Sample size and type-token ratios for oral language of preschool children.

This study investigated the stability of five type-token ratios (TTRs) in 50-utterance oral language samples segmented into nine lengths. The samples were obtained from 83 children, 3, 4, and 5 years of age. The five TTRs included the basic type-token ratio, the corrected type-token ratio, the root type-token ratio, the bilogarithmic type-token ratio, and the Characteristic K. The sample segment sizes consisted of the first, second, third, and fourth 50-word segments; the first and second 100-word segments; the first 150-word segment; the first 200-word segment; and the total 50-utterance sample. Based on the results of this study, TTR measures on the language of young children should not be compared for samples that differ in number of words. Further, TTR measures for sample size of 50 and 100 words have reliabilities that are judged to be inadequate for research or clinical purposes. The size of the language sample needed for minimum reliability of .70 is 350 words. Greater reliability requires larger word-sample size.