Speech Away

How practice becomes skill

The feedback that makes you better today can stop you learning

Salmoni, A. W., Schmidt, R. A., & Walter, C. B. (1984). Knowledge of Results and Motor Learning. Psychological Bulletin, 95(3), 355-386. See also Winstein & Schmidt (1990), JEP:LMC 16(4).

Here is the most replicated - and most counter-intuitive - finding in skill learning. Give feedback after every single attempt and performance climbs while the feedback is there, then falls apart the moment it is withdrawn. Cut the feedback to half, fade it, or delay it, and practice looks worse but the skill that survives is real. The metric you can watch improve the fastest is often the one teaching the least.

50% reduced feedback frequency that improved long-term retention (Winstein & Schmidt 1990)
On this page
  1. Why it happens
  2. It holds for speech too
  3. What it means for Speech Away
Figure · Schematic ยท shape illustrativePractice score vs what you keep
0100feedback frequency โ†’performanceDuring practice (with feedback)Later retention (feedback gone)
As feedback gets more frequent, in-practice performance rises (solid) but later retention falls (dashed). The two lines cross - which is why optimizing the live number quietly sabotages the skill. Shape is schematic; the crossover itself is the robust, replicated result.

Why it happens

Constant feedback becomes a crutch. If a coach corrects every rep, you never have to build your own error-detection - the internal sense of "that one was rushed" that you need when the coach is gone. Reduced, faded, or summarized feedback forces you to generate that sense yourself. Delivered instantly - before you can even self-assess - feedback degrades learning outright (Swinnen et al. 1990).

Figure · Reported valuesReduced vs constant feedback - retention test
50% feedback frequency82100% feedback frequency64
Winstein & Schmidt (1990): giving knowledge-of-results on only half the trials produced better retention than giving it every time. Less guidance, more learning.
Data table
Itemretention performance (relative)
50% feedback frequency82
100% feedback frequency64

It holds for speech too

A small speech-specific study (n=39) replicated the pattern: immediate knowledge-of-results gave better acquisition, but delayed feedback gave better retention and better transfer to untrained words. The direction is the same one the motor-learning labs have found since 1984.

What it means for Speech Away

This is why heavy analysis lands after the speech, not during it, and why every fifth take is a silent "cold check" with no feedback shown - we judge your progress on the unaided rep, not the coached one. As you improve, the plan is to fade the report into periodic summaries. The number you should care about is how you do when nothing is helping.