View Content #17517

Contentid17517
Content Type3
TitleLearner-computer Interactions: NLP-generated Feedback
Body

Trude Heift is a professor of linguistics at Simon Fraser University. Her research interests include Computer-Assisted Language Learning, and Applied and Computational Linguistics

Of all of the types of interactions that occur in language learning, particularly interesting in CALL development and research are the instances where learners are able to obtain help or feedback on their language as well as to engage in conversation and negotiation of meaning with another speaker of the language. Both cognitive interactionist (e.g., Long, 1983, 1996; Gass 1997; Pica 1994) and input processing theories (VanPatten, 2007) focus on how learners process L2 input. The task for language learners is to make form-meaning mappings that are essential for ultimately incorporating new items into their interlanguage knowledge and ability for use. For CALL, this means that learner-computer interactions must support students with their language comprehension and production by, for instance, 1) making key linguistic characteristics salient by highlighting and providing opportunities for repetitions and modifications for particular forms, and 2) supporting modified interaction between the learner and the computer by providing the learner with control over when to request help, modify responses, and get access to repetition and review (see Chapelle, 1998, 2011; Heift & Chapelle, 2011). Developers of CALL materials can intentionally provide these opportunities for learners.

In the past, form-focused learning environments such as commonly found with Tutorial CALL (Levy, 1997) have offered language students a relatively low level of interactivity and a limited ability to construct meaning independently. A heavy emphasis was placed on the production of accurate language with limited learner feedback and little flexibility of student responses. However, by now and due to advances in technology and CALL pedagogy, there is a lot of potential in providing students with more valuable computer interactions in the form of informative feedback to a variety of learner input and individualization of the learning experience. Indeed, research has sought evidence that learner feedback in CALL makes a difference, and more specifically what kind of feedback makes a difference. One of the early studies investigating different feedback types for Japanese grammar instruction found that 'intelligent' feedback (with a metalinguistic explanation) was more effective than traditional feedback (e.g., wrong, try again!) (Nagata, 1993). A number of studies followed and they generally support the claim that students benefit from explicit feedback because they subsequently perform better on particular target language structures and/or because students' grammatical awareness is subsequently raised. Evidence for effects of computer-generated feedback has also been sought in studies examining learner error correction behavior, referred to as learner uptake(Lyster, 2007), in response to distinct feedback types (e.g., Heift & Rimrott, 2008). Here, the findings report significantly more learner uptake for feedback that provides detailed corrections.  But what are the technical requirements for 'intelligent' feedback?

If one does not compare the students' input string to the strings of anticipated answers, the computer needs to be capable of a much more sophisticated linguistic analysis of student input to detect errors and provide corrective feedback and contextual instructional guidance in an individualized learning environment. This research approach is taken in Intelligent CALL (ICALL). One of the key features of an ICALL system resides in the detailed and individualized level of feedback that the program offers the student, along with keeping track of each student's most common mistakes. In addition, help tools and scaffolding are driven by complex underlying linguistic tools and analyses which help to provide a richer learning environment (see Heift, 2011). ICALL systems integrate natural language processing (NLP) and artificial intelligence (AI) modeling into CALL. NLP techniques model "understanding" of human language by computer, while AI techniques can be used to model the individualized learning experience, thus aiming at learning programs that come closer to natural language interaction between humans than has been the case in traditional CALL. The development of an NLP system, however, along with its integration into a CALL package is a very complex, onerous, and extremely time-consuming endeavor, largely due to its sophisticated underlying technology. There are currently a number of ICALL applications used in foreign language classrooms, most notably a system for Japanese (Robo-Sensei), for Portuguese (Tagarela) and for German (E-Tutor). In addition, we find an increase in more general NLP tools used and applied to online language learning environments (e.g., Compleat Lexical Tutor).

References

Chapelle, C. A. (2011). The relationship between second language acquisition theory and computer-assisted language learning. The Modern Language Journal, 93, 741-753.

Chapelle, C. (1998). Multimedia CALL: Lessons to be learned from research on instructed SLA. Language Learning & Technology, 2(1), 22-34.

Heift, T. (2010). Developing an Intelligent Language Tutor. CALICO, 27(3), pp. 443-459.

Heift, T. & Chapelle, C. (2011). Learning through Technology. In Gass, S. & Mackey, A. (eds.). Handbook of Second Language Acquisition, Routledge, pp. 555-570.

Heift, T., & Rimrott, A. (2008). Learner responses to corrective feedback for spelling errors in CALL. System, 36(2), 196-213.

Levy, M. (1997) CALL: Context and conceptualization. Oxford: Oxford University Press.

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

Long, M. H. (1996). The role of the linguistic environment in second language acquisition. In Ritchie, W. C., & Bahtia, T. K. (eds.), Handbook of second language acquisition (pp. 413-68). New York: Academic Press. 

Lyster, R. (2007). Learning and Teaching Languages Through Content: A Counterbalanced Approach. John Benjamins.

Pica, T. (1994). Research on negotiation: What does it reveal about second-language learning conditions, processes, and outcomes? Language Learning, 44(3), 493-527.

VanPatten, B. (2007). Input processing in adult second language acquisition. In B. VanPatten & J. Williams (Eds.), Theories in second language acquisition (pp. 115–135). Mahwah, NJ: Erlbaum.

SourceCASLS Topic of the Week
Inputdate2014-04-04 16:24:18
Lastmodifieddate2014-04-28 03:09:04
ExpdateNot set
Publishdate2014-04-28 02:15:02
Displaydate2014-04-28 00:00:00
Active1
Emailed1
Isarchived0