It is easy to imagine machines that can communicate using spoken
natural language. Constructing such machines is more difficult. The
available methods for development of interactive speech applications
are costly, and current research is mainly focused on producing more
sophisticated systems, rather than on making it easier to build them.

This thesis describes how components used in interactive speech
applications can be automatically derived from natural language
grammars written in Grammatical Framework (GF).  By using techniques
borrowed from the field of programming language implementation, we can
generate speech recognition language models, multimodal fusion and
fission components, and support code for abstract syntax
transformations.

By automatically generating these components, we can reduce duplicated
work, ensure consistency, make it easier to build multilingual
systems, improve linguistic quality, enable re-use across system
domains, and make systems more portable.