@phdthesis{01J8TMB4TKFW0VG82NQGG87R1X,
  abstract     = {{In the rapidly evolving digital era of the 21st century, language and technology are growing closer together than ever. Language technology tools have become particularly adept at performing so-called “transactional” language operations. Translating text from one language into another, for example, is a transaction that machine translation tools such as DeepL and Google Translate will generally perform well. A direct result of this ever-improving technological support for language tasks is that the main reasons for people to learn a foreign/second language (L2) – such as travelling or boosting one’s career opportunities – become less incentivising. Yet, human conversations and social interactions in foreign languages do not exclusively revolve around performing practical language transactions, but also around establishing relationships. For technological applications to assist in building these relationships, they face tasks that appear much more difficult to achieve, such as interpreting facial expressions. It is therefore likely that relying on technological tools alone will not suffice for anyone who wants to establish sustainable international contacts and fully engage in a foreign culture: being able to understand and speak foreign languages remains – even in the current digital era – an indispensable human skill. What language technology tools do possess, however, is the ability to play the role of a valuable assistant in the L2 learning process. It is this research area, the interface between L2 learning and computer assistance, to which the present PhD dissertation will contribute. More specifically, we conduct research into how relevant vocabulary items and example sentences (that illustrate how these vocabulary items are used) can be automatically selected from large collections of texts (“corpora”) to facilitate the development of learning materials for L2 Spanish. To perform these operations in an “intelligent” way (i.e. more efficiently and more tailored to the needs of L2 learners), we resort to natural language processing (NLP) techniques as our source of “intelligence”, making this research fall in the scientific domain of “Intelligent Computer-Assisted Language Learning” (ICALL). In our experiments, we show (1) that computer-assisted methods for vocabulary retrieval are highly performant but require further research to ensure the error-free output necessary in pedagogical settings; (2) that automatic vocabulary selection methods show moderate to strong correlations with L2 learners’ intuitions and needs; and (3) that artificially generated example sentences are found more suitable by L2 learners than sentences retrieved from corpora. Regarding difficulty-based vocabulary selection, we present LexComSpaL2, a first-of-its-kind dataset that can be used to train individualised vocabulary difficulty classifiers. Finally, we show by means of three case studies that learning in an adaptive and interactive online learning environment (called an “ICALL ecosystem”) does not necessarily lead to a better user experience or an increased interest in language technology, but that it does enhance students’ insights into NLP and increase their confidence in the computer as a learning assistant.}},
  author       = {{Degraeuwe, Jasper}},
  language     = {{eng}},
  pages        = {{XVIII, 317}},
  publisher    = {{Ghent University. Faculty of Arts and Philosophy}},
  school       = {{Ghent University}},
  title        = {{IVESS : intelligent vocabulary and example selection for Spanish vocabulary learning}},
  year         = {{2024}},
}