8.4.3 Children overextend both in comprehension and production tasks

The present study demonstrated that overextensions occurred both in production (e.g, naming
task) and comprehension tasks (e.g., definition task). Existence of overextension errors were
also found by Gelman et al., (1998) in a multiple choice task. The overextensions that
occurred in the present study both in naming and comprehension tasks could be explained
in two possible ways. First, one possibility may be that overextensions in the naming task are
in part usage errors and they may reflect a retrieval problem (Huttenlocher, 1974; Thompson
and Chapman, 1977). Another possibility may be that overextensions reflect aspects of the
child’s underlying semantic representations. For example, if a child calls an ostrich duck,
then the child assumes duck refers to ostrich as well as to duck. Unlike adults, who have
constructed two separate lexical entries, the child has a single lexical entry. On this view,
overextensions indicate a truly broader semantic category (Mervis, 1987).

Maybe the children who are in the process of developing a semantic representation about the
new word make these overextension errors. These errors will disappear later on when they
will have developed a correct semantic representation. That pattern was evident from the
analysis of the naming tasks in both Experiments. Thus, in Experiment 1, during the delayed
post test, and in post tests 2 and 3, they used less basic level words than in the immediate post
test and post test 1.

8.4.4 Children’s word knowledge of the target words changes over time

Both experiments demonstrated the extent to which children’s word knowledge changed
over time. Particularly, Experiment 1, showed that children’s performance on the naming,
multiple choice, analogy and contrast tasks was better during the immediate than the delayed
post test. However, when the analysis was repeated for each group separately, it was found
that the Analogy group performed significantly better on the analogy task during the delayed
than the immediate post test. That finding demonstrates how the influence of the linguistic
input can change the pattern. Moreover, the Inference group’s performance on the inference,
and sentence generation tasks was better during the delayed than the immediate post test.
Experiment 2, also demonstrated that children’s performance across most of the tasks
differed significantly over time. Children’s performance on the naming, definition, short
questions and story generation tasks improved over time. No significant differences, over
time were found for the multiple choice, association and contrast task.

315

<<   312   313   314   315   316   317   318   >>

More intriguing information