Explaining coherence in coherence masking protection for adults and children

Abstract

Coherence masking protection (CMP) is the phenomenon in which a low-frequency target (typically a first formant) is labeled accurately in poorer signal-to-noise levels when combined with a high-frequency cosignal, rather than presented alone. An earlier study by the authors revealed greater CMP for children than adults, with more resistance to disruptions in harmonicity across spectral components [Nittrouer and Tarr (2011). Atten. Percept. Psychophys. 73, 2606Psychophys. 73, -2623. That finding was interpreted as demonstrating that children are obliged to process speech signals as broad spectral patterns, regardless of the harmonic structure of the spectral components. The current study tested three alternative, auditory explanations for the observed coherence of target þ cosignal: (1) unique spectral shapes of target þ cosignal support labeling, (2) periodicity of target þ cosignal promotes coherence, and (3) temporal synchrony across target þ cosignal reinforces temporal expectancies. Adults, eight-year-olds, and five-year-olds labeled stimuli in five conditions: F1 only and F1 þ a constant cosignal (both used previously) were benchmarks for comparing thresholds for F1 þ 3 new cosignals. Children again showed greater CMP than adults, but none of the three hypotheses could explain their CMP. It was again concluded that children are obliged to recognize speech signals as broad spectral patterns.

Developmental Effects in Children’s Ability to Benefit From F0 Differences Between Target and Masker Speech

“…This could suggest that harmonicity is a not salient for children when integrating spectral components of speech. This finding has been interpreted as showing that children pay attention to broad spectral structure instead of the fine spectral detail present in F0 (Nittrouer & Miller, 1997;Nittrouer & Tarr, 2011;Tarr & Nittrouer, 2013).…”

“…Studies of coherence masking protection (CMP) in children and adults have investigated whether children are obliged to process speech signals as broad spectral patterns (Gordon 1997;Nittrouer & Tarr 2011;Tarr & Nittrouer 2013). These studies involved labeling a phonetically-important low frequency formant (F1) presented in noise either alone or in combination with a higher frequency cosignal (a second and third formant, F2 and F3) which does not provide any additional vowel information and is outside the critical band of the target.…”