Three phonological theories at test about language impairment in French-speaking children
Résumé
This study put to the test three theories about language deficit in children with SLI. The three theories –pure phonological deficit (Owen, Dromi, & Leonard, 2001), surface (Leonard, 1992), mapping (Chiat, 2002)– have in common the fact that the existence of a phonological impairment is a primary cause for the children's language disorder. The three theories differ about the way syntactic impairment is linked to phonological impairment. This allows to make predictions about the relationship between phonological mismatches and syntactic categories produced by children. As predictions differs from one theory to another, theories can be rated by how well they predict the children's difficulties for each syntactic category.
The spontaneous language of 16 children with SLI and of 16 control children matched on MLU was analysed. In both SLI and NLD groups, the children were distributed in two different subgroups, on the basis of their MLU: 8 SLI and NLD children with low MLU (2.4-2.7), 8 SLI and NLD children with high MLU (3.7). The children production was phonetically transcribed. Percentages of correct word production (CWP) and percentage of phonemes correct (PPC) was computed for each of the following nine syntactic categories: adverb, determiner, noun, preposition, strong pronoun, subject pronoun, unmarked lexical verb, auxiliary, marked lexical verb. The results obtained were compared to the predictions of each theory. Mapping theory always presented the worse match between prediction and results. CWP results gave a slight advantage to phonological theory other surface theory, whereas PPC results didn't give an advantage to any of the two theories. Surface theory predicted results better for children with SLI than for control children, whereas phonological theory did not fare differently in the two populations.
Analysis for each syntactic category showed that, for CWP, predictions were better for determiners, nouns, prepositions, and non-marked lexical noun, whereas for PPC, predictions were better for personal subject pronouns and determiners. For both measures, predictions were worse for strong pronouns, which children produced not as correctly as predicted. Adverbs fared similar to strong pronouns, but to a lesser degree. On the contrary, determiner, auxiliaries and personal subject pronouns were produced by children better than predicted. There was no developmental effect.
These results show that, although most theories are tailored to existing data (often English data), they seemed to explain the data only in some syntactic categories. French auxiliaries fared much better than English auxiliaries, probably because they are phonologically very simple. However, no such explanation is readily available for adverbs, and strong pronouns, which are not phonologically complex and should have been better produced. It remains to decide whether the inadequacy of the theories was due to their focusing on the English language or their focusing on the (English) verb category. This calls for a fine grained analysis of the results obtained for French. It also shows that theories about language disorders that focus mainly on the disorders with verbs may be insufficient to explain the whole pattern of phonological disorders in children with SLI.
The spontaneous language of 16 children with SLI and of 16 control children matched on MLU was analysed. In both SLI and NLD groups, the children were distributed in two different subgroups, on the basis of their MLU: 8 SLI and NLD children with low MLU (2.4-2.7), 8 SLI and NLD children with high MLU (3.7). The children production was phonetically transcribed. Percentages of correct word production (CWP) and percentage of phonemes correct (PPC) was computed for each of the following nine syntactic categories: adverb, determiner, noun, preposition, strong pronoun, subject pronoun, unmarked lexical verb, auxiliary, marked lexical verb. The results obtained were compared to the predictions of each theory. Mapping theory always presented the worse match between prediction and results. CWP results gave a slight advantage to phonological theory other surface theory, whereas PPC results didn't give an advantage to any of the two theories. Surface theory predicted results better for children with SLI than for control children, whereas phonological theory did not fare differently in the two populations.
Analysis for each syntactic category showed that, for CWP, predictions were better for determiners, nouns, prepositions, and non-marked lexical noun, whereas for PPC, predictions were better for personal subject pronouns and determiners. For both measures, predictions were worse for strong pronouns, which children produced not as correctly as predicted. Adverbs fared similar to strong pronouns, but to a lesser degree. On the contrary, determiner, auxiliaries and personal subject pronouns were produced by children better than predicted. There was no developmental effect.
These results show that, although most theories are tailored to existing data (often English data), they seemed to explain the data only in some syntactic categories. French auxiliaries fared much better than English auxiliaries, probably because they are phonologically very simple. However, no such explanation is readily available for adverbs, and strong pronouns, which are not phonologically complex and should have been better produced. It remains to decide whether the inadequacy of the theories was due to their focusing on the English language or their focusing on the (English) verb category. This calls for a fine grained analysis of the results obtained for French. It also shows that theories about language disorders that focus mainly on the disorders with verbs may be insufficient to explain the whole pattern of phonological disorders in children with SLI.