What type of interruption interrupts? An investigation of the processing of centre embedded subject relatives with adults and children

Anna Weighall, Gerry Altmann and Alison McManus
Dept. of Psychology, University of York, UK.

In contrast to earlier findings using the act out task (Sheldon, 1977; Tavakolian, 1981), it has recently been shown that 6-7 year old children find centre embedded subject relative sentences (SS) more difficult to comprehend than right branching object relatives (OS) (Weighall & Altmann, 2001; Kidd & Bavin, in press). It has also been shown that children have problems comprehending the interrupted main clause of the embedded structure, not the relative clause itself (Weighall & Altmann, 2001). Experiment 1 used an auditory comprehension task to establish that adults exhibit the same pattern of results as children. 30 adults heard embedded and right branching relative sentences. Very few comprehension errors were made for either clause of the right branching structures, or for the relative clause part of the embedded sentences. Performance on the main clause of the embedded sentences was significantly worse than in the other three conditions. Single embeds have a processing cost for the adult, and developing, parser. Experiment 2 investigated the nature of the disruption with 30 6-7 year olds using three types of embedded structures (8 sentences per condition). The children responded to a comprehension question about the main clause for each one.

(A) Subject Relative The cat that bumped the bear will hug the cow
(B) Prepositional Phrase The cat next to the bear will hug the cow
(C) Adjective Relative The cat with the striped tail will hug the cow

Question: What sort of animal will hug the cow?

Performance in conditions A and B was equally poor (25% and 28% correct respectively), but performance in condition C was significantly better (66% correct). This suggests that not all interruptions disrupt processing equally. Furthermore, the difficulty with embedded structures is not limited to those structures that contain a relative clause. The difference between conditions A and C cannot be explained in purely structural terms. This pattern of results is explained by the "attentional shift" hypothesis which asserts that there is a processing cost when a structure requires a shift in attention from one agent to another and back again. Conditions A and B require this shift (from the cat to the bear and back to the cat) condition C does not, hence performance is superior. Alternative explanations will also be considered, as will a further study controlling for the effect of number of animate objects within each condition. We suggest that children's processing of the embedded structures is qualitatively similar to that of adults.

References

Sheldon, A. (1977). On strategies for processing relative clauses: A comparison of children and adults. Journal of Psycholinguistic Research, 6, 305-318.

Tavakolian, S. L. (1981). The conjoined-clause analysis of relative clauses. In S. L. Tavakolian (Eds.), Language acquisition and linguistic theory. (pp. 167-187) Cambridge, Mass.: MIT Press.

Kidd, E. & Bavin, E.L. (to appear). English speaking children's understanding of relative clauses: Evidence for the universal-cognitive and language specific constraints on development. Journal of Psycholinguistic Research.

Weighall, A. & Altmann, G.T.M. (2001). When two cats are better than none: Children's interpretation of relative clauses (revisited). Paper presented at AMLaP 2001, Saarbrucken, Germany.