columns (କାରଣ, ପ୍ରାଣ against ନାନା-type words) and into ଡ଼/ଢ଼ against ଡ/ଢ. The physical sorting plus saying the word aloud is what builds the distinction; copying the correct form ten times does not.
• Minutes 5–9, dictation of six words and two short sentences using only taught vocabulary, then immediate self-check against a wall chart, then peer check with a partner. Immediate feedback matters more than the volume of practice.
• Minutes 10–13, proofreading a deliberately flawed four-line paragraph written by the teacher with exactly the target errors plus one error of a different type, so children must discriminate, not just pattern-match.
• Minutes 14–15, one-sentence dictation kept in a log.

Evidence of transfer: (a) error rate in a fresh unseen dictation in week 3 compared with a baseline dictation in week 1, scored by error type; (b) whether children correct their own target errors during editing rather than only on request; (c) the frequency of the same confusions in the next piece of free writing, which is the only measure that matters. If free-writing errors do not fall while dictation improves, the routine is producing test performance and needs a proofreading-in-context step added.

Handle it separately from the marks, and handle it privately. The composition is a learning task; the summative examination three weeks later is where integrity rules must be unambiguous, so the response should teach the norm now rather than punish it retrospectively.

Steps: speak to each student individually, ask them to show where the material came from, and give them the chance to write a short replacement piece on the same topic in class time. Record the incident in your own notes. Do not read the copied scripts aloud, do not put a zero in the language column, and do not use them as an example to the class, because that teaches fear rather than authorship. Follow whatever the school's written policy says if one exists - and if none exists, that gap is worth raising at the staff meeting, since three weeks later the same ambiguity will recur under examination conditions.

For the other 38, the decision protects marking validity: if copied scripts were quietly scored low for 'language', the language sub-score would silently carry a moral judgement. Keep the 10 marks about language, and let plagiarism be answered by a redone task. Then in the next unit, teach the difference between using a source and copying it, with a two-line quoting exercise as part of the letter-writing unit.