The research team, led by Laura M. Justice of the Crane Center for Early Childhood Research and Policy at The Ohio State University, set out to answer a deceptively simple question: do children with language difficulties experience peer talk differently from their classmates? Language difficulties are among the most common neurodevelopmental concerns of early childhood. Late language emergence, marked by delays in vocabulary and word combinations, affects about 13 percent of two-year-olds, while developmental language disorder, a condition impairing language comprehension and production, affects an estimated 7 to 10 percent of four- and five-year-olds. The causes remain unknown, though the condition is highly heritable, pointing to a genetic etiology.

Why does peer talk matter so much? Decades of developmental science show that adverse peer relations in childhood predict internalizing and externalizing behavior problems, conduct difficulties, and risk behaviors extending into adolescence and adulthood. For young children specifically, exposure to talk from others provides the linguistic stimuli that shape the neural pathways supporting language ability. Research in inclusive preschool classrooms has even shown that language growth over an academic year accelerates for children with language difficulties when they interact frequently with peers who have advanced language skills, a phenomenon known as the peer effect. If children with language difficulties are talking less with peers, they may be missing a critical developmental engine.

Previous work had already hinted at trouble. A study of 56 inclusive preschool classrooms using teacher reports found that children with language difficulties had significantly fewer ties in the classroom social network and were substantially more likely to be isolated from their peers. But teacher ratings and brief observations carry well-known biases. The new study took a different route: objective, continuous measurement using the Interaction Detection in Early Academic Settings, or IDEAS, sensing system, which combines voice recorders with Bluetooth proximity beacons worn by teachers and children throughout the school day.

The technology is technically sophisticated. Each participant wore a pouch or vest housing a voice recorder and a location-tracking beacon, with an accelerometer triggering data capture once the wearer was moving around the classroom. Wall-mounted antennae communicated with the beacons via radio frequency signals, classifying participants as being in proximity when they were within 1.5 meters of one another and oriented within 45 degrees. A MATLAB-based pipeline then applied automatic speech recognition to the audio, counting the words exchanged between people who were simultaneously in proximity. A support vector machine filtered out idle periods, and all interaction measures were standardized to words per minute of wear time. Validation work shows the word counts correlate with ground truth at r = .99 for teacher talk and r = .93 for child talk.

The study spanned 29 preschool classrooms in two mid-western states, followed across two consecutive academic years, with up to eight daylong observations per classroom between December and June. Of 181 children in the analytic sample, 36, or 20 percent, were identified as having language difficulties using multiple criteria: a standard score at or below 81 on the expressive communication subscale of the Preschool Language Scale Fifth Edition, an Individualized Education Program, and/or teacher-reported concern about language development. The researchers deliberately avoided relying solely on IEP status, since many children with language needs are never formally identified during the preschool years.

The headline finding is stark. Across all children, peer talk dwarfed teacher talk: children received an average of 31.33 words per minute from peers but only 6.75 from teachers, and produced 21.93 words per minute to peers versus just 2.55 to teachers, effects so large they correspond to standardized differences of 1.61 and 2.23 respectively. Against this backdrop, children with language difficulties received 3.66 fewer peer words per minute and produced 2.50 fewer peer words per minute than their classmates, differences that were statistically significant in multilevel models accounting for observations nested within children and classrooms. Scaled up, the researchers calculate, that gap amounts to about 7,200 fewer words from peers per week, or roughly 260,000 words across an academic year.

Just as striking was what the study did not find. Verbal interactions with teachers occurred at nearly identical rates for children with and without language difficulties, with no significant differences in either incoming or outgoing teacher talk. This surprised the team, given prior evidence that teachers interact more frequently with children with disabilities in inclusive settings. The authors suggest the null result may reflect the more objective and sustained nature of sensing-based observation compared with earlier methods, or the broader, less clinically defined way language difficulties were identified in this sample. Either way, the finding underscores that the distinctive experience of these children lies squarely in the peer domain, not the adult one.

Why do children with language difficulties talk less with peers? The study cannot establish causes, but the authors offer two leading interpretations. The first is homophily, the well-documented tendency of children to prefer playmates who share characteristics such as gender, age, or skill level. Recent research across 28 preschool classrooms found that children receiving special education were 11 percent more likely to play with other special-education children than with those outside that designation. Since children with language difficulties made up only 20 percent of each classroom, such preferences would sharply restrict the pool of available conversational partners. The second interpretation involves pragmatic language, the social use of language. Skills such as taking conversational turns, entering ongoing play, asking questions, and resolving conflicts verbally are often underdeveloped in children with language difficulties, and weak pragmatics has been specifically linked to social exclusion in preschool settings.

The developmental stakes are considerable, though the authors are careful about directionality. Less incoming peer talk could deprive children of linguistic input that fuels vocabulary and grammar growth, compounding their initial difficulties in a feedback loop. Less outgoing talk could further marginalize them from classroom social networks, with downstream consequences for behavior and learning that peer-rejection research has documented for decades. Alternatively, the authors note, the consequences could be neutral or even positive in some respects, and determining which is the case requires future classroom-based research. The study also has limits the authors acknowledge: 29 classrooms limit generalizability, the language-difficulty group may not map neatly onto clinical diagnoses, and the sensors capture frequency but not quality, missing gestures, facial expressions, perspective-taking, and real-time repair of communication breakdowns. Out-of-classroom settings such as playgrounds, where social networks may form differently, were also not observed. Still, the findings mark a technical milestone, demonstrating that wearable sensing can quantify, with unprecedented granularity, the everyday verbal lives of young children, and they signal an urgent need for interventions, such as peer-mediated approaches, that promote peer-to-peer talk for the children who need it most.

Subject of Research: Peer verbal interactions of preschool children with language difficulties measured with wearable sensing technology

Article Title: Peer Talk in the Preschool Classroom: Are the Experiences of Children With Language Difficulties Distinctive?

Article References: Justice, L. M., Pelfrey, G. L., Jiang, H., Foster, T. J., Irwin, D., & Villasanti, H. G. (2026). Peer Talk in the Preschool Classroom: Are the Experiences of Children With Language Difficulties Distinctive?. Journal of Autism and Developmental Disorders. https://doi.org/10.1007/s10803-026-07436-8

Image Credits: AI Generated

DOI: 10.1007/s10803-026-07436-8

Keywords: developmental language disorder, preschool, peer interactions, peer talk, wearable sensors, early childhood education, language development, social exclusion, pragmatic language, teacher-child interaction, inclusive classrooms, child development

The study, conducted by researchers at Michigan State University including Kang Gao, Chandani Bhandari, Tianxiang Zhang, Sarah N. Douglas, and Subir Biswas, focused on a single target child in an inclusive preschool classroom who had been identified by his head teacher as facing both behavioral challenges and English language learner (ELL) status. The child, given the pseudonym Doruk, was a four-year-nine-month-old White male who spoke Turkish at home and used conversational English at school. His teacher reported that he engaged with peers in aggressive ways, such as hitting and biting, and that classmates had begun avoiding him as a result. Despite concerns shared with the family, no special education services were pursued, and Doruk carried no formal diagnosis related to his behavior.

The stakes of such early social difficulties are considerable. The authors note that nearly 22 percent of young children in the United States speak a language other than English at home, with 15 percent of kindergarten children identified as English language learners, while behavioral challenges affect approximately 14 percent of children in early childhood. Prior research has shown that children with behavioral challenges tend to engage in fewer peer interactions, are often excluded from peer play, experience more peer conflicts, have fewer friends, and more frequently play alone. For children who are ELL, peer interactions are often short-lived, and a lack of English proficiency can itself produce challenging behaviors when peers fail to understand a child’s messages—sometimes leading to special education referrals or exclusionary discipline.

The theoretical grounding for the work draws on complementary frameworks. Social engagement theory holds that active participation in social settings benefits both the child and peers, while a lack of engagement is associated with poorer mental health and behavioral difficulties. Vygotsky’s sociocultural theory emphasizes that peer interaction is crucial for cognitive and language development, and Bandura’s social learning theory highlights how children learn by observing and imitating their social environment—processes that can be hampered when language barriers exist. Together, these perspectives suggest that children with behavioral and language challenges may need additional, well-timed teacher support to build sustainable social engagement in preschool settings.

Measuring that engagement, however, has long been a practical bottleneck. Continuous observation is time-consuming, can introduce bias, and reduces the direct teaching time teachers have available. When teachers lack dedicated observation time, they may rely on memory, which the researchers note can be inaccurate. The study therefore employed the Wearable Human Interaction Tracker, or WHIT, a sensor system built around Ultra-Wideband (UWB) radio-based location tracking, integrating localization hardware from Eliko Systems within a custom wearable package. UWB localization was chosen because its performance has been validated in prior research and industrial applications.

The technical architecture of WHIT is elegantly simple in concept. Participants carry tag sensors that transmit UWB blink messages received by wall-mounted anchor sensors; using signal timing, the system performs triangulation to determine each child’s precise real-time location, forwarding computed positions to wall-mounted receiver sensors. Each sensor is the size of a standard credit card, weighs 12 grams, and carries a rechargeable battery lasting seven hours per charge. The electronics are sealed in plastic packaging for safety and secured with snaps into pockets of custom-made t-shirts to prevent removal. Doruk wore two sensors, one in each sleeve pocket, to detect the directionality of his movements, while his five participating peers wore a single front-pocket sensor each.

Proximity served as the indicator of social engagement. A threshold of one meter was chosen empirically based on prior studies: proximity within one meter maintained for at least 15 seconds was coded as social engagement, ending only when the distance exceeded one meter for 10 seconds or more—an approach that accommodated natural movement while preserving measurement accuracy. Calibration preceded data collection, and at least two research assistants independently logged each child’s location and behaviors every minute from an observation booth to validate the sensor data. Agreement between observational and sensor data reached 93.73 percent, with most disagreements occurring during transitions between classroom locations or near zone boundaries. Data were collected across six observation days over two months in late Fall 2024, during both structured teacher-led activities and unstructured free play, with sessions averaging 80 minutes.

The results painted a nuanced and somewhat surprising portrait of Doruk’s social world. His overall social connectivity was 45.1 percent, actually higher than the average of his peers, yet his average connectivity duration was shorter at 48.5 seconds compared with the peer average of 54.4 seconds. His connectivity with any individual peer never exceeded 19 percent, and he showed a preference for one peer, Otto, over the other four. The most striking finding emerged when activity type was considered: during structured, teacher-guided activities, Doruk’s social connectivity soared to 91.6 percent, exceeding every peer, while during unstructured free play it fell to 33.7 percent—the largest structured-versus-unstructured gap of any child in the study.

Initiation and termination data added another layer of insight. Doruk’s initiation rate was 53.1 percent, slightly above the peer average, and his initiation-to-termination ratio of 1.09 exceeded the classroom average of 0.97. In practical terms, Doruk frequently started social interactions, but it was usually his peers who ended them—a pattern the authors interpret as a less stable social engagement style, with limited peer interest in sustaining interactions. A spatial heatmap of his movements revealed a strong preference for the area between the kitchen and the center table, along with time spent at the craft and sensory tables, and comparatively little time in the book nook, animals, and blocks areas.

For the researchers, these findings carry clear implications for practice. The data suggest that Doruk may have a strong need for structure in his social interactions, and that teacher guidance allows him to engage more appropriately—while also indicating he may need support learning to sustain peer engagement and build stronger individual relationships. Teachers could use initiation and termination data to design interventions promoting meaningful peer interaction, scaffold social engagement activities strategically, and leverage area preferences when planning targeted activities. More broadly, the study demonstrated that individual differences in social connectivity, peer preference, and engagement patterns are common even among children without identified challenges—differences that continuous observation alone would rarely detect. The authors acknowledge limitations, including the single-classroom exploratory design, the inability to separate the effects of ELL status from behavioral challenges, and the asymmetric sensor configuration, and they call for larger samples and two-sensor configurations for all participants. With further development, they conclude, wearable sensor systems like WHIT could provide the objective, real-time data teachers need to complement their own observations and inform targeted interventions in inclusive preschool classrooms.

Subject of Research: Wearable sensor-based measurement of social engagement patterns in preschool children with behavioral challenges and English language learner status

Article Title: Sensor Technology to Measure Social Engagement for Children with Behavioral Challenges

Article References: Gao, K., Bhandari, C., Zhang, T., Douglas, S. N., & Biswas, S. (2026). Sensor Technology to Measure Social Engagement for Children with Behavioral Challenges. Early Childhood Education Journal. https://doi.org/10.1007/s10643-026-02353-y

Image Credits: AI Generated

DOI: 10.1007/s10643-026-02353-y

Keywords: wearable sensors, social engagement, preschool, behavioral challenges, English language learners, ultra-wideband localization, inclusive classrooms, early childhood education, peer interaction, child development, WHIT sensor system, teacher intervention