Emotional Speech Processing in 3- to 12-Month-Old Infants

• In Event: Poster Session 02
  In Poster Session: PS 02 Topic - Language, Communication

Abstract

Language development takes place in a socio-affective environment (Kuhl, 2007). In daily speech, emotional prosody delivers a speaker’s emotional state and social intention through altered pitch, loudness, and speaking rate. This vocal emotion guides infants’ selective attention towards salient social cues in their environment (Grossmann, 2010), and the varying affective prosody of the same spoken word also provides diverse exemplars that facilitate infants’ word recognition (Singh, 2008). Unlike phonetic development, there is a lack of developmental research on early emotional speech processing. Furthermore, limited studies have yielded conflicting findings on the facilitative role of affective speech in language acquisition (Bhullar, 2008; Singh et al., 2004). To understand the developmental trajectory of processing of emotional prosody in the context of speech and language development, the present study employed a four-way preferential looking paradigm (Shultz & Vouloumanos, 2010) to examine attentional processing of four basic emotions (happy, angry, sad, and neutral) in 46 3-to-12-month-old infants. We presented each emotion for four times in a randomized block design, with a total of 16 trials and an additional pre-test trial. In each block, non-repeating spoken words were used to ensure that infants responded to the target prosodic categories. For each trial, experimental sounds were presented with a static checkerboard image on the screen, and the trial was terminated once the infant looked away for more than 2 seconds or when the sound file ended. The looking time for each trial was calculated by offline video coding (PsyCode, http://psy.ck.sissa.it/) and log-transformed to be the dependent variable in a mixed effect model. The model included participant as the random effect factor, and emotion (neutral, happy, sad, and angry), trial number (1 – 16), and first trial (second trial – last trial were coded as 0; first trial was coded as 1) as the trial-level fixed effect factors. Participant-level fixed factors included age (younger group was coded as 0, mean age = 6 months; older group was coded as 1, mean age = 9 months; cut-off age = 8 months) and sex (female was coded as 0; male was coded as 1). Cross-level interactions of age and emotion, and sex and emotion were also included. The parameters of the fitted model were reported in Table 1. The younger infants (6-month-olds) showed reduced looking time to sad-sounding words relative to the other emotions, which indicates that lower pitch and pitch variations are less attractive to younger listeners. However, the older infants (9-month-olds) responded equally to all emotional categories (Figure 1), suggesting increasing attention to spoken words over the paralinguistic cues with increasing age. Neither group showed a clear preference for happy-sounding words, which challenges the view that infant-directed speech (IDS) consists only of a happy voice that draws infants’ attention (Benders, 2013; Trainor et al., 2000). No sex differences were observed in our data. The current results demonstrate developmentally adaptive changes in infants’ emotional speech listening preferences. Future research is needed to better understand the mechanisms at work in early vocal emotional processing at both behavioral and neural levels.

Authors

• Chieh Kao, University of Minnesota - Twin Cities
  Presenting Author

• Maria D Sera, University of Minnesota
  Non-Presenting Author

• Yang Zhang, University of Minnesota
  Non-Presenting Author