Chapter 8 Hypothesis check

Here I revisit my pre-analysis hypotheses.

Children’s accuracy and efficiency of recognizing words will improve each year.

Yes. Their curves reached higher heights and showed steeper slopes each year.

There are stable individual differences in lexical processing of familiar words such that children who are relatively fast at age 3 remain relatively fast at age 4 and age 5.

Yes. The rankings of children by lexical processing measures (peak probability, average probability, linear slope) were concordant over the three years.

However, the magnitude of these individual differences diminishes over time, as children converge on a mature level of performance for this paradigm.

Yes. I simulated new longitudinal participants based on what the model learned about the observed children. The range of plausible looking proportions narrowed each year, so individual differences became less variable each year.

Consequently, individual differences in word recognition at age 3, for example, will be more discriminating and predictive of age-5 language outcomes than differences at age 4 or age 5.

Yes. Correlations between growth curve features with future vocabulary measures were strongest for the age-3 growth curve features.

Children will become more sensitive to lexical competitors as they age, based on the hypothesis that children discover similarities among words as a consequence of learning more and more words.

Yes. The advantage of the phonological competitor and semantic competitor over the unrelated word increased with development.

Children will differ in their sensitivity to lexical competitors, and these individual differences will correlate with other child-level measures.

No evidence to support or refute this hypothesis. I did not find a relationship between age-3 measures with the phonological or semantic competitors. In principle, however, one could design a task and derive a measure from the competitor looking curves that does correlate with child-level measures.