Document Type

Honors Thesis

Publication Date



This paper reviews research on mean length of utterance (MLU) and longest length of utterance (LLU) as measures of child language development. Although there has been much research on MLU, there is some controversy as to its validity, especially across different data gathering procedures and clinical populations. In contrast, LLU has rarely been the focus of research, but the limited amount of information available indicates that it is a potentially useful way to analyze child speech. This study provides a detailed analysis of 1655 utterances from three children at ages 3;10, 4;1 and 4;4, and 2938 utterances from adults interacting with the children. Variables included MLU, LLU, Fluharty-2 scores, utterance length distributions, and utterance intention codes. LLU did not co-vary strongly with either MLU or Fluharty-2 scores. Distributions of utterance lengt! hs were highly skewed, with single-morpheme utterances occurring most frequently. LLU was highly dependent on language sample size, as shown in analyses of fixed-length samples. Longest utterances were nearly all topic expansions and responses to questions. Adults tended to use topic expansions and questions immediately before children’s longest utterances. Overall, the results suggest that information from LLU may supplement MLU and standardized testing. However, LLU was dependent on a small number of utterances and very susceptible to change according to language sample size. Advantages, disadvantages, and a possible solution are discussed.