History of ICALL

Most of this part was taken from Levy, Michael, 1997. As opposed to the previous chapters, from here on the focus is on Intelligent Computer-Assisted Language Learning (ICALL). Even though there may also be references to traditional CALL here the part about ICALL starts.

Prototypical ICALL-programs were developed as early as the late 70s, e.g. Weischedel, Ralph M. ; Voge, Wilfried M. et al., 1978. These programs ran on large mainframe-computers in the computing-labs of the universities. One of the large projects of the last century was the ATHENA Language Learning Project (ALLP) at the Massachusetts Institute of Technology (MIT) which started in 1983 (Murray, Janet H. in: Holland, V.M. ; Kaplan, J.D. et al. (Ed.), 1995). In the project the method of communicative language learning was to be applied to a computer program. Three main technologies were used to develop the system: Natural language processing, speech processing and interactive video. One part of the material presented to the learner was a story about an amnesiac Columbian scientist. Interactive video was used to present some part of the story and then the learner had to complete the information through interactive dialogs with characters from the plot. Another task was the communication with a Poltergeist, who could be orderd to clean or mess up a room according to the instructions given by the learner. This included not only natural language parsing, but also discourse processing (focus) and a model of the world, which could be altered according to given limits. It has to be mentioned, that the ALLP was only in certain parts used with language learners for an evaluation. More information about this specific system can be found in the chapter » Example System I: ALLP.

Another project, which was developed a little later, was FLUENT-1 and FLUENT-2 by Hamburger, e.g. Hamburger, Henry in: Appelo, L. ; de Jong, F. (Ed.), 1994 and Hamburger, Henry in: Holland, V.M. ; Kaplan, J.D. et al. (Ed.), 1995. Here a similar approach was chosen, in that the learner was to clean up a kitchen, in which a graphical hand moved according to the learner's input. The input could be done in natural language or via moving the graphicals. The concept behind this system was to integrate the visual approach with the language approach to 'immerge' the learner into the situation without recursion to grammar, vocabulary or mother tongue. More information about this specific system can be found in the chapter » Example System II: FLUENT.

According to Levy, Michael, 1997 one problem of these systems was, that they were only tested in artificial situations, where the developers tried to forsee the reactions of the students. More recent developments try to incorporate more "real" data to make the systems capable of handling data, which occurs in real language learning situations e.g. the ICICLE-project. Even though the mentioned programs ALLP and FLUENT were largely prototypical systems I do not know, why e.g. the mentioned commercial program "Herr Kommissar" (deSmedt, William H. in: Holland, V.M. ; Kaplan, J.D. et al. (Ed.), 1995) is not available anymore, even though it seems to have been used successfully.

As far as I know there is not a single "intelligent" language learning program on the market available in Germany right now. The only true advancement which can be identified is that there seems to be a trend to incorporate more multimedia capabilities into the programs which supports the multi channel requirement. The disregard of methods of computational linguistics might also be due to the increased usage of the internet to stimulate the general use of computers in language learning. Some programs now include sections, where single words can be spoken into a microphone and the program tries to evaluate the correctness of the pronunciation. However this should probably not be called an "intelligent" program, since these programs do not include any mechanism to identify words of a language as e.g. speech recognition software does.

It seems worth to mention in this chapter, that resources and linguistic tools have become a valuable part of the concepts in CALL and also ICALL. In the 1980ies computers were not fast enough and/or did not have enough memory to store and provide language resources like lexica and corpora. This has changed dramatically in the last few years and is likely to continue. There have been a number of concepts being developed to use resources such as WordNet or CELEX as well as corpora for language teaching/learning purposes.

In recent years several programs have been developed in research projects which on the one hand include techniques from CL but on the other hand are actually used in language teaching, such as the E-Tutor and Glosser. These projects demonstrate that computers now have matured enough to be able to run complex NLP-applications.

An important external link for more information is The history of CALL, even though it contains a lot of information about the general development of IT and ends around the year 2000. All important research initiatives during the last decades are mentioned and set into context with language teaching and IT-development.

In the next sections two systems are presented in more detail to give an idea of the advantages and diffculties which show up in the development and usage of ICALL-systems.