The VL3: A Project at the Crossroads between Linguistics and Computer Science

The VL3: A Project at the Crossroads between Linguistics and Computer Science
Print Friendly
XML
Nuñez, Camelia Gianina, The CulturePlex Laboratory, The University of Western Ontario, Canada, cnunez3@uwo.ca
Mavillard, Antonio Jiménez, The CulturePlex Laboratory, The University of Western Ontario, Canada, ajimene6@uwo.ca
The current study will introduce the VL3 (Virtual Language Learning Lab) a multidisciplinary project that brings together a variety of disciplines: from Theoretical Linguistics, Second Language Acquisition and Second Language Pedagogy to Computer Science and Natural Language Processing (NLP). The VL3 stands as a great example of how humanistic knowledge, when combined with the power of the technology, can lead to the creation of innovative tools, beneficial to society at large.

Specifically, the VL3 aims to provide a virtual environment where (Spanish) second language learners can work towards improving their communicative skills in the target language, by participating in a set of predefined conversation scenarios that closely mimic real life situations.

As mentioned earlier, the VL3 is heavily based on research within the fields of Linguistics, Second Language Acquisition and Second Language Pedagogy; as such an important part of this presentation will focus on introducing and explaining the theoretical context for the development of the current project. We will then proceed to show how this research has been applied in order to further develop and refine the existing spoken dialogue system technology used by our partner, Natural Language Inc.

Although the importance of linguistic theory (phonology, morphology, syntax) to the advances in the field of NLP has been pointed out by many (Matthews1993; Schultze & Gupta 2010), it continues to be a relationship that needs to be explored in more detail. Without a doubt, greater attention paid to linguistic research would lead to even more significant developments for natural language processing technologies. Skeptics of such technologies claim that computer programs will never be able to account for the complexity of the human language (Salaberry 1996) and although it may well be the case that we will not be able to fully describe something as unpredictable as human language, it is only by maintaining a close relationship with linguistics, its discipline of study, that any real attempt can be made.

As such, in order to meet our objective of creating a highly developed technology capable of processing human language as a whole with the highest degree of accuracy possible, we will recur to recent theories of linguistics and second language acquisition.

Language technologies such as ones described above, are used in telephone conversations where a person needing information is speaking to a machine trained to understand human language (call centers, for example). Although these technologies have enjoyed great success, they are most often limiting, in sense that they are only trained to deal with highly specialized language. For example a NL software employed by an internet service provider will only ‘understand’ language that is specific to this context. In fact, it is often the case that these technologies are trained to only pick up on certain key words (home internet, high speed, technical problems, billing, etc.) uttered by the person on the phone and from that on, infer the purpose of the call. Should the caller ask a question ‘outside’ this scenario, the machine will most likely not understand. As such, it can be said that most of these technologies have divided language into specific contexts and have focused on teaching computers to ‘understand’ scenario-specific language. That is, they are trained to deduce the meaning of an utterance by picking up on certain key words.

The VL3 thus proposes to adopt and further adapt these technologies to the field of foreign language learning and offer students much needed opportunities to practice their target language in one-on-one conversations that imitate real life situations. This of course implies the need of less restrictive NL technology that is able to understand human language as a whole rather than only context-specific language. Consequently, such a project demands a NL technology that is based on grammatical analysis strategy rather than keyword spotting.

The technology used by Natural Language Inc., our partner in this project, is already superior to other existing ones mainly because it employs grammatical strategies for understanding human language. Nonetheless, our research experience within the fields of Linguistics, Second Language Acquisition and Second Language Teaching, will allow us to further develop the existing technology of Natural Language Inc. in order to successfully use it as a second language-learning tool. Any speaker of a foreign language is well aware of the challenge that lays in the process of learning a language other that one’s own as well as the many different factors that come at play in the learning process. The field of Second Language Acquisition has shown us that one such factor is input in the foreign language. In fact, input has been described as ‘the single, most important concept of second language acquisition’ (Gass 1997). Input specifically refers to exposure to the target language (oral, written, formal, informal etc.) and no one would argue that a second language can be learned in isolation and with no exposure to it.

Furthermore, it has been suggested that aside from input, interaction with a native speaker also plays an important role in the process of second language development. (Long 1981, 1983, 1996). As such, producing the target language as part of the learning process (Swain 1985, 1993, 1995, 1998, 2005) has been suggested as beneficial to the learner. One of the benefits of interaction as part of the process of second language learning is what is known as ‘negative evidence’ That is, when the learner says something that their interlocutor does not understand, after some ‘negotiations for meaning’ take place, the interlocutor may model the correct language form. Consequently, this feedback received by the learner on their production will turn into more positive input. Interaction strategies between native and non-native speakers such as modifications, simplifications, paraphrasing etc. have been vastly studied as well and have clearly shown their benefits to the language learner.

Evidently, linguistic research can provide us with valuable information for the development of a project such as the VL3. Throughout this presentation you will have the opportunity to see how such research has been applied to the creation of an effective online tool that will significantly impact the field of second language pedagogy.

references
Gass, S. (1997). Input, Interaction and the second language learner. Mahwah, NJ: Erlbaum.

Gupta P., and M. Schultze (2010). Human Language Technologies (HLT). Module 3.5. In G. Davies (ed.), Information and Communications Technology for Language Teachers (ICT4LT), Slough, Thames Valley University [Online]. Available from: http://www.ict4lt.org/en/en_mod3-5.htm [Accessed 10 December, 2010].

Long, M. (1981). Input, interaction and second language acquisition. In H. Winitz (ed.), Native Language and Foreign Language Acquisition. Annals of the New York Academy of Sciences 379: 259-278.

Long, M. (1983). Native speaker/non-native speaker conversation and the negotiation of comprehensible input. Applied Linguistics 4(2): 126-141.

Long, M. (1996). The Role of the Linguistic Envronment in Second Language Acquisition. In W. R. Ritchie and T. J. Bhatia (eds.), Handbook of Second Language Acquisition. San Diego: Academic Press, pp. 413-468.

Matthews C. (1993), Grammar frameworks in Intelligent CALL. CALICO Journal 11(1): 5-27.

Salaberry R. (1996). A Theoretical foundation for development of pedagogical tasks in computer mediated communication. CALICO Journal 14(1): 5-34.

Swain, M. (1985). Communicative Competence: Some roles of comprehensible input and comprehensible output in its development. In S. Gass and C. Madden (eds.), Input in second language acquisition. Rowley, MA: Newbury House, pp. 235-253.

Swain, M. (1993). The Output Hypothesis: just speaking and writing aren’t enough. The Canadian Modern Language Review 50: 158-164.

Swain, M. (1995). Three functions of output in second language learning. In G. Cook and B. Seidlhofer (eds.), Principles and Practice in the Study of Language. Oxford: Oxford UP.

Swain, M. (1998). Focus on form through conscious reflection. In C. Daughty and J. Williams (eds.), Focus on Form in Classroom Second Language Acquisition. New York: Cambridge UP.

Swain, M. (2005). The Output Hypothesis: Theory and Research. In E. Hinkel (ed.), Handbook of Research in Second Language Teaching and Learning. New Jersey: Lawrence Erlbaum, pp. 471-483.