Language Support Models for AAC

There are multiple areas of AAC (Augmentative and Alternative Communication) where programmatic language processing can provide value. Here we address two of those areas: word identification and basic inflection. By providing standardized datasets we hope to make it easier for AAC systems to support language-aware solutions for multiple locales. These standards are defined in the words & inflections documentation.

By keeping a dataset of common words in a language, AAC systems can automatically colorize buttons based on parts of speech, offer alternative word forms (verb tenses, pluralization, etc.), and more effectively handle advanced features like automatic or popup inflections. (see an example here - words-en.json)

By tracking for common inflection rules, AAC systems can automatically offer more grammatically correct sentences for users. (see an example here - rules-en.json)

We need more languages! If you know and are willing to contribute to a language file, please join our chat group at OpenAAC.org and we can give you directions how to contribute!

Below is a simple inflections lookup tool you can use to try out the words and rules datasets for yourself. You can enter a word (i.e. "think"), a prior ("tell me if he"), or both and the inflections library will suggest a possible inflection if it matches any of the defined rules. Note that in ambigous cases, where multiple inflections could be correct, a judgment call has to be made and either one "more common" inflection will be proposed, or none will be.