This proposal focuses on a specific field of educational research: the analysis of interactional processes in hypertextual online learning environments. These processes can be taken as e-learning, as the interaction of users with the content of an online-learning environment. In general these interactional processes are described by the nautical metaphor of „navigation“ as the linear unfolding of a non-linear hypertext (Kuhlen 1991). Exactly this unfolding is pronounced by characterizing hypertext as a pull-medium. By selecting specific links from a multitude of possible links, a navigational sequence or path is established. On this account hypertextual learning environments can be referred to as multisequential or multilinear (Landow, 1997). The multiplicity of hypotheses about interactional processes and especially about processes of e-learning conflicts with a general lack of quantitative methods for its analysis. So the main research question is how to analyze these interactional processes: How can patterns, regularities and structures of interactional processes be identified? How can these processes be interpreted in a pedagogical perspective, i.e. against the background of general learning strategies? On the one hand this proposal presents a quantitative and exploratory methodology for analyzing sequences by means of optimal-matching in combination with methods of cluster analysis. The overall aim of optimal-matching is the identification of patterns, regularities and structures in sequences (i.e. user – learning-environment interaction).(see below, methods). On the other hand results of an empirical study are presented to demonstrate the potential of the described approach. This study refers to on an optimal-matching analysis of 1500 sequences (including 4700 informational units), which represent the usage of a hypertextual online learning environment. In general, the analyzed interactional processes are are interpreted as autodidactic acting (autodidactics). This term stresses the specific navigational path of different users as the succession of selected informational units. In this perspective autodidactics is defined as the transformation of temporal figures into spatial figures: Learners acquire knowledge by developing a non-linear mental model of an specific issue based on their temporal-linear interactional process. The pros and cons of the presented approach will be discussed and areas of application will be outlined.

The presented approach to analysis of interactional processes is based on the method of optimal matching in combination with methods of cluster analysis. The optimal-matching algorithm is well known in applied and natural sciences (i.e. molecular biology) but is hardly used in social sciences. Optimal Matching (see Kruskal 1999) starts with a pairwise comparison of every interactional sequence of the dataset. The distance of two sequences as the degree of similarity is determined depending on the number of operations (insertion, deletion and substitution of elements ) to achieve an alignment of origin and target sequence (levenshtein-distance, see Levenshtein 1966). According to Erzberger (2001) the exploratory-heuristic sequence analysis by means of optimal matching can be characterized as an exploratory and heuristic, quantitative and case-oriented, inductive approach that allows the identification of typical patterns in empirical sequences, without using theoretical models or reference sequences From a technical point of view the interaction of users with online environments can be reconstructed by server-sided log files. As a specific potential Priemer (2004) describes the process of collecting log data as unobtrusive, detailed, objective and non-reactive. Priemer emphasis that log data refers to the action of users in authentic situations (see Webb et al.1966). In contrast to common analyses of aggregated log file data the analysis of elearning processes is based on sequenced log file data. In general the approach to analyze interactional sequences by means of optimal matching follows the research tradition of Flanders Interaction Analysis Categories (FIAC). Flanders (1970) focuses on the observation of dyadic teacher – pupil interaction in primary school classrooms. The described approach differs in respect to the underlying methodology: complete interactional sequences are analyzed by optimal matching in contrast to the analysis of dyadic probablitiies of teacher behavior and pupil reaction by markov-chains.

The analysis of sequenced logdata within the scope of optimal-matching starts at a point, where the analysis of aggregated logdata ends: the interactional process as a sequence. By this hypotheses of navigational processes and their effects are becoming object of empirical educational research. Moreover the present and dominant focus on the result of e-learning can be put into perspective. This focussing on the outcome of navigational processes and at the same time suppressing of the whole process is dissatisfactory from a pedagogical point of view: processes of learning and formation are a fundamental topic of pedagogy and knowledge about the processes allows a multitude of pedagogical practices and interventions, i.e. the pedagogical design of an online learning environment and the support of learners. In general the analysis of e-learning processes by means of optimal-matching refers to the heart of didactics, which analyses the empirical transformation of spatial figures into temporal figures (didactics) as well as the transformation of temporal figures in spatial figures (autodidactics) and at the same time reflects these processes in regard to appropriateness and alternative possibilities. In particular hypertext as the fundamental technology of online learning environments implies a radical structural change, where existing processes of transformation are up for discussion. Analyzing navigational processes in an online learning environment which is based on the ontology of Web-Didatics (Meder 2006) about 1500 navigational sequences including about 4700 informational units where analyzed (Iske 2007). As a result specific navigational patterns and structures were identified, i.e. linear patterns of „exploration“ and „examination“, as well as non-linear patterns of direct and selective navigation. These patterns as specific strategies can be interpreted as „focussed on explanations“, „focussed on examples“ and „focussed on tests“. Besides theses specific patterns especially the diversity and plurality of empirical navigational paths become apparent.