Exploring the Use of the Language Environment Analysis (LENA) System in Preschool Classrooms of Children with Autism Spectrum Disorders

Background: Language deficits are a core behavioral feature of autism spectrum disorders (ASD).  Traditionally, language progress is measured with standardized tests. However, researchers debate the validity of using standardized language assessments on young children with ASD because most measures are normed on samples of typically developing children (Charman, 2004; Koegel, Koegel, & Smith, 1997). Researchers are promoting the use of varied methods of language assessment for children with ASD, including data collection in natural contexts (Luyster et al., 2008; Tager-Flusberg et al., 2009). The Language Environment Analysis (LENA) system is a language recording device that collects and analyzes child and adult language in the natural environment.  

Objectives: To explore the utility of using LENA with children with ASD in preschools by: a) examining LENA data, which includes child vocalizations, adult word counts, and child-adult conversational turns; and b) assessing relationships between these variables and standardized assessments of language, cognition, and autism severity.

Methods: Twenty-one children with ASD, ages 3 to 5, enrolled in self-contained classrooms from a Southeastern school district participated in this study. Fifteen of the 21 children were in five classrooms which followed the TEACCH approach; 6 were in two “business as usual” (BAU) classrooms, which used an eclectic approach. All children met ASD diagnostic criteria and had an educational or clinical diagnosis of ASD or DD. Children wore the LENA device one day for an average of 2 hours, 40 minutes. LENA data were converted to rates (frequency/minute), since language samples varied in length.  LENA data and baseline measures of language (PLS-4), cognition (Mullen), and autism severity (CARS) will be reported.

Results: Descriptive language data for 21 children include rates of child vocalizations (CV) (M=3.2, SD=1.6) and child-adult conversational turns (CT) (M=1.0, SD=.5). Adult word counts (AWC) averaged 28.8 words per minute (SD=11.9). Language age-equivalents on the PLS-4 were significantly correlated with LENA variables: CV (r=.45), AWC (r=.44), and CT (r=.52). Mullen Visual Receptive scores were significantly correlated with AWC (r=.45).   No significant correlations were found between LENA measures and CARS scores.  Data will be analyzed for 17 additional children.

Conclusions: This study has produced three compelling findings. First, LENA language data indicate low rates of child vocalizations and conversational turns for children with ASD at the beginning of the school year. Second, the three LENA variables were correlated with a standardized measure of language development (PLS-4), demonstrating that standardized and naturalistic language data collection methods captured similar information. Finally, there was a correlation between children’s cognitive abilities and adult word counts, which provides support for the transactional nature of teacher-child interactions (Yoder & McDuffie, 2006).  Overall, LENA appears to be a feasible tool for research and clinical use in the classroom.  Some of the inherent challenges of using LENA in classroom settings also will be shared.