Sixty-Two Scripts, Two Weekends: What an Essay Score Should Represent
Madam Owusu teaches English to a Form Three class of 62. She sets a 350-word argumentative essay: 'Should mobile phones be allowed in secondary schools?' She marks at home over two weekends, correcting every error in red and writing one score out of 30. Students glance at the score and put the essay away. Two scripts trouble her. Akosua's essay has six errors in 350 words, five short paragraphs, one reason ('phones are bad because they distract'), no counter-argument, and a conclusion that repeats the introduction. Yaw's essay has nineteen errors, one long paragraph, three reasons with examples, an acknowledged counter-argument, and a vivid opening. Madam Owusu scored Akosua 16/30 and Yaw 11/30. A colleague says this is fair because English is about grammar. Madam Owusu has 62 scripts, two weeks and no photocopier.
1. Identify the construct problem in these two scores and explain what the marking practice is actually measuring.
Read the worked answer
The intended construct. The task is an argumentative essay. It asks students to take a position, support it with reasons and examples, anticipate an objection, and organise the whole for a reader. Quality on that construct is what the score is supposed to represent.
What the scores actually represent. Akosua's six errors are being allowed to outweigh the fact that she offers one reason, ignores the counter-argument and ends where she began. Yaw's nineteen errors are outweighing three supported reasons, a handled objection and a deliberate opening. Reversing their positions means the score is tracking error density, not argument. This is construct underrepresentation: the trait the task targets is barely sampled. It also introduces construct-irrelevant variance, since error counts are affected by essay length, handwriting and how tired the marker is at script forty.
Why the red pen is not neutral. Correcting roughly twenty errors across 62 scripts produces well over a thousand corrections that no student reads. More seriously, it teaches a rule: writing is the avoidance of visible error. The predictable response is what Akosua has already done - short sentences, safe vocabulary, minimal risk. That is negative washback, and it works against the very reasoning the examination rewards.
The colleague's argument. 'English is about grammar' confuses one component of writing with the whole. Accuracy matters, but an essay is a rhetorical act. A marker who scores only accuracy should not set an argumentative task in the first place.
2. Propose a revised marking scheme with weightings and justify the weights against the purpose of the task. Show how the two scripts would be scored.
Read the worked answer
A four-trait analytic scheme (30 marks).
- Content and argument (10): clear position, relevant reasons, use of examples, acknowledgement of the other side.
- Organisation and cohesion (8): paragraph function, ordering, linkers, introduction and conclusion that do work.
- Language range and accuracy (8): sentence variety, vocabulary choice, control of tense and agreement.
- Mechanics (4): spelling, punctuation, capitalisation.
Justification of weights. The task asks students to persuade a reader. Content and organisation therefore carry 18 of the 30 marks: they are the point of the exercise. Language carries a substantial 8, because accuracy and range affect whether the argument is understood and because the syllabus requires it. Mechanics carries only 4, because spelling and punctuation errors rarely block comprehension and are better handled as a separate editing habit; weighting them heavily rewards caution.
Applying the scheme. Akosua: content 4, organisation 4, language 7, mechanics 3 = 18. Yaw: content 8, organisation 6, language 4, mechanics 2 = 20. Yaw now leads, but narrowly, and the profile shows exactly why each lost marks. That profile matters more than the total: Akosua needs a reason and a counter-argument, Yaw needs sentence boundaries and paragraphing.
Moderation. Write one-sentence descriptors for each band, mark five scripts together with a colleague, and compare. A single total hides the pattern; a profile plus one strength and one target makes the score teachable.
3. How should Madam Owusu handle 62 scripts in two weeks so that feedback actually changes what students write next?
Read the worked answer
Move some of the work before submission. A 40-minute peer response session with a three-item checklist - is there a counter-argument, does each paragraph have one job, is the conclusion doing new work - removes the largest problems before marking. Students submit a clean copy, so the teacher is not correcting the same weakness 62 times.
Stagger collection. Take half the class each week. This halves the pile at any moment and leaves time for the whole-class error clinic, which reaches every student more efficiently than individual red ink.
Mark in two passes. First, read the whole script quickly for meaning and assign content and organisation. Second, scan for the two most frequent error patterns and code them in the margin rather than rewriting. Correcting every slip produces fatigue in the marker and passivity in the writer.
Run a weekly error clinic. From the fifty most common errors across all scripts, pick five patterns, put anonymised sentences on the board and have students repair them. This is where grammar teaching belongs, and it costs ten minutes.
Conference the weakest ten. Five ten-minute conferences a day for two days. Oral feedback on one target transfers faster than a page of symbols.
Return within a week. A script returned after a fortnight teaches nothing. Also keep a class error log and a small portfolio of three pieces, so that the final judgement rests on evidence rather than one pressured essay.