The พริกไทย Trap: Separating Vocabulary Gaps from Pragmatic Gaps in Short Dialogues
A learner preparing for CU-TFL Listening Section 1 - ten short dialogues, each played once, one question per dialogue - has scored 6/10 on two consecutive mock tests. Reviewing item 6, the transcript reads: customer, "ขอผัดไทยเผ็ดน้อยนะครับ"; waitress, "ได้ค่ะ แต่พริกหมดนะคะ ใส่พริกไทยแทนได้ไหมคะ"; customer, "เอ่อ…งั้นเอาอย่างอื่นก็ได้ครับ". The question asks what the customer will eat; the options are (A) pad thai, (B) pad thai with pepper, (C) a different dish, (D) nothing. The learner chose B. On item 9 the speaker says "ไม่เป็นไร เดี๋ยวฉันไปเอง" and the learner chose "the speaker accepts the offer of help." In feedback the learner says the audio is "too fast", asks for longer vocabulary lists, and wants the transcript played twice. The real test is in five weeks and the target is 8/10.
1. Is the item-6 error a vocabulary problem? Explain what actually failed, using evidence from the transcript and from item 9.
Read the worked answer
No. He already knows ผัดไทย, พริก, พริกไทย, เผ็ดน้อย - every content word in the exchange. What failed is pragmatic tracking across a repair sequence.
The waitress's turn is a conditional substitute (ใส่พริกไทยแทนได้ไหม), not an announcement of what will be served. The customer's "เอ่อ…" is a hesitation marker, and "เอาอย่างอื่นก็ได้" closes the whole dish, not the pepper question. ก็ได้ signals reluctant concession, not enthusiasm. Option B is a deliberate lexical lure: it rewards recognising พริกไทย while ignoring who accepted what.
Item 9 shows the same failure: ไม่เป็นไร + เดี๋ยวฉันไปเอง is a polite refusal of help, but he read ไม่เป็นไร as "fine, yes". Both errors sit at the level of speaker stance, not word recognition.
Evidence that vocabulary is not the bottleneck: he selected the option containing the one word he heard clearly, and his proposed remedy (word lists) targets a level that is already sufficient. More vocabulary will not fix a stance-tracking problem; only one-play inference practice will.
2. Name the distractor type at work here and design three solo practice routines that target it. Specify what the learner does, for how long, and with what material.
Read the worked answer
Distractor type: keyword bait embedded in a negotiation/repair sequence - the correct answer is carried by the final stance move, the wrong answer by a salient content word.
Routine 1 - Speech-act dictation (15 min/day). Take 10 short exchanges from Thai call-centre clips, cooking shows or drama scenes. Listen once only, no pausing, and write one word per turn: offer / refuse / hedge / accept / complain. Compare against a second listen. This trains turn function rather than lexis.
Routine 2 - Pre-answer lure prediction (10 min, 3 items). Before playing, read the options and mark which one is "keyword-true but inference-false". After listening, name the exact turn that overturned it. This makes the trap conscious.
Routine 3 - Final-turn shadowing (10 min). Replay only the last five seconds and paraphrase it aloud in Thai. Thai dialogues usually resolve in the final turn; learners who over-weight the middle miss the resolution.
Build a stance bank, not a vocabulary list: ก็ได้, ไม่เป็นไร, เดี๋ยว, คง, น่าจะ, ไม่ค่อย, พอไหว, เอ่อ. Every drill runs under one-play conditions; the transcript is opened only after the answer is locked in.
3. How would you know, within five weeks, that the fix is working rather than that scores drifted up by luck? Give measurable indicators and one thing to stop doing.
Read the worked answer
Track four things weekly, on 10-item sets of unseen material:
- Accuracy split by item type. Tag every item as literal detail vs inference/stance. Target: literal stays ≥80% while inference rises from ~40% to ≥70%. Whole-score gains alone are ambiguous; the split tells you where the change happened.
- Error signature. Code each wrong answer as lure (chose the keyword option), mishearing (wrong word), or no-parse (could not follow). A real fix shows lure errors falling while the others stay flat or fall too.
- Self-report timing. After each item, note whether the decision came from the middle or the final turn. Increasing "final turn" decisions is the behavioural marker of the new strategy.
- Delayed retest. Re-run one set from three weeks earlier; retention, not same-day performance, is the evidence.
Stop doing: replaying audio and re-reading transcripts immediately after an item. It converts a listening-inference task into a reading task and hides the deficit that the exam actually measures.