Not many teachers use spoken corpora in their teaching (Chen & Tian, 2020), because of reasons such as a general discomfort with technological innovations (O’Brien et al., 2018) and lack of familiarity with corpus tools and search techniques (Mauranen, 2004). But research on uses of spoken corpora in language instruction has shown promising results in the teaching of prosody. For example, Staples (2019) created an ESP pronunciation curriculum for internationally educated nurses, relying on corpus-based needs analysis, material development, and assessment. Chen and Han (2020) used a phonologically annotated corpus of Mandarin to identify common areas of segmental and suprasegmental difficulties in the speech of L1 Cantonese learners of Mandarin and develop a platform to provide learners with perception and production activities on these areas. 

Corpora have also been used in training pronunciation teachers. In a workshop for pre-service pronunciation teachers in an EFL context, Chen and Tian (2020) introduced teachers to a number of English learner corpora and corpus-based lesson plan and activity development and provided them with more extensive training on using corpora in their teaching.

For pronunciation-focused courses, corpus data can be employed at different stages of curriculum development and lesson planning (see Staples, 2019). CoTACS can be used by ITA trainers before developing a course to conduct needs analysis aimed at identifying language functions that ITAs are expected to perform in the target situation, i.e., teaching undergraduate courses (“Target Situation Analysis”; Flowerdew, 2013). Instructors can also benefit from CoTACS in assessing the gap that exists between learners’ current stage of knowledge and what is expected at the end of the course (“Present Situation Analysis”; Flowerdew, 2013). Similarly in teacher training, CoTACS can be used to introduce instructors to the linguistic objectives ITAs are expected to reach by the end of the course, as well as areas of pronunciation that ITAs from different L1 backgrounds need to improve on.

In developing content for pronunciation lessons, CoTACS offers a wide range of speech data to choose from depending on the learners’ needs and the instructional context. Segments of speech that include the target pronunciation features for each lesson can be extracted from both the native and non-native samples of the corpus. While native speech samples can help illustrate prosodic patterns that facilitate listeners’ understanding, it is also important for pronunciation teachers to make use of non-native speech samples to illustrate correct uses of pronunciation features. This suggestion is supported by almost three decades of pronunciation research, which has shown the importance of intelligibility over native-like pronunciation (Levis, 2005; 2020; Munro and Derwing, 1995). Instructors should consider using intelligible non-native speech samples in their teaching and open up discussions of why being intelligible should be prioritized over native-speaker standards. Pronunciation instructors can use the corpus audio files either with screenshots of Praat to provide a visual representation of the target pronunciation feature, or with short TextGrid files extracted from the original TextGrid files. The latter provides learners with the opportunity for a more in-depth analysis of speech and is recommended with advanced learners who have received extensive training on how to use speech corpora .

Prior to the lessons, learners need to be familiarized with basic pronunciation concepts related to prosody, especially thought grouping or tone units. This will prepare them for analyzing other intonational features that occur within a tone unit, such as prominence and tone choice.

Before tasking learners with exercises using corpus data, introduce them to the speech analysis software, Praat, and how its basic features work.

Provide learners with guidance about the conventions used in the corpus annotations, such as double slashes for tone units and capital letters for prominence, and what information is presented in each tier of the TextGrid files.

Learners can be provided with transcripts of short segments of native and non-native speech from CoTACS and asked to mark tone units (thought groups) on the transcripts while listening to the audio. They can then compare their observations of the differences between thought grouping in native and non-native speech. More advanced learners can be asked to mark thought groups on the transcripts before listening to the audio, allowing them to assess their understanding of the concept before focusing on their perception skills. This can be followed by an activity focused on correctly producing the non-native examples in thought groups.

Learners can be asked to open excerpts of TA speech in Praat, identify short and long pauses, measure their length, and compare these traits between native and non-native speech samples. This can be followed by a discussion of the impact of pause length on how listeners understand speech. Advanced learners can also be prompted to examine pause placement as it relates to thought grouping. 

Instructors can provide a lesson focused on how falling, rising, and level tones and pitch contours can be used in interactions to indicate friendliness and build rapport. In the case of ITAs, for example, the instructor can provide examples from CoTACs to explain how tone choice can help elicit linguistic or non-linguistic response from undergraduate students, resulting in more interaction in the classroom. In such a lesson, ITAs learn about or improve their understanding of using a rising tone to raise a question or to speak a statement in such a way that elicits students’ response.