Recasts vs prompts, the corrective feedback you give without knowing it.

Roy Lyster sat in French immersion classrooms in Montreal and counted every correction the teachers gave for a year. Fifty-five percent of those corrections came in one form. It was also the form that produced the least learning.

The Lyster and Ranta data

Lyster and Ranta's 1997 study (Corrective Feedback and Learner Uptake, Studies in Second Language Acquisition) is the paper that gave the field its categories. They watched grade four and five immersion classrooms, transcribed everything, and sorted teacher corrections into six buckets.

The buckets that mattered:

Recast

The teacher reformulates the student's error inside their own next turn. Student says Yo va al baño. Teacher says Ah, tú vas al baño, perfecto, ve. The correction is folded into a continuation. The conversation flows.

Prompt

The teacher signals there is an error and pushes the student to self-correct. Four flavors. Elicitation: Yo…? with the rising tone that says finish it. Clarification request: ¿Cómo? Metalinguistic clue: Cuidado, esa es la forma de él. Repetition with intonation: ¿Yo va? The student has to fix it.

What the recordings showed

Recasts were by far the most common feedback move. They were also the easiest one for students to miss. Lyster and Ranta tracked uptake (any student response that engaged with the feedback) and repair (a student response that fixed the error). Recasts produced uptake 31 percent of the time. The prompt categories produced uptake between 78 and 100 percent of the time, with repair rates roughly four times higher than recasts.

Read that again. When the teacher reformulated, the kid kept talking and rarely registered the correction. When the teacher pushed back, the kid almost always fixed the error on the spot.

The 2004 study that closed the deal

Lyster published a follow-up experiment in 2004 (Differential Effects of Prompts and Recasts in Form-Focused Instruction, Studies in Second Language Acquisition). Three intact French immersion classes. Same instructor input on French grammatical gender. One class got recast feedback. One got prompt feedback. One got no feedback, only the input. He tested before, immediately after, and eight weeks later.

The prompt group beat both other groups on every measure and held the advantage at eight weeks. The recast group was not much better than the no-feedback group. The conclusion was not that recasts are useless. The conclusion was that recasts work best where students are already noticing form, and prompts work better where they are not.

Prompts had larger effects than recasts overall, and the difference grew on freer outcome measures and at delayed post-tests.

Why teachers default to recasts

Three reasons, none of them stupid.

First, recasts are polite. You do not interrupt the kid. You do not stop the conversation to correct the verb ending. You slide the corrected form in and let them carry on. In a student-centered classroom that values risk-taking, the recast feels like the right move. It probably is, for a beginner who is still building courage.

Second, recasts are fast. A prompt requires the student to stop, reconsider, and self-repair. That takes time. With twenty-five students and forty-three minutes, recasts let you move through more turns.

Third, recasts are the default move in fluent adult conversation. We do this with toddlers learning their first language. We do this with friends who mispronounce a word. Reformulating without making a deal of it is the socially smooth thing to do. Classroom feedback is not adult conversation, but it inherits the habit.

The counterbalanced approach

Lyster's 2007 book (Learning and Teaching Languages Through Content: A Counterbalanced Approach, John Benjamins) argues that immersion and content-based programs need to deliberately swing the pendulum in the direction the lesson is not naturally going. A meaning-heavy lesson needs form-focused feedback to keep grammar accuracy from drifting. A grammar-heavy lesson needs meaning-focused tasks to keep the form connected to communication.

For most Spanish teachers in the US, the lesson is naturally meaning-heavy and the feedback is naturally recast-heavy. The counterbalance is to be more deliberate with prompts when a rule-based feature is the lesson's target. Not all the time. When the target form is on the table.

When recasts are still the right move

Lyster and Saito's 2010 meta-analysis (Oral Feedback in Classroom SLA, Studies in Second Language Acquisition) ran the numbers across fifteen years of feedback research. Both prompts and recasts produced significant gains, with prompts having a larger effect overall. The size of the recast effect depended on three things: how salient the recast was, how much the student noticed, and whether the structure was rule based or lexical.

For lexical errors (wrong vocabulary, wrong collocation), a recast that emphasizes the correct word is often enough. For rule-based features (verb agreement, gender, preterite versus imperfect, ser versus estar), the student needs to do the work. Prompts make them do the work.

What this means for your Monday

Pick the two features your Spanish students get wrong most often. For my Spanish 3 class it is ser versus estar and preterite versus imperfect. For your class it might be different.

For those two features only, commit to prompting instead of recasting for one week. When a student says Yo soy en la casa, do not say Ah, tú estás en la casa. Say ¿Yo soy? with the eyebrow. Wait. Let them fix it. If they cannot, give one metalinguistic clue. If they still cannot, then recast, but make a note that the form is not yet learnable for that student.

For everything else, recast as you normally would. You will not have time to prompt every error, and the research does not say you should. The research says that for the rule-based features you actually want students to acquire, the polite move is not the effective one.