The reliability of language performance measurement in language sample analysis of children aged 5-6 years

The language sample analysis (LSA) is more common in other languages than Persian to study language development and assess language pathology. We studied some psychometric properties of language sample analysis in this research such as content validity of written story and its pictures، test-retest reliability، and inter-rater reliability. We wrote a story based on Persian culture from Schneider’s study. The validity of written story and drawn pictures was approved by experts. To study test-retest reliability، 30 children looked at the pictures and told their own story twice with 7-10 days interval. Children generated the story themselves and tester did not give any cue about the story. Their audio-taped story was transcribed and analyzed. Sentence and word structures were detected in the analysis. Mean of experts'' agreement with the validity of written story was 92. 28 percent. Experts scored the quality of pictures high and excellent. There was correlation between variables in sentence and word structure (p<0. 05) in test-retest، except complex sentences (p=0. 137). The agreement rate was 97. 1 percent in inter-rater reliability assessment of transcription. The results of inter-rater reliability of language analysis showed that correlation coefficients were significant. The results confirmed that the tool was valid for eliciting language sample. The consistency of language performance in repeated measurement varied from mild to high in language sample analysis approach.