Moderating a Twelve-Mark Composition Item: Case, Not Recitation
A Class X terminal examination at a Sambalpur high school carries a twelve-mark Sanskrit composition item: four English sentences to be rendered into Sanskrit, three marks each. Thirty scripts are in and the answer key is being moderated. The key gives (1) ahaṃ bhrātre pustakaṃ dadau; (2) tau ācāryau asmān pāṭhayataḥ; (3) he rājan, gāḥ pāhi. On paper the correct and the wrong forms differ by one or two letters, yet three students have lost marks. Sanjay wrote bhrātuḥ in sentence 1; Meera wrote asmākaṃ pāṭhayataḥ and ācāryāḥ in sentence 2; Rina wrote he rāja, gāvaḥ in sentence 3. All three protest that their forms appear in the textbook's declension tables and are therefore correct. The head examiner must decide each case, name the vibhakti and vacana of both the correct and the wrong form, and frame a marking rule that would survive a parent's appeal.
1. Adjudicate each of the three disputed renderings. For every form, give the stem, vibhakti and vacana, and identify the reasoning error rather than only the correct answer.
Read the worked answer
Sentence 1 - bhrātuḥ. The stem is bhrātṛ (ṛ-kārānta). bhrātuḥ is ṣaṣṭhī (and, with the same shape, pañcamī) ekavacana - 'of/from the brother'. The sentence needs the recipient, caturthī ekavacana: bhrātre. Sanjay has produced a real form of the right stem in the wrong case; the error is case selection, not declension memory.
Sentence 2 - asmākaṃ pāṭhayataḥ; ācāryāḥ. The pronoun asmad has dvitīyā bahuvacana asmān ('us'), which is the karma of pāṭhayataḥ; asmākam is ṣaṣṭhī bahuvacana ('our'). Separately, tau fixes the subject as exactly two, so ācāryau (prathamā dvivacana) is required; ācāryāḥ is plural. Meera's error is twofold: a genitive substituted for an accusative, and a plural for a dual.
Sentence 3 - he rāja, gāvaḥ. The stem is rājan, a halanta stem; its sambodhana ekavacana is he rājan, not he rāja (which belongs to the different stem rāja). And pāhi ('protect') governs a karma in dvitīyā bahuvacana: gāḥ, not gāvaḥ, which is prathamā bahuvacana. Rina has used nominatives where the verb requires an object.
In every case the forms are attested in tables - which is exactly why 'it is in the book' is not a defence. The item tests karaka-driven selection, not reproduction.
2. What systematic error connects the three scripts, and how should the feedback and re-teaching be designed so the same errors do not recur?
Read the worked answer
Two patterns run through all three scripts.
First, case is being copied from English word order rather than computed from the verb. English marks 'to my brother' and 'us' by position and preposition; the students mapped 'my brother's' and 'our' onto ṣaṣṭhī forms and left the object in the nominative. The remedy is not more declension recitation - they know the tables - but substitution drills that start from the verb: dadau takes a caturthī recipient and a dvitīyā thing given; pāhi takes a dvitīyā object. Ask 'which karaka slot is empty?' before 'which ending?'.
Second, number is being read off the English noun rather than off the qualifier. tau and 'those two' were ignored, and gāvaḥ was pulled in because 'cows' looks like a plural subject. Drill the rule that tau/te, 'two', and dual verb endings jointly force dvivacana.
For feedback, mark the karaka slot wrong in a different colour from the ending wrong, so the student sees the two separate decisions. Keep the stem-plus-ending tables available during the re-test: if the errors vanish when tables are allowed, the problem is analysis, not memory - which is what these scripts suggest.
3. Design the three-mark rubric for each sentence so that partial credit is principled and defensible on appeal.
Read the worked answer
A workable split for a three-mark rendering is: 1 mark for stem and pada (parasmaipada/ātmanepada and person-number of the verb), 1 mark for karaka-correct vibhakti, 1 mark for vacana and agreement across the sentence.
Applied to the disputed scripts: Sanjay keeps the stem mark for bhrātuḥ (right stem, right number) but loses the vibhakti mark, and loses the agreement mark only if the rest of his sentence is defective - 1 or 2 out of 3. Meera loses the vibhakti mark on asmākam and the vacana mark on ācāryāḥ, keeping the verb and stem mark; 1 out of 3. Rina loses vibhakti on gāvaḥ and both stem and sambodhana marks on he rāja, since rāja is a different stem; 0 or 1 out of 3.
The appeal-proof feature is that the rubric names the decision, not the form: the examiner can say precisely which of the three decisions failed and show the student the same three-decision grid for a sentence they got right. It also prevents the opposite injustice, where a single wrong vowel costs a student all three marks even though the karaka analysis was sound. Record on the key, for each sentence, the verb, the karaka slots it opens, and the required vibhakti - so two examiners marking independently reach the same number.