Listeners beware: Speech production may be bad for learning speech sounds

Abstract

Spoken language requires individuals to both perceive and produce speech. Because both processes access lexical and sublexical representations, it is commonly assumed that perception and production involve cooperative processes. However, few studies have directly examined the nature of the relationship between the two modalities, particularly how producing speech influences speech perception. In a series of experiments, we examine the counter-intuitive finding that learning perceptual representations can be disrupted by producing tokens during training. We investigate whether this disruption can be alleviated by prior experience with the speech sounds, and whether the cause of the disruption is production of the particular sound being learned, or is a more general conflict between the production system and the system that develops new perceptual representations. Our results paint a more competitive relationship between perception and production than might be assumed and suggest that both demands inherent to production and cognitive demands modulate this relationship.

Perceptual statistical learning over one week in child speech production

“…When the restrictions in production and perception conflicted, participants' errors only corresponded to what they produced, suggesting that the act of production was more relevant to learning the co-occurrence constraints. Baese-Berk and Samuel (2016) asked Spanish-speaking participants to learn the / a/ - / ∫a/ contrast from Basque. Some participants completed an initial perceptual learning phase, while other participants produced the contrast concurrently to the perceptual training.…”

Reverse production effect: children recognize novel words better when they are heard rather than produced

“…But when the novel words comprised low phonotactic probabilities, the effect was attenuated and there was no production advantage in recall. Baese‐Berk and Samuel () found that production disrupted adults’ learning of a novel fricative contrast compared to a heard condition. In Kaushanskaya and Yoo () and Baese‐Berk and Samuel (), adults did not always produce target‐like productions when they repeated the stimuli aloud.…”

“…Baese‐Berk and Samuel () found that production disrupted adults’ learning of a novel fricative contrast compared to a heard condition. In Kaushanskaya and Yoo () and Baese‐Berk and Samuel (), adults did not always produce target‐like productions when they repeated the stimuli aloud. Therefore, part of the decreased learning in the overt production condition may have stemmed from the mismatch between the auditory target and participants’ non‐target‐like productions.…”

“…Interestingly, Baese‐Berk and Samuel also found a reduction in learning of a novel fricative contrast when the production system was engaged in another task—participants named letters on the screen during training. Baese‐Berk and Samuel suggest that part of the disruption in learning may also stem from “cognitive mechanisms such as selective attention and task switching” (Baese‐Berk & Samuel, , p. 31).…”

In search of a learning model

“…This point is particularly intriguing. Experimental studies with adults have revealed a disruptive effect of production on the learning of both novel lexical items and non‐native sound contrasts (Baese‐Berk & Samuel, ). Similarly, perhaps, Zamuner, Morin‐Lessard, Strahm, and Page () found that in learning non‐words children are actually hampered by production.…”