The present study adopted Articulatory Phonology as a theoretical framework to investigate the aticulatory timing of English consonant clusters. Both native and non-native (Mandarin ESL learners) speakers’ performances were of interest. An acoustic approach was taken to explore the consonantal overlap in both native and non-native English speakers’ production. Also investigated in the present study were the factors that influence the overlap between consonants. Thirty-one native Mandarin speakers and eight native Canadian English speakers participated in the study. The thirty-one native Mandarin speakers were divided into three proficiency groups according to a pretest which evaluated Mandarin speakers’ English speaking proficiency. The experiment of the study was a reading task. Participants were instructed to put the words ending with the target consonant clusters in four carrier sentences and read them aloud. In total, 256 tokens (20 clusters×2 vowel environments×4 carrier sentences+12 consonants×2 vowel environments×4 carrier sentences) were recorded for each participant. The duration of each segment in the word was measured in the phonetic software, Praat. Three timing ratios: consonant to cluster, cluster to a pair of individual consonants, and cluster to rime were calculated. Repeated measures ANOVAs showed significant effect for place of articulation and manner of articulation on the articulatory timing of English consonant clusters in the coda position. Meanwhile, voicing feature, as an unexpected factor, was also reported to be an influential factor. More detailed analysis revealed that heterorganic clusters have more overlap than homorganic clusters. Within heterorganic clusters, a tongue tip consonant is more overlapped by a following tongue body consonant than a tongue body consonant is by a following tongue tip consonant. For manner of articulation, stops are found to be more overlapped by a following stop than by a following fricative. Overlapping caused by voicing feature was an unexpected outcome found in the present study. Voiceless consonant clusters have more overlap than voiced clusters. The difference between native and non-native speakers is also of interest. With respect to the amount of overlap, native speakers have more overlap than non-native speakers. Moreover, statistic tests reported a significant effect for proficiency group. From the comparison of mean values of three ratios, the performance of advanced group was close to native speakers. And intermediate and low groups exhibited similar performance. / Graduate

This study was an exploration of the effects of a pedagogical intervention on the development of Japanese university students’ oral performances. In task-based language teaching (TBLT), developing speaking proficiency is a major learning goal. However, research examining the effect of a focus on linguistic form in TBLT is limited. One way to balance communication and attention to linguistic form in TBLT is to add form-focused instruction to the communicative tasks. This study is an exploration of the longitudinal effects of form-focused instruction in a speaking task on the development of speaking proficiency. The current study was conducted for the following research purposes. The first purpose was to explore the longitudinal development of CALF (complexity, accuracy, lexis, and fluency) through form-focused intervention. A one-semester form-focused intervention was conducted to investigate how L2 learners develop or change their linguistic performance as measured by the CALF variables. The second purpose was to explore proceduralization through the 3/2/1 task. The third purpose was to investigate the relationship between communicative adequacy and CALF in the 3/2/1 task. This purpose was addressed by comparing human raters’ perceptions of communicative adequacy with the CALF analyses. The final purpose was to qualitatively investigate what the participants prioritized during their task performances. The participants were 48 first-year Japanese university students attending a private university in eastern Japan. A shortened version of the 4/3/2 task, the 3/2/1 task, was implemented 10 times for 13 weeks in one academic semester. In the 3/2/1 task, students talk about the same topic for 3 minutes, then 2 minutes, and finally 1 minute. The participants were divided into three groups: the comparison group, the teacher-led group, and the teacher and peer group. Two types of form-focused instruction were implemented, teacher-led planning and a peer-check activity. The participants in the comparison group started the 3/2/1 speaking task immediately, those in the teacher-led group read a teacher-model passage with the target formulaic language underlined prior to beginning the 3/2/1 task, and those in the teacher and peer group received a peer-check treatment while doing the 3/2/1 task in addition to teacher-led planning. Listener partners checked to see if the speakers used the target formulaic language during the 3/2/1 task. The target forms were (a) stating opinions (e.g., In my opinion), (b) giving reasons (e.g., It is mainly because…), (c) giving examples (For example…), and (d) expressing possibilities (If…). Speaking data were collected at Time 1 (Week 2), Time 2 (Week 8), and Time 3 (Week 14), transcribed, and analyzed for syntactic complexity, morphosyntactic accuracy, lexical diversity, fluency and communicative adequacy. This result showed that form-focused instruction with the target formulaic language improved the Japanese university students’ speaking fluency such as mean length of run and phonation time ratio. The participants also improved human raters’ perceptions of communicative adequacy over one academic semester. There was a significant and strong positive relationship between utterance fluency and human raters’ evaluation of communicative adequacy. In addition, the peer-check enhanced the learners’ usage of a wider variety of the target formulaic language. The results indicated that including formulaic language instruction can enhance learners’ mean length of run, which is a measure of speaking fluency, while teacher-led planning can help learners notice target forms. The peer-check can pressure learners to use the target forms during the 3/2/1 task and provide feedback so that speakers know what form should be used in the next 3/2/1 task performance. Suggestions for future studies regarding the use of formulaic language in TBLT tasks are proposed. / Teaching & Learning

This study explores, from a test validity perspective, the extent to which the two variables of topic and background knowledge of topic have an effect on spoken performance in language proficiency interviews. It is argued that in assessment contexts where topics are randomly assigned to test takers, it is necessary to demonstrate that topics of tasks and the level of background knowledge that test takers brings to these topics do not exert an undue influence on test results. Otherwise, a validity threat may be introduced to the test. Data were collected from 82 Farsi speakers of English who performed on ten different topics, across three task types. Participants’ background knowledge of topics was elicited using self- report questionnaires while C-tests were used as a measure of general English language proficiency. Four raters assigned scores to spoken performances using rating scales. Semi- structured interviews were carried out with raters, upon completion of the rating process. A mixed- methods strategy of inquiry was adopted where findings from the quantitative analyses of score data (using Multi-Faceted Rasch Measurement, multiple regression and descriptive statistics) were synthesised with the results of the qualitative analyses of rater interviews and test takers’ content of speech in addressing the foci of the study. The study’s main findings showed that the topics used in the study exhibited difficulty measures which were statistically distinct i.e. topics, within a given task type, could not be considered parallel. However, the size of the differences in topic difficulties was too small to have a large practical effect on scores. Participants’ different levels of background knowledge were shown to have a consistent, systematic and statistically significant effect on performance with low levels of background knowledge posing the highest level of challenge for test takers and vice versa. Nevertheless, these statistically significant differences in background knowledge levels failed to translate into practically significant differences, as the size of the differences were too small to have a large impact on performance scores. Results indicated that, compared to general language proficiency which accounted for approximately 60% of the variance in spoken performance scores, background knowledge only explained about 1-3% of the variance. Qualitative analyses of data suggested lack of background knowledge to be associated with topic abandonment, disengagement from topic-related questions, and fewer opportunities for test takers to elaborate on topics. It was also associated with negative affective influence on test takers, particularly lower proficiency individuals. Taken together, the findings have theoretical, methodological and practical implications for second language speaking performance assessment.