Tuning the blueprint: how studies of implicit learning during speaking reveal the information processing components of the production system

ABSTRACT

The “blueprint for the speaker” (Levelt, 1989) claimed that production is carried out by autonomous processing components. We present a framework for how those production components may change with experience and review studies that demonstrate limitations in what can be easily learned by adult speakers. Such limitations have been observed in the acquisition of new phonotactic constraints and in speakers’ ability to alter particular verbs’ biases for particular syntactic structures. We attribute these limitations to the learned componential structure of the production system.

KEYWORDS:

• Language production
• implicit learning
• speech errors
• phonotactics
• verb bias

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 One could propose that sleep is needed to “wall off” learning the experimental rule from English phonotactics. But this view incorrectly predicts that 1st order rules should also require sleep.

2 In American English, the vowel /iew/ as in “few” does not allow some consonants (e.g. /d/) to be onsets, but it does allow them to be codas (e.g. feud)

3 It is true that very fast speech can change the phonetic properties of the output. But these effects are not such that rate can determine whether a particular consonant, say, /f/ is always an onset or always a coda.

Additional information

Funding

This research was supported by National Institutes of Health (NIH) [grant number HD- R01 HD086099].