Research in science education echoes that students come to science classrooms with everyday conceptions that may differ from school scientific views (Harlen, 2003; Driver, 1985). In order to develop scientific views among learners, it seems essential for the science teacher to provide opportunities to students to express their existing schemas during classroom talk. Mortimer and Scott (2003, p. 1) point out, the "key feature of any science lesson (...) is the way in which the teacher orchestrates the talk of the lesson, in interacting with students, to develop the scientific story being taught" (italics in original). However, many researchers such as, Treagust and Duit (2008), Mortimer and Scott observe that systematically utilization of the learning potential of ‘talk' in science classrooms has been "somewhat neglected" (p. 2). This is the case as research on teachinglearning in science primarily focuses on teachers' views and instructional practices without exploring the main tool by which these practices are implemented: classroom talk. Through this work, I tend to (a) examine the patterns of classroom talk in an educational context of bilingual secondary science classroom, (b) find out various ways in which differing patterns of classroom talk enable the development of scientific epistemic skills among learners, and (c) establish "discourse-cognition" (Westgate & Hughes, 1997) relationships in a science classroom. This study took place in three urban schools located in New-Delhi, India. Three teachers were involved in the study who taught secondary classrooms science curriculum in their respective schools. The average class size was 40 students per class. Qualitative methods of data gathering, such as, classroom observations, semi-structured interviews with teachers were used. The lessons were audiotaped or videotaped or both. Using a ‘talk agenda' analytical framework developed in this study, patterns of discourse were identified. It is followed by coding the discourse moves with respect to the cognitive categories which represented the students' epistemic skills expressed within the classroom discourse. It was observed through the study that "dialogic" moves in classroom discourse inadvertently supported the growth of epistemic functions in science classroom. It is argued that embedding dialogicity in classroom discourse is neither an optional nor an extra aspect of science teaching-learning rather it is mandatory for development of conceptual as well as epistemic skills among learners. Appropriate repertoire of skills for enhancing dialogic discourse in science classrooms is identified for teachers.