Electrophysiological Responses to Audiovisual Speech in Hearing and Deaf Children With Cochlear Implants

• In Event: Poster Session 09
  In Poster Session: PS 09 Topic - Language, Communication

Abstract

Naturalistic speech occurs under conditions in which access to both auditory (i.e. acoustic) and visual (i.e. facial) information is available. A large body of behavioral research shows benefits of A/V over A-only presentations in word recognition, lexical decision and sentence processing (Bernstein et al., 2004). Speech understanding in noise is further enhanced with audiovisual presentation (Ma et al., 2009). Electrophysiological studies have sought to better understand the behavioral enhancements of A/V presented speech. Current work has shown modality modulations of early ERP components (N1/P2) and later effects such as the N400 (Basirat et al., 2018; Brunellière et al., 2020; Pilling, 2009). These components are consistently responsive to speech, with amplitudes and latencies that are modulated by the presence of visual information. On balance, these data indicate that there is an attenuation of the ERP components associated with speech processing under condition of A/V versus A-only presentations. Developmental changes in these effects have also been reported, with younger children showing less influence of visual speech cues on auditory ERP components compared to older children (Knowland et al., 2014). We are investigating the contributions of AV speech processing in a population with known sensory deficits, congenitally deaf children who have received a cochlear implant (CI). We have collected data from deaf children with CI (n = 28, mean age = 83 mos) and typical hearing children (n = 18, mean age = 75 mos) in a word-picture priming paradigm in which audio-visual presented word primes preceded picture targets. Prior reporting of N-400 ERP effects to picture targets indicates significance group differences, suggesting difference in semantic integration in normally hearing and deaf children with CIs (Pierotti et al., 2020). In the current work we are exploring the effects of processing of the spoken Audio-Visual primes in this data set. We predict that relative to normally hearing children, children who rely upon CI’s will show differences in visual sensory (P1) and early auditory (N1/P2) components. However inspection of these data indicate that the typically a-priori defined windows of interest for (N1/P1/P2) components may mask more subtle group differences given the variability observed in this population. We are actively exploring a data-driven approach to better characterize these data. Specifically we will make use of Permutation Tests for Time Series Data (Voeten, 2019) to help identify the timepoints where significance of an effect begins and ends in this multi-channel EEG data. These analyses will be used to evaluate regional effects of audiovisual speech processing. Statistical analysis of these data will make use of Linear Mixed-Effects (LMER) Modeling to quantify effects based on chronological age and age of implantation. LMER modeling is particularly advantageous for designs such as the current one, where there are multiple, potentially collinear, continuous variables. These data will help enrich our understanding attentional and perceptual factors of speech processing in children with cochlear implants.

Authors

• Elizabeth Pierotti, University of California - Davis
  Presenting Author

• Sharon Coffey-Corina, University of California, Davis
  Non-Presenting Author

• Tristan D Schaefer, University of California - Davis
  Non-Presenting Author

• David P. Corina, University of California - Davis
  Non-Presenting Author