Bio
Piet Desmet is Full Professor of French and Applied linguistics and Computer-Assisted Language Learning at KU Leuven and KU Leuven KULAK. He coordinates the iMinds-research team ITEC (Interactive Technologies), focusing on domain-specific educational technology with a main interest in language learning & technology. He leads a range of research projects in this field devoted amongst others to the integration of HLT into CALL and to the effectiveness of adaptive and personalized learning environments. He is the National Representative of EUROCALL for Belgium.
NLP-enhanced CALL: challenges and opportunities
Even though the academic discipline of computer-assisted language learning (CALL) has made considerable progress in the exploitation of natural language processing (NLP) techniques for second and foreign language (L2) teaching and learning, the use of NLP in classrooms is hardly mainstream practice. The adoption of NLP in L2 teaching and learning has been limited by a complex mix of technological and pedagogical factors, and requires more research on the one hand, and teacher and learning training on the other hand. Nevertheless, the broader field of intelligent computer-assisted language learning (ICALL) has made substantial progress and some of the results of ongoing research and development are now making their way towards more widespread integration into L2 teaching and learning.
This talk offers a tentative typology of the possible roles NLP may play in CALL, focusing exclusively on written language input and output. We see at least seven possible functions for ICALL applications.
1. Providing target language input: (semi-)automatic selection of comprehensible & authentic text material based on readability & formal complexity, analysis of meaning or text categorization.
2. Providing access to resources: creating reference materials such as search engines on bilingual corpora or corpus-enriched learner dictionaries.
3. Accompanying and supporting the reading process: helping students understand L2 materials through annotation layers, both on a formal and semantic level.
4. Generating exercises and tests: (semi-)automatic generation of exercise and test items based on the analysis of L2 text materials and/or on the analysis of learner errors.
5. Detecting errors and providing feedback in semi-open practice tasks: analysis of learner output using rule-based or data-driven statistical NLP-approaches, in order to go beyond (more limited) approximate string matching techniques.
6. Supporting the writing process: supporting the second language user in writing a functional, well-formed text.
7. Adaptive item sequencing: creating adaptive learning environments based on student modelling.
For each of these functions we will offer a conceptual outline as well as illustrative applications that result from academic research and development. This includes, but is not limited to, results from our own research team. This overview will allow us to give a balanced picture of the challenges and opportunities of ICALL.
While Nyns (1989: 46) was still “pessimistic about the possibility of ICALL”, we will argue that, nowadays, there is cause for careful optimism about the affordances of NLP for L2 teaching and learning. Key advancements include the improved accuracy and recall of NLP-technologies, more possibilities for rich and meaning-focused language use, and the utilisation of machine learning and statistical data analysis in NLP-enhanced CALL.