Dr. Judy Harvey
Professor Alicia Davis
Date of this Version
Javernick, Q. (2021). The Evaluation of Comprehensibility in Non-Native English Speakers [Unpublished master's thesis]. University of Nebraska-Lincoln
As of 2020, accent modification instruction was added to a speech-language pathologist's Scope of Practice. This elective service is for individuals who wish to change or modify their speech patterns to increase speech clarity. In developing assessment and instructional programs geared towards accent modification instruction, considerations must be made for the listener and the speaker. A range of accent modification assessment materials and instruction programs currently exist. The purpose of this study was to determine if a simplified Likert-type rating scale, ranging from "not comprehensible" to "comprehensibility unaffected," could serve as a quick tool to measure a non-native English speaker's degree of comprehensibility. A series of untrained listeners evaluated ten connected speech samples of Spanish-influenced English across three simplified Likert-type rating scales. The following research questions were formed (1) What is the inter-rater reliability for a 3-, 5-, and 7-point Likert-type rating scale? (2) How consistent are the ratings across the 3-, 5-, or 7-point Likert-type rating scale? (3) What is the relationship between the comprehensibility ratings and the number of affected speech features (i.e., vowel distortions, dropped endings, and voice/voiceless substitutions)? (4) What features did the raters report as affecting speaker comprehensibility? The research questions were answered by running two analyses, Interclass Correlation Coefficient and Pearson's correlation, both measurements of consistency and interrater reliability. The results indicate that both a 3-point and 5-point Likert-type rating scale were reliable in measuring comprehensibility. Additional analysis revealed that comprehensibility ratings correlated with the frequency of speech features (dropped endings, distorted dentals, distorted vowels, and total speech features) but not with rate of speech. Further, raters generally expressed terminology surrounding the generalized definition of accentedness rather than utilizing verbiage reflecting specific speech features in describing the rating process. The results support that perceptual judgments regarding the comprehensibility of Spanish-influenced English align with identified speech features previously associated with increased accentedness.
Advisor: Judith Harvey