Developing a Core Vocabulary for a Mandarin Chinese AAC System Using Word Frequency Data

Abstract

Creating a generative language system in augmentative and alternative communication (AAC) for both literate and non-literate users requires three fundamental design steps: analyzing the grammar of the language; developing an efficient, culturally relevant pictorial representation system for encoding the language; and selecting a list of target words to encode. This paper will focus on the problem of selecting vocabulary for the AAC system. This is an important problem because the selection of the encoded vocabulary will ultimately determine the language coverage of the AAC system. The vocabulary list should provide coverage on three fronts: it should contain the most frequently used vocabulary of the spoken language; it should allow description of all the universal, fundamental semantic concepts; and it should meet users' expectations of what they should be able to say on a daily basis, in any given situation. Mandarin Chinese, as a relatively new target language for AAC, presents some unique challenges to the vocabulary designers. For example, the conceptual differences between the Chinese word and character can be difficult to pin down; this highlights the importance of a consistent and scientific approach to vocabulary selection. This paper describes the research process used in the creation of a core vocabulary list for a new icon-encoded Mandarin Chinese AAC system, starting with a basic corpus-derived word-frequency list, and continuing with the modification and supplementation of this list with other linguistic and anecdotal knowledge.