This chapter will address both theoretical and methodological issues in pause analysis. First, the theoretical assumptions concerning how pause analysis can contribute towards our understanding of written language production are discussed. Subsequently, some measures, methods and tools for automatic pause analysis are presented and some empirical analyses are given. Most of the analyses and tools presented will be based on the concept of a “micro-context”. A micro-context is defined here as the context around a certain transition between two keystrokes. By categorising micro-contexts in terms of categories such as “within word”, “between words”, “between letter and punctuation mark”, pause frequencies and durations in certain types of micro-contexts can easily be analysed automatically and we can quickly gain an impression of how writers use their pauses. Finally, the complexity of pause definitions and operationalisations of the same are addressed. In typing, each letter is produced as a discrete unit and each transition between two keystrokes is a possible candidate for a pause because there will always be a short inactivity between keystrokes. However, obviously not every transition should be defined as a pause. Intuitively, a pause is a transition time between two keystrokes, which is longer than what can be expected for the time needed to merely find the next key. To make a pause writers have to “interrupt” their typing considerably longer than the “normal” transition time between two keystrokes. The problem is how to define a good pause criterion, taking into account the different typing speeds of individual writers.
Writing is a dynamic activity. In order fully to appreciate that fact, we have to analyse writing as it unfolds in real time. The present paper gives a bird’s eye perspective on the budding paradigm of using keystroke logging as a window on the online process of writing. First, methods and tools for designing writing experiments, recording writing activity, and for analysing recordings of writing activity are reviewed. Second, an overview of different lines of research is presented. Here, rather than attempting an exhaustive review of the state-of-art, we present selective illustrations from four areas of research where keystroke logging by means of the research tool ScriptLog was used as a basis for the analysis of writing: the contrastive study of speech and writing, the study of learning to write, the study of writing difficulties, and the experimental study of spelling. Also, the combination of keystroke logging with eye-tracking technology is illustrated. The paper ends with a discussion of future directions for research and applications, including test, diagnosis, evaluation, writing support, and a new generation of corpora of computer-logged writing and their use in research and education.