The Rich Language of Book Sharing within English- and Spanish-Speaking Families

• In Event: Opportunities for learning from picture books and shared reading in monolingual and bilingual environments

Abstract

The benefits of shared book reading with children are well-established. Children who are read to frequently tend to have more advanced vocabularies and stronger pre-literacy skills (e.g., Bus et al., 1995). This may be grounded in the types of language input that books themselves, as well as the book-sharing context, provide. For example, large-scale corpus analyses show that the language in book texts is more lexically diverse and more syntactically complex than the language in child-directed speech (CDS) (Dawson et al., 2021). Moreover, caregivers’ spontaneous speech related to the book in book-sharing interactions is more complex than spontaneous speech in other contexts (Demir-Lira et al., 2019). Thus, findings suggest ‘book language’ is richer than other sources, though that language could be further parsed into the text of the book, what is actually read of that text, and then the additional spontaneous speech that occurs alongside it. Expanding on prior work, our study directly compares all three types of ‘book language’ with other CDS within the same samples of both English- and Spanish-speaking families.

In this study, we first collected all-day naturalistic recordings using LENA from 22 monolingual English-speaking and 20 Spanish-speaking families with 24-mo-old children from diverse socioeconomic backgrounds. For each family, we used the LENA automated adult word counts to sample the six densest 10-min segments containing CDS. Then we coded all activities in which caregivers were verbally engaged, including book-sharing, play, routines, meals, and unstructured conversation. All CDS was transcribed verbatim. For book-sharing episodes, we annotated all utterances as either read-aloud CDS or spontaneous–book CDS. For all non-book-sharing activities, we annotated all spontaneous–other CDS. Finally, all books chosen by families and read in the recorded book-sharing episodes were transcribed as book text in their entirety.

These four sources of linguistic input were analyzed by: (1) concatenating all transcripts within categories to form “mega-transcripts”, and examining the type and type–token ratio curves; and (2) calculating by-transcript measures of lexical diversity (measure of textual lexical diversity [MTLD]), lexical density (proportion of lexical words), lexical sophistication (mean token frequency in SUBTLEXus), and grammatical complexity (mean length of utterance in words [MLU-w]). We compared these measures across sources via mixed-effects regressions with family as a random effect.

Replicating results of previous studies, the “mega-transcript” measures showed that book text had more word types and higher type–token ratios than spontaneous–other CDS. We also found that book texts and read-aloud CDS were more lexically diverse than spontaneous talk during both book and non-book activities (Figure 1). The by-transcript measures showed that across both language groups read-aloud CDS was similar to book text in all measures, while spontaneous–book CDS was similar to spontaneous–other CDS. Further, book text and read-aloud CDS were richer than spontaneous CDS.

We conclude that the richness of language during book sharing may largely be driven by caregivers’ speech as they read book texts aloud with their children, rather than by the spontaneous speech that also occurs during book-sharing. The latter may serve important but different functions.

Authors

• Kirsten Read, Santa Clara University
  Presenting Author

• Janet Yougi Bang, San José State University

• Alvin Tan

• Virginia A. Marchman, Stanford University