Hugo Gonzalez Villasanti

Abstract Classrooms are dynamic spaces where student learning emerges from interactions among peers, teachers, and the environment. We present a computational framework based on agent-based modeling to explore how physical layout, peer influence, and teacher guidance affect equity in skill development. The model represents classroom elements as potential fields and updates each student’s skill state from daily interactions, using equity metrics inspired by distributive justice. Comparing a barrier-free layout with one featuring physical obstacles, the simulation shows that unobstructed environments promote better access to learning resources, peer and teacher interactions, and ultimately reduce skill disparities. This work lays the groundwork for a sociotechnical digital twin that uses real-time data, and its flexibility in defining skill and equity criteria enables educators and stakeholders to test and refine classroom designs collaboratively. In doing so, it offers the foundation for a transparent and adaptable tool aimed at optimizing classroom conditions so that every learner gains equitable opportunities to learn in their classroom.

Curriculum planning for many educators and educational researchers takes into account the use of the classroom environment. This is true across most age groups, but becomes even more important for early childhood learning for which the design of the space, the activities associated with a particular location, and the learning objectives are tightly coupled. Existing approaches to understand classroom usage are based on human-based observational methods and qualitative assessments. However, these methods may result in misallocation of the spatial usage when multiple students are engaged in different areas within the environment and the human observer may not be present. To provide a more quantitative approach, this work investigates the use of non-invasive wearable sensors and Markov chains to analyze proximity data collected for 8 days over a span of 4 months. Probabilistic maps representing the transition sequences between different location and task states for 7 users are presented. Variations between users and from day-to-day interactions are discussed. The Markov chains derived from these data illustrate that students preferred the snack area when the teacher was not engaged within the decision-making process, with a shift to drama time or circle time when the teacher was engaged with the students.

Control-based adaptive interventions are promising tools to promote equitable psychomotor skill development in classrooms. This work develops a novel multi-agent model tailored for such interventions, where students are represented as agents in a cooperative task involving simultaneous learning and goal achievement. The model’s low dimensionality facilitates validation, while its mathematical tractability enables analysis and control-based task design. Two students interacting with a ball-on-plate task are simulated to investigate the sensitivity of inequity and learning outcomes to diversity and collaboration. The results reveal that the outcomes depend nonlinearly on diversity and collaboration, suggesting the need for optimal design of learning tasks.

In this paper, we present an agent-based model (ABM) to study equitable interventions in cyber-physical-social systems (CPSIS) within educational environments. With the use of influence derived from physical and cognitive factors, we can capture the nuances of social interactions and peer learning. Simulation results demonstrate how different layouts and instructional strategies impact equity, providing valuable insights for educators aiming to optimize classroom environments for diverse learners. The model represents an explainable framework for participatory modeling and intervention design.

PURPOSE: This study examines the accuracy of Interaction Detection in Early Childhood Settings (IDEAS), a program that automatically transcribes audio files and estimates linguistic units relevant to speech-language therapy, including part-of-speech units that represent features of language complexity, such as adjectives and coordinating conjunctions. METHOD: measure determines the accuracy of IDEAS diarization (i.e., speech segmentation and speaker classification). Two additional evaluation metrics, namely, median absolute relative error and correlation, indicate the accuracy of IDEAS for the estimation of linguistic units as compared with two conditions, namely, Oracle (manual diarization) and Voice Type Classifier (existing diarizer with acceptable accuracy). RESULTS: measure for SLP talk data suggests high accuracy of IDEAS diarization for SLP talk but less so for child talk. These differences are reflected in the accuracy of IDEAS linguistic unit estimates. IDEAS median absolute relative error and correlation values for nine of the 10 SLP linguistic unit estimates meet the accuracy criteria, but none of the child linguistic unit estimates meet these criteria. The type of linguistic units also affects IDEAS accuracy. CONCLUSIONS: IDEAS was tailored to educational settings to automatically convert audio recordings into text and to provide linguistic unit estimates in speech-language therapy sessions and classroom settings. Although not perfect, IDEAS is reliable in automatically capturing and returning linguistic units, especially in SLP talk, that are relevant in research and practice. The tool offers a way to automatically measure SLP talk in clinical settings, which will support research seeking to understand how SLP talk influences children's language growth.

Young children's language and social development is influenced by the linguistic environment of their classrooms, including their interactions with teachers and peers. Measurement of the classroom linguistic environment typically relies on observational methods, often providing limited 'snapshots' of children's interactions, from which broad generalizations are made. Recent technological advances, including artificial intelligence, provide opportunities to capture children's interactions using continuous recordings representing much longer durations of time. The goal of the present study was to evaluate the accuracy of the Interaction Detection in Early Childhood Settings (IDEAS) system on 13 automated indices of language output using recordings collected from 19 children and three teachers over two weeks in an urban preschool classroom. The accuracy of language outputs processed via IDEAS were compared to ground truth via linear correlations and median absolute relative error. Findings indicate high correlations between IDEAS and ground truth data on measures of teacher and child speech, and relatively low error rates on the majority of IDEAS language output measures. Study findings indicate that IDEAS may provide a useful measurement tool for advancing knowledge about children's classroom experiences and their role in shaping development.

Young children spend substantial portions of their waking hours in noisy preschool classrooms. In these environments, children's vocal interactions with teachers are critical contributors to their language outcomes, but manually transcribing these interactions is prohibitive. Using audio from child- and teacher-worn recorders, we propose an automated framework that uses open source software both to classify speakers (ALICE) and to transcribe their utterances (Whisper). We compare results from our framework to those from a human expert for 110 minutes of classroom recordings, including 85 minutes from child-word microphones (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathrm{n}=4$</tex> children) and 25 minutes from teacher-worn microphones (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathrm{n}=2$</tex> teachers). The overall proportion of agreement, that is, the proportion of correctly classified teacher and child utterances, was. 76, with an error-corrected kappa of. 50 and a weighted F1 of. 76. The word error rate for both teacher and child transcriptions was. 15, meaning that 15% of words would need to be deleted, added, or changed to equate the Whisper and expert transcriptions. Moreover, speech features such as the mean length of utterances in words, the proportion of teacher and child utterances that were questions, and the proportion of utterances that were responded to within 2.5 seconds were similar when calculated separately from expert and automated transcriptions. The results suggest substantial progress in analyzing classroom speech that may support children's language development. Future research using natural language processing is under way to improve speaker classification and to analyze results from the application of the automated framework to a larger dataset containing classroom recordings from 13 children and 3 teachers observed on 17 occasions over one year.

Young children spend substantial portions of their waking hours in noisy preschool classrooms. In these environments, children's vocal interactions with teachers are critical contributors to their language outcomes, but manually transcribing these interactions is prohibitive. Using audio from child- and teacher-worn recorders, we propose an automated framework that uses open source software both to classify speakers (ALICE) and to transcribe their utterances (Whisper). We compare results from our framework to those from a human expert for 110 minutes of classroom recordings, including 85 minutes from child-word microphones (n=4 children) and 25 minutes from teacher-worn microphones (n=2 teachers). The overall proportion of agreement, that is, the proportion of correctly classified teacher and child utterances, was .76, with an error-corrected kappa of .50 and a weighted F1 of .76. The word error rate for both teacher and child transcriptions was .15, meaning that 15% of words would need to be deleted, added, or changed to equate the Whisper and expert transcriptions. Moreover, speech features such as the mean length of utterances in words, the proportion of teacher and child utterances that were questions, and the proportion of utterances that were responded to within 2.5 seconds were similar when calculated separately from expert and automated transcriptions. The results suggest substantial progress in analyzing classroom speech that may support children's language development. Future research using natural language processing is under way to improve speaker classification and to analyze results from the application of the automated framework to a larger dataset containing classroom recordings from 13 children and 3 teachers observed on 17 occasions over one year.

Sensing technologies that provide continuous, real-time information about teachers’ and students’ individual experiences are increasingly being applied to classroom-based research. Sensing technologies provide a possible alternative to costly and time-intensive in-person or hand-coded observations and have the potential to increase our present understanding of the vastly different experiences students within the same classroom often have. The goal of the present article is to provide an overview of sensing technologies, an explanation of how these technologies can be applied in early childhood classroom-based research, and examples of existing studies that have successfully implemented sensing technologies in the classroom environment.

PURPOSE: The purpose of this study was to examine the measurement structure of the linguistic features of speech-language pathologists' (SLPs) talk during business-as-usual therapy sessions in the public schools and to test the longitudinal stability of a theorized dimensional structure consisting of quantity, grammatical complexity, and lexical complexity. METHOD: Seventy-five SLPs' talk during therapy sessions with primary-grade students was automatically transcribed and coded for linguistic features from a corpus of 579 videotaped therapy session videos collected at the beginning, middle, and end of one school year with an approximately 12-week interval. We explored video characteristics and conducted descriptive statistics on eight linguistic indices of SLP talk to examine the variability in SLP talk between therapy sessions. Confirmatory factor analyses were used to explore the dimensional structure of SLP talk at each time point separately for the theorized three dimensions, and we conducted longitudinal measurement invariance analyses to test the stability of the three-factor structural model across the academic year. RESULTS: There were considerable variabilities among SLPs in the characteristics of SLP talk during therapy sessions. The proposed three-factor structure of SLP talk consisting of quantity, grammatical complexity, and lexical complexity had good model fit at all three time points. The linguistic measurement properties representing the three factors were invariant over time. CONCLUSIONS: Results provided robust evidence of between-SLP variability in their child-directed talk, established a three-dimensional structure of the linguistic features in SLP talk, and identified that the linguistic features in SLP talk stably measured the same constructs across one school year, based on measurement invariance. The dimensions of SLP talk during therapy with students may represent important, malleable features of therapy that influence child language gains.

PURPOSE: A common goal of peer-mediated interventions is to train peers to successfully initiate and maintain social and linguistic interactions with a target child in the classroom. Ample evidence indicates that peer-mediated interventions improve social and linguistic outcomes for students in the primary and later grades with developmental disabilities; however, relatively little work has focused on applying these approaches to socially isolated preschoolers who are vulnerable to academic and social challenges. In this early-stage feasibility study, we examined the potential of Promoting Early Engagement, Relationships, and Socialization (PEERS), a peer-mediated intervention designed to support socially isolated preschoolers. The main goal was to consider whether the intervention showed promise for improving the social and linguistic experiences of young, isolated children and was appropriate for implementation in a preschool setting. METHOD: In one preschool classroom, two peers who were well connected to their classmates were trained via storybook-based activities to provide social support to an isolated classmate. In-person and sensing technology observations were used to examine the target student's incoming peer interactions, outgoing peer interactions, and physical proximity to peers. RESULTS: In-person observations showed a trend toward increased interactions; sensing technology data, which provided longer snapshots of classroom experiences, were more variable. In addition, teachers perceived PEERS to be appropriate and effective. CONCLUSION: The findings of this study support further investigation of storybook-based peer-mediated interventions in preschool settings to improve the social and linguistic experiences of children who are socially isolated.