Extensive eye-tracking dataset derived from Japanese L2 English learners' text reading

Language processing is a highly intricate human function, and it has been extensively studied in the fields of psychology and education. Eye movement measurement, particularly for written language, has proven to be a beneficial method for language processing.

In recent years, numerous initiatives have been undertaken to collect, organize, and publish large-scale eye-movement data during reading in first and second/foreign languages. However, these efforts have primarily focused on English learners in Europe and its surrounding regions. To date, there has been a lack of focus on text reading among Japanese speakers learning English.

In a study, published in Research Methods in Applied Linguistics, researchers constructed an eye-movement dataset, comprising more than 410,000 words, for Japanese students engaged in text reading in English.

The dataset includes eye-movement data for approximately 10,000 words per individual from 41 Japanese undergraduate and graduate students. These students read passages used for Eiken Grade Pre-2 to Grade Pre-1.

From these data, the researchers computed nine eye-tracking measures for each word in the text, including fixation count, fixation duration, word skipping, and regressions. They conducted various analyses and confirmed the high reliability and validity of the collected data.

The Tsukuba Eye-tracking Corpus (TECO) is now publicly available on the Open Science Framework. This dataset is beneficial for research on reading comprehension and second language acquisition.

Furthermore, it holds potential for natural language processing and artificial intelligence research and is expected to have a considerable impact on future research across a broad spectrum of academic fields.