Early Language in Blind, Deaf/Hard-of-Hearing, and Typically-Developing Infants

• In Event: The Role of Language in Concept Acquisition in Typical and Atypical Populations

Abstract

Infants learn about the world through direct perceptual experience and linguistic input. Infants who are blind, deaf/hard of hearing (DHH), and typically-developing infants have different access to sensory input. Is this accompanied by differences in spoken language experiences overall or within perceptually-centric adjectives? We use naturalistic recordings to query this in DHH infants with aided hearing, blind infants, and typically-developing controls.

Analysis 1: We measured language in the home environment using LENA daylong audio-recordings from blind infants and age-matched controls (n=12), and DHH infants with cochlear implants and age- and hearing-age-matched controls (n=33; see Table for details). We analyzed LENA's Adult-Word-Count and Child-Vocalization-Count estimates based on this sample of ~25,000 minutes of audio across these 28 children. Adult-Word-Count did not differ across groups (ps> .05 by Kruskal-Wallis Test; Fig.1). Child-Vocalization-Count did not differ between blind infants and the other groups (p>.05 Bonferroni-corrected Dunn test), while DHH children produced significantly more vocalizations than hearing-age-matched controls (MDHH=2674(1559) vs. MTD-HA=1035(498); p<.05 Bonferroni-corrected Dunn test), but not age-matched controls (MTD-A=1982(665); p>.05 Bonferroni-corrected Dunn test).

Analysis 2: We used 30-minute transcribed video-recordings of parent-child free play interaction from 13-month-old DHH infants with hearing aids or cochlear implants and age-matched typically-developing controls (n=36; see Table) from the Ambrose-Moeller corpus (Ambrose, 2016). This analysis focused on sense-specific adjectives to probe whether parents provide linguistic descriptions of perceptual features that their children have limited access to (e.g. adjectives like ‘loud’ for DHH infants). Adjectives were tagged for sensory modality (touch, hearing, smell, taste, vision, and amodal). For each child, we calculated total adjective tokens, the proportion that are sensory, and the proportion that are sound-specific. In ~1,000 total minutes of video recordings across 36 children, there were 1428 adjective tokens (587 types), 298 (160 types) of which were sensory (e.g., loud, blue, smelly). DHH and typically-developing infants’ parents used statistically equivalent quantities of adjectives across these measures: (Total adjectives: MDHH=38(24) vs. MTD=40(22); Proportion of Sensory Adjectives: MDHH=23%(18%) vs. MTD=18%(9%); Proportion of Sound Adjectives: (MDHH=.3%(.9%) vs. MTD=.7%(2%)).

Across analyses, despite differences in sensory experience, blind and DHH infants’ spoken language input is highly comparable to their age-matched typically-developing peers, amidst large variability. This held both for overall Adult-Word-Counts and adjective input, and for (relatively rare) sensory adjectives. While not all perceptual experiences are accessible to infants with sensory impairment, these results suggest that parents do not appear to tailor their input to their child’s sensory abilities. Indeed, the only reliable difference was between DHH infants’ vocalization counts relative to hearing-age matched (thus, much younger) infants, suggesting that age plays a much stronger role in early vocalizations than sensory experience. These results converge with work showing that blind individuals and Deaf individuals (Landau & Gleitman, 1985; Kim et al., 2019; Rosen, 2007) have rich concepts of perceptually-inaccessible (or less accessible) domains, suggesting that language input itself provides ready support for learning.

Authors

• Erin Campbell, Duke University
  Presenting Author

• Martin Zettersten, Princeton University
  Non-Presenting Author

• Molly Lewis, Stanford University
  Non-Presenting Author

• Elika Bergelson, Duke Univeristy
  Non-Presenting Author