This three-paper dissertation uses social network analysis to examine the literacy development and small group discourse of multilingual students in Grades 4 and 5. The data come from a randomized-controlled trial testing the effectiveness of homogeneous and heterogeneous small groups for English learners. The first paper tests whether two types of social relationships, and how multiple languages are used in those relationships, are associated with reading comprehension directly, or indirectly via word recognition/decoding and English language comprehension. This paper uses a sample of multilingual students (N = 249) screened for participation in the larger study, and examines hang-out ties (i.e., who students like to hang-out with) and classwork ties (i.e., who students prefer to work on classwork with). Path model analyses indicated that the number of classwork ties, and the proportion of both hang-out and classwork ties that used multiple languages, negatively and indirectly associated with reading comprehension via English language comprehension. In the second study, I examine whether the number of hang-out ties among students participating in a small group literacy curriculum predicts the later reading comprehension of multilingual participants. I also test whether groups with low and high numbers of hang-out ties differ in interaction quality. This study uses a sample of multilingual students (N =143) participating in the larger study. Autoregressive path models indicated that the number of group hang-out ties negatively predicted later reading comprehension. Exploratory analyses of interaction quality suggested that teachers ask more open-ended questions and students give more elaborated responses in groups with low numbers of hang-out ties. In the third study, I model small group discourse as a social network to examine how discourse patterns differ by teacher, time, lesson type, and grouping. This paper uses a sample of linguistically diverse students (N = 18) participating in small groups using the same language and literacy curriculum. I transform transcripts into network edgelists of speakers and intended recipients, which I use to create visualizations and to estimate network statistics. Results indicated wide heterogeneity, with some groups adopting the dialogic principle of the curriculum quickly and others doing so by the end of implementation. Across all three papers, results suggested that social ties do matter for language and literacy learning of multilingual students, and that social network analysis can shed light on the important social factors that predict literacy. [The dissertation citations contained here are published with the permission of ProQuest LLC. Further reproduction is prohibited without permission. Copies of dissertations may be obtained by Telephone (800) 1-800-521-0600. Web page: http://bibliotheek.ehb.be:2222/en-US/products/dissertations/individuals.shtml.]