Extraction of transforming sequences and sentence histories from writing process data: a first step towards linguistic modeling of writing

Abstract Producing written texts is a non-linear process: in contrast to speech, writers are free change already text at any place point time. Linguistic considerations likely play an important role, but so far, no linguistic models of the writing process exist. We present approach for analysis processes with focus on structures based novel concepts transforming sequences, history, and sentence history. The processing raw keystroke logging data application natural language tools allows extraction filtering product be stored hierarchical structure. This structure used re-create visualize genesis history its individual sentences. Focusing sentences as primary building blocks full texts, we aim complement established analyses and, ultimately, interpret timecourse respect structures. To enable researchers explore this view, provide fully functional implementation our open-source software tool visualizations results. report small scale exploratory study German where tool. results indicate both feasibility that actually revise level. latter confirms need modeling production from perspective beyond word