Grammatical morphology is not a sensitive marker of language impairment in Icelandic in children aged 4–14 years

Abstract

PurposeGrammatical morphology continues to be widely regarded as an area of extraordinary difficulty in children with Specific Language Impairment (SLI). A main argument for this view is the purported high diagnostic accuracy of morphological errors for the identification of SLI. However, findings are inconsistent across age groups and across languages. Studies show morphological difficulty to be far less pronounced in more highly inflected languages and the diagnostic accuracy of morphology in such languages is largely unknown. This study examines the morphological use of Icelandic children with and without SLI in a cross-sectional sample of children ranging from preschool age to adolescence and assesses the usefulness of morphology as a clinical marker to identify SLI. MethodsParticipants were 57 monolingual Icelandic-speaking children age 4–14 years; 31 with SLI and 26 with typical language development (TD). Spontaneous language samples were coded for correct and incorrect use of grammatical morphology. The diversity of use of grammatical morphemes was documented for each group at different age and MLU levels. Individual accuracy scores were plotted against age as well as MLU and diagnostic accuracy was calculated. ResultsMLU and morphological accuracy increased with age for both children with SLI and TD, with the two groups gradually approaching each other. Morphological diversity and sequence of acquisition was similar across TD and SLI groups compared based on age or MLU. Morphological accuracy was overall high, but was somewhat lower in the SLI group, in particular at ages below 12 years and MLU levels below 6.0. However, overlap between the groups was important in all age groups, involving a greater tendency for errors in both groups at young ages and scores close to or at ceiling at older ages. Sensitivity rates as well as likelihood ratios for each morpheme were all below the range considered acceptable for clinical application, whereas better specificity rates in some age groups for some morphemes indicated that very low scores are indicative of SLI whereas high scores are uninformative. Age effects were evident in that the morphemes varied in the age at which they separate the groups most accurately. ConclusionsThe findings of this study show that Icelandic children with SLI are somewhat more prone to making morphological errors than their TD counterparts. However, great overlap exists between the groups. The findings call into question the view that grammatical morphology is a central area of deficit in SLI.