The words used for comparing child and adult phonotactic probability and neighborhood density were the 380 nouns on the MacArthur-Bates Communicative Development Inventory: Words and Sentences (Fenson et al., 1993 ) because these words are likely known by both children and adults and have been used in past word learning research (Storkel, 2004a , 2009 (link)). In addition, the words in this set vary in the sounds targeted (i.e., all English sounds present), syllable structure (e.g., CV, CCV, CCCV, VC, VCC, CVC, CCVC, CCCVC, CVCC), and word length (i.e., M = 4.42 sounds, SD = 1.56 sounds, range = 1 – 10 sounds), suggesting that the set represents a wide range of word structures learned by young children.
For each of the 380 nouns, positional segment average, biphone average, and neighborhood density were computed using the on-line calculator and the child corpus. In addition to these raw values, z scores were computed for each word following the procedures of Storkel (2004b , (obtained value - mean)/standard deviation) and using the child means and standard deviations for the appropriate word length (see Appendix A). The same raw values also were computed for each word using the on-line calculator and the adult corpus, and then z scores were calculated using the adult means and standard deviations for the appropriate word length (see Appendix A).
A parallel analysis was completed for a set of 310 nonwords, with results shown in Appendix B. Generally, the nonword analysis produced similar results to the real word analysis.