Project main goal
Key issues and technical approach
Expected achievements and impact


  • Project main goal
  • The FreeText project will develop a hypermedia computer assisted language learning system for intermediate to advanced learners of French that relies on natural language processing and communicative approaches to second language acquisition. The software will incorporate authentic documents and various sentence processing tools to involve learners in the foreign language structure discovering process. Thanks to an efficient and robust error diagnosis system that incorporates error typologies based on statistical French learner corpus analysis, FreeText will handle free productions and provide the learners with meaningful feedback. The system will also include reference aids such as a linguistically motivated reference grammar, teacher-oriented authoring capabilities, and an evaluation tool. The complete system will be available both on CD-ROMs and via Internet/intranets.

     

  • Key issues & technical approach
  • The FreeText project is highly innovative at the scientific and technological level:

    1. Instead of traditional grammar teaching, FreeText follows resolutely communicative approaches to second language acquisition (SLA). Thus, it will focus explicitly on the use of grammatical forms through meaningful tasks-based activities to drive the learnersí attention to the input ("noticing" process, a necessary condition for language acquisition). As a consequence, the types of exercises provided in FreeText will be varied. Well-known exercise types such as multiple choice questions and cloze tests will be used with parsimony, to leave room for newer types of exercises in CALL soliciting more open-ended answers from the users. Such exercises (i) will be automatically corrected by the system, (ii) facilitate language acquisition according to the SLA theory in use, and (iii) effectively check all areas (productions as well as comprehension skills) of the learnersí knowledge of the target language. All the partners will be involved in the realisation of the exercises. UMIST will provide the didactic content. UGEN will provide the diagnosis tool based on the error typology compiled by UCL. Softissimo will work on the shape and realisation of these exercises.
    2. The system integrates various NLP tools, namely (i) a sentence analyser from which will be derived both a diagnosis tool for written input and a sentence structure viewer, (ii) a speech synthesiser, and (iii) a sentence reformulation tool and a translator. Few NLP tools are currently used in CALL systems, mainly because they lack reliability, coverage, and/or efficiency. Thus, language learners cannot trust the systemís output, cannot try out words and sentence structures that are not directly covered in the systemís tutorials, and become easily bored when the system does not provide error feedback or speech synthesis in real time. These drawbacks will be lessened in FreeText, since UGEN is a specialist of reusable, large coverage NLP systems and has already developed NLP tools such as analysers, speech synthesisers and sentence generators.
    3. FreeText incorporates error typologies based on the analysis of extended corpora from learners of different linguistic background, both to make the error diagnosis system more effective and to prepare relevant feedback information and links to specific parts of the tutorials or of the reference grammar, in order to help the learner review the misapplied rules. The corpus analyses and the error typologies will be performed by UCL, a specialist in error analysis and coding.
    4. The FreeText CALL system comprises an authoring tool that allows language teachers to tailor the software to their studentsí needs by adding their own series of exercises. The possibility of easily adding exercises to a CALL system allows teachers to feel less excluded from the teaching process and therefore more likely to use the system. At the same time, a system customised to a particular group of learners has better chances of responding to their needs. This difficult task will be supervised by Softissimo, a specialist in natural language software.

     

  • Expected achievements & impact
  • The FreeText project is expected to result in a complete hypermedia NLP-based CALL system for French which will include tutorials exploring different text types, a reference grammar, NLP-based tools, and numerous exercises. This system will be authorable, allowing teachers to complement it with their own contributions. It will also make the most effective use of all the NLP-based tools available, for tutorials and exercises alike. The system will be able to run on two different platforms with the same capabilities: on the Internet for remote learning, and on PCs for individual and closed-environment uses.

    Given its characteristics, the FreeText system will directly contribute to various EU policies currently being implemented, among which the education, training, language and information policies, by promoting life-long, remote training, multilingualism, and hypermedia mastering. FreeText will also directly contribute to some major Community social objectives, mainly by promoting cultural awareness through foreign language study, trans-European mobility, and employment prospects.



    Last modified: Monday, 13-Sep-2004 10:49:27 MEST, Webmaster