Purpose: It is well documented that object relatives are more difficult to process than subject relatives. However research by Kidd, Brandt, Lieven and Tomasello (2007) showed that when typically developing children were tested on object relatives they most often say and hear in naturalistic speech the subject/object asymmetry disappears. This study compares the performance of children with SLI on subject relatives and two types of object relatives – those that have been typically used in previous research studies and those that are reflective of the relative clauses in the ambient language.
Method: Children with SLI (mean age = 6;10, n = 32) and two control groups - an AM-TD group (mean age = 6;11, n = 32) and a YTD group, (mean age = 4;9, n = 20) repeated sentences containing subject and object relative clauses controlled for length. The object relatives were manipulated for animacy of the head noun and the type of subject within the relative clause (lexical noun phrase or personal pronoun).
Results: The children with SLI showed significantly greater difficulty than both control groups. Both the AM-TD group and the children with SLI performed significantly better on object relatives that contained inanimate head nouns and pronominal relative clause subjects (Oi). There was no difference in their performance on intransitive subject relatives and Oi-relatives and they performed significantly better on Oi relatives than on transitive subject relatives. There were no significant differences between subject and object relatives for either of the control groups.
Conclusion: Our findings show that when children were asked to repeat object relatives that they most often hear and say in naturalistic speech, not only does the subject – object asymmetry reported in the literature disappear but the children with SLI perform significantly better on object relatives. It is argued that this is reflective of the importance of structural frequency in the ambient language.