Theory 037/Mind & Behaviour /Developmental cognition
Language Learning Is Gated by Echo Memory
Vocabulary is predicted from how many words a child hears. The bottle neck may be how many sounds can be held at once.
Vocabulary acquisition is gated by phonological working memory capacity rather than by exposure, so children with the same input can differ several-fold in vocabulary depending on their echo memory span.
01The evidence
Nonword repetition, a measure of phonological storage, predicts vocabulary size better than input quantity in many studies.
02The tension
Input-based models predict vocabulary from exposure. That predicts similarity among children in the same environment, which is not observed.
03The mechanism
A new word must be held in phonological store long enough to be consolidated. Capacity sets how long and how many, so it sets the rate at which words can be added.
That makes span the rate-limiting variable, and predicts that vocabulary growth should track span rather than input once input is above a minimum.
04What this is not
Input models predict from exposure. This predicts from storage, which explains variance within shared environments.
05Predictions
- P1Vocabulary should track span better than input.
- P2Training span should accelerate vocabulary growth.
- P3Children with equal input should diverge by span.
06Refutation
Vocabulary tracking input at controlled span would restore the exposure model.
07What it changes
It would change early language interventions.