R. J. Blake, D.W. Fahy, and R.F. Walters

University of California, Davis

* (Accepted for publication on-line, December, 1998)

Learning to speak a second language (L2) is both an intensive and time-consuming process, taking normally takes about five years to reach functional fluency. Difficult writing systems, like Japanese, can further exacerbate the process. Going to the region(s) where the target language is spoken and immersing oneself in the language and culture remains, of course, the preferred but most expensive method of acquiring linguistic competence in another language. For those who are unable to do this, how can we increase meaningful contact with and interactions in the target language, the main mechanisms for nurturing second language acquisition (SLA)?

SLA research has demonstrated the importance of oral interactions that require the negotiation of meaning with native speakers (Hatch 1978, Long 1980, Holliday 1993), but the same benefits appear to accrue in the interactions among non-native L2 learners as well (Pica 1994, Gass and Varonis 1994). Could technological advances allow the instructor to extend these benefits beyond the walls of the classroom and the limits of a seat-time driven curricula?

This project report describes the implementation of a network-based communication or "chat" program into intermediate Spanish and Japanese language courses at UC Davis, using a package we have developed called Remote Technical Assistance (RTA) which permits synchronous transfers of text, sound, graphs, and provides a collaborative writing tool, TEXTPAD, (For more information on RTA, see the RTA home page: http://escher.cs.ucdavis.edu/).

Preliminary results indicate a high degree of success in the use of this concept in language learning at the intermediate level. These early experiments also pointed out the need to refine the method, a process that is ongoing at the present time.

One frustrating aspect of acquiring a second language is the relatively long period in which students communicate in a degenerate code described as "interlanguage," an inbetween linguistic state where students repeatedly make mistakes and borrow structures from their native language. The important research question here is how do L2 students liberate themselves from the interlanguage quagmire and truly acquire the target language. In order to do this, research suggests that they must first focus on their own deficiencies vis-à-vis the target language and develop a metalinguistic awareness (Schmidt 1990). When working in pairs to solve a real communication task, students typically "push-down" from the discourse of the task to negotiate problems in meaning (i.e. lexical, grammatical, and phonological confusions). These negotiation events result in the correction of specific mistakes and promote the evolution of their interlanguage toward the target (Pellettieri 1998). In this project, we hypothesize that these types of negotiation events constitute the principle stimulus for language acquisition and the gradual remediation of interlanguage. If this metalinguistic awareness comes about by working in pairs or groups to negotiate meaning while carrying out specific tasks, we postulated that network-based communication would facilitate this process in a similar fashion to face-to-face negotiations found in the classroom environment.

Our first pilot project in Spring, 1998, confirmed our expectations, but the proof can only be seen in a close examination of the transcripts and not in the results of the more standardized proficiency tests and attitudinal surveys.

In Spring Quarter, 1998, one intermediate Spanish and one Japanese class (n=25 in each) were asked to work in dyads once a week over a period of six weeks using the RTA Chat program to communicate synchronically with their partners. The pairs were to accomplish tasks similar to those listed below (the specifics differed slightly by language). The results from the language measures were then compared with a control class that did passive lab with no interaction with partners.

Of these four tasks, only the first one implied an open-ended solution; task #2-4 were closed tasks in that they guided students toward a unique solution, although not necessarily an identical one for all the dyads in the class. We also predicted that the closed tasks would stimulate more negotiation events, as described above, because the tasks' meaning had to be fully understood in order to arrive at a solution.

The Remote Technical Assistance (RTA) package was conceived approximately three years ago, with a goal of enhancing learning in all modes of instruction. In 1996, we received support from the UC Office of the President (IAPIF) to experiment with RTA between the UC Davis and UC Berkeley campuses. These experiments were less than satisfactory; network communication failures and other technical bugs raised many problems, as did unpredictable support from technical staff at Berkeley. The tasks were incompletely defined by the instructors, and they were not fully understood by the students. Because of the client-server design, which monitored all traffic, it was possible to detect and correct most of the technical problems that interfered with student communication, so that by Fall, 1997, the package was reasonably robust.

In 1997, we received support from the Fund for Improvement of Post-Secondary Education (FIPSE). We have completed the first year of this pilot study and, based on promising initial results, we are proposing slight modifications of the approach for the final two years of this research project.

When RTA was first designed, it was intended to be a platform-independent, multimedia tool to enhance interaction between humans. Interaction was envisaged as including either synchronous, live chats between pairs or groups, or asynchronous messaging, both supporting complex messages. The process was monitored by a server-manager that routed all communication and also stored whatever was needed to be saved for subsequent use or analysis. Hence, sound and images could be transmitted, but also saved as needed (i.e., voice messages were not telephonic; they could be replayed as often as required).

Early in the research, the need for foreign language character sets was addressed. RTA now supports Spanish, Russian and Japanese character sets on multiple computer platforms (mainly PC and Macintosh), and could easily be adapted to accommodate other character sets as well. Language input for each was based on keyboard adaptations widely used in each language, minimizing the amount of learning specific to RTA interaction.

FIPSE requires extensive evaluation, both formative (evaluating the process) and summative (evaluating the results). We were fortunate to have a consultant who is expert in the design of evaluation measures, and with his help we developed pre- and post-questionnaires, attitude surveys and other written instruments testing language proficiency to assist us in the collection of data.

By the start of the Spring Quarter, we were ready to test the new course content using RTA. Two sections of each course (intermediate Japanese and Intermediate Spanish) were selected, one for control, the other for use of RTA. An electronic pre-test was administered, the students in the treatment group received four tasks involving pairing with another student to resolve some challenges requiring close interaction in the target language. At the end of the course, student performances on standardized tests were compared, and attitudinal post-tests administered, again electronically.

The results of this study showed, as is usually the case, no perceived significant difference in test scores. The electronic tests proved to be a mistake, since few students in the control groups bothered to answer them, skewing those results to a probable bias in favor of the more electronically adept. Attitudinal responses of the students were uniformly highly enthusiastic about the RTA-mediated experience. It is our perception that we would probably note a significant difference in cultural acclimation to the second language over a longer period of time (terms at Davis are ten weeks long with a one-week finals period), but we also recognize a real problem in measuring this type of skill and subtle changes in the students' interlanguage. However, we did find that the stored dialogues of conversations between students offered a wealth of material suitable for analysis of negotiated breakthroughs in understanding.

To date, an analysis of this type of data has been completed for Spanish. Table 1 summarizes the number of negotiation events by task as compared to the total number of conversational turns or exchanges generated by the Spanish RTA dyads. As predicted, the closed tasks produced more negotiations than the open task (#1).

Contrary to fears that conversations among L2 learners will only reinforce the substandard forms of their interlanguage (Kern 1995), the dyad work with RTA shows that students actively correct and teach themselves when performing closed tasks in a network-based communication environment. Returning to the Spanish RTA transcripts, 75% of these corrections consisted of lexical confusions, as illustrated in the following transcript of a negotiation event (the student's dialogue was done entirely in Spanish):

The L2 students involved in the Spanish RTA project also corrected each others' grammar mistakes.

The results from this project have shown that network-based communication in conjunction with a closed-task curriculum does allow L2 learners to raise their metalinguistic awareness and, consequently, promotes acquisition of the target language. In some cases, chat communication permits many students to interact more freely than in the often anxiety-charged oral classroom. The RTA log files provide instructor s a window to observe students’ interlanguage and progress and, thereby, an opportunity to provide positive feedback for them.

We also found that standardized proficiency measures and questionnaires shed little light on the students' successes because of the highly process-oriented nature of these prolonged interlanguage phase. Further testing will have to be based on longer period of observation. Future implementations of RTA in the foreign-language curriculum will include interactions between native-speakers and L2 learners as well.

RTA has been developed entirely on the Davis campus and is the property of the Regents of the University of California. We wish to make it available without charge to not-for-profit educational institutions, and we have already negotiated several licenses for such use with universities and colleges in the United States and in Australia. We welcome opportunities to work collaboratively with other interested institutions. Further information about RTA and it use may be found at the home page:http://escher.cs.ucdavis.edu/

We gratefully acknowledge the University of California and the UC Office of the President (IAPIF grant), the Fund for Improvement of Post-Secondary Education (FIPSE), and the Apple Corporation for their support of this research.