Reading-Time Annotations for “Balanced Corpus of Contemporary Written Japanese”

Proceedings of COLING 2016, the 26th International Conference on Computational Linguistics: Technical Papers, pages 684–694, Osaka, Japan, December 11-17 2016. Reading-Time Annotations for Balanced Corpus of Contemporary Written Japanese Masayuki Asahara National Institute of Japanese Language and Linguistics, National Institute for the Humanities, Japan masayu-a.ninjal.ac.jp Hajime Ono Faculty of Liberal Arts, Tsuda College. Edson T. Miyamoto Graduate School of Humanities and Social Science, University of Tsukuba Abstract The Dundee Eyetracking Corpus contains eyetracking data collected while native speakers of English and French read newspaper editorial articles. Similar resources for other languages are still rare, especially for languages in which words are not overtly delimited with spaces. This is a report on a project to build an eyetracking corpus for Japanese. Measurements were collected while 24 native speakers of Japanese read excerpts from the Balanced Corpus of Contempo- rary Written Japanese Texts were presented with or without segmentation (i.e. with or without space at the boundaries between bunsetsu segmentations) and with two types of methodologies (eyetracking and self-paced reading presentation). Readers’ background information including vocabulary-size estimation and Japanese reading-span score were also collected. As an exam- ple of the possible uses for the corpus, we also report analyses investigating the phenomena of anti-locality. 1 Introduction Corpora of naturally-produced texts such as newspapers and magazines marked with detailed morpho- logical, syntactic and semantic tags, are often used in human language-production research. In contrast, texts created by psycholinguists exclusively for research purposes, are commonly used in language- comprehension research. We introduce a reusable linguistic resource that can help bridge this gap by bringing together tech- niques from corpus linguistics and experimental psycholinguistics. More concretely, we have collected reading times for a subset of texts from the Balanced Corpus of Contemporary Written Japanese (BC- CWJ) (Maekawa et al., 2014), which already contains syntactic and semantic types of annotations. The goal is to produce a resource comparable to the Dundee Eyetracking Corpus (Kennedy and Pynte, 2005), which contains reading times for English and French newspaper editorials from 10 native speakers for each language, recorded using eyetracking equipment. The English version of the Dundee Eyetracking Corpus is composed of 20 editorial articles with 51,501 words. The Dundee Eyetracking Corpus does not target a specific set of linguistic phenomena; instead, it provides naturally occurring texts for the testing of diverse hypotheses. For example, Demberg and Keller (2008) used the corpus to test Gibson’s Dependency Locality Theory (DLT), (Gibson, 2008), and Hale’s surprisal theory (Hale, 2001). The corpus also allows for replications to be conducted, as in Roland et al. (2012), who concluded that previous analyses (Demberg and Keller, 2007) had been distorted by the presense of a few outlier data points. Our goal is to produce a similar resource that can serve as a shared, available foundation for research in Japanese text processing. Once completed, the corpus will allow us to address two issues that are specific to Japanese. The first issue is related to two types of reading-time measurements commonly used, namely, eyetracking and self-paced reading. …

Content truncated to 500 words