Sentence Repetition as a Measure of Early Grammatical Development in Italian

Devescovi, A., & Caselli, M.C. (2007). Sentence repetition as a measure of early grammatical development in Italian. International Journal of Language & Communication Disorders, 42(2), 187–208. 

Early identification of language impairment is essential in order to provide timely intervention, however, few standardized measures exist for children under four years of age. Devescovi and Caselli suggest that the Sentence Repetition Test (SRT) could serve as a suitable assessment of language development for young children based on the premise that a child’s ability to repeat a sentence is representative of her spontaneous language abilities. Consensus has not been reached on whether the SRT is an accurate measure of language ability; some say it overestimates ability, while others argue it underestimates it. Most significantly though, SRT has been recognized as a tool for identifying language impairment (e.g., Conti-Ramsden et al., 2001).

In the first of two studies, Devescovi and Caselli examine test-retest reliability of SRT in 100 Italian-speaking children aged 2;0 to 4;0. The authors find that calculating the mean length of utterance in words is sensitive to linguistic development between the ages of 2;0 and 2;6, and that number of sentences repeatedly correctly better discriminates the older age bands. It is important to note that MLU is calculated in words in this study by counting not morphemes, but whole words, which is common in studies of the development of Italian. Perhaps calculation of MLU using morphemes would offer better discrimination beyond 2;6.

In the second study, Devescovi and Caselli examine the correlation between SRT performance, verbal memory span (VMS), and spontaneous speech (structured conversation and picture description), noting that verbal memory span has been shown to predict MLU in spontaneous speech (Blake et al., 1994). SRT responses and spontaneous utterances are analyzed for three factors: number of verbs, omitted articles, and MLU (in words). Analysis revealed no correlation between any of these measures and VMS after controlling for age. Article omission and number of verbs in SRT were both related to spontaneous speech, while correlation between MLU (in words) of SRT and spontaneous speech was not significant.

The sentences used in the SRT included simple sentence structure and only those morphemes expected to develop by 4;0. While it is important to include sentences that could be repeated by the youngest children in the age span, it is possible that limiting the complexity of the stimuli may have limited the performance of the older children, causing a ceiling effect. Despite limitations of the task, positive correlations shown in the findings suggest that some form of SRT may be beneficial in assessing language of young children.

Blogger: Laura Pauls