Meta-modeling of Tense and Aspect in a Cross-linguistic Grammar Engineering Platform

This work describes the implementation of a meta-modeling approach to morphologically marked tense and aspect within the LinGO Grammar Matrix customization system. The Matrix customization system creates a small hpsg grammar for a language based on answers to questions about particular linguistic phenomena. In the original system, these answers indicated a choice of pre-determined analyses of the various phenomena. The complexity of tense and aspect systems, the cross-linguistic variation in those systems, as well as the variation in linguists’ analyses of tense and aspect and their underlying assumptions, make tense and aspect difficult to implement in this way. Meta-modeling is proposed as an answer to the question of how tense and aspect can be implemented in the customization system while imposing a minimal amount of analysis on the linguist-user. The meta-modeling approach supports the linguist-user in developing their own analytical model of tense and aspect for a particular language. Within the framework of hpsg, as implemented in the LinGO Grammar Matrix, the meta-modeling of tense and aspect provides flexibility for the linguist-user with regards to the definition of types, hierarchies and features. In addition, this approach allows the user to determine how tense and aspect features relate to specific morphemes as well as how the features contribute to the definition of grammatical constraints.