Research (R)
Kendra Marks, AuD
Research Coordinator
Northwestern University
Evanston, Illinois, United States
Pamela Souza, PhD
Professor and Chair
Northwestern University
Evanston, Illinois, United States
Lakshmi Nanduri, BS
Northwestern University
Palatine, Illinois, United States
Marisha Speights Atkins, PhD
Assistant Professor
Northwestern University
Evanston, Illinois, United States
Rationale: Older adults frequently report increased difficulty understanding speech, and this can be especially problematic if the speaker is a young child. In clinic, it is not unusual for patients to report difficulty understanding their young grandchildren as a motivator for seeking intervention. Despite this, most research on speech recognition relies on adult speech materials. In this study, we examine listener and talker factors related to intelligibility of child talkers.
Design: Adults aged 18-80 years with a range of hearing thresholds are recruited for this study. Speech recognition is measured for speech drawn from the Speech Exemplars and Evaluation Database (SEED) (Speights Atkins et al. 2020). Six different child talkers (3 male, 3 female; mean age 4.4 years) produced a set of single-syllable words. Words are played to the listener after a carrier prompt and were presented at an audible and comfortable level. Listeners respond by repeating the presented word. Words are scored as either correct or incorrect, and errors are recorded for later analysis. Vowel space is quantified using Linear Predictive Coding-based measurements of formant frequencies. Recognition scores for each talker are related to talker acoustic properties, specifically to vowel space.
Results: Analysis reveal that listener age, hearing status and acoustics of the child talker contribute to speech intelligibility. As age of the listener increases, overall speech recognition (i.e. mean % correct across all talkers) decreases. Results show variable intelligibility across talkers. Vowel space based on the frequencies of the first and second formants also vary across talkers.
Conclusion: Older adult listeners have increased difficulty recognizing child talkers compared to younger adult listeners. Individual talker intelligibility varies, with some children being easier to understand than others. Improving our understanding of the factors which contribute to older adult recognition of child speech may improve counseling approaches regarding how best to communicate with young children.