Pedagogy · ELT methodology
Corrective Feedback
When learners get something wrong, what’s the most useful response? Corrective feedback (CF) is the everyday craft of helping learners notice gaps between what they said and what they meant — without breaking the flow of communication or the learner’s confidence.
“Errors are not signs of failure. They are evidence that the learner is testing hypotheses about the language.”
— S. Pit Corder, 1967
Why it matters
Four core commitments
Good corrective feedback isn’t a reflex — it’s a series of judgement calls about timing, technique, and tone.
Notice, don’t punish
An error is data. The teacher’s first job is to make the learner aware that something went off-track, not to make them feel bad about it.
Form vs. meaning
Decide what the moment calls for. A fluency activity wants a light touch; an accuracy task wants surgical precision.
Output, then input
The most useful feedback prompts the learner to produce the corrected form themselves. Telling them is faster; eliciting them is stickier.
Selectivity beats saturation
Correcting everything teaches nothing. Pick a focus — pronunciation, a tense, a register — and let the rest go for now.
Calibrate to the learner
A nervous beginner needs different treatment than a confident B2. Read the room, read the person.
Make it loop back
One-off correction rarely sticks. Build the target form into the next activity, the next lesson, the next homework.
Taxonomy
Six types of corrective feedback
The classic Lyster & Ranta (1997) framework. Each technique sits somewhere on a spectrum from implicit (subtle, communicative) to explicit (direct, instructional).
Recast
What it is: The teacher reformulates the learner’s utterance correctly, without flagging the error.
Learner: “I go yesterday to the cinema.”
Teacher: “Oh, you went to the cinema yesterday — what did you see?”
When to use: During fluency work, when interrupting would derail meaning. Best for lower-level or anxious learners.
Clarification request
What it is: The teacher signals that something didn’t make sense, prompting the learner to repair their own utterance.
Learner: “I have hungry.”
Teacher: “Sorry — you have?”
When to use: When you’re confident the learner can self-correct. Forces them to monitor and repair.
Repetition
What it is: The teacher repeats the error back, often with rising intonation on the problem word.
Learner: “She don’t like coffee.”
Teacher: “She don’t?”
When to use: Good for highlighting a single, isolatable error without giving the answer away.
Elicitation
What it is: The teacher prompts the learner to produce the correct form by pausing, asking a direct question, or rephrasing.
Learner: “He goed home.”
Teacher: “He… ?” / “How do we say go in the past?”
When to use: When the learner has the form somewhere in their head and just needs to retrieve it. Strongest evidence base for uptake.
Metalinguistic feedback
What it is: The teacher names the rule or grammatical category without giving the corrected form.
Learner: “Yesterday I am tired.”
Teacher: “Watch your tense — yesterday is past.”
When to use: With learners who know the rule but slip in production. Pairs well with grammar-aware adults.
Explicit correction
What it is: The teacher names the error and provides the correct form directly.
Learner: “She have three childrens.”
Teacher: “We say has, and children — no s. Try again.”
When to use: Sparingly — for high-stakes errors, fossilised mistakes, or when implicit techniques have failed. Quick, but lowest uptake of the six.
Timing
When to correct — and when to wait
The most underrated CF skill is knowing when not to correct. Match technique to the activity’s purpose.
| Activity type | Default approach | Why |
|---|---|---|
| Fluency speaking | Note errors, correct after | Interrupting kills the willingness to take risks. Capture errors silently, address 2–3 at the end. |
| Accuracy drill | Correct immediately | The learner is here to get the form right — let them know in the moment. |
| Pronunciation | Recast immediately | Bad pronunciation hardens fast. Model correctly within seconds, then have them repeat. |
| Free writing | Coded correction, delayed | Mark error type with a code (T for tense, WO for word order). Learner self-corrects in the rewrite. |
| Role-play | Stay out | The role-play is the data. Save feedback for the post-mortem. |
| Presentation | Detailed delayed feedback | Watch the video, write notes, debrief next class. In-the-moment correction would crush them. |
In the room
What ifs — common scenarios
Specific moments you’ll meet and what to actually do about them.
Stop hoping it’ll fix itself. Make it the focus of one activity next class — explicit attention, controlled practice, then the same form embedded in a freer task. Track it on a Post-it on your desk.
Use a generic recast: address it to the whole class as a “common error” without looking at the source. Or note it and bring it up one-to-one at break.
Don’t argue — verify together. Open a dictionary, a corpus tool, or a reliable site and let the evidence do the talking. Saves face on both sides.
Say so. “Let me check that.” Modelling intellectual honesty is more valuable than pretending to be infallible. Look it up before next lesson.
Smile, nod, let them finish. Then briefly acknowledge: “Good catch.” Self-correction is the holy grail — don’t over-celebrate it but do reinforce that monitoring works.
Name the interference explicitly: “In Spanish you say tengo hambre; in English we say I’m hungry, not I have hungry.” Contrast helps the rule stick.
Make register the lesson. “That’s grammatically fine, but it sounds very formal — what would you say to a friend?” Sociolinguistic awareness is part of the curriculum.
Switch techniques. Recasts are easy to miss. Move to elicitation or explicit correction — they need a clearer signal.
Stop the activity. Re-teach for two minutes — board the rule, two examples, two contrasts — then resume. A class-wide error is a teaching gap, not a student gap.
Move on. Check in privately afterwards. CF only works in a safe affective climate — if you’ve damaged that, repairing it is the priority, not the grammar.
Self-audit. Pick one focus per lesson and let the rest go. “Today we’re listening for past tense” is more useful than chasing every slip.
“It’s a collocation — that’s just the verb that goes with that noun.” Native speakers have intuition, not always rules. Teach the chunk; explain it later if a rule exists.
In practice
Six techniques that actually work in class
Short, repeatable routines you can drop into almost any lesson.
Error-of-the-day board
One persistent class error stays on the board all lesson. Anyone catching themselves or others using it gets a tally. Builds attention without singling out individuals.
Reformulation
Take 2–3 sentences from a learner’s writing. Rewrite them as you would. Show both. They notice the gap themselves — far more powerful than red ink.
Coded marking
Use a small set of symbols (T = tense, WO = word order, SP = spelling, ART = article). Learner does the actual fix. Effort shifts from teacher to learner — where it should be.
Delayed feedback slot
Last 5 minutes of class, board 5 anonymised sentences from the lesson — some right, some wrong. “Find the errors, fix them.” Group memory work.
Peer correction
Pairs swap written work with a checklist (3 things to look for). Slows the teacher down, sharpens learner attention. Works once trust is established.
Recast + check
Recast the error, then 30 seconds later ask a question that requires the same form. “When did you go to the cinema?” If they produce went, it landed.
Snapshot
A correction that worked
B1 conversation class. Topic: holidays. Recurring error: past simple of irregular verbs.
Two errors, two elicitations, two repairs — under fifteen seconds, no flow lost. The teacher said almost nothing. Marco did the work, which is why the next time he reaches for an irregular past, his monitor will catch it a fraction faster.
What NOT to do
Common pitfalls
- Correcting everything. The learner stops talking. Pick a focus.
- Recasting at lower levels and assuming they noticed. They probably didn’t. Use a more explicit technique.
- Long explanations mid-task. A two-minute grammar tangent kills the activity. Note it, address it later.
- Correcting in front of a low-confidence student. Use private channels — written notes, one-to-one moments, anonymous board work.
- Getting irritated. Errors are evidence the learner is trying. Tone of correction is more memorable than content.
- Repeating the same correction technique forever. Vary it. Recast, elicit, prompt, ignore, board it. Predictability dulls attention.
- Correcting accent. Unless it impedes intelligibility, leave it. Accent is identity. Pronunciation that causes misunderstanding is fair game; everything else is taste.
- Marking writing in red without follow-up. Without a rewrite, the correction is just decoration.
Theory & research
Where corrective feedback comes from
CF sits inside a wider conversation about how second languages are actually learned. The headlines:
The Interaction Hypothesis (Long, 1996)
Learners acquire language through negotiation of meaning — the back-and-forth that happens when communication breaks down. Clarification requests, recasts, and confirmation checks are the engine of this process.
The Output Hypothesis (Swain, 1985)
Comprehensible input alone isn’t enough. Learners need to produce language and have their output pushed for accuracy. CF — especially elicitation — is what does the pushing.
Noticing (Schmidt, 1990)
For input to become intake, the learner has to notice the gap between their version and the target. CF’s whole job is to make that gap noticeable.
Skill Acquisition Theory (DeKeyser, 2007)
Procedural knowledge develops through deliberate practice with feedback. CF is the feedback loop in this model — without it, practice just rehearses the error.
Lyster & Ranta (1997) and the meta-analyses
The original Montreal classroom study found that prompts (elicitation, clarification requests, metalinguistic cues) led to far more learner uptake than recasts. Twenty years of meta-analyses (Li 2010, Lyster & Saito 2010) broadly confirm this — though recasts still beat doing nothing, and effects depend heavily on context, level, and salience.
Affect matters (Krashen, 1985)
The Affective Filter Hypothesis reminds us that anxiety blocks acquisition. The tone of correction — calm, matter-of-fact, oriented to the work not the person — is part of the technique, not separate from it.
Bring this into your teaching
If you’d like to see corrective feedback in action — calibrated to your level, focused on the things you actually need — try a conversation lesson at ABC English Online.
Explore coursesLyster & Ranta (1997) · Long (1996) · Swain (1985) · Schmidt (1990) · DeKeyser (2007) · Li (2010) · Lyster & Saito (2010) · Krashen (1985) · Corder (1967)