Elicited imitation of grammatical and ungrammatical utterances in children with SLI
Résumé
Syntax is known to be a major source of difficulties for children with SLI. It has been widely studied, using various experimental settings such as spontaneous language production, language comprehension, or sentence completion. Elicited imitation has not been much used to study syntax, although it is quite common in phonology. It could be a powerful tool, especially as children with SLI often exhibit delays or deficit in working memory. It also allows targeting specific items or structures and is easy to use in experimental settings.
The current study used an elicited imitation paradigm with 140 utterances. Half the utterances proposed for repetition contained a single grammatical error and the other half were fully grammatical. Grammatical errors covered seven syntactic categories: subject pronoun, object pronoun, auxiliary, verb, determiner, preposition, and noun. For each category, ten utterances were designed with a single grammatical error and ten control utterances were designed with the error corrected, thus testing specific competence for each category. The hypothesis was that it would be more difficult for children with SLI than for the control children (CTR) to restrain from correcting the ungrammatical utterances, as a consequence of their reduced ability to manipulate language.
The current study compared the performance of children with SLI with CTR, matched by language development. Matching was done using a language comprehension tool, the ECOSSE (adaptation of TROG for French). Children were also assessed for language production using the morphosyntactic sub-test of a French test battery, N-EEL. Eleven children with SLI (ages 8;5 to 12;8) and eleven CTR (ages 4;11 to 5;11) were recruited for the study. The task consisted of repeating utterances produced by a personal computer. Children were asked to repeat the utterances as heard, even if some sounded funny to them. All utterances had been checked for syllable length, ranging from 10 to 15 (M. = 12.4).
Two types of analysis were performed on the utterances produced. Analysis #1 checked only the ungrammatical utterances. It attributed a score for each utterance, 1 for very poor repetition, 2 for good repetition but correction of the grammatical error, 3 for exact reproduction of the grammatical error. A significant difference was found, F(1,20)=23.605, p=.00095. Children with SLI produced significantly more poor repetitions and less perfect repetitions than CTR, but a similar percentage of automatic corrections. There was also an effect of syntactic category. Noun was correctly processed by children with SLI who had many difficulties with object pronoun, verb, preposition and determiner. Although performances were significantly different between children with SLI and CTR for subject pronoun and auxiliaries, the difference was smaller, CTR having most of their problems with those two categories. For all other categories, exact reproduction of the grammatical error by CTR was frequent. Analysis #2 checked all utterances, correct and incorrect, using a six values score. A significant difference was found between children with SLI and CTR, children with SLI producing more poor quality repetitions and less perfect repetitions. A significant difference was found between repetition of correct vs. incorrect utterances, but only for repetitions that contained one or two errors or on the contrary, for those that were perfect. Incorrect repetitions often contained a single error, which was usually the correction of the original ungrammatical error. This tendency was the same in both groups.
Elicited imitation proved to be a highly sensitive tool for differentiating between children with SLI and CTR, even with children perfectly matched in language development. This type of test could be included in assessment tools and could also be used to study specific syntactic phenomena, as testing material can be designed easily and with great specificity.
The current study used an elicited imitation paradigm with 140 utterances. Half the utterances proposed for repetition contained a single grammatical error and the other half were fully grammatical. Grammatical errors covered seven syntactic categories: subject pronoun, object pronoun, auxiliary, verb, determiner, preposition, and noun. For each category, ten utterances were designed with a single grammatical error and ten control utterances were designed with the error corrected, thus testing specific competence for each category. The hypothesis was that it would be more difficult for children with SLI than for the control children (CTR) to restrain from correcting the ungrammatical utterances, as a consequence of their reduced ability to manipulate language.
The current study compared the performance of children with SLI with CTR, matched by language development. Matching was done using a language comprehension tool, the ECOSSE (adaptation of TROG for French). Children were also assessed for language production using the morphosyntactic sub-test of a French test battery, N-EEL. Eleven children with SLI (ages 8;5 to 12;8) and eleven CTR (ages 4;11 to 5;11) were recruited for the study. The task consisted of repeating utterances produced by a personal computer. Children were asked to repeat the utterances as heard, even if some sounded funny to them. All utterances had been checked for syllable length, ranging from 10 to 15 (M. = 12.4).
Two types of analysis were performed on the utterances produced. Analysis #1 checked only the ungrammatical utterances. It attributed a score for each utterance, 1 for very poor repetition, 2 for good repetition but correction of the grammatical error, 3 for exact reproduction of the grammatical error. A significant difference was found, F(1,20)=23.605, p=.00095. Children with SLI produced significantly more poor repetitions and less perfect repetitions than CTR, but a similar percentage of automatic corrections. There was also an effect of syntactic category. Noun was correctly processed by children with SLI who had many difficulties with object pronoun, verb, preposition and determiner. Although performances were significantly different between children with SLI and CTR for subject pronoun and auxiliaries, the difference was smaller, CTR having most of their problems with those two categories. For all other categories, exact reproduction of the grammatical error by CTR was frequent. Analysis #2 checked all utterances, correct and incorrect, using a six values score. A significant difference was found between children with SLI and CTR, children with SLI producing more poor quality repetitions and less perfect repetitions. A significant difference was found between repetition of correct vs. incorrect utterances, but only for repetitions that contained one or two errors or on the contrary, for those that were perfect. Incorrect repetitions often contained a single error, which was usually the correction of the original ungrammatical error. This tendency was the same in both groups.
Elicited imitation proved to be a highly sensitive tool for differentiating between children with SLI and CTR, even with children perfectly matched in language development. This type of test could be included in assessment tools and could also be used to study specific syntactic phenomena, as testing material can be designed easily and with great specificity.