engleski

Generalizations in verb paradigms in Croatian typically developing children and a child with developmental language disorder

One of the key questions in language acquisition is how children form generalizations that allow them to produce novel utterances, while avoiding ungrammatical forms. This process is gradual, and children overcome morphologically complexity by overgeneralization of rules. Probably mostly researched is a case of English past tense where children use morpheme -ed for verbs such as hit, sit or go (hitted*, sitted*, goed*). The entrenchment hypothesis (Braine and Brooks, 1995) states the role of frequency in retreating from overgeneralization errors. Studies have shown that production probability of such errors decreases with increasing verb frequency (overview: Ambridge, 2013). Morphologically rich languages comprise broad array of partial regularities and additional rules that must be included in producing morphological forms, differing from languages the acquisition of which has been mostly studied (namely, English). Ambridge et al. (2013) suggest that mechanism which govern the acquisition of morphology might differ, i.e. children have to rely on some kind of probabilistic generalization mechanism working at the levels of distribution, morphophonology, and semantics simultaneously. Croatian with its system of 10 verb classes, at least one of them with additional subclasses (Jelaska, 2005) presents challenges for the acquisition. Research has shown that overregularized forms persist up to at least 7 years of age with typically developing children (Hržica & Lice, 2012), although the number of errors diminishes with age. Children with developmental language disorder (DLD ; previously known as specific language impairment – SLI) display differences in language production skills when compared to the children with typical language development (TD). Research made so far found that DLD children produce more grammatical errors and more ungrammatical utterances overall (e.g. Dunn et al., 1996, Eisenberg and Guo, 2013), and that certain markers can serve as an indicators of language impairment (overview: Wittke et al., 2017). This study comprises two questions: a. can the entrenchment explain retreat from overgeneralizations in verbal morphology in Croatian as a morphologically rich language and b. do children with DLD benefit from the frequency effect. Parental diary of one child with DLD was analysed, in addition to Croatian corpus of child language (Kovačević, 2002) and standardization data of one language test. Fifty overgeneralized verbs were selected from the parental diary and other sources. All verbs were divided to subtypes and their frequency was calculated using Croatian Web Corpus (Ljubešić & Klubička, 2014). Results have shown that the child with DLD produced overgeneralized verbs in more subtypes when compared to TD children (although the difference in source of data should be considered: diaries might have denser data in some respect – see Teakstorn et al., 2012). Also, relative frequency of verbs was from less than 10 to more than 600 per million. There was no obvious effect of faster recovery with more frequent verbs. However, subtypes of errors, based on transparency of verb classes, provided more ground for conclusions. This was true for all participants. More research in morphologically rich languages is needed in order to understand the acquisition of complex morphological systems.

verb paradigm, overgeneralization, language acquisition, TLD children, DLD children

nije evidentirano