Music, Lyrics and Working Memory in Three Classes
A secondary school in Nairobi is revising its study-skills programme. A teacher runs a lunchtime experiment with three intact Grade 11 classes (n = 28, 31 and 26). Each class studies a 600-word passage for 12 minutes under one condition: silence, instrumental music, or music with lyrics. Immediately afterwards, students complete 10 multiple-choice comprehension questions and a 15-item recall test. The lyrics group scores highest on comprehension but lowest on recall; the silence group is most consistent. Students were not randomly assigned; the classes differ in prior attainment, and the music was chosen by the teacher. Several students in the lyrics group said they "knew the songs" and sang along. The head of department wants to announce that music with lyrics improves comprehension and should be allowed in all lessons. A student with ADHD says silence helps her concentrate.
1. Using the working memory model, explain the pattern of results.
Read the worked answer
The working memory model separates the phonological loop, visuospatial sketchpad, central executive and episodic buffer. Lyrics contain verbal information that competes with reading and rehearsal for the phonological loop. If the task requires holding words or sentences in memory, the lyrics create dual-task interference, especially when students sing along, because articulation consumes the loop. Instrumental music has less phonological content and may mainly affect arousal or mood.
Comprehension scores can be higher in the lyrics group for a different reason: the music may raise arousal or create a preferred context, or the groups may differ in prior knowledge. The model predicts a trade-off. Immediate comprehension of a passage can rely on gist and prior knowledge, whereas the 15-item recall test demands verbatim storage, which is exactly what the lyrics disrupt. The student with ADHD reminds us that individual differences in distractibility matter; the model describes a general architecture, not a fixed capacity prediction for every person. The explanation is therefore plausible but not conclusive without controlling prior knowledge and singing along.
2. Evaluate the proposed announcement. Identify two threats to validity and suggest one redesign.
Read the worked answer
The announcement would overgeneralise. Threat 1: selection and intact classes. The three classes were not randomly assigned and differ in prior attainment, teacher, time of day and syllabus coverage. Any difference could be a class effect, not a music effect. Threat 2: lack of control over the music and demand characteristics. The teacher chose the music, students knew the hypothesis, and the lyrics group sang along. Singing is a confound because it changes both attention and rehearsal. A third issue is that comprehension was measured immediately, with no delayed test.
Redesign: randomly assign students within each class to conditions, use the same piece of music across sessions, counterbalance order, measure prior knowledge with a pre-test, and include a delayed recall test. Analyse with the individual as the unit, not the class. A manipulation check such as "how much did the music distract you?" would show whether the independent variable worked. Until then, the school should not make a whole-school policy. A small within-subjects pilot would be a fair next step.
3. Explain how schema theory or levels of processing could provide an alternative explanation. What would you measure to distinguish between explanations?
Read the worked answer
Schema theory predicts that comprehension depends on prior knowledge: students with more background knowledge about the passage topic can fill gaps and score well even while distracted. Levels of processing predicts that deep, semantic processing produces better recall than shallow processing; lyrics may encourage shallow, sound-based processing, or may simply reduce available capacity.
To distinguish these, measure prior knowledge with a topic-specific test and give a recognition test as well as free recall. If prior knowledge predicts comprehension but not recall, schema effects are likely. If a semantic orienting task improves recall in all conditions, depth of processing is implicated. Also measure arousal and self-reported distraction. This matters because the original conclusion attributes the effect to music when it may be a group difference in knowledge or strategy. The cognitive approach favours controlled experiments with clear operationalisation, not whole-class comparisons, so the alternative explanations are testable rather than merely rhetorical.