Hard to Say, Hard to See? Speech-in-noise discrimination at different levels of sensorimotor proficiency

Sensorimotor processing strengthens speech-in-noise perception (e.g. Skipper, Devlin & Lametti, 2017). Yet, sensorimotor knowledge can only give its contribution as long as it is accumulated through experience with speech production. Relevantly, full speech motor control is only reached at the end of childhood (e.g. Cheng et al., 2007). Thus, during typical ontogenesis, it should be possible to detect the effects of the gradual emergence of sensorimotor knowledge on speech-in-noise perception. Moreover, in the case of developmental speech production disorders, the more severe consequences of the lack of such information should be observable.
This research tested such hypotheses. The participants (98) were as follows: 32 adults; 25 typically developing preschoolers; 23 typically developing school-aged children; 18 children with Childhood Apraxia of Speech.

The perception experiment assessed the in-noise discrimination (Gaussian white noise, 0dB) of non-sense CV syllables varying along the degree of motor proficiency required by the production of the C: one half of the trials contrasted ‘easy-to-produce’ consonants (plosives and nasals), the other half ‘difficult-to-produce’ consonants (affricates).
The production assessment measured speech sensorimotor skills by means of the Maximum Performance Rate Task. This was run within an auditory masking paradigm (white noise, 70 dB) preventing the participants to perceive their own auditory feedback (thus forcing them to rely on proprioception of the speech movements).

The tasks were analyzed in Mixed-Effect Models; a correlational analysis was run on the two (Kendall’s Tau). As to perception, the adults obtained high scores with all of the trial-types. Conversely, the typically developing children displayed an advantage for the discrimination of the ‘easy-to-produce’ Cs, close to significance in the school-aged group (z=1.8; p=0.07) and significant in the preschoolers (z=2; p= 0.02). Speech-in-noise discrimination was uniformly below chance levels in the Apraxia group.
Furthermore, specific correlational patterns emerged. In the adults, production skills only correlated with discrimination accuracy for the ‘difficult’ consonants (z=3 ; p=0.002) while, in both groups of typically developing children, production skills only correlated with the ability to discriminate the ‘easy’ consonants (z=2 ; p=0.02 and p=0.006). In the Apraxia group, the ability to discriminate the ‘easy’ consonants only correlated with the duration of the therapy targeting speech production (z=4; p<0.0001).
Such results support the hypothesis of a gradual emergence of the sensorimotor contribution during typical ontogenesis and argue for the perceptual relevance of the lack of sensorimotor information in Apraxia. This was the first study to analyze speech-in-noise perception in this disorder.