A Social Network Analysis of Discourse in Linguistically Diverse Small Groups

• In Event: Exploring Heterogeneity, Discourse, and Linguistic Awareness in Multilingual Literacy Learning

Abstract

This paper models discourse as a social network to examine how discourse patterns of multilingual small groups change by teacher, lesson type, and grouping (i.e., homogeneous or heterogeneous grouping of multilingual students) during a language and literacy curriculum implementation. Some research on classroom discourse has conceptualized discourse as a network, with a tie representing an utterance directed from one student to the teacher, another student, or the whole group (e.g., Wagner & González-Howard, 2018). These discourse networks could be heavily hierarchical, with teachers leading much of the conversation, or more distributed, with teachers making space for students’ voices. Small group discourse networks might change based on the types of lessons being taught, the teachers, and the group composition (i.e., whether multilingual students are grouped homogeneously with other multilingual students, or heterogeneously with mixed multilingual and monolingual students). This paper examines discourse networks in small-group literacy instruction using the Cultivating Linguistic Awareness for Voice and Equity in Schools (CLAVES) curriculum to see whether multilingual students’ patterns of participation do change based on teacher, lesson type, and grouping.

The participants were a pilot sample of students (N = 18) from a district located in a large metropolitan area who participated in small groups (small groups N = 6) with their teachers (teacher N = 3). The majority of the pilot sample were female (76%) and were identified by parents as Hispanic/Latino (96%). This study uses video observation data from the 3 teachers with their 2 groups on 3 occasions per teacher (18 videos total). These videos capture the teachers delivering lessons at the beginning, middle, and end of study implementation to both linguistically heterogeneous and homogeneous groups. These videos also capture the teachers implementing each of three types of lessons present in the first two lesson cycles of a CLAVES unit (focal text reading, extended language work/multimodal texts, dialogic reasoning). The videos were transcribed and transformed into networks representing patterns of participation in small groups. Each turn at talk was coded as a directed tie from the speaker to the directed recipient of the talk. Turns at talk addressed to the group were coded as a directed tie to all participants. Thus, the resulting network data are edgelists of all speakers and recipients in a given small group observation. I compared teachers, lesson types, and groups (homogeneous or heterogeneous groups) on the out-degree centrality strength (i.e., the number of turns-at-talk) of multilingual students. Preliminary results from a repeated-measures analysis of variance and follow-up comparisons using Tukey’s test indicated that multilingual students initiate more turns-at-talk during discussion lessons (p < .001) compared to other lesson types, that one teacher allowed for more student participation (p < .001) compared to both other teachers, but that there was no significant difference for homogeneous compared to heterogeneous groups (p = .7518). These findings indicate teachers and lesson structure both impact how multilingual students participate in small group discussions. This study also highlights the potential of using social network approaches to provide insights about discourse data.

Authors

• Andrew Weaver, Marist University

• Michael J. Kieffer, New York University

• Patrick Proctor, Boston College

• Elise Cappella, New York University

• Yoav Bergner, New York University