Automatic Generation of Exercises for Learning Chinese

Description

In recent years, there has been a surge in the number of non-Chinese students in Hong Kong schools. We propose to develop a mobile application that automatically generates exercises for these learners of Chinese as a foreign language (CFL).The exercises are to be generated from online Chinese texts and from the LIVAC (Linguistic Variations in Chinese Speech Communities) synchronous corpus. The application will apply natural language processing techniques to identify sentences that provide good examples of word usage. It will then classify these sentences in terms of its subject area and difficulty level. Lastly, it will turn them into fill-in-the-blank exercises by inserting blanks and generating distractors.This approach has the potential to deliver highly personalized language instruction that is infeasible in the traditional classroom setting. The application will log all activities to analyze what the user already knows and what she finds difficult. Based on this analysis, it can select exercises that are most likely to engage the user and stretch her linguistic proficiency --- exercises that fit the user's reading ability, match her personal interests in terms of subject area, and review unfamiliar or recently learned vocabulary items.In summary, this project will create a "dynamic textbook": a textbook that can be tailored on-the-fly with respect to the ability, learning needs, and personal interests of individual learners. This new type of textbook is expected to be useful not only to non-Chinese students in Hong Kong, but also to CFL learners at large.