Language Processing and Intelligent Computer-Assisted Language Learning

Introduction and Motivation
In the introduction the question of what the advantages of computers in language learning and teaching are, is answered and the content of the following chapters is described briefly. Learning Goals are the fundamental advantages of using computers in a learning scenario.

Overview
The following parts contain general remarks about the way language learning should be taught based on didactics research and how this is connected to using computers in language learning. The second part is about the ICALL-research, which has taken place over the years. There have been various important initiatives to develop intelligent language learning systems, but the outcome of these projects rarely managed to surface as a commercial product. Two example ICALL-systems are presented in more detail. Learning goals are a general "feel" for the issues involved in the didactics of CALL and a glimpse into the history of ICALL.

Categorization and Evaluation
In this chapter a categorization scheme for the various types of programs which can be used for language learning and teaching is suggested. This is (at first) done without regard of methods of CL and AI. The next chapter on Functionality will build on this structure. In the second part some aspects of program evaluation are explored. Both aspects are important as most systems are not complete language trainers but focus on certain aspects of language learning. This applies to ICALL systems even more so.

Functionality
A functional perspective on CALL-systems allows an assessment of the possibilities for CL integration. It seems clear that the ultimate goal of applying CL methods should be the improvement of functionality. In taking this perpective two main areas of application can be identified. Either the program may be improved without considering language in the first place or the "content" can be improved by using CL methods for the analysis of language data either from the learner or for the learner.

Error-Analysis
One of the main areas of applying CL methods to CALL has been the analysis of learner input. The idea is that the "deep" linguistic analysis of the input allows a precise characterization and adequate feedback of the errors the learner has possibly made. This has been tried on almost all linguistic levels from phonology to semantics. Additionally the deep analysis may allow an advanced type of exercise with almost free formed learner input. In this chapter the use of statistical methods such as the mentioned LSA for an evaluation of learner input is excluded.

Intelligent Tutoring and User Modelling
In order to provide an individualized learning environment the CALL systems needs to be able to adapt the system to the learner. This is done on some general assumptions about language learners in general and also on specific information collected during the interaction of the learner with the system. A number of program modules are suggested to process the information from the learner, to draw conclusions from it and to adapt the system accordingly. A central aspect is the learner's model and the possibilities of adaptation.

WWW-accessible programs
In this chapter three example systems are listed. On the one hand the all of them are accessible via the WWW and on the other hand they represent the different categories used in the previous chapters. Ther parser from the PromisD project is used in a tutorial system. The LogoTax system can be considered a tool as it serves as a personal electronic lexicon and finally Glosser is an information system providing data on a fixed set of texts.