Writing and revising : Didactic and Methodological Implications of Keystroke Logging

Abstract: Keystroke logging records keyboard activity during writing. Time and position of all keystrokes are stored in a log file, which facilitates detailed analysis of all pauses, revisions and movements undertaken during writing. Keystroke logging further includes a replay function, which can be used as a tool for reflection and analysis of the writing process. During writing, writers continuously plan, transcribe, read, and revise in order to create a text that meets with their goals and intentions for the text. These activities both interact and trigger one another.This thesis includes studies in which keystroke recordings are used as bases for visualisation of and reflection on the cognitive processes that underlie writing. The keystroke logging methodology is coupled with Geographical information systems (GIS) and stimulated recall in order to enhance the understanding of keystroke logged data as representations of interacting cognitive activities during writing. Particular attention is paid to writing revision and a taxonomy for analysis of on-line revision is proposed. In the taxonomy, revisions made at the point of inscription are introduced as ‘pre-contextual’ revisions, and highlighted as potential windows on cognitive processing during transcription. The function of pre-contextual revisions as revisions of form and concepts was ascertained in an empirical study, which also showed that 13-year-old writers revised more form and concepts at the point of inscription when they wrote in English as a foreign language (EFL) than in Swedish as a first language (L1).In this thesis, a learning method, Peer-based intervention (PBI), is introduced and examined through case studies and statistical analysis. PBI is based on theories about cognitive capacity, noticing, individual-based learning and social interaction. In PBI, the keystroke-logging replay facility is used as a tool for reflection on and discussion of keystroke logged data, i.e. representations of cognitive processes active during writing. In the studies presented in this thesis, teen-aged and adult writers’ texts, written before and after PBI, were analysed according to text quality and revision. Descriptive and argumentative texts in both L1 and EFL were included in the studies. The results showed that PBI raised adult and teen-aged writers’ awareness of linguistic and extra-linguistic features, and that the effect varied across levels of learner ability, text type and language.