Study of the Validity of Japanese Word Lists for Cognitive Tasks to Evaluate Attentional Bias towards Sleep-related Stimuli Utilizing Japanese Word Vectors

Attentional bias has been described as a characteristic of cognitive information processing in insomnia, in which processing resources are preferentially allocated to sleep-related stimuli. Cognitive tasks (e.g., dot probe tasks and emotional Stroop tasks) that depend on reaction time differences between sleep-related words and neutral words have been developed to evaluate the degree of attentional bias. To develop a Japanese version of attentional bias tasks, Yamamoto et al. (2024) utilized an “Association Frequency Table” to identify candidate sleep-related words. Using the “Short Unit Vocabulary List Database” of the “Balanced Corpus of Contemporary Written Japanese,” they identified candidate sleep-related words, described the appearance frequency of such words, and identified neutral words that appeared at the same frequency as each candidate sleep-related word. As a result, 24 candidate sleep-related words and five neutral words with identical appearance frequency for each of these words (144 words, composed of two kanji characters) were identified. They also performed an impression evaluation for the degree to which each word was related to sleep and confirmed significant impression evaluation differences for sleep-related words compared to neutral words in all lists. The present study aimed to investigate the validity of the word lists created by Yamamoto et al. (2024) from a linguistic perspective leveraging “chiVe” (Sudachi Vector; Works Applications, 2024), a Japanese word vector based on a large-scale language corpus and multigranularity segments, to calculate cosine similarity in semantic relevance between the word “睡眠 (sleep)” with all sleep-related words and neutral words on the lists.