Towards WordBricks - a Virtual Language Lab for Computer-Assisted Language Learning

intelligent computer-assisted language learning environment, recently initiated at our institution. WordBricks is intended to serve as a " virtual language lab " that supports open experiments with natural language constructions. Being based on dependency grammars, this instrument illustrates the use of modern natural language processing technologies in language learning. The use of computer-assisted language learning (CALL) instruments is now widespread and well recognized both by language teachers and language learners. Past decades brought more powerful and accessible computers and numerous CALL software packages; the level of technological awareness among teachers has also increased greatly. At this point, it seems natural that researchers are often more focused on the integration of existing technologies into language curricula and the development of well-balanced teaching methods that combine theory, technology, and pedagogy, rather than on purely technological advancements for CALL systems [1]. However, popular CALL systems still rarely incorporate modern achievements of natural language processing technologies. For example, language learning software, recently reviewed in PC Magazine [2], at best provide the following capabilities: lessons with multimedia content, word-based memory games, online tutoring, and pronunciation training. Some packages were characterized as being brilliantly designed, nicely organized (as a combination of traditional lessons, word drills, scenario-based lessons/dialogues, etc.), or based on innovative educational concepts, such as involving a learner into a real text translation project. Undoubtedly, these features are crucial for a language learner, but in most cases they do not make use of recent research advancements (probably, the only exception is high-quality speech recognition). The lack of intelligence in CALL systems is a well-known problem, clearly formulated at least as early as in 1992 [5]. It has been suggested that a hypothetical intelligent CALL (ICALL) system can be based on both technical (natural language processing, speech recognition, feedback generation) and theoretical (pedagogy, cognitive science) advancements. The review of ICALL instruments conducted in 2002 identified at least 40 systems that use artificial intelligence (AI) technologies to a certain extent [6]. The same paper admits that many capabilities of ICALL systems cannot be reliably addressed with state-of-the-art technologies. This is a likely reason for a low interest in AI technologies for CALL today. As noted in [7], " the development of systems using NLP technology is not on the agenda of most CALL experts, and in-terdisciplinary research projects integrating computational linguists and foreign language teachers remain very rare ". Examples of ICALL systems provided in [6] and …