We believe we can promote the positive effect of picture book reading by finding a picture book that is suitable for children's interest and development. For this purpose, we are conducting researches such as similarity search algorithm that can be applied to various features such as words and pictures, readability estimation method based on textual features such as age of acquisition of each words, and morphological analysis for hiragana-dominant picture book sentences. Based on the findings obtained from these researches, we have developed a picture book search system "Pitarie."
More than 20 years ago, NTT conducted psychological experiments to investigate word familiarity and thus construct a Japanese lexicon of about 77,000 words. This lexicon is still used in many fields. Now, we reinvestigated and reconstructed an unparalleled-scale Lexicon of 163,000 words through crowdsourcing. By applying careful screening, we succeeded in obtaining highly reliable results. This makes it possible to make comparisons with results of 20 years ago. Based on this, we created the Reiwa edition of a vocabulary-size estimation test. Furthermore, this lexicon allows us to estimate the vocabulary size appropriate for the present day. We are working on vocabulary-size investigations for a wide range of ages, including elementary school to high school students. In the future, we will investigate and analyze the relationship between vocabulary size, reading comprehension, and academic ability, aiming to achieve effective educational support.