Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Congenitally blind participants show enhanced prosody sensitivity in garden path sentences.
Poster Session A, Friday, September 12, 11:00 am - 12:30 pm, Field House
Emily Silvano1, Ziqi Chen1, Zaida McClinton1, Marina Bedny1; 1Johns Hopkins University
The study of language processing in blind individuals provides insight into the interface between sensory experience and language. Prior studies find that people born blind recruit “visual” cortices for language and outperform sighted peers on some tasks (e.g., Röder et al., 2002; Bedny et al., 2011). Blind adults are resilient to garden path ambiguities in spoken sentences (Loiotile et al., 2020). One hypothesis is that congenitally blind individuals have superior sentence processing abilities, potentially due to greater cortical resources devoted to language. Another, not mutually exclusive possibility, is that they rely more on prosodic cues to avoid garden-paths. The prior study used human-recorded sentences with subtle, but measurable, prosodic cues to sentence structure and could not distinguish between these alternatives. 23 congenitally blind and 22 sighted age-and-education matched adults performed a self-paced listening task (Speech Experiment) and a self-paced reading task (tactile for blind, visual for sighted; Reading Experiment). In the Speech Experiment, the sentence were constructed from words recorded in isolation and therefore had disrupted prosody. In the Reading Experiment, explicit prosody is unavailable. Participants heard/read sentences segment-by-segment and advanced to the next segment by pressing a button; half were syntactically ambiguous (garden path - GP), half were controls (NGP). Fillers were added to minimize strategic adaptation. Participants answered a yes/no comprehension question after each sentence. Offline accuracy and reaction times (RTs) and online RTs were analyzed using mixed-effects models. Speech Experiment: When prosody was disrupted, blind participants exhibited greater GP costs in offline comprehension accuracy (complexity × group interaction: B=-1.11, SE=0.44, p =.01; main GP effect, p’s<0.001; main group effect: B=-0.05, SE=0.55, p = 0.9) and a marginally larger RTs (group x complexity interaction: B=0.07, SE=0.04, p = .09; main group effect: B=-0.05, SE=0.06, p = 0.3, main GP effect B=-0.14, SE=0.02, p<0.001). These results suggest that distorted prosody has a larger effect on blind listeners, who are sensitive to prosodic cues. Both groups also showed a significant GP effect at the critical segment in the online measure (B=-0.01, SE=0.008, p=0.04). Blind participants overall listened faster than sighted participants (p’s<0.001). Thus, blind participants performed similarly on comprehension questions despite listening nearly twice as fast, possibly due to experience with rapid screen-reader speech, an assistive technology. Reading Experiment: GP ambiguity was associated with reduced accuracy across groups (main GP effect: B=1.28, SE=0.32, p<0.001). Blind participants performed similarly to sighted in offline comprehension accuracy (main group effect: B=-0.26, SE=0.42, p=0.5) and showed a marginally smaller GP cost (group x complexity interaction: B=-0.49, SE=0.26, p=0.06) but were somewhat slower to answer the comprehension questions (main group effect B=-0.1, SE=0.05, p=0.01). Both groups showed a significant GP effect on critical segment in the online reading-time measure (B=-0.08, SE=0.01, p<0.001). Together, these findings suggest that when prosodic cues are available, blind individuals outperform sighted listeners, but blind people are also more negatively affected by disrupted prosody. We hypothesize that blind individuals rely heavily on prosodic information during sentence parsing, likely due to habitual reduced access to visual referential context.
Topic Areas: Language Development/Acquisition, Syntax and Combinatorial Semantics