Teacher AI Literacy for Multilingual Learner Instruction: Scale Development and Initial Validity Evidence

Higher education institutions require a methodologically sound measure of teachers’ ability to utilize GenAI in the classroom. Current measures focus on general competencies and fail to address the specific pedagogical requirements, instructional judgments and ethical considerations. The purpose of this study is to develop, test and analyze the factor structure of a measure of teacher AI literacy for ML instruction and examine its relationship with teacher responsible-use intentions and vignette-based decision making. Utilizing a convergent mixed-methods design, this study surveys 32 in-service teachers. Descriptive statistics and reliability are calculated for all quantitative scales. EFA is conducted on the AI literacy scale. Pearson correlations are calculated to examine relationships among the quantitative variables. Regression models are also estimated to predict responsible-use intentions and vignette-based decision-making. Thematic analysis is employed to analyze the qualitative responses. The results indicate moderate to high levels of AI literacy, high internal consistency reliability and high levels of responsible-use intentions. The exploratory factor analysis results in a three-factor solution explaining 69.1% of the variance. Finally, AI literacy is positively correlated with both responsible-use intentions and vignette-based decision-making. These findings provide empirical support for developing domain-specific assessments and underscore AI literacy as more than just general AI confidence.