Description

A series of self-recording tasks were designed and incorporated into the assignments of the target class. An example task was to record oneself reading a preassigned word list or paragraph in a designated language (English, Mandarin, Cantonese, or other languages/dialects). Students either recorded with their own recording device at home or used one of the language labs on campus, and then uploaded the audio file to a Blackboard-based online system. After the recordings were processed and organized by the project team, they were shared back with students in the teaching of CBS 1900, inviting students to participate in the linguistic analysis of the speech data. Thus, we created a cycle of data collection - sharing - analyzing, where students play an important role in each critical step.

Implementation

The project was implemented between June 1, 2013 and May 31, 2015.

• June 1, 2013 - September 1, 2013 Developing the online platform and designing recording projects
  • Develop an online platform where students can upload and share speech data. The platform will be based on the Blackboard online learning system, utilizing the new podcast feature of Blackboard.
  • Develop a preliminary set of recording projects (about 3-4 per class) in relation to the content ofCBS1900.
• September 1,2013 - December 31, 2013 Collecting the first batch of speech data
  • Adjust (if necessary) and announce the recording projects to students of CBS 1900 during the fall semester of2013/14.
  • Collect the consent from students.
  • Provide technical support to students for uploading and sharing speech data on the online platform.
  • Maintain the online platform.
• Jan 1, 2014 - May 31, 2014 Develop speech databases
  • Download the first batch of speech data from the online platform.
  • Convert file formats (if necessary) and add necessary annotation.
  • Organize the audio files and data provider information of each class into a searchable database that can be used for future research.
• September 1, 2014 - December 31, 2014 Collecting the second batch of speech data
  • Adjust (if necessary) and announce the recording projects to students of CBS 1900 during the fall semester of2013/14.
  • Collect the consent from students.
  • Provide technical support to students for uploading and sharing speech data on the online platform.
  • Maintain the online platform.
• Jan 1,2015 - May 31, 2015 Updating the speech databases and creating the language/dialect map
  • Download the second batch of speech data from the online platform.
  • Convert file formats (if necessary) and add necessary annotation.
  • Add the new data and information to the speech databases.
  • Create an interactive language/dialect map based on the speech data in the databases

Actual implementation was largely in accordance with the project plan, though with some modifications. After consulting the EDC colleagues and the IT technicians in CBS, we adopted Learn@PolyU (i.e. Blackboard) for uploading and sharing the recordings instead of developing a new online platform. Four recording tasks were developed and announced for the class CBS1900 in the fall semester of2013. Each student submitted four recordings of reading paragraphs (two in their L1 - usually Cantonese - and two in English) to fulfill partial credit ofthe course. An online language background survey and a consent survey were also administered on Learn@PolyU, which collected some background information from the students as well as allowed the students to indicate their consent to different ways of using their recordings for teaching and research purposes.

A full-time project assistant (PA) was hired to assist with the project. The PA was in charge of maintaining the recording task on Learn@PolyU, downloading the recordings, preprocessing the speech data, converting file format and organizing a speech database. While doing so, the PI and PA worked together and established a pipeline of scripts and protocols that would facilitate the processing of speech data in the future. 

In fall, 2014, when CBS1900 was re-run, the second batch of recordings were collected from students of the 2014 cohort, using the same methodology as in 2013. More importantly, in addition to creating speech samples, students ofthe 2014 cohort also participated in the analysis of the 2013 data.

Lastly, after both rounds of recordings were organized and preprocessed, the project team developed a website with an interactive language/dialect map.

Evaluation

As for project evaluation, an online questionnaire was administered to collect students' feedback on the recording tasks after the course ended. A total of 18 responses (out of 97 students) were successfully collected. The questionnaire results showed that all students spent no more than 30 minutes on completing the four recording tasks and about 15 minutes on filling out the language background survey and consent survey. Most students (N = 11) did the recording at home, while a small number of students recorded in their dorm (N = 2) or in a language lab on campus (N = 5). The tools that were most often used by the students for recording were computers (N = 12; either desktop or laptop), followed by mobile phones (N = 6) and recording pens (N = 2). All the students reported that they found the recording instructions clearly stated and easy to follow. One student reported that the recording passages contained some infrequent words which were hard to pronounce (which was part of the task design). A couple students reported technical difficulties in saving and uploading recording files (e.g. due to slow network speed). Overall, the students' feedback was highly positive and they considered the recording tasks easy to manage.