Abstract
The mean length of utterance (MLU; Brown, 1973) is an established tool for assessing children's language development based on spontaneous language production. Although many studies have shown a high correlation between MLU calculated in words, morphemes, or syllables (Parker and Brorson, 2005; Rice et al., 2010), in some languages a certain type of MLU is preferred (Allen and Dench, 2015; Tomas and Dorofeeva, 2019).
Research results also disagree on how long a transcript should be used to calculate the MLU - although some authors suggest a minimum of 75 (Tomas and Dorofeeva, 2019), 100 (Oosthuizen and Southwood, 2009) or 175 utterances (Gavin and Giles, 1996), others show the possibility of calculating MLU from shorter transcripts (Casby, 2011). In practice, then, MLU is often calculated from all intelligible utterances (Dethorne et al, 2005; Rice et al. 2006, 2010).
This paper compares the results of three different types of MLU calculated based on different length transcripts and verifies the validity of the tool in Czech by comparing it with the results of tests of vocabulary and grammar comprehension in children aged 2.5 and 3-4 years.