The Start of VISL
VISL, which stands for "Visual Interactive Syntax Learning", is a research and development project at the Institute of Language and Communication (ISK), University of Southern Denmark (SDU) - Odense Campus. Since September 1996, staff and students at ISK have been designing and implementing Internet-based grammar tools for education and research.
At the start of the project, four languages were involved: English, French, German, and Portuguese, soon to be followed by Danish, Spanish and Esperanto. Since then, many additional languages have joined the project - as evidenced by the growing number of entries in the language list. Teaching treebanks have been developed for all participating languages, and most of VISL's grammar teaching tools and games work with all languages. However, most linguistic research, CG/NLP and corpus annotation focus on VISL's core languages, covering the Romance and Germanic language families.
VISL is being continually updated
The VISL project is highly product- and process-oriented in the sense that new grammar and language tools are made freely available on the Internet as soon as an operational prototype is available. The result of this strategy is that existing modules are continually being improved and updated - not least in response to suggestions from users worldwide. Given the system's highly dynamic architecture, however, it would be difficult to maintain parallel user interfaces, so in spite of the large number of target languages, one language, English, has been chosen as the site's meta-language. Exceptions are grammatical "technical" terms and abbreviation symbols, where target language translations are provided, as well as text books and course material for Danish schools (e.g. Grammy and the URKAS course), which are offered in Danish. Likewise, a Norwegian teaching interface has been established by VISL's Norwegian partners (GREI project).
Constraint Grammar and treebanks
Building on a complex web of HTML-pages, CGI-scripts, Java- and Perl-programs, manually annotated text data bases, and Constraint Grammar (CG) tools for automatic analysis and corpus building, the VISL site offers a graphic interface which allows the user, for a wide variety of languages, to analyse corpus examples, textbook material and free running text in an interactive way - choosing between full automatic parsing and guided manual analysis on various levels of complexity. At the core of the grammatical analyses, whether manual or automatic, is a clear distinction between form and function on the word, group, and clause levels.
Although the automatic grammatical description is based on CG, it can be transformed to different user-specified notational systems, such as constituent or dependency tree structures, tagged running text or in-text color codes. The core of VISL's language data bases are its treebanks. Research treebanks are derived from automatically annotated CG-corpora by using add-on PSG-grammars. Teaching treebanks are smaller and manually made, with pedagogically chosen sentences (Sentence lab).
The VISL teaching interface
Four core concepts have guided the design of the VISL teaching interface: Flexibility, interactivity, naturalness and tutoring. The interface is flexible with regard to notational conventions, textual input and learning level. Grammatical analysis can be run in an interactive mode, where the user analyzes a sentence step by step, selecting form or function readings for words or constituents - thereby building a syntactic tree from scratch and measuring it against the computer's background knowledge. Unlike most grammar teaching software, VISL avoids the limitations of "toy lexica" and "toy grammars", allowing - for a number of the languages - user input of free, natural text. Though a human teacher's intuitive understanding of a student's problems is, of course, difficult to build into a program, tutoring is nevertheless an important aspect of the VISL interface, which offers "clickable" definitions of grammatical terms, error comments in interactive analyses and quizzes, visual and oral feedback in games, as well as live access to corpus examples.
An important VISL feature is the use of one and the same interface for multiple purposes: a) to access explanatory metatext on grammatical terms and symbols, b) to enjoy language games and quizzes, and c) to search the system's corpus databases. This makes it easy for users to create personalized learning sessions which bring together information from definitions, examples, analyses, and games - thus improving individual control of grammatical concepts and terminology through a variety of means using a single interface.
In addition to its use as as self-tutoring system, VISL provides far-reaching opportunities for class-room teaching and for scholarly research. At University of Southern Denmark (SDU), VISL has been employed by many students in preparation for the final exam in, for example, English grammar. VISL has also contributed in various ways to a number of student research projects at the B.A., M.A. and Ph.D. levels. The VISL system is also being used to provide grammatical form and function tags for large text corpora, such as the Danish Korpus90/2000/2010, the Portuguese Público and Folha newspaper corpora, the English BNC or the multi-lingual Europarl and Internet corpora.
Staff and students connected with VISL have conducted numerous workshops, demonstrations, seminars and courses dealing with the VISL tools - both inside and outside Denmark. VISL's natural language processing research has spawned a considerable number of papers at international conferences. Outside Denmark, such activities have included Brazil (Sao Paulo, Curitiba), Finland (Helsinki), Greece (Athens), Holland (Groningen), Lithuania (Vilnius), Mozambique (Maputo), Norway (Oslo, Bergen), Portugal (Lisbon, Evora, Faro), Scotland (Edinburgh), Estonia (Tartu), France (Paris, Nancy), Germany (Tübingen, Freiburg), Iceland (Reykjavik), Marocco (Fez) and Sweden (Stockholm, Göteborg, Växjö). As a result, the VISL system is being used increasingly in research and in the classroom, from grade school to university level.
Offline parsers and language technology licensing
Although the VISL tools are freely available worldwide via the Internet, some institutional or commercial users, as well as individual researchers, have shown interest in the underlying language technology as such, for instance taggers, parsers, dictionaries, treebanks and other corpus annotation, named entity-recognizers etc. If you are interested in an offline installation of any of these, please contact firstname.lastname@example.org for further information on licensing and prices.
Support for VISL
VISL has received generous financial support from a number of sources, including:
These grants have made it possible for VISL to employ a salaried project director, Ph.D. students, full-time research assistants, and student help for computer programming, grammatical analyses, corpus tagging, and web design.
Information to users
All rights to this website are administered by the University of Southern Denmark. The website may not, in part or in whole, without the University's prior written consent, be reproduced, presented, or distributed to a third party.
Unauthorised use of this website and its contents may constitute an infringement of the Danish Copyright Act and may have legal consequences.