Date of Award

Summer 2021

Document Type

Open Access Thesis

First Advisor

Daniel Fogerty


Purpose: The purpose of this error analysis was to analyze the lexical properties of misperceptions made when listening to speech in multitalker babble.

Methods: Twenty young adults with normal hearing (YNH), 20 older adults with normal hearing (ONH), and 22 older adults with hearing impairment (OHI) completed a speech-in-babble task. Participants were asked to repeat the final word in 25 high and 25 low context sentences. On each trial, participants either responded with the correct target word, a misperception error, or skipped the trial response. Misperceptions were compiled and analyzed according to their lexical properties.

Results: Results of this study showed that neighborhood density, the proportion of misperceptions within the same phonological neighborhood as the target, and phonologic and semantic similarity measures all demonstrated significant effects of context. Specifically, the older adult groups (ONH and OHI) responded with words in significantly smaller phonological neighborhoods in low context sentences compared to the target, phonological similarity was found to be greater in low context sentences than high, and semantic similarity was found to closer to the target in high context sentences. In addition, group differences were observed in that the YNH group was more likely to guess, resulting in more misperceptions, while the OHI group was more likely to skip response opportunities. Similarly, group differences were noted where older adults produced misperceptions from smaller phonological neighborhoods than the target in low context sentences and the ONH group, specifically, produced more errors that were less phonologically similar to the target compared to the YNH group in high context sentences.

Conclusions: All groups performed similarly in regard to overall accuracy in recognizing speech in multitalker babble. However, systematic differences were noted between participant groups in the types of misperceptions made across sentence contexts. These patterns suggest that age, hearing loss, and context affect the lexical properties of misperceptions made when listening to speech in multitalker babble.