The Degraded Audio File and the Confident Speaker
SMK Seri Murni runs the SPM Arabic listening and oral components in the same week. Cikgu Hafiz recorded a 95-second radio interview in which a volunteer coordinator describes a coastal clean-up; the file was captured on a phone in the school hall, and the last twelve seconds carry a heavy echo. Item 7 asks what the coordinator did after the campaign ended, and the key is «نظّمنا ورشةً للطلاب». In the degraded audio the م of نظّمنا sits close to a fricative, and 19 of 34 candidates chose the distractor «نظّف الشاطئ» - an option made attractive by the campaign topic itself. In the oral, Siti speaks for three minutes with real confidence, but code-switches twice («gotong-royong», «kebajikan»), says «إنّ الطلابُ مجتهدون» twice, and drops most case endings elsewhere. The head of department wants item 7 voided and Siti capped at a mid band.
1. Should item 7 be voided? Justify your decision by reference to what the item is actually measuring and to the acoustics of the keyed phrase.
Read the worked answer
Decision: do not void; replay once and re-record for the next cohort.
The reasoning turns on the difference between a hard item and a failed item. A failed item has no defensible key; a hard item has one key that is recoverable by a competent candidate. Here the key is recoverable, for two reasons:
- Syntactic cue. «نظّف» is transitive and needs an object. The object the distractor supplies, الشاطئ, is precisely what the campaign itself was doing - so «نظّف الشاطئ» after the campaign ended is redundant, not sequential.
- Cohesive cue. The stem asks what happened after the campaign. «نظّمنا ورشةً للطلاب» answers the sequence; the distractor repeats the campaign.
What is genuinely damaged is the phonetic cue: م and ف are both labial and both follow the emphatic ظ, and echo smears the distinction. That is an acoustic problem, so the remedy is acoustic - one replay of the segment, which is standard accommodation for degraded audio.
Two cautions. First, the distractor is topic-primed, so part of what the item discriminates is resistance to priming rather than listening. Note the facility: 19/34 is 56% on a four-option item, which is above chance but not comfortable. If a re-run after replay still shows more than 40% on the distractor, drop the item and redistribute its mark - that is the threshold to apply, not intuition. Second, write the next item so that the keyed word is not phonologically adjacent to the passage's own topic vocabulary.
2. How should Siti's three minutes be scored, and in what order should the errors be addressed in her feedback?
Read the worked answer
Score the criteria separately and refuse to let one error be charged twice.
- Fluency and interaction: high. Three minutes sustained, self-repaired, no collapse - that is the behaviour the oral component rewards.
- Lexical range: split the two switches. «gotong-royong» names a Malaysian practice with no single Arabic equivalent; rendering it as a loan item or glossing it («عملٌ تطوعيّ جماعيّ») is a legitimate strategy, not a gap. «kebajikan» is a gap, because Arabic offers «العمل الخيريّ», «الرفاه الاجتماعيّ», «الإحسان». Mark the switch, not the concept.
- Accuracy: «إنّ الطلابُ مجتهدون» is a governance error, not a slip - it occurred twice. Dropping case endings in unscripted speech is normal even among native speakers and should not be scored as if it were the same phenomenon.
Feedback order: (1) confirm the communicative success first, because the student's risk-taking is the asset you do not want to extinguish; (2) teach circumlocution as a strategy so that a lexical gap does not force a switch; (3) reteach إنّ وأخواتها with three substitution drills; (4) only then mention pausal case endings, as a register point.
On the head of department's request: capping her at a mid band because of code-switching misreads the construct. The oral tests communicative Arabic, and one cultural loan word is not a failure of that. Deduct under the accuracy and range criteria where the evidence sits, and leave the fluency band alone.