Protocol for a Scoping Review of Task Design in Automated Assessment of L2 Speaking and Pronunciation

With improvements in automatic speech recognition (ASR) technology, there has been an increase in the use of automated scoring of pronunciation and speaking assessments. Test takers’ speech is elicited through controlled (e.g., read aloud tasks) or spontaneous (e.g., picture descriptions) tasks. It has been established that task type can influence pronunciation performance, yet controlled tasks tend to be used more frequently than spontaneous tasks, both in pronunciation instruction and pronunciation assessment studies. Despite the impact the task type can have on test taker output, there has been no systematic examination of the use of task type in research focusing on automated scoring of pronunciation and speaking. This scoping study will lay the groundwork for such an examination. We used a structured search strategy across seven databases to identify relevant studies. We will apply forward and backward citation tracking to identify any missing relevant studies. Data regarding task type, as well as other relevant assessment information, will be extracted. This scoping review will provide a descriptive summary of task type use, such as trends over time, through the use of descriptive statistics and visuals. This will provide insight into how task type shapes findings in automated pronunciation and speaking research.