CALL related to linguistic 1

1. Electronic Tandem Language Learning (eTandem): A Third Approach to Second Language Learning for the 21st Century

GARY CZIKO
University of Illinois at Urbana-Champaign

Abstract:
Tandem language learning occurs when two learners of different native languages work together to help each other learn the other language. First used in face-to-face contexts, Tandem is now increasingly being used by language-learning partners located in different countries who are linked via various forms of electronic communication, a context that has become known as eTandem. In addition to providing a brief account of the history of Tandem and eTandem, this paper provides a comprehensive overview of electronic communication media suitable for eTandem, both asynchronous and synchronous, from the telephone and email to amateur radio and business-quality videoconferencing. Finally, the Electronic Network for Language And Culture Exchange (ENLACE) is introduced, a Web-based medium through which users of Windows, Macintosh, and Unix-based computers can find language learning partners and engage in eTandem language learning using synchronous text chat alone or in combination with video and/or audio conferencing.

KEYWORDS

CALL, Tandem, eTandem, CMC, Videoconferencing

INTRODUCTION

Second language (L2) acquisition has traditionally been conceptualized as occurring in one of two different environments: formal and informal. Formal environments typically involve a classroom where a teacher (who may or may not be a native speaker of the target language) provides instruction and practice opportunities for a group of students using activities that are grammar based (form focused), communicative based (functional-notional), or some combination of the two. In contrast, informal (or “natural”) environments are those in which an individual acquires a second language as it is used as a medium of authentic communication outside of classroom settings, as when an individual moves to a country where the target language is spoken or lives with a native speaker of the target language.

Much research in L2 acquisition has attempted to provide a better understanding

25

of the process, potential, and limitations of L2 acquisition in these two settings. Although some studies have found that formal L2 instruction can have a positive impact on L2 acquisition (Long, 1983, 1988; Norris & Ortega, 2000) there remain lingering doubts concerning the effectiveness of formal L2 instruction (Doughty, 2003). What is clear, however, is that there are several limitations to L2 acquisition in a typical classroom setting. Five obvious limitations are (a) limited exposure to the L2, (b) limited opportunities for L2 production, (c) exposure to inaccurate, nonnative L2 as produced by fellow students, (d) limited opportunities for authentic L2 communication in a wide range of physical and sociolinguistic settings, and (e) the limited language ability and cultural knowledge of many nonnative L2 teachers who provide a less-than-ideal model of the L2 and its associated culture for learners. These limitations impose serious restrictions on the quantity, quality, and variety of L2 exposure, production, and practice opportunities that students have access to in typical L2 classroom settings.

In contrast, most informal, naturalistic environments provide extensive exposure to and interaction with native speakers of the L2 in a variety of authentic communicative settings and thus provide both a large quantity and variety of native-speaker L2 exposure and opportunities for L2 production. However, many adult learners placed in such settings do not develop very high levels of proficiency in the L2, and those who do typically retain nonnative phonological, lexical, syntactic, and/or sociolinguistic aspects in their L2 (Meisel, Clahsen, & Pienemann, 1981; Schumann, 1978a, 1978b). One reason for this may be that such L2 learners typically do not receive corrective feedback, either explicit or implicit, for any of the L2 errors they make.

There is, however, a less well known third environment for L2 acquisition, called Tandem language learning, that shares aspects of both natural settings and formal instruction and has the potential of combining the best aspects of both. Like informal L2 environments, it provides extensive exposure to the L2 as spoken by native speakers (and not other L2 learners) within authentic communicative settings with many opportunities for productive use of the L2. Like formal L2 environments, it also provides focus on form and corrective feedback to the L2 learner. And in addition to providing a promising context for L2 acquisition, it also has important potential benefits for developing intercultural understanding, autonomous learning skills, and technical abilities.

This paper has two primary purposes. The first is to introduce this third approach to L2 learning, Tandem language learning, to the many L2 students, teachers, and researchers who are unfamiliar with it, particularly those living outside of Europe. The second purpose is to show how recent technological advances have now made an electronic form of Tandem, known as eTandem, available to the countless number of language learners having access to the Internet.

TANDEM AND eTANDEM: DEFINITION, HISTORY, AND RESEARCH

Tandem language learning may take many forms (as we will see), but by definition it always involves language-based communication between two learners who are native speakers of different languages and who are learning each other's

26

language as a L2. As defined by Brammerts (2001), “language learning in Tandem occurs when two language learners with different native languages communicate with one another sharing the common objective of learning from each other” (p. 10, translated by d'Atri, 2002).

Brammerts further defines the two major principles that have guided Tandem: reciprocity and learner autonomy. Concerning reciprocity, “language learning in Tandem occurs in a learning partnership in which each partner contributes knowledge and skills that the other person wishes to acquire and in which both partners simultaneously support each other's learning. The reciprocal dependence of each partner on the other requires that they work for each other and that both profit as much as possible from their common efforts” (p. 10, translated by Cziko). With respect to autonomy, “Each of the two partners is responsible for his own learning. He decides what, how and when he wants to learn and what kind of help he would like to have from his partner” (p. 10, translated by Cziko).

Tandem language learning does not have a very long history, that is, if we disregard the unorganized, informal reciprocal language learning that must have occurred even in prehistoric times when two speakers of different languages came into contact and felt a need to learn each other's language for reasons related to cooperation, trade, religion, or marriage. As can be seen from the chronology of Tandem developments available at www.tandemcity.info/index2.html? direccion=general/en_history.htm, an organized form of Tandem first appeared in Europe in 1968 and was first put into practice as part of a French-German youth exchange program. Tandem originally involved face-to-face Tandem in which partners met in the same physical space and was often included in exchange programs involving students from two European countries with different languages.

In 1979, Jürgen Wolff implemented Spanish-German Tandem partnerships in Madrid, forming the basis of what was to become the TANDEM® Network in 1983. Originally involving language schools in 16 countries, these schools offered Tandem learning experiences in conjunction with the classroom language courses offered to their students.

Another major development occurred in 1992 when the Internet was first employed for Tandem, initially in the form of an English-German online discussion group created by Helmut Brammerts who, in 1994, founded the International Email Tandem Network in which 11 European universities originally participated. This network became the International eTandem Network in 1996 (see www.slf.ruhr-uni-bochum.de/etandem/etindex-en.html) in which the use of other electronic communication media was also explored. Now known as the International Tandem Network, it includes a total of 12 European universities located in Denmark, Germany, Portugal, Spain, France, Great Britain, Ireland, Sweden, the Netherlands, and Italy (see www.tcd.ie/CLCS/tandem/email/infeng01.html).

In 2000, a majority of the 23 member schools of the TANDEM^® Fundazioa (established in 1994 by Jürgen Wolff in San Sebastián/Donostia in the Basque Country, see www.tandem-f.org) formed Tandem International (www.tandem-schools.com), a new version of the Tandem Network whose member schools continue to offer Tandem learning experiences to their language students.

27

Finally, in 2004, the Electronic Network for Language and Culture Exchange (ENLACE) was established by this author. The purpose and operation of this system for worldwide synchronous eTandem learning is described later in this paper.

Despite its rather short history, a large number of research studies and reports have been published on Tandem language learning, both in its face-to-face and eTandem forms. An extensive, up-to-date bibliography of Tandem publications maintained by Brammerts (available at www.slf.ruhr-uni-bochum.de/learning/tanbib.html) includes well over 400 references. As an indication of Tandem's European roots and its relative obscurity in North America, it is of interest to note that when searched in June 2003, not one of the above references appeared in any of the major North American second/foreign language research journals, including Applied Linguistics, Language Learning, Foreign Language Annals, Modern Language Journal, Studies in Second Language Acquisition, and TESOL Quarterly.

While space does not allow here a thorough review of research on Tandem, descriptions of the major research studies by institutions associated with the International Tandem Network can be found in the German-language volume edited by Brammerts and Kleppin (2001). These include descriptive studies of Tandem in both face-to-face and electronic forms undertaken in England, Germany, Ireland, and Spain, as well as evaluative studies conducted in England, France, Ireland, Mexico, the Netherlands, and Portugal (an English translation of the Brammerts and Kleppin volume is due soon). The only known study of Tandem conducted in the U.S. was an investigation of methods of coaching eTandem learners of Farsi, French, and German (d'Atri, 2002).

FROM TANDEM TO eTANDEM: THE USE AND POTENTIAL OF ELECTRONIC TANDEM IN THE 21^ST CENTURY

While Tandem was first employed in face-to-face situations where pairs of learners shared the same physical space, recent technological developments now make it possible for Tandem language learning to take place between two individuals located anywhere in the world. Electronic communication media now makes eTandem available to millions of language learners living in places where face-to-face interaction with native speakers of the target language is unavailable or inconvenient. The emergence of eTandem is a particularly important development for the spread of Tandem language learning in the Americas where, compared to Europe, there are fewer major languages (Spanish, English, Portuguese, and French) spoken over much wider areas, making face-to-face Tandem less practical. This section will provide an overview of some of the many electronic media that can now be used for eTandem.

Telephone

Invented by Alexander Graham Bell in 1876, the traditional wired telephone can hardly be considered new technology. Nonetheless, the telephone can be a practical

28

and useful tool for synchronous audio eTandem for many learners, even those with access to newer communication technologies.

The wired telephone has a number of important advantages for eTandem. First, it is widely available. Just about every home and workplace in the developed world has one or more telephones, and telephones are also widely available in the developing world. Second, the telephone can provide good quality, full-duplex audio communication with no perceptible delay between two eTandem partners anywhere in the world. The term “full duplex” refers to the fact that most telephone communication employs two channels of communication so that each party can speak and hear at the same time. This makes it possible for interlocutors to provide vocal feedback to and interrupt each other, as is done in normal face-to-face communication. Third, while the telephone is limited to audio communication, it can be combined with other media such as text chat and video conferencing to be discussed below.

The primary disadvantage to the use of the telephone for long-distance communication is that there is usually a cost associated with each minute of communication. But while long-distance telephone charges remain high in many parts of the world, low-cost telephone service is now possible between North America and Europe and from these two continents to many other countries. For example, using one of the many “virtual phone cards” marketed by Zaptel (www.zaptel.com) someone in the U.S. can call any country in Western Europe for as low as 5 cents a minute. This means that a U.S.-Europe tandem pair can talk for one hour for only $3.00, which is effectively reduced by half to $1.50 if each partner takes turns calling the other. Even cheaper rates are available for long-distance telephone service via a carrier that makes use of the Internet to connect users' telephones (which is transparent to the user). For example PennyTalk (www.getpennytalk.com) charges a connection fee of 49 cents for each call. But per minute rates are as low as 2 cents a minute between the U.S. and all of Europe (including Moscow).

One potential disadvantage of the telephone for long conversations is that eTandem telephone partners may find it uncomfortable to use a standard telephone handset for long periods of time. The use of a telephone headset (available at electronic stores such as Radio Shack) eliminates this discomfort and allows the user to talk hands free. Hands-free telephone also allows eTandem partners to more easily combine the convenience and quality of telephone with synchronous Internet text chat (see below).

Amateur Radio

It was at the beginning of the 20^th Century that amateur radio operators (“hams”) discovered that radio waves at high frequencies from 3.5 to 28 MHz could be used for worldwide radio communication. It was found that these high-frequency (HF) radio waves can be reflected by both the earth's ionosphere and the earth's surface so that under certain conditions even low-power signals could be used for Morse code and voice communication between two radio amateurs many thousands of miles apart. Amateur radio operators have since used the magic of HF radio

29

communication to create an international community of hams involving every nation on earth who engage in contests, share technical information, and “rag chew” (talk at length) on topics of mutual interest.

It is not surprising that as in many other areas, English has become the international language of ham radio operators worldwide and many hams from non-English-speaking countries have developed impressive proficiency in English through the medium of amateur radio. Many English-speaking hams have also exploited the global communications potential of amateur radio to learn other languages, including many U.S. hams using Spanish with Spanish-speaking hams residing either in the U.S. or south of the border.

It is unlikely, however, that the typical language learner will turn to traditional amateur radio for eTandem language learning opportunities. In all countries an examination must be passed to obtain a license to use amateur HF radio communication, an examination that tests technical radio and operating knowledge and may include a test of Morse code operating skill. Fairly expensive and complicated radio equipment along with sophisticated radio operator knowledge is necessary in order to establish reliable communication with distant stations. And even a well equipped, high-power amateur radio station is at the mercy of sunspot cycles, geomagnetic storms, and man-made sources of interference that can disrupt radio communication for extended periods of time.

However, new technological developments have begun to change all this as hams have now linked their radios to the Internet in order to establish reliable worldwide communication that does not require advanced technical knowledge, expensive equipment, or cooperative space weather. Two of the most popular radio-Internet linking systems are the Internet Radio Linking Project (IRLP) and EchoLink

IRLP (www.irlp.net) now links over 1,300 two-way radios and repeaters in 34 countries plus Antarctica. A repeater is a radio relay station that simultaneously receives weak signals and retransmits them with more power using a well located antenna. Repeaters have been used for decades to provide reliable local and regional communication to hams using two-way mobile and low-power hand-held radios. These same repeaters can now be used for reliable worldwide communication if they are linked to the Internet via IRLP. This means that a ham in central Illinois can use a small hand-held radio to communicate for free with another ham in Mexico, Germany, or Tokyo who is also located near an IRLP-linked repeater.

EchoLink (www.echolink.org) is similar to IRLP in that it can also link distant repeaters. But EchoLink is more flexible by making it possible for users to link directly to repeaters using EchoLink software from a Windows or Macintosh computer connected to the Internet. It also allows two or more computer-equipped users to connect to each other directly without using a radio or repeater at all (this mode also provides synchronous text chat). At last count there were more than 126,000 registered EchoLink users in 149 countries. So while Echolink use is still restricted to licensed amateur radio operators, a user with access to an Internet-connected computer does not even need a radio to make use of this system for worldwide audio communication.

30

Used in its computer-to-computer mode, EchoLink software is much like the synchronous audio Internet conferencing software described below. There are important differences, however. One is that all users must be licensed and identifiable on the system. Another is that all users have a common interest in amateur radio and communication technology and form part of a worldwide amateur radio community that prides itself on international goodwill, cross-cultural understanding, and courteous communication practices.

The author (licensed in the U.S. as N9MJZ) has made extensive use of IRLP via radio and Echolink via both radio and computer for language learning. As a native speaker of English with knowledge of French, German, and Spanish, the author has been able to find native speakers of his foreign languages on IRLP or EchoLink at just about any time of the day with whom to practice his second language and who are often keen to practice their English with him. This contrasts with public-access chat and audio-conferencing software (described below) where it is usually more difficult to find a speaker of a desired language for one-on-one communication and where interactions are too often of a defamatory, sexual, or obscene nature.

So while the typical language learner is unlikely to turn to traditional HF amateur radio for Tandem language learning opportunities, these new radio-Internet linking systems are worth considering by anyone seriously interested in connecting to friendly other-language speakers worldwide at little or no cost. The license required in the U.S. to operate the frequencies used for the radio-Internet links requires only very basic knowledge of radio and operating procedures, and no Morse code test is required. (Information about licensing requirements in the U.S can be found at the web site of the American Radio Relay League at www.arrl.org/hamradio.html.) A basic VHF and/or UHF amateur hand-held radio can be obtained for about $100. And a radio is not even needed to connect to other users on EchoLink if an Internet-connected computer is available. The disadvantage of this mode of communication is that the communication is half-duplex with a noticeable delay when changing speakers. Another possible disadvantage (depending on one's point of view) is that the majority of amateur radio operators are older and retired men.

Computer-mediated Communication

While telephone has ruled for well over a century as the predominant person-to-person telecommunications tool, the development of the personal computer combined with the emergence and rapid spread of the Internet now provides a much wider range of telecommunication tools useful for eTandem. The two main advantages of computer-mediated communication (CMC) is low or no per-minute cost (that is, in addition to the cost of the Internet connection) and the wide range of multimedia possibilities, including text, audio, and video in both asynchronous and synchronous modes. The disadvantages of CMC are the necessary computer equipment, Internet connection costs, and the complications of installing the necessary hardware and software.

31

Email

The earliest and still most widely used form of CMC is the asynchronous text medium known as email. And so it is not surprising that the first eTandem projects involved the use of email (the International Email Tandem Network begun in 1994 by Brammerts mentioned above).

eTandem involving email exchanges usually takes the form of language learners spending half their time writing each language. This gives each partner practice in writing the L2 and provides written models by a native speaker. Alternately, partners may decide to change language after each pair of email exchanges.

eTandem via email has some clear advantages. These include email's wide availability and widespread use even by those who are not highly skilled computer users. Free email software and accounts are easily obtained and in many locations Internet-connected computers are provided at no cost by public libraries and educational institutions. Due to its asynchronous nature, email Tandem partners do not have to be online at the same time and can take the time needed to read, review, and respond to email messages. Email also provides a permanent record of exchanges that learners can later review and that researchers can analyze (Ushioda, 2000; Wooding, 1997).

Electronic Bulletin Boards

Internet-accessible bulletin boards are another possibility for asynchronous text-based eTandem learning. Probably the most popular collection of such bulletin boards are those provided by Yahoo Groups (groups.yahoo.com). Anyone with Internet access can set up a Yahoo Group for free. In addition to sharing text messages, a Yahoo Group also allows the sharing of files and photos. A search for Yahoo Groups using the terms “tandem” and “language” turned up two Yahoo Groups mentioning Tandem learning in their description: “languageeducation” and “tandemenglishfrancais” with 19 and 49 members, respectively. Neither group appeared very active. Because Yahoo Groups is an “advertising supported service,” users are subjected to frequent ads when using the system.

Another possibility for bulletin-board based Tandem learning is the Tandem Community (www.tandem-schools.com/index.php) developed by Tandem International (www.tandemcity.com). With over 400 members, the Tandem Community provides a forum for finding and communicating with Tandem partners using synchronous text chat and asynchronous text messages. Perhaps the most useful feature of the Tandem Community is its member list which includes information on members (including native and foreign languages, age, location, occupation, hobbies, and interests) and the ability to send a private message to any member (who must log on to the Tandem Community to see and respond to messages).

Chat

While the word “chat” originally referred to face-to-face vocal communication, in the world of CMC it describes synchronous text communication. First popularized by AOL Instant Messenger, chat involves two or more persons engaged in

32

text communication that appears on the other person's screen either while it is being composed or (more usually) after the message has been typed and sent by the writer. eTandem via Internet chat involves two partners who attempt to use each language approximately 50% of the time. In addition to AOL Instant Messenger, many other free chat services are available, including Yahoo Messenger, MSN and Windows Messenger, PalTalk, and iVisit.

Internet chat is a communication medium that combines aspects of synchronous oral communication and asynchronous written communication. Like synchronous oral communication, chat occurs in “real time” between two participants and, in this sense, resembles an oral conversation, albeit in text form. But having to enter text slows down the pace of the conversation and leaves a visible record of the language used to which participants can easily refer both during and after the chat. However, no paralinguistic communication is possible via chat and the turn-taking rules present in oral communication may be different or entirely absent (see Blake, 2000; Fernández-García & Martínez-Arbelaiz, 2002; Smith, 2003; Lee, 2004). Like email, chat can provide a permanent visual record of eTandem interactions that can be reviewed by learners and analyzed by researchers. One important advantage of chat (and MOOs discussed below) is that because it is text based, no special computer equipment is needed.

MOOs

MOOs provide a unique context for CMC that has also been used for eTandem (MOO stands for “Multi-user; Object Oriented”). A MOO provides users with a virtual environment that they can explore and manipulate using synchronous text communication. A MOO usually has a theme, such as a city, college campus, or building. In addition to text communication, users can often enter text commands for navigating through the virtual environment and for displaying information concerning emotions. Although MOOs are being increasingly used for Tandem (see Schwienhorst, 1998), sophisticated computer skills may be needed to set up a MOO or obtain access to one (for more information about MOOs and their use in education, see www.marshall.edu/commdis/moo ). Two popular language learning MOOs are MundoHispano (www.umsl.edu/~moosproj.mundo.html) and Diversity Education (www.du.org).

Audio and Video

Whereas text-based CMC programs use asynchronous or synchronous text as the medium of communication, audio-based CMC programs allow eTandem learners to engage in oral communication. Like text-based CMC, audio-based CMC also exists in both asynchronous and synchronous forms.

Asynchronous Audio CMC

One way of using asynchronous audio communication is to attach a voice recording to an email message. Although there are many ways to do this, the popular email program Eudora (which exists for both Windows and Macintosh in both

33

free and paid versions, see www.eudora.com) makes this especially easy by integrating the PureVoice player/recorder within its email functions. This means that Eudora users do not have to leave the email application in order to record and attach an audio message or to play a message sent by a Tandem partner.

In addition to the convenience of using PureVoice within Eudora, the PureVoice player/recorder also provides flexibility in recording and playing audio messages. When recording, users can listen to their recording and revise it before attaching it and sending it. The editing can involve recording over it completely or appending to what has already been recorded. In playback mode, PureVoice allows users to vary the speed of the recording, thereby allowing language learners to slow down the recording (without changing its pitch) to facilitate comprehension. And, of course, the recording can be played back unlimited times. The PureVoice software can also be used as a stand-alone program to record messages that can be sent using any email that permits attachments.

Another way to use asynchronous audio communication for Tandem is to make use of a Wimba voice board (www.wimba.com). A Wimba voice board allows users with a Java-enabled browser and microphone to leave and read/playback both text and audio messages. However, the Wimba voice board server software must be purchased and installed on a server and doing so is not inexpensive—the server software alone costs about $2,000 a year.

A Wimba voice board is more limited in its recording and playback capabilities than PureVoice described above in that users cannot append to an existing audio message and cannot vary the speed of playback. And like text-only message boards, the multiparty character of a Wimba voice board is not entirely consistent with the one-to-one philosophy of Tandem language learning, although it could be useful in matching two classes of students from different countries who are learning each other's languages and cultures.

Synchronous Audio CMC

There are several free commercial programs that allow synchronous audio communication over the Internet. The best known and most used are AOL Instant Messenger (AIM), Yahoo Messenger, MSN Messenger, Windows Messenger, PalTalk, and iVisit. Cziko and Park (2003) recently reviewed these programs to determine their usefulness for language learners “and found that although the audio (and video) quality varied both among and within the programs reviewed, the communicative contexts created by these programs can provide second language learners with an inexpensive means for useful audio interaction with native speakers of their second language” (p. 15).

Any of these programs could be used for satisfactory one-to-one eTandem communication for Windows users, while Macintosh users are limited to AIM and iVisit. Each of these programs also allows synchronous text chat and could probably be used (in audio plus text chat mode) successfully for users with even relatively slow telephone modem connections to the Internet.

34

Synchronous Video CMC

Of the six free Internet audio programs mentioned above and reviewed by Cziko and Park (2003), four also provide synchronous videoconferencing capability (the term “videoconferencing” as used here includes synchronous audio communication, even though video can be used without audio, as is done by deaf users of sign language). Yahoo Messenger, MSN Messenger, Windows Messenger, and AIM permit one-to-one video (along with text chat) for Windows users (Windows Messenger and AIM require Windows XP for video). iVisit is the most flexible program in that it permits one-to-one and multiparty videoconferencing (maximum of eight participants) for users of either Windows or Macintosh computers. Another free videoconferencing program not reviewed by Cziko and Park (2003) is the Virtual Room Videoconferencing System or VRVS (www.vrvs.org). Used for many years by physicists, VRVS provides excellent audio and video quality, even for large groups of users, and can be used with Windows, Macintosh, and Unix-based computers.

In order to have a satisfactory videoconferencing experience, users should have a high-speed Internet connection at least as fast as those typically provided by cable modems and DSL connections. And in addition to a microphone, a web cam is required in order for the user's image to be seen by the eTandem partner. Usually connected to one's computer via the USB port, web cams have been steadily dropping in price with good quality, medium resolution (at least 640 x 480 resolution is recommended) now available from manufacturers such as Logitech for as little as $30 to $40.

A major development in the evolution of consumer-oriented audiovideo CMC software that has occurred since Cziko and Park's (2003) review is the introduction of Apple's iChat AV software. iChat AV provides text, audio, and video communication with quality comparable to the much more expensive and bandwidth-hungry business-oriented H.323 products described below. It is also the only consumer-oriented audiovideo CMC program that provides echo-cancellation, permitting the use of loudspeakers and full-duplex audio with little or no audio feedback.

While delivering remarkable audio and video quality, Apple's iChat AV does have some limitations. These include the fact that it is compatible only with the Macintosh 10.2 and higher operating systems, handles only one-to-one connections, and requires a FireWire web cam (such as Apple's iSight) or video camera making it incompatible with most existing web cams having a USB connection. Some of these limitations can be circumvented by using iChatUSBCam, a $10 program available from Ecamm Network (www.ecamm.com) that allows the use of USB web cams with iChat AV. iChatUSBCam also makes it possible to use slower Macintosh G3 computers (below 500 MHz) with iChat AV. iChat AV 2.1 can also be used to communicate with Windows XP users using AIM 5.5. iChat AV requires a broadband Internet connection on each end and does not allow the creation of user rooms as do Yahoo Messenger, PalTalk, and iVisit (see Cziko &

35

Park, 2003). iChat AV is included as part of the Macintosh 10.3 operating system (“Panther”) and costs $30 for users of Macintosh OS 10.2 (“Jaguar”).

In its current form, iChat AV is clearly the synchronous CMC product of choice for existing eTandem partners who both have fast Internet connections as well as the required Macintosh hardware and software. And while iChat AV has no integrated directory for finding potential Tandem partners, a number of independent directories have been created, including MyiSight (www.myisight.com) and iChatFinder (www.ichatfinder.com). MyiSight is particularly useful for eTandem learners in that it allows one to search iChat AV users from specified countries having video and audio capabilities (for example, using MyiSight, the author found over 30 iChat users located in France having video and audio capability). eTandem learners with iChat AV capability may also find the list of chat rooms provided by dotmac.info of some use (www.dotmac.info/index.html/chat/rooms) where group text chats in English, French, Spanish, German, Japanese, Hebrew, and Polish can be found. It is possible to enter these chat rooms and inquire if anyone present who has iChat AV capability would like to establish a one-to-one eTandem session.

Another option for synchronous video CMC over the Internet is the use of the protocol known as H.323. This protocol can provide “business-quality” video (full-screen and 30 frames per second) and good quality audio that is superior to anything yet available via the programs designed to run on a personal computer (with the possible exception of Apple's iChat AV). Because of the high processing demands of H.323, these systems are usually either stand-alone “appliances” (made by companies such as Tandberg, Polycom, and D-Link) or use a camera unit with a built-in processor (such as Polycom's ViaVideo) that attaches to a Windows computer. This results in much higher cost than using free software with an inexpensive web cam.

The Internet connection speed requirements for H.323 are also quite high so that even cable modems and DSL connections can be too slow (H.323 usually requires at least 384 kbps on both uplink and downlink for “business-quality” results). Linking more than two users using the H.323 protocol can be quite complex and usually requires additional hardware units called multipoint control units (MCUs). Finally, hardware-based H.323 videoconferencing does not provide text chat capability. For these reasons, H.323 videoconferencing is currently not a practical medium for most eTandem pairs despite the high-quality video and audio it can provide.

THE ELECTRONIC NETWORK FOR LANGUAGE AND CULTURE EXCHANGE (ENLACE)

For language learners with access to the Internet, CMC has great potential for eTandem learning. Two serious obstacles, however, are finding appropriate language partners with compatible technology and scheduling synchronous sessions with one's partner. For a fee, MyLanguageExchange.com (mylanguageexchange.com) allows registered users to find partners. (The organization boasts having members from 130 countries learning 115 languages.) However, MyLanguageExchange

36

offers no tools for scheduling meetings nor does it provide a common technology for synchronous communication. (The organization recommends PalTalk which cannot be used by Macintosh users.)

In order to make synchronous eTandem accessible to language learners worldwide, the author has begun the development of the Electronic Network for Language And Culture Exchange (ENLACE). Developed with support from the National Center for Supercomputer Applications and the University of Illinois at Urbana-Champaign, the goal of ENLACE is to make it easier for language learners to find and synchronously communicate with eTandem language partners via the Internet.

ENLACE consists of several interrelated components that are accessible to users of just about any modern Web browser. Users must first register on ENLACE and provide information concerning their first language and culture and the target (second) language(s) and culture(s). In addition, users indicate their preferred synchronous communication media, which can be text chat alone or text chat in combination with audio and/or video.

Users can then search the ENLACE database for an appropriate eTandem learning partner among other registered users. Once an appropriate eTandem partner is found, up to three meeting times can be proposed and sent to the selected partner. The chosen partner can then either agree to one of these times (with time zone differences computed by ENLACE) or propose up to three alternative meeting times. When a mutually convenient meeting time is found, ENLACE reserves a meeting room and later sends a reminder notice via email to the two eTandem partners as the meeting time draws near.

Macromedia's Flash Communication Server is the communications heart of ENLACE, providing Web-based synchronous audio, video, and text chat communication for users of Windows, Macintosh, or Unix-based computers. Internet-connected users require no software other than a Web browser and the Flash Player which is included in many browsers and can be downloaded free from Macromedia at www.macromedia.com/software/flash. A registered user entering ENLACE within 15 minutes of a scheduled eTandem session will be ushered to a waiting room and then automatically connected to the reserved meeting room at the scheduled time. The two eTandem partners then have use of the room for up to one hour for any combination of text chat, audio, and video conferencing for their eTandem session.

ENLACE is currently available to the language students of cooperating instructors at institutions of higher education worldwide. Language instructors who are interested in making ENLACE available to their students should contact the author for access. Future plans call for making ENLACE available to primary and secondary school students as well as to autonomous adult learners of foreign languages and cultures.

CONCLUSION

The 21^st Century dawns as a particularly exciting time for students, teachers, and researchers interested in the learning of foreign languages and cultures.

37

As international trade, travel, and business increase, so does the need for knowledge of foreign languages and cultures. And as the Internet continues its growth, linking institutions, businesses, and homes around the globe, so grows as well the need for knowledge of foreign languages and cultures.

It can be considered a lucky coincidence that as the growth of technology has increased the need for knowledge of foreign languages and cultures, this same growth has also provided new resources and means by which this knowledge can be obtained. Electronically linking distant pairs of learners who are learning each other's language is an obvious way to improve knowledge of foreign languages and cultures. It is therefore surprising that eTandem is not more widely known and used by learners and teachers or investigated by researchers in the Americas.

The major goal of this article is to make eTandem both better known and more accessible to students and encourage researchers to investigate its potential. It is hoped that ENLACE will be the first of several global Internet-based projects that will be successful in attracting large numbers of language learners to an exciting new development in eTandem—synchronous text, audio, and video conferencing that is now available to anyone with a personal computer connected to the Internet.

REFERENCES

Blake, R. (2000). Computer-mediated communication: A window on LS Spanish interlanguage. Language Learning & Technology, 4 (1), 120-136. Retrieved May 17, 2004, from http://llt.msu.edu/vol4num1/blake/default.html

Brammerts, H. (2001). Autonomes Sprachenlernen im Tandem: Entwicklung eines Konzepts. In H. Brammerts & K. Kleppin (Eds.), Selbstgesteuertes Sprachenlernen im Tandem: Ein Handbuch (pp. 9-16). Tübingen: Stauffenburg Verlag.

Brammerts, H., & Kleppin, K. (Eds.). (2001). Selbstgesteurtes Sprachenlernen im Tandem: Ein Handbuch. Tübingen, Germany: Stauffenberg Verlag.

Cziko, G. A., & Park, S. (2003). Internet audio communication for second language learning: A comparative review of six programs. Language Learning & Technology, 7 (1), 15-27. Retrieved May 17, 2004, from http://llt.msu.edu/vol7num1/review1/default.html

d'Atri, D. (2002). Final Report: SBIR Phase I: eTandem distance coaching (DMI-012870).

Doughty, C. J. (2003). Instructed SLA: Constraints, compensation, and enhancement. In C. J. Doughty & M. H. Long (Eds.), The handbook of second language acquisition (pp. 256-310). Malden, MA: Blackwell.

Fernández-García, M., & Martínez-Arbelaiz, A. (2002). Negotiation of meaning in nonnative speaker-nonnative speaker synchronous discussions. CALICO Journal, 19 (2), 279-294.

Lee, L. (2004). Learners' perspectives on networked collaborative interaction with native speakers of Spanish in the US. Language Learning & Technology, 8 (1), 83-100. Retrieved May 17, 2004, from http://llt.msu.edu/vol8num1/lee/default.html

38

Long, M. H. (1983). Does second language instruction make a difference? A review of research. TESOL Quarterly, 17 (3), 359-382.

Long, M. H. (1988). Instructed interlanguage development. In L. Beebe (Ed.), Issues in second language acquisition: Multiple perspectives (pp. 115-141). Rowley, MA: Newbury House.

Meisel, J., Clahsen, H., & Pienemann, M. (1981). On determining developmental stages in natural second language acquisition. Studies in Second Language Acquisition, 3, 109-135.

Norris, J., & Ortega, L. (2000). Effectiveness of L2 instruction: A research synthesis and quantitative meta-analysis. Language Learning, 50, 417-528.

Schumann, J. (1978a). The acculturation model for second language acquisition. In R. Gingras (Ed.), Second language acquisition and foreign language teaching (pp. 27-50). Washington, DC: Center for Applied Linguistics.

Schumann, J. (1978b). The pidginization process: A model for second language acquisition. Rowley, MA: Newbury House.

Schwienhorst, K. (1998). The “third place”—virtual reality applications for second language learning. ReCALL, 10 (1), 118-126.

Smith, B. (2003). Computer-mediated negotiated interaction: An expanded model. Modern Language Journal, 87 (1), 38-57.

Ushioda, E. (2000). Tandem language learning via email: From motivation to autonomy. ReCALL, 12 (2), 121-128.

Wooding, J. (1997). Email tandem learning and the communicative curriculum. ReCALL, 9 (1), 22-23.

AUTHOR'S BIODATA

Gary A. Cziko (garycziko.net) is Professor of Educational Psychology and of English as an International Language at the University of Illinois at Urbana-Champaign. His interests include the development and use of autonomous technology-assisted language learning (ATALL). He is the founder of the Electronic Network for Language And Culture Exchange (ENLACE).

AUTHOR'S ADDRESS

Gary A. Cziko

Department of Educational Psychology

University of Illinois at Urbana-Champaign

1310 S. Sixth St, Room 210F

Champaign, IL 61820

Telephone: 217-333-8527

Fax: 217-244-7620

Email: g-cziko@uiuc.edu

2. An Investigation of the Fundamental Characteristics in Quality Online Spanish Instruction

MARGARET ROSE DON
College of the Bahamas

Abstract:
This article presents research findings investigating the fundamental characteristics in online Spanish instruction (at the university level in the United States) designed to maximize learning outcomes. The researcher collected data to develop a rubric of the fundamental characteristic in online Spanish instruction and then to determine whether the characteristics were currently being included in online Spanish instruction and to what level of success, according to students taking online Spanish courses. The analysis in the study followed Moore's principles of distance education as a theoretical base and tested the research findings against these theories.

KEYWORDS

Online Spanish Instruction, Online Foreign Language Instruction, Quality Online Instruction

INTRODUCTION

Online instruction is an area of rapid growth in education, and with this rapid growth comes mixed opinions. Some educators embrace the potential of online education, while others fear these changes will compromise the quality of education. For the most part, however, their ideas are based on personal views and/or experience and are not supported by quality research. While comparative research that reports no significant difference between online and on-ground instruction does exist across disciplines, the quality of this research has been called into question (Institute for Higher Education, 1999; Saba, 2000).

The purpose of the study is to contribute to, and improve upon, the small body of existing research that investigates the quality of online Spanish language instruction by addressing the following research questions:

Question 1: What are the fundamental characteristics of a quality online foreign language course as determined by an expert pool comprised of online foreign language course developers, online Spanish language instructors, and master Spanish classroom instructors?

285

Question 2: What additional characteristics, not previously identified by the expert pool, might also be important to the quality of online foreign language instruction as determined by students taking online Spanish courses at the university level?

Question 3: Which of the characteristics identified in question number one above are currently in use in Spanish online instruction and to what level of success in students' perceptions of learning?

Question 4: What relationship exists, if any, between Moore's theories of distance education (theory of transactional distance and theory of interaction) and the research findings of this study?

Current Research in Technology-Enhanced Spanish Language Instruction

In general, most research in technology-enhanced language instruction and computer-assisted language instruction has found that they both have positive consequences in foreign language instruction. The computer can serve many uses in language teaching.

It can be a tutor which offers language drills or skill practice; a stimulus for discussion and interaction; or a tool for writing and research. With the advent of the Internet it can also be a medium of global communication and a source of limitless authentic material. (Warschauer, 2000, p. 46)

This idea is expressed repeatedly in a compilation of articles entitled Technology-enhanced language learning (Bush, 1997). The editor of this volume also holds the same view that technology can help language learning but also notes that language teaching falls behind other disciplines in the use of technology.

The available research in technology-enhanced language instruction is also sparse and lags behind similar research available across disciplines. Four currently available studies will be outlined below along with their findings and recommendations. While these studies do not exactly match the topic addressed in this study, and while none of them are conducted entirely online, their findings cannot be ignored in light of the absence of more pertinent, applicable research.

The first study involves elementary school students and Spanish distance education (Glisan, Dudt, & Howe, 1998). This study used student assessment results and student feedback to evaluate findings and concluded that “it is `feasible' to use video conferencing technology to teach Spanish to students at two different sites.” They found that in the absence of a qualified teacher, the distance education program provided opportunities for students to learn another language that they would not have otherwise received. Evidence of student learning was found in the areas of listening comprehension, new language production, and awareness of cultural concepts. The findings also suggest that time on task was key to achievement and that more time on task was needed in both speaking and cultural exploration.

286

The second study, done at Florida State University, entitled Student perceptions of language learning in a technological environment: Implications for the new millennium gathered data from university-level beginning Spanish classes that used a combination of technologies such as CD-ROM, electronic pen pals, and threaded discussions (Stepp-Greany, 2002). This study aimed to determine the role and importance of the instructor in technology-enhanced language learning, the importance of the lab and the individual components in student learning, and the effects of technology on student experience with foreign language learning. The study found that

Students attributed an important role to instructors and perceived that cultural knowledge, listening and reading skills, and independent learning skills were enhanced but were divided in their perceptions about the learning or interest values of the individual components (p. 65).

The third study was undertaken by the Department of Instructional Technology at the University of Georgia to determine whether online reading, writing, and research could enrich the university language classroom and the quality of interaction between students and instructors (Benson & Wright, 1999). To this end, an online learning module was designed to improve student skills in reading and writing Spanish. It is important to note that this research fused the traditional classroom with online activities and that it was not an exclusively online course. The course was already in place at the University, and the online assignments were integrated into it. The researchers found that the online component of learning fostered higher order thinking skills along with the capacity to process information. Their findings also suggested that the online learning module was effective but that some students still experienced technical difficulties when trying to complete assignments. Finally, the researchers ended with a cautionary note regarding access to technology, saying that it is important for institutions implementing online components or courses to ensure that all students have up-to-date hardware and software. Technology should enhance communication, not differentiate socioeconomically among students.

The final study also employed the use of the Internet and included an online component as part of an existing first-year Spanish course. This study was carried out by the University of Albany and involved the “piloting of five Spanish language activities using the Internet” (Osuna

287

& Meskill, 1998, p. 2). The Internet activities corresponded with the textbook's activities, and the researchers wanted to assess the learners' perceptions of their learning experience. The study involved 13 subjects, and each of the subjects completed a postactivity questionnaire after finishing each of the five Internet activities. The questionnaire (6-point Likert-scale format) elicited information regarding learning outcomes, effectiveness of task design, and the technology used to perform the task. At the end of the course, students also filled out a final questionnaire regarding the activities as a whole.

The overall conclusion of the study was that “this small pilot demonstrates that the Internet is an excellent tool for teaching foreign language and culture” (Osuna & Meskill, 1998, p. 2). The researchers “tentatively” attributed their success to a balanced interaction of course textbook, language and culture, purposeful use of the target language, variety and relevance of tasks, and authenticity of the Spanish speaking culture provided by the Internet.

Overview of Research Comparing Online Spanish Courses to On-ground Spanish Courses

Studies comparing online education to on-ground education are currently being done across disciplines in order to determine how online education compares to on-ground education in different areas. The results of these studies indicate what has been labeled “the no significant difference phenomenon.” Thomas Russell (1999) has compiled a total of 335 research reports conducted over the past 70 years comparing distance education to traditional on-ground education with results of no significant difference. This compilation of studies, according to Russell, offers evidence that “technology does not denigrate instruction” (p. xiii). The compilation of the 335 studies also indicates that distance education compares at least equally to traditional forms of education, which, of course, is good news for the field of education. However, recently, the research design of these types of comparative studies has been called into question, thereby also calling their results into question.

Both Saba (2000) and the Institute for Higher Education (1999) have recently found fault with the overall quality of comparative research in distance education and have strongly suggested that in order to come to any firm and reliable conclusions, comparative research designs of the past need to be improved. Saba's main argument was that the research studies were not sound due to the transformation from controlled laboratory settings to uncontrolled settings in the field, especially with respect to the random selection of subjects in laboratory settings versus the impossibility of random selection in naturalistic settings. He also criticized the absence of any discussion of theoretical foundations in most of these studies. The authors of the Institute for Higher Education's report were in complete agreement with Saba. They concluded that there was a need for research based on theory and also pointed out the disadvantage of not using random selection.

The following section describes the research that has been done specifically comparing online Spanish instruction to on-ground Spanish instruction. This research, compatible with research done across disciplines, finds “no significant difference” between the two forms of instruction, but it, too, falls in the same category of research design criticized by Saba and The Institute for Higher Education.

Example 1: University of Arizona

A study done by Kartchner and Judge (2000) at the University of Arizona sought to determine whether a second year Internet-based Spanish course could produce results comparable to those of a regular classroom in the four language skills of reading, writing, speaking, and listening. The methodology included comparing

288

pretest and posttest results in reading, writing, speaking, and listening from a regular classroom (control group) to an Internet class (experimental group). Two foreign language specialists not connected to the study evaluated the data using the ACTFL guidelines (Kartchner & Judge, 2000).

The sample frame consisted of 24, fourth-semester Spanish students: 12 students in a regular classroom and 12 in the Internet class. Three of the Internet students dropped out, and, accordingly, 3 classroom students were randomly chosen to be excluded from the study. Students were permitted to self-select to be in the Internet class. The Internet program was designed to be user friendly and manageable by even students with very limited computer skills.

The following pretest and posttest results were obtained:

1. Writing

The students in the control and experimental groups had very similar scores on the pre- and posttest.

2. Speaking

Students in the control group started with a higher mean score than those in the Internet group (4.67 vs. 3.67). Students in both groups performed at comparable levels on the posttest.

3. Listening and reading

Posttests for both groups showed a decrease in mean score from the pretest, but difference was small.

In summary, the data analysis suggested that teaching via the Internet using programs developed at the university were equally effective in teaching Spanish to students compared to the regular classroom approach.

Example 2: Christopher Newport University

Cahill and Catanzaro (1997) found similar results in a study on first-year online Spanish students at Christopher Newport University. After researching other online studies, they felt confident that previous research had proven that online and traditional courses are comparable in learning outcomes. However, they found no research proving the same for language courses and had no models to work from. Still, they needed an online language course for their online degree programs. This situation left them with no alternative but to develop and evaluate their own course. Their findings represent three years of updating and refining the two online language programs they developed.

Their overall suggestions included the following:

1. The syllabus should emphasize cooperative learning, student interaction, pair-group work, peer editing, and peer learning.

2. Students should have an area for public messages and questions in order to reduce their sense of isolation and frustration.

3. The instructor needs to offer prompt and consistent feedback.

289

4. The instructor should make assignments clear and precise and post strict date and time deadlines for all assignments. They should also provide short, repetitive instructions to keep students on track.

Their overall conclusions were

1. Online instruction is effective in second language acquisition due to the interactive nature of online learning.

2. A determining factor for success is that both instructors and students log on regularly.

3. Reading and writing are more easily developed in an online format.

4. Online students produced better quality essays (online mean 3.21 versus 1.96 in the traditional class) with fewer incorrect words (online 18.45% incorrect, traditional class 28.56% incorrect).

5. The teacher-centered nature of the traditional class is replaced by a more student-centered process in online courses.

6. Online courses are more time consuming than traditional classes for the students and the instructor as a result of the technical aspects of electronic communication.

The researchers' ultimate conclusion was that “on-line foreign language courses are an effective alternative to meet the needs of students. Our experience… demonstrated that students' development of writing skills, in particular, can exceed those obtained in a traditional class” (Cahill & Catanzaro, 1997, p. 111).

Overall, the two research studies above indicate that online instruction can equal on-ground instruction. While more studies across disciplines are available, only two studies could be found comparing Spanish online instruction to Spanish on-ground instruction. Because the research design of such studies has been questioned (e.g., Saba, 2000), additional research in this area is necessary.

THEORETICAL BASE

Quality and Moore's Theories of Distance Education

Moore's theory of transactional distance and his theory of interaction are perhaps the most widely accepted and cited theories in the field of distance education. For this reason, they were chosen as the theoretical base for this research study.

Moore's Theory of Transactional Distance

Moore's (1989) theory of transactional distance offers a different view of distance and suggests that distance is not always a function of space, but rather of dialogue and structure. According to this theory, when dialogue (the extent to which an educational program allows educator and learner to respond to each other) is high and structure (level of structure which results in instructional methods that are decided for the learner) is low, distance is small.

Dialogue is determined by various factors such as subject matter, instructor

290

philosophy, educator and learner personality, environmental factors, and, most importantly, the medium of communication being used. For example, if the educational medium is television, dialogue is not possible, whereas, if the medium is correspondence, dialogue possibilities are heightened.

Structure is also determined by a variety of factors. In a highly structured course, objectives and methods are predetermined for learners and do not change to meet their needs. Teaching objectives, strategies, and evaluation methods are not adaptable to student needs; instead, students must adapt to the structure of the course. In contrast, a low structure course offers greater possibilities for flexibility and variation according to the needs of individual students.

The most desirable combination of dialogue and structure is one in which dialogue is high and structure is low, which provides great possibilities for student and instructor to respond to each other (high dialogue) and also a great opportunity to adapt the course to the needs of the student (low structure). Moore's theory of transactional distance has been empirically tested by Saba and Shearer (1994) with results that favorably support the theory.

Moore's Theory of Interaction

Moore's (1989) theory of interaction includes and defines three different types of learner interaction: learner-instructor, learner-content, and learner-learner.

Learner-instructor Interaction

Learner-instructor interaction is the “interaction between the learner and the expert who prepared the subject material, or some other expert acting as instructor” (Moore, 1989, p. 4). The aim of the instructor in this type of interaction is to stimulate, maintain, and/or enhance learners' interest in the subject matter being taught, including self-motivation and self-direction, and to motivate student learning. Instructors do this by presenting information, skills, or modeling; organizing students' application of what is learned; organizing evaluation of student progress; and providing counseling, support, and encouragement.

Learner-content Interaction

Learner-content interaction is the “interaction between the learner and the content or subject of the study” (Moore, 1989, p. 2). According to Moore learning cannot take place without learner-content interaction because learning is the process that causes changes in the learner's understanding, perspective, and cognitive structures. He also points out that some learning programs completely rely on learner-content interaction and area self-directed study.

Learner-instructor Interaction

Learner-learner interaction is the “inter-learner interaction between one learner and other learners, alone or in group setting, with or without the real-time presence of an instructor” (Moore, 1989, p. 4). This type of interaction is a valuable

291

resource for learning and, in some cases, even necessary. Moore points out that the desirability of learner-learner interaction depends upon factors such as learner age, experience, and level of learner autonomy.

Moore believes all three of the above interactions are important to the success of traditional education and distance education and states

Educators need to organize programs to ensure maximum effectiveness of each type of interaction, and ensure they provide the type of interaction that is most suitable for the various teaching tasks of different subject areas, and for learners at different stages of development. The main weakness of many distance education programs is their commitment to only one type of medium. (Moore, 1989, p. 5).

Thus, not only should dialogue be high and structure low (rendering distance small), but interaction should also be high and include the “maximum effectiveness” of the three interactions in order to achieve heightened success in distance education.

The study presented here focuses on what characteristics are essential in a quality online Spanish course by surveying an expert pool of Spanish language instructors, online Spanish instructors, and expert online course developers and then investigates whether these characteristics are currently being put into practice, and with what level of success. Finally, it examines these findings in light of Moore's theories of distance education.

RESEARCH DESIGN

All of the data for this study were gathered from two surveys: an expert pool survey and a student survey. The expert pool survey consisted of three open-ended questions and was administered to experts in the fields of traditional Spanish instruction and online Spanish instruction. The student survey was developed from the information provided by the expert pool survey and was administered to students taking online Spanish courses.

Subjects: Expert Pool

The expert pool was derived from three main sources: online foreign language course developers, online Spanish instructors, and master Spanish classroom instructors. The online foreign language course developers were chosen for inclusion only if they had experience developing an online foreign language course on their own, or in cooperation with others, and if they had either completed, or nearly completed, their doctoral degree in a related field, had published an article in a related field, or had made some other significant contribution to the field.

Due to the small quantity of Spanish online courses currently available, the only criteria for inclusion in the Spanish online instructor category were that the instructor currently be teaching a first- or second-year Spanish course online or had taught a first- or second-year Spanish course online within the past year and

292

that he/she have a minimum of one quarter or one semester of experience teaching Spanish online. The master Spanish language instructors were chosen from private universities, public universities, and community colleges. To be included in the pool, they needed to have taught first- or second-year Spanish at some time during the past 2 years and to have a minimum of five successful years of teaching first- or second-year Spanish at the college level with an average of 85% or higher in their teaching evaluations. The 85% criterion was established to include only instructors who had been rated as top instructors.

Thirteen potential expert online course developers who were also online course instructors and five master Spanish instructors were chosen for inclusion in the expert pool.¹ All subjects in the expert pool were informed that their responses would be used anonymously in a doctoral study to develop a rubric of fundamental characteristics for quality online foreign language courses. While the expert pool subjects were encouraged to participate, they were under no obligation to do so. They were also encouraged to forward the survey to anyone they considered an expert in Spanish instruction or online Spanish instruction.

Twenty-one people responded to the expert pool survey. Two were eliminated because they did not meet the criteria to be considered an expert, and two were eliminated for not supplying enough information to be analyzed. The 17 remaining respondents were categorized as follows:

8 master Spanish classroom instructors

3 online Spanish instructors

1 online Spanish course developer

1 online Spanish course developer and Spanish online instructor

4 online Spanish course developers, online Spanish instructors, and master classroom Spanish instructors

Subjects: Students

The 28 participating student subjects in the study were college and university students taking first- or second-year Spanish online courses during spring/summer quarter/semester of 2003 at five different colleges and universities in the United States. Seventeen of the students were taking first-year courses, and 11 of the students were taking second-year courses. All of the participants were 17 years of age or older; 9 participants were between the ages of 17 and 23, 7 participants were 31-40 years old, and 8 were over 40. Twenty-one of the participants were female, and 7 were male. The students completed an online version of the survey via a link in the course they were taking or an email version sent by their instructor.

Instrumentation

Two surveys were used to collect data: a set of open-ended questions for the pool of experts and a set of more structured questions derived from the expert pool results for the pool of students. The student survey also contained questions that allowed the students to evaluate the instructional effectiveness of the characteristics

293

of quality online courses identified by the experts and to list other characteristics of quality online courses not identified by the experts.

Expert Pool Survey

The expert pool survey was designed to gather information regarding fundamental characteristics in foreign language online courses and interval data ranking the importance of these fundamental characteristics. To this end, the survey asked three main questions.

1. In your opinion, what characteristics are fundamental in a quality online language course (first- or second-year university level) designed to achieve maximum learning outcome? Include as many (or as few) as you consider fundamental.

2. Which of these characteristics do you consider the most important? Please rank the characteristics you identified in Number 1 above in order of importance. Put a Number 1 next to the most important, a Number 2 next to the second most important, etcetera.

3. Why are these characteristics fundamental to maximum learning outcome?

The number of times each characteristic was listed and the ranking each one was given were used to determine it as a fundamental characteristic. If a characteristic was mentioned more than once, it was grouped with other similar responses and then given one to five points according to its ranking by the respondent. If a characteristic was considered the most important characteristic (received a rank of 1), it received five points. If it was considered the second most important characteristic (received a rank of 2) it received four points. If it was considered the third most important (received a rank of 3), it received three points, and so on. If a respondent listed fewer than five characteristics, each characteristic was ranked and weighted beginning with the most important receiving five points and continued in the manner explained above only for the characteristics listed. This ranking system accommodated 16 of the 17 respondents' responses. Only one respondent included a sixth most important characteristic, which received one point also.

Student Survey

All of the student survey questions, except one, were directly derived from the responses gathered from the expert pool survey (see the student survey in the Appendix). The top five fundamental characteristics identified by the experts were converted into a rubric of fundamental characteristics. Each item in the rubric was then used to develop a corresponding question in the student survey. The students were asked to identify whether each of the characteristics was present in their online course. If so, the students were subsequently asked to rate on a 5-point Likert scale to what extent this characteristic contributed to their success in learning.

The student survey included an additional question not derived from the expert pool results. This question asked students to identify any characteristic not present in their online course that would have contributed to their learning.

294

FINDINGS

Analysis of the responses from the expert pool survey revealed that the five characteristics the experts found to be fundamental in an online foreign language course designed for maximum learning outcome were (a) clear instructions, (b) student-instructor contact, (c) an audio component, (d) the presence of all four language skills (reading, writing, speaking, and listening), and (e) student-student interaction. The above five characteristics were then divided into levels of importance. Level 1, the most important characteristics, contained (a) clear instructions, (b) student-instructor contact, (c) all four language skills, and (d) an audio component, while level 2 (secondary importance) contained student-student interaction. These findings indicate that online foreign language course developers and course instructors should do everything possible to ensure that their courses include these five characteristics as part of the course curriculum in which the first four characteristics are of primary importance and the fifth of secondary importance.

In addition, a third level of importance was also identified by the expert pool and contains characteristics that should be taken into consideration by online course developers and instructors. These characteristics are, in order of importance, (a) culture, (b) clear understanding of instructor expectations, (c) good organization, (d) sample tests and exercises, (e) competent instructor, (f) prompt replies to students, and (g) instructor telephone contact with students. Table 1 summarizes the difference between the three levels and lists additional information regarding the number of responses and weighted results relevant to the ranking and levels.

Table 1

Rubric of Fundamental and Additional Characteristics Identified by Expert Pool

	Level of importance	Characteristics	Number of responses	Weighted results
Fundamental characteristics	Level 1	Clear instructions	5	18
		Student-instructor contact	4	19
		All four language skills	4	18
		Audio component	4	16
	Level 2	Student-student interaction	3	10
Additional characteristics		Culture	2	9
		Clear understanding of instructor expectations	2	7
		Good organization	2	7
		Sample tests and exercises	2	7
		Competent instructor	2	6
		Prompt replies to students	2	6
		Instructor phone contact with students	2	3

Each of the Level 1 and Level 2 characteristics are discussed below.

295

Clear Instructions

Clear instructions were identified as the most important fundamental characteristic because it was the only characteristic that was listed five times by the participants in the expert pool. All of the other characteristics were included four or fewer times. Clear instructions are therefore the characteristic that should be given the most attention.

Student-instructor Contact

Student-instructor contact was identified as the second most important characteristic and should be given nearly as much priority as clear instructions because, although it was listed by one less expert pool survey respondent, it received one more weighted point than clear instructions. Therefore, although it was identified as the second most important characteristic, it ranks only barely below clear instructions.

All Four Language Skills (reading, writing, speaking, and listening)

All four language skills (reading, writing, speaking, and listening) were also identified by the expert pool as being of great importance when designing an online foreign language course. Although this characteristic ranks third in importance, it received the same weighted importance as clear instructions. However, this characteristic ranks below clear instructions because one less survey respondent listed it as important. It also ranks below student-instructor contact because it received one less weighted point than student-instructor contact. Again, although it is identified as the third most important characteristic, it ranks only barely below clear instructions and student-instructor contact.

Audio Component

The presence of an audio component was also identified as a fundamental characteristic by the expert pool. The same number of respondents identified it as important as student-instructor contact and all four language skills, but it ranks below these two characteristics because it received fewer weighted points. It was, however, still determined as one of the most fundamental characteristics in an online foreign language course.

Student-student Interaction

Student-student interaction was also included in the rubric of fundamental characteristics, but it was categorized in importance at Level 2 because it was identified by only three respondents, while the Level 1 characteristics were listed four or five times. It also received a weighted score of 10, while the Level 1 characteristics all received a weighted score between 16 and 19.

The level 3 characteristics can also be considered for possible inclusion in an online foreign language course. While these characteristics were classified as Level 3 in importance, their inclusion in the expert pool reports gives them enough

296

weight to carry importance. The experts had of course a large number of characteristics they could have included in their responses, and the fact that even 2 of 17 agree on the importance of these characteristics lends them sufficient importance to be considered for appropriate inclusion in online foreign language courses. Future studies should investigate the role of these Level 3 characteristics.

Finally, question number three of the expert pool survey was designed to gather information as to why the experts considered the characteristics they identified as fundamental in a quality online foreign language course. Analysis of the responses to question three provided only one small piece of additional information beyond what was reported in questions one and two, and this piece of information was directly related to Moore's theories of distance education and will be discussed below. The responses to question three were otherwise simply elaborations of what is listed in Table 1.

Presence of Fundamental Characteristics in Online Spanish Courses

The student survey was designed to determine whether the five fundamental characteristics identified by the expert pool were employed in first- and second-year online Spanish courses currently being offered by colleges and universities in the United States. The results of the students' responses to each characteristic are discussed below.

Clear Instructions

Twenty-seven of 28 student survey respondents responded yes to the question that asked whether their course had clear instructions.

Student-instructor Contact

Twenty-eight of 28 student survey respondents responded yes to the question that asked whether there was student-instructor contact in their course.

All Four Language Skills (Reading, Writing, Speaking, and Listening)

Twenty-seven of 28 student survey respondents responded yes to the question that asked whether all four language skills were included in their course.

Audio Component

Twenty-eight of 28 student survey respondents responded yes to the question that asked whether there was an audio component in their course.

Student-student Interaction

Twenty-four of 28 student survey respondents responded yes to the question that asked if there was student-student interaction in their course.

Overall, the students' responses provided strong evidence that the five fundamental characteristics identified by the expert pool as necessary to achieve maximum

297

learning outcomes were currently in use in online Spanish courses offered by universities and colleges in the United States.

Success of the Fundamental Characteristics at Contributing to Student Learning

The student survey was also designed to determine how successful the five fundamental characteristics were at contributing to students' perceptions of learning in the online courses. Each characteristic was ranked on a scale of one to five, in which one was low and five was high, to indicate to what degree the characteristic contributed to the students' learning. A mean score of four or above was determined to be a minimum for making a contribution to student learning. The results of the analysis of students' responses for each characteristic are summarized, in rank order of importance in Table 2.

Table 2

Mean Scores and Differences Between Mean Scores and High Contribution to Learning

Variables	Mean	SD	Difference from 4.0 minimum	t	p
Clear instructions (n = 27)	4.3704	0.967	0.3704	1.99	.057
All four language skills (n = 27)	4.1852	0.786	0.1852	1.22	.232
Audio component (n = 28)	4.1786	1.056	0.1786	0.89	.379
Instructor contact (n = 28)	4.0357	0.962	0.0357	0.20	.864
Student interaction (n = 24)	2.8333	1.308	1.308	4.37	.000

Clear Instructions

Clear instructions received the highest mean score of 4.3704. This mean score provided evidence that, in students' opinions, clear instructions contributed to learning in an online course. Although the difference between this mean score and the minimum score of 4.00 is not significant, it shows a clear trend that clear instructions were very important for the students. This finding reinforces the ranking of the expert pool finding that clear instructions are likely the most important characteristic in a quality online Spanish course.

All Four Language Skills (Reading, Writing, Speaking, and Listening)

The inclusion of all four language skills received the second highest mean score (4.1852), indicating that students believed that this characteristic contributed to learning in an online Spanish course. This finding also supports the finding in the expert pool that identifies all four language skills as a fundamental characteristic in an online Spanish courses.

298

Audio Component

The presence of an audio component received the third highest mean score (4.1786), indicating that students considered the presence of an audio component as contributing to their learning in their online course. This finding also supports the expert pool's identification of an audio component as a fundamental characteristic in an online course.

Student-instructor Contact

Student-instructor contact received the fourth highest mean score (4.0357), also providing evidence that students' opinions showed instructor contact as contributing to learning in an online course and again supporting the finding from the expert pool.

Student-student Interaction

Student-student interaction received the lowest mean score of 2.8333 and was significantly lower than the 4.0 minimum (p < .001). This result indicates, in the students' opinion, student-student interaction did not contribute to learning in their online course. Student-student interaction was also ranked fifth by the expert pool and received a Level 2 importance. Thus, the students' responses once again corroborated the expert's views of characteristics of a quality online course.

Question Number Six of the Student Survey

Question number six of the student survey sought to determine whether there were any additional characteristics not identified by the expert pool that students taking online courses would find important to their learning of Spanish in the online course. Students responses to this question focused on two common themes: an audio component and communication. These comments complement the expert pool findings that identified an audio component and student-instructor/student-student contact as fundamental characteristics. The students added that not only does there need to be an audio component but that it should also be of high quality and of sufficient quantity to compensate for the online environment.

Moore's Theories of Distance Education and the Survey Findings

This study used Moore's theories of distance education as a theoretical base and aimed to test the research findings against Moore's theories of interaction and transactional distance. The findings reported here do not support Moore's theory of transactional distance. However, the results of the expert pool survey do support Moore's theory of interaction. Two of the five fundamental characteristics identified by the expert pool (student-instructor contact and student-student interaction) are directly related to Moore's theory of interaction. (The other three characteristics identified by the expert pool do not support Moore's theory of interaction). Student-instructor contact was identified by the expert pool as the

299

second most important characteristic and student-student interaction as the fifth most important characteristic in a quality online Spanish course. In addition, the responses of the expert pool to question number three, why they considered the characteristics identified in the previous two questions as fundamental in an online foreign language course, also support Moore's theory of interaction. Four of the 17 respondents reported student-instructor and/or student-student contact as a reason why they considered the characteristics fundamental.

These results offer evidence in support of Moore's theory of interaction and suggest that it is applicable to and compatible with online foreign language instruction. The responses in the student survey also support Moore's theory of interaction, but not as strongly as those of the expert pool survey. The student survey responses indicate that student-instructor contact is very important but that student-student interaction is not. Thus, the results of both surveys generally indicate that online course developers and instructors may be well advised to consider aspects of Moore's theory of interaction as an appropriate theoretical framework for their online courses.

SUMMARY AND CONCLUSIONS

The expert pool identified the following five characteristics (in order of importance) as fundamental to an online foreign language course designed for maximum learning outcome: (a) clear instructions, (b) student-instructor contact, (c) the inclusion of all four language skills, (d) an audio component, and (e) student-student interaction. Although this study focused on Spanish online instruction, it is of importance to note that the fundamental characteristics identified by the expert pool are likely not exclusively related to the Spanish language. They should be considered important for online course developers and instructors teaching other foreign languages as well, and it would be interesting to compare the findings of similar studies in other foreign languages.

The student respondents unanimously agreed that the above five characteristics are currently employed in their online Spanish courses. In addition, the student respondents reported that all of the characteristics contributed highly to their learning with the exception of one characteristic: student-student interaction. They ranked the five fundamental characteristics in the following order of importance: (a) clear instructions, (b) the inclusion of all four language skills, (c) an audio component, (d) student-instructor contact, and (e) student-student interaction.

These results indicate that online course developers and instructors should pay special attention to including clear instructions, student-instructor contact, all four language skills (reading, writing, speaking, and listening), and an audio component in an online foreign language course and take note that student-student contact, while important, is of secondary importance. In addition, because both groups of respondents agreed on the importance of including an audio component, clear instructions, and all four language skills (reading, writing, speaking, and listening), these three characteristics should be considered particularly important when developing or teaching online foreign language courses. Both groups also showed agreement on the most important fundamental characteristic (clear instructions)

300

and the least important fundamental characteristic (student interaction). Finally, the student comments on other characteristics they would find important (Question 6) enhanced the importance of the audio component and communication for student learning. Three of the six responses stated that either more or better quality audio would have contributed to student learning, and two of the six responses stated that communication (with instructor and students) would have contributed to their learning.

Thus, according to the findings of the expert pool and the student surveys, current Spanish online courses contain the fundamental characteristics necessary to contribute to maximum learning outcomes. As foreign language course developers/instructors begin or continue to develop/teach online, they should feel confident that the base for the courses described here is strong; they should take the fundamental characteristics identified by the expert pool and the students into account and may well wish to consider Moore's theory of interaction as an appropriate theoretical framework.

NOTES

¹ The potential expert online Spanish course developers and Spanish online instructors were from the following colleges and universities: Cerro Coso Community College, Empire State College, Florida Community College, Foothill College, Monash University, North Carolina State College, Portland Community College, Rio Salado Community College, Rogers State University, Saint Petersburg College, University of Maryland University College, Utah Valley State College, and The University of Washington.

The potential master Spanish instructors were from the following colleges and universities: Mesa Community College, Miramar Community College, San Diego State University, Southwestern Community College, and The University of San Diego.

REFERENCES

Benson, A., & Wright, E. (1999). Pedagogy and policy in the age of the wired professor: A case study of teaching Spanish online. T.H.E. journal, 27 (4), 60-68. Retrieved March 13, 2001, from http://ericae.net.ericdc/EJ601909.htm

Bush, M. D. (Ed.). (1997). Technology-enhanced language learning. Lincolnwood, IL: National Textbook Company.

Butler-Pascoe, M. E. (1999). Effective uses of computer technology in the development of writing skills of students enrolled in college-level English as a second language writing program. Unpublished doctoral dissertation, United States International University, San Diego, CA.

Cahill, D., & Catanzaro, D. (1997). Teaching first-year Spanish on-line. CALICO Journal, 14 (2-4), 97-114.

Glisan, E., Dudt, K. P., & Howe, M. S. (1998). Teaching Spanish through distance education: Implications of a pilot study. Foreign Language Annals, 31 (1), 49-56.

301

Harrell, W. L., Jr. (1998). Computer assisted instruction: Enhancements for language-learning applications (distance education) (Doctoral dissertation, Louisiana State University). Dissertations Abstracts International, 60 (03A), ED 355 835.

Institute for Higher Education Policy. (1999). What's the difference? A review of contemporary research on the effectiveness of distance learning in higher education. Retrieved March 5, 2002, from http://www.ihep.com/Pubs/PDF/difference.pdf

Kartchner, R. E., & Judge, P. (2000). Foreign language learning through the Internet: A pilot study. Retrieved November 7, 2000, from http://www.wpine.Com/sub/DLRC /study-UARIZ.html

Kennedy, G., David, L., & Omeadhra, B. (1989). Computers in language teaching. Dublin: Proceedings from IRAAL Seminar. (ERIC Document Reproduction Service No. ED 349 794)

Larson, S. (1999). It can be done—It's this simple: Implementing a basic, streamlined methodology for teaching a foreign language to adults on the Internet. Unpublished master's thesis, Northern Kentucky University, Highland Heights, KY.

Moore, M. G. (1983). The individual adult learner. In M. Tight (Ed.), Adult learning and education (pp. 153-168). London: Croom Helm.

Moore, M. G. (1989). Three types of interaction. The American Journal of Distance Education, 3 (2), 1-6.

The National Council of State Supervisors of Foreign Languages (NCSSFL). (1997). Characteristics of effective foreign language distance learning programs. (ERIC Document Reproduction Service No. ED 419 387)

North Carolina State Department of Public Instruction. (1997). Cover to cover: A guide to foreign language programs, instruction and resources. (ERIC Document Reproduction Service No. ED 419 387)

Osuna, M., & Meskill, C. (1998). Using the World Wide Web to integrate Spanish language and culture: A pilot study. Language Learning and Technology, 1 (2), 71-92. Retrieved April 17, 2000, from http://llt.msu.edu/vol1num2/article4/default.htm

Palloff, R. M., & Pratt, K. (1999). Building learning communities in cyberspace. San Francisco: Jossey-Bass Publishers.

Russell, T. L. (1999). The no significant difference phenomenon. Chapel Hill: North Carolina State University Press.

Saba, F. (2000). Research in distance education: A status report. International Review of Research in Open and Distance Learning, 1 (1), 1-9.

Saba, F., & Shearer, R. (1994). Verifying key theoretical concepts in a dynamic model of distance education. The American Journal of Distance Education, 8 (1), 36-59.

Stepp-Greany, J. (2002). Student perceptions on language learning in a technological environment: Implications for the new millennium. Language Learning and Technology, 6 (1), 165-180. Retrieved February 11, 2002, from http://llt.msu.edu/vol6num1/steppgreany

Warschauer, M. (2000). On-line learning in second language classrooms: An ethnographic study. In M. Warschauer & R. Kern (Eds.), Network-based language teaching: Concepts and practice (pp. 41-58). Cambridge: Cambridge University Press.

302

APPENDIX

Student Survey

303

304

Thank you for your participation in completing this survey.

305

AUTHOR'S BIODATA

Dr. Margaret Rose Don has 7 years of experience teaching part-time Spanish at the university level for San Diego State University and United States International University in San Diego, CA. She received her doctoral degree in Education and Technology from United States International University in San Diego, CA, two master's degrees from San Diego State University (English and Spanish), and her bachelor's degree in Spanish from Occidental College in Los Angeles, CA. She is currently a full-time professor of Spanish at the College of the Bahamas in Nassau, The Bahamas.

AUTHOR'S ADDRESS

Margaret Rose Don

Lecturer of Spanish

The College of The Bahamas

P.O. Box N-4912

Nassau, Bahamas

Phone: 242/302-4559

Email: mdon@cob.edu.bs

3. Monitoring Bilingualism: Pedagogical Implications of the Bilingual Tandem Analyser

KLAUS SCHWIENHORST
Trinity College, Dublin
ALEXANDRE BORGIA
Longueuil, Québec

Abstract:
Tandem learning is the collaborative learning partnership of two language learners with complementary language combinations, for example an Irish student learning German and a German student learning English. One of the major principles in tandem learning, apart from reciprocity and learner autonomy, is balanced bilingualism. While learners may find it relatively easy to control their bilingualism in email exchanges, it is not so easy to do so in synchronous text-based exchanges in object-oriented multi-user domains (MOOs), where measurements of bilingualism are often crude and inaccurate. The purpose of this paper is twofold. First, we wanted to develop and test a computerized tool, the Bilingual Tandem Analyser, that automatically analyzes and provides feedback on the languages that are used during a �live� exchange. Second, we wanted to implement the tool in a bilingual exchange between German and Irish students to see whether the balance in bilingualism improved. Our results show that the Bilingual Tandem Analyser is quick, reliable, and highly accurate for the four languages tested. When implementing the tool, we noticed that there is a noticeable improvement towards more balanced exchanges but that more work is needed on the pedagogical implementation.

KEYWORDS

Learner Autonomy, MOO, Tandem Learning, Bilingualism, Text-based Communication

INTRODUCTION

Tandem learning is the collaborative learning partnership of two language learners with complementary language combinations, for example an Irish student learning German and a German student learning English. One of the major principles in tandem learning, apart from reciprocity and learner autonomy, is bilingualism. The principle of bilingualism suggests that a successful partnership requires both

349

learners to use both languages equally; on the one hand, to provide sufficient target language input for their partner and, on the other hand, to be “pushed” to produce target language output themselves. While learners may find it relatively easy to control their bilingualism in email exchanges, it is not so easy to do so in synchronous text-based exchanges in object-oriented multi-user domains (MOOs), where measurements of bilingualism are often crude and inaccurate.

The purpose of our research was twofold:

1. How accurately can a computerized tool analyze the language proportions in synchronous text-based, bilingual-learner exchanges, compared to manual analyses of transcripts?

2. When learners are given a tool that accurately analyzes their bilingual proportions in exchanges, is there a noticeable effect on their bilingual behavior, that is, do learners try to work towards a more balanced use of L1 and L2?

We have structured the paper in the following manner. First, we will give a short overview of tandem-learning principles and the rationale behind this study. Second, we will describe the design specifications for the Bilingual Tandem Analyser (BTA) and examine how accurately it is able to analyze text-based input in real time and provide feedback to learners. In our third section, we will look at the pedagogical framework for our study and the implementation of the BTA tool. We will focus on comparing bilingual proportions in two project groups, one that used the BTA and one that did not. In our conclusion, we will summarise the main results and suggest where the pedagogical and technological framework needs to be improved.

THREE PRINCIPLES OF TANDEM LEARNING AND THE RATIONALE BEHIND THE BTA

In a tandem learning partnership, two learners with complementary L1 and L2 combinations work together (e.g., an Irish learner of German works with a German learner of English), either face-to-face, via asynchronous email, or synchronous tools such as text-based object-oriented multi-user domains (MOOs). Tandem learning, as one of the pedagogical implementations of learner autonomy principles (see Little, 1991), is based on the three principles of (a) bilingualism, (b) reciprocity, and (c) learner autonomy (Little & Brammerts, 1996). Reciprocity requires both partners to help each other and adjust to each other's proficiency levels; this could take the form of error correction, modification of input, the use of repair strategies, and so forth. In such a partnership, learners need to understand that the success of the partnership relies on an equal effort by both partners. Learner autonomy, as a principle in tandem learning, implies that both learners need to take responsibility for their exchange (and, thus, their language learning) by negotiating topics, arranging working methods, and generally planning, monitoring, and evaluating learning processes and outcomes. In this way, learners are not only

350

responsible for the success of their own learning but are also in part responsible for the success of their partner's learning. Face-to-face and email tandem learning partnerships have been organized and coordinated by the ETandem Network (see http://www.tcd.ie/CLCS/tandem), but in recent years, tandem partnerships have also been implemented using synchronous environments such as text-based object-oriented multi-user domains (MOOs) (see Kötter, 2002, 2003; Schwienhorst, 2003a, 2003b, 2004). MOOs have established themselves over the last few years as a valuable language learning tool (see Beatty, 2003; Shield, 2003; Shield, Davies, & Weininger, 2000; Sotillo, 2000; Trebbi, Jopp, & Coco, 2003).

A recurrent problem in both email and MOO tandem partnerships is that the more proficient L2 often takes over as the exclusive language of communication. However, a balanced and reciprocal partnership is vital for the success of these exchanges to ensure both exposure to comprehensible input and the production of “pushed” comprehensible output. It is common sense that the problem of imbalanced bilingualism is more prominent in synchronous exchanges than in asynchronous exchanges. Learners will naturally find it easier to arrive at a reasonable estimate of their bilingual proportions in an email to their partners and then act upon it. However, in transcripts of synchronous bilingual 1-hour sessions, learners (and even researchers) would find it difficult to arrive at a reasonable estimate regarding the proportions of the two languages used. In synchronous exchanges, the problem is also further complicated by the fact that code switching is much more likely to occur in synchronous exchanges than in asynchronous exchanges where learners have more time for editing and revising their utterances.

At the Centre for Language and Communication Studies (CLCS) at Trinity College, Dublin, we have conducted tandem exchanges in synchronous, text-based MOOs since 1998. My colleague Breffni O'Rourke and I have always noted that the stronger L2 was used much more than the weaker L2. In our German-Irish exchanges, this meant that English became the dominant language of communication between students (see O'Rourke, 2002, 2005; Schwienhorst, 2000). This is no surprise because students had no tool at their disposal to control their bilingualism other than a rough guess on whether they had used both languages equally.

For email, Christine Appel developed the Electronic Tandem Resources web site for e-mail tandem exchanges (Appel, 1999; Appel & Mullen, 2000), which uses n-gram analysis to analyze learners' emails and provide detailed feedback on language statistics. We wondered whether a combination of pedagogical and technological tools could alleviate the problem of imbalanced bilingualism in synchronous environments like MOOs. First, this required developing a system that can accurately analyse which language is being used at any given moment, and display this statistical information to each individual learner. Second, this system had to be accepted by students as a tool that would help them with their exchanges, that is, a tool that would give them control over their bilingualism at any time and then allow them to react accordingly rather than make them follow the dictates of a control mechanism imposed by the teachers/administrators.

351

DESIGN SPECIFICATIONS FOR THE BTA

In developing, implementing, and evaluating the BTA, we had to take into account both pedagogical and technological considerations. On the one hand, we had to develop and implement a system that accurately and reliably analyzed which language was being used at any given time, store that information, and present feedback to learners on an individual basis. The statistical presentation of data should include “live” statistics (so learners could adjust their bilingualism during sessions with their tandem partner), as well as daily, weekly, monthly, and global statistics on their bilingualism. Also, the system should not be perceived by learners as an intruder in their tandem exchange but, instead, as a tool to help them with their exchange, not a mechanism created by the teachers/administrators of the MOO to control their behavior.

Nevertheless, we were certain that students had to be automatically presented with statistics on a regular basis. After all, this is not a tool found in traditional face-to-face communication, and learners need to become accustomed to it. We consider it important, especially in the framework of learner autonomy, that learners accept tools as personally meaningful, as devices that offer them better ways of learning a language. Only then can we expect learners to accept responsibility for and take control of their learning. In practice, the design and implementation of such a tool can be a balancing act, where, in one extreme, the tool is too far removed from learners' stage of autonomy that they do not use it, or, in the other extreme, the tool is forced on learners and impinges on their autonomy, in which case the tool may be used but without becoming personally meaningful and useful to the individual learner (for a more in-depth discussion of this issue, see O'Rourke, 2002; Schwienhorst, 2002; O'Rourke & Schwienhorst, 2003).

The first problem we encountered in the design phase was deciding on what basis we should assess bilingualism. There are three aspects that should be mentioned. First, in previous projects, we had simply told our students to spend 30 minutes talking in one language and 30 minutes in the other. However, that did not work since students came in late, took some time to find their partner, left early, and so on. In addition, the better L2 speaker would, of course, produce much more text in half an hour than the weaker L2 speaker (and there usually were big differences in L2 proficiency between German students of English and Irish students of German). One of the decisions that had to be made was, therefore, whether we should measure bilingualism by the amount of text students produce or the time they spend processing and producing utterances in the L2. We decided that the amount of text produced was a more reliable and useful option because it was less dependent on network delays, time spent on looking up words in online dictionaries, and other external factors. Second, O'Rourke (2002), which provided comparison data for our BTA, had analyzed bilingualism on the basis of partnerships rather than individual learners. For our purposes, we found it more useful for the system to analyze bilingualism in individual learners; otherwise, the extreme case of one learner speaking one language and the other speaking the other language all the time would end up in a 50:50 balance. Third, bilingualism could be measured by counting words or utterances. From a sociopragmatic perspective, we

352

thought it was more useful to measure bilingualism on the basis of utterances. As mentioned earlier, code switching would negatively influence a word-by-word analysis, so we decided for the utterance as the framework for analysis.

On the basis of these three decisions, we developed the BTA tool. The main components of the system are the Language Categorisation Tool (LCT) and the Student Journal. The LCT is the “heart” of the BTA because it is able to guess which language is being used at any given time in whole MOO utterances. The LCT was inspired by TextCat (Van Noord, 1999) and follows n-gram analysis (Appel & Mullen, 2000; Cavnar & Trenkle, 1994). Although we decided to start with four languages (English, German, French, and Italian), the LCT allows for new language profiles to be developed and parameters to be adjusted (e.g., if the accuracy of the tool is affected).

The Student Journal simply displays the results of the LCT analysis to the learners in the form of overall (or global) statistics, monthly, weekly, daily, and “live” statistics—to check bilingualism during a MOO session.

Figure 1

Display Options of Language Statistics in the Student Journal

The Student Journal is displayed automatically with global statistics whenever learners connect to the MOO. After that, it is completely up to students when and how often they want to display it by typing in <@journal>. The Student Journal is also available to the MOO administrators and teachers for research purposes, and administrators can receive regular overviews by email of how the LCT analyzes input, as shown in Figure 2.

353

Figure 2

Excerpt from Administrator's Log File

** [Irish student] - Thu Mar 27 11:40:34 2003 GMT

** German

(bleh! wir haben kein umlauten )“

** [Irish student] - Thu Mar 27 11:40:36 2003 GMT

** English

heh heh cool snow :)“

** [Irish student] - Thu Mar 27 11:40:39 2003 GMT

** German

Hi Stephanie wie gehts?“

** [Irish student] - Thu Mar 27 11:40:45 2003 GMT

** English

Hmm, would it be Ok if we changed the language now?“

** [Irish student] - Thu Mar 27 11:41:33 2003 GMT

** too short

Ok“

** [Irish student] - Thu Mar 27 11:42:12 2003 GMT

** ambiguous

yah ! Umlauten!“

Our first question was whether the LCT can perform the analysis accurately and efficiently. The next section examines in more detail the concepts behind n-gram analysis and a comparison between machine analysis and manual (human) analysis of bilingual input.

THE LANGUAGE CATEGORIZATION TOOL

N-gram analysis is a fast and adaptive way to determine general characteristics of unknown media (Cavnar & Trenkle, 1994). Since n-gram analysis has already been proven to be efficient for document analysis over the years, this section will focus on the main concerns when implementing it for synchronous chat environments, starting with a brief overview of the method. The Language Categorization engine works by doing statistical comparisons between texts of a known and unknown nature. Profiles are built by parsing documents into small tokens that are sorted by frequency of appearance, basically resulting in a list of most commonly used combinations of letters. Similarities found between such profiles suggest similarities between their original documents: a common topic, for instance, if looked at closely—but more obviously, a common language. The actual categorization is done by simply gathering a database of representative languages using literary works as a default comparison base. The engine queues user inputs and processes

354

them one after another against all previously built language profiles, keeping individual statistics of the most relevant matches. The engine is fast enough to provide real-time feedback to the Student Journal, inputs taking about one second each to process on a standard workstation with four concurrent languages.

By abstracting language into statistics, the determination process can handle streams of text very well even if the stream contains broken syntax, misspellings, or typographical mistakes, all of which are commonly found in chat transcripts. There are, however, a few drawbacks in not using sets of rules or a lexicon for formal analysis. Since the engine relies on accumulation and comparison of data to make decisions, it can only analyze text as blocks and will usually remain clueless when dealing with very short inputs. This is a main concern when monitoring chat activities because the speakers will often break up their sentences into smaller utterances, providing sparse information about logical breakpoints (e.g., uppercase characters or punctuation marks) to combine them into bigger, less ambiguous blocks that are more likely to contain a single language.

One way around this problem was to reinforce the recognition scheme with mechanisms that keep erroneous matches out of the analysis. Small utterances (less than around 10 characters long) are automatically blocked by the engine; also, at some point, matches will be considered ambiguous if the results are too close between languages. Furthermore, if users are known only to deal with specific languages in their learning activities, the other languages will be ignored along with those already marked erroneous when balancing user statistics. As shown by the results in Table 1, this method is quite efficient and accurate, the ratio of language usage among tandem users in session transcripts being similarly detected by automated and human analysis, with a difference equal to or less than 2%. This difference is not statistically significant (p > 0.05).

Table 1

Comparison of Machine and Manually Parsed Results (O'Rourke, 2002)*

Session	# English utterances	English (%)		# German utterances	German (%)
		Automatic	Manual		Automatic	Manual
A	460	75%	76%	143	24%	24%
B	440	73%	73%	166	26%	27%
C	331	86%	87%	48	13%	13%
D	604	96%	97%	20	3%	3%
E	416	84%	86%	69	15%	14%
F	292	87%	90%	33	12%	10%
Pooled across sessions	2,543	83%	84%	479	16%	16%

*Analysis based on 97 transcripts.

355

Language activity monitoring has thus been successfully tested in a real-time chat application. The Language Categorization Tool is currently available for use in a MOO environment, bundled with the Journal service, and provides easy commands to create and manage language databases and to fine tune resource consumption (for free download and documentation, see Borgia, 2003). Due to the nature of the parsing, the engine is also able to gather basic sentence statistics, such as word count or average word length, along the way which are available for display. These statistics may prove useful for future research agendas.

PEDAGOGICAL FRAMEWORK FOR THE IMPLEMENTATION OF THE BTA AND METHODOLOGY

Before looking at the implementation of the BTA in our tandem projects, it would be helpful to describe the pedagogical framework of our language courses in which the projects took place. The subjects in our study consisted of Information and Communication Technology students from Trinity College Dublin, who were studying German as part of their degree program and Information Technology students from the Fachhochschule Bonn-Rhein-Sieg, who were studying English as part of their degree program. The German students had completed an English entry exam and could be rated as lower advanced English speakers, while the Irish students' proficiency in German could be described as higher beginning or lower intermediate. The students formed learning partnerships on the basis of introductory emails sent by the Irish students to the German students and from which the students selected their partners. Since the number of German students was much higher than the number of Irish students, some 1+2 partnerships had to be formed in addition to the 1+1 partnerships. Students worked together with their selected partner(s) for 1 hour per week in scheduled class sessions. We compared bilingualism in this exchange (2002-03) with similar exchanges from 2000-01 (see Table 2), mainly for two reasons. First, the number of transcripts was almost identical and therefore provided us with a reasonable amount of comparable data. Second, the 2000-01 exchanges had been the subject of O'Rourke's (2002) manual analysis of bilingualism, and the original data were available to us with permission by the students.

Table 2

Overview of the Subjects in the 2000-01 and 2002-03 Studies

Year	Student groups	Tandem set up	Number of weeks/transcripts
2000-01	26 Irish, 34 German	26 pairs, 8 extra German	6/298
2002-03	12 Irish (1 native German) 18 German	6 pairs, 6 groups of 1+2	12/286

Students in all groups worked on bilingual tasks with a final assessed product, which they were able to access in the MOO and on a paper print out (see Table 3)

356

Table 3

Sample Activity and Task

Topic 5: Work experience	Thema 5: Erfahrungen in der Arbeitswelt
AIM:	ZIEL:
to discuss previous work experience	Jobs diskutieren
to develop vocabulary for a CV	Vokabeln für einen Lebenslauf sammeln
TIME: 1-2 sessions	ZEIT: 1-2 Stunden
Discuss with your partner:	Diskutiere mit Deinem Partner:
What kind of jobs have you been doing?	Welche Jobs hast Du schon gehabt_ Als was hast Du schon gearbeitet?
Which jobs did you particularly like/dislike? Why?	Welche Jobs haben Dir besonders gefallen/nicht so gut gefallen? Warum?
Do you have plans for other jobs? Describe them!	Hast Du schon Pläne für andere Jobs? Beschreibe sie!
What information is vital for a CV? Are there differences between German and Irish CVs?	Welche Information sollte in einem Lebenslauf stehen? Gibt es Unterschiede zwischen deutschen und irischen Lebensläufen?
Final task: write a CV and send it to your partner for correction.	Abschluss: schreibe einen Lebenslauf und schicke ihn an Deinen Partner zur Korrektur.

Within class sessions, we organized training sessions to introduce learners to the tools and tandem principles and, half way through the project, reflection sessions to have them reflect on their behavior. We further encouraged learners to keep online learner diaries in the form of Dam's class diaries (Dam, 1995) and to work on their individual session transcripts that were sent to them automatically after each session.

We accessed the statistics for all students taking part in the 2002-03 exchange directly in the MOO database. For the 2000-01 exchange, transcripts were edited using Textpad (http://www.textpad.com) and then individually sent through the MOO for analysis by the BTA. Thus, the transcripts from the 2000-01 group were analyzed in the same way as “live” data from the 2002-03 group, and we were able to collect the data for individual and global statistics. (We did not simulate weekly or monthly statistics because we did not consider them relevant for the analysis here.)

We determined a threshold of 70% to 30% (or 30% to 70%) as a useful measure for balanced bilingualism, that is, if an individual MOO session or the global statistics of a student were 64% English and 38% German, it was deemed balanced, whereas 74% English and 26% German were deemed imbalanced. We are aware that setting this threshold could be considered arbitrary, but it seems like a useful threshold for our purposes. We then counted the amount of individual MOO sessions and global statistics that were within that range and the average percentages across all sessions/global statistics.

357

COMPARING THE RESULTS OF BILINGUALISM IN TWO MOO TANDEM PROJECTS

We examined in detail the individual sessions by learners and their global statistics (i.e., their bilingual proportions over the whole exchange). In this section, we will first present the analysis of the individual sessions and then look at the global statistics.

Table 4 presents an overview of bilingualism in the individual MOO sessions. We can see that the 2002-03 average of all sessions, with the BTA in use, approached balanced bilingualism and was just within the 70:30 range, whereas the 2000-01 group without the BTA was considerably imbalanced in favor of the stronger L2 (English). The number of sessions that were within the range is slightly higher in the 2002-03 group (32%) than that in the 2000-01 group (25%).

Table 4

Bilingual Proportions of English (E) and German (G) in Individual MOO Sessions

Student group	Individual sessions average (E%:G%)	Number of individual sessions in 70:30 range
2000-01	78%:22%	75 out of 298 (25%)
2002-03	67%:32%	92 out of 286 (32%)

We then looked at the results in the global statistics (i.e., daily, weekly, and monthly statistics). These were the statistics that were presented to the students automatically when they connected to the MOO. These more broadly based statistics can be considered more reliable than the results from individual sessions (see Table 5).

Table 5

Bilingual Proportions of English (E) and German (G) in the Global Statistics

Student group	Global statistics average (E%:G%)	Number of global statistics in 70:30 range
2000-01	76%:23%	15 out of 60 (25%) 6 Irish (23%) 9 German (26%)
2002-03	63%:36%	16 out of 30 (53%) 8 Irish (67%) 8 German (44%)

Similar to the results of the analysis of the individual sessions, the results of the analysis of the global statistics show a greater balance of bilingualism in the 2002-03 group than in the 2000-01 group, but the improvement is dramatically greater: 16 out of 30 sessions (53%) in 2002-03 versus 15 out of 60 sessions (25%) in 2000-01. Considering the 70:30 ratio as a criterion of success, the difference

358

between the groups is significant (chi square = 7.11, p = .008). Although the German students improved in the number of successful sessions, the far greater contribution to improvement came from the Irish students. Finally, when we looked at the individual partnerships, we noticed that in three out of the six 1+2 partnerships all participants had imbalanced bilingualism, whereas in the 1+1 partnerships there was at least always one learner within the 70:30 range, suggesting that 1+1 partnerships offer the most effective framework for balanced bilingualism in the MOO.

CONCLUSION

In our design, implementation, and evaluation of the BTA we learned some valuable lessons for future bilingual exchanges in text-based synchronous environments. The major results of our study are

1. the BTA provides learners with a very accurate tool to control bilingualism in synchronous text-based environments,

2. L1/L2 proportions are more balanced with the BTA tool than without it,

3. global and individual stats are substantially more balanced with the BTA tool, and

4. closer examination showed that three out of six 1+2 groups in 2002-03 were all imbalanced, whereas in pairs at least one learner was balanced.

Still, we need to consider other factors that may have played a role and ask how we can further improve balanced bilingualism in tandem exchanges. Thus, for example, training and reflection sessions can raise learners' awareness of the principles of tandem learning and subsequently influence their bilingualism, but the number of sessions needed remains an unanswered question. In order to increase the balance of bilingualism, we could implement technological improvements, such as “friendly reminders” that caution students when they leave the specified zone of balanced bilingualism. These reminders have already been built into the email web site by Appel (personal communication), and it would be easy to do the same for synchronous MOOs. We could also experiment with the default display of the statistics: would it maybe be more beneficial to students to always or periodically be informed of statistics during a session, after a certain time, or after students leave a certain zone of balanced bilingualism? We also need to consider other forms of support for the weaker L2 speakers such as integrating dictionaries into the MOO. In terms of pedagogical improvements, we should definitely avoid 1+2 partnerships in future projects. It appears that one student is simply overwhelmed by the input and the decisions of two other students.

In summary, we can say that the development of the BTA has been successful, although the results of its implementation suggest the need for more research to increase its instructional value. Now, to our knowledge, for the first time, learners have an accurate tool (so far, for English, German, French, and Italian) that gives them full control over their bilingualism in synchronous text-based tandem exchanges. The tool is freely available on our web site (http://kontakt.tcd.ie/BTA)

359

and may become an integrated part of the EnCore MOO database in the future. The fact that learners use the tool to adjust their language proportions indicates that they see it as a valuable addition to the language learning tools at their disposal during exchanges (see also Kapec & Schwienhorst, 2005). As such, the BTA can become a useful tool to develop learner autonomy because it gives learners control over an important aspect of bilingual language learning exchanges.

REFERENCES

Appel, M. C. (1999). Tandem learning by e-mail: Some basic principles and a case study (Vol. 54, CLCS Occasional Paper). Dublin: Trinity College, Centre for Language & Communication Studies.

Appel, C., & Mullen, T. (2000). Pedagogical considerations for a web-based tandem language learning environment. Computers and Education, 34 (3-4), 291-308.

Beatty, K. (2003). Teaching and researching computer-assisted language learning. Harlow: Longman.

Borgia, A. (2003). Moo Utilities [Web page]. Retrieved August 24, 2003, from http://kontakt.tcd.ie/BTA

Cavnar, W. B., & Trenkle, J. M. (1994). N-gram-based text categorization. In Proceedings of third annual symposium on document analysis and information retrieval (pp. 161-175). Las Vegas, NV: UNLV Publications/Reprographics.

Dam, L. (1995). Learner autonomy 3: From theory to classroom practice. Dublin: Authentik.

Kapec, P., & Schwienhorst, K. (2005). In two minds? Learner attitudes to bilingualism and the Bilingual Tandem Analyser. ReCALL, 17 (2), 254-268.

Kötter, M. (2002). Tandem learning on the Internet. Frankfurt: Peter Lang.

Kötter, M. (2003). Negotiation of meaning and codeswitching in online tandems. Language Learning & Technology, 7 (2), 145-172. Retrieved October 12, 2005, from http://llt.msu.edu/vol7num2/kotter/default.html

Little, D. (1991). Learner autonomy 1: Definitions, issues, and problems. Dublin: Authentik.

Little, D., & Brammerts, H. (Eds.). (1996). A guide to language learning in tandem via the Internet (Vol. 46, CLCS Occasional Paper). Dublin: Trinity College, Centre for Language & Communication Studies.

O'Rourke, B. (2002). Metalinguistic knowledge in instructed second language acquisition: A theoretical model and its pedagogical application in computer-mediated communication. Unpublished doctoral dissertation, Trinity College, Dublin.

O'Rourke, B. (2005). Form-focused interaction in online tandem learning. CALICO Journal, 22 (3). 433-466

O'Rourke, B., & Schwienhorst, K. (2003). Talking text: Reflections on reflection in computer-mediated communication. In D. Little, J. Ridley, & E. Ushioda (Eds.), Learner autonomy in the foreign language classroom: Teacher, learner, curriculum and assessment (pp. 47-60). Dublin: Authentik.

360

Schwienhorst, K. (2000). Virtual reality and learner autonomy in second language acquisition. Unpublished doctoral dissertation, Trinity College, Dublin.

Schwienhorst, K. (2002). Pressures, potentials, and affordances: The role of tools in CALL environments. Communication & Cognition—Artificial Intelligence, 19 (3-4), 133-149.

Schwienhorst, K. (2003a). Learner autonomy and tandem learning: Putting principles into practice in synchronous and asynchronous telecommunications environments. Computer-Assisted Language Learning, 16 (5), 427-443.

Schwienhorst, K. (2003b). Neither here nor there? Learner autonomy and intercultural factors in CALL environments. In D. Palfreyman & R. C. Smith (Eds.), Learner autonomy across cultures: Language education perspectives (pp. 164-180). London: Palgrave Macmillan.

Schwienhorst, K. (2004). Native-speaker/non native-speaker discourse in the MOO: Participation and engagement in a synchronous text-based environment. Computer-Assisted Language Learning, 17 (1), 35-50.

Shield, L. (2003). MOO as a language learning tool. In U. Felix (Ed.), Language learning online: Towards best practice (pp. 97-122). Lisse: Swets & Zeitlinger.

Shield, L., Davies, L. B., & Weininger, M. J. (2000). Fostering (pro)active language learning through MOO. ReCALL, 12 (1), 35-48.

Sotillo, S. M. (2000). Discourse functions and syntactic complexity in synchronous and asynchronous communication. Language Learning & Technology, 4 (1), 82-119. Retrieved October 12, 2005, from http://llt.msu.edu/vol4num1/sotillo/default.html

Trebbi, T., Jopp, C., & Coco, M. (2003, 16-18 October). Didaktisk rollespill i en MOO: Er læringen satt på spill? Paper presented at the Digital Dannelse Conference, Oslo.

Van Noord, G. (1999). TextCat [computer software]. Retrieved 25 March, 1999, from http://odur.let.rug.nl/~vannoord/TextCat

AUTHORS' BIODATA

Klaus Schwienhorst works as co-ordinator of extracurricular language modules and Lecturer in Applied Linguistics at the Centre for Language and Communication Studies at Trinity College, Dublin, Ireland. He has published mainly in the area of learner autonomy and synchronous text-based communication tools. His main research interests lie in virtual reality, computer mediated communication, and learner autonomy for second language acquisition.

Alexandre Borgia is an independent software developer whose main interests revolve around computer-assisted pedagogy. He has worked on many private eLearning solutions and research tools for language studies in collaboration with Trinity College Dublin. He is co-founder of MOOFrançais, a virtual French community, which has the goal of promoting French and helping second language learners.

361

AUTHORS' ADDRESSES

Klaus Schwienhorst

CLCS, Arts Building

Trinity College Dublin,

Dublin 2

Ireland.

Phone: +353-1-608 3316

Fax: +353-1-608 2941

Email: kschwien@tcd.ie

Alexandre Borgia

3554 Rigaud

Longueuil

QC

Canada

J4L 4K7

Phone: 450/442-9168

Fax: 450/442-4301

Email: alexborgia@hotmail.com

4. A Tale of Two Communities: Group Dynamics and Community Building in a Spanish-English Telecollaboration

MARK DARHOWER
North Carolina State University

Abstract:
This study provides a theory-driven account of community building in a bilingual telecollaborative chat setting. A symmetrical arrangement of 70 L1 English learners of Spanish and L1 Spanish learners of English engaged in weekly Internet chat sessions in small groups. The learning metaphors of community and participation serve as the theoretical framework to describe linguistic and social behaviors and interpersonal relationships among participants in two ongoing chat groups, while, at the same time, discourse data are used to build upon theory of (virtual) community. Based on Brown's (2001) classification of levels of online community, the findings illustrate the discursive construction of one community that reached the third, cooperation/camaraderie, level and another that struggled to maintain the second, membership, level.

KEYWORDS

Chat Communities, ESL, Participation Metaphor, Spanish, Telecollaboration

These are early days in the exploration of the concept of community … in the emerging hybrid worlds in which people live … . Conversations and relationships are, for a growing number of people, mediated through new tools enabled by computing and telecommunications. These are exciting times, akin to the first decades of the written word … (Barab, Kling, & Gray, 2004, p. xiii)

INTRODUCTION

'Telecollaboration' (known as 'tandem learning' in Europe) has been an active area of second language acquisition (SLA) research in recent years. A telecollaboration consists of groups of geographically separated learners in which half of the learners are native speakers of the language the other half are learning, and vice versa. Telecollaborations are by nature virtual learning communities (Renninger

561

& Shumar, 2002). 'Community,' a metaphor oft employed in scholarship throughout the humanities and social sciences, is especially robust in SLA research as the field increasingly embraces the social, as well as cognitive factors involved in learning languages (Block, 2003; Firth & Wagner, 1997). Much of this research has been cast in the Vygotsky-inspired sociocultural theoretical framework (for an overview, see Lantolf & Appel, 1994) complemented with constructs such as 'community of practice' (Lave & Wenger, 1991). Although recent SLA studies have been carried out in the context of virtual learner communities (e.g., Belz, 2001, 2002b), research has yet to sufficiently define a role for community in L2 learning (Darhower, 2006). Studies are needed to enlighten the specific processes by which learners become full fledged members of L2 discourse communities (or not).

The current study analyzes a virtual community named the¹ Bilingual Chat Community (BCC). This collaboration between the North Carolina State University and the University of Puerto Rico unites L1 English speakers learning Spanish with L1 Spanish speakers learning English. The purpose is to initiate learners into a bilingual discourse community in which they can coconstruct meaning with native speakers in the L2 and share with each other aspects of their respective cultures. This study explores linguistic and social behaviors involved in community building in two groups of the BCC who chatted for a 10-week period. One of the groups constructed a cooperative, cohesive communicative environment in their chat room, whereas the other group established a very different type of communicative environment. Discourse analysis will illuminate the trajectory of each group as it constructs the communicative norms of its community.

LITERATURE REVIEW

Community as Language Learning Metaphor

According to Lantolf (1996), SLA research is largely metaphorical. Metaphors applied to the field range from single constructs (e.g., input/output) to theories of language and language acquisition (e.g., information processing or mind = computer). Metaphors help researchers conceptualize abstract ideas in a more concrete way. Community is a metaphor currently of great interest to academics across the humanities and social sciences. A related metaphor in learning theory is participation (Lave & Wenger, 1991), which views learning as a process of becoming a participant in a community of practice (CoP). This contrasts with other common learning metaphors, such as the acquisition metaphor which equates the human mind with a container to be filled with certain materials of which the learner is then the owner (Sfard, 1998, p. 5).

The CoP is perhaps the most developed concept of community in psychology and learning theory. A CoP consists of individuals who interact on a regular basis around a common set of issues, interests, or needs. Wenger (1998) established three criteria for defining a CoP. When members of the CoP accomplish something on an ongoing basis, they have a 'joint enterprise.' Members have 'mutual engagement' when they interact with one another to clarify their work and to

562

define and even change how the work is done. Through this mutual engagement, members establish their identities relative to the community. A 'shared repertoire' refers to the methods, tools, techniques, language, and behavior patterns that comprise the cultural context for the members' work.

In the participation view of SLA, learners ideally integrate themselves as full participants in some type of community that employs the L2 as its means of communication. As Sfard (1998, p. 6) puts it, "learning a subject is now conceived of as a process of becoming a member of a certain community. This entails, above all, the ability to communicate in the language of this community and act according to its particular norms." Defining whether a collection of human beings constitutes a community or not is an elusive endeavor, however. Some theorists establish specific criteria for community, such as Wenger's CoP. Riel and Polin (1994, p. 18) make the distinction that "a community differs from a mere collection of people by the strength and depth of the culture it is able to establish and which in turn supports group activity and cohesion."

The social activity that occurs in CoPs is inherently tied to group membership and identity (Riel & Polin, 1994). Norton (2001) argues the importance of understanding how learners develop identities as 'legitimate speakers,' that is, how they come to be accepted as fully functioning members of different CoPs with which they engage. The process of becoming a "fully functioning member" implies that learners evolve over time in their social roles and identities relative to the group. Rogoff (1994, p. 210) considers this process of "transformation of participation" an essential part of learning.

However, not all members of a community actively participate or necessarily enjoy participating. Some group members might not develop strong connections with each other, especially if the groupings are involuntary. As Riel and Polin (1994) point out, communities do not always entail healthy contexts of close interpersonal relationships but rather may be dysfunctional, scattered, or otherwise troubled.

As participant roles develop and are played out through discourse practices, the social history of the group is created and stored (Hall, 1993). Such history can then be brought forth for reconstruction. 'Historicity' (and futurity) can be noted in groups that have a mutual engagement as they build on their past and plan and prepare for their future.

Linguistic and Social Affordances

According to Wenger (1998, p. 72), "The negotiation of meaning is the level of discourse at which the concept of practice should be understood." In the L2 context, learners attempt to coconstruct meaning with each other or with more proficient speakers of the L2. Proficient speakers can offer linguistic resources to learners. Borrowing from psychologist James Gibson, Van Lier (2000) refers to linguistic resources available to learners as 'affordances.' Affordances can include such familiar SLA constructs as comprehensible input (Krashen, 1981, and elsewhere), negative feedback ( Aljaafreh & Lantolf, 1994), and scaffolding

563

(Donato, 1994; Guerrero & Villamil, 2000; Wood, Bruner, & Ross, 1976). Additionally, affordances are aspects of the linguistic world which can be "demands and requirements, opportunities and limitations, rejections and invitations, enablements and constraints (Shotter & Newson, 1982, as cited in Van Lier, 2000, p. 253). These can be considered 'social affordances.' In a social view of SLA, participation in a CoP is the activity of language learners. As learners participate in CoPs, affordances are available for them to make use of--or not.

Virtual Communities and Telecollaboration

Rheingold (1993) is credited with pioneering the concept of 'virtual community,' which he describes as social aggregations that emerge on the Internet when enough people carry on discussions long enough to form personal relations in cyberspace. In her extensive research into online community-building, Brown (2001) delineates three levels in the evolution of online communities. In the 'acquaintance' level, participants get to know each other. Brown refers to the 'community conferment or membership' level as the "membership card" for the community. Learners usually feel that they are members when they are part of long, thoughtful discussions with each other. The final level is 'camaraderie,' which is achieved after "long-term and/or intense association with others involving personal communication" (Brown, 2001, p. 24). These levels are not necessarily linear. In other words, a community that reaches camaraderie level could regress to membership level, and individual community members can be on different levels at different times. Brown also pointed out that community can be present for some individuals but not for others who for whatever reasons do not engage well with their counterparts.

In observing hundreds of online communities, Kim (2000) noted that participants always fit into some type of role, ranging from 'visitor' to 'novice,' 'regular,' 'leader' and eventually to 'elder.' Many participants evolve from the visitor or novice role to regular member, and most communities have one or more individuals in a leader role at any given time. Renninger and Shumar (2002, p. 298) note that "learners play a variety of roles and may participate in various ways, from active to passive."

SLA scholars remain intensely interested in the emergent culture of electronically mediated discourse (e.g., Belz & Thorne, 2005), although studies that specifically address community as a construct are relatively few. Darhower (2006) made a case for a stronger role for community in National Standards-based L2 learning. Arnold, Ducate, Lomicka, & Lord (2005) demonstrated the unique types of learning that take place when teachers in training from geographically dispersed places unite in an online community.

Studies of telecollaborative learning communities highlight linguistic, social and cultural factors of the communities, providing information on both the benefits and potential disadvantages of telecollaborations (Belz, 2001, 2002a; Kramsch & Thorne, 2002; Schneider & von der Emde, 2005). In telecollaborations between American and German students, for example, Belz (2001) found that such factors

564

as language valuation, access to technology, and matching of proficiency levels have a profound effect on the linguistic and interpersonal nature of telecollaborative learning environments. Belz's conclusion was that "telecollaboration does not unproblematically afford target language interaction in all cases" (p. 229).

The current study seeks to complement extant research on telecollaborations by illustrating the specific social discursive processes involved in participating in and acquiring membership to virtual bilingual language communities.

THE STUDY

Context and Participants: The Bilingual Chat Community (BCC)

The BCC is a web-enabled community which provides a forum for English-speaking learners of Spanish to communicate in weekly chat sessions with Spanish-speaking learners of English. The physical space of the BCC resides in the chat rooms hosted by the WebCT server at the North Carolina State University and the BCC web site containing photographs and biographical information on the participants, as well as information they need to be productive members of the community, such as weekly topic assignments (see topics list in Appendix A) and the chat schedule. The participants in the semester's BCC were approximately 35 Spanish students from North Carolina State University (NCSU) and 35 English students from the University of Puerto Rico (UPR). Those included in the study were 12 members of two different groups. Four of the 12 were male, and 8 were female. All 12 members were the traditional university age (19-22), and all were estimated to be somewhere in the intermediate-mid to intermediate-high ACTFL oral proficiency level.

Research Question

The general research question driving this case study is: What linguistic and social behaviors define the process of forming a community and becoming full participants in the community?

Data Collection

Weekly transcripts collected by the chat server were the principal data source. In qualitative research, it is often necessary to reduce large quantities of data. With 11 groups chatting for 10 weeks, there were a total of 110 chat episodes, each consisting of anywhere from 5 to 11 pages of text. To reduce the enormous amount of data collected, the researcher employed purposive sampling (Miles & Huberman, 1994), selecting two groups for analysis: Group E and Group D. The participants in Group E appeared to form a cohesive community throughout their ten weeks of chat, developing meaningful interpersonal relationships mediated by electronic discourse. Group D, at the other extreme, did not appear to form a tight social bond. The two groups were selected to contrast levels of community development. The researcher does not wish to give the impression that these two groups are representative of the BCC as a whole. The remainder of the 11 groups

565

fell somewhere between Group E and Group D in terms of a cohesion continuum, based on the researcher's impressionistic review of the 110 chat logs.

To answer the research question, the researcher made reiterative passes through the data, viewing the chat scripts through the lens of the three-level framework established by Brown (2001) (i.e., acquaintance, membership, and camaraderie levels). Within that framework emerged several categories of analysis: experiences with the L2, historicity and futurity, integration of new members, member roles and regulatory behaviors, members versus nonmembers, and linguistic and social affordances. Discourse excerpts were selected to illustrate each category.

DATA ANALYSIS: GROUP E

Five learners were assigned to Group E: two from NCSU and three from the UPR. There were also six visitors who joined the group for one to three sessions, but the analysis focuses on members originally assigned to the group. Group E had an attendance rate of 88%, that is, a total of only six absences dispersed among the five participants over the course of 10 weeks. Table 1 shows which participants attended each weekly chat session.²

Acquaintance Phase (Weeks 1 and 2)

Experiences with the L2

In Week 1, Bob, Manuela, and Ricardo become acquainted with each other. (It is noteworthy that these three participants had excellent attendance, so they chatted with each other all 10 weeks.) In Excerpt E1, the chatters discuss the extent of their experience with their L2 (lines 1-7) and aspects of using the L2 that are most challenging for them (lines 8-20). In lines 22-24, they look favorably upon their chat experience in helping them develop their L2 skills.³

566

Excerpt E1 (Week 1): Experiences with the L2

Discussing their experiences with their L2 sets the stage for the bilingual social relationships that the participants will develop throughout ten weeks of chatting together.

Historicity and Futurity

After the first moments of their initial chat, the chatters not only have a present but also a past and a future. In other words, they possess historicity and futurity. An example of this is found in Excerpt E2 in which Ricardo remembers that Bob had a birthday (line 2), then both Bob and Lisa wish Ricardo a happy birthday (line 10). (Coincidentally, three of the group members had a birthday during the semester.) Later, in Week 10 (not included in this excerpt), the other birthdays are remembered.

Excerpt E2 (Week 10): Historicity/Futurity

567

The group's sense of historicity and futurity is a good indicator that by Week 2 they were already developing a sense of community.

Integration of New Members and Establishment of Roles

In Week 1, Bob, Manuela, and Ricardo are novices since they had never chatted together before. Since the BCC has preestablished rules governing chat topics and equal use of English and Spanish, somebody must take on the role of initiating and maintaining the topic, switching languages, and ending the chat session. Bob fulfills this role in the first chat. Interestingly, as shown in Excerpt E3, Manuela (the only female in the group) asks the two gentlemen to change the topic of discussion to something to which she can relate (line 1). In doing so, she asserts her rights as a full member while still a novice in the community. In lines 3, 5, and 6, Bob demonstrates that he is happy to accommodate Manuela's request.

Excerpt E3 (Week 2): Manuela's request to change topics

Later in Week 2 (not illustrated in the excerpt), Lisa and Nilsa integrate themselves into the community. Nilsa enters and greets the others, then asks to be initiated: es la primera vez que entro asi que ubiquenme 'it's the first time I'm entering, so orient me.' Lisa enters and greets the community members, then instantly integrates herself into the discussion by exchanging introductory information with her fellow chatters. Manuela is pleased that she is no longer the only female in the chat: chicas nuevas en la sala … al fin no me siento sola jejeje 'new girls in the room … at last I don't feel alone hahaha.' Bob directs the group to get to know the newcomers before continuing the conversation: Parece que tenemos que empzar a conocernos otra vez! 'It looks like we have to get to know each other again!'

All assigned participants have chatted at least once by the end of Week 2. The requirements for membership seem to be clear to them, so the community enters the community membership level.

568

Community Membership Level

By the third chat session, most members of Group E become regulars of the community (except Nilsa, who is absent the third and fourth weeks), meaning that they are initiated into the norms and communicative patterns of the community.

Roles and Regulatory Behaviors

Wenger (1998, p. 74) notes that "The kind of coherence that transforms mutual engagement into a community of practice requires work. The work of 'community maintenance' is thus an intrinsic part of any practice." During the membership level, participant roles evolve. As shown in Table 2, Lisa picks up the regulatory functions soon after she enters her first chat (Week 2, to which she was 19 minutes late) and maintains these functions almost exclusively throughout the 10 weeks, except for the week she is absent (Week 6). It can be said, then, that Lisa fulfills the leader role, taking this over from Bob after Week 1.

Members and Nonmembers

After a few weeks of chatting, the group members know who is part of their community and who is not. Excerpt E4 shows what happens when a visitor (Adalia) enters.

Excerpt E4 (Week 3): Entrance of a visitor

569

Ricardo simply greets Adalia (line 2), but Manuela must make it a point to find out who Adalia is and if she is male or female (lines 3-8) before she and Ricardo can integrate Adalia into their discussion.

Linguistic Affordances

Beginning the first week, the chatters provide linguistic affordances to each other, which have the potential to aid in the development of their L2. In just one example of this, Excerpt E5 shows Ricardo requesting the meaning of the expression "lift weights" (line 2). Bob explains the meaning in lines 4 and 5, and by line 6, Ricardo seems to understand and thanks Bob for the linguistic affordance.

Excerpt E5 (Week 1): Request for meaning (I)

In Excerpt E6, line2, Ricardo returns the favor, although he provides the meaning in Bob's L1 (English) instead of explaining it in Spanish. The provision of such linguistic affordances is an important function of the reciprocity of the chat community. That is, each member of the community can alternate between the language expert and language learner roles.

Excerpt E6 (Week 6): Request for meaning (II)

Throughout the 10 weeks, one of the chatters in particular (Bob) makes abundant use of affordances provided by the Spanish speakers. In Week 2, for example, Bob begins using the conditional tense verb debería 'ought,' a tense generally not yet employed by learners at this (intermediate) level, after seeing Ricardo use the verb form. In Excerpt E7, Bob repeats the conditional tense verb twice in response to Ricardo's question and then uses the form once again in a question directed back to Ricardo. Bob eliminates the preposition de from the verb form;

570

presumably because he does not think it necessary, and, in fact, it is not. He may have lifted the word debería from Ricardo as a lexical item rather than analyzing it as a conjugated verb form because he did not change the verb ending to the second person familiar form (deberías).

Excerpt E7 (Week 2): Bob's use of a linguistic affordance

Bob also employs circumlocution to help the Spanish speakers provide him the linguistic affordance he needs. In Excerpt E8, Bob negotiates the word anestesiológo 'anesthesiologist.' In lines 5 and 6, Josefina and Manuela indicate that they do not understand what Bob means in line 2 by "she makes sure you don't have pain." When Bob adds "person who gives medicine" to this in line 7, the Hispanophones figure out what he means and give him the word in Spanish (line 10).

Excerpt E8 (Week 2): Negotiation of a lexical item

571

This circumlocution, coupled with Bob's independent use of affordances as illustrated in Excerpt E6, demonstrates that Bob has appropriated chat room discourse as a mediator of his L2 development, a desirable accomplishment for chatters to attain.

One type of linguistic affordance widely believed to aid acquisition is negative feedback or error correction. BCC members were not specifically instructed on how and when to correct L2 errors because the instructors wanted to allow learners the autonomy to determine for themselves what the communicative norms would be in their chat rooms. As a result, a range of error correction patterns emerged from group to group. In Group E, there seemed to be an implicit norm that errors would not be corrected unless the speaker requested correction or if the error impeded comprehension. While the lack of negative feedback makes for fewer interruptions to the flow of conversation, there were a number of opportunities to provide such affordances which could have helped the learners in their L2 development. For example, Manuela says in Week 8, "lisa did u saw havanna nights?" The same week, Ricardo says, "Did every vary saw "Dance with me"?" Neither time was "saw" corrected, and, in fact, Ricardo misspells every "vary" (everybody) throughout the entire chat without anyone bringing it to his attention. This is one area, provision of negative feedback, in which instructors might want to intervene to guide chatters in providing and taking advantage of linguistic affordances in their chats.

Cooperation/Camaraderie Level

Social Affordances

As the chatters in Group E communicate with each other on a weekly basis, they develop enjoyable social relationships and exhibit community building behaviors, such as cooperating and collaborating with each other, using humor, and expressing enjoyment of their chat experience. These behaviors can be referred to as social affordances.

By Week 5, it is noticeable that the community has reached the cooperation/camaraderie level. The topic for this week consisted of a Puerto Rican legend of which the Anglophone students read the first half, and the Hispanophone students read the second half. The task was a jigsaw task which required the chatters to reconstruct the story by relating their half to those who had read the opposite half. Bob suggests a way to carry out the English part of the task: Yo puedo empezar y hacer la mitad y tu puedes continuarlo Lisa 'I can start and do half and you can continue it Lisa.' The group begins to proceed in that way, assisting each other in reconstructing the story. When Adalia interrupts the storytelling with a question, Ricardo suggests Porque no esperamos que nos cuenten toda la historia y luego hacemos las preguntas, les parece? 'Why don't we wait until they tell us the whole story and later we'll ask questions, OK?' Later, Manuela verifies whether it is an appropriate time to ask questions and then asks Bob and Lisa if they are ready to be informed about the other half of the story. They reconstruct the story completely and express their reactions to it. They say that they enjoyed the story.

572

Adalia, who is assigned to Group D but visiting Group E this day, says that she enjoyed the story and the chat session. As will be seen later, she never expresses enjoyment of the chats in her regular group. In Week 7 the chatters are given a similar task and proceed with the task in much the same manner.

Agreement

Participating, according to Lave and Wenger (1991), does not mean that participants always have to cooperate and agree with each other. In Group E, however, it appears that the members feel the need not only to cooperate but also to agree with each other most of the time. As reflected in all 10 chat logs, the participants are courteous to each other in every chat session, and conflicts never arise. In fact, cooperation is taken to a bit of an extreme in this group. For example, in Week 9, Lisa says at the end of the episode, "I agree with you all," even though she had (quite politely) expressed differences of opinion throughout the episode. The chatters use polite expressions such as "with all due respect …" before they disagree with each other. The politeness maintains a spirit of cooperation and camaraderie in the community; however, being willing to disagree might have pushed the community members' social and L2 linguistic skills a bit further (Schneider & von der Emde, 2005).

Leave Takings

After completing discussions of their assigned topic each week, members of Group E engaged in leave takings in which they talked about their plans for the week and wished each other well until their next chat. Leave takings averaged about 30 lines of text from week to week. In the final one, Week 10, leave takings consisted of nearly 50 lines. At the end of the last chat session, the participants express sadness about not being together anymore. In Excerpt E9 (lines 2-4 and 6), they mention how much they enjoyed the chat sessions and forming social relationships with each other. They exchange email addresses and mention the possibility of meeting in person (lines 1, 5, 7, 12).

Excerpt E9: Final leave taking (Week 10)

573

The chatters' expression of enjoyment and their intention to maintain contact after the semester is over are indicators that they truly formed a community and benefited socially, and perhaps linguistically, from the community experience.

Summary of Group E

In summary, Group E quickly established their community norms and followed the norms throughout, with very few exceptions. They were able to achieve a membership and a camaraderie level in their community. They cooperated with each other and appeared to have enjoyed the chat experience from beginning to end.

DATA ANALYSIS: GROUP D

Group D, with seven assigned members (two from NCSU and five from the UPR), manifested an attendance rate of 60% (28 absences among the seven members during the 10 weeks). Table 3 shows the participants present in each chat session.

574

Acquaintance Phase (Weeks 1-3)

As can be seen in the table, only two of the seven chatters assigned to this group (Beth and Will) attended more than half of the 10 scheduled chat sessions. In the first week, Ana, Beth, Iliana, Marta, and Will get acquainted with each other. These four do not appear simultaneously again until Week 5, which is the last time they all see each other. Iliana does not return until Week 4, and Marta returns Week 5. Unlike Group E in its first week, this group does not enter into discussion of their experiences with their L2, nor do they make mention of their future as a group. They do provide information about themselves such as their age and what they study, and they talk about their photos which are posted on the web site. Much of the discussion in Week 1 is focused on what they look like, whether Will (the only male in the room) has a girlfriend or whether any of the females have a boyfriend.

The bulk of the discussion in Week 2 also revolves around girlfriends and boyfriends. (This happened to be the week of St. Valentine's Day). Week 2's discussion is extremely short, less than half the length of Week 1's. Week 3 is also very short. Adalia appears for the first time in Week 3, and Beth and Will get acquainted with her. By Week 3, there is still one assigned member who has not attended a chat session, José. The group is not prepared to discuss the assigned topic in Week 3 (news articles from each country). Instead, they discuss the weather and Beth's computer problems. This group remains in the acquaintance level for a week longer than Group E, taking longer to establish themselves as a community. In fact, as will be seen later, the group at times reverts back to an acquaintance level, as they barely maintain community membership.

Community Membership Level

Members versus Nonmembers

By the fourth week, it is clear who is a member of this community and who is not. Ana, Beth, and Will can be considered full members because they all attended at least two of the three previous chats and established community procedures. Iliana is still a novice, having missed two weeks of chats and entered 26 minutes late in Week 4. In fact, Iliana never makes it to be a regular member of the group. After weeks 1 and 4, she only attends three more chat sessions (Weeks 5, 8, and 10). In Week 8, Iliana does not know the topic or that the language in which to start the chat is assigned, and, in Week 10, she does not know the topic. Although also assigned to this group, José never becomes more than a visitor. He enters the chat for the first time in Week 4, 8 minutes before the end of the session, and only attends two more subsequent chat sessions. José makes a positive contribution to the discussions in Week 6 and Week 7, but he never returns after Week 7. The other members, Adalia, Ana, and Marta, do not particularly disrupt the community,

575

but they only attend five of the ten chat sessions. Therefore their role can be classified as novice, almost members.

José's visitor-like behavior is called into question in Week 5. As shown in Excerpt D1, José is chastised by Iliana and Will for reporting late for the second consecutive time (lines 6-7). José's entrance causes an interruption, after which Mónica tries to bring the group back on topic (line 11). When Will chastises José for being late (line 7), a conflict is produced as Beth and Mónica both admonish Will for picking on José (lines 9 and 16). Will is apparently put in his place by line 17, after which their discussion continues.

Excerpt D1 (Week 5): Violation of group norms

The visitor and novice member status of the majority of the students assigned to Group D will have profound negative effects on the dynamics of the community, as will be shown in the sections that follow.

Roles and Regulatory Behaviors

Kim (2000) notes that, in most communities, at least one person takes on the leader role at any given time. Unlike Group E in which one of the members served as a leader most of the time, leadership in Group D was distributed. Table 4 indicates which members initiated the necessary group regulatory behaviors.

A number of community members take on the leader role at different times and some of the regulatory behaviors are negotiated among group members before action is taken. Beth often takes the leader role, initiating about one third of the topics and tasks, language changes, and endings of sessions (mostly in Weeks 1-4). In Week 1, for example, Beth uses capital letters (the chat equivalent of shouting) to get group members' attention: ESCUCHEN!!!! 'LISTEN!!!' In Week 3, she keeps the group on track: "heard anything in the news in NC lately?" and "so

576

what recent news has anyone heard lately?" It is noteworthy that most of Beth's leadership activity was carried out in the first four weeks of the chats and drops off after Week 4. The lack of a consistent leader, whether Beth or someone else, might very well be a function of the poor attendance habits of several group members, which surely contributed to the group's difficulty in attaining cohesiveness.

Linguistic Affordances

During their weekly conversations, the group members occasionally provided each other with linguistic affordances. An example of this can be seen in Excerpt D2, when José helps Adalia generate the word "opponent" in order to make her question in English understandable to Bob (lines 4-6).

Excerpt D2 (Week 6): Negotiation of a lexical item

Negative Feedback (Error Correction)

Chatters in the BCC are not typically instructed as to how and when to provide negative feedback, so it is not counterintuitive that each group might handle this

577

issue differently. It was seen that members of Group E did not correct each others' errors directly. Error correction became an issue in Group D, however, due to conflicting expectations that were not resolved. From the first chat session on, Ana and Marta apparently believed that their role in the chat was to correct L2 Spanish errors committed by the Anglophones. Excerpt D3 shows Ana's direct correction of Will's errors (line 2).

Excerpt D3 (Week 1): Direct Error Correction

After making a direct correction, Ana wants to verify whether others share her view on error correction (line 3). Will agrees that Ana should correct him and thanks her for it. Later in the chat session, Will indicates that he has difficulty with the gender of nouns in Spanish (see Table 5), which might be interpreted as a request to correct him when he makes such errors. Beth misspells a word and is corrected by Marta. Will makes another gender error and is corrected by Marta. When the conversation switches to English, the Anglophones do not overtly correct the Spanish speakers. Table 5 provides an overview of error correction in Group E in all ten chat sessions.

578

As can be seen in the table, correction activity increases and becomes interesting in Week 5. Ana corrects Will and Beth once, and Marta corrects Will once and

579

Beth three times. When the chat switches to English, Will and Beth apparently seek revenge. Although these two chatters had never directly corrected their Hispanophone counterparts before, they correct them six times in just a few minutes (shown in lines 2, 6, 7, 10, and 12 of Excerpt D4) and include sarcasm in their corrections (lines 3 and 14).

Excerpt D4 (Week 5): More direct error corrections

In Week 6, Adalia, who was not present in Week 5 and thus did not witness Will's and Beth's revenge, corrects Will three times, after one of which Will says, mi culpa 'my fault.' Week 7 begins in English. Excerpt D5 shows Beth's and Will's impatience and sarcasm directed toward their counterparts. In line 5, Beth pressures the Puerto Ricans to continue narrating their half of the story. In line 9, Will insinuates that Ana's statement in line 8 was copied from the story they were assigned to read. When challenged by Ana about what he meant by "copy/paste" in line 9, Will replied that it was a "typo" (line 16). Beth continues her impatience in line 15 when she tells Ana and Mónica "continue."

Excerpt D5 (Week 7): Sarcasm and impatience among the Anglophones

580

The sarcasm continues in the English part of Week 7's chat. In the Spanish part, the situation reaches a climax after Ana corrects Beth once, Marta corrects Will once, and Adalia corrects Will five times. As shown in Excerpt D6, Will erupts directly after the fifth correction (line 3).

Excerpt D6 (Week 7): Will's eruption over error correction

Adalia defends herself against Will's eruption in line 4, to which Will replies that he does not correct every mistake that Adalia makes in English (line 5). José and Ana interject (lines 7 and 8), asking Adalia not to mistreat Will. Adalia defends herself again and asks to be forgiven in line 9. This ends the conflictive part of their chat session.

581

There are a number of interesting facets to error correction in Group D. First, it did not become a problem until Week 5, by which time the community had already passed its acquaintance level (for the members who did pass this level). Marta and Ana did all of the correcting in Week 5, but Will and Beth vented their frustrations mainly on Iliana. Beth and Will allowed their frustration to mount, provoking their sarcasm in Week 5 and Will's attack on Adalia in Week 7 (who only started correcting him in Week 6).

In their work on group dynamics in the L2 instructional setting, Dornyei and Murphey (2003) claim that norms need to be discussed and willingly accepted by group members. Will apparently resented being directly corrected so many times. Instead of negotiating the norms on error correction in the earlier chat sessions, Will and Beth allowed the problem to progress. After Will's outburst, the issue does not reoccur in subsequent chats; Adalia only corrects Will one time and Alice (a visitor) one time in Week 8. In Weeks 9 and 10, no corrections are made.

Error correction as an affordance can be considered a form of scaffolding, or specialized assistance offered to a learner (Wood, Bruner, & Ross, 1976). Aljaafreh and Lantolf (1994) purport that scaffolding should always be contingent (being offered only when requested or needed), negotiated between the novice and the expert, and withdrawn as soon as the novice shows signs of self-control and ability to function independently. In Group E, the participants asked for specific help when they wanted it and were usually not offered it otherwise. Group D never established such norms, even though the conflicts produced in Week 5 and Week 7 presented them an opportunity to do so. Schneider and von der Emde (2005) determined that participants in telecollaborations would benefit from appropriating conflicts as learning opportunities rather than avoiding them. Group D might have benefited both socially and linguistically from dealing with their conflict and taking it as a learning experience.

Ironically, most corrections made in Group D were to errors which had minimal communicative value and did not appear to impede communication. The only errors likely to impede communication in Table 5 were the lexical items (2 out of 30 corrections). On the other hand, several more serious errors which would have impeded communication went uncorrected. Additionally, error corrections were often not taken up by the recipients. One example of this occurred in Week 1, when Will used the object pronoun mi 'me' instead of the subject pronoun yo 'I' and was corrected, but in Week 10 he made the same mistake. He also continues to use cuál and qué incorrectly and to mismatch the gender of definite articles and nouns. In this type of situation, it might be advisable for the teacher to intervene by instructing chatters on how to better provide and make use of linguistic affordances in chats.

Disruptive Behaviors

If chat participants do not establish and follow norms of attendance and participation, there might not be sufficient cohesiveness to consider the group a community. A number of norm violations affected the cohesiveness of Group D. Some

582

participants did not check the assigned topic ahead of time, an important part of preparing for the chat. In Weeks 1-4, at least one person did not know the assigned topic for the week. In Week 3, it appeared that none of the chatters had prepared for the chat by completing the assigned reading ahead of time. As a result, most of Week 3's chat was off topic. In Week 4, one of the chatters entered 38 minutes late and interrupted the discussion to find out what the topic was. Interestingly, most of the lack of preparation occurred during the acquaintance level of the community, in Weeks 1-4, although absenteeism and tardiness continued throughout the 10 weeks.

Absenteeism and tardiness also negatively affected the community. José only attended four chats and was significantly late for two chats of the four. Of the five English learners assigned to the group, none attended more than half of the chat sessions. Most of them did not follow the norms of the community enough to become more than novice members. Since the absences were so dispersed, many of the community members never got to know each other well, some having chatted together only one or two times, never to see each other again. Due to so many disruptions, Group D seems to have barely achieved a membership level, sometimes reverting to acquaintance level and only rarely manifesting aspects of a camaraderie level.

Camaraderie Level

Humor as Social Affordance

Because most of the participants in this group scarcely made it to the regular level of membership, the community did not enjoy much of a camaraderie level. This is not to say that there was no camaraderie in the community at all. Although the chatters did not get to know each other enough to establish close social bonds, there were a few social affordances provided in the chat sessions. Most of these were in the form of humor, expressed by Will (the jokester of the community), and mainly in the first few weeks of the chats. Excerpts D7 and D8 illustrate Will's use of humor.

Excerpt D7 (Week 1): Humor

It is obvious to the chatters what their names are, as the names appear before each of their messages. It is interesting that Will seems to attempt to establish camaraderie from the start.

Excerpt D8 (Week 2): More of Will's humor

583

Will was apparently trying to evoke a shock reaction from Ana and Beth by insinuating that he might be homosexual (line 1). Humor as a social affordance can serve to lighten the atmosphere and bring about cohesiveness in the chat community. In Group D, however, the use of humor occurred mainly during the first two weeks of the chats and then died off, presumably because tensions were mounting among group members as the weeks went by.

Leave Takings

Leave takings at the end of chat sessions in this group are often abrupt. In comparison to Group E's average of 30 lines of leave taking, Group D averaged 10 lines. In the final chat some of the participants said it was a pleasure to meet the others before saying good-bye, and some simply said good-bye. Mysteriously, Will dropped out of the conversation a few minutes before the end and therefore did not participate in the leave takings. Nobody mentioned the possibility of staying in touch after the chats, nor did they exchange email addresses and instant messenger names, as the members of Group E had. If Group D would have manifested more regular attendance and active participation, they might have reached a more cohesive membership level of community and perhaps enjoyed more of a camaraderie level.

DISCUSSION

The discourse data in this study corroborate Kling and Courtright's (2004, p. 91) claim that "Developing a group into a community is a major accomplishment that requires special processes and practices." Brown's (2001) taxonomy of levels of online community, coupled with Kim's (2000) participant roles, provided the lens through which to closely view the participants in Group E and Group D as they defined their communities during 10 weeks of bilingual chat. There were clear signs that all of the participants assigned to Group E progressed from novice to regular members, and a few of them took on the leader role. Group E moved through Brown's acquaintance level in just 2 weeks, establishing a pattern of communicative and social norms, then moved into membership level as they engaged in long, thoughtful discussions with each other. Their continuous cooperation with each other and expression of enjoyment indicates that they also experienced a camaraderie level as early as Week 4.

Group D's chat community evolved in a different manner. Mainly due to inconsistent attendance, most of the seven members assigned to this group failed to become regular members of the community, and one of them (José) was in effect only a visitor. The group did not establish and follow cohesive communicative

584

and social norms, which impeded them from reaching a camaraderie level of community and caused conflicts due to inconsistent expectations.

This telecollaboration was named Bilingual Chat Community to encourage cooperation and belonging among participants. However, possessing the community label does not ensure community-like behavior on the part of all members, as witnessed in this study. Brown's (2001) taxonomy of levels of community proved to be a robust conceptual framework in which to trace the evolution of both chat groups in this study, enabling the researcher to delimit levels of cohesiveness experienced in each group.

Of the 12 learners assigned to Group D and Group E, it can be said that seven of them met the goal of becoming fully integrated into the community of bilingual language practice. Four of the participants were novice members, and one was a visitor. With more motivation and commitment, these last five learners might have become regular members. Both Norton (2001) and Lave and Wenger (1991) cogently point out that in addition to participation another possible relation to the CoP is nonparticipation. "Our identities are constructed not only by what we are but also by what we are not" (Lave & Wenger, 1991, p. 168). Wenger (1998, p. 85) further points out that "CoPs are not intrinsically beneficial or harmful." Thus, it cannot be assumed that learners will achieve full participation in a CoP because of the mere existence of the CoP.

LIMITATIONS OF THE STUDY

As with any research project, this study has some limitations. First of these would be that, with its number of participants (12) and unique context, the findings should be cautiously generalized to other learning contexts. Also, data extracted from learner questionnaires would have complemented the subjective discourse analyses presented as the primary source of data.⁴ Such data would help the researcher to understand the participants' points of view regarding their relation to the CoP.

CONCLUSION

This study has aimed to contribute to the literature on community-based L2 learning by illustrating the participation and community metaphors of learning in action in a bilingual chat telecollaboration. The community metaphor enabled the researcher to capture the nature of the discursive construction of interpersonal relationships in the chat groups involved in the study. It was demonstrated that some learners can and do form a community and integrate themselves as full members of the community in a telecollaborative context. Because of the hard work and challenges involved in becoming a member of a bilingual chat community, not all groups and participants in a telecollaboration will contribute to a camaraderie level of community, and, in fact, their membership level can be delicate and unstable.

Although the Bilingual Chat Community was designed to belong to its participants, teacher intervention might have been advisable at times to help learners

585

make better use of the linguistic and social affordances available to them in the chats. The teacher's presence might not often be necessary in groups such as Group E, but, in groups like Group D, the teacher can serve as a bilingual resource person who helps to bring the community together into a more cohesive communicative environment. The teacher can also assist members of groups such as Group E to become more aware of providing and making use of linguistic affordances.

If the L2 teaching profession is to adopt the community as a metaphor for language learning, it behooves researchers and practitioners to delineate what full participation looks like in discursive practice, as well as the optimal conditions for promoting such participation. More research into the specific mechanisms of community building is necessary. SLA research needs to demonstrate in finer detail what counts as participation, how learners become participants, and the causes and effects of nonparticipation in the community structure. It is hoped that this study will stimulate further research into community-based L2 learning.

NOTES

¹ The use of the definite article does not imply that this is the only bilingual chat community in existence, but rather that it designates the proper name of this particular community.

² All names used in this study are pseudonyms.

³ In some excerpts, lines of unrelated text were removed due to space limitations.

⁴ Electronic questionnaires were administered to the participants in this study. However, due to technical issues, the data set was not complete enough to render it a useful research instrument.

REFERENCES

Aljaafreh, A., & Lantolf, P. (1994). Negative feedback as regulation and second language learning in the zone of proximal development. Modern Language Journal, 78 (4), 465-483.

Arnold, N., Ducate, L., Lomicka, L., & Lord, G. (2005). Using computer-mediated communication to establish social and supportive environments in teacher education. CALICO Journal, 22 (3), 537-566.

Barab, S., Kling, R., & Gray, J. (Eds.). (2004). Designing virtual communities in the service of learning. New York: Cambridge University Press.

Belz, J. A. (2001). Institutional and individual dimensions of transatlantic group work network-based language teaching. ReCALL, 13 (2), 129-147.

Belz, J. A. (2002a). Identity, deficiency, and first language use in foreign language education. In C. Blyth (Ed.), The sociolinguistics of foreign-language classrooms (pp. 209-248). Boston: Heinle.

586

Belz, J. A. (2002b). Social dimensions of telecollaborative foreign language study. Language Learning & Technology, 6 (1), 60-81. Retrieved March 4, 2007, from http://llt.msu.edu/vol6num1/belz

Belz, J. A., & Thorne, S. (Eds.). (2005). Internet-mediated intercultural foreign language education. Boston: Heinle.

Block, D. (2003). The social turn in second language acquisition. Washington, DC: Georgetown University Press.

Brown, R. E. (2001). The process of community building in distance learning classes. Journal of Asynchronous Learning Networks, 5 (2), 18-35.

Darhower, M. (2006). Where's the community?: Bilingual Internet chat and the "fifth C" of the National Standards. Hispania, 89 (1), 84-98.

Donato, R. (1994). Collective scaffolding in second language learning. In J. Lantolf & G. Appel (Eds.), Vygotskian approaches to second language research (pp. 33-56). Norwood, NJ: Ablex.

Dornyei, Z., & Murphey, T. (2003). Group dynamics in the language classroom. New York: Cambridge University Press.

Firth, A., & Wagner, J. (1997). On discourse, communication and (some) fundamental concepts in SLA research. Modern Language Journal, 81 (3), 285-300.

Guerrero, M. C., & Villamil, O. (2000). Activating the ZPD: Mutual scaffolding in L2 revision. Modern Language Journal, 84 (1), 51-68.

Hall, J. K. (1993). The role of oral practices in the accomplishment of our everyday lives: The sociocultural dimension of interaction with implications for the learning of another language. Applied Linguistics, 14 (2), 145-166.

Kim, A. J. (2000). Community building on the web. Berkeley, CA: Peachpit Press.

Kling, R., & Courtright, C. (2004). Group behavior and learning in electronic forums. In S. Barab, R. Kling, & J. Gray (Eds.), Designing for virtual communities in the service of learning. New York: Cambridge University Press.

Kramsch, C., & Thorne, S. (2002). Foreign language learning as global communicative practice. In D. Block & D. Cameron (Eds.), Globalization and language teaching (pp. 83-100). London and New York: Routledge.

Krashen, S. (1981). Principles and practice in second language acquisition. New York: Pergamon Press.

Lantolf, J. (1996). SLA theory building: "Letting all the flowers bloom!" Language Learning, 46 (4), 713-749.

Lantolf, J., & Appel, G. (Eds.). (1994). Vygotskian approaches to second language research. Norwood, NJ: Ablex.

Lave, J., & Wenger, E. (1991). Situated learning: Legitimate peripheral participation. New York: Cambridge University Press.

Miles, M. B., & Huberman, A. M. (1994). An expanded sourcebook: Qualitative data analysis (2nd ed.). Thousand Oaks, CA: Sage Publications.

Norton, B. (2001). Non-participation, imagined communities and the language classroom. In M. Breen (Ed.), Learner contributions to language learning: New directions in research (pp. 159-171). New York: Pearson Education.

587

Renninger, K. A., & Shumar, W. (Eds.). (2002). Building virtual communities. New York: Cambridge University Press.

Rheingold, H. (1993). The virtual community: Homesteading on the electronic frontier. Reading, MA: Addison-Wesley.

Riel, M., & Polin, L. (1994). Online learning communities: Common ground and critical differences in designing technical environments. In S. Barab, R. Kling, & J. Gray (Eds.), Designing for virtual communities in the service of learning (pp. 16-50). New York: Cambridge University Press.

Rogoff, B. (1994). Developing understanding of the idea of communities of learners. Mind, Culture, and Activity, 1 (4), 209-229.

Schneider, J., & von der Emde, S. (2005). Conflicts in cyberspace: From communication breakdown to intercultural dialogue in online collaborations. In J. A. Belz & S. Thorne (Eds.), Internet-mediated intercultural foreign language education (pp. 178-206). Boston: Heinle.

Sfard, A. (1998). On two metaphors for learning and the dangers of choosing just one. Educational Researcher, 27 (2), 4-13.

Shotter, J., & Newson, J. (1982). An ecological approach to cognitive development: Implicate orders, joint action and intentionality. In G. Butterwood & P. Light (Eds.), Social cognition: Studies of the development of understanding. Chicago: University of Chicago Press.

Van Lier, L. (2000). From input to affordance: Social-interactive learning from an ecological perspective. In J. Lantolf (Ed.), Sociocultural theory and second language learning (pp. 245-259). New York: Oxford University Press.

Wenger, E. (1998). Communities of practice: Learning, meaning, and identity. New York: Cambridge University Press.

Wood, D., Bruner, J. S., & Ross, G. (1976). The role of tutoring in problem solving. Journal of Child Psychology and Psychiatry, 17 (1), 89-100.

588

Appendix A

AUTHOR'S ADDRESS

Dr. Mark Anthony Darhower

Department of Foreign Languages and Literatures

Campus Box 8106, North Carolina State University

Raleigh, NC 27695

Phone: 919 513 4807

Fax: 919 515 6981

Email: mark_darhower@ncsu.edu

AUTHOR'S BIODATA

Mark Anthony Darhower is Assistant Professor of Spanish Linguistics at the North Carolina State University in Raleigh, NC. He obtained his Ph.D. in Spanish Applied Linguistics at the University of Pittsburgh and previously taught English as a second language and linguistics at the University of Puerto Rico. Currently he teaches undergraduate and graduate courses in Spanish language, linguistics, and second language acquisition. His research interests include computer-mediated communication, Vygotsky-inspired sociocultural theory, and other social discursive approaches to second language acquisition. His publications appear in the CALICO Journal and Hispania.