Page 42 - Spring2019

Hearing in the Classroom
subtle F0 differences between target and masker speech. These data highlight the importance of auditory experience and maturational effects in learning how to segregate target from masker speech.
In addition to relying on acoustic voice differences between talkers when listening in complex auditory environments, adults with normal hearing take advantage of the differ- ences in signals arriving at the two ears. These differences provide critical information regarding the location of sound sources in space, which, in turn, facilitates segregation of target and masker speech (e.g., Bregman, 1990; Freyman et al., 2001). The binaural benefit associated with separating the target and masker on the horizontal plane is often called spatial release from masking (SRM). In the laboratory, SRM is typically estimated by computing the difference in speech recognition performance between two conditions: the co- located condition, in which the target and masker stimuli are presented from the same location in space, and the spatial separation condition, in which the target and masker stimuli are perceived as originating from different locations on the horizontal plane. For adults with normal hearing, SRM is substantially larger for speech recognition in a masker com- posed of one or two streams of speech than in a noise masker (reviewed by Bronkhorst, 2000).
Several studies have evaluated SRM in young children and demonstrate a robust benefit of spatially separating the tar- get and masker speech (e.g., Litovsky, 2005; Yuen and Yuan, 2014). Results are mixed, however, regarding the time course of development for SRM. Although Litovsky (2005) observed adult-like SRM in 3-year-old children, other studies have re- ported a smaller SRM for children compared with adults, a child/adult difference that remains until adolescence (e.g., Yuen and Yuan, 2014; Corbin et al., 2017). In a recent study, Corbin et al. (2017) assessed sentence recognition for chil- dren (8-10 years) and adults (18-30 years) tested in a noise masker and in a two-talker masker. Target sentences were always presented from a speaker directly in front of the lis- tener, and the masker was either presented from the front (co-located) or from 90° to the side (separated). Although a comparable SRM was observed between children and adults in the noise masker, the SRM was smaller for children than adults in the two-talker masker. In other words, children benefitted from binaural difference cues less than adults in the speech masker. This is important from a functional per- spective because it means that not only are children more detrimentally affected by background speech, but they are
also less able to use spatial cues to overcome the masking associated with speech.
In addition to sound source segregation, auditory scene analy- sis depends on the ability to allocate and focus attention on the target. Findings from studies using behavioral shadowing procedures provide indirect evidence that selective auditory attention remains immature well into the school-age years (e.g., Doyle, 1973; Wightman and Kistler, 2005). In a typical shadowing task, listeners are asked to repeat speech presented to one ear while ignoring speech or other sounds presented to the opposite ear. Children perform more poorly than adults on these tasks, with age-related improvements observed into the adolescent years (e.g., Doyle, 1973; Wightman and Kistler, 2005). Moreover, children’s incorrect responses tend to be in- trusions from speech presented to the ear they are supposed to disregard. For example, Wightman and Kistler (2005) asked children (4-16 years) and adults (20-30 years) to attend to tar- get speech presented to the right ear while disregarding mask- er speech presented to both the right and left ears. Most of the incorrect responses made by adults and children older than 13 years of age were due confusions with the masker speech that was presented to the same ear as the target speech. In contrast, incorrect responses made by the youngest children (4-5 years) tested were often the result of confusions with the masker speech presented to the opposite ear as the target speech. This result is interpreted as showing that young children do not re- liably focus their attention on the target even in the absence of energetic masking.
Although behavioral data suggest that selective auditory at- tention remains immature throughout most of childhood, a key limitation of existing behavioral paradigms is that we cannot be certain to what a child is or is not attending. Poor performance on a shadowing task might reflect a failure of selective attention to the target but is also consistent with an inability to segregate the two streams of speech (reviewed by Sussman, 2017). This issue is further complicated by the bi- directional relationship between segregation and attention; attention influences the formation of auditory streams (e.g., Shamma et al., 2011). Researchers have begun to disentangle the independent effects of selective auditory attention by measuring auditory event-related brain potentials (ERPs) to both attended and unattended sounds (e.g., Sussman and Steinschneider, 2009; Karns et al., 2015). The pattern of re- sults observed across studies indicates that adult-like ERPs associated with selective auditory attention do not emerge until sometime after 10 years of age, consistent with the time
40 | Acoustics Today | Spring 2019