Figure 5.19 Total number of correct responses on the sentence generation task by age
for both post tests
і Delayed post test
Significant correlations were found between children’s age and their performance on the
sentence generation task for both post tests. The same pattern was evident when their
vocabulary and memory scores were controlled for (see Appendix 5.26 for correlations). As
the above figure shows the older children performed better than the younger ones for both
post tests. Two one way ANOVAs with age as the independent factor and scores on the
sentence generation task as the dependent measure for the both post tests were carried out to
investigate the differences. Significant differences were found during the immediate post test
(Kruskal-Wallis I-Way ANOVA: X2= 44.2, df=2, p<.0000) and the delayed post test
(Kruskal-Wallis I-Way ANOVA: X2= 37.4, df=2, p<.0000).
Post-hoc analyses revealed that the 5 year-olds performed significantly better than the 4 year-
olds during the immediate (Mann-Whitney: Z=3.2, p<.005) and the delayed post test (Mann-
Whitney: Z=3.7, p<.0005). Additionally, the 6 year-olds performed significantly better than
the 4 year-olds during the immediate (Mann-Whitney: Z=6.4, p<.0000) and the delayed post
test (Mann-Whitney: Z=5.9, p<.0000). Furthemiore, the 6 year-olds performed significantly
better than the 5 year-olds during the immediate post test (Mann-Whitney: Z=4.04, p<.0005)
and the delayed post test (Mann-Whitney: Z=2.4, p<.05).
Whether the same pattern was evident within each linguistic condition was also investigated.
The same pattern was found for the inference condition [(immediate: Kruskal-Wallis I-Way
Anova: X2= 10.7, df=2, p<.005); (delayed: Kruskal-Wallis I-Way Anova: X2= 12.4, df=2,
p<.005)], Definition condition [(immediate: Kruskal-Wallis I-Way Anova: X2= 12.8, df=2,