Margaret Cychosz, Ph.D.
Meet the Researcher
Cychosz received her doctorate in linguistics from the University of California, Berkeley, and completed postdoctoral training at the Center for Comparative and Evolutionary Biology of Hearing/University of Maryland Cochlear Implant Center for Excellence. She is currently an assistant professor of linguistics at UCLA where her lab specializes in spoken language development in infants and children, especially those who receive cochlear implants. A 2025 Emerging Research Grants scientist, Cychosz is the recipient of an Elizabeth M. Keithley, Ph.D. Early Stage Investigator Award.
The Research
University of California, Los Angeles
Leveraging automatic speech recognition algorithms to understand how the home listening environment impacts spoken language development among infants with cochlear implants
To develop spoken language, infants must rapidly process thousands of words spoken by caregivers around them each day. This is a daunting task, even for typical hearing infants. It is even harder for infants with cochlear implants as electrical hearing compromises many critical cues for speech perception and language development. The challenges that infants with cochlear implants face have long-term consequences: Starting in early childhood, cochlear implant users perform 1-2 standard deviations below peers with typical hearing on nearly every measure of speech, language, and literacy. My lab investigates how children with hearing loss develop spoken language despite the degraded speech signal that they hear and learn language from.
This project addresses the urgent need to identify predictors of speech-language development for pediatric cochlear implant users in infancy. In accordance with social interaction theories of child development, infants with typical hearing who receive more language input have better speech-language outcomes. We amend this theory for the unique listening experience of cochlear implant users and instead hypothesize that infants with cochlear implants who receive more audible language input in the home will be more advanced on two critical precursors to spoken language: speech production and vocabulary development. Study results may translate into inexpensive, caregiver-centered therapies that could be administered during aural rehabilitation to improve children’s spoken language. Furthermore, the large-scale annotated corpus of child speech may serve as training data for a novel speech classification algorithm that can be used to diagnose speech-language delay for infants with cochlear implants, years before many current testing regimes permit.
Long-term goal: Data from this project will allow us to (1) understand how different learning environments (the home, daycare) impact spoken language development in young children with hearing loss and (2) develop a novel child speech classification algorithm that will identify children with cochlear implants who are at risk of language delay early in infancy. Speech-language delays among children with cochlear implants are common, but are often not diagnosed until the preschool years, or later, when the child can respond to standardized behavioral testing. If children could be identified earlier, ideally using inexpensive methods that are behaviorally-tailored to infants, we could intervene earlier and provide a more systematic precision medicine approach to each child.