Abstract
Traditionally, emotion recognition research has primarily used pictures and videos, while audio test materials are not always readily available or are not of good quality, which may be particularly important for studies with hearing-impaired listeners. Here we present a vocal emotion recognition test with pseudospeech productions from multiple speakers expressing three core emotions (happy, angry, and sad): the EmoHI test. The high sound quality recordings make the test suitable for use with populations of children and adults with normal or impaired hearing. Here we present normative data for vocal emotion recognition development in normal-hearing (NH) school-age children using the EmoHI test. Furthermore, we investigated cross-language effects by testing NH Dutch and English children, and the suitability of the EmoHI test for hearing-impaired populations, specifically for prelingually deaf Dutch children with cochlear implants (CIs). Our results show that NH children’s performance improved significantly with age from the youngest age group onwards (4–6 years: 48.9%, on average). However, NH children’s performance did not reach adult-like values (adults: 94.1%) even for the oldest age group tested (10–12 years: 81.1%). Additionally, the effect of age on NH children’s development did not differ across languages. All except one CI child performed at or above chance-level showing the suitability of the EmoHI test. In addition, seven out of 14 CI children performed within the NH age-appropriate range, and nine out of 14 CI children did so when performance was adjusted for hearing age, measured from their age at CI implantation. However, CI children showed great variability in their performance, ranging from ceiling (97.2%) to below chance-level performance (27.8%), which could not be explained by chronological age alone. The strong and consistent development in performance with age, the lack of significant differences across the tested languages for NH children, and the above-chance performance of most CI children affirm the usability and versatility of the EmoHI test.
Highlights
Development of emotion recognition in children has been studied extensively using visual stimuli, such as pictures or sketches of facial expressions (e.g., Rodger et al, 2015), or audiovisual materials (e.g., Nelson & Russell, 2011), and in some clinical groups, such as autistic children (e.g., Harms, Martin & Wallace, 2010)
Model comparison showed that the full model with random intercepts per participant and per stimulus was significantly better than the full models with only random intercepts per participant [χ 2(1) = 393, p < 0.001] or only random intercepts per stimulus [χ 2(1) = 51.9, p < 0.001]
The results of the Dunnett’s tests showed that the accuracy scores of Dutch NH children of all tested age groups differed from Dutch NH adults (4–6 years difference = −0.47, p < 0.001; 6–8 years difference = −0.31, p < 0.001; 8–10 years difference = −0.19, p < 0.001; 10–12 years difference = −0.15, p < 0.001), and the accuracy scores of English NH children of all tested age groups differed from English NH adults (4–6 years difference = −0.43, p < 0.001; 6–8 years difference = −0.27, p < 0.001; 8–10 years difference = −0.20, p < 0.001; 10–12 years difference = −0.12, p < 0.01)
Summary
Development of emotion recognition in children has been studied extensively using visual stimuli, such as pictures or sketches of facial expressions (e.g., Rodger et al, 2015), or audiovisual materials (e.g., Nelson & Russell, 2011), and in some clinical groups, such as autistic children (e.g., Harms, Martin & Wallace, 2010). Not much is known about the development of vocal emotion recognition, even in typically developing children (Scherer, 1986; Sauter, Panattoni & Happé, 2013). Children and adults both are better at facial emotion recognition than vocal emotion recognition (Nelson & Russell, 2011; Chronaki et al, 2015). All of these observations combined indicate that the formation of robust representations for vocal emotions is highly complex and possibly a long-lasting process even in typically developing children
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call