Child-adult speech diarization in naturalistic conditions of preschool classrooms using room-independent ResNet model and automatic speech recognition-based re-segmentation.

Speech and language development are early indicators of overall analytical and learning ability in children. The preschool classroom is a rich language environment for monitoring and ensuring growth in young children by measuring their vocal interactions with teachers and classmates. Early childhood researchers are naturally interested in analyzing naturalistic vs controlled lab recordings to measure both quality and quantity of such interactions. Unfortunately, present-day speech technologies are not capable of addressing the wide dynamic scenario of early childhood classroom settings. Due to the diversity of acoustic events/conditions in such daylong audio streams, automated speaker diarization technology would need to be advanced to address this challenging domain for segmenting audio as well as information extraction. This study investigates alternate deep learning-based lightweight, knowledge-distilled, diarization solutions for segmenting classroom interactions of 3-5 years old children with teachers. In this context, the focus on speech-type diarization which classifies speech segments as being either from adults or children partitioned across multiple classrooms. Our lightest CNN model achieves a best F1-score of ∼76.0% on data from two classrooms, based on dev and test sets of each classroom. It is utilized with automatic speech recognition-based re-segmentation modules to perform child-adult diarization. Additionally, F1-scores are obtained for individual segments with corresponding speaker tags (e.g., adult vs child), which provide knowledge for educators on child engagement through naturalistic communications. The study demonstrates the prospects of addressing educational assessment needs through communication audio stream analysis, while maintaining both security and privacy of all children and adults. The resulting child communication metrics have been used for broad-based feedback for teachers with the help of visualizations.

Exploring Growth in Expressive Communication of Infants and Toddlers With Autism Spectrum Disorder.

One of the earliest indicators of autism spectrum disorder (ASD) is delay in language and social communication. Despite consensus on the benefits of earlier diagnosis and intervention, our understanding of the language growth of children with ASD during the first years of life remains limited. Therefore, this study compared communication growth patterns of infants and toddlers with ASD to growth benchmarks of a standardized language assessment. We conducted a retrospective analysis of growth on the Early Communication Indicator (ECI) of 23 infants and toddlers who received an ASD diagnosis in the future. At 42 months of age, children with ASD had significantly lower rates of gestures, single words, and multiple words, but significantly higher rates of nonword vocalizations. Children with ASD had significantly slower growth of single and multiple words, but their rate of vocalization growth was significantly greater than benchmark. Although more research is needed with larger samples, because the ECI was designed for practitioners to monitor children's response to intervention over time, these findings show promise for the ECI's use as a progress monitoring measure for young children with ASD. Limitations and the need for future research are discussed.

Exploring a Novel Tool to Measure Wandering Behavior in the Early Childhood Classroom.

Wandering, or random movement, affects cognitive and social skills. However, we lack methods to objectively measure wandering behavior. The purpose of this pilot study was to explore the use of the Ubisense real-time location system (RTLS) in an early childhood setting to explore wandering in typically developing (TD) children (n = 2) and children with or at risk for developmental disabilities (WA-DD; n = 3). We used the Ubisense RTLS, a tool for capturing locations of individuals in indoor environments, and Fractal Dimension (FD) to measure the degree of wandering or the straightness of a path. Results of this descriptive, observational study indicated the Ubisense RTLS collected 46,229 1-s location estimates across the five children, and TD children had lower FD (M = 1.36) than children WA-DD (M = 1.42). Children WA-DD have more nonlinear paths than TD children. Implications for measuring wandering are discussed.

Automated Measures to Understand Communication Opportunities for Young Children With Autism in the Community: A Pilot Study.

Community participation is vital to children's development and provides opportunities to practice social communication skills. Although previous studies suggest that young children with autism spectrum disorders (ASD) experience decreased community participation, there is little empirical evidence on the precise patterns of participation that may influence social communication opportunities. Therefore, this pilot study investigated the communication among families of children with ASD ( n = 5) versus typical development (TD; n = 5) across various community locations. We used automated, objective measures: the Language ENvironmental Analysis (LENA) system™ to measure the amount of communication and integrated this with a Global Positioning System (GPS; that is, Qstarz™) to measure community location. Results showed that families of children with ASD and TD spent a similar amount of time in community locations; however, there were differences in the amount of adult talk directed toward children with ASD versus TD across community locations. Findings suggest that automated measures may be successfully integrated to quantify social communication during community participation.

Automated Language Environment Analysis: A Research Synthesis.

Purpose: The Language Environment Analysis (LENA®) represents a breakthrough in automatic speech detection because it makes one's language environment, what adults and children actually hear and say, efficiently measurable. The purpose of this article was to examine (a) current dimensions of LENA research, (b) LENA's sensitivity to differences in populations and language environments, and (c) what has been achieved in closing the Word Gap. Method: From electronic and human searches, 83 peer-reviewed articles using LENA were identified, and 53 met inclusionary criteria and were included in a systematic literature review. Each article reported results of 1 study. Results: Originally developed to make natural language research more efficient and feasible, systematic review identified a broad landscape of relevant LENA findings focused primarily on the environments and communications of young children but also older adults and teachers. LENA's automated speech indicators (adult input, adult-child interaction, and child production) and the audio environment were shown to meet high validity standards, including accuracy, sensitivity to individual differences, and differences in populations, settings, contexts within settings, speakers, and languages. Researchers' own analyses of LENA audio recordings have extended our knowledge of microlevel processes in adult-child interaction. To date, intervention research using LENA has consisted of small pilot experiments, primarily on the effects of brief parent education plus quantitative linguistic feedback to parents. Conclusion: Evidence showed that automated analysis has made a place in the repertoire of language research and practice. Implications, limitations, and future research are discussed.

Exploring paraprofessional and classroom factors affecting teacher supervision.

BACKGROUND: Paraprofessionals serve a primary role in supporting students with disabilities in the classroom, which necessitates teachers' supervision as a means to improve their practice. Yet, little is known regarding what factors affect teacher supervision. AIMS: We sought to identify how paraprofessional competence and classroom type affected the levels of teacher direction. METHODS AND PROCEDURES: We administered an adapted version of the Paraprofessional Needs, Knowledge & Tasks Survey and the Survey for Teachers Supervising Paraprofessionals to teachers supervising paraprofessionals in elementary schools. Structural Equation Modeling was used to examine the link between paraprofessional competence and classroom factors affecting the level of teacher supervision. OUTCOMES AND RESULTS: Our results indicated that when teachers perceived paraprofessionals as being more skilled, they provided more supervision, and when more supervision was provided the less they thought paraprofessionals should be doing their assigned tasks. Additionally, paraprofessionals working in classrooms with more students with mild disabilities received less supervision than paraprofessionals working in classrooms with more students with moderate-to-severe disabilities. Those paraprofessionals in classrooms serving mostly children with mild disabilities were also perceived as having lower levels of skill competence than those serving in classrooms with students with more moderate-to-severe disabilities. CONCLUSION AND IMPLICATIONS: By understanding the factors that affect teacher supervision, policy and professional development opportunities can be refined/developed to better support both supervising teachers and paraprofessionals and, in turn, improve the outcomes of children with disabilities.

An automated approach to measuring child movement and location in the early childhood classroom.

Children's movement is an important issue in child development and outcome in early childhood research, intervention, and practice. Digital sensor technologies offer improvements in naturalistic movement measurement and analysis. We conducted validity and feasibility testing of a real-time, indoor mapping and location system (Ubisense, Inc.) within a preschool classroom. Real-time indoor mapping has several implications with respect to efficiently and conveniently: (a) determining the activity areas where children are spending the most and least time per day (e.g., music); and (b) mapping a focal child's atypical real-time movements (e.g., lapping behavior). We calibrated the accuracy of Ubisense point-by-point location estimates (i.e., X and Y coordinates) against laser rangefinder measurements using several stationary points and atypical movement patterns as reference standards. Our results indicate that activity areas occupied and atypical movement patterns could be plotted with an accuracy of 30.48 cm (1 ft) using a Ubisense transponder tag attached to the participating child's shirt. The accuracy parallels findings of other researchers employing Ubisense to study atypical movement patterns in individuals at risk for dementia in an assisted living facility. The feasibility of Ubisense was tested in an approximately 90-min assessment of two children, one typically developing and one with Down syndrome, during natural classroom activities, and the results proved positive. Implications for employing Ubisense in early childhood classrooms as a data-based decision-making tool to support children's development and its potential integration with other wearable sensor technologies are discussed.

Measuring Social Communication in the Community: Novel Tools for Advancing Family Participation.

Natural contexts and family involvement are key features of effective intervention approaches. However, the measurement of community participation and parent engagement with children remains complex. Therefore, we examined the feasibility of combining use of the Language ENvironment Analysis (LENA®) system and a global positioning system (GPS; i.e., Qstarz® BT-Q1000XT). The LENA is a small speech recognition device that captures and quantifies full-day recordings of the natural language environment. The Qstarz BT-Q1000XT is a wearable GPS data logger that allows identification of the locations a person visits. The marrying of these measures allows for an accurate representation of community settings that afford children greater social communication opportunities. Our results show that the combination of measures provides meaningful social communication location data. Also, the participating caregiver reported that the collection of measures was feasible across community settings.

Long-Term Impact of a Cell Phone-Enhanced Parenting Intervention.

Home visiting programs support positive parenting in populations at-risk of child maltreatment, but their impact is often limited by poor retention and engagement. The current study assessed whether a cellular phone-supported version (PCI-C) of the Parent-Child Interactions (PCI) intervention improved long-term parenting practices, maternal depression, and children's aggression. Low-income mothers ( n = 371) of preschool-aged children were assigned to one of the three groups: PCI-C, PCI, and a wait-list control (WLC) group. Parenting improved in both intervention groups between baseline and 12-month follow-up compared to the WLC. Children in the PCI-C group were rated to be more cooperative and less aggressive than children in the WLC. The results offer evidence of the long-term effectiveness of PCI and the additional benefits of cellular phone supports for promoting intervention retention and improving children's behavior.

The questions verbal children with autism spectrum disorder encounter in the inclusive preschool classroom.

This study investigated questions adults asked to children with autism spectrum disorder in inclusive pre-kindergarten classrooms, and whether child (e.g. autism severity) and setting (i.e. adult-to-child ratio) characteristics were related to questions asked during center-time. Videos of verbal children with autism spectrum disorder (n = 42) were coded based on the following question categories adapted from the work of Massey et al.: management, low cognitive challenging, or cognitively challenging. Results indicated that management questions (mean = 19.97, standard deviation = 12.71) were asked more than less cognitively challenging questions (mean = 14.22, standard deviation = 8.98) and less cognitively challenging questions were asked more than cognitively challenging questions (mean = 10.00, standard deviation = 6.9). Children with higher language levels had a greater likelihood of receiving cognitively challenging questions (odds ratio = 1.025; p = 0.007). Cognitively challenging questions had a greater likelihood of being asked in classrooms with more adults relative to children (odds ratio = 1.176; p = 0.037). The findings present a first step in identifying the questions directed at preschoolers with autism spectrum disorder in inclusive classrooms.

Child and setting characteristics affecting the adult talk directed at preschoolers with autism spectrum disorder in the inclusive classroom.

Difficulty with social competence is a core deficit of autism spectrum disorder. Research on typically developing children and children with disabilities, in general, suggests the adult talk received in the classroom is related to their social development. The aims of this study were to examine (1) the types and amounts of adult talk children with autism spectrum disorder are exposed to in the preschool classroom and (2) the associations between child characteristics (e.g. language), activity area, and adult talk. Kontos' Teacher Talk classification was used to code videos approximately 30 min in length of 73 children with autism spectrum disorder (ages 3-5) in inclusive classrooms (n = 33) during center time. The results indicated practical/personal assistance was the most common type of adult talk coded, and behavior management talk least often coded. Child characteristics (i.e. age and autism severity) and activity area were found to be related to specific types of adult talk. Given the findings, implications for future research are discussed.

Using the Language Environment Analysis (LENA) system in preschool classrooms with children with autism spectrum disorders.

This study describes the language environment of preschool programs serving children with autism spectrum disorders (ASDs) and examines relationships between child characteristics and an automated measure of adult and child language in the classroom. The Language Environment Analysis (LENA) system was used with 40 children with ASD to collect data on adult and child language. Standardized assessments were administered to obtain language, cognitive, and autism severity scores for participants. With a mean of over 5 hours of recording across two days several months apart, there was a mean of 3.6 child vocalizations per minute, 1.0 conversational turns (in which either the adult or child respond to the other within 5 seconds) per minute, and 29.2 adult words per minute. Two of the three LENA variables were significantly correlated with language age-equivalents. Cognitive age-equivalents were also significantly correlated with two LENA variables. Autism Diagnostic Observation Schedule severity scores and LENA variables were not significantly correlated. Implications for using the LENA system with children with ASD in the school environment are discussed.