A new approach to CALL content authoring [Elektronische Ressource] / von Judith Knapp

Informations

• Publié par: gottfried_wilhelm_leibniz_universitat_hannover
• Publié le: 01 janvier 2004
• Nombre de lectures: 30
• Langue: English
• Poids de l'ouvrage: 2 Mo

Extrait

A New Approach to CALL Content Authoring
Vom Fachbereich Informatik der Universitat¨ Hannover
zur Erlangung des Grades einer
Doktorin der Naturwissenschaften
Dr. rer. nat.
genehmigte Dissertation
von
Dipl. Ing. JUDITH KNAPP
geboren am 21.03.1972 in Bruneck, Italien
20042
Referent: Prof. Dr. Wolfgang Nejdl
Korreferent: Prof. Dr. Bernardo Wagner
Tag der Promotion: 15. Oktober 2004Preface
Nowadays science is becoming a working field in which knowledge of several disci
plines is required from the researchers. The world cannot be divided into domains
within which pure specialists could reach increasingly higher results. Reality shows
that forgetting one aspect of a related field may significantly influence the entire out
come. Even a team of several specialists may not be able to cover an entire field,
namely when exchanging knowledge between them does not work. Hence, more and
more people with interdisciplinary knowledge are required. Such people can con
tribute to communication and network construction. If in a group of researchers the
co operation does not only consist of distributing tasks but of real openness and true
mutual interest, it is much more likely that communication is a fruitful part of the com
mon work, and that all facets of a given problem are discovered and considered within
the development process.
This thesis is an interdisciplinary work which covers aspects of different disci
plines: The main focus has been to find a way for developing a highly sophisticated
electronic language learning system with realistic resources. For this task it was neces
sary to create a new method which facilitated the implementation and made it possible
to join all the ideas. Work on the content side was initially carried out solely by lin
guists and language teachers. However, the techniques used allowed providing many
more features than linguists and language teachers had demanded. Hence in order to
be able to make meaningful suggestions about possible improvements of the system,
it was necessary for the computer scientists to well understand the main principles of
didactics, as well as the problems the linguists tried to resolve in their research.
Consequently, beyond technological issues, this thesis also contains a lot of lin
guistic, didactic, and sociological elements: the sociological background of the province
of South Tyrol, which is home to the main target group of our work, has been exam
ined; linguistic and didactic demands on language teaching have been studied; work
on putting these ideas into practice in computer assisted language learning has been
reviewed; and modern technologies such as networking, computational linguistics, and
Artificial Intelligence have been explored.
The overall outcome is a new approach to CALL content development. Its im
plementation and evaluation was realized within the ELDIT language learning project
carried out at the European Academy Bozen/Bolzano.
34Acknowledgments
First of all, I would like to thank my advisor Prof. Wolfgang Nejdl of the University
of Hanover for his support. I am grateful for the time he had for me during his visits to
Bolzano and for the invitation to the University of Hanover, where I spent three weeks
in summer 2000. During this period, I had some helpful discussions with Nicola Henze
and Martin Wolpers.
The closest collaboration was with Dr. Johann Gamper, who, as my local advisor,
provided feedback and stimulating discussions on all different aspects of my work.
When I started my work at the European Academy of Bolzano, he gave me the oppor-
tunity to become acquainted with a field that was very new to me at the time. Through
all these years, he always had time to meet with me and to answer my questions.
Furthermore, I would like to thank my team colleagues Andrea Abel, Vanessa
Weber, Claudia Richter, Stefania Campogianni, Julia Reichert, Sara Campi and Chiara
Vettori for their openness and interest in an interdisciplinary collaboration. We had
many vivacious discussions. I particularly want to thank Andrea for her optimism and
never ending energy to search for funds, collaborations and possibilities to realize our
ideas.
I remember also a lot of stimulating discussions with Oliver Streiter, Paolo Dongilli
and Nikolaus Augsten. Their comments were helpful and stimulating for my work.
Moreover, they always had time for me when I needed any kind of help.
Additionally, I would like to thank our project partners Pius ten Hacken and Sandro
Pedrazzini for their first rate collaboration. I am grateful to Pius for many interesting
discussions we had, not only within the meeting rooms, but also during various walks
and dinners in Bolzano and Lugano. I also thank Sandro for his constant availability
and for answering my email questions almost always immediately.
Thank you also to Peter Warasin and Lorenzo Donati, who programmed some parts
of the system and to the Institute for Computational Linguistics of the University of
Stuttgart for POS tagging our text corpus.
I am also very grateful to Christa Knapp and Albert Mairhofer for proofreading
my English text. They did an invaluable service for me.
Last but not least, I would like to thank the European Academy of Bolzano for
financing the ELDIT project, for the large amount of freedom we have, for the oppor-
tunities to participate in international workshops and conferences and for all kinds of
support.
56Abstract
The ELDIT project carried out at the European Academy Bozen/Bolzano aims at creat
ing an innovative language learning system for the population of South Tyrol, allowing
them to prepare for the exams in bilingualism. The system is furthermore sufficiently
general, so that everybody interested in learning the German or Italian languages can
use it. The project is carried out by linguists, language teachers, computational lin
guists, and computer scientists in interdisciplinary research work. The program can be
freely used via the Internet (http://www.eurac.edu/eldit).
Efficient knowledge engineering is known to be a difficult task. On the one hand
the linguists and language teachers demanded the implementation of very innovative
but complex didactic ideas; on the other hand, approaching the realization in the usual
way was judged as not feasible by them, since it would have meant the manual input
and a detailed encoding of a large amount of data. Hence a new approach to content
authoring was elaborated by the computer scientists which made it possible to develop
the system with realistic resources.
The main solution we applied was (1) to support the linguists in the detailed data
encoding process by electronically rewriting a hand made semi structured version of
the data into the final extensive version needed by the system and (2) to help them in
the extensive data collection process by electronically adding as much information as
possible to the manually collected educational data.
The approach we applied provided some more advantages: (3) the realization of
some more innovative features than the linguists had originally demanded, became
possible, (4) the inclusion of externally developed products that worked only on spe
cific language features became feasible which considerably augmented the innovative
ness of our system, and (5) the tools developed for data authoring were incorporated
into the system and reused during runtime to provide online for teachers.
ELDIT is an ongoing research project that started in autumn 1999. The approach
was appreciated and judged as feasible by the linguists in our team. Hence it was ap
plied throughout the development of the system. Its flexibility for the implementation
of didactic demands has been confirmed each time we programmed a new module for
the program.
Keywords Computer assisted language learning, Computational Linguistics, Adap
tive Hypermedia
78Kurzfassung
Das Projekt ELDIT hat zum Ziel, ein innovatives Sprachlernprogramm fur¨ die
Bevolk¨ erung von Sudtirol¨ zur Vorbereitung auf die Zweisprachigkeitsprufung¨ zu ent
wickeln. Das System ist jedoch sehr allgemein und kann von jedem sinnvoll verwendet
werden, der die deutsche oder italienische Sprache lernen mochte.¨ Das Programm wird
¨in interdisziplinarer Zusammenarbeit zwischen Linguisten, Sprachlehrern, Computer-
linguisten und Softwareentwicklern durchgefuhrt¨ und steht im Internet kostenlos zur
Verfugung¨ (http://www.eurac.edu/eldit).
Eine effiziente Wissensverarbeitung war eine schwierig zu losende¨ Aufgabe in
diesem Projekt. Auf der einen Seite verlangten die Linguisten und Sprachlehrer in
unserem Team die technische Umsetzung von zwar sehr innovativen, jedoch auch sehr
komplexen didaktischen Ideen. Auf der anderen Seite wurde es von ihnen als fur¨
nicht machbar gehalten, die Umsetzung dieser Ideen mit traditionellen Methoden und
Ansatzen¨ zu realisieren, da dies bedeutet hatte,¨ dass eine sehr große Menge an Daten
manuell eingegeben und codiert werden hatte¨ mussen.¨ Aus diesem Grund entwickel
ten die Softwareentwickler einen neuen Ansatz zur Datenverarbeitung, der es moglich¨
machte, das System mit realistischen Ressourcen zu entwickeln.
Die Hauptlosung¨ war (1) die Linguisten im Kodierungsprozess zu unterstutzen,¨
indem eine manuell erarbeitete, teilstruktur