A pragmatic assessment of intelligibility and comprehensibility

This dissertation is a study of how different types of intelligibility and comprehensibility assessments correlate with each other as well as measures of vowel space and performances in a referential communication task. This study elicited speech samples from four English speakers (three of these speakers were non-native speakers of English whose first language was Mandarin) of varying proficiency as judged by independent experts in the field. These speakers were recorded reading a list of 16 vowels in a constant phonetic frame (/h/ V/d/) a list of twenty sentences, a narrative passage, and describing ten abstract figures. These recordings were played to 97 undergraduate students who were asked to identify the vowels, write the last word of each sentence, answer true/false statements regarding the narrative passage, and order an out of sequence set of abstract figures based on a spoken description by each of the four speakers. The undergraduate students were also asked to judge each of the four speakers on five subjective Likert-type rating scales. These scales asked the students to make judgments regarding how easy the speaker was to understand, rate, clarity, and enthusiasm of the speech and how good (or bad) the speaker would be as a teacher of a college course like history or psychology. The measure of vowel space as used in this study treated the first and second formants of the three corner vowels /i/, /a/, and /u/ as x-y coordinates in a Cartesian plane. The area of the triangle created by these three points was calculated and then compared to each of the other measures by means of factor analysis. The results showed that the vowel space of speakers used in this study was significantly correlated with each of the other measures used in this study (with the exception of the Likert-type scales). The measure of vowel space correlated significantly with respondents' performances on the referential picture matching task, and both tasks loaded highly on the same factor indicating a stronger relationship between these types of measures than had heretofore been supposed.