Date of Award
Master of Arts
Chang, Soo Jung
AN ABSTRACT OF THE THESIS OF Ousmane Sawadogo, for the Master of Arts degree in Applied Linguistics and TESOL, presented on May 4, 2015, at Southern Illinois University Carbondale. TITLE: PERCEPTION OF /q/ IN THE ARABIC /q/-/k/ CONTRAST BY NATIVE SPEAKERS OF AMERICAN ENGLISH: A DISCRIMINATION TASK MAJOR PROFESSOR: Dr. Soo Jung Chang Studies on speech perception have suggested that non-native sound perception is influenced by the listener’s native language. Non-native sound contrast perception depends on a given sound’s similarity or dissimilarity to the listener’s equivalent native language sound. The Perceptual Assimilation Model (PAM) posits that it is difficult to distinguish non-native sounds when they are perceived as very similar to native sounds and are thus assimilated to a native sound category, but identification is easier when the non-native is sound is dissimilar to a native sound (Best, 1994a). The present study investigated whether native speakers of American English would display very good discrimination of /q/ in the Arabic /q/-/k/ contrast as predicted by the PAM. The Speech Learning Model (SLM) posits that non-native perception is position-sensitive and hypothesizes that the listener’s perception of non-native sounds would vary from one position to another (Flege, 1995). The current study also aimed to investigate whether the discrimination of the Arabic /q/-/k/ contrast would be position-sensitive. The current study also investigated the effect of the vocalic context on the discrimination of /q/. Participants consisted of 22 (6 male and 16 female) native speakers of American English who were students or faculty members at Southern Illinois University. Their ages ranged between 19 and over 50. The data were collected through an online AXB discrimination task survey. Target sounds were represented in 108 pseudowords so that the sounds could be contrasted in minimal pairs. The environments were word-initially followed by /i/, /u/, and /a/; word-medially, between two instances of /i/, two instances of /u/, and two instances of /a/; and word-finally, preceded by /i/, /u/, and /a/. Two pseudoword pairs were selected for each contrast. Four AXB combinations (AAB, ABB, BAA, and BBA) were generated for each of the nine contrasts, which resulted in a total of 36 stimuli. The participants were requested to click on a button to listen to the recordings of these word pairs and check the right answer. The findings were consistent with predictions made by PAM that native speakers of American English would have a very good discrimination of /q/ in the Arabic /q/-/k/ contrast. The results suggested that the uncategorized versus categorized (UC) type could also be of excellent discriminability. SLM was not totally supported because the differences were not statistically significant. However, the data indicated that some positions resulted in better discrimination scores than other positions and that certain vowels likewise resulted in better vocalic discrimination scores. Keywords: Discrimination task, non-native sound perception, Arabic /q/-/k/ contrast, position-sensitive, context-sensitive, vocalic context, UC type assimilation, Speech Leaning Model (SLM), Perceptual Assimilation Model (PAM).
This thesis is Open Access and may be downloaded by anyone.