Connected speech versus the phoneme: a B2 pronunciation judgement
You teach a 90-minute evening class of 16 B2 learners: 11 L1 Spanish speakers, 3 Korean and 2 French. A diagnostic recording shows that all 16 can produce /θ/ and /ð/ accurately in word lists, but in a role-play 9 said /fɪŋk/ for 'think' and 12 dropped the /ð/ in 'that' after a consonant, as in 'I think that' → /aɪ fɪŋk æt/. The coursebook unit, 'The sounds of English', provides a minimal-pair discrimination task (/θ/–/t/, /ð/–/d/) and a repetition drill. Your plan is to spend 25 minutes on the drill because the diagnostic flagged the phoneme, then move to a fluency role-play. You still have the original role-play recording and can re-record. The department expects a 20-minute phonology focus in the lesson and a short evaluation note.
1. Evaluate the diagnosis and the planned 25-minute drill. What does the evidence actually show, and what would you prioritise?
Read the worked answer
What the evidence shows. Accurate production in word lists means the phonemes are already in the learners' inventory. The role-play errors are therefore performance and automaticity problems, not a phoneme gap. /fɪŋk/ is a substitution, while dropping /ð/ after a consonant is connected-speech elision. A 25-minute minimal-pair drill targets perception and isolated production, so it is unlikely to transfer to fluent speech. Priorities should be the chunks that recur in the role-play and carry interactional weight ('I think that…', 'Do you think…?'), together with rhythm and prominence. Intelligibility, not accent eradication, is the criterion.
What to keep. The discrimination task can stay as a brief noticing stage, but only if it is tied to the chunks learners actually use. The drill should not dominate the lesson. The planned fluency role-play is valuable and should remain, because it supplies the context in which the target forms need to work.
2. Propose a revised 20-minute phonology focus with concrete stages and a rationale grounded in phonology and classroom practicality.
Read the worked answer
Revised 20 minutes.
- 4 min – noticing from their own data. Learners listen to two short extracts from the original recording and mark where they hear 'think that' and what happens to the sounds. This makes the gap salient without teacher explanation.
- 5 min – dictogloss of four target chunks. Reconstruct 'I think that…', 'Do you think…?', 'I don't think that…' and 'What do you think…?'. Attention is on /θ/ and /ð/ in context, not in isolation.
- 6 min – rehearsal under communicative pressure. Pairs rehearse a 30-second transactional turn using the chunks and record it on a phone.
- 5 min – listen back. Pairs identify one target they achieved and one thing to keep.
The rationale is that connected-speech forms are best noticed and rehearsed inside meaningful chunks; isolated drills do not build automatic access. The stage is practical: it uses the existing recording, needs only phones, and fits the department's 20-minute requirement.
3. How would you evaluate whether the focus worked, and what would count as evidence?
Read the worked answer
Re-record the same role-play after two weeks and compare like with like: count substitutions of /θ/ and elisions of /ð/ in the target chunks, not in word lists. Ask learners to identify their own tokens from the recording; self-monitoring is part of the target. Evidence of success would be fewer substitutions in connected speech and more accurate chunk production under time pressure, plus learner comments showing awareness of the chunk rather than the isolated sound. A gain in word-list accuracy would not be sufficient evidence, because it was already high at baseline. Keep the evaluation small and realistic: two or three target chunks, a before-and-after count, and a short learner reflection.