Linguistic Qualities of Teacher and Child Contextualized and Decontextualized Talk in Norway

• In Event: Children’s decontextualized talk in ethnically and socioeconomically diverse communities in the U.S., China and Norway

Abstract

Engagement in decontextualized talk – that is, talk that extends beyond the immediate context – is a significant predictor of children’s linguistic skills as well as their proficiency with academic language (Demir, Rowe, Heller, Goldin-Meadow, & Levine, 2015, Rowe 2012, Uccelli, Demir-Lira, Rowe, Levine, & Goldin-Meadow, 2019). It is hypothesized that the linguistic qualities inherent in this type of talk may explain the positive impact of decontextualized input on child language outcomes. For example, decontextualized conversations about the nonpresent, requires explicit references to people, objects, and events, rather than the use of gestures or deictic cues to convey meaning. Consequently, decontextualized input may expose children to more elaborate vocabulary and structurally advanced sentences compared to talk that is grounded in the here-and-now (i.e., contextualized talk). Yet, few studies have compared characteristics of contextualized and decontextualized language, and there is a need to learn more about the linguistic features of these types of talk. In the present study, we observed Norwegian preschoolers and their teachers in three different conversational settings and investigated the syntactic and lexical qualities of child and teacher contextualized and decontextualized talk.
Thirty-eight dyads (sample demographics are provided in Table 1) of preschool children and their teachers were filmed while engaging in three different activities: a) play with animal figures, b) picture book reading, and c) story card conversation. Seven minutes of conversation from each activity were transcribed orthographically following CHAT conventions (MacWhinney, 2020). The unit of transcription was the utterance, segmented into communication units. Utterances were subsequently coded as contextualized or decontextualized depending on whether or not they were related to observable entities or information otherwise provided by the context. Reliability was achieved by having two coders independently code 24 % of the transcripts (Cohen’s kappa ranging between .79-.95). To assess the linguistic qualities of contextualized and decontextualized talk, we calculated child and teacher mean length of utterance in words (MLU-W), lexical density (the number of content words per utterance), and the number of elaborated noun phrases (ENP) per utterance.
As seen in Table 2, child and teacher decontextualized utterances were, on average, longer than contextualized utterances and included a higher density of content words. The rate of ENPs was also higher in decontextualized utterances, although this trend was less pronounced. Next, the linguistic quality measures for children and teachers were z-standardized and collapsed into one dependent variable, allowing us to fit a single multivariate regression model and account for multiple hypothesis testing. Dummy variables for the three linguistic quality measures and type of utterance (i.e., contextualized/decontextualized) were included as predictors in the model. The analysis revealed higher outcomes on the composite measure of linguistic quality for decontextualized utterances (B = 0.89; CI 0.77-1.02; p < .001). Significant interaction effects indicated that differences between contextualized and decontextualized talk were smaller for ENPs than for measures of lexical density and MLU. The findings of the present study support the hypothesis that decontextualized talk is more structurally complex and lexically dense compared to contextualized talk.

Authors

• Ellen Brinchmann, University of Oslo
  Presenting Author

• Hanne Roe-Indregard, University of Oslo
  Non-Presenting Author

• Jannicke Karlsen
  Non-Presenting Author

• Stefan Schauber, University of Oslo
  Non-Presenting Author

• Bente Hagtvet
  Non-Presenting Author