Errors are evidence of an interlanguage system, not only failures. Respond according to their source (transfer / developmental / teaching-induced) and whether they impede communication.

Where the sources disagree

  • Transfer vs developmental: Thornbury doubts that most errors are L1 interference. Teachers in single-L1 classes (Jaworski) see transfer everywhere.
  • Written CF: contested (Truscott vs Ferris). See ELT Research Bites - reading queue.
  • Synthesis across sources (Claude): Lyster & Ranta (1997) found recasts rarely produce uptake, while prompts (elicitation, metalinguistic cues) do. Tsateri says connect first, correct later (on intuition). Ellis & Shintani (via Fluency First) say both immediate and delayed correction can help. The best-supported principle is to be selective (few, global, developmental errors) and noticeable, more than any rule about timing.
  • In-flow vs delayed: Underhill (Demand High) upgrades language during fluency work with silent cues; commenters and the delayed-feedback tradition say interrupting fluency harms it, especially with anxious learners.
  • Prevention vs correction: Dellar: “prevention is infinitely desirable to cure” (design tasks within learners’ priming). Pinard: a structured response sequence with explicit, varied techniques. These are complementary: Dellar is about planning, Pinard about the moment.

Notes

More notes (by source)

After Octopus (Vedrana Vojkovic Estatiev)

Key terms (DELTA M1)

error vs mistake (Corder) · global vs local error · contrastive analysis · error analysis · interlanguage · fossilisation / stabilisation · overgeneralisation · CLI · recast · elicitation · metalinguistic feedback · clarification request (Lyster & Ranta 1997 typology)