Members of the project

Objectives

The overall aim of this project was to make a contribution to our growing understanding of the complex relationship between the range of linguistic constructions that children hear in the input and those that they produce in their own speech. 

The specific objectives of the project were:

1) To establish a baseline for the use of indirect speech, and temporal and causal subordinates in the early school years 

2) To evaluate the effects of indirect language training on children's short-term use and longer-term acquisition of subordinate clauses.

3) To provide primary school teachers and designers of language materials with research-based evidence on how to incorporate the training of complex syntax in language and literacy activities in the early school years. 

Methodology

The project included two studies: Study 1 in which we primed the use of indirect speech subordinates, and Study 2 in which we primed the use of temporal and causal subordinates; fifty Year1 primary school children from Greater Manchester participated in each study. 

The experimental design of both studies included priming condition as a between-subjects variable (subordinate primes, coordinate primes), and phase as a within-subjects variable (post-test1, post-test2). A pre-test was included to assess children’s baseline production of subordinates.

Unlike in traditional priming experiments, where participants describe a target picture after hearing a description of an unrelated prime picture, in this project we embedded primes in a narrative task. The inspiration for this methodological innovation came from a study of passive priming by Vasilyeva, Huttenlocher, & Waterfall (2006), and was motivated by the need to assess the effects of priming in a more ecologically valid communicative setting. However, unlike Vasilyeva et al., we used narratives both for the priming phase and for the pre- and the post-tests to maximize the naturalness of the testing situation. 

Children’s use of subordination was individually assessed at pre-test (before priming), at post-test1 (1 week after priming), and at post-test2 (10 weeks after priming). During the pre-test, children were administered tests of receptive vocabulary, (the British Pictures Vocabulary Scale 3rd Edition; Dunn, Dunn, Sewell, and Styles, 2009), and of expressive language, (the Formulated Sentences sub-test of the Clinical Evaluation of Language Fundamentals UK, 4th Edition; Semel, Wiig, and Secord, 2006). 

For each study, we wrote thirteen different stories loosely inspired by children’s books. Each story was digitally recorded by a native speaker and was accompanied by visual props in the form of a video with toy figurines, or a PowerPoint presentation with ClipArt pictures. Three stories were used for the one-to-one testing sessions and the ten remaining ones were used for the priming phase that was delivered in group sessions. The ten prime stories had both a subordination and a coordination version that were presented to the children in the two different priming conditions.

The children’s task in the three testing sessions was to listen to a story and then re-tell it to an experimenter while looking at the accompanying PowerPoint slides. To assess the baseline for subordination, in the pre-test all children listened to the same prime story including only coordination. The rationale for this choice was to see whether children would spontaneously use subordinates to connect events. In the two post-tests we used two different stories containing only subordinates to evaluate the extent to which children would respond to priming as a function of training.

During priming, two Year1 classes were randomly allocated to either the subordination condition, where they heard stories containing 20 examples of subordinates, or to the coordination condition, where they heard 20 examples of coordinates in each story; the stories were between 30 and 40 sentences long. Both groups of children participated in daily group story-telling sessions lasting approximately 40 minutes over a period of two consecutive weeks. 

Summary of main findings

To assess the statistical significance of children's use of subordinates in the two studies we performed ANOVAs with priming condition (subordination, coordination) as a between-subjects independent variable, and phase (post-test1, post-test2) as a within-subjects independent variable. The dependent variable in Study 1 was the proportion of indirect speech clauses over the total number of direct and indirect speech clauses. This measure showed how often children chose an indirect speech clause when they decided to report a speech event. In Study 2 the dependent measure was the number of subordinate clauses over the total number of utterances. Unlike in Study 1, where direct and indirect speech are in complementary distribution, the use of a subordinate clause is not necessarily in direct competition with a coordinate clause, we therefore decided to use the total number of utterances as a denominator in this case. 

The results of the statistical analyses for both studies showed a main effect of priming condition: children in the subordination condition produced significantly more indirect speech clauses (Study 1) and more temporal and subordinate clauses (Study 2) than children in the direct speech/coordination condition. In Study 1 there was also a significant main effect for phase showing a proportional increase in the use of indirect speech clauses between post-test1 and post-test2.; no such difference was found in Study 2 where the proportion of subordinate clauses remained constant over the two post-tests. In both studies children in the subordination condition produces significantly more subordinates than the children in the coordination condition at both time points.

No significant associations were found in either study between BPVS standard scores, or Formulated Sentences sub-test standard scores, and the production of subordinate clauses at pre-test or post-test1. For Study 1, at post-test2 CELF subtest scores for children in the subordination condition were positively correlated with the number of indirect speech clauses and with the proportion of indirect speech clauses over the total number of utterances. In Study 2, at post-test2 CELF subtest scores for children in the subordination condition were positively correlated with the number of subordinate clauses. 

These results show long-term effects of priming (up to 10 weeks after the end of the priming phase) beyond clause level and in a very open-ended narrative task. They extend previous findings from lab-based picture description tasks to a more realistic communicative situation and provide new evidence for the effect of input manipulation on children's use of language. The correlational findings between the CELF scores and the effect of priming are somewhat limited but they indicate a positive role for expressive language in predicting sensitivity to priming. 

Outputs

Serratrice, L., Hesketh, A. & Ashworth, R. (2011). The use of indirect speech in a narrative context: A priming study. In Proceedings of the 35th Boston University Conference on Language Development.