This paper addresses a fundamental dilemma in the design of intelligent language learning environments: the more freedom a system offers to learners in the use of the target language, the more unwieldy the data is which the learners produce and the less able the system is to support inferences about learners from that data. It is shown how in a platform where learners and teachers interact, the teachers' feedback which is archived in the system and indexed to the learners' target language production can constitute affordances that support a process of bootstrapping from raw language output to potential insights into the learners' interlanguage and gaps in their grasp of the target language. The approach is illustrated with three types of learner errors uncovered in the corpus of learner English through this bootstrapping heuristic.
One of the most common and persistent error types in second language writing is collocation errors, such as learn knowledge instead of gain or acquire knowledge, or make damage rather than cause damage. In this work-inprogress report, we propose a probabilistic model for suggesting corrections to lexical collocation errors. The probabilistic model incorporates three features: word association strength (MI), semantic similarity (via Word-Net) and the notion of shared collocations (or intercollocability). The results suggest that the combination of all three features outperforms any single feature or any combination of two features.
The shift towards communicative language teaching in recent decades has created pressure towards individualized pedagogy that arises from the diversity found within any group of learners. One of the richest areas of diversity in target language needs across learners is the lexis of the various discourse communities that different learners are attempting to enter. This paper elucidates one way that the Web and the new practices that it has engendered has created possibilities for individualizing vocabulary learning in context and for doing this incidentally and in real time. We propose an approach to individualization that leverages the navigation level of the Web. This approach is illustrated with a browser-based agent that first detects collocations incidentally within the web pages that the user freely browses, and then unobtrusively offers to bring these to the learner's attention. The novel challenge for the technology was that the natural language processing techniques must perform reliably in real time, in unscripted noisy contexts. Two empirical studies are reported. One shows that the tool enhances collocation learning and the other that unrestricted learners using the tool encounter divergent sets of collocations, resulting in individualization.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.