Using Multimodal Discourse Analysis to Identify Patterns of Problem Solving Processes in a Computer-Supported Collaborative Environment

Introduction

The proliferation of technologies designed to provide learners with an immersive and enriched learning environment presents new challenges for practitioners in understanding the behavior of learners. In parallel, the ability to collect complex datasets (including audio, video and screen capture, for example) means that researchers have the information to answer these complex questions. What is needed is a process of visualizing and analyzing multiple sources of data, from complex instances of learning. Given the progress in the theories of how and what people learn, particularly in fields such as the learning sciences, what is needed is to draw from traditionally opposing perspectives to establish a more systematic approach to understanding these complex learning environments. In order to appropriately answer questions at this level, complex analytical procedures, with multiple coding schemes (Evans, Feenstra, Ryon, & McNeill, 2011) and theoretical frameworks (Suthers et al., 2011) need to be applied. Knowledge of learners’ behavior while using innovative tools can inform the design of the tools as well as the tasks.

In research that examines technology-mediated learning, more traditional forms of measurement, such as pre-and post-tests, surveys and interviews, have provided the base data (Barab et al., 2009; Jacobson, Lim, Lee, & Low, 2007; Ketelhut & Dede, 2006). These studies provide minimal information about the processes of learning, what learners are doing, saying and responding to in an immersive environment. Reporting on the processes of collaboration and the development of theory and methods of analysis has seen this area develop in the last five years. Reporting on multiple measures of process using group decision making and discourse analysis in computer supported collaborative learning (CSCL) is becoming more common (van Aalst, 2009; Evans et al., 2011; Weinberger & Fischer, 2006). One advantage of the use of multiple measures is the ability to report on interaction effects (Weinberger & Fischer, 2006) and relationships between data at different levels, particularly individual contributions to group processes (Ding, 2009). Discussion of how best to analyze and describe the relationship between the individual processes of learning within a collaborative process is ongoing (Friend Wise, Hsiao, Marbouti, Speer, & Perera, 2012; Clara & Mauri, 2010). However, it is generally agreed upon that there is an interaction between the individual, small group, and meso levels of collaboration (Goodyear, Jones & Thompson, forthcoming).

The use of Systemic Functional Linguistics (SFL) has been prominent in computational linguistics as a means of analyzing text-based corpora. In this area, there has been a growing interest in the role of features of language, such as modality, negation and cue words, which complement the already robust body of theoretical studies in this area (Gravano, Hirschberg, & Beňuš, 2012; Moreante & Sporleder, 2012). As Moreante and Sporleder (2012, p. 223) explain, the “emergence of this area is a natural consequence of the consolidation of areas that focus on the computational treatment of propositional aspects of meaning, like semantic role labeling, and a response to the need for processing extra-propositional aspects of meaning as a further step towards text understanding.”