Children's books for general and African American audiences: Comparisons of Language and Content

Children's books for general and African American audiences: Comparisons of Language and Content

First Author: Mark Seidenberg -- UW-Madison
Additional authors/chairs: 
Rubiarbriana Enjoli Korenuk; Matt Cooper Borkenhagen
Keywords: Early Literacy, Book choice, African Americans, Text Characteristics, semantic learning
Abstract / Summary: 

Purpose: Books for children are an important source of knowledge about language and the world. We compared American children's books written for general audiences and ones written for African Americans. The principal question concerns the information these books convey about language and people, including racial identity and gender.
Method: Children's books (ages 5-8) written for general (n = 120) and African American (AA) audiences (n = 99) were transcribed to create a database including information about words (e.g., AoA, affective valence) and main characters' age, race and gender (if identifiable). Initial comparisons focused on word frequencies, affective valence, genderedness, emotional intensity (arousal); semantic relations among words as indicated by word2vec analyses of each corpus; and words associated with race and gender.
Results: Words in the two types of books are similar on many standard measures yet differ substantially. There were 7238 and 8141 distinct word types in the AA vs. (slightly larger) general corpus. 3771 words occurred in both corpora, ~3400 occurred only in the AA corpus and ~4300 only in the general corpus.
Words found in both corpora scored similarly on measures of AoA, concreteness, contextual diversity, valence, and other properties. However semantic representations derived using word2vec for each corpus were correlated at .26 suggesting that many words have different semantic associations across corpora. Content differences were seen in analyses of the words associated with main characters' race and gender, and emphases on different topics (e.g., a higher proportion of words related to family and identity in the AA books).
Conclusions: The two types of books are similar in lexical complexity but differ in content, conveying different information about racial identity, gender, and other human characteristics. Children's books may be an important source of differing knowledge in these areas relevant to child socialization.