A comparison of learner characteristics, beliefs, and usage of ASR-CALL systems

Wall Street English3 has built online activities that allow students to record phrases and receive word-level Automatic Speech Recognition (ASR) driven pronunciation feedback. Students in language centres in China, Vietnam, Saudi Arabia, and Italy (N=2,867) used ASR-Computer Assisted Language Learning (CALL) activities, and some (N=482) completed a questionnaire. A high number of students reported that ASR-CALL activities helped them to improve their pronunciation. However, the study found remarkable differences in usage of product features across countries, with students from Vietnam and China using more retries than Saudi Arabia, and students from Italy using the fewest retries. Students from China, Vietnam, and Saudi Arabia more frequently listened to model audios than students from Italy. A series of Kruskal-Wallis tests revealed significant group differences between dominant L1 and students’ beliefs and perceptions using ASR, and between age groups and students’ beliefs and perceptions using ASR. This study points to the importance of considering regional differences, and suggests that learner engagement may depend not only on the effectiveness of the technology, but also on learner beliefs and perceptions.


Introduction
ASR-CALL activities offer considerable opportunities for individualised practice and personalised feedback on pronunciation (Levis, 2007), and recent studies demonstrate that ASR-CALL activities can have a measurable impact on learning (inter alia, Golonka et al., 2014).
Perceptions regarding the difficulty of achieving English pronunciation skills may be linked to L1/nationality (cf. Cenoz & Lecumberri, 1999;Simon & Taverniers, 2011), and while there are commonalities, beliefs regarding effective learning strategies have been found to differ according to learner L1/nationality (Nowacka, 2012). Such beliefs may lead to pre-conceptions relating to the effectiveness of ASR-CALL activities, which may affect student engagement. This study explores adult students' engagement with ASR-CALL activities and aims to address the following research questions.
• Do students think pronunciation activities with ASR help them improve their pronunciation?
• Do students in four countries (China, Vietnam, Italy, and Saudi Arabia) make different use of ASR-CALL activity features?
• Are there differences between L1s and age groups in students' beliefs and perceptions on learning pronunciation using ASR?

Context and participants
Wall Street English uses proprietary learning content to deliver a blended-learning course, offering a combination of multimedia self-study lessons and face-to-face teacher-led classes. It incorporated ASR word-level pronunciation feedback into three multimedia activity types: (1) repeat and practise, (2) read and record, and (3) conversation. Students were exposed to these ASR activities as part of their course over six weeks. Researchers had access to anonymised back-end data at the end. Participants (N=2,867) completed ASR-powered course activities, and 482 of these responded to an optional, anonymous online questionnaire. There was an unequal age range distribution in questionnaire participants in terms of the most common age groups: [16][17][18][19][20][21][22][23][24][25][26][27][28][29][30] in Saudi Arabia, and 23-60 in Italy (Table 1).

Instruments
Learners completed the activities as part of their normal studies. They were familiar with them, only ASR feedback was new. Questionnaire items (see Table 2 and Table 3) were based on previous research. A six-step Likert scale (from 1=not at all to 6=fully agree) was used.

Results and discussion
Ninety-five percent of students believed that pronunciation activities with ASR helped improve their pronunciation. Vietnamese students were the most enthusiastic (98.8%), closely followed by students in Italy (98.5%), then in Saudi Arabia (95.2%), and, finally, students in China (91.5%).
Concerning the usage of ASR activity features, students in all territories used a much lower number of attempts than they were allowed. Students in China, Saudi Arabia, and Italy only used one attempt (Mdn=1) and those in Vietnam used two (Mdn=2), whereas they were allowed three attempts in Activities 1 and 2, and four in 3. For listening to model audio recordings, students in China and Vietnam reported using this feature the most (82%), less so for students in Saudi Arabia (57%) and in Italy (41%). Finally, students in Vietnam reported listening to own utterance recordings the most (80%), less so for students in China (66%) and in Saudi Arabia (61%), and remarkably lower usage was reported by students in Italy (57%).
Finally, we investigated potential between-group differences in age groups and L1s in students' beliefs and perceptions about L2 pronunciation items. The analyses for age revealed statistically significant differences between age groups and all items but one (Item 4 -see Table 2). Differences were favourable to the youngest age group (16)(17)(18)(19)(20)(21)(22), suggesting that older learners may have less overall confidence in their ability to acquire pronunciation skills (Marinova-Todd, Marshall, & Snow, 2000). Results for L1 analyses are displayed in Table 3. The analyses yielded statistically significant differences between L1s and all beliefs and perceptions about L2 pronunciation statements, in line with some findings reported in Nowacka (2012).
The effect sizes were noticeably larger than for age groups, which may help to explain some differences found in feature usage. For example, students from Italy, who reported the lowest instances of listening to both the sample and their own recorded audios, also demonstrated lowest self-belief in ability to be able to speak English very well, and placed lowest importance on being able to speak with excellent pronunciation.

Conclusions
Learners were overwhelmingly positive towards ASR-CALL's potential in helping improve their pronunciation. However, differences in feature usage were observed between students of different L1s/nationalities, which may be related to differences in learner beliefs.

5.
Disclaimer: Research-publishing.net does not take any responsibility for the content of the pages written by the authors of this book. The authors have recognised that the work described was not published before, or that it was not under consideration for publication elsewhere. While the information in this book is believed to be true and accurate on the date of its going to press, neither the editorial team nor the publisher can accept any legal responsibility for any errors or omissions. The publisher makes no warranty, expressed or implied, with respect to the material contained herein. While Researchpublishing.net is committed to publishing works of integrity, the words are the authors' alone.
Trademark notice: product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe.
Copyrighted material: every effort has been made by the editorial team to trace copyright holders and to obtain their permission for the use of copyrighted material in this book. In the event of errors or omissions, please notify the publisher of any corrections that will need to be incorporated in future editions of this book.