Supporting Language Learners by Intelligent and Efficient Use of Technology

Introduction
CALL is an interdisciplinary research field, which exploits the use of computer technology for language learning. A survey of CALL systems (Gamper & Knapp, 2002b) revealed the increasing tendency of language learning systems to adopt computational linguistics, artificial intelligence and network technologies in order to improve the learning process. However, many systems are driven more by technological possibilities than by a proper analysis of the problem to be solved. What is needed is a more integrated and comprehensive approach to language learning, which considers didactic requirements and uses different technologies that are tailored for the training of specific skills. Clearly, developing such a system is not an easy task: a lot of expertise out of different disciplines as well as time and money are needed. However, the development process can be facilitated if methods are found to join independently developed mature products to a single learning environment.
In this paper, the results of a collaborative project of the European Academy of Bolzano and the Scuola Universitaria Professionale della Svizzera Italiana (Funded by the European Union, Interreg IIIA, Italia-Svizzera) are presented, which involves combining the systems of Word Manager (WM) and ELDIT. WM is a system for reusable morphological dictionaries. Its lexical resources currently include morphological and orthographic knowledge for three languages, namely English, German and Italian (for German see http://www.canoo.net). ELDIT, the "Electronic learners' dictionary for German and Italian" (see http://www.eurac.edu/eldit) is a flexible language learning system easily enlargeable by external products. As it turned out, ELDIT could profit from the WM components in several ways. In this paper we describe one of the new features that resulted from the collaboration, namely how we can provide the entire inflection paradigm for each single word in the ELDIT dictionary. The flexibility of ELDIT, the reusability of the WM modules (WMTrans modules), and the resulting creation of an innovative language learning environment are also described.
Word Manager - A reusable lexical database
Work on the Word Manager project is carried out by an interdisciplinary team consisting of linguists, computational linguists and computer scientists. In Word Manager, lexical resources are developed in two stages. In the first stage, the morphological system of a language is described. This results in a morphological rule database, a description of the inflectional classes and the word formation processes of the language. In the second stage, lexemes are entered, and classified through the morphological rule database. The result is a reusable lexicon database. On the basis of this lexicon database, specialized tools can be developed for any individual task using the knowledge made available in the database. Ten Hacken & Domenig (1996) describe the general architecture and the rule types involved.
ELDIT - a Web-based Learners' Dictionary
At the European Academy Bozen-Bolzano an interdisciplinary research team consisting of linguists, language teachers and computer scientists is developing a Web based language learning system. Its core module is a union of a German and an Italian learners' dictionary (Abel & Weber, 2000). The information is organized in a modular way as far as system development and presentation are concerned. On the left-hand side the lemma together with different word meanings is shown. On the right-hand side a tab metaphor is used to illustrate the use of the word, word relations, etc. The tab "Verwendung" enabled in figure 1 shows typical collocations, which are described by a pattern, a translation equivalent, a lexicographic example, and optional explanations or linguistic hints. The basic architecture of the system is a client-server model exploiting the standard Internet and WWW protocols (Gamper & Knapp, 2002a). The server is implemented using Java Servlet technology, a method that allows the dynamic creation of HTML pages. Java APIs can be easily integrated into an existing system. On the client side, the dictionary can be accessed by any Web-Browser.
Figure 1: Screenshot of the Italian dictionary entry "Haus" (house) in ELDIT
Providing Inflection in ELDIT by reusing Word Manager
Conventional dictionaries in paper or electronic form usually provide the inflection paradigm for some example words. The learner should then be able to apply the rules to similar cases and notice or learn irregular cases. This may work for someone intentionally studying a language, but often causes problems when a learner just picks up an unknown word in a dictionary. The small hint that the word under consideration is irregular may be overseen, or - if noticed - it may be a lot of work to derive the appropriate form from the list with exceptional rules and forms.
One way ELDIT profits from being joined with WM is that it is now able to provide the entire inflection paradigm for each single dictionary entry. The modular structure of ELDIT and the tab metaphor made it easy to include the new module. In the interface a tab was added where the inflectional information is shown., At the server side, WM dictionaries were integrated in the form of so-called WMTrans Finite State Transducers (WMTrans). A WMTrans is produced by compiling the WM lexicon entries from a WM dictionary database into a finite state transducer (Pedrazzini, 1999; 2000). Currently, a whole range of WMTrans products have been developed (see http://www.canoo.com/wmtrans): WMTrans word recognizers, lemmatizers, analyzers and generators for inflection and word formation. As an example we describe the WMTrans Lemmatizer: a module that returns the citation form of any valid word for a specified language. The result of a query is a list of citation forms followed by the corresponding category:
query -> ging result -> gehen (Cat V)
In order to integrate explicit information on inflection in ELDIT, we needed the WMTrans inflection generator, which is able to generate the whole inflection paradigm starting from a citation form. The information is returned as a list of word forms, followed by a list of morphosyntactic features related to each single word form.
The transducers have been implemented in Java and can hence be easily integrated into our Java-Servlet program through a class API. When a user wants to retrieve the inflection of a particular dictionary entry, he or she activates the tab "conjugation" or "declension". In this way the client sends a request to the server. The request is processed in several steps. First the requested dictionary entry is identified. Then the inflection paradigm is retrieved from the corresponding Word Manager Transducer. Finally, the HTML answer page containing the inflection data is dynamically compiled. The presentation of the information is realized by CSS. The screenshot in figure 2 shows the inflection paradigm for the German word "Haus" (house).
Figure 2: The inflection paradigm of the German word "Haus" (house)
Conclusion
A lot of knowledge out of different disciplines is necessary to provide an innovative language learning system. A system can greatly be improved, if methods are found to join independently developed mature products to a single learning environment. We have presented two interdisciplinary research projects out of the domain of lexicography, which are both ready to be integrated within other e-learning resources. Word Manager is a reusable lexical database that provides information about inflection, word formation and orthography. ELDIT is a web-based dictionary for language learners. Both systems can be integrated into other learning environments as well as easily extended by new modules. Joining the two products efficiently supports the creation of innovative features in a language learning environment.
References
1. Abel, A. & Weber, V. (2000): ELDIT - A Prototype of an Innovative Dictionary. In: Heid, U. et al. (Eds.): Proceedings of the Ninth Euralex International Congress, Euralex 2000. Stuttgart. Vol. II, S. 807-818.
2. Gamper, J., & Knapp, J. (2002a). A Data Model and its Implementation for a Web-Based Language Learning System. In Proceedings of the Twelfth International World Wide Web Conference, WWW2003. Budapest, Hungary. S. 217-225.
3. Gamper, J. & Knapp, J. (2002b). A review of intelligent CALL systems. Computer Assisted Language Learning, 15(4):329-342, Oct. 2002.
4. Pedrazzini, S. (1999): The Finite-State Automata's Design Patterns, WIA'98, Third International Workshop on Implementing Automata, Rouen, France, September 1998. Published in J.-M. Champarnaud, D. Maurel, D. Ziadi (Eds.): Automata Implementation, Lecture Notes in Computer Science, p. 213, Springer.
5. Pedrazzini, S. (2000): The Jacaranda Framework, available at http://b.die.supsi.ch/~pedrazz/jacaranda.
6. ten Hacken, P. & Domenig, M. (1996), 'Reusable Dictionaries for NLP: The Word Manager Approach', Lexicology 2:232-255.