Interactive learning and natural language systems

In this paper lnteractive Learning is presented as a research application to apply and
experiment with N atural Language processing.
Computer-Assisted Instruction (CAI) systems are useful and applied successfully in
many domains now. CAI systems provide instruction in a restricted knowledge
domain mostly without using a sophisticated Natural Language (NL) interface. The
CAI interface only supports the use of predefined natural language utterances that are derived directly from the knowledge domain.
The biggest problem with applying natural language processing tools to CAI is the
lack of a direct link between the architecture of the NL interface and the architecture
of a CAI system. This paper is a proposal for solving it.
There are two approaches presented. One approach focusses on defining a NL
interface on top of a CAI system. The other approach focusses on Interactive
Language Learning.
The first approach stresses that applying Natura} Language dialogues to CAI
impoverishes the way instruction is provided by the system, unless the system is
based on discourse planning.
The second approach stresses that Natural Language processing benefits CAI systems
for Language Learning. It offers a learning environment for different language skills,
based on a variety of naturallanguage processing tools.