Date of Award
Speech and Hearing Sciences
Language and languages -- Variation, Children -- Language -- Ability testing, Spanish language -- Lexicology
Purpose: various studies have assessed whether lexical diversity in language samples is a useful clinical tool for assessing language abilities. However, not all studies used the same measure of lexical diversity indices or sample demographic. Therefore it is unclear which lexical diversity indices work best when analyzing Spanish language samples. The current study aimed to compare validity evidence for five lexical diversity measures based on Spanish language samples. Specifically it assessed how changes in language-sample length influenced each measure. The most accurate lexical diversity measures were expected to yield similar estimates for a given child across language samples that differed in length.
Method: sixty monolingual Spanish-speaking children from Mexico participated in the study. The five lexical diversity measures evaluated included TTR, MATTR, MTLD, HDD, and Maas. Each lexical diversity measure was estimated at language-sample lengths of 50, 100 and 200 words for each child to assess whether the estimates changed across different lengths. The coding program R was used to truncate the language samples and estimate the lexical diversity indices. To determine whether variation in language-sample length yielded differences in lexical diversity estimates, we analyzed scores from each index using a repeated measure ANOVA with three levels (50 Words, 100 Words, and 200 Words).
Results: results suggested that Maas and MTLD were the most effective measures of lexical diversity in the Spanish language, followed by HDD. The least effective measures were MATTR and TTR.
Stills, Morgan, "Language Sample Length Effects on Various Lexical Diversity Measures: An Analysis of Spanish Language Samples from Children" (2016). University Honors Theses. Paper 233.