MCQs, translations and mini-whiteboard checks show performance in that task, not acquisition. Knowing a rule, recognising it, processing it for meaning and using it spontaneously are four different achievements.
Claim
“A correct answer in a controlled task is evidence of successful performance in THAT task, not proof that the construction has been acquired.” Teachers get “systematically reassured by the wrong kinds of success”. Result: weak listening and spontaneous production that only shows up at high-stakes exams.
Mechanism / theory
Automaticity and practice (DeKeyser & Criado 2013; DeKeyser 2013); input processing and structured input (VanPatten 1996, 2004); interleaving and delayed retention (Nakata & Suzuki 2019); corrective feedback (Lyster & Saito 2010); Teachability (Pienemann 1989, 1998).
Five low-prep checks that probe acquisition
- Listen and choose: form–meaning in connected speech (A/B on whiteboards), then with temporal clues removed.
- Complete and transform: stem → change the reason → remove the stem.
- Say something about yourself: no model; does the form appear unprompted?
- Timed sentence challenge: 5 seconds; measures retrieval efficiency.
- The unexpected comeback: revisit the form 3 weeks later through a different topic, without naming it.
Limits & critique
- Polemical tone; targets a named blogger (Bastow).
- Speeded written retrieval ≠ oral automaticity (he concedes).
- Conceptual. His checks aren’t shown to predict outcomes either.
Apply in my context (Claude)
- This is the implicit vs explicit knowledge distinction (R. Ellis 2005, measuring with timed vs untimed tasks). Use those terms in DELTA M1.
- RS Insight diagnostic: add an “unexpected comeback” item type (target form elicited 3 weeks later, unnamed). It would make the platform measure acquisition, not coverage. This is a clear differentiator.
- Ties to your thesis video “Covering the syllabus ≠ learning”.