Language in real time: What can process corpus data reveal about cognition?
(2026) 10th international conference of the French Cognitive Linguistics Association (AFLiCo10) — Location: Campus Condorcet Paris – Aubervilliers, France (22.June.2026)
Files
No attached file found for this publication.
Details
- Authors
- Abstract
- The linguistic study of cognitive phenomena has mostly relied on two types of data: constrained data of the experimental type and naturally-occurring data taken from corpora. Corpus data have the advantage of presenting a high degree of ecological validity. In addition, it is possible (and relatively easy) to collect corpus data from a large number of subjects, which enhances the reliability and generalizability of corpus-based findings. This had led to the popularity of cognitive corpus linguistics (Grondelaers et al. 2007). Despite their obvious advantages, it should be acknowledged that corpus data can only provide indirect information about cognition, which very much remains a black box (Pederson & Nuyts 1997). For example, frequency of occurrence in corpora has often been used to make claims about cognitive entrenchment (e.g. Langacker 1987) or mental storage (e.g. Durrant & Mathews-Aydinli 2011), but such claims are still largely speculative. Part of the speculation comes from the fact that corpora tend to represent the final product and do not say anything about how language is actually produced. In this presentation, it will be argued that corpus data showing language production in real time help come closer to the reality of cognitive phenomena. The focus will be on second language acquisition and on writing process data, which represent the writing process as it unfolds. Illustrations will be taken from the Process Corpus of English in Education (PROCEED; Gilquin 2022), a learner corpus including, in addition to the final texts, keylog files and screencast videos reproducing the writing process. It will be shown how the process data from PROCEED can shed light on cognitive phenomena typical of second language acquisition such as errors, avoidance or automaticity. A case study will explore how pausing in writing can be relied on to tackle the issue of holistic storage of multiword units (see Gilquin 2024). The presentation will also briefly demonstrate how additional data included in PROCEED, namely eye-tracking, stimulated recall and metadata, can complement the keylog files and screencast videos, before highlighting some caveats and offering some concluding remarks.
- Affiliations
Citations
Gilquin, G. (2026, June 23). Language in real time: What can process corpus data reveal about cognition? 10th international conference of the French Cognitive Linguistics Association (AFLiCo10), Campus Condorcet Paris – Aubervilliers, France. https://hdl.handle.net/2078.5/277797