Chapter 14 Hypothesis check

Children’s accuracy and efficiency of recognizing real words and fast-associating nonwords will improve each year.

Yes and no. There were obvious gains for nonwords from 3 to 4 and from 4 to 5 for real words. There are gains, but not “each year”. One complication here is that many children started to hit ceiling performance by age 4, so there was not much of a developmental gradient for these conditions.

Performance in real word recognition and fast association of nonwords will be highly correlated, based on the hypothesis that the same process (referent selection) operates in both situations.

Yes, there were correlated. But not “highly”.

Under the alternative hypothesis, real word recognition and fast referent selection reflect different skills with different developmental trajectories. Thus, if there is any dissociation between recognition of real words and nonwords, it will be observed in younger children.

No—not at all. I observed a dissociation between the two conditions, but it worked in the opposite direction and at the older ages. That is, children at age 4 and age 5 demonstrated an apparent nonword advantage.

Although these two measures will be correlated, I predict performance in the nonword condition will be a better predictor of future vocabulary growth than performance in the real word condition. This hypothesis is based on the idea that fast referent selection is a more relevant skill for learning new words than recognition of known words.

Yes, peak looking probabilities for the nonword condition at age 3 predicted expressive vocabulary at age 5, and the real word looking probabilities did not. The size of this effect was rather small, however.

For the mispronunciations, I predict children with larger vocabularies (that is, older children) will be more likely to tolerate a mispronunciation as a production of familiar word compared to children with smaller vocabularies.

Yes, older children looked more to the familiar word on average on the mispronunciation trials.

Mispronunciations that feature later-mastered sounds (e.g., rice-wice) will be more likely to be associated to novel objects than earlier-mastered sounds (duck-guck).

Not answered. I present the item-level results in in Appendix E. I chose not to formally analyze the items because there were only 6 mispronunciations each year, making it too difficult to generalize about different kinds of mispronunciations. Plus, children seemed to not know rice as well as the other familiar words which makes the problem of measuring a mispronunciation penalty more difficult.