Ludwigsburg University of Education

Background Enhancing student learning with the use of cognitively and linguistically challenging questions is considered key for fostering student learning in classroom discourse. Theoretical models suggest that teachers’ use of questions is related to their achievement expectations for students. However, evidence for the interplay of teacher expectations, teacher questions, and student learning is scarce. Aims This study (1) describes the use of teacher questions in whole-classroom discourse, (2) examines relations between teacher expectations and teacher questions, (3) and investigates the effects of teacher questions on student achievement in reading, vocabulary, and mathematics. Sample The sample includes 329 first grade students and 17 teachers (nschools = 13) from 15 German language and 14 mathematics classrooms in Germany. Methods First, we coded the language-promoting quality of teacher questions based on video recordings of whole-classroom discourse. Second, we employed multilevel modeling to examine links between teachers’ (inaccurate) expectations, teacher questions, and later student achievement. Results Teacher questions were mostly of low language-promoting quality. Teachers asked significantly more questions in mathematics, including more open-ended questions. For mathematics, (inaccurately) low expectations were linked with higher frequencies of questions that promote elaboration and description. A positive effect on vocabulary emerged from questions that motivate students to hypothesize and conclude. Conclusions Overall, language-promoting questions occur comparatively rarely in content lessons. Evidence for class-level expectancy effects on teacher behavior was limited. Our findings suggest that the effectiveness of teacher questioning is domain-specific.

Generative Artificial Intelligence (GenAI) is reshaping education by introducing tools that enhance teaching methodologies, personalize learning, and streamline administrative tasks. However, adoption of these tools remains uneven, raising concerns about disparities in AI literacy and competency across geographic regions and educational contexts. Here we investigate the adoption of GenAI tools among second language (L2) educators in the United States, Colombia, Germany, and Macau—professionals uniquely positioned to benefit from and highlight barriers to GenAI integration. Using survey data, we assess four areas: accessibility of GenAI tools, teacher knowledge of potential applications, integration in teaching practices, and the nature of professional development provided. Our results indicated substantial intra- and inter-country variance, with U.S. and Colombian educators reporting higher familiarity and usage compared to those from Germany and Macau. Additionally, university and high school teachers were more likely to access professional development and leverage GenAI for tasks like assessment and differentiation than elementary or middle school educators, regardless of geographic setting. These disparities align with broader trends in AI adoption, reflecting heterogeneity in cultural attitudes, systemic barriers, and institutional support. Our findings highlight the critical need for targeted strategies that mitigate these emerging gaps in AI literacy, competency, and professional development.

Large language models (LLMs) are increasingly used in research as both tools and objects of study. Much of this work assumes that LLM performance under fixed conditions (identical model snapshot, hyperparameters, and prompt) is time-invariant, meaning that average output quality remains stable over time; otherwise, reliability and reproducibility would be compromised. To test the assumption of time invariance, we conducted a longitudinal study of GPT-4o's average performance under fixed conditions. The LLM was queried to solve the same physics task ten times every three hours over approximately three months. Spectral (Fourier) analysis of the resulting time series revealed substantial periodic variability, accounting for about 20

Abstract Artificial Intelligence (AI) has become an important modeling tool in physics. This article is intended as a research-informed instructional proposal on teaching about AI use cases in physics for upper-level undergraduate physics students. Four exemplary use cases are presented, where students can learn in what ways AI can facilitate data-driven and physics-informed modeling for physical systems with an emphasis on instructional goals and limitations of AI. Our work builds on prior curriculum suggestions for data-driven physics and knowledge on AI usage in physics. We seek to provide guidance for physics instructors and students on potentials and limitations of AI in physics, and how to implement AI models for the respective use cases.

ABSTRACT Background Measurement of adaptive competencies is important in the assessment, educational planning and progress monitoring of children with intellectual disabilities. The responsible use of diagnostic instruments in clinical and educational settings requires a comprehensive psychometric knowledge base for all intended uses of test scores. The Vineland‐3 is a widely used instrument that incorporates three dimensions of adaptive behaviour: Communication, Daily Living Skills and Socialization. Despite their widespread use, few studies have addressed the reliability of the Vineland‐3 Domain‐Level Teacher Form in children with intellectual disabilities. Method Reliability, known‐groups validity, distributional characteristics and intercorrelations of the domain scales from the German adaption of the Vineland‐3 Domain‐Level Teacher Form were examined in a sample of 600 pupils with intellectual disabilities. Results Reliability coefficients for the domain scales and the Adaptive Behavior Composite (ABC) were excellent (0.95–0.98), and known‐groups validity was as expected, with most ABC scores (88.0%) falling below 70. The test floor of 20 (corresponding to a z score of −5.33 with an expected occurrence of approximately 1 in 20 million cases) was reached in 15.8% of the sample. Despite the assumed conceptual distinctiveness of the scales intercorrelations of domain scales were substantial ( r = 0.76–0.91) in all age groups. All domain scales correlated very strongly with the ABC score ( r ≥ 0.90). Conclusion The German Domain‐Level Teacher Form of the Vineland‐3 demonstrated high reliability. However, our data strongly advise caution when interpreting standard scores in the very low range. We suggest that test publishers verify the accuracy of the norms of the Domain‐Level Teacher Form.