Paper short abstract:

This paper explores the question of ordering kanji using artificial intelligence method, a collection of decision-based algorithms called an expert system. This research will present the expert system for ordering kanji according to chosen criteria, for purposes of linguistic analysis or education.

Paper long abstract:

Kanji can be observed in various ways: as a logographic script, as vocabulary items, or as a complex system which can be analysed and ordered. In this paper, kanji will be analysed as a collection of components, readings and statistical information of their frequency in various sources. Using that data, I will present the desicion-based algorithm for ordering kanji for various purposes based on user's input.

I purport that the human-made kanji orders such as on-reading order, grade school order, JLPT test, or simple one-source frequency orders are not optimised or useful for either educational or research purposes. A script as complex as kanji needs to take into account more than one criteria to produce an optimised order. I constructed an expert system that bases the decision an all mentioned criteria, and weighs them differently according to user's priorities.

The system is based on topological sort algorithm with customisable weights assigned to each character. The weights are affected by character frequency, components or radicals, position in grade school order and other commonly used orders, and usage in common textbooks. The frequency of a character is computed using various sources (newspaper, wikipedia, literature and Twitter) since it varies according to the source.

The user is asked to chose relevant criteria, e.g. optimisation for university level study; optimised order of JLPT N3 characters which are previously unordered; or optimised order of cross-section of two popular textbooks. In addition to that, the custom criteria can be added. The resulting order can be used for construction of teaching materials, self-study, and various kinds of linguistic analysis.

Paper short abstract:

In this paper, we present an interdisciplinary project dedicated to the creation of a bilingual English Japanese language learning environment. The project incorporates the expertise of specialists in app development, natural language processing, linguistics, digital humanities and psychology.

Paper long abstract:

In psychology and linguistics, the mental lexicon is defined as information on the meaning, syntactic features, pronunciation, and sociolinguistic knowledge humans have about words. Previous research (De Deyne, Verheyen, and Storms 2016) suggests that word associations are an effective tool to measure the mental lexicon, in a way that complements text-corpus based resources. The ability to collect word associations at a large scale provides new ways to measure meaning in the mental lexicon and study how language shapes and is shaped by our mental representations. Such topics as demographic-dependent differences in language use (Garimella, Banea, and Mihalcea 2017), lexical centrality, and semantic similarity (De Deyne et al. 2019) are investigated with word associations as the main material for the studies.

In this paper, we present an interdisciplinary project dedicated to the creation of a bilingual English Japanese language learning environment based on the Small World of Words project, a large-scale crowd-sourced word association project in 17 languages that aims to measure common sense word meaning. The project is conducted as a collaboration between researchers and developers at the University of Tokyo, Japan, and the University of Melbourne, Australia. It incorporates the skill and expertise of specialists in app development, natural language processing, linguistics, digital humanities and psychology.

The learning environment is currently under development, a prototype beta version was launched in 2022. The environment aims to provide research-driven, clear, and concise visualizations to learn about what meaning is shared and what is unique in English and Japanese. It uses word association data from the Small Word of Words project in Japanese and English. It highlights what associations are common across speakers of these languages and which are language-specific by aligning word association networks in these languages. We expect this project to be of interest to language teachers and learners as well as researchers interested in comparative analyses of mental lexicon across languages. In future releases, new words will be added and the project will be extended to other languages covered as a part of the Small Word of Words project.