Reading the marks in a translation section
Ten Assamese sentences were set for translation into Tai, two marks each, in the half-yearly paper. In a class of 34 the section average is 9.4 out of 20. The scripts show a clear pattern. Twenty-one students put the verb at the end of the sentence. Fourteen translated Assamese light-verb phrases word for word, producing a verb plus object string where the answer key has a single verb. Six used the polite sentence-final particle in sentences between close friends. Two teachers marked the section using different keys, and in 11 scripts the ticks match neither key. The department meets this afternoon, before the marks go on the register, and the principal wants to know whether the section should be re-taught, re-tested, or both.
1. Diagnose the error pattern. What does it say about the teaching rather than the students?
Read the worked answer
Three of the four patterns are not separate mistakes; they are one habit showing up in three places.
Verb at the end (21 students). Assamese puts the verb last. Tai does not. A student translating under time pressure falls back on the clause shape she has used all her life. This is transfer, not carelessness, and it will reappear in every free composition until the clause pattern is taught explicitly.
Light-verb phrases (14 students). Translating word for word from Assamese produces a verb-plus-object string where Tai uses one verb. The students are not inventing; they are copying the shape of the source sentence, which is precisely what a translation exercise is meant to break.
The polite particle (6 students). Overgeneralisation from a recent lesson - the class learned the polite form and applied it everywhere.
What this says about the teaching: the class has been given vocabulary lists and sentences to translate, but not the two or three patterns the sentences are testing. Thirty-four students did not independently fail to notice word order. The lesson was about words; the paper asked about structure. That is a planning problem, and it should be said in the meeting, plainly, before anyone blames the class.
2. Design the marking rubric for this section and say how it should be announced.
Read the worked answer
Two marks per sentence are enough if each mark has a named job:
- 1 mark - message. All the content of the source sentence is there and nothing is distorted. A different but grammatical choice of words keeps this mark, provided the word is in the primer.
- 1 mark - the pattern the item was set to test. Each sentence in the key must be labelled with its target: sentence 3 targets clause order; sentence 5 targets the light-verb phrase; sentence 8 targets the level of politeness. A student who gets the message but puts the verb last loses this mark and nothing else.
Announce the rubric before the next test, on the board, with one worked example of a sentence that is grammatical but not the key's wording, and show that it earns 2. Otherwise students optimise for copying the key's vocabulary instead of building the clause.
Also state the negative rule explicitly: no mark is deducted for a legitimate alternative, and no mark is given for a correct pattern wrapped around a distorted message. A rubric that only punishes is a rubric students game; one that names what is rewarded teaches.
3. What do you do about the 11 scripts whose ticks match neither key, and the marks already shown to students?
Read the worked answer
Do not quietly re-mark and do not average the two teachers' totals - an average of two inconsistent keys is still an inconsistent key.
Procedure:
- Build one key this afternoon, with each item labelled by its target pattern, as above.
- Re-mark all 34 scripts blind to the earlier marks, by one teacher, with the second teacher moderating a sample of ten.
- Record both marks. Where the new mark is lower, keep the higher of the two, because the students are not responsible for the department's disagreement.
- Tell the class what happened and why: two keys were used, one key is now in force, several marks changed. Naming it costs ten minutes of embarrassment and buys years of trust.
Then use the re-marked set as the baseline. If the class average moves only slightly, the diagnosis stands - the problem is structural, not a marking artefact. Re-teach the clause pattern and the light-verb phrase over three periods, then re-test only the pattern items, not the whole paper, and enter the better of the two scores for those items. A full re-test punishes the class twice for a planning error that was not theirs.