The Copenhagen Corpus of Eye Tracking Recordings from Natural Reading of Danish Texts

Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 1712–1720 Marseille, 20-25 June 2022 © European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0 1712 The Copenhagen Corpus of Eye Tracking Recordings from Natural Reading of Danish Texts Nora Hollenstein1, Maria Barrett2, Marina Bj¨ornsd´ottir1,2 1 Center for Language Technology, University of Copenhagen 2 Computer Science Department, IT University of Copenhagen nora.hollenstein@hum.ku.dk, mbarrett@itu.dk, marina.bjorns@gmail.com Abstract Eye movement recordings from reading are one of the richest signals of human language processing. Corpora of eye movements during reading of contextualized running text is a way of making such records available for natural language processing purposes. Such corpora already exist in some languages. We present CopCo, the Copenhagen Corpus of eye tracking recordings from natural reading of Danish texts. It is the first eye tracking corpus of its kind for the Danish language. CopCo includes 1,832 sentences with 34,897 tokens of Danish text extracted from a collection of speech manuscripts. This first release of the corpus contains eye tracking data from 22 participants. It will be extended continuously with more participants and texts from other genres. We assess the data quality of the recorded eye movements and find that the extracted features are in line with related research. The dataset available here: https://osf.io/ud8s5/. Keywords: eye tracking, reading, Danish, corpus, psycholinguistics, cognitively-inspired NLP 1. Introduction Records of eye movements during reading have been studied since decades in controlled psycholinguistic studies. The advantages of eye movement data for studying cognitive language processing are well estab- lished. For example, there is a thoroughly examined link between the fixation duration on a word and the cognitive effort required to process this word. During skilled reading, the eyes fixate sequentially through the text but approximately 10–15% of the fixations occur in a previously read part of the text to further process it. Most often, the reader is not aware of this. There- fore, eye movements allow us to study both early and later stages of cognitive text processing (Rayner et al., 1989).1 In controlled studies, participants typically read con- structed individual sentences. Often, pairs of sentences with minimal differences are used to study one particu- lar linguistic phenomenon (Brennan, 2016). However, as highlighted by Demberg and Keller (2019), there is a need for evaluating psycholinguistic theories on natural reading corpora rather than on highly infrequent sen- tence constructions specifically designed for the pur- pose of one experiment. Eye tracking and brain activ- ity studies during unpaced reading of naturally occur- ring, contextualized text have recently gained more at- tention in the research community thanks to advances in recording technologies (Hamilton and Huth, 2018; Sato and Mizuhara, 2018). While natural eye track- ing corpora cannot replace controlled psycholinguistic studies, it is a complementary method of studying some of the same phenomena with higher ecological validity. A naturalistic reading setup allows us to analyze real- 1For a review, see Clifton et al. (2007). time language processing from real-world text (Hasson and Egidi, 2015). An obvious advantage of larger …

Content truncated to 500 words