Examination of Test-Retest and Split-Half Reliability for Measures Derived from Language Samples of Young Handicapped Children

Abstract

Test-retest and split-half reliability for measures derived from language samples were examined for 10 children with developmental delays, including the area of language development. The children were between the ages of 52 and 80 months. MLU and specific categories of lexical and morphological production were examined. It was found that test-retest samples resulted in greater variability for lexical and morphological measurements than split-half transcripts. In addition, a trend was noted wherein 50 utterance sample segments contained 70–80% of the lexical information found in 100 utterance samples. Implications for gathering and interpreting language samples are discussed, including recommendations for length of samples and number of samples required to gather representative information.