Search for: All records

An important dimension of classroom group dynamics & collaboration is how much each person contributes to the discussion. With the goal of distinguishing teachers' speech from children's speech and measuring how much each student speaks, we have investigated how automatic speaker diarization can be built to handle real-world classroom group discussions. We examined key design considerations such as the level of granularity of speaker assignment, speech enhancement techniques, voice activity detection, and embedding assignment methods to find an effective configuration. The best speaker diarization system we found was based on the ECAPA-TDNN speaker embedding model and used Whisper automatic speech recognition to identify speech segments. The diarization error rate (DER) in challenging noisy spontaneous classroom data was around 34%, and the correlations of estimated vs. human annotations of how much each student spoke reached 0.62. The accuracy of distinguishing teachers' speech from children's speech was 69.17%. We evaluated the system for potential accuracy bias across people of different skin tones and genders and found that the accuracy did not show statistically significantly differences across either dimension. Thus, the presented diarization system has potential to benefit educational research and to provide teachers and students with useful feedback to better understand their classroom dynamics. 

With the aim to provide teachers with more specific, frequent, and actionable feedback about their teaching, we explore how Large Language Models (LLMs) can be used to estimate "Instructional Support" domain scores of the CLassroom Assessment Scoring System (CLASS), a widely used observation protocol. We design a machine learning architecture that uses either zero-shot prompting of Meta's Llama2, and/or a classic Bag of Words (BoW) model, to classify individual utterances of teachers' speech (transcribed automatically using OpenAI's Whisper) for the presence of Instructional Support. Then, these utterance-level judgments are aggregated over a 15-min observation session to estimate a global CLASS score. Experiments on two CLASS-coded datasets of toddler and pre-kindergarten classrooms indicate that (1) automatic CLASS Instructional Support estimation accuracy using the proposed method (Pearson R up to 0.48) approaches human inter-rater reliability (up to R=0.55); (2) LLMs generally yield slightly greater accuracy than BoW for this task, though the best models often combined features extracted from both LLM and BoW; and (3) for classifying individual utterances, there is still room for improvement of automated methods compared to human-level judgments. Finally, (4) we illustrate how the model's outputs can be visualized at the utterance level to provide teachers with explainable feedback on which utterances were most positively or negatively correlated with specific CLASS dimensions.