Phrase-based Coding Framework for Teachers’ Motivational Language Based On Self-Determination Theory

Teachers’ motivational behaviours play a critical role in shaping student outcomes, but commonly used measurement approaches rely on self-report or observation. These are vulnerable to bias and practical constraints, limiting rigorous theory testing and intervention research. To overcome these limitations, and building on advances in language-based behavioural analysis, we developed a phrase-based coding framework for classifying teachers’ motivational language based on Self-Determination Theory. We first compiled a pool of candidate phrases. Experts reviewed this pool to ensure face validity and refinement, resulting in the full, unfiltered dictionary. In the training set, transcripts from the most and least need-supportive teachers—as determined by observers—were compared using weighted log odds ratios to identify phrases more characteristic of each group. We then filtered the dictionary using these odds ratios, resulting in the filtered dictionary. The expert-derived dictionary consisted of 227 words (149 need-supportive and 82 need-thwarting). Correlations between dictionary-based and observer ratings were moderate and significant for the unfiltered dictionary (r = .34), comparable to inter-observer agreement, rating the same lesson (r = .32). Filtering the dictionary improved accuracy on the training set (r = .49) but decreased accuracy on the test set compared to the unfiltered dictionary (r = .64 vs. .73). This phrase-based framework provides a cost-efficient and reliable method for analysing teachers’ motivational language. It appears to perform comparably to human observers in predicting need-supportive teaching and offers a practical tool for teacher education, as well as a theoretically grounded basis for training more advanced models.