Accuracy scoring of elicited imitation: A tutorial of automating speech data with commercial NLP support
Kathy MinHye Kim, Xiaobin Chen, Xiaoyi Liu · Research Methods in Applied Linguistics · 2024
This tutorial demonstrates how to automate the scoring of two oft-used English morphosyntactic forms, be-passive and third person singular -s, using commercial Natural Language Processing services. It focuses specifically on the context of elicited imitation (EI) tests drawing on previously web-collected EI data (Kim & Godfroid, 2023; Kim et al., 2024). We provide step-by-step instructions and example codes covering three key stages of data processing: (1) speech-to-text transcription, (2) identification of morphosyntactic structures, and (3) the scoring algorithm. This method can be applied to various form-based EI scoring schemes or other form-based automatic scoring tasks, enhancing the broader adoption and practical application of automated scoring in both research and educational settings.