Columns Reviews Volume 12, Number 1 February 2008
Transcription
Columns Reviews Volume 12, Number 1 February 2008
Columns From the Editors by Dorothy Chun & Irene Thompson pp. 1-2 On the Net You've Got Some GALL: Google-Assisted Language Learning by George Chinnery pp. 3-11 Emerging Technologies Of Elastic Clouds and Treebanks: New Opprtunities for Content-Based and DataDriven Language Learning by Robert Godwin-Jones pp. 12-18 Announcements News from Sponsoring Organizations pp. 19-22 Reviews Edited by Sigrun Biesenbach-Lucas Review of Five English Learners' Dictionaries on CD-ROM Reviewed by Alfonso Rizo-Rodríguez pp. 23-42 Volume 12, Number 1 February 2008 Peer Feedback on Language Form in Telecollaboration Paige Ware, Southern Methodist University Robert O'Dowd, Universidad de Léon pp. 43-63 The Role of Offline Metalanguage Talk in Asynchronous Computer-Mediated Communication Keiko Kitade Ritsumeikan University, Kyoto pp. 64-84 Methodological Hurdles in Capturing CMC Data: The Case of the Missing Self-Repair Bryan Smith Arizona State University pp. 85-103 Commentary: Can Free Reading Take you All the Way? A Response to Cobb (2007) Jeff McQuillan, Center for Educational Development Stephen D. Krashen, University of Southern California pp. 104-108 Commentary: Response to McQuillan and Krashen (2008) Tom Cobb Université du Québec à Montreal pp. 109-114 Contact: Editors or Editorial Assistant Copyright © 2007 Language Learning & Technology, ISSN 1094-3501. Articles are copyrighted by their respective authors. About Language Learning & Technology Language Learning & Technology is a refereed journal which began publication in July 1997. The journal seeks to disseminate research to foreign and second language educators in the US and around the world on issues related to technology and language education. • Language Learning & Technology is sponsored and funded by the University of Hawai'i National Foreign Language Resource Center (NFLRC) and the Michigan State University Center for Language Education And Research (CLEAR), and is co-sponsored by the Center for Applied Linguistics (CAL). • Language Learning & Technology is a fully refereed journal with an editorial board of scholars in the fields of second language acquisition and computer-assisted language learning. The focus of the publication is not technology per se, but rather issues related to language learning and language teaching, and how they are affected or enhanced by the use of technologies. • Language Learning & Technology is published exclusively on the World Wide Web. In this way, the journal seeks to (a) reach a broad audience in a timely manner, (b) provide a multimedia format which can more fully illustrate the technologies under discussion, and (c) provide hypermedia links to related background information. • Beginning with Volume 7, Number 1, Language Learning & Technology is indexed in the exclusive Institute for Scientific Information's (ISI) Social Sciences Citation Index (SSCI), ISI Alerting Services, Social Scisearch, and Current Contents/Social and Behavioral Sciences. • Language Learning & Technology is currently published three times per year (January, May, September). Copyright © 2007 Language Learning & Technology, ISSN 1094-3501. Articles are copyrighted by their respective authors. Sponsors, Board, and Editorial Staff Volume 12, Number 1 Sponsors University of Hawai`i National Foreign Language Resource Center (NFLRC) Michigan State University Center for Language Education and Research (CLEAR) Co-Sponsor Center for Applied Linguistics (CAL) Advisory and Editorial Boards Advisory Board Susan Gass Richard Schmidt Michigan State University University of Hawai`i [email protected] [email protected] Sigrun Biesenbach-Lucas Thierry Chanier Graham Crookes Robert Godwin-Jones Lucinda Hart-González Philip Hubbard Michelle Knobel Marcus Kötter Marie-Noelle Lamy Lara Lomicka Allan Luke Mary Ann Lyman-Hager Alison Mackey Carla Meskill Denise Murray Noriko Nagata John Norris Georgetown University Université de Franche-Comte University of Hawai`i Virginia Commonwealth Univ. Second Language Tesing, Inc. Stanford University Montclair State University University of Münster The Open University University of South Carolina University of Queensland San Diego State University Georgetown University SUNY-Albany San Jose State University University of San Francisco University of Hawai`i [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] Lourdes Ortega Jill Pellettieri Joy Kreeft Peyton University of Hawai`i Santa Clara University Center for Applied Linguistics, Washington, DC University of Cal., Berkeley Montclair State University Monterey Institute of International Studies Univ. of California, Irvine [email protected] [email protected] [email protected] Editorial Board Maggie Sokolik Susana Sotillo Leo van Lier Mark Warschauer [email protected] [email protected] [email protected] [email protected] Editorial Staff Editors Dorothy Chun Irene Thompson Associate Editors Richard Kern Editorial Assistant Web Production Editor Book & Multimedia Review Editor On the Net Editors Emerging Technologies Editor Copy Editors Batia Laufer Hunter Hatfield Carol Wilson-Duffy Sigrun BiesenbachLucas Jean W. LeLoup Robert Ponterio Robert Godwin-Jones Stephanie Alexis Balunda Matthew Buscemi Elizabeth Pfaff Lavolette Suann Robinson University of CA, Santa Barbara The George Washington University (Emerita) University of CA, Berkeley University of Haifa University of Hawai`i Michigan State University Georgetown University [email protected] SUNY at Cortland SUNY at Cortland Virginia Commonwealth University Indiana University [email protected] [email protected] [email protected] University of Hawai`i University of Hawai`i [email protected] [email protected] University of Hawai`i [email protected] thompson@roadstarinternet. net [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] Copyright © 2007 Language Learning & Technology, ISSN 1094-3501. The contents of this publication were developed under a grant from the Department of Education (CFDA 84.229, P229A60012-96 and P229A6007). However, the contents do not necessarily represent the policy of the Department of Education, and one should not assume endorsement by the Federal Government. Information for Contributors Language Learning & Technology is seeking submissions of previously unpublished manuscripts on any topic related to the area of language learning and technology. Articles should be written so that they are accessible to a broad audience of language educators, including those individuals who may not be familiar with the particular subject matter addressed in the article. General guidelines are available for reporting on both quantitative and qualitative research. Manuscripts are being solicited in the following categories: Articles | Commentaries | Reviews Articles Articles should report on original research or present an original framework that links previous research, educational theory, and language teaching practices that utilize technology. Articles containing only descriptions of software, classroom procedures, or those presenting results of attitude surveys without discussing data on actual language learning outcomes will not be considered. Full-length articles should be no more than 8,500 words in length, including references, and should include an abstract of no more than 200 words. Appendices should be limited to no more than 1,500 words. We encourage articles that take advantage of the electronic format by including hypermedia links to multimedia material both within and outside the article. All article manuscripts submitted to Language Learning & Technology go through a two-step review process. Step 1: Internal Review. The editors of the journal first review each manuscript to see if it meets the basic requirements for articles published in the journal (i.e., that it reports on original research or presents an original framework linking previous research, educational theory, and teaching practices), and that it is of sufficient quality to merit external review. Manuscripts which do not meet these requirements or are principally descriptions of classroom practices or software are not sent out for further review, and authors of these manuscripts are encouraged to submit their work elsewhere. This internal review takes about 1-2 weeks. Following the internal review, authors are notified by e-mail as to whether their manuscript has been sent out for external review or, if not, why. Step 2: External Review. Submissions which meet the basic requirements are then sent out for blind peer review from 2-3 experts in the field, either from the journal's editorial board or from our larger list of reviewers. This second review process takes 2-3 months. Following the external review, the authors are sent copies of the external reviewers' comments and are notified as to the decision (accept as is, accept pending changes, revise and resubmit, or reject. Titles should be concise (preferably fewer than 10 words) and adequately descriptive of the content of the article. Some good examples are • • • Social Dimensions of Telecollaborative Foreign Language Study "Reflective Conversation" in the Virtual Language Classroom Teaching German Modal Particles: A Corpus-Based Approach Copyright © 2007 Language Learning & Technology, ISSN 1094-3501. Articles are copyrighted by their respective authors. Commentaries Commentaries are short articles, usually no more than 2,000 words, discussing material previously published in Language Learning & Technology or otherwise offering interesting opinions on theoretical and research issues related to language learning and technology. Commentaries which comment on previous articles should do so in a constructive fashion. Hypermedia links to additional information may be included. Commentaries go through the same two-step review process as for articles described above. Submission Guidelines for Articles and Commentaries Please list the names, institutions, e-mail addresses, and if applicable, World Wide Web addresses (URLs), of all authors. Also include a brief biographical statement (maximum 50 words, in sentence format) for each author. (This information will be temporarily removed when the articles are distributed for blind review.) Articles and commentaries can be transmitted in either of the following ways: 1. By electronic mail, send the main document and any accompanying files (images, etc.) to [email protected] 2. By mail, send the material on a Macintosh or IBM diskette to LLT NFLRC University of Hawai'i at Manoa 1859 East-West Road, #106 Honolulu, HI 96822 USA Please check the General Policies below for additional guidelines. Reviews Language Learning & Technology publishes reviews of professional books, classroom texts, and technological resources related to the use of technology in language learning, teaching, and testing. Reviews should normally include references to published theory and research in SLA, CALL, pedagogy, or other relevant disciplines. Reviewers are encouraged to incorporate images (e.g., screen shots or book covers) and hypermedia links that provide additional information, as well as specific ideas for classroom or research-oriented implementations. Reviews of individual books or software are generally 1,200-1,600 words long, while comparative reviews of multiple products may be 2,000 words or longer. They can be submitted in ASCII, Rich Text Format, Word, or HTML. Accompanying images should be sent separately as jpeg or gif files. Reviews should include the name, institutional affiliation, e-mail address, URL (if applicable), and a short biographical statement (maximum 50 words) of the reviewer(s). In addition, the following information should be included in a table at the beginning of the review: Books Author(s) Title Series (if applicable) Publisher City and country Software Title (including previous titles, if applicable) and version number Platform Minimum hardware requirements Publisher (with contact information) Support offered Copyright © 2007 Language Learning & Technology, ISSN 1094-3501. Articles are copyrighted by their respective authors. Year of publication Number of pages Price ISBN Target language Target audience (type of user, level, etc.) Price ISBN (if applicable) LLT does not accept unsolicited reviews. Contact Sigrun Biesenbach-Lucas if you are interested in having material reviewed or in serving as a reviewer ([email protected]). Sigrun Biesenbach-Lucas 21333 Comus Court Ashburn, VA 20147 General Policies The following policies apply to all articles, reviews, and commentaries: All submissions should conform to the requirements of the Publication Manual of the American Psychological Association (4th edition). Authors are responsible for the accuracy of references and citations, which must be in APA format. Manuscripts that have already been published elsewhere or are being considered for publication elsewhere are not eligible to be considered for publication in Language Learning & Technology. It is the responsibility of the author to inform the editor of any similar work that is already published or under consideration for publication elsewhere. Authors of accepted manuscripts will assign to Language Learning & Technology the permanent right to electronically distribute their article, but authors will retain copyright and, after the article has appeared in Language Learning & Technology, authors may republish their text (in print and/or electronic form) as long as they clearly acknowledge Language Learning & Technology as the original publisher. The editors of Language Learning & Technology reserve the right to make editorial changes in any manuscript accepted for publication for the sake of style or clarity. Authors will be consulted only if the changes are major. Authors of published articles, commentaries, and reviews will receive 10 free hard-copy offprints of their articles upon publication. Articles and reviews may be submitted in the following formats: HTML files Microsoft Word documents RTF documents ASCII text If a different format is required in order to better handle foreign language fonts, please consult with the editors. Copyright © 2007 Language Learning & Technology, ISSN 1094-3501. Articles are copyrighted by their respective authors. Language Learning & Technology http://llt.msu.edu/vol12num1/editors/ February 2008, Volume 12, Number 1 p. 1-2 FROM THE EDITORS It is our pleasure to introduce Volume 12, Number 1 of Language Learning & Technology, a regular issue of our journal. We want to take this opportunity to wish you a happy and productive 2008, a year proclaimed by the United Nations General Assembly as the International Year of Languages. We are proud to be a part of this international effort to promote the study of languages and cultures worldwide. We want to thank our contributors, reviewers, and readers for making 2007 a very successful year for our journal. The number of subscribers grew from 8,500 in 2006 to 10,600 in 2007. We received a record number of 144 submissions from 31 countries in 2007, up from 105 in the previous year. We are looking forward to 2008 being an even better year. This issue features three articles and two commentaries in addition to our regular columns. The three articles coincidentally all deal with various issues involved in computer-mediated communication. "Peer feedback on language form in telecollaboration" by Paige Ware and Robert O’Dowd explores corrective peer feedback on form in asynchronous discussions. Their findings indicate that such feedback occurred only when students were explicitly required to provide it. Pedagogical implications include the need to situate peer feedback on form within current models of telecollaboration and to assist students in finding feedback strategies that do not require a sophisticated understanding of L1 or L2 grammar. "The role of metalanguage talk in asynchronous computer-mediated communication" by Keiko Kitade examines the benefits of offline dialogue in an asynchronous computermediated communication (ACMC) activity. The study suggests that offline dialogue may compensate for lack of instant tailored feedback in ACMC. The author recommends further investigation of the potential of offline interactions for creating a collaborative context, not only among online interlocutors but also among offline peers. "Methodological hurdles in capturing CMC data: The case of the missing self-repair" by Bryan Smith studies the use of self-repair among learners of German in a task-based CMC environment in order to (1) establish how potential interpretations of CMC data may depend on the method of data collection and evaluation and (2) explicitly examine the nature of CMC self-repair in the task-based foreign language CALL classroom. His results show that the interpretation of the chat interaction is a function of the data collection and evaluation methods employed. The findings also suggest a possible difference in the nature of self-repair across face-to-face and SCMC environments. In view of the results, this paper calls for CALL researchers to abandon the reliance on printed chat log files when attempting to interpret SCMC interactional data. The commentary "Can free reading take you all the way: A response to Cobb (2007)" by Jeff McQuillan and Stephen Krashen argues that in "Computing the Demands of Vocabulary Acquisition from Reading" (Language Learning & Technology, October, 2007), Cobb underestimated the amount of reading that even a very modest reading habit would afford L2 readers, and therefore underestimated the impact of free reading on L2 vocabulary development. In addition, the authors point out that Cobb’s own data show that free reading is a very powerful tool in L2 vocabulary acquisition. Copyright © 2008, ISSN 1094-3501 1 From the Editors In his commentary "Response to McQuillan and Krashen (2008)", Cobb questions the adequacy of free reading for vocabulary development in the typical time frame of instructed L2 acquisition. He suggests that the development of an adequate L2 lexicon results from well-designed L2 instruction that includes, but is not limited to, reading. The "On the Net" column "You’ve got some GALL: Google-assisted language learning" by guest contributor George Chinnery proposes a number of interesting ways in which the power of Google can be harnessed for pedagogical purposes. The "Emerging Technologies" column "Of elastic clouds and treebanks: New opportunities for content-based and data-driven language learning" by Robert GodwinJones describes a plethora of new technical developments that make it possible to use large data sets for language learning. Our Reviews column edited by Sigrun Biesenbach-Lucas contains a detailed review of five English learners’ dictionaries on CD ROM by Alfonso Rizo-Rodríguez. Please take a look at the updated list of PhD dissertations dealing with language learning and technology. We would like to thank Dr. Evelyn Reder Wade of UC Santa Barbara who compiled the original list and provided the current updated one. Please take a few minutes to complete the LLT subscription form if you have not already done so. Sincerely, Dorothy Chun and Irene Thompson, Editors Language Learning & Technology 2 Language Learning & Technology http://llt.msu.edu/vol12num1/net/ February 2008, Volume 12, Number 1 pp. 3-11 ON THE NET You’ve Got some GALL: Google-Assisted Language Learning George M. Chinnery University of Maryland – Baltimore County INTRODUCTION "Just google it!" "Have you googled it yet?" "I'll google it later." Commands, inquiries, and intentions of this sort have become so commonplace in class discussions, during meetings, over dinner, and on the phone as to approach cliché. One article making the rounds on the AP wire even investigated "googling your date" (Irvine, 2007). The impact of the internet on the English, and global, lexicon is nothing new. It has become habitual to send e-mails or text messages in lieu of using snail-mail or calling on the phone. Many other forms of computer-mediated communication have similarly found themselves both publicly and officially recognized. In 2004, blog was named Merriam Webster's word of the year ("Blog Picked," 2004; Merriam-Webster Online, 2005). Likewise podcast, which the New Oxford American Dictionary named as its 2005 word of the year ("Wordsmiths Hail Podcast," 2005). Google entered Merriam-Webster the next year, though only as runner-up for word of the year, losing out to truthiness (Ahrens, 2006; Merriam-Webster, 2006). What is unique about Google’s cross-over is not only the fact that its brand name has trumped its function, in the same way many of us blow our noses in kleenex, toss frisbees, and dress our wounds with band-aids, but that it is this function with which it is synonymous (i.e., it’s a verb). As such, no longer do we simply ‘search’ for something online. Now we google it. A BRIEF HISTORY OF GOOGLE In 1998, co-founders Larry Page and Sergey Brin launched their newly renamed search engine, Google. Acknowledging the mathematical origins of its moniker (a ‘googol’ is 1 followed by 100 zeros), a statement on its website indicates that "Google's play on the term reflects the company's mission to organize the immense amount of information available on the web." In the eyes of the public, this mission was seemingly accomplished very quickly, such that in 2003, a New York Times columnist somewhat sarcastically asked "Is Google God?" (Friedman, 2003). In 2004, Wired magazine celebrated Googlemania, chronicling the site’s rise to the summit of search and its impending death-match with Microsoft (Malone, 2004). In 2005, the fictional retrospective documentary, EPIC 2015, mockingly documented the Evolving Personalized Information Construct, portending a mammoth cyber-merger with Amazon which would deliver its GoogleZon progeny to the world. In 2006, ‘Google’ itself was the most searched term on AOL’s search engine during a three month period, i.e., Google itself was apparently being googled (Nakashima, 2006a). Come 2007, half of all US searches were being conducted on Google (comScore, 2007a). By March of the same year, Google was reportedly the "world’s most-visited site" (Kopytoff, 2007). And at least two major universities were offering courses in ‘Google’ (McCloskey, 2007). Numerous books have been published on the subject of Google, covering it both as a successful business model and a powerful internet tool. Following the theme of the latter, Google offers a range of practical applications for language instructors and learners alike. Copyright © 2008, ISSN 1094-3501 3 George M. Chinnery Google-Assisted Language Learning A GOOGOL OF PEDAGOGICAL USES FOR GOOGLE Since its inception, language instructors have recognized the informational potential offered by the internet. Corpus linguistics, for instance, went online with web-based linguistic corpora and KWIC (key word in context) concordancers (e.g., MiCASE). Google has itself even been proposed as a ‘quick ‘n dirty’ concordancer (Robb, 2003; Rundell, 2000). But it also has the capability to do much more than simply facilitate basic Boolean-type searches. Google as Informative Tool At a basic level, Google, by default, checks for and corrects spelling errors, such that a query for ‘cofee’ proffers ‘Did you mean: coffee.’ Beyond superficial form, however, learners can discover meaning by appending a dictionary command to the start of a term (e.g., ‘define:coffee’). Google can also focus on usage. The 'define:coffee' command offers several common collocations (e.g., coffee break, Turkish coffee) at the top of the page. Typing into Google Suggest will preview similar collocations. And in using a wildcard command such as ‘I drink * coffee’, the asterisk acts as a placeholder for a gap-fill, and results in a range of potential responses. This is also useful for phrasal verbs, such that by typing ‘come * with’, learners discover ‘come up with’, ‘come away with’ and more. Another way to maintain context is to search authentic texts in Google Books, where a search for ‘coffee’ introduces learners to rich prose describing ‘roasted coffee’ and ‘steaming coffee’. Learners curious about the different synonyms for coffee can compare the regional popularity of their usage at Google Trends. A search for ‘cup of coffee, cup of joe, cup of mud, cup of java’ will not only inform the inquisitive learners of the global popularity of ‘cup of coffee’, but also that ‘cup of joe’ is not uncommon in the United States, particularly in New York. Conducting such comparative searches on Google Fight, a Google hack (an unauthorized modification by a third party), provides a more animated and entertaining display of the results. To discover synonyms in the first place, learners can prefix a given term with a tilde (e.g., ~coffee), which searches not only for said term, but also popular related terms. Vocabulary development can be encouraged more interactively through use of the Google Image Labeler, a real-time two-player game where the goal is to reciprocally label a series of digital photos. Google Sets provides another option useful even to beginning learners: an opportunity for listing and brainstorming. A search for ‘black’, for instance, displays an extensive list of other colors. Google also offers several tools for beginning learners’ numeracy work. Typing ‘3 x 2’ into Google turns it into an instant calculator. Queries patterned after official exchange lingo (e.g., ‘3.99 USD in RUB’) offer updated currency conversions. Adapting ‘weather Seattle’ displays local forecasts both textually and graphically. And simply typing an accurate address into Google directs learners to Google Maps. Another option well-suited to beginning, as well as more advanced, language learners is the Google Language Tools page, which provides interfacing in over 100 legitimate and faux (e.g., Elmer Fudd) languages. Interface and search language ‘preferences’ can also be set from Google’s home page, such that all results are restricted to the language of choice. Also available for many of these are search and translation services. Entire websites can be translated in mere seconds. The creative 1888usa Google hack combines Google's translations with AT&T’s speech synthesis (a.k.a., text-to-speech) demo. More advanced learners can be encouraged to manipulate and interact with their target language by conducting creative webquests on Google. For example, learners can type in a few random ingredients (e.g., ‘black beans brown rice tomatoes cilantro’) to see what recipes Google can concoct. The Cookin' with Google hack performs similar searches, exclusively on several popular recipe sites. Google can also be used to guide learners in more traditional webquests. Returning to the coffee illustration, Google would enable learners conducting research on the history of coffee to search for information on preselected sites (e.g., ‘coffee site:www.ethiopianembassy.org’). By clicking the ‘Cached’ link under any of the search results, the search terms are brightly highlighted. And if said learners are in need of more Language Learning & Technology 4 George M. Chinnery Google-Assisted Language Learning information, use of the link command (e.g., link:www.ethiopianembassy.org) will provide referral to sites linked to the given source. Instructors desiring greater control over learners’ search activities can tailor their own search engines using Google Coop. For example, this Ethiopian Coffee search engine will only search pre-selected sites identified on that page. Instructors and learners unable or uninterested in recalling the aforementioned commands can access Google’s Advanced Search page, which provides a more user-friendly interface for many of these. Though Google’s range of search tools is in and of itself impressive, perhaps what makes it all the more powerful is its recognition of the internet’s potential. e-Language Learning describes the use of modern web-based tools for learning opportunities which are informative, productive, collaborative, communicative, and aggregative. The preceding examples illustrate how Google successfully provides myriad opportunities for the first of these, in essence employing its most traditional use as an information provider. Google also offers a comprehensive suite of free programs (to anyone who registers for an account) which help facilitate the remainder. Google as a Productive Tool An article in the New York Times once declared that "all the Internet’s a stage" (Stross, 2006). Thus, whereas the heavily informative quality of Google can be aligned with language input, its productive tools foster opportunities for output. This reality is reflected in blogging sites like Google’s Blogger, which allows learners to instantly author, publish, and syndicate their own textual, audiovisual, and generally multimedia productions for a global audience. Google Docs offers collaborative web-based word processing. Essentially, it is like a free web-based version of Microsoft Word. One of its key strengths is ability to be shared, like a wiki (incidentally, Google has acquired Jotspot, a popular wiki), a feature which allows for a plethora of creative applications. An instructor might post a text, intentionally replete with errors, for learners to correct. Likewise, learners can easily peer-edit, as this program leaves an editing trail. Another option is chain storytelling, where an instructor begins a story which each learner contributes to in turn. Such a tool is useful in group projects in general. Another feature of Google Docs is web-based spreadsheets, similar to Microsoft Excel, that instructors can use for attendance-keeping and grading. Google as a Collaborative Tool Vygotskyan constructivism (Vygotsky, 1978) posits that knowledge construction and meaning-making are best facilitated via scaffolded collaboration. The aforementioned tools clearly provide for such collaborative opportunities. This potential is further enhanced through use of Google Calendar, which can be used for scheduling and sending out reminders, and Google Groups, which can also be used to send out announcements, as well as to facilitate asynchronous class discussions. Google as a Communicative Tool Google also offers its own versions of some of the more standard communicative tools, which provide opportunities for interaction/negotiation in the target language. Gmail is Google’s email program, and Google Talk is its instant messenger-cum-internet telephony service, that allows users to save, print and email text chats. The latter can be used as the medium of communication between pairs of learners engaging in classic cooperative activities such as jigsaw tasks. The benefits of doing so via chat have been summarized by Swaffar (1998), who indicates that chats "seem to help all individuals in language classes engage more frequently, with greater confidence, and with greater enthusiasm in the communicative process than is characteristic for similar students in oral classrooms" (p. 1). Another communicative tool with which to focus on form is GOOG-411, an automated telephone directory which integrates speech recognition and text messaging. Used effectively, it can be used to aid in the development of learners’ pronunciation. Language Learning & Technology 5 George M. Chinnery Google-Assisted Language Learning Google as an Aggregative Tool In addition to providing learning opportunities that are informative, productive, collaborative and communicative, Google offers several tools that recognize linguistic, visual, audio, gestural and spatial literacies in aggregate (New London Group, 1996). On iGoogle, for example, learners can create ‘start pages’ that collect many of the aforementioned Google tools, as well as many others. Google Reader is a web feed aggregator which can be used by learners and instructors to collect updated news feeds, blogs, podcasts and vidcasts together into a single interface. Google Gears allows them to view this content offline, in the same way that podcasting allows audiovisual content to be downloaded from the internet for later use. Google Page Creator is a deceptively simple webpage creation tool. The ‘My Maps’ feature of Google Maps is a mash-up tool which allows learners to tailor-make maps, which they can embed with descriptive text, and digital images and drawings. Google Earth, which has the ability to take learners home with satellite precision, can be similarly utilized. Video mash-ups can be created using the Googleowned YouTube Remixer. And another feature of Google Docs is a presentation tool tantamount to Microsoft PowerPoint. All of these can be used to promote digital storytelling by language learners. On an iGoogle start-page, for example, learners can present their hometowns, complete with digital images, weather forecasts, current events, and more. Using the Google Maps mash-up tool, immigrants and sojourners alike can imaginatively narrate their travels. CRITICAL THINKING Despite the benefits its tools offer to those involved in language instruction and learning, as well as the population-at-large, Google is not without its critics. Publishers are worried about the repercussions of the Google Books Library Project, which aspires to create a digital archive of the world’s books, public domain and otherwise (Ekman, 2006; Sipress, 2007). Spurred by similar copyright concerns of newspaper editors, Belgian courts ordered Google to stop posting headlines from its national papers on Google News (White, 2007). Google has also been accused of selective censorship. For a time, Google blocked web content critical of the Church of Scientology due to pressure from the Church (Gallagher, 2002). It removed YouTube videos which the Thai government alleged were insulting to its monarchy (Vandenberghe, 2007). And it voluntarily agreed to censor itself in China (Crampton, 2006). More globally, it has been argued that sites ranked highest by Google tend to remain the most popular, thereby restricting public exposure to new sites, essentially a rich-get-richer phenomenon or ‘googlearchy’ (Hindman, Tsioutsiouliklis, & Johnson, 2003). Though this unique form of technological determinism has been accused of widely influencing the media (Lohr, 2006), allegations have not been Google-specific (Introna & Nessenbaum, 2000). Furthermore, these findings have been disputed by others who portray the search engine as more of an egalitarian ‘googleocracy’ (Menczer, Fortunato, Flammini, & Vespignani, 2006). Then there is the question of its expansion, prompting discussions over "How much more should it be allowed to grab?" (Pearlstein, 2007), "Is Google too big?" (Spanbauer, 2007), and even "Is Google’s data grinder dangerous?" (Keen, 2007); echoing comparison to monopolistic Microsoft (Rivlin, 2005). Perhaps the most widely publicized concerns over Google pertain to privacy. Despite having refused similar requests from US authorities (Mohammed, 2006), it handed over identifying information of its users to Brazilian Courts (Nakashima, 2006b). This issue was also raised with the advent of the Street View feature on Google Maps, which—as the name implies—posts street-level screen shots of certain locations, complete with unsuspecting passersby (Liedtke, 2007). Google’s decision to save all search queries by default and for an indefinite period of time prompted calls "to shift the default when storing personal information back to where it has been for millennia, from remembering forever to forgetting over time" (Mayer-Schönberger, 2007, p. 17). This all culminated in a 2007 report evaluating internet privacy, in which Google ranked worst amongst a group of popular websites (Privacy International, Language Learning & Technology 6 George M. Chinnery Google-Assisted Language Learning 2007). Still, it has requested assistance from the US government to battle international censorship (Rugaber, 2007) and has agreed to ‘anonymize’ search histories after 18 months and auto-delete cookies after 24 ("Google Cookies," 2007; Wearden, 2007). Though these controversies might discourage some from using Google, intrepid instructors can pedagogically transform them into opportunities for critical thinking, akin to the higher levels of Bloom’s Taxonomy (Bloom, 1956). When learners come across these issues during an assignment, the instructor could turn the issue into a class discussion or writing assignment. Moreover, already cognizant of these issues, the instructor might intentionally plan these ‘teachable moments’ as part of the lesson. In addition to the aforementioned topics of inquiry, for instance, learners might do some of the following: 1) Compare the results of a Google search with those of Yahoo or another search engine, or with a search conducted on Scroogle, which alleges to ‘scrape’ Google of all its tracking potential. 2) Send emails on pre-selected topics to one another over Gmail and analyze the forthcoming advertisements embedded at the bottom of the message for their relevance to the original message. 3) Search for Tibet on Google Maps or Google Earth, and use its absence as a discussion prompt over Google’s policy with China? 4) Google themselves and write a paper based on the results. 5) Hold a mock trademark trial between Google and the inventor of the number ‘googol’ (using stories from NPR or The Inquirer as prompts). 6) Debate the response of Google’s Advertising Team to the release of Sicko, Michael Moore’s cinematic attack on the US health care system, using Google’s Health Advertising Blog as a prompt. 7) Discuss the notion of a ‘Google Generation’ and develop a concerted and comprehensive definition of cyberplagiarism (using this BBC News article as a prompt). Given Google’s prominence, there is unlikely to be a shortage of provocative issues. Any number of other controversies can be culled from perusals of Google Watch, Google Blogoscoped, Googlified, and Google Operating System. SEARCH PARTY Remember Excite? AltaVista? Even Yahoo—the one-time premiere search tool—ultimately suffered with the arrival of Google, and for a while even adopted the latter’s search technology (Perrone, 2004). Google has grown so exponentially as to surpass popular estimations on the advancement of technology (Kurzweil, 2001; Moore, 1965), and the prevailing signs indicate continued development and acquisition. Co-founder Larry Page has been quoted as stating that "[t]he ultimate search engine would understand everything in the world. It would understand everything that you asked it and give you back the exact right thing instantly" (Wray, 2006). Google’s CEO, Eric Schmidt, has more explicitly envisaged Google’s role in said engine’s development, indicating that "[t]he goal is to enable Google users to be able to ask the question such as ‘What shall I do tomorrow?’ and ‘What job shall I take?’" (Daniel & Palmer, 2007). Such statements are suggestive of an ambition to rise from mere search engine to total internet engine. It could be argued, however, that such efforts to dominate will ultimately destroy the minimalist appeal which attracted its legions of fans in the first place. There are indeed hints that Google may falter. Despite its continued dominance in the US and Europe (comScore, 2007a; 2007b), according to Hargittai (2004), "many people do not use it, do not know about it, or even if they use it they may not know how to do so well." China’s Baidu, for instance, remains the nation's most popular search engine, despite Google’s attempts to gain ground (La Monica, 2007). There are also signs that Google recognizes its own mortality. Concerned over the potential ‘genericide’ of its name, and its resultant loss of prestige—and even trademark—it has actively canvassed for an end to such genericized usage, and a return to its status as a proper adjective (a la Xerox) (Ahrens, 2006; Duffy, 2003; Sturgeon, 2006). Language Learning & Technology 7 George M. Chinnery Google-Assisted Language Learning In the meantime, a horde of ‘Google killers’ is looming. And the latest of these search engines are born more finely-tuned than their forbears, so as to be more accurate and useful. Yahoo’s Mindset, for example, allows users to quantify the degree to which their intentions are commercial or informational. Natural language processing (NLP) search engines such as Powerset take into account ‘stop words’ (e.g., prepositions, articles) which Google ignores by default, thereby being more likely to consider the difference between ‘taking off’ and ‘taking in’ a shirt. Along the same lines, Q & A search engines like Hakia allow users to ask questions directly indicative of their meaning. Some, including ChaCha, even provide live guides. Clustered or federated search engines such as Clusty and the more visually stimulating KartOO utilize semantic data-mining technologies. Social search engines (e.g., Swicki) take a ‘wisdom of crowds’ approach, learning from the search strategies of their community. Social bookmarking sites (e.g., del.icio.us) similarly utilize user-generated ‘tags’. A modern version of keywords, tagging is a system of classification which employs an information retrieval method known as folksonomy, a portmanteau of folks and taxonomy, literally a classification system by and for the people. Personalized search engines (e.g., Rollyo) allow users to create their own search engines, using sites of their choice. Yet others (e.g., Collarity) combine features of clustered, social, and personalized search engines. Search providers aspire to offer services that are not only more useful than their competitors, but also more convenient. Some, such as Snap, recognize that many of today’s searchers demand immediate gratification and may have limited attention spans, and therefore provide instant previews of search results. Others (e.g., Riya) take the notion of visual literacy to the extreme, excluding text altogether by searching only images. Also in existence are a host of other search tools—so-called vertical search engines—that directly target a specific range of topics (e.g., WebMD for health information), media (e.g., EveryZing searches content within podcasts and web-based videos), and populations (e.g., cRANKy for people over 50). And the list goes on… For more information on this topic, just google it. ABOUT THE AUTHOR George M. Chinnery is an English language instructor, e-teacher trainer, and PhD candidate in Language, Literacy and Culture at the University of Maryland Baltimore County (UMBC). His research and practical interests include the cross-cultural uses of information and communication technologies; elanguage learning and e-teaching; and the global digital divide. REFERENCES Ahrens, F. (2006a, July 7). goo-gle (goo'gul). The Washington Post. Retrieved June 30, 2007, from http://www.washingtonpost.com/ Ahrens, F. (2006b, August 5). So Google is no Brand X, but what is 'Genericide'? The Washington Post. Retrieved June 22, 2007, from http://www.washingtonpost.com/ Blog picked as word of the year. (2004, December 1). BBC News. Retrieved June 30, 2007, from http://news.bbc.co.uk/ Bloom, B. S. (Ed.). (1956). Taxonomy of educational objectives. New York: Longman. comScore. (2007a, December 21). comScore releases November US search engine rankings. Retrieved January 6, 2008, from http://www.comscore.com/ Language Learning & Technology 8 George M. Chinnery Google-Assisted Language Learning comScore. (2007b, June 4). comScore releases first comprehensive review of Pan-European online activity. Retrieved July 1, 2007, from http://www.comscore.com/ Conan, N. (2004, May 3). Talk of the Nation. NPR. Retrieved June 30, 2007, from http://www.npr.org/ Crampton, T. (2006, January 25). Google puts muzzle on itself in China. International Herald Tribune. Retrieved June 22, 2007, from http://www.iht.com/ Daniel, C., & Palmer, M. (2007, May 22). Google’s goal: To organize your daily life. FT.com. Retrieved July 14, 2007, from http://www.ft.com/ Duffy, J. (2003, June 20). Google calls in the ‘language police’. BBC News. Retrieved June 22, 2007, from http://news.bbc.co.uk/ Ekman, R. (2006, August 22). The books Google could open. The Washington Post. Retrieved June 17, 2007, from http://www.washingtonpost.com/ Fortune. (2007, January 22). 100 best companies to work for 2007. Retrieved June 9, 2007, from http://money.cnn.com/ Friedman, T. L. (2003, June 29). Is Google God? The New York Times. Retrieved June 30, 2007, from http://www.cnn.com/ Gallagher, D. F. (2002, April 22). A copyright dispute with the Church of Scientology is forcing Google to do some creative linking. The New York Times. Retrieved June 22, 2007, from http://www.nytimes.com/ Google cookies will ‘auto delete’. (2007, July 17). BBC News. Retrieved July 17, 2007, from http://news.bbc.co.uk/ Hargittai, E. (2004). Do you google? First Monday, 9(3). Retrieved July 1, 2007, from http://www.firstmonday.org/issues/issue9_3/hargittai/index.html Hindman, M., Tsioutsiouliklis, K., & J. A. Johnson. (2003). 'Googlearchy': How a few heavily-linked sites dominate politics on the web. Paper presented at the Annual Meeting of the Midwest Political Science Association (Chicago, IL; April 4, 2003). Retrieved June 22, 2007, from http://www.cs.princeton.edu/~kt/mpsa03.pdf Inquirer Staff. (2004, May 18). Googol may sue Google. Inquirer. Retrieved June 30, 2006, from http://theinquirer.net/ Introna, L. D. & Nessenbaum, H. (2000). Shaping the web: Why the politics of search engines matters. The Information Society, 16(3). Retrieved June 30, 2007, from http://www.indiana.edu/~tisj/ Irvine, M. (2007, April 9). Love in the age of Google. CBS News. Retrieved July 16, 2007, from http://www.cbsnews.com/ Keen, A. (2007, July 12). Is Google’s data grinder dangerous? Los Angeles Times. Retrieved January 6, 2008, from http://www.latimes.com/ Kopytoff, V. (2007, April 25). Google surpasses Microsoft as world's most visited site. San Francisco Chronicle. Retrieved June 10, 2007, from http://www.sfgate.com/ Kurzweil, R. (2001, March 7). The law of accelerating returns. Retrieved January 6, 2008, from http://www.kurzweilai.net/articles/art0134.html La Monica, P. R. (2007, June 28). Here comes China 2.0. CNNMoney.com. Retrieved July 1, 2007, from http://money.cnn.com/ Language Learning & Technology 9 George M. Chinnery Google-Assisted Language Learning Liedtke, M. (2007, June 1). Google hits streets, raises privacy concerns. MSNBC. Retrieved June 30, 2007, from http://www.msnbc.msn.com/ Lohr, S. (2006, April 9). This boring headline is written for Google. The New York Times. Retrieved June 30, 2007, from http://www.nytimes.com/ Malone, M. S. (2004, March). The complete guide to Googlemania! Wired, 12(3). Retrieved June 9, 2007, from http://www.wired.com/ Mayer-Schönberger, V. (2007, April). Useful void: The art of forgetting in the age of ubiquitous computing. Retrieved June 17, 2007, from http://ksgnotes1.harvard.edu/Research/wpaper.nsf/rwp/RWP07-022 McCloskey, P. (2007, March 6). Google 101 courses: From hardcore to high concept. Campus Technology. Retrieved June 8, 2007, from http://campustechnology.com/ Menczer, F., Fortunato, S., Flammini, A., & Vespignani, A. (2006, February). Googlearchy or Googlocracy? IEEE Spectrum Online. Retrieved June 30, 2007, from http://www.spectrum.ieee.org/ Merriam-Webster Online. (2005). Previous words of the year. Retrieved June 30, 2007, from http://www.m-w.com/info/05prevwords.htm Merriam-Webster Online (2006). Merriam-Webster’s words of the year 2006. Retrieved January 6, 2008, from http://www.m-w.com/info/06words.htm Mohammed, A. (2006, January 20). Google refuses demand for search information. The Washington Post. Retrieved June 30, 2007, from http://www.washingtonpost.com/ Moore, G. E. (1965). Cramming more components onto integrated circuits. Electronics, 38(8), 114-117. Retrieved January 6, 2008, from http://www.intel.com/technology/mooreslaw/index.htm Nakashima, E. (2006a, August 17). AOL search queries open windows onto users' worlds. The Washington Post. Retrieved June 8, 2007, from http://www.washingtonpost.com/ Nakashima, E. (2006b, September 2). Google to give data to Brazilian court. The Washington Post. Retrieved June 30, 2007, from http://www.washingtonpost.com/ New London Group. (1996). A pedagogy of multiliteracies: Designing social futures. Harvard Educational Review, 66(1), 60-92. Retrieved July 17, 2007, from http://wwwstatic.kern.org/filer/blogWrite44ManilaWebsite/paul/articles/A_Pedagogy_of_Multiliteracies_ Designing_Social_Futures.htm Pearlstein, S. (2007, April 22). How much more should it be allowed to grab? The Washington Post. Retrieved June 30, 2007, from http://www.washingtonpost.com/ Perrone, J. (2004, February 19). Yahoo! challenges for Google's crown. Guardian Unlimited. Retrieved July 1, 2007, from http://www.guardian.co.uk/ Privacy International. (2007, June 9). A race to the bottom: Privacy ranking of internet service companies. Retrieved July 1, 2007, from http://www.privacyinternational.org/ Rivlin, G. (2005, August 24). It's Google's turn as the villain. The New York Times. Retrieved June 30, 2007, from http://www.nytimes.com/ Robb, T. (2003, September). Google as a quick 'n dirty corpus tool. TESL-EJ, 7(2). Retrieved June 23, 2007, from http://www-writing.berkeley.edu/TESl-EJ/ej26/int.html Rugaber, C. S. (June 25, 2007). Google fights global internet censorship. Forbes. Retrieved January 6, 2008, from http://www.washingtonpost.com/ Language Learning & Technology 10 George M. Chinnery Google-Assisted Language Learning Rundell, M. (2000, May 17). The biggest corpus of all. Humanising Language Teaching, 2(3). Retrieved June 23, 2007, from http://www.hltmag.co.uk/may00/idea.htm Sipress, A. (2007, March 7). Microsoft attacks Google over book search. The Washington Post. Retrieved June 23, 2007, from http://www.washingtonpost.com/ Spanbauer, S. (2007, June 19). Is Google too big? The Washington Post. Retrieved June 30, 2007, from http://www.washingtonpost.com/ Stross, R. (2006, June 30). All the internet's a state: Why don't CEO's use it? The New York Times. Retrieved June 2, 2007, from http://www.nytimes.com/ Sturgeon, W. (2006, August 16). Google wants people to stop googling. CNET News.com. Retrieved June 17, 2007, from http://news.com.com/ Swaffar, J. (1998). Networked language learning: Introduction. In J. Swaffar, S. Romano, P. Markley, & K. Arens (Eds.), Language learning online: Theory and practice in the ESL and L2 computer classroom. Austin, TX: The Daedalus Group. Retrieved July 17, 2007, from http://www.daedalus.com/downloads_public/llo/llo_standard.pdf Vandenberghe, M. (2007, May 11). YouTube to remove some clips mocking Thai king. Reuters. Retrieved June 30, 2007, from http://www.reuters.com/ Vygotsky, L. S. (1978). Mind and society: The development of higher mental processes. Cambridge, MA: Harvard University Press. Wearden, G. (2007, June 12). Google to cut time it holds data. Guardian Unlimited. Retrieved July 1, 2007, from http://www.guardian.co.uk/ White, A. (2007, February 13). Court orders Google to pull Belgian news. The Washington Post. Retrieved June 17, 2007, from http://www.washingtonpost.com/ Wordsmiths hail podcast success. (2005, December 7). BBC News. Retrieved June 30, 2007, from http://news.bbc.co.uk/ Wray, R. (2006, May 23). Google users promised artificial intelligence. Guardian Unlimited. Retrieved June 17, 2007, from http://www.guardian.co.uk/ Language Learning & Technology 11 Language Learning & Technology http://llt.msu.edu/vol12num1/emerging/ February 2008, Volume 12, Number 1 pp. 12-18 EMERGING TECHNOLOGIES OF ELASTIC CLOUDS AND TREEBANKS: NEW OPPORTUNITIES FOR CONTENT-BASED AND DATA-DRIVEN LANGUAGE LEARNING Robert Godwin-Jones Virginia Commonwealth University Creating effective electronic tools for language learning frequently requires large data sets containing extensive examples of actual human language use. Collections of authentic language in spoken and written forms provide developers the means to enrich their applications with real world examples. As the Internet continues to expand exponentially, the vast "cloud" of Web pages created provides a nearly inexhaustible and continuously updated language bank, particularly in English. The issue remains, however, of how to make practical use of large amounts of data for language learning, given storage and data processing demands. Recently, new developments in storage virtualization and distributed computing offer practical solutions, as demonstrated by Amazon's Elastic Computer Cloud and SimpleDB. At the same time, the move to XML encoding of language corpora and text collections provides the compatibility and interchange which has hampered their practical exploitation for language learning. Tools are also being created to facilitate the transformation of text collections into more usable formats, particularly into syntactically annotated corpora called treebanks. These developments offer opportunities for content-based language learning in particular. GROWTH IN CONTENT-BASED LANGUAGE LEARNING Rich data collections are especially important for development of learner focused language applications. In recent years there has been a sharp increase in the development of language learning tools for specific learner populations. Not surprisingly, this has been most in evidence in Europe, as the European Union has continually added new member nations bringing with them additional official languages. The EU Europa Web site list 171 different projects in the area of content-based language learning that have, since 1999, earned the "European Language Label", awarded for creative applications in language learning. A number of these projects have been created with funding supplied by EU grant programs, including Lingua, Leonardo, and Socrates. Most involve the creation of electronic tools and multimedia and increasingly are using the Web for delivery. Many are designed for use in either instructor-led or selfstudy settings, or both. The EU site highlights a variety of projects in language learning for special purposes, including such diverse targeted areas as agricultural workers, apprentices, architectural workers, automotive workers, building maintenance workers, computer scientists, construction workers, customs officers, dock workers, entrepreneurs, hospital patients, insurance industry workers, isolated rural inhabitants, teachers, prison officers, the unemployed, and young immigrants. Some projects are even more narrowly focused, such as French for racing apprentices, Polish for missionaries, or English for ski lift cashiers. The largest number of projects targets the hospitality sector, where the need for multi-lingual workers is evident. The VIRTEX project was recently awarded first place in the European Language Label competition and is designed for workers in the hotel and restaurant industries learning English or German. Originally a CDROM project, it now incorporates a rich set of online tools, including streaming video. Several of the vocational language projects make use of a full-fledged virtual learning environment. The EUROVOLT project, which offers vocationally-oriented language learning in a variety of languages for many industries, is implemented in Moodle and makes extensive use of new media and collaborative tools. It also incorporates language e-portfolios. Interesting projects in this area also include BeCult and Online VoCAL/Weblingua, both of which have richly developed tools and media. Copyright © 2008, ISSN 1094-3501 12 Robert Godwin-Jones Content-Based and Data-Driven Language Learning Not enough information on the projects listed above is given by their Web sites to know to what extent they make use of word sets or data collections. An example that shows the benefit of word sets for content-based language learning is the Academic Word List (AWL)for English, developed by Averil Coxhead. The 570 words on the list (sub-divided into ten categories) were compiled from a corpus of 400 written academic texts. It excludes the most common 2000 English words. The list targets students entering an English-speaking university and provides an efficient base on which to create language learning exercises such as matching or cloze. The AWL Highlighter offers a nice example of the benefits of having such a list: it allows users to enter an arbitrary text, which is then parsed for AWL words and returned as a new document with the AWL items in bold, allowing students to work with the words in context. This helps guide the students to focus on vocabulary likely to be found in the text repeatedly, rather than learning items that are unlikely to be encountered again. Content-based language learning is inherently learner-centered, focusing as it does on the specific context in which the target language will be used. It also lends itself well to task-based learning activities. Many of the projects targeting language for special purposes are built around real-life scenarios, often delivered through digital video clips, as an example from the Virtex project demonstrates. The students watch a real or simulated conversational exchange or an on-the-job interaction and are provided with comprehension aids such as full/partial transcripts, isolated audio playback, cultural notes, or lists of idiomatic expressions. Students are then asked to use the expressions from the dialogues in on-line exercises, written assignments, or group work. The importance of vocabulary development in content-based language learning necessitates that the vocabulary items chosen are those needed by the learners. Developing content-specific word lists in the manner of AWL would be highly beneficial, assuming enough texts can be found to build a specialized corpus. One of the advantages of having a corpus to draw from is the possibility of using concordances as a vocabulary and grammar learning tool. Concordances are not effective for all learners, but for many motivated students it can provide a means for working with language structures through real world use. Students using concordances can be asked to reflect on areas such as inflections and collocations involving core vocabulary for the areas they are studying. Since the materials are tailored specifically for students' needs, it is more likely that such efforts will be successful. Some interesting examples of the use of concordances are collected by Bernd Rüschoff based on workshops and other sources. Tom Cobb's lextutor enriches the use of concordances by linking the found items to the on-line WordNet dictionary. Wordnet is a large lexical database of English that was first made public in 1991 and has since inspired similar collections in other languages. LANGUAGE CORPORA AND XML The percentage of Web-based vocabulary and discrete grammar exercises based on language corpora is quite low. There are many understandable reasons for this, including lack of access to appropriate corpora, incompatibility of the data with authoring tools, ignorance of how to incorporate data sets, and the need to focus on vocabulary prioritized in textbooks. The process could be made considerably easier for the average language instructor if available tools interfaced more readily with language corpora or text collections. Many popular tools for creating Web-based exercises, such as Hot Potatoes, allow for importation of text files for creating cloze or gap exercises. However, they do not allow for retrieval and incorporation of texts from large data sets or concordances. This situation is largely a by-product of the proprietary format in which language corpora and text collections have traditionally been encoded. Data with idiosyncratic encoding schemes and interfaces does not lend itself to searching or sharing. In many cases tools created in conjunction with the data have not been designed to be interoperable. Fortunately, the widespread use of XML for encoding corpora and text collections is moving towards a resolution of this problem. XML has become the de facto standard for encoding of language corpora. XML recommends itself because of its platform independence, extensibility, and widespread acceptance Language Learning & Technology 13 Robert Godwin-Jones Content-Based and Data-Driven Language Learning by software companies and researchers. Standardizing text encoding in XML greatly facilitates data interchange. Since structural and semantic information about a text is separated from its presentation in XML, the same encoded text can be displayed in multiple ways, using CSS style sheets or XSLT transformations. With the advent of XML as the preferred system for representation of corpus resources, existing tools have been modified to work with XML, while new applications have been created that are designed to be XML ready. The Linguist's Toolbox, for example, now features export to XML. The text searching software, Xaira, designed to be used with the British National Corpus, has been re-written as a general purpose XML search engine with full Unicode support. The Unicode editor CLaRK has been designed specifically to work with XML. Language archives can now be submitted to OLAC (Open Language Archives Community) by uploading a single XML file containing the necessary metadata information about the resource. Tools for the semi-automatic annotation of corpus data are being developed, such as @nnotate from the University of Saarland. DepAnn is a treebank creation tool, which uses Tiger-XML, the accepted standard for treebank encoding. EULIA, from the University of the Basque Country, provides a graphical Web interface for editing annotated corpora. These kinds of tools will become increasingly important as language data sets increase in size, since manually annotating texts to create treebanks is a slow and expensive process. One of the most widely used XML encoding schemes for text archives is TEI, Text Encoding Initiative. A new version of the TEI Guidelines was released in November, 2007. It offers a number of enhancements, including more support for manuscript descriptions and better support for multimedia and graphics. Additionally, a Web application called Roma has been developed which provides a visual editor for working with TEI. An example of the power and versatility of TEI is the Henry III Fine Rolls project, from the British National Archives. These are fiscal and administrative records in Latin from the 13th century. The site provides user-friendly access to graphic representations of the original parchment rolls, as well as the original texts, translations, and notes/annotations. TEI allows the Henry III project to be included in general searches and to be easily referenced within other text projects. Version 4 of the Perseus Digital Library, a collection of classics texts, also uses TEI encoding and adds a set of XMLbased Web services which allow for chunking larger texts into smaller units, as well as for sophisticated morphological analysis. NEW OPTIONS FOR DATA STORAGE AND PROCESSING Projects that house discrete, well-defined collections of texts can usually manage storage and delivery resources using traditional options, namely one or more servers housing a database, a Web server, and any associated Web services. If the site is popular, redundant servers might be needed. However, if the project is unusually large, such as the American National Corpus, being created as an American English cousin to the British National Corpus, the traditional project paradigm may not suffice. This is particularly the case if the goal is not just to deliver static text selections, but to allow for dynamically generated resources selected by sophisticated search, retrieval, and concatenating options such as are available with the Perseus project. In this scenario, there are significant demands in terms of processing which may well overwhelm the traditional setup for a text repository. In recent years, some new options have emerged which make it easier to set up and manage a large-scale text project. The technical means have been available for some time to enable load balancing and parallel processing, but traditionally such systems have been difficult to create and run and tended to be so expensive as to be beyond the means of most academic projects. Today, through tools and services originating with Google and Amazon, there are ways for programmers without experience with parallel or distributed systems to use the resources of a large distributed environment to achieve high performance with off-the-shelf PC's that are linked together. Large Web companies such as these, as well as Yahoo and eBay, have established developer outreach programs, through which they hope to drive more users to visit their site. As part of that program, these companies provide application programming interfaces (APIs) Language Learning & Technology 14 Robert Godwin-Jones Content-Based and Data-Driven Language Learning which instruct developers on how to write Web applications that take advantage of their sites and services. Google's MapReduce is one example which has generated considerable interest. It is a programming model and associated code library for processing and generating large data sets. The design simplifies the process of enabling multiple computers to process information and then collect back the results centrally. MapReduce assigns program instructions to multiple computers to be accomplished in parallel. It breaks down the calculations into two steps. In step one (the "map" function) a key/value pair is processed, providing a set of intermediate results. In step two (the "reduce" function), these intermediate results themselves are merged to compute a final answer. An example of MapReduce from a Google developer presentation shows how the phrase "to be or not to be" would be processed in the MapReduce model: Figure 1: MapReduce processing of "to be or not to be" This seems very simple, and it is, but by extending the process to several levels of analysis (i.e. further mapping of reduced results) it allows for very complex calculations to be broken down into simple steps. The general technique can be applied to many analytical problems. MapReduce includes its own middleware that automatically breaks down computing jobs, doles out tasks to multiple computers, and collects the results. It also creates duplicate copies of each map-and-reduce function, finds idle machines to which to assign the tasks, and tracks the results. The worker machines load their individual piece of data processing, do the work, and notify the master machine when the work is completed ("mapped") and ready to be collected ("reduced"). If a machine freezes or breaks down, the master re-assigns that task after a specific period of not being able to communicate with the worker. The process is used by Google in many different ways, including machine translation between languages. While MapReduce itself is proprietary to Google, an open source implementation, Hadoop, which implements the MapReduce method, has been released. Recently, the New York Times used Hadoop as the basis for creating a system to serve up archived newspaper articles. It needed to implement a largescale operation as the decision had been made to make all the Times archives from 1851 to 1980 publicly available for free. In addition to Hadoop, the project was implemented using several Web services available through Amazon, namely Amazon Simple Storage Service (S3) and the Amazon Elastic Compute Cloud (EC2). S3 is an archive storage service that uses the same scalable system as is implemented in Amazon's retail site. EC2 is a computing service on which one can load and run applications. Both use a standard Web services interface, as does the recently announced Amazon SimpleDB, a database service. Collectively, these services provide the ability to store, process and query data sets residing on the Internet. Traditionally, this would require a relational database (such as Oracle or mySQL) and a dedicated database administrator. In contrast, the Amazon system is designed to be relatively easy to use. While it is not free, its pricing is low enough that it may be cheaper than operating a home server, let alone setting up a cluster-based computing environment. The Amazon services used by the New York Times work well not only with text and graphics, but with other media as well. For Language Learning & Technology 15 Robert Godwin-Jones Content-Based and Data-Driven Language Learning example, CastingWords, a podcasting transcription service, stores audio files and transcribed text on S3. Clearly, this could be an interesting option for large-scale language projects. LEARNING OBJECTS REPOSITORIES AND METADATA One could envision something like the Harvard Text Annotator, an authoring tool for creating online glossed texts, running under Amazon and serving up vast quantities of on-the-fly annotated texts culled from Internet sources. For such a project to be successful, however, more than just text searching would have to be possible, even if sophisticated search options are available. Items collected in large data sets also need accompanying metadata to allow for more efficient narrowing of searches. This is important as well for finding and retrieving structured language learning resources, often labelled "learning objects" (LO). The OLAC metadata set implements a consensus approach among language corpora researchers. However, the modified Dublin Core metadata used in OLAC does not fulfil all the needs for materials to be used in language learning. One project that moves in this direction is the FLORE learning objects repository (LOR) for teaching and learning French. FLORE takes advantage of the French/English CanCore Learning Resource Metadata Initiative, a collaborative Canadian project, itself based on the IEEE Learning Object Metadata (LOM) standard. FLORE leverages a number of the LOM elements to provide additional information important for judging the appropriateness of a resource for language learners, including level of language proficiency and type of language learning environment targeted (i.e., immersion, self-study, etc.) The FLORE project is noteworthy also because it supports the Open Archive Initiative's Metadata Harvesting Protocol (OAI-MHP), which allows FLORE's metadata records to be shared with other repositories and to allow its metadata records to be linked directly with other systems. There are, in fact, more and more collections of learning resources on the Web. A recent study features an extensive international listing. However, relatively few of the LORs include standard metadata such as that provided by OAI-MHP. The GLOBE initiative (Global Learning Objects Brokered Exchange) is an effort to move repositories in this direction. The CORDRA project is also attempting to standardize LO encoding. Including standard identifying information with learning resources would help enormously in making searches across multiple data sets, known as federated searches, faster and more efficient. Federated searches for learning objects are now available from LOR sites such as Merlot and Ariadne (which even include searching of sites such as Flickr and YouTube), but the search results are inconsistent and incomplete and do not allow for advanced search options. A language learning LOR that exemplifies best practices in this area is the L2 O project out of the University of Southampton. This is a collaborative project building on the work of the eLanguage group, which produced a set of lessons for English for Academic Purposes. The L2 O project has been generating reusable LOs created mostly from existing materials. The project has developed a metadata set based on the LOM, but which adds contextual information important for language learning such as accent/region and subtitles/transcript. It complements the work done in this area by the FLORE group. The tagged LOs are retrievable from the project's repository, CLARe (Contextualized Learning Activity Repository). CLARe is currently being expanded to include social networking tools such as tag clouds and ratings. A related project, MURLLO, has begun to develop a user-friendly LO editor. One of the features that would be helpful to see included in both LO editors and repositories is support for RSS feeds. The required information for the feeds could be automatically collected from the LO metadata and used by teachers or learners to be notified whenever new learning resources in targeted areas become available. Developing easy-to-use tools for LO editing is a high priority if there is an expectation that subject matter experts such as language teachers create the resource, rather than it being created by technical specialists. A Swiss project from the University of Zurich is developing a tool for use with its LO model known as eLML. One of the better-known open source LO editors, eXe, has recently released a new version available for Windows, Mac, and Linux. A commercial LO editor, the SoftChalk LessonBuilder, is also Language Learning & Technology 16 Robert Godwin-Jones Content-Based and Data-Driven Language Learning about to see a new version with additional features including more support for multiple languages. These editors support SCORM, an LO standard that originated with the U.S. Department of Defense but which has recently been transferred to a new international organization, LETSI, Learning-Education-Training Systems Interoperability. These and other editors will likely support the new IMS standard, Common Cartridge. This is a project designed to combine e-learning standards including SCORM, LOM, and IMS QTI (Question and Test Interoperability), along with other Web services, to create a fully developed learning module which can be imported into learning management systems such as moodle or Blackboard. In the US, it is generating considerable interest as an electronic alternative to traditional textbooks. This is also the thrust of the new Digital Marketplace initiative, an outgrowth of the Merlot project based at California State University. This has been hailed as a possible model for a "national digital marketplace," advanced recently by a US government study on the price of textbooks. The Global Text Project and wikibooks are non-commercial efforts in this direction. RESOURCE LIST Content-based Language Learning • E-Lingua European project for learning the language of hotel service and management • BeCult European language project for students in hospitality industries • EUROVOLT European Vocational Online Language Teaching and Language learning via a VLE • VIRTEX Project for Hotel and Catering • Education & Training Programs in the EU List of projects related to content-based language learning • MapReduce: Simplified Data Processing on Large Clusters By Jeffrey Dean and Sanjay Ghemawat Corpora and Data-driven Language Learning • • • • • • • • The Compleat Lexical Tutor Example of data-driven language learning WordNet Lexical database for English Wordnets in the world Wordnets in multiple languages Academic Word List From Averil Coxhead Sample Exercises Data-driven Language Learning examples Best Practice Recommendations for Language Resource Description For language archives Penn Treebank Online Searchable tagged corpora in English Xaira Corpus search engine • The Linguist's Toolbox and XML Technologies By Chris Hellmuth, Tom Myers & Alexander Nakhimovsky CanCore Metadata system from Canada Raise and Rise Example of learning object from wisc online The Open Archives Initiative Protocol for Metadata Harvesting OAI guidelines Harvesting Issues About implementing OAI metadata harvesting Open Archives Initiative Metadata Harvesting Project University of Illinois project Exposing information resources for e-learning Combining OAi and IMS metadata harvesting Digital Repositories Specification From IMS Real-time demonstration of interoperability between Learning Object Repositories Interoperability demonstration involving the ARIADNE network and the FIRE federation • • • • • • • • Language Learning & Technology 17 Robert Godwin-Jones • • • Content-Based and Data-Driven Language Learning CORDRA Content Object Repository Discovery and Registration/Resolution Architecture GLOBE Global Learning Objects Brokered Exchange Federated Search Through Ariadne Sharing Language Learning Objects Example walk through the technological and pedagogical 'process models' of L2O project • A Typology of Learning Object Repositories By Rory McGreal • EML eLesson Markup Language for creating structured eLessons using XML • Common Cartridge: e-Learning Made Easy IMS Standard in place of textbooks • LETSI The international group now in charge of the SCORM standard • TEI Text Encoding Initiative Distributed Computing • • • • • • • • • Self-service, Prorated Super Computing Fun! NY Times archive use of Amazon S3 hadoop Open source implementation of MapReduce MapReduce The Google white paper Running Hadoop MapReduce on Amazon EC2 and Amazon S3 From Amazon development services S3 Amazon Simple Storage Service EC2Amazon Elastic Compute Cloud SimpleDB Amazon data base service Windows Live Web services from Microsoft Language Learning & Technology 18 News From Our Sponsors NEWS FROM SPONSORING ORGANIZATIONS Sponsors University of Hawai`i National Foreign Language Resource Center (NFLRC) Michigan State University Center for Language Education and Research (CLEAR) Co-Sponsor Center for Applied Linguistics (CAL) University of Hawai'i National Foreign Language Resource Center (NFLRC) The University of Hawai‘i National Foreign Language Resource Center engages in research and materials development projects and conducts workshops and conferences for language professionals among its many activities. FILIPINO AS A GLOBAL LANGUAGE: FUTURE DIRECTIONS & PROSPECTS This conference, to be held on March 17-19, 2008 at the University of Hawai‘i at Manoa, is a venue for bringing together scholars, writers, language teachers, researchers and other practitioners from around the world to discuss issues pertaining to the role of Filipino as a global language. Participants can be teachers, researchers, program administrators/coordinators and other practitioners who are directly involved in the promotion and nurturing of the Filipino language, literature and culture. This first conference is geared towards establishing a tradition of scholarly meetings of this kind among practitioners in the field of Filipino language, literature and culture studies. (The NFLRC serves as co-sponsor for this event). 2008 SECOND LANGUAGE RESEARCH FORUM Just in! With the theme, Exploring SLA: Perspectives, Positions, and Practices, the Second Language Research Forum (SLRF) returns to the University of Hawai‘i at Manoa for the third time on October 1719. 2008 (with the NFLRC serving as co-sponsor). Check out our website as more information becomes available. NEW NFLRC PUBLICATIONS Selected papers from Pragmatics in the CJK Classroom: The State of the Art This online collection (http://nflrc.hawaii.edu/CJKProceedings/) presents 10 selected papers from the forum on Pragmatics in the CJK Classroom: The State of the Art held from June 5 to June 7, 2006 at the University of Hawai‘i-Manoa in Honolulu, Hawai‘i. The papers are representative of the many outstanding contributions to the field of L2 pragmatics that were presented at the gathering. The papers are also representative of the diverse range of research interests and pedagogical issues taken up by the conference presenters. Cumulatively, the papers in this volume address current concerns in L2 pragmatics that range from the development of pragmatic competence by children and college-age students in both foreign and second language settings to pragmatics-focused instruction in the foreign language classroom for students at all levels of foreign language learning. The pedagogically-oriented contributions are diverse in their scope: from innovative approaches for teaching true beginners to the specialized curriculum of students receiving post-graduate professional training. Instructional innovations for L2 classroom pragmatics-focused teaching from each of the language groups — Chinese, Japanese, and Korean — are included. Check out our many other publications. OUR ONLINE JOURNALS SOLICIT SUBMISSIONS Language Learning & Technology 19 News From Our Sponsors Language Learning & Technology is a refereed online journal, jointly sponsored by the University of Hawai`i NFLRC and the Michigan State University Center for Language Education and Research (CLEAR). LLT focuses on issues related to technology and language education. For more information on submission guidelines, visit the LLT submissions page. Language Documentation & Conservation is a fully refereed, open-access journal sponsored by NFLRC and published exclusively in electronic form by the University of Hawai‘i Press. LD&C publishes papers on all topics related to language documentation and conservation. For more information on submission guidelines, visit the LD&C submissions page. Reading in a Foreign Language is a refereed online journal, jointly sponsored by the University of Hawai`i NFLRC and the Department of Second Language Studies. RFL serves as an excellent source for the latest developments in the field, both theoretical and pedagogic, including improving standards for foreign language reading. For more information on submission guidelines, , visit the RFL submissions page. Michigan State University Center for Language Education and Research (CLEAR) CLEAR's mission is to promote the teaching and learning of foreign languages in the United States. Projects focus on materials development, professional development training, and foreign language research. MATERIALS DEVELOPMENT Selected Products The list below comprises just some of our free and low-cost materials for language educators. Be sure to visit our website occasionally for updates and announcements on new products: http://clear.msu.edu. • • • • NEW! Celebrating the World’s Languages: A Guide to Creating a World Languages Day Event (guide) – This publication provides a step-by-step guide to planning "World Languages Day," a university event for high school students designed to stimulate interest in learning languages and to highlight the importance of cultural awareness. NEW! La phonétique française (CD-ROM) – Now available in beta version, this cross-platform multimedia program consists of interactive lessons that can be used by French teachers to learn how to teach pronunciation, or by advanced students working independently. CLEAR’s Rich Internet Applications initiative has been underway for over a year. RIA is a research and development lab where our programmers are working on free tools that language teachers can use to create online language teaching materials – or have their students create activities themselves! o NEW! Audio Dropboxes (put a dropbox in any web page; students’ recordings get put into your dropbox automatically) o NEW! Conversations (record prompts for students to do virtual interviews and conversations) o Mashups (combine media elements to create a new resource for language teaching) o Viewpoint (record or upload videos to link from other sites or embed inside your own web pages) o SMILE (tool for creating interactive online exercises) MIMEA: Multimedia Interactive Modules for Education and Assessment (German, Chinese, Arabic, Vietnamese, Korean, Russian; online video clips and activities) Language Learning & Technology 20 News From Our Sponsors • Language Learning Materials for Russian: A Content-Based Course Pack (online learning modules) Coming Soon! • Introductory Business German (CD-ROM) PARTNERSHIP WITH ACTFL In January and February 2008, CLEAR collaborated with the American Council on the Teaching of Foreign Languages (ACTFL) to manage their student video podcast contest, “Not Just a Language Class!” This contest was part of the Discover Languages… Discover the World! (http://www.discoverlanguages.org) national public awareness campaign to build public support for language education. Students were asked to create a two-minute video podcast depicting how the study of other languages had had an impact on their lives. ACTFL contacted CLEAR and requested that we create an online submission and storage system for the podcast entries based on our Rich Internet Application called Viewpoint. We were able to tailor the contest website to ACTFL’s needs, and look forward to future collaboration – watch for this annual contest! (ACTFL is using our Rich Internet Applications… are you? http://ria.clear.msu.edu) CONFERENCES CLEAR exhibits at local and national conferences year-round. We hope to see you at ACTFL, CALICO, MiWLA, NCOLCTL, Central States, and other conferences. UPCOMING CONFERENCE PRESENTATIONS • • Central States Conference - Dearborn, MI - March 6-8, 2008 • Workshop: Using Rich Internet Applications in Your Classroom • Session: Report from the R&D Lab: Rich Internet Applications for Language Learning • Session: Reaching Out and Building Enrollment through a "World Languages Day" CALICO - San Francisco, CA - March 18-22, 2008 • Session: Learners' Perception and Preference of Audio Stimuli During an Online Pragmatics Test NEWSLETTER CLEAR News is a biyearly publication covering FL teaching techniques, research, and materials. Contact the CLEAR office to join the mailing list or check it out on the Web at http://clear.msu.edu/newsletter/. We welcome your submissions! The Center for Applied Linguistics (CAL) The Center for Applied Linguistics is a private, nonprofit organization that promotes and improves the teaching and learning of languages, identifies and solves problems related to language and culture, and serves as a resource for information about language and culture. CAL carries out a wide range of activities in the fields of English as a second language, foreign languages, cultural education, and linguistics. Featured Resources: • CAL News Language Learning & Technology 21 News From Our Sponsors CAL News is our new periodic electronic newsletter created to provide periodic updates about our projects and research s well as information about new publications, online resources, products, and services of interest to our readers. Visit our Web site to sign up. • Alliance for the Advancement of Heritage Languages The Alliance for the Advancement of Heritage Languages (the Alliance) consists of individuals and organizations who share a commitment to advancing language development for heritage language speakers in the United States. The Alliance is committed to fostering the development of the heritage language proficiencies of individuals in this country as part of a larger effort to educate members of our society who can function professionally in English and other languages. The Alliance has revamped its Web site to offer expanded content and improved navigation. • National K-12 Foreign Language Survey Underway CAL conducts a national survey of foreign language instruction in elementary and secondary schools every decade to gain greater understanding of current patterns and shifts over time in enrollments, the number of schools offering foreign language classes, the types of foreign language offerings, foreign language curricula and methodologies, teacher qualifications and training, and the effects of NCLB, among other issues. We are currently conducting the third survey to be able to show trends in foreign language education at three points in time (1987, 1997, 2007). For further details, see the fall 2007 update. • Center for Research on the Educational Achievement and Teaching of English Language Learners (CREATE) Visit the newly expanded CREATE Web site to learn more about CREATE, its research and upcoming events. To keep current on CREATE activities, sign up to receive an electronic newsletter and periodic announcements. • Spotlight on Language Series In support of the Discover Languages campaign led by ACTFL, CAL has developed a regular Web series to provide information about specific languages. These language spotlights are introductory in nature and are intended to encourage readers to explore these languages and CAL’s work with them in more detail. Different languages will be highlighted periodically. • CAL Services CAL provides a variety of professional development and technical assistance services related to language education and assessment needs. Recent Publications: • • • • • • Guiding Principles for Dual Language Education, Second Edition Developing Reading and Writing in Second Language Learners Refugees from Burma: Their Backgrounds and Refugee Experiences An Insider’s Guide to SIOP Coaching Realizing the Vision of Two-Way Immersion: Fostering Effective Programs and Classrooms What’s Different About Teaching Reading to Students Learning English? Visit CAL’s Web site to learn more about our projects, resources, and services. Language Learning & Technology 22 Language Learning & Technology http://llt.msu.edu/vol12num1/review1/ February 2008, Volume 12, Number 1 pp. 23-42 REVIEW OF FIVE ENGLISH LEARNERS' DICTIONARIES ON CD-ROM Title Platform Cambridge advanced learner’s dictionary on CD-ROM. (2nd Ed., version 2.0a, 2005) Collins Cobuild advanced learner’s English dictionary on CD-R4OM. (5th Ed., 2006) Windows 98, NT4, 2000, ME, or XP Windows 98, 2000, ME, XP, or NT 350 MHz 64 MB RAM Free hard disk: 120 MB Longman dictionary of contemporary English, writing assistant edition CDROM. (Updated 4th Ed., 2005) Windows 98, NT, ME, 2000, or XP. Mac 10.2. or higher. Linux Redhat 9, Mandrake 10/10.1, Suse 9.1, Debian 3.0 Windows 2000 or XP 350 MHz 128 MB RAM Free hard disk: 460 MB Pearson Education, Ltd. www.pearso ned.co.uk 300 MHz 128 MB RAM Free hard disk: 450 MB Macmillan Publishers, Ltd. www.macmi llan.com Windows 98, NT, ME, 2000, or XP. Mac 10.2 or higher. Linux Redhat 9 or higher 350 MHz 128 MB RAM Free hard disk: 110 MB Oxford University Press www.oup.co .uk Macmillan English dictionary for advanced learners CD-ROM. (2nd Ed., version 2.0.0702, 2007) Oxford advanced learner’s compass. (7th Ed., 2005) Minimum hardware requirements 300 MHz 128 MB RAM Free hard disk: 200 MB Publisher Support offered Price ISBN Cambridge University Press www.cambri dge.org Harper Collins Publishers, Ltd. www.harper collins.com Brief manual attached to the disk. Guided tour $21.07 0521604 990 Brief manual attached to the disk. Help menu. Phone: 44(0)1413063 322 Brief manual attached to the disk. Guided tour. www.longman .com/dictionari es/ $15.94 0007210 124 $64.79 9781405 811286 Brief manual attached to the disk. Guided tour. www.macmill andictionary.c om Graphic tutorial $41.48 9781405 025263 $34.07 0194316 491 Review by Alfonso Rizo-Rodríguez, Department of English, University of Jaén, Spain English learners’ dictionaries on CD-ROM are attracting more and more attention, given the enormous potential afforded by the new technologies to enhance language description. McCorduck (1996) states that this "shows the exciting promise of the application of multimedia computer technology to lexicography and language learning" (p. 225). As an electronic resource, a dictionary on CD-ROM is based on its printed counterpart, "a synchronic monolingual dictionary intended to meet the demands of the foreign user" (Herbst, 1990, p. 1379). This review focuses on the latest editions of five advanced learners’ dictionaries of English on CD-ROM, each of which comes packaged with their printed editions. The review highlights their most outstanding characteristics and constraints and compares them over ten dimensions: Copyright © 2008, ISSN 1094-3501 23 Alfonso Rizo-Rodríguez Review of Five English Learners' Dictionaries on CD-ROM graphical user interface, accessibility and information retrieval, macrostructure, microstructure, thesauruslike consultation, complex searches, copy and print functions, extras, multimedia resources, and customization. The comparison also addresses the advantages of computer-aided lookup over paper-based consultation methods. GRAPHICAL USER INTERFACE (GUI) The GUI is a key feature of electronic dictionaries (EDs) since users expect to gain access to every function in an electronic dictionary in a simple, direct manner. A GUI is graphics-based, rather than character-based, although it reproduces the entries content of paper-based dictionaries. Consequently, GUI design may make consultation easier while extra attributes (e.g., color and clear typography) can act as psychological incentives for users. Corris, Manning, Poetsch, and Simpson (2000) note that "electronic interfaces still possess the charm of novelty" (p. 178), and this helps explain users’ satisfaction with EDs and their preference for them (Nesi, 2000b). A clear evolution is obvious in the graphical interfaces of current EDs, compared to their earlier editions: graphical innovations have been added to make the interface more modern and stylish and to enable access to menus and options easier; further links to extras have been included; and small pop-up windows have been designed for joint use of EDs with other computer applications (e.g., word processors, hypertext on the Internet, or e-mail). These elements, and others mentioned below (pop-up menus, tool bars, on-screen buttons, and dialogue boxes), add to the user-friendliness of EDs and ensure "fail-safe" lookup procedures (de Schryver, 2003, p. 182). Figures 1 through 5 in the Appendix illustrate the GUIs of the five EDs examined here, all showing the same search word, catch. All five EDs have similar interfaces, although certain differences are apparent. As far as layout is concerned, three dictionaries—Cambridge Advanced Learner’s Dictionary (henceforth CALD2), Collins Cobuild Advanced Learner’s English Dictionary (COBUILD5), and Macmillan English Dictionary for Advanced Learners (MED2)—present a long, narrow panel on the left displaying an alphabetical index list and a results list of all entries containing the search word. This design might be more informative than the interfaces of the Oxford Advanced Learner’s Compass (OALD7) and the Longman Dictionary of Contemporary English (LDOCE4), which simply include a drop-down menu next to the search box showing the search word within a limited list of words beginning with the letters keyed in. Neither of the last two dictionaries displays the full A-Z index of entries; however, OALD7 offers a results list window showing a word in four sections: headwords, idioms, phrasal verbs, and structures. The second element on the screen is a definition (or entry) window showing the entry for the word. The design of this window is different in each of the dictionaries. In paper dictionaries, "all the information [is presented] in a linear order on the same level (unless using different typesets or colours)" (Tono, 2000, p. 855); in contrast, three electronic dictionaries—CALD2, MED2, and LDOCE4—opt for the so-called "layered" presentation (p. 857). They do not display all information straightforwardly, but rather, by clicking on different tabs, the user can retrieve further details about the search word not supplied directly in the text of the definition window. This utility adds to the simplicity of entries and facilitates customization, that is, adaptation to the lookup aims of each user. In contrast, the other two EDs examined have a "traditional interface", where "information is provided in a similar way to that in a paper dictionary" (Tono, 2000, p. 856), and, hence, they reproduce the entry text of their printed counterparts accurately. As a result, the definition window is packed with information and might be a little hard to process. Standard elements of the interface of EDs are a menu bar and a tool bar. The former includes Windowslike menus, such as File, Edit, Options, History, or Help, and is found in two dictionaries, CALD2 and COBUILD5. In contrast, MED2, LDOCE4 and OALD7 opt for Tool bars of various kinds and designs, including options such as Back, Forward, Copy, Print, Paste, History, Help, and Quick Search, also used Language Learning & Technology 24 Alfonso Rizo-Rodríguez Review of Five English Learners' Dictionaries on CD-ROM in CALD2 and COBUILD5. Tool bars may prove more user-friendly than Windows-like menus since they provide quicker, more direct access to the software utilities. A more recent feature of the interface of these EDs is the inclusion of on-screen buttons, or tabs, which give direct access to dictionary extras, as well as to other complementary books accompanying the dictionary proper. These elements are very practical for users because, by simply clicking a button, they can easily consult other resources. Some of the EDs examined also incorporate additional components on the right-hand side of the interface. LDOCE4 offers access to a Phrase Bank, an Examples Bank, and an Activate-Your-Language option. OALD7 offers Word Origin, Example Sentences, and a Wordfinder tool. The interfaces of CALD2, COBUILD5 and MED2 do not contain these types of panels, and, hence, LDOCE4 and OALD7 can be considered superior in this regard. From the user’s point of view, all five interfaces allow quick and easy access to the desired information (see Table 1). The interfaces of LDOCE4 and OALD7 stand out among the others since they enable access to the largest number of dictionary extras and complementary books. MED2 and CALD2 are unique due to their modern, layered interface. COBUILD5’s interface (see Figure 2) proves rather plain in its content and visual appeal. Table 1. Comparison of Graphical User Interface Features Index list panel Results list panel Definition window Layered interface Traditional interface Menu bar Tool bar Access to complementary books Extra type of panels CALD2 + + + + + + - COBUILD5 + + + + + + - LDOCE4 + + + + + MED2 + + + + + - OALD7 + + + + + + ACCESSIBILITY AND INFORMATION RETRIEVAL EDs surpass hard copy dictionaries in their search potential. The software locates every occurrence of a word in the entire dictionary. Thus, users automatically get information about lexical items as they appear not only in main entries, but also in derived words, compounds, phrasal verbs, idioms, collocations, definitions, and examples. However, the search capabilities of the five EDs are not identical. MED2, CALD2 and COBUILD5 (in that order) offer the widest range whereas OALD7 is slightly less informative because its simple search function locates a word only in headwords, idioms, phrasal verbs, and collocations. LDOCE4 looks for a word only in main entries. Which search function is most practical to a given user depends on his/her needs and proficiency level. A wide search range is most useful to advanced users, researchers, and EFL teachers, who may want to seek detailed information about a lexical item as it is used in idioms, examples, collocations, or definitions. Conversely, language learners may just need to access a word in a quick manner by referring directly to its main entry. CALD2 uses a special color code for entries, which helps users identify the type of entry appearing in the results list, e.g., headwords are displayed in dark blue, phrasal verbs in green. This feature might be particularly helpful if the user needs to identify the specific location of a search term in the text of the dictionary quickly, e.g. in a definition or in an idiomatic construction. In contrast, OALD7 uses explicit labels (headwords, idioms, phrasal verbs, and collocations) for that purpose. This practice, typical of OALD7, is even more helpful to users because it is more self-explanatory. None of the other three dictionaries feature this type of color code or label convention, which makes them less user-friendly. Language Learning & Technology 25 Alfonso Rizo-Rodríguez Review of Five English Learners' Dictionaries on CD-ROM Accessibility is a major and distinctive attribute of EDs, realized in different word look-up methods (Corris et al., 2000). Users may key in a word in the search box, they may choose a lexical item from a word list after typing in "the first three or so letters of a word" since "the word list automatically scroll[s] down to that point", or they may locate it by means of "fuzzy spelling options" (p. 175). This last search method is made possible as a result of a very helpful feature of the five dictionaries, their spell-check function. When a word typed in the search box is spelled incorrectly, a small window is automatically activated providing the correct form, as in COBUILD5. The four other EDs also display other lexical items with a similar spelling. To illustrate, if the user types *assesment, the EDs list assessment, abasement, assortment, amusement, amazement, and so forth. This fuzzy spelling option helps language learners retrieve words from the A-Z list when they are not certain about correct spelling and are guided merely by the way the words sound. Other instances of the flexibility of EDs as search tools are their "hyperlinking" function, "a search mechanism by which a double click on a word on screen will call up a dictionary entry for that word" (Nesi, 1999, p. 61) and instant retrieval of fixed expressions and idioms, which can be looked up with great ease. The accessibility of the entry text in the five EDs is superior to that of print editions due to the design of the entry window. The entry word is clearly highlighted; extensive use is made of indentation in order to separate meanings, example sentences, derivatives, and so forth; spacing is generous; and use of color and fonts adds to the clarity and neatness of the text (see Figures 1 – 5). As a result, the layout of entries and their visual impression is entirely distinct from that characteristic of print editions. Some variation among the five EDs is, nevertheless, evident in the presentation of entries. As shown in Figures 1 through 5, only MED2 offers a comprehensive list, or menu, of the various meanings of polysemous words at the top of the definition window—e.g. catch (verb). This facilitates the lookup process for language learners since the whole range of uses of the word can be seen at a glance. Moreover, each of the numbered meanings provides hyperlinks to the word’s definition and examples. LDOCE4 optionally displays this type of menu by allowing the user to click on the appropriate tab on the toolbar at the top of an entry. The other EDs, like their printed versions, do not feature this element, and hence the user inevitably has to scroll through the text, which proves cumbersome at times. Finally, access to further lexicographical information is enhanced through inclusion of a link to various online electronic dictionaries. This is a feature exclusive to CALD2 and MED2. Users may thus obtain details about a word not supplied by the dictionary on CD-ROM. More specifically, in CALD2, users may conduct searches in various works published by Cambridge University Press, such as the Dictionary of American English, the International Dictionary of Phrasal Verbs, the International Dictionary of Idioms, or the Diccionario Klett Compact Spanish-English. In MED2, the link enables free access to the British and American editions of this work, as well as regular updates of new words. Table 2. Accessibility and Information Retrieval Simple A-Z search in dictionary text Color/label-coded results list Spell-check function Hyperlinking function Easy-to-read entry text List of meanings Online dictionaries link CALD2 ++++ + + + + + COBUILD5 +++ + + + - LDOCE4 + + + + + - MED2 +++++ + + + + + OALD7 ++ + + + + - Table 2 summarizes the characteristics of the five EDS as far as accessibility and information retrieval are concerned. The first criterion in the table is satisfied by each dictionary to various degrees, so that a Language Learning & Technology 26 Alfonso Rizo-Rodríguez Review of Five English Learners' Dictionaries on CD-ROM gradation is established by means of one or more plus signs. On the whole, MED2 and CALD2 have the greatest number of accessibility and information retrieval devices. MACROSTRUCTURE The macrostructure of the five EDs, that is, the alphabetical list of entries, is identical to that of their printed counterparts. Moreover, the latest editions of these printed works, based on solid corpus evidence, include a large variety of newly coined terms, as well as subject-specific vocabulary, especially from the areas of science, computing, and communications technology. For example, the terms ozone-friendly and neural network are included in all five dictionaries. However, the coverage of regional varieties of English is not uniform in the five EDs. MED2, CALD2 and OALD7 (in this order) stand out in this respect because they include words characteristic of different varieties, such as American, Australian, or Indian English, such as anchorman (American), bonzer (Australian), or prepone (Indian). In contrast, regionalisms, with the exception of American English, are scarce in LDOCE4 and COBUILD5. Similarly, the treatment of spoken and informal registers is much more exhaustive in MED2, CALD2, OALD7, and LDOCE4 than in COBUILD5. For example, the terms nohow, irregardless, foul something up, allnighter, and argy-bargy are included only in the former four dictionaries. Representative coverage of regionalisms and register differences adds to the usefulness of a learner’s dictionary for decoding purposes. Interestingly, CALD2 has maintained an approach characteristic of the first and second editions of its printed versions: the different meanings of polysemous words are described in separate entries. For example, the noun line is described in fifteen different entries, and its derivative line (verb) appears three times under specific meanings of its noun. Similarly, the description of very common functional words, e.g., in, off, and lexical ones, e.g., go, put, requires a very large number of entries. For example, the verb go itself (not its use in a phrasal verb) is described in twenty-six different entries accompanied by semantic indications such as move, leave, become, weaken, and happen. This proliferation of entries may be a serious obstacle for users interested in one particular meaning of a word because, instead of finding all the information at a glance within the same entry, users will have to scroll down on the screen. Another unfortunate consequence of this practice is that the pronunciation and irregular forms (in the case of verbs) of words are repeated in every single occurrence of each entry (Rizo-Rodríguez, 2005). On a positive note, the normal macrostructure of the printed edition of LDOCE4 is interspersed with 9,000 encyclopaedic entries taken from the Longman Dictionary of English Language and Culture (2nd revised edition, 1999). Similarly, the alphabetical list of OALD7 words is supplemented with 10,000 cultural entries from the Oxford Guide to British and American Culture (1999). This is particularly advantageous because learners can gain access not only to the core vocabulary of English, as is expected in a learner’s dictionary, but also to a significant number of terms (typical of an encyclopaedia) which may enrich their background and cultural knowledge. Table 3. Macrostructure A-Z list of entries: Identical to printed edition Regional varieties of English Spoken and informal registers Addition of encyclopaedic entries Language Learning & Technology CALD2 + COBUILD5 + LDOCE4 - MED2 + OALD7 - ++++ +++ - + ++ - ++ +++ + +++++ +++ - +++ +++ + 27 Alfonso Rizo-Rodríguez Review of Five English Learners' Dictionaries on CD-ROM Table 3 depicts the macrostructure features of the five EDs. Two or more plus signs indicate differences in coverage. Accordingly, MED2, CALD2 and OALD7’s A-Z list of entries is rather representative of the different uses and varieties of the English language compared to LDOCE4 and COBUILD5. Besides, the inclusion of encyclopaedic entries in LDOCE4 and OALD7 constitutes a significant addition to the macrostructure of their printed counterparts. In contrast, CALD2, COBUILD5, and MED2 are still heavily dependent on the A-Z list of their printed editions, an obvious shortcoming. MICROSTRUCTURE The microstructures of the five EDs, that is, the content of entries, are exact replicas of their hard-copy versions. Two dictionaries, COBUILD5 and OALD7, opt for the so-called traditional interface, which faithfully reproduces the entry text of the book. In contrast, the microstructure of CALD2, MED2, and LDOCE4 exhibits a layered presentation which permits users to select extra details not supplied directly in the definition window. As a result, entries are less compact and users can customize their searches (see Table 4). The entry window in CALD2 displays links to pictures, study pages, and terms related to the target word, as well as the on-screen buttons: Smart Thesaurus, Word Building, Verb Endings, Extra Examples, Collocations, Common Learner Errors, and Usage Notes. Both the links and the notes on Collocations, Common Learner Errors, and Usage are appended to respective entries in the printed version, but the other buttons supply additional information found only in the electronic version. Similarly, MED2 includes different tabs in some entries—Am/BrE Differences, Animations, Avoiding Offense, Collocations, Cultural Notes, Exercises, Expressing Yourself, Get-It-Right Notes, Illustrations, Metaphors, Sound Effects, Synonyms, Usage Notes, Weblinks, Word Sets, and Word Stories. All of these, except Exercises, Sound Effects, and Weblinks, also form part of the print dictionary. Moreover, every meaning of an entry on the CD-ROM is connected to a thesaurus button, and entries for verbs, nouns, and adjectives include an Inflections button which can be clicked to get the inflected forms. LDOCE4 includes a toolbar showing extra details about a lexical item: Pronunciation, Menu, Word Family, Word Origin, Verb Forms, Word Sets, and Frequency of Use. These details, except the last, are available only in the electronic version, which has maintained three features of the hard-copy edition— Word Choice, Word Focus, and Collocations—by appending them to some entries. Finally, OALD7 supplements entry information with the addition of three small panels (exclusive to the CD-ROM): Word Origin, Example Sentences, and Wordfinder. On the whole, as Table 4 illustrates, the microstructure of LDOCE4 offers the largest number of extras, as compared with its paper-based edition, while COBUILD5 contains the least amount of extra information. This variety of supplements adds to the usefulness of an ED since they are intended to complete the description of a search term. Users can thus obtain a wealth of information about a word, and this extra knowledge may significantly contribute to enhancing their command of the language. Table 4. Microstructure Traditional presentation of entry text Layered presentation of entry text Extra information CALD2 + + COBUILD5 + - LDOCE4 + + MED2 + + OALD7 + + THESAURUS-LIKE CONSULTATION Four EDs—CALD2, MED2, LDOCE4, and OALD7—feature thesaurus-like "onomasiological" resources not furnished by their book counterparts, apart from their A-Z "semasiological" dictionaries (RizoRodríguez, 2004, p. 37). Every meaning of a word in CALD2 is accompanied by a button which opens up Language Learning & Technology 28 Alfonso Rizo-Rodríguez Review of Five English Learners' Dictionaries on CD-ROM a Smart Thesaurus, whose internal semantic typology of concepts, drawn up very much in the style of Roget’s Thesaurus of English Words and Phrases (150th anniversary edition, 2002), though less exhaustive, constitutes the basis of a very detailed onomasiological classification of English words that proceeds from concepts to words (Rizo-Rodríguez, 2004). The Smart Thesaurus utility in CALD2 is perfectly integrated with the A-Z list, so that the user can obtain a large variety of expressions semantically related to each meaning of a search word, as well as definitions and examples. Finally, the user can consult an index to the Smart Thesaurus categories for words classified in related categories. The second edition of MED also includes a Thesaurus that supplies synonyms, antonyms, related words, and their definitions next to every meaning of a word. Conceptual categories semantically associated with a search word can be looked up, but they are not organized into an index. Another shortcoming of this Thesaurus is that it can only be consulted jointly with the A-Z dictionary, not as an independent resource. LDOCE4 features the Longman Language Activator (2nd edition, 2002). Conceived as a production dictionary, the Activator organizes vocabulary around 866 key words denoting basic concepts, which in turn are expressed by more specific lexical items. The internal organization of the Activator is alphabetical with each particular word explained and illustrated. Users can refer to this onomasiological dictionary separately or can browse through it in conjunction with the A-Z dictionary. Similarly, OALD7 incorporates the Oxford Learner’s Wordfinder Dictionary (1997), an onomasiological lexicon which classifies vocabulary into 630 keywords in alphabetical order. Each of these semantic spheres comprises semantically related subareas and their corresponding terms, all of them accompanied by a definition and example sentences, as in the Activator. The Wordfinder is available as an independent resource on this compact disk, or it can be looked up in combination with the A-Z dictionary. In summary (see Table 5), the most complete onomasiological information is found in CALD2, whose Smart Thesaurus is easy to use. Equally accessible is LDOCE4’s Activator, another excellent source of terms semantically related to a search word. OALD7’s Wordfinder is clearly inferior in coverage to the Activator since it targets intermediate learners. MED2’s Thesaurus is a user-friendly tool, but, as explained above, it has some limitations. Finally, COBUILD5 does not include thesaurus resources. Instead, many of its entries list synonyms and antonyms of a search word; however, the electronic version contains a higher number of these types of terms than its printed edition. Table 5. Thesaurus-Like Consultation Thesaurus-like resources Definition of related words Exemplification of related words Index of semantic categories Alphabetical organization of concepts Thesaurus as independent resource CALD2 + + + + + COBUILD5 - LDOCE4 + + + + + MED2 + + - OALD7 + + + + + COMPLEX SEARCHES A feature exclusive to EDs is their capacity to carry out complex word searches in a manner that exceeds the capabilities of the most meticulous paper-based dictionary user. It is one of the clearest indications of the potential of EDs as language learning/teaching tools. Only four of the EDs examined incorporate an advanced function that allows users to carry out complex lexical searches: CALD2, LDOCE4, MED2 and OALD7 (see Table 6). In order to compensate for the absence of this function, COBUILD5 enables users to conduct two extra searches: a phonetic search (e.g., by entering the word rite, one obtains rite, right and write) and a morphological search (e.g., the program gives the singular form of an irregular plural noun keyed in (e.g., mouse, mice) or the bare infinitive of any inflected verbal form (e.g., lie, lying)). Language Learning & Technology 29 Alfonso Rizo-Rodríguez Review of Five English Learners' Dictionaries on CD-ROM Three of the EDs, CALD2, MED2 and LDOCE4, possess an elaborate system of filters for customizing complex searches. For example, users can look for all the adverbs ending in –ly, all the verbs followed by a that-clause, or all the Spanish loan words. While the search window of each of these EDs shows its own distinctive design, the searching mechanisms are similar. They are based on filters and are graphically displayed in a dialog box where the filters are conveniently organized and ready to be selected individually or in combination with others. Searches can be conducted by entering a word in an appropriate box and by selecting one or more filters. The software then looks for that word according to the conditions established by the filters selected. For instance, in CALD2, after entering the word get and marking the option Linking verb in the grammar filter, the software identifies only the copulative uses of that verb. Alternatively, a search can be launched simply by selecting some of the filters. In that case, the program returns all the words in the dictionary that match the filter criteria. For example, in MED2, by choosing the option Australian in the filter Region, the user obtains all the terms registered in this dictionary typically used in that variety of English. From the user’s perspective, these sophisticated searching mechanisms exceed the demands of the average learner, who will not normally make use of them. However, these tools do serve the needs of the EFL teacher or the language researcher concerned with the description of English or the retrieval of very specific linguistic data which can be used in the classroom. MED2 and CALD2’s search systems are quite exhaustive. They include six filters: part of speech, grammar, region, style (or usage), frequency, and subject (or topic). CALD2 is more meticulous than MED2 in its grammar filters. It includes refined options, such as "+ two objects", "+ object + toinfinitive" for the description of verb complementation patterns. Conversely, the region filter is clearly more specific in MED2 (e.g., Indian, New Zealand, Scottish, and Irish English) than in CALD2 (which includes only American English, British English, and Other Regions). In contrast, LDOCE4 has only three filters—frequency, part of speech, and style— but it features its own original types of searches for multimedia additions, word origin, subject, and pronunciation. Multimedia, like MED2’s Extra Features Search, locates all the terms that are accompanied by illustrations or sound effects in the dictionary. Word Origin returns lexical borrowings that entered English and classifies them by century. Subject search finds terms specific to a discipline. Finally, pronunciation locates all the words containing a sequence of phonemes, and, hence, it is useful in locating words whose exact spelling is not known, or in retrieving homophones and words that rhyme. MED2 and COBUILD5, unlike CALD2 and OALD7, possess identical pronunciation functions. In contrast, the advanced search utility in the Oxford Advanced Learner’s Dictionary is quite different. Searches have to be conducted within a blank window without the help of any filter menus. Instead, searches require the use of specific labels listed in the advanced search window. After a test of this system, it was found that it is in need of substantial revision and simplification, mainly because queries must be formulated in a syntax barely explained in the Help menu, and also because many of the labels are, unfortunately, interpreted literally by the software. For example, after keying in the label "computing", the user does not only retrieve terms relative to this discipline, but rather all the occurrences of this lexical unit in the dictionary text. Finally, advanced searches can be formulated with the help of wild cards: the symbols ? (standing for one letter) and * (standing for zero or more letters) and Boolean operators AND, OR, and BUT. This is possible only in MED2, LDOCE4 and OALD7. This sophisticated tool can be exploited by EFL teachers to obtain supplementary classroom materials. For instance, by typing in will / would AND if, teachers can retrieve a large number of examples of conditional sentences. Similarly, language researchers may want to expand the scope of a search in order to look for certain types of lexical items. The search NOUN:*ee returns all the nouns ending in –ee recorded in the dictionary. Language Learning & Technology 30 Alfonso Rizo-Rodríguez Review of Five English Learners' Dictionaries on CD-ROM On the whole, MED2 has the most powerful advanced search instrument. CALD2 and LDOCE4’s are also very efficient. The three of them, unlike OALD7, are equipped with an elaborate system of filters which simplify searches. In contrast, the advanced search function of OALD7 is extremely intricate and actually hinders complex searches while COBUILD5 does not incorporate this utility at all. Table 6. Complex Searches Advanced search function Filter system Pronunciation search Wild cards search Boolean operators search CALD2 + + + - COBUILD5 + - LDOCE4 + + + + + MED2 + + + + + OALD7 + + + COPY AND PRINT FUNCTIONS Much of the usefulness of the results obtained with the advanced search function described above resides in their readiness to be exported to a word processor. The copy and print functions will be particularly appreciated by EFL teachers and language researchers, who can easily employ the results furnished by the sophisticated search mechanism. CALD2 and MED2 can be successfully exploited for this purpose. With CALD2, both the hits listed in the results window after performing a normal search and those appearing in the advanced search window can be copied (up to 1,000 hits) or printed. MED2 also permits copying, printing, and saving with the advanced search utility (up to 200 hits). The other dictionaries, however, fall short in this respect. LDOCE4 and OALD7’s complex search results cannot be copied or printed at all. This is a shortcoming which significantly lessens the utility of the advanced search mechanism. Another restriction of OALD7 is that the text from the Word Origin window and the content of the Example Sentences window cannot be exported or printed. In contrast, COBUILD5’s Wordbank, LDOCE4’s Examples Bank and Phrase Bank, and MED2’s extra examples can be copied or printed, while those in CALD2 can only be printed. Moreover, COBUILD5 and MED2 are the only dictionaries reviewed which allow copying of a complete entry without altering its text properties (font, color, phonological transcription, indentation, spacing) so that it can be used for teaching purposes after pasting it into a word processor document. The other EDs can copy an entry but turn it into unformatted, plain text. To conclude, as shown in Table 7, as far as the handling and export of search results and dictionary text are concerned, the best dictionaries are MED2 and CALD2. COBUILD5 and LDOCE4 are less flexible, and OALD7 is deficient in this regard. Table 7. Copy and Print Functions Powerful copy facility Copying and printing results list items Copying and printing advanced search hits Copying extra examples Printing extra examples Copying entry text without alterations Language Learning & Technology CALD2 + + + COBUILD5 - LDOCE4 - MED2 + + OALD7 - + - + + + + + - + + + - 31 Alfonso Rizo-Rodríguez Review of Five English Learners' Dictionaries on CD-ROM EXTRAS The latest versions of EDs all contain a number of extra features to cater to users’ different needs. According to Varantola (2002), "the future dictionary is […] an integrated tool or a number of tools in a professional user’s toolbox where it coexists with other language technology products such as encyclopedic [sic] sources of reference, different types of corpora, corpus analysis tools" (p. 35). However, an ED should not just be a mere amalgam of extra materials (Nesi, 2000a); instead, "there is need for ‘multi-referencing’: for simultaneous signalling to the user that the same query item is to be found in a number of different resources" (Leech & Nesi, 1999, p. 301). An examination of the five EDs under review reveals that four of them—LDOCE4, OALD7, MED2, and CALD2—stand out in terms of their extras (see Table 8). Both LDOCE4 and OALD7 are collections of reference books: the former includes the text of its printed counterpart as well as the Longman Language Activator. Similarly, OALD7 comprises the 7th edition of its printed version, The Oxford Guide to British and American Culture (1999), The Oxford Learner’s Wordfinder Dictionary (1997), and also Oxford Genie. All these books work in perfect conjunction, as advocated by Leech & Nesi (1999), and guarantee multi-referencing. The other three EDs do not include further reference works. An exclusive feature of OALD7 is its Word Origin window that furnishes etymological information on 20,000 words. LDOCE4 is notable for its Phrase Bank (a vast collection of phrases containing the search word and collocates used with the entry word, all of them illustrated with additional examples), its lesson plans for teachers (a mixture of notes for teachers and language activities intended to promote familiarization with the dictionary and its use), and its grammar section (an appendix offering brief summaries of various grammar points). All these extra elements constitute "evidence of pedagogical design and consideration of the learners’ needs" (Seedhouse, 1997, p. 63). They will benefit the advanced user who wants to obtain further details about a search term and also the teacher concerned with promoting dictionary skills. A similar attempt on the part of LDOCE4, OALD7, CALD2 and MED2 to serve those needs is the inclusion of a writing assistant. But the capabilities of this tool are not identical in the four EDs. The most practical writing assistant is that of LDOCE4. It provides thesaurus-like details and diverse grammatical information, as well as typical learner mistakes and their correct versions. In contrast, it is questionable whether the Know-how utility in OALD7 is equally useful for every user. Learners must type a sentence in order to check its acceptability against various example sentences, but the effectiveness of this resource ultimately depends on the users’ ability to infer grammatical information from linguistic chunks. Despite its name, CALD2’s Superwrite is not a genuine writing assistant since it simply displays any word in a text pointed at with the cursor with its complete entry. Finally, MED2 does not feature a writing assistant proper either, but its sections Improve Your Writing Skills and Expand Your Vocabulary offer appendixlike explanations on a number of communicative functions and expressions accompanied by interactive writing exercises. Additional extra features are MED2’s language awareness articles (fourteen contributions by leading specialists on idioms, metaphors, pragmatics, or word formation), its atlas featuring geographical information, and its links to websites offering cultural and encyclopaedic information. Similarly, CALD2 contains study pages about grammar, vocabulary, and pronunciation. These will all be helpful to language learners, who can easily access and use these resources through on-screen buttons. LDOCE4, CALD2, OALD7, and MED2 also provide a large variety of exercises (accompanied by both check-answer and show-answer buttons) intended for upper intermediate and advanced level learners of English. LDOCE4 offers the widest range on grammar, vocabulary, culture, listening comprehension, intonation, sentence dictation, and word dictation. OALD7 incorporates a variety of vocabulary exercises, as does CALD2, which additionally includes grammar exercises. Finally, MED2 features lexical and Language Learning & Technology 32 Alfonso Rizo-Rodríguez Review of Five English Learners' Dictionaries on CD-ROM grammatical activities, matching Get-It-Right explanatory notes on typical learner errors (appended to 104 entries) as well as writing exercises. Moreover, LDOCE4, CALD2, and OALD7 include exam papers from a variety of language certificates, e.g., Certificate of Proficiency in English, Test of English for International Communication, or Business English Certificate, which should be useful especially for advanced learners of English preparing for these examinations. Another outstanding feature of the latest editions of EDs on CD-ROM is the inclusion of extra examples, or small language corpora, a component metalexicographers find useful (de Schryver, 2003; Svartvik, 1999) as they provide ample evidence of language use. This will prove most advantageous for encoding tasks, mainly writing, and also for teaching purposes. Four of the EDs feature this component, but to different degrees (indicated in Table 8 by means of multiple plus signs). COBUILD5 stands out with its Wordbank, a 5-million-word representative sample of British and American English, both written and spoken. Likewise, LDOCE4’s Examples Bank window offers a wealth of example sentences (76,000 from other dictionaries published by Longman and over one million examples from the Longman Corpus Network, http://www.longman.com/ldoce/about_cd.html). OALD7 contains 200,000 extra example sentences (http://www.oup.com/elt/catalogue/isbn/6712?cc=es&sk=19676491). These examples have been taken from the entire text of the dictionary, as is the case with MED2’s Example Sentence search. Finally, CALD2 gives just five extra examples, apart from those included in the entry itself, for the most important words, marked with the symbols E(ssential), I(mprover), and A(dvanced). Finally, CALD2, MED2, and LDOCE4 feature a Guided Tour of the dictionary. The first two are particularly effective because their tutorials are actual videos combining animation and narration. OALD7 has a graphic tutorial, and COBUILD5 offers a Help menu. This type of graphical element is in line with our present-day predominantly visual culture nowadays as it may help learners to quickly become familiar with the content and use of the dictionary. In contrast, a typical traditional paper dictionary merely includes an introduction and a brief key to its entries, which proves much less user-friendly than a visual tutorial. In brief, LDOCE4 is the most outstanding dictionary in terms of the usefulness of its extras (Table 8). MED2, CALD2, and OALD7 are also characterized by the inclusion of abundant supplementary materials, and COBUILD5 exceeds the others only with its Wordbank. Table 8. Extras Multi-referencing CD-ROM as collection of reference books Etymological information Phrase and collocations bank Lesson plans for teachers Grammar section Writing assistant Language awareness articles Atlas Weblinks Study pages Language exercises Exam practice Extra examples Guided tour Language Learning & Technology CALD2 + + + + + + + + COBUILD5 + +++++ - LDOCE4 + + + + + + + + +++ + MED2 + + + + + + + ++ + OALD7 + + + + + + ++++ - 33 Alfonso Rizo-Rodríguez Review of Five English Learners' Dictionaries on CD-ROM MULTIMEDIA RESOURCES A distinctive feature of EDs is their integration of various kinds of multimedia resources (see Table 9). The five dictionaries offer both American and British English recorded pronunciation of each entry, as well as pronunciation practice.. In LDOCE4, even example sentences are accompanied by recordings, and users can practice reading them aloud and comparing their pronunciation to that of the original. This ED, like MED2, also features the inclusion of recordings of some musical instruments and of onomatopoeic verbs and nouns denoting types of sounds difficult to define (e.g. bleat, chirp). Moreover, all of the EDs, except OALD7, include an option for "automatic pronunciation replay" or "always play pronunciation/sound," which activates the recorded sound of every word keyed in the search box. These types of multimedia resources add to the usefulness of an electronic dictionary, since exposure to the spoken language is always beneficial for language acquisition (Brown & Yule, 1983). Interestingly, video clips are no longer included in the latest EDs, except in MED2, whose animation videos illustrate hard to define verbs or nouns (e.g. juggle, lob). Earlier works, such as the Longman Interactive English Dictionary (2nd edition, 2000) and the Oxford Advanced Learner’s CD-ROM Dictionary (6th edition, 2000), made use of video clips. This feature might have been eliminated from the majority of the EDs in order to save space on the CD-ROM and to give priority to the type of original extras discussed above, which may prove more informative to users. As far as illustrations or pictures are concerned, images and color also play a prominent role in the EDs examined, except in COBUILD5. LDOCE4 uses illustrations and photographs exclusively to support the definition of certain terms and, hence, they form an integral part of the entries. In CALD2, pictures are presented in an appendix accessed by means of an on-screen button, and they serve to enrich lexical description by grouping illustrations of semantically related terms (for example, in the office). In MED2, illustrations accompany some entries, or alternatively, they can be viewed separately as members of a list. Finally, in OALD7 users have no direct access to pictures, which accompany only some entries in a small window, but illustrations labelled expand open in a larger window depicting items semantically associated (for example, vegetables). MED2 and LDOCE4 are the only EDs whose complete list of illustrations and photographs can be retrieved by means of the advanced search mode. This might be helpful for teaching purposes in order to present semantically related vocabulary. In addition, the pictures in CALD2 and MED2 can be rendered interactive through "hot spots": by holding the cursor over some parts of a drawing, users can activate terms denoting items in a picture. In contrast, in LDOCE4 and OALD7, illustrations include labels denoting specific vocabulary. All in all, multimedia resources are abundant in MED2 and LDOCE4 but not in the other EDs examined. Table 9. Multimedia Resources Recorded pronunciation Recorded example sentences Sound effects Automatic pronunciation replay Animation videos Illustrations appended to entries Illustrations in appendix CALD2 + + + COBUILD5 + + - LDOCE4 + + + + + - MED2 + + + + + + OALD7 + + + CUSTOMIZATION One recurrent demand from metalexicographers is that EDs should allow customization in terms of the lookup aims of particular users (Atkins, 2002; Corris et al., 2000; de Schryver, 2003; Varantola, 2002). Language Learning & Technology 34 Alfonso Rizo-Rodríguez Review of Five English Learners' Dictionaries on CD-ROM The five examined EDs allow some degree of customization (see Table 10). Interactivity, a feature of the advanced search mode, is typical of all the dictionaries except COBUILD5. Users can carry out complex searches according to their individual reference needs. Customization and flexibility are also evident in the layered presentation characteristic of the graphical interface of CALD2, MED2, and LDOCE4. This ensures that users can decide what type of information is of interest to them. Other customization elements are restricted to the display style of entries on the screen and recorded pronunciation. All dictionaries offer very similar alternatives: selection of font size, American or British English pronunciation, and display options (e.g., phonetic transcription, grammar labels, spell check, and quick/full view). MED2 also features an annotation facility, which allows users to add personal notes to any entry. For example, a translation equivalent or a list of synonyms previously retrieved with the advanced search function can be permanently appended to an entry and, afterwards, they can be edited or removed. COBUILD5 allows customization through its "My Dictionary" function, which enables the creation of a personal lexicon. On the whole, the five EDs offer very similar customization options. Table 10. Customization Interactive advanced search On-screen buttons access to specific information Display style options Full / compact entry view Annotation facility Personal lexicon CALD2 + + COBUILD5 - LDOCE4 + + MED2 + + OALD7 + - + - + + + + - + + + - + - SUMMARY Based on the preceding comparison of five EDs several conclusions may be drawn. The electronic versions of these works are superior to their hard-copy counterparts in terms of accessibility and flexibility of information retrieval (enhanced by their graphical user interface), wider macrostructure (in some EDs), more detailed microstructure, thesaurus-like resources, complex search mechanisms, copy and print functions, extra components, multimedia resources, and the degree of customization. Additionally, the comparison shows that all EDs are equally easily accessible to users and that they all provide a large variety of lookup operations. At a more specific level, however, the preceding analysis suggests that MED2 ranks highest because it exhibits the largest number of functions and innovative features. It is closely followed by LDOCE4 and CALD2. OALD7 also possesses a large number of valuable features, while COBUILD5 is the most basic. In particular, MED2, LDOCE4, CALD2, and OALD7 stand out due to their modern graphical interface, addition of entries from other reference books (LDOCE4 and OALD7), supplementary information in their entries, inclusion of abundant thesaurus-like information, powerful search utilities (MED2 and CALD2), excellent copy and print functions (MED2 and CALD2), high number of extra elements, recorded pronunciation and practice of every example sentence (LDOCE4), and layered presentation of content (CALD2, MED2, and LDOCE4). Nevertheless, this comparison has also exposed some weak points in the EDs. For example, the macrostructure of COBUILD5, CALD2, and MED2 merely reproduces that of their hard-copy counterparts; LDOCE4’s dictionary search function proves insufficient for retrieving very specific types of language data (it consists of only three filters); and OALD7’s search function (non-filter based) would benefit from revision and simplification. In addition, all dictionaries, except MED2 and CALD2, which already possess it, should incorporate a flexible, unrestricted copy and print function in order to help users utilize the findings of the advanced search function. Furthermore, a first-rate attribute of LDOCE4 – its Language Learning & Technology 35 Alfonso Rizo-Rodríguez Review of Five English Learners' Dictionaries on CD-ROM recorded pronunciation and sentence practice functions – might well be a "must" in subsequent editions of other EDs as they facilitate exposure to the spoken language. Furthermore, these dictionaries might also look toward including instrumental animated videos featured by MED2 since they facilitate learning of words that defy semantic description. Finally, the EDs might benefit from extending customization features to suit users’ reference needs. These additions would result in more efficient, powerful language tools that meet the increasing demands of learners, teachers, and metalexicographers. Finally, although the macrostructure and microstructure of these electronic works depend to a considerable extent on those of their printed versions, these EDs also possess properties which set them apart from paper dictionaries and which make them very effective. The present survey of ED features has shown that significant moves are being made by lexicographers and publishers to produce versatile, multipurpose electronic dictionaries that clearly surpass their printed editions. APPENDIX Figure 1. Cambridge advanced learner’s dictionary on CD-ROM (2nd Ed., 2005) Language Learning & Technology 36 Alfonso Rizo-Rodríguez Review of Five English Learners' Dictionaries on CD-ROM Figure 2. Collins Cobuild advanced learner’s English dictionary on CD-ROM (5th Ed., 2006) Language Learning & Technology 37 Alfonso Rizo-Rodríguez Review of Five English Learners' Dictionaries on CD-ROM Figure 3. Longman dictionary of contemporary English, writing assistant edition CD-ROM (Updated 4th Ed., 2005) Language Learning & Technology 38 Alfonso Rizo-Rodríguez Review of Five English Learners' Dictionaries on CD-ROM Figure 4. Macmillan English dictionary for advanced learners on CD-ROM (2nd Ed., 2007) Language Learning & Technology 39 Alfonso Rizo-Rodríguez Review of Five English Learners' Dictionaries on CD-ROM Figure 5. Oxford advanced learner’s compass (7th Ed., 2005) NOTES 1. Research leading to this review has been sponsored by the Spanish Ministry of Science and Technology under I+D contract HUM2007-61766/FILO entitled "ADELEX: Assessing and Developing Lexis through New Technologies." ACKNOWLEDGMENTS The author is deeply grateful to Dr. Sigrun Biesenbach-Lucas, Georgetown University, Washington, DC, USA, and Dr. Carmen Pérez-Basanta, University of Granada, Spain, for their insightful comments on a previous version of this review. Language Learning & Technology 40 Alfonso Rizo-Rodríguez Review of Five English Learners' Dictionaries on CD-ROM ABOUT THE REVIEWER Dr. Alfonso Rizo-Rodríguez (Ph.D., University of Granada, Spain) is senior lecturer in English Linguistics at the University of Jaén, Spain. His research interests comprise English grammar and lexicographical theory. His publications include a monograph on English catenative verbs, as well as numerous articles on diverse grammatical aspects and on dictionary use and criticism. Email: [email protected] REFERENCES Atkins, B. T. S. (2002). Bilingual dictionaries: Past, present and future. In M.-H. Corréard (Ed.), Lexicography and natural language processing. A Festschrift in honour of B. T. S. Atkins (pp. 1-29). Stuttgart: Euralex. Brown, G. & Yule, G. (1983). Teaching the spoken language. Cambridge: Cambridge University Press. Corris, M., Manning, C., Poetsch, S., & Simpson, J. (2000). Bilingual dictionaries for Australian languages: User studies on the place of paper and electronic dictionaries. In U. Heid, S. Evert, E. Lehmann, & C. Rohrer (Eds.), Proceedings of the ninth Euralex International Congress, EURALEX 2000 (pp. 169-181). Stuttgart: Universität Stuttgart. De Schryver, G.-M. (2003). Lexicographers’ dreams in the electronic-dictionary age. International Journal of Lexicography, 16(2), 143-199. Herbst, T. (1990). Dictionaries for foreign language teaching: English. In F. Hausmann, O. Reichmann, H. Wiegand, & L. Zgusta (Eds.), Dictionaries. An International Encyclopaedia of Lexicography. Volume 2 (pp. 1379-1385). Berlin: De Gruyter. Leech, G., & Nesi, H. (1999). Moving towards perfection: The learners’ (electronic) dictionary of the future. In T. Herbst and K. Popp (Eds.), The perfect learners’ dictionary (?) (pp. 295-306). Tübingen: Max Niemeyer Verlag. McCorduck, E. (1996). Review article of the Longman interactive English dictionary on CD-ROM. Dictionaries: Journal of the Dictionary Society of North America, 17, 225-235. Nesi, H. (1999). A user’s guide to electronic dictionaries for language learners. International Journal of Lexicography, 12(1), 55–66. Nesi, H. (2000a). Electronic dictionaries in second language vocabulary comprehension and acquisition: The state of the art. In U. Heid, S. Evert, E. Lehmann, & C. Rohrer (Eds.), Proceedings of the ninth Euralex International Congress, EURALEX 2000 (pp. 839-847). Stuttgart: Universität Stuttgart. Nesi, H. (2000b). On screen or in print? Students’ use of a learner’s dictionary on CD-ROM and in book form. In P. Howarth & R. Herington (Eds.), EAP learning technologies (BALEAP Conference Proceedings) (pp. 106-114). Leeds: Leeds University Press. Rizo-Rodríguez, A. (2004). Current lexicographical tools in EFL: Monolingual resources for the advanced learner. Language Teaching 37(1), 29-46. Rizo-Rodríguez, A. (2005). Advanced monolingual learners’ dictionaries of English in book form: A preliminary state-of-the-art survey. In J.-L. Martínez-Dueñas, C. Pérez-Basanta, N. McLaren, & L. Quereda (Eds.), Towards an understanding of the English language: Past, present and future. Studies in honour of Fernando Serrano (pp. 565-580). Granada, Spain: Editorial Universidad de Granada. Seedhouse, P. (1997). Review article of Collins Cobuild on CD-ROM (1995). ReCALL, 9(1), 61-63. Language Learning & Technology 41 Alfonso Rizo-Rodríguez Review of Five English Learners' Dictionaries on CD-ROM Svartvik, J. (1999). Corpora and dictionaries. In T. Herbst & K. Popp (Eds.), The perfect learners’ dictionary (?) (pp. 283-294). Tübingen: Max Niemeyer Verlag. Tono, Y. (2000). On the effects of different types of electronic dictionary interfaces on L2 learners’ reference behaviour in productive/receptive tasks. In U. Heid, S. Evert, E. Lehmann, & C. Rohrer (Eds.), Proceedings of the ninth Euralex International Congress, EURALEX 2000 (pp. 855-861). Stuttgart: Universität Stuttgart. Varantola, K. (2002). Use and usability of dictionaries: Common sense and context sensibility? In M.-H. Corréard (Ed.), Lexicography and natural language processing. A Festschrift in honour of B. T. S. Atkins (pp. 30-44). Stuttgart: Euralex. Language Learning & Technology 42 Language Learning & Technology http://llt.msu.edu/vol12num1/wareodowd/ February 2008, Volume 12, Number 1 pp. 43-63 PEER FEEDBACK ON LANGUAGE FORM IN TELECOLLABORATION Paige D. Ware Southern Methodist University Robert O'Dowd Universidad de Léon, Spain We performed a two-phase, year-long research project that explored the impact of peer feedback on language development. We investigated specifically how and when postsecondary learners of English and Spanish provide corrective feedback on their partners' use of the target language in weekly asynchronous discussions by assigning them to one of two conditions: e-tutoring, in which students were asked to provide peer feedback on any linguistic form they perceived as incorrect; and e-partnering, in which students were not required to provide peer feedback but could do so on their own initiative. We examined the frequency and type of language use by coding the feedback for language-related episodes (Swain & Lapkin, 1998) and for feedback strategies (Ros i Solé & Truman, 2005). The findings indicate that students in both conditions preferred an inclusion of feedback on form as part of their exchange, but such feedback only occurred when explicitly required in the e-tutoring condition. Pedagogical implications include the need to situate peer feedback on form within current models of telecollaboration and to assist students in using feedback strategies such as reformulations, which do not rely on a deep understanding of the target or native language grammar. INTRODUCTION Online communication tools have been taken up eagerly by the foreign language teaching community. An early focus on within-class communication among foreign language students was quickly followed by a second stage of network-based language teaching in the late 1990s in which language students were linked with learners in other contexts to form international partnerships (Kern, 1995, 1996; Tella, 1991; Warschauer, 1997). Goals of such partnerships, often called telecollaboration, include assisting students' linguistic and pragmatic development and intercultural awareness (Belz, 2003; Kern, Ware, & Warschauer, 2004; Thorne, 2006). In recent years, research has explored in greater depth how different configurations of telecollaboration, from real-time chatting to videoconferencing, have impacted students' language development through online interaction with peers using the target language (Bauer, deBenedette, Furstenberg, Levet, & Waryn, 2006; Belz, 2003; Belz & Kinginger, 2003; Belz & Vyatkina, 2005; Dussias, 2006; Kern, 1996; Kinginger, 1998; Kinginger & Belz, 2005; Lee, 2004). A smaller number of studies within this paradigm (Belz, 2006; Lee, 2006; Levy & Kennedy, 2004; Sotillo, 2005) have focused on the value of having students actively reflect on language form for linguistic development in telecollaborative exchanges. We build on this growing research base by reporting on a two-phase, year-long research project that explores the impact of peer feedback on language development. We investigated specifically how and when post-secondary learners of English and Spanish provide corrective feedback on their partners' use of the target language in weekly asynchronous discussions by assigning them to one of two conditions: etutoring, in which students were asked to provide peer feedback on any linguistic form they perceived as incorrect, and e-partnering, in which students were not required to provide peer feedback but could do so on their own initiative. We examined the frequency and type of language use by coding for languagerelated episodes (Swain & Lapkin, 1998) and for feedback strategies (Ros i Solé & Truman, 2005), both of which are discussed in detail in the methods section. Copyright © 2008, ISSN 1094-3501 43 Paige Ware and Robert O’Dowd Peer Feedback on Language Form in Telecollaboration LITERATURE REVIEW Research on language use in telecollaboration has drawn on several areas of applied linguistics research. With this in mind, we review both sociocultural and interactionist interpretations of telecollaborative language learning, and we pay particular attention to how a focus on form has been integrated into online exchanges to date. Sociocognitive and Sociocultural Perspectives Researchers have studied a range of issues in synchronous and asynchronous exchanges, such as intercultural exploration and understanding (Belz, 2003; Furstenberg, Levet, English, & Maillet, 2001; Liaw, 2006; O'Dowd, 2003, 2006), the role of the instructor (Belz & Müller-Hartmann, 2003; MüllerHartmann, 2006; O'Dowd and Ritter, 2006; Ware & Kramsch, 2005), cultural patterns of use (Kramsch & Thorne, 2002; Thorne, 2003), and the influence of socioinstitutional contexts on students' participation patterns and attitudes toward online correspondence (Belz, 2002; Ware, 2005). Much of this research has yielded rich analyses of language development, including the acquisition of pronouns of address (Belz & Kinginger, 2003; Kern, 1996), the development of modality and expressions of appraisal (Belz, 2003), the development of null-overt subject use and gender agreement (Dussias, 2006), and the acquisition of modal particles (Belz & Vyatkina, 2005). Interactionist Perspective Research examining how online interaction can contribute to learners' grammatical competence and syntactic complexity stems from the literature base of task-based learning, focus on form, and negotiation of meaning in second language acquisition. These studies are often based on the application of Long and Robinson's (1998) interaction hypothesis to online environments. This hypothesis proposes that negotiation of meaning in interaction exposes learners to input that is both linguistically and interactionally modified. Such input is expected to draw learners' attention towards grammatical form and to push them to modify their own output. Negotiation of meaning is seen as a natural and automatic process as interlocutors seek to understand and clarify each others' utterances. Studies in the interactionist tradition have tended to focus on synchronous online interaction, for example, MOO's (Multi-User Domain Object Oriented applications) and chats, either between students within the same classroom (Blake, 2000; Pellettieri, 2000; Smith, 2005) or between native speakers and learners of the target language (Dussias, 2006; Kötter, 2003; Lee, 2004, 2006; Tudini, 2003). Lee (2004) demonstrated that native speakers of Spanish assisted non-native speakers in composing their ideas and in improving their grammar, although she found that language proficiency, computer skills, and age also impacted the nature of the interactions. In a later study using the Blackboard virtual learning platform (2006), Lee focused on open-ended and goal-oriented tasks in synchronous interactions between native speakers of Spanish and American students of Spanish as a foreign language. She found that the Spanish native speakers provided mostly recasts and focused mainly on lexical rather than syntactical errors. Tudini (2003) examined Italian language learner interaction in native Italian Web-based chat rooms and found that negotiation sequences in synchronous interaction occurred in over 9% of total turns and that language learners received both implicit and explicit feedback on their language from their native speaker interlocutors. In short, work in the interactionist tradition has shed much-needed light on how real-time written interaction can support language development in online interactions. However, it focuses mainly on interactions involving negotiation of meaning, not on additional ways that students can support one another when attending to form. Focus on Form Focus on form in online interaction is considered important for several reasons. First, Lee (2004) and Levy and Kennedy (2004) have argued that computer-mediated communication should balance fluency and linguistic accuracy. Second, studies of foreign language students in the US have found that students Language Learning & Technology 44 Paige Ware and Robert O’Dowd Peer Feedback on Language Form in Telecollaboration often consider the "real" part of language learning to involve the study of grammar (Chavez, 2002) and that a focus on culture takes away from the primary goals of classroom instruction (Kubota, Austin, & Saito-Abbott, 2003). In a study of telecollaboration by one of the authors (Ware, 2005), many students cited their preference for focusing on language. The noticing of language forms can occur through ongoing interactional support provided during the normal flow of conversation (Foster & Ohta, 2005) and in explicit feedback in electronic tandem (etandem) partnerships (Appel & Mullen, 2000; Brammerts, 1996; O'Rourke, 2005). Foster and Ohta (2005) provide an example of how the cognitivist approach of the interactionist tradition can be combined with a sociocultural lens to explore data on oral negotiated interaction among English and Japanese learners. They found that students helped one another not only through negotiation of episodes that focused on clarifying meaning, but also through assistance in formulating their messages even when a communication breakdown did not occur. This type of interaction draws students' attention to language form by providing opportunities to discuss language choices, to play with language, and to notice the difference between their own linguistic formulations and those of native speakers. Research on the etandem approach focuses on one-on-one partnerships in which learners provide feedback on one another's errors whether or not they impede meaning. These take place either outside of a traditional classroom (Brammerts, 1996) or within a classroom (O'Rourke, 2005). Students can refer to L2 structures and vocabulary that were used earlier by their partners and reuse them in other situations and contexts. More recent work has examined how telecollaboration can help students to actively notice, process, and discuss specific language forms and functions (Belz, 2006; Dussias, 2006; Levy & Kennedy, 2004). For example, Dussias (2006) compared the linguistic gains of U.S. students of Spanish in a treatment group who were each paired in telecollaborative partnerships with students in Spain against the gains of U.S. students in a control group who performed the same tasks with non-native speaking peers. She found greater gains in the treatment group in overt-null subjects, gender agreement, and communicative fluency. Belz (2006) proposed using learner corpus analysis to assist learners in examining their own patterns of error and in tracing their language development. In a study of "stimulated reflection," Levy and Kennedy (2004) examined how teachers used online communication tools to engage their students in reflection on form. Students of Italian engaged in audio-conferencing with various interlocutors including classmates and Italian native speakers. The recordings of the audio interaction and the shared screen content were then analysed together by the teacher and students with a focus on grammar, vocabulary, pronunciation, and register. The sessions served to focus on the process of interaction in the L2 and to encourage learners to reflect on the accuracy and complexity of their target language and on their communication strategies, including social appropriateness. A focus on the social aspects of language use stems from the potential of telecollaboration to provide opportunities for students to see language and culture as two sides of the same coin (Belz, 2003; Furstenberg et al., 2001; Kern, 1996; Thorne, 2003, 2006; Ware & Kramsch, 2005). Therefore, the tasks given to the students during both phases of this study focused on highlighting the link between language and culture and on developing learners' intercultural awareness (see Appendices A and B). Our study contributes to the above research base by examining peer feedback and attention to language form in asynchronous writing. It is theoretically grounded in a sociocultural approach that views language learning as embedded in a particular sociocultural context (Lantolf, 2000). This implies that any study of focus on form in an online intercultural exchange must take into account sociocultural factors such as the attitudes of each set of learners to the culture of their interlocutors and issues of face and communication breakdowns that regularly occur in intercultural interaction. Sociocultural issues identified as particularly relevant in this study included cultural differences in the techniques used by Spanish and North American students to correct their partners and how previous experiences of formal language learning shaped students' attitudes towards the importance of a focus on form in online intercultural exchange. Language Learning & Technology 45 Paige Ware and Robert O’Dowd Peer Feedback on Language Form in Telecollaboration METHODOLOGY Background Information This two-phase study investigated the integration of peer feedback on language into classroom-based adult foreign language learning using qualitative and quantitative methods. We examined the type and frequency of language-related episodes, feedback strategies students used to focus on morphosyntactic forms, and students' attitudes toward the presence or absence of an explicit focus on language in their online interactions. Students were assigned to one of two conditions: • • e-tutoring, in which they were asked to provide corrective feedback to their partners on language errors or, in the absence of errors, to provide suggestions for language improvement such as different wording or increased vocabulary. Students in this condition received training from their teachers in how to provide such feedback and suggestions (see Appendix C). e-partnering, in which students were not explicitly encouraged or trained to provide corrective feedback to their partners. Instead, they were told that they could provide feedback or suggestions if they chose to or if their partners asked them to do so. Research Questions The following questions guided our study: 1) What are the types and frequencies of language-related episodes in each of the two online conditions of e-partnering and e-tutoring? 2) What feedback strategies did participants use when integrating a focus on morphosyntactic form into their online interactions? 3) What were the attitudes of the participants in each condition toward the presence or absence of a focus on language form in their online interactions? Stages and Procedures To answer these questions, the research was conducted in two phases (Table 1). Table 1. Organization of the Two-Phase Study Phase I E-tutoring E-partnering Phase 2 E-tutoring E-partnering 2 U.S. students & 11 Spanish students 2 U.S. students & 11 Spanish students 8 weeks 14 U.S. students & 14 Spanish students 22 U.S. students & 22 Chilean students 8 weeks 8 weeks 10 weeks Phase I Phase I took place during the spring semester of 2006 as a monolingual online exchange in English between 22 EFL students in Spain and 4 post-secondary students in the US. Conducting this pilot phase in only one language allowed us to control for the effect of instructor, syllabus, classroom, and semester as we explored the potential for conducting a larger follow-up study involving more students and instructors. In this first phase, we randomly assigned 22 post-secondary advanced EFL students (ages 19-22) at a university in Spain to either the e-tutoring or e-partnering condition. All of the students were in the same language course conducted by the second author. Their online partners were a cohort of four post- Language Learning & Technology 46 Paige Ware and Robert O’Dowd Peer Feedback on Language Form in Telecollaboration secondary students (ages 19-21) enrolled in a small university in the US. Two of the U.S. students were required to provide weekly feedback to 11 Spanish students in the e-tutoring condition, and two were asked to provide feedback only when solicited by their EFL partners in the e-partnering condition. The four U.S. students met weekly with the first author to discuss the tasks and to review the research protocol, so the pilot phase functioned not as a typical classroom telecollaborative project, but as a small controlled experiment, in which the four U.S. students were responsible for maintaining the distinctions between the control and treatment groups. All students completed the same task cycle (see Appendix A) using a course management system called Moodle, an open source platform similar to commercial course management systems such as Blackboard and WebCT, that allows for data storage, file sharing, and asynchronous and synchronous interaction (Robb, 2004). Phase II Phase II, in the fall of 2006, was a shift from the more tightly controlled design to an implementation phase, in which we examined the presence or absence of peer feedback on form in a condition more typical of bilingual classroom-based telecollaborative projects. The same two conditions were established: e-tutoring and e-partnering. In the e-tutoring condition, 14 students who were enrolled in an advanced Spanish grammar course at a university in the US were matched with 14 students in the second author's Advanced EFL course in Spain. These 28 students were assigned to the e-tutoring condition for eight weeks. In the e-partnering condition, 22 U.S. students enrolled in an advanced Spanish conversation course were paired with 22 students enrolled in an advanced EFL course in Chile. They participated in the e-partnering condition for 10 weeks. The differences in the lengths of the exchanges were due to differing institutional constraints at the three universities. The students in each telecollaborative project were required to write at least 300 words in each language weekly. In both projects, students were placed into pairs (one native English speaker and one native Spanish speaker), and these pairs remained constant for the duration of the exchange. Students in the etutoring condition were allowed to choose among different tasks (see Appendix B), and students in the epartnering condition wrote on themes related to movies they watched as part of their coursework. All students in Phase II communicated in asynchronous interactions on Blackboard, a widely used, licensed, password-protected course management system. Data Collection and Analysis Language Related Episodes The data sources were a database of weekly online transcripts, surveys that provided descriptive information on students' attitudes, and student-produced writing such as language reflection essays and term papers. To answer the research questions related to the frequency and type of corrective feedback and feedback strategies the students used (i.e., questions 1 and 2), language-related episodes (LREs) were used as a unit of analysis. These are described by Swain and Lapkin (1998) as "any part of a dialogue where the students talk about the language they are producing, question their language use, or correct themselves or others" (p. 326). The online written dialogue was coded for any evidence of writing that focused on language use including mechanics, vocabulary, grammar, style, and other types of corrections and feedback. The total number of words written in the LREs was divided by the total number of words to provide the percentage of writing that focused on language in the LREs. The LREs were categorized as three types of feedback: morphosyntactic, lexical, and affective (see Table 2). In addition to these three categories, we further sub-coded the morphosyntactic LREs using a coding scheme of Ros i Solé and Truman (2005). Sub-codes for lexical items and affective feedback were not needed because no apparent patterns emerged within those categories. Feedback in the morphosyntactic LREs, however, was provided in two ways: specific feedback, in which partners provided the correct answer for mistakes or made suggestions for improving style and syntactic complexity, and Language Learning & Technology 47 Paige Ware and Robert O’Dowd Peer Feedback on Language Form in Telecollaboration commentaries, in which partners not only corrected or pointed out errors but also provided extended metalinguistic commentaries justifying the suggested revisions (see Table 3). Table 2. Examples of Coding for LREs LRE Code Morphosyntactic Lexical Affective Examples From Interactional Data a. "We use the word transport as a verb, and transportation is the noun." b. "I can see why you thought it could be used there, but if the two sentences it connected had been about the same subject, then it would have worked, but they are two completely different sentences and should be separate." "Also 'dumb' is like saying she becomes stupid. If that's what you meant, fine, but it may be better understood if you said ‘dumbfounded' or ‘speechless.'" "Anyways, I thought now would be a good opportunity to tell you about some of your English. Overall it sounds very nice and can be read smoothly. There are just a few changes you should make." Table 3. Examples of Coding for Feedback Strategies Specific Feedback Commentaries a. "Instead of 'Forest fires every year devastate north Spain…' you should say 'Forest fires devastate northern Spain every year' (order of words)." b. "Instead of saying 'In add,' say ‘In addition.'" c. "'She changes of topic' should be said like 'she changes the topic' or just 'she changes topic.'" "I don't know if I told you about the trick of using 'FANBOYS' or not .... Adding commas and semicolons in long sentences makes the sentence more understandable and easier to read. This is when you should use commas in a sentence, when you have any of the FANBOYS: For, And, Nor, But, Or, Yet, So." All transcripts in the e-tutoring and e-partnering conditions of both phases were coded using these categories. To analyze the attitudes of the participants to an absence (or presence) of a focus on form (i.e., research question 3), we used traditional qualitative research methods (Bogdan & Biklen, 1998; Erikson, 1986) including interviews and surveys, which were structured around these areas: background information concerning experience with the target language and with technology, preference for task types in the exchange, perceptions of the usefulness of partners' feedback, self-reported increased use of new forms, and level of interest in online exchanges. FINDINGS Frequency and Type of Language-Related Episodes Phase I Analysis of the data from Phase I in which students interacted only in English reveals that a much greater percentage of LREs occurred in the e-tutoring condition, in which the students were asked to provide feedback on their partner's language whether it was solicited or not (see Table 4). This is not surprising as those students had been asked to provide such feedback, while it was optional for students in the epartnering condition. We did not expect, however, so few LREs in the e-partnering condition because student surveys had earlier revealed their preferences for having at least some focus on form. Language Learning & Technology 48 Paige Ware and Robert O’Dowd Peer Feedback on Language Form in Telecollaboration Table 4. Percentage of Total Interactions Related to Language Form in Phase I Condition E-tutoring Total number of words across all interactions 68,015 Total number of words related to LREs 9,159 Percentage of words related to LREs 13.5% E-partnering 53,367 1398 3% Phase II The results of Phase I led us to expect that students in the e-partnering condition of the second phase would most likely not provide or elicit feedback unless explicitly directed to do so by their instructors. Analysis of the data for Phase II confirmed these expectations in that only the students in the e-tutoring condition tended to provide language-related feedback (see Table 5). Again, while this finding is not surprising, note that students in the e-partnering phase also indicated in their final surveys a preference for having a language focus. Given this preference, why they did not actively elicit such language feedback is unclear to us; possible explanations include a real or perceived lack of time, reluctance to switch the focus from fluency and conversation, lack of confidence in knowing what feedback to provide, or discomfort with taking on a role they might see as more fitting for a teacher. Table 5. Percentage of Total Interaction Related to Language Form in Phase II Condition E-tutoring (Spain-US) E-partnering (Chile-US) Total number of words across all interaction Total number of words related to LRE's Percentage of words related to LREs 112,126 16,511 14.7% 171,457 567 0.003% In our analysis of the type of feedback provided (see Table 6), students assigned to the e-tutoring condition of the bilingual exchange in Phase II of our project put a major focus on morphosyntactic LREs and a secondary focus on affective moves such as praise and mitigation. Lexical items received the least emphasis. Table 6. Focus of LREs in E-Tutoring Bilingual Forums Focus Morphosyntactic LREs Number (percentage) of words in English forums 8,110 (77.1%) Number (percentage) of words in Spanish forums 4,075 (68.0%) Lexical LREs 968 (9.2%) 749 (12.5%) Affective LREs 1,441 (13.7%) 1,168 (19.5%) Total 10,519 (100%) 5,992 (100%) While the English forums included a slightly higher focus on morphosyntactic LREs and the Spanish forums a slightly higher focus on lexical LREs, both forums have an overall higher focus on Language Learning & Technology 49 Paige Ware and Robert O’Dowd Peer Feedback on Language Form in Telecollaboration morphosyntactic LREs. The results also show that the students in both conditions produced affective feedback at a higher rate than they produced lexical feedback. Students' Strategies for Focusing on Language Form in Online Discussions Based on the coding categories from Ros i Solé and Truman (2005), the most frequent type of feedback provided by the U.S. students to the Spanish students in Phase I was that of commentaries (provision of metalinguistic explanations). This was also the case in Phase II as can be seen in Table 7. Table 7. Feedback Strategies Used in the Morphosyntactic LREs in Phase II Focus Specific Feedback Commentaries Total Number (percentage) of words in morphosyntactic LREs in Spanish 1,625 (39.9%) Number (percentage) of words in morphosyntactic LREs in English Total number (percentage) of morphosyntactic LREs 7,663 (94.5%) 9,288 (76.2%) 2,450 (60.1%) 447 (5.5%) 2,897 (23.8%) 4,075 (100%) 8,110 (100%) 12,185 (100%) An interesting distinction emerged between the morphosyntactic LREs written by the U.S. students (in English) and those written by the Spanish students (in Spanish). The Spanish students used more metalinguistic commentaries (60.1%) than did the American students (5.5%). When the American students did provide commentaries, the accuracy and depth of their explanations tended to be limited and not quite accurate, as in these two examples: " ... 'is visited' is passive voice, and is generally frowned upon in the English language. Also we say during weekends because it is a period of time." "Many of the verbs in English are followed by 'to' or take the 'ing' ending as you talked about. The 'ing' form can be used in multiple tenses as well, such as: 'I was playing soccer' and 'I am playing soccer.'" Additionally, the students in the US did not seem as well versed in metalinguistic knowledge or terminology as their EFL partners in Spain. When his partner requested feedback on phrasal verbs, for example, this student in the US tried to be helpful but was unsure how to proceed: "I am not totally sure what you mean by phrasal verbs. If you give me a few phrases of phrasal verbs and then your own I can correct them again for you. I just looked phrasal verbs up on the internet, are they verbs like, add up, and act up etc. I haven't taken grammar in a really long time but if you give me an example I can make up a few phrases. I can also make up some incorrect ones and you can try to fix them." In stark contrast to their U.S. peers, however, the Spanish students were much more familiar with metalinguistic terminology and explanations. They provided significantly more commentaries (60.1% of the total), most of which were accurate, even if sometimes only partial, explanations as these examples demonstrate: "Y por fin cuidado con el verbo 'saber' que es irregular: 'sepa' al subjuntivo y 'supieron' al pretérito" [And finally be careful with the verb 'saber' because it is irregular: 'sepa' in the subjunctive and 'supieron' in the preterit.] Language Learning & Technology 50 Paige Ware and Robert O’Dowd Peer Feedback on Language Form in Telecollaboration "Cuando dices 'querréis' sería queréis ya que es el verbo querer en tiempo presente. Yo quiero, tú quieres ... vosotros queréis. Sólo llevaría dos erres en el condicional: querriáis." [When you say 'querréis' it would be 'queries' because it's the verb querer in the present tense. I want, you want. It would only have two rr's in the conditional: querriáis.] The Spanish students' greater familiarity with metalinguistic terminology may be related to their participation in foreign language classes throughout elementary and secondary education. In contrast, students in the US often only take two or three years of language before post-secondary education. While the mismatch in students' access to the language about language did not impede their attempts at providing feedback, more research would be needed to determine if shared terminology and grammatical awareness might enhance the type of feedback provided and the manner in which it could be acted upon. Turning to the code of specific feedback, the American students relied mostly on this strategy (94.5%), whereas the Spanish students used it less frequently (39.9%). Almost all instances of specific feedback took the form of reformulations of their partner's original message, which parallels the findings in Lee's (2006) study on synchronous interaction among learners. In the case of asynchronous interaction, however, students would first restate the original phrasing, then indicate how it might be better expressed. A typical episode, for example, would be initiated by the non-native speaker in the form of a generic request for feedback, which would be coded as an "affective LRE": "I'm looking forward to learning more through these emails with you. Please tell me about anything that doesn't sound quite right to you!"; "Well, I'm sorry if I have mistakes, I would like to hear your suggestions to how the text can be improved." Subsequently, the native speaker would choose several specific areas from the non-native speaker's message on which to provide feedback and then offer specific feedback: "Everything you wrote was really good, but I have a few suggestions. Instead of saying 'To put out the fires more or less 1,200 soldiers have been deployed in Galicia region' you can say 'About 1,200 soldiers have been deployed in Galicia region to put out the fires.'. Instead of 'Forest fires every year devastate north Spain…' you should say 'Forest fires devastate northern Spain ever year' (order of words)." To end the episode, the non-native speaker would either provide a general acknowledgement of the advice or simply request more feedback on the new message, once again coded as an affective LRE ("Thank you for correcting my english [sic] mistakes, it really helps me."). In the lexical LREs, feedback tended to come in two forms, either by providing a definition with examples or by exemplifying the word's use in different contexts: "When you say 'of the taste' you should say 'with the taste.' Also 'dumb' is like saying she becomes stupid. If that's what you meant, fine, but it may be better understood if you said 'dumbfounded' or 'speechless.'" "Cuando dices que vistéis una peli sería el personaje principal ya que no es una persona sino un actor que hace de esa persona." [When you say that you saw a movie, it would be 'the main character' since it's not a 'person' but rather an actor who acts out this person.] Such reformulations, with a secondary focus on vocabulary, were more time-efficient for the students, and they were less likely to lead to inaccurate explanations of grammar. Using these reformulations, their target partners could use the strategy of noticing (Schmidt, 1993) to compare their own original writing against the more "native-sounding" rephrasing. Participant Attitudes Toward Presence or Absence of Focus on Language Form The role and status of grammar in foreign language education among the different groups of learners in this study differed slightly. Spanish students taking an English philology degree at the university in this study tended to attribute considerable importance to the grammar aspects of their language courses Language Learning & Technology 51 Paige Ware and Robert O’Dowd Peer Feedback on Language Form in Telecollaboration despite their open preference for day-to-day class activities based on the development of communicative skills and intercultural awareness. Informal comments by students often gave the impression that while a language was best learned by practicing speaking and listening, the real business of language learning in educational contexts involved the study and mastery of grammatical forms and vocabulary: "Teachers only just do the textbook or give us photocopies. I think there should be more grammar from which to take notes. And apart from that there should be interesting exercises from which we can learn." Clearly reflecting these attitudes to language learning, many of the Spanish students who had participated in online exchanges in the past had complained of the lack of a clear focus on elements of form in their collaborative work. In contrast, students in the United States who were assigned to the e-tutoring condition in which they were asked to provide corrective feedback were initially hesitant to write commentaries about their partner's language use. In the U.S. students' institutional context, online learning is frequently a part of their regular university coursework, and students often participate in student-based discussion boards as part of their out-of-class coursework. These boards are often informal spaces for sharing ideas, and most evaluative feedback remains the role of the course instructor, so the U.S. students' concerns centered mainly on fears of transforming their online conversations into less informal sessions. At the end of Phase I, Spanish students assigned to both the e-tutoring and e-partnering conditions of the exchange reported seeing their participation not only as an opportunity to get to know and understand members of the target culture, but also as a way to improve their English and be exposed to informal English language from native speakers. Many of those students who had been assigned to the e-partnering condition were disappointed when the American partners did not explicitly provide language feedback and concluded the exchange with feelings of frustration: "No, she's too polite [to comment on grammar]. But I prefer it if she does because if they don't correct you, you can't improve. It [participating in an exchange] is useful because you see how language works but it's not enough because you can't improve your writing because they don't say to you what you are doing wrong." Interestingly, in the second phase of our study, surveys distributed at the end of the exchange found no significant differences between the two conditions of e-tutoring and e-partnering in student attitudes toward language feedback (see Table 8). Table 8. Student Attitudes Toward Language Feedback in Phase II Condition* Writing to native speakers should be a part of all language classes. When writing to native-speaking peers, it is important to include a focus on grammar. I prefer my partners to be primarily conversation partners without a strong focus on grammar. E-tutoring E-partnering E-tutoring Disagree (%) 4.3 0 8.7 Neutral (%) 13.0 11.1 26.1 Agree (%) 82.6 88.9 65.2 E-partnering 14.8 14.8 70.4 E-tutoring 65.2 26.1 8.7 E-partnering 51.9 14.8 33.3 *Note. E-tutoring, n = 23; e-partnering, n = 27. Clearly, both groups of students strongly favored writing to native speakers as part of their language classes. Both groups of students also favored including grammar in the exchange. Students' attitudes differed slightly in the degree to which they believed a grammar focus was important. Those students who had participated in the e-tutoring condition tended to favor a stronger focus on grammar than those in the e-partnering condition. Although the survey indicated no clear consensus as to why, we speculate that the Language Learning & Technology 52 Paige Ware and Robert O’Dowd Peer Feedback on Language Form in Telecollaboration e-tutoring group had concrete, positive experiences with the language focus, whereas for the e-partnering group, the question was hypothetical because they had primarily focused on conversational fluency. In summary, students in the e-tutoring condition who did receive feedback on the accuracy of their writing spoke very positively about this feature of the exchange. These students highlighted the difference between focusing on form with their online partners compared to the traditional grammar focus in their contact classes with their teachers. They mentioned, for example, that the corrections they received from their online partners made a greater impact on their learning than normal classroom feedback and that the corrections were experienced in a more personalized and unthreatening manner: "In class you write down notes about grammar and vocabulary and it stays in your notebook. With an exchange partner she corrects and the information stays with you .... You learn more from mistakes in the forums than from reading rules from the blackboard …. Maybe it's more interesting by the net. You are chatting so you are enjoying. If the teacher gives me a corrected essay, I just read it and that's all." "My partner helped me with sentence structure because in his emails I saw how he wrote and I try to learn with his emails and another thing was the vocabulary, because I want to write something and he had already written another words and it was very useful." Students viewed online correction as a more contextualized way of learning about grammar and vocabulary. From the students' perspective, the discussion forums provided them with a springboard for reflecting on language form that differed from the classroom-based style to which they were accustomed, and they appreciated the newer style. DISCUSSION Several findings that are worthy of further discussion and analysis emerge from the data. First, note that the limited focus on feedback (3% in Phase I and 0.003% in Phase II) in the e-partnering conditions of these asynchronous exchanges replicates the findings from similar research on synchronous interactions. In his extensive study of a MOO-based synchronous tandem exchange between students of German and English, for example, Schwienhorst (2000) found that even though students were encouraged to correct their partners' grammatical errors, as was the case with our e-partnering condition, very little evidence of error correction appeared in the transcripts1. The author suggested that this was due to the students perceiving the point of the activity as being primarily one of communicating and establishing relationships with their online partners. Focusing on grammatical corrections was considered of only secondary importance to the learners. Several explanations are possible for the primary focus on morphosyntactic LREs and secondary focus on affective LREs during the exchange. First, because the students had more time to compose their messages in asynchronous forums, they were able to look up vocabulary instead of relying on synchronous negotiation of meaning to clarify unfamiliar terms. With the extra time available to read, interpret, and respond to messages, they were better positioned to infer vocabulary from the larger context of the message. Students might also have understood "grammar" to exclude a focus on vocabulary and thereby focused their attention on morphosyntactical forms, even though they were told to focus on whatever aspects of language they deemed important, including lexical items. Another possibility is that because all of the students were in advanced language courses, they might not have had any immediate difficulty understanding the gist of the messages, thereby eliminating the need to negotiate meaning. The higher proportion of affective LREs than lexical LREs suggests that students in telecollaborative exchanges might not feel comfortable providing corrective feedback (Lee, 2004) and therefore want to mitigate or contextualize their language-related feedback. Students in this study used various ways of talking about the process of focusing on language, including offering praise on one another's use of the Language Learning & Technology 53 Paige Ware and Robert O’Dowd Peer Feedback on Language Form in Telecollaboration target language, mitigating the importance of their language-related comments, and thanking one another for the language-related feedback: "Concerning your grammar, you did a great job in this forum! There are only a very few mistakes that I saw and they were very small." "With this phrase, the only problem is ... " "The sentence is just a little awkward ... " Attempts to engage in grammar correction as sensitively as possible through the use of praise and mitigation strategies were well received by the participants. Belz's (2003) findings were similar in that the American students in her study used positive appraisal with their German partners. Comments from the Spanish students in their interviews and portfolios confirmed that such affective strategies were appreciated and a key factor in the success of the exchange: "I found her very helpful, she was really nice to me and I'm very grateful. The corrections have helped me not to commit so many errors when writing. And I think if a native speaker corrects you, you'll pay more attention since they do it in an informal way." Finally, in relation to students' strategies for focusing on form, it is important to note that individual students were differently equipped to provide accurate feedback. As mentioned earlier, the Spanish students as a general rule used greater metalinguistic terminology and typically provided more substantive feedback than the American students. Even so, the feedback students provided was often considerably less complete than what a trained teacher could provide. In short, the feedback was sometimes very well intentioned but misleading. Pedagogical Implications The findings of this study raise several issues for instructors and researchers interested in exploring an explicit focus on peer feedback on language form in online exchanges. First, our research indicates that language learners do appreciate their partners' active attempts to provide them with individualized feedback. However, even though they favor this aspect of telecollaboration, they do not integrate it into their online interactions unless given explicit directions to do so by the language instructor. To counteract this avoidance of focusing on form, teachers may therefore have to go further than merely encouraging students to correct their partners. Strategies could include dedicating sufficient class time to modeling effective feedback strategies and requiring that parts of students' portfolios or final essays be dedicated to reflecting on how error correction was dealt with during their online interaction. In our study, students claimed that they used the online discussions to notice how their partners used language and then re-used that language themselves later. However, we found little evidence of this reuse within the transcripts themselves, as was also the case in the empirical work in the e-tandem tradition by Little et al. (1999) and Schwienhorst (2000). We speculate that, from a student's perspective, online exchanges are likely "forward-oriented" toward the next message containing new information, unlike, perhaps, teacher-directed class assignments that can be iterative products that are revised multiple times for accuracy (and a grade). Therefore, we would suggest that teachers structure carefully sequenced tasks so that they build on the previous interaction. We have evidence that the feedback provided by peers is often limited in scope or accuracy. The limitations of peers' metalinguistic comments may well be an indication that peer feedback, in the sense of asking students to provide accurate explanations of their native language grammar, may not be an appropriate use of telecollaboration. A more effective frame for peer feedback in telecollaboration could be to request that language learners provide one another with reformulations as they tended to do naturally both in our asynchronous study and in the synchronous one conducted by Lee (2006). In this Language Learning & Technology 54 Paige Ware and Robert O’Dowd Peer Feedback on Language Form in Telecollaboration way, the online forum can serve as an alternative type of language learner reflection journal, in which students document what they notice about the target language. Instructors can use transcripts, or as Belz (2006) suggests, a learner-based corpus stemming from the transcripts, as a starting point for reflecting on language use and form. To give students ample opportunities to reflect on their online interaction and to study new linguistic structures and lexical items, using portfolios and learner diaries as proposed by researchers in the e-tandem tradition (Little et al., 1999) are invaluable. Learner' diaries, for example, can be used by students to maintain an ongoing record of their experiences of the online exchange and to reflect on what they are learning, both culturally and linguistically, from their interaction with their partners. When teachers use portfolios as part of the evaluation process for telecollaboration, students also have an opportunity to show how they have benefited from their exchange using presentations in which they demonstrate their use of the feedback. Limitations and Recommendations for Future Studies Until more studies are undertaken that can replicate our findings with different groups of students across other online learning contexts, our conclusions are limited to this particular context of Spanish and English post-secondary learners. Because the two cohorts of EFL learners were located in Spain and in Chile, there are potential sociocultural factors at the national and institutional levels (e.g., different emphasis placed on language or culture at the tertiary level, different secondary educational experiences with English, etc.) that might have affected the students' interactions. Second, measures of student uptake of and acquisition of particular language forms are needed to determine whether the online feedback has an impact on either language learning or metalinguistic awareness beyond the positive evaluation given to it by the participants in this study. A final limitation is the difficulty of attributing effects to any single factor, such as the use of asynchronous instead of synchronous forums, the type of in-class instruction used, or the assignment to e-tutoring or e-partnering. What this study does provide is rich descriptive data on how peer feedback on form plays out under two types of telecollaboration, those of e-tutoring and epartnering. Opportunities for future research are multiple. As mentioned previously, strong measures are needed to examine if and how specific language forms are taken up and acquired in the short and long term as the result of peer feedback in asynchronous writing. For example, in a recent example of learner uptake in synchronous chat, Smith (2005) cautions that a "diminished role" (p. 33) is possible for uptake in online contexts because he found no relationship between uptake and the acquisition of lexical items. Similar rigorous methods need to be applied to asynchronous contexts and to other aspects of language use such as morphosyntactic complexity. This could be done using a pre- and post-test design targeting specific items or through researcher-derived instruments that monitor the ongoing progress of individuals on items specific to the interactions of each partnership. Research is also needed that continues to explore the role of task type in promoting attention to language form along with intercultural learning (Müller-Hartmann, 2000). To create a greater number of online sequences that involve either negotiation of meaning or peer feedback on language, specific tasks may need to be adopted that enhance the amount of negotiation between partners or reflection on language use (Lee, 2006; Pellettieri, 2000). Finally, more research needs to investigate the extent to which foregrounding a focus on language form might impact the ways in which students establish working relationships with their partners and grapple with intercultural learning online. CONCLUSION Taking into account the quantitative and qualitative findings of our research project, the students clearly favored an integration of language form into their online exchanges, but they were not always equipped with a strong enough understanding of the structure of their native languages to provide quality metalinguistic explanations. Therefore, telecollaborative projects that intend to have a language focus Language Learning & Technology 55 Paige Ware and Robert O’Dowd Peer Feedback on Language Form in Telecollaboration need to borrow both principles and techniques from various models of online exchange, with special emphasis given to the role of the instructor. Instructors must not only make clear their expectations that students provide feedback, but they must also provide examples of when and how to provide feedback. Students will learn how to work with their partners in the second language in a sensitive and efficient way when course instructors provide their students with appropriate training and awareness-raising activities in their contact classes. For this reason, the principle of carefully integrating and linking contact classes with online activities as proposed by intercultural telecollaborative models (e.g., Belz, 2003; Furstenburg et al., 2001; Thorne, 2006) is highly recommended for an approach that integrates peer feedback on language form. NOTES 1. Although asynchronous computer mediated communication (ACMC) and its synchronous equivalent (SCMC) may differ in some ways, these forms of communication share many key characteristics that justify taking into account research findings from both contexts in the area of online foreign language learning research. Both ACMC and SCMC, for example, are text-based forms of communication that provide learners with a level of anonymity that would not usually be possible in face-to-face learning contexts. The fact that both are text-based also means that learners have the opportunity to focus on the written form of their own and their partners' output to a greater extent that they would in oral forms of interaction. This can encourage learners to reflect on accuracy and content, especially when extracts of interactions from either form of CMC are saved, printed, and reflected on by learners and teachers at a later point. Nevertheless, it is important to recognize that the immediate nature of SCMC may lead learners to engage more regularly in negotiation of meaning to resolve misunderstandings that arise in their interactions. In ACMC, learners usually have more time to reflect on their partners' texts and to decide what was meant without actually needing to ask them to clarify or reformulate their ideas. APPENDICES Language Learning & Technology 56 Paige Ware and Robert O’Dowd Peer Feedback on Language Form in Telecollaboration Appendix A. Tasks for Phase I Date Tasks (Task 1) Introductions: 1: February This task involves getting to know your partner and their local culture. You should create background texts on two themes: 1) Personal biographies -- describing who you are (100-150 words) and 2) An introduction to life in your town and university -- taking into account the aspects which be of particular interest to someone from the other culture(400-500 words). Students should (in the following week) respond to their partners' texts, asking questions and making comments on the original posts. Also this month, post your first draft of your film review so your Spanish and American partners can make suggestions for improvements. 2: March (Task 2) Both the United States and Spain have experienced periods of incredible economic growth and social change over the past 15-20 years. Even if you are too young yourself to remember what life was like in your country 15 years ago, the media is constantly reminding us about how life has changed so dramatically. In relation to this, you have two 'sub-tasks' to do this month: Firstly, discuss with your partner how life has changed in your country for young people in recent years. How have young people's lifestyles changed? What are their 'new' interests and hobbies? What are the main worries and problems of your generation? Are young people better off now than 15/20 years ago? (Students should write a minimum of two posts on this part of the task.) (Task 3) Each group will find on our platform a set of graphs and statistics which show developing attitudes of your society to certain topical issues, such as immigration and the death penalty. Describe the graphs to your partner and then compare how these different issues are viewed in the USA and Spain. (Students should write a minimum of two posts on this part of the task.) 3: April Language Learning & Technology (Task 4) In this round you have the opportunity to carry out an ethnographic interview with your partner(s) on the topic of your choice. (This involves two separate interviews: One in which the American student interviews the Spanish partners and a second where the Spaniards interview their American partner.) The two interviews do not need to be based on the same theme. You can choose any topic related to your partner's culture which you are particularly interested in (e.g. the education system in the other country, the issue of immigration etc etc). In each interview you should send three ‘rounds' of questions to your partners finding out about their attitudes and experiences in relation to your theme. Each round of questions should expand on the responses you receive from your partner. 57 Paige Ware and Robert O’Dowd Peer Feedback on Language Form in Telecollaboration Appendix B. Sample Tasks From the E-Tutoring Condition in Phase II Advertisement Adaptation Listening Translation Help Creative Expression Idiomatic Expressions Other Choose an advertisement (for example about Coke or some other product aimed at young people) and write an adaptation of the script in your target language. You can change the content as well as the language style, so that the ad is appropriate for the other culture. Your partners should comment on the language, style, and cultural appropriateness. Suggest that your partner listen to a song or radio station in Spanish or English and use that as a basis for talking about music in your life, in your generation. A useful website: http://www.multilingualbooks.com/online-radio-spanish.html You can also use other websites to download songs or podcasts Choose a short text (song lyrics, article, letter, etc.) in your native language and translate it into your target language. Without seeing the original text, your partner needs to correct the translation to make it as appropriate and "natural-sounding" as possible. Discuss the errors your partner made in their translation and try to explain why it "sounds wrong." Express yourself in a creative genre (poem, song, story) and share it with your partner; have your partner comment on the way you are using Spanish and/or provide you with tips on making it more colloquial, more formal, or more culturally relevant Compose a text (or texts) in which you use at least 5 idiomatic expressions that your partner has asked you to explain and have your partner help you make sure you are using the appropriately useful website: http://www.caslt.org/research/spidiom.htm You and your partner can propose a different activity as one of your four choices. Appendix C. Suggestions for Language-Related Commentaries 1) Distinguish between "global errors" and "local mistakes": Local mistakes are typically small mistakes that language learners make when they are in a hurry. Often, the learners know the rules they are breaking but they are so focused on writing or speaking fluently, that they overlook them.Sometimes they are easy to identify: misspelled words, missing articles, missing accent marks, or the occasional wrong verb tense. In contrast, global errors are identified as sentences or phrases that sound awkward to your nativespeaking ear. 2) Use specific strategies for providing feedback: It is often helpful to use these strategies: * Provide feedback: Look for patterns in the errors and provide feedback. Instead of simply writing in the correct answer for your partner, go back through their text and highlight with a different font all of the errors of a particular type. * Selective correction: It is important to focus on just one or two types of errors per message (for example, focus on verb tenses or on comma usage but not on both at once) * Reformulation: You can rewrite one or two sentences for your partners so they can compare the "native-sounding" version to their own. This is a useful technique! * Give examples: When you explain a grammar rule or a vocabulary word, give multiple examples so your partner has a context for using the new expression. * Ask clarification questions: If you do not understand a particular sentence or think there might be multiple meanings, ask your partner directly what they mean by such-and-such. * Provide "mini-grammar lessons": If you feel comfortable explaining your native language, try giving your partner short lessons. Think of these mini-lessons as teaching patterns and reasons, not necessarily rules. Language Learning & Technology 58 Paige Ware and Robert O’Dowd Peer Feedback on Language Form in Telecollaboration 3) Ask your partners what they would like help with (and specify this for yourself, too): It is often easier to provide feedback when your partner tells you specifically what they would like help with. Here are some sample requests when asking for focused feedback: * Could you please read this and comment on how I'm using the subjunctive? * As you read this, will you write down any more sophisticated vocabulary words that come to mind? I think mine are still very simple. Please ignore accent marks this time! * I'd like to learn more idioms -- as you read this, do any come to mind that I might use? 4) Keep the tone positive: Upbeat comments certainly help encourage your partner to take risks in trying out more complicated and sophisticated target language writing. * Praise specific points; don't just make general comments: Good: "I really like how descriptive you are -- you have such a wide range of vocabulary!" Not-so-good: "Good work!" 5) Don't worry if you don't know how to explain something in your native language: Even language teachers have to look up language explanations some of the time. You can always help out by looking up resources online or by asking your own teacher to explain something to you. * Remember that there are regional and national differences in language. Look at these layers of language as potential areas to explore, not to "correct." * Realize that context often influences grammar or vocabulary choice. Help your partners to obtain a more complex view of how English or Spanish is used by pointing out differences between "registers" of language. ACKNOWLEDGMENTS We are grateful for the funding support of The International Research Foundation for English Language Education (TIRF) and for the invaluable suggestions made by the LLT reviewers. ABOUT THE AUTHORS Paige Ware is an assistant professor at Southern Methodist University. She has a Ph.D. in Language, Literacy, and Culture from the University of California at Berkeley. Her publications include research on the use of new technologies to support adolescent language learners and the integration of telecollaboration into ESL and EFL courses. She currently directs and teaches in Project CONNECT, a program for secondary teachers to work with English language learners. Email: [email protected] Robert O'Dowd teaches EFL and Foreign Language Methodology at the University of León in Spain and has a Ph.D. on the development of intercultural competence through the use of networked technologies. He is currently on the executive committees of both Eurocall and IALIC and has published widely on the themes of on-line foreign language education and on the role of culture in foreign language learning. Email: [email protected] Language Learning & Technology 59 Paige Ware and Robert O’Dowd Peer Feedback on Language Form in Telecollaboration REFERENCES Appel, C., & Mullen, T. (2000). Pedagogical considerations for a web-based tandem language learning environment. Computers and Education, 34, 291-308. Bauer, B., deBenedette, L., Furstenberg, G., Levet, S., & Waryn, S. (2006). The Cultura project. In J. Belz & S. Thorne (Eds.), Internet-mediated intercultural foreign language education (pp. 31-62). Boston: Heinle & Heinle. Belz, J. (2002). Social dimensions of telecollaborative language study. Language Learning & Technology, 6(1), 60-81. Retrieved January 8, 2008, from http://llt.msu.edu/vol6num1/belz/default.html. Belz, J. (2003). Linguistic perspectives on the development of intercultural competence in telecollaboration. Language Learning & Technology, 7(2), 68-117. Retrieved January 8, 2008, from http://llt.msu.edu/vol7num2/belz/default.html. Belz, J. (2006). At the intersection of telecollaboration, learner corpus analysis, and L2 pragmatics: Considerations for language program direction. In J. Belz & S. Thorne (Eds.), Internet-mediated intercultural foreign language education (pp. 207-246). Boston: Heinle & Heinle. Belz, J., & Kinginger, C. (2003). Discourse options and the development of pragmatic competence by classroom learners of German: The case of address forms. Language Learning, 53(4), 591-647. Belz, J. A., & Müller-Hartmann, A. (2003). Teachers as intercultural learners: Negotiating GermanAmerican telecollaboration along the institutional fault line. The Modern Language Journal, 87(1), 71-89. Belz, J., & Vyatkina, N. (2005). Learner corpus analysis and the development of L2 pragmatic competence in networked intercultural language study: The case of German modal particles. Canadian Modern Language Review/Revue canadienne des langues vivantes, 62(1), 17-48. Blake, R. (2000). Computer mediated communication: A window on Spanish L2 interlanguage. Language Learning & Technology, 4(1), 120-136. Retrieved January 8, 2008, from http://llt.msu.edu/vol4num1/default.html. Bogdan, R. C., & Biklen, S. K. (1998). Qualitative research for education: An introduction to theory and methods (3rd ed.). Boston: Allyn & Bacon. Brammerts, H. (1996). Language learning in tandem using the Internet. In M. Warschauer (Ed.), Telecollaboration in foreign language learning (pp. 121-130). Honolulu, HI: University of Hawaii Press. Chavez, M. (2002). We say "culture" and students ask "What?": University students' definitions of foreign language culture. Die Unterrichtspraxis, 35(2), 129-140. Dussias, P. E. (2006). Morphological development in Spanish-American telecollaboration. In J. Belz & S. Thorne (Eds.), Internet-mediated intercultural foreign language education (pp. 121-146). Boston: Heinle & Heinle. Erikson, F. (1986). Qualitative methods in research on teaching. In M. Wittrock (Ed.), Handbook of research on teaching. New York: Macmillan. Foster, P., & Ohta, A. (2005). Negotiation for meaning and peer assistance in second language classrooms. Applied Linguistics, 26(3), 402-430. Furstenberg, G., Levet, S., English, K., & Maillet, K. (2001). Giving a virtual voice to the silent language of culture: The CULTURA project. Language Learning & Technology, 5(1), 55-102. Retrieved January 8, 2008, from http://llt.msu.edu/vol5num1/furstenberg/default.html. Kern, R. (1995). Restructuring classroom interaction with networked computers: Effects on quality and quantity of language production. Modern Language Journal, 79(4), 457-476. Language Learning & Technology 60 Paige Ware and Robert O’Dowd Peer Feedback on Language Form in Telecollaboration Kern, R. (1996). Computer-mediated communication: Using e-mail exchanges to explore personal histories in two cultures. In M. Warschauer (Ed.), Telecollaboration in foreign language learning (pp. 105-120). Honolulu, HI: University of Hawaii Press. Kern, R., Ware, P., & Warschauer, M. (2004). Crossing frontiers: New directions in online pedagogy and research. Annual Review of Applied Linguistics, 24(1), 243-260. Kinginger, C. (1998). Videoconferencing as access to spoken French. Modern Language Journal, 82(4), 502-513. Kinginger, C., & Belz, J. (2005). Socio-cultural perspectives on pragmatic development in foreign language learning: Microgenetic and ontogenetic case studies from telecollaboration and study abroad. Intercultural Pragmatics, 2(4), 369-422. Kötter, M. (2003). Negotiation of meaning and codeswitching in online tandems. Language Learning & Technology, 7(2), 145-172. Retrieved January 8, 2008, from http://llt.msu.edu/vol7num2/kotter/default.html. 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: Routledge. Kubota, R., Austin, T., & Saito-Abbott, Y. (2003). Diversity and inclusion of sociopolitical issues in foreign language classrooms: An exploratory survey. Foreign Language Annals, 36, 12-24. Lantolf, J. P. (2000). "Introducing sociocultural theory." In Sociocultural theory and second language learning (pp. 1-26). Oxford, UK: Oxford University Press. Lee, L. (2004). Learners' perspectives on networked collaborative interaction with native speakers of Spanish in the U.S. Language Learning & Technology, 8(1), 83-100. Retrieved January 8, 2008, from http://llt.msu.edu/vol8num1/lee/default.html. Lee, L. (2006). A study of native and nonnative speakers' feedback and responses in Spanish-American networked collaborative interaction. In J. Belz & S. Thorne (Eds.), Internet-mediated intercultural foreign language education (pp. 147-176). Boston: Heinle & Heinle. Levy, M., & Kennedy, C. (2004). A task-cycling pedagogy using stimulated reflection and audioconferencing in foreign language learning. Language Learning & Technology, 8(2), 50-69. Retrieved January 8, 2008, from http://llt.msu.edu/vol8num2/default.html. Liaw, M-L. (2006). E-learning and the development of intercultural competence. Language Learning & Technology, 10(3), 49-64. Retrieved January 8, 2008, from http://llt.msu.edu/vol10num3/liaw/default.html. Little, D., Ushioda, E., Appel, C., Moran, J., O'Rourke, B., & Schwienhorst, K. (1999). Evaluating tandem language learning by e-mail: Report on a bilateral project. Dublin, Trinity College: CLCS Occasional Paper. Long, M. H., & Robinson, P. (1998). Focus on form: Theory, research and practice. In C. Doughty & J. Williams (Eds.), Focus on form in second language acquisition (pp. 15-41). Cambridge, UK: Cambridge University Press. Müller-Hartmann, A. (2000). The role of tasks in promoting intercultural learning in electronic learning networks. Language Learning & Technology, 4(2), 129-147. Retrieved January 8, 2008, from http://llt.msu.edu/vol4num2/default.html. Language Learning & Technology 61 Paige Ware and Robert O’Dowd Peer Feedback on Language Form in Telecollaboration Müller-Hartmann, A. (2006). Learning how to teach intercultural communicative competence via telecollaboration: A model for language teacher education. In J. Belz & S. Thorne (Eds.), Internetmediated intercultural foreign language education (pp. 63-84). Boston: Heinle & Heinle. O'Dowd, R. (2003). Understanding the "other side": Intercultural learning in a Spanish-English email exchange. Language Learning & Technology, 7(2), 118-144. Retrieved January 8, 2008, from http://llt.msu.edu/vol7num2/odowd/default.html. O'Dowd, R. (2006). Telecollaboration and the Development of Intercultural Communicative Competence. Munich, Germany: Langenscheidt-Longman. O'Dowd, R. and Ritter, M. (2006). Understanding and Working with "Failed Communication" in Telecollaborative Exchanges. CALICO Journal, 61(2), 623-642. O'Rourke, B. (2005). Form-focused interaction in online tandem learning. CALICO Journal, 22(3), 433466. Pellettieri, J. (2000). Negotiation in cyberspace: The role of chatting in the development of grammatical competence in the virtual foreign language classroom. In M. Warschauer & R. Kern (Eds.), Networkbased language teaching: Concepts and practice (pp. 59-86). Cambridge: Cambridge University Press. Robb, T. (2004). Moodle: A virtual learning environment for the rest of us. TESL-EJ, 8(2), 1-8. Ros i Solé, C. & Truman, M. (2005). Feedback in distance language learning: Current practices and new directions. In B. Holmberg, M. Shelley, & C. White (Eds.). Distance education and languages: evolution and change (pp. 72-91). Clevedon, UK: Multilingual Matters. Schmidt, R. (1993). Consciousness, learning, and interlanguage pragmatics. In F. Kasper and S. BlumKulka (Eds.), Interlanguage Pragmatics (pp. 21-42). Oxford, UK: Oxford University Press. Schwienhorst, K. (2000). Virtual reality and learner autonomy in second language acquisition. Unpublished manuscript, Trinity College, Dublin, Ireland. Smith, B. (2005). The relationship between negotiated interaction, learner uptake, and lexical acquisition in task-based computer-mediated communication. TESOL Quarterly, 39(1), 33-58. Sotillo, S. M. (2005). Corrective feedback via Instant Messenger learning activities in NS-NNS and NNSNNS dyads. CALICO Journal, 22(3), 467-496. Swain, M., & Lapkin, S. (1998). Interaction and second language learning: Two adolescent French immersion students working together. Modern Language Journal, 82(3), 320-337. Tella, S. (1991). Introducing international communications networks and electronic mail into foreign language classrooms: A case study in Finnish senior secondary schools. Helsinki, Finland: Yliopistopaino. Thorne, S. (2003). Artifacts and cultures-of-use in intercultural communication. Language Learning & Technology, 7(2), 38-67. Retrieved January 8, 2008, from http://llt.msu.edu/vol7num2/thorne/default.html. Thorne, S. (2006). Pedagogical and praxiological lessons from Internet-mediated intercultural foreign language education research. In J. Belz & S. Thorne (Eds.), Internet-mediated intercultural foreign language education (pp. 2-30). Boston: Heinle & Heinle. Tudini, V. (2003). Using native speakers in chat. Language Learning & Technology, 7(3), 141-159. Retrieved January 8, 2008, from http://llt.msu.edu/vol7num3/default.html. Language Learning & Technology 62 Paige Ware and Robert O’Dowd Peer Feedback on Language Form in Telecollaboration Ware, P. (2005). "Missed communication" in online communication: Tensions in fostering successful online interactions. Language Learning & Technology, 9(2), 64-89. Retrieved January 8, 2008, from http://llt.msu.edu/vol9num2/default.html Ware, P., & Kramsch, C. (2005). Toward an intercultural stance: Teaching German and English through telecollaboration. Modern Language Journal, 89(2), 190-205. Warschauer, M. (1997). Computer-mediated collaborative learning: Theory and practice. Modern Language Journal, 81(4), 470-481. Language Learning & Technology 63 Language Learning & Technology http://llt.msu.edu/vol12num1/kitade/ February 2008, Volume 12, Number 1 pp. 64-84 THE ROLE OF OFFLINE METALANGUAGE TALK IN ASYNCHRONOUS COMPUTER-MEDIATED COMMUNICATION Keiko Kitade Ritsumeikan University, Kyoto In order to demonstrate how learners utilize the text-based asynchronous attributes of the Bulletin Board System, this study explored Japanese-as-a-second-language learners' metalanguage episodes (Swain & Lapkin, 1995, 1998) in offline verbal peer speech and online asynchronous discussions with their Japanese key pals. The findings suggest the crucial role of offline collaborative dialogue, the interactional modes in which the episodes occur, and the unique discourse structure of metalanguage episodes concerning online and offline interactions. A high score on the posttest also suggests the high retention of linguistic knowledge constructed through offline peer dialogue. In the offline mode, the learners were able to collaboratively construct knowledge with peers in the stipulated time, while simultaneously focusing on task content in the online interaction. The retrospective interviews and questionnaires reveal the factors that could affect the benefits of the asynchronous computer-mediated communication medium for language learning. INTRODUCTION Asynchronous computer-mediated communication (ACMC) enables language learners to actively engage in interactions with a wider range of interlocutors because the interactions are both place-independent and time-independent. In addition to the accessibility for learners' engagement in real online communities, the unique interactional features of ACMC are considered to facilitate second language (L2) learning. By reexamining the potential of text-based interactions and the time interval between messages within a sociocultural perspective, this study attempts to investigate learners' behaviors in ACMC activities beyond the period of online interaction. Text-Mediated Interactive Features Studies of both SCMC (Synchronous CMC) and ACMC indicate the significant potential of text-based interaction within a sociocultural perspective, based on the work of Vygotsky (1978). Warschauer (1997) employs this framework in computer-mediated communication (CMC) to stress the role of text-mediation and the context for collaborative learning. From a sociocultural viewpoint, language is one of the semiotic tools that mediate both higher mental functioning and actions. Considering such cognitive and selfregulative functions of language, text is viewed as a "thinking device," since the writer or reader is able to describe and reflect upon its immediate interpretation and extract new meanings on the basis of its written representation (Lotman, 1988). Chang-Wells and Wells (1992) observe children's engagement in writing activities and indicate that text-based activity fosters the development of "literate thinking." Through this engagement, children are required to explicitly posit their arguments, keep their arguments consistent with their own position, consider alternatives and justify them, and carefully evaluate the consequences of their stance. Text-based communication allows learners to store, edit, reevaluate, revise, and perform such activities that may enhance their reflective process. Additionally, CMC's interactive dimension promotes a collaborative context for learning. Vygotsky (1978) claims that in the process of higher cognitive development in an individual, knowledge is first constructed through social interaction and then internalized through private speech. According to this view, learning occurs in collaborative dialogues where learners, with their partners' assistance, are able to bridge the "zone of proximal development" (ZPD)—the gap between the level of development that learners are capable of independently attaining and the level that they can achieve with guidance or Copyright © 2008, ISSN 1094-3501 64 Keiko Kitade The Role of Offline Metalanguage Talk in Asyncrhonous CMC collaborative assistance. Second language acquisition (SLA) studies based on a sociocultural perspective (e.g., Lantolf, 2000; Lantolf & Appel, 1994) agree with the significant role played by collaboration in expert-novice and peer interactions in the L2 learning context. By combining the text-based nature of communication with interactive attributes, CMC may enhance collaborative activities. Kitade (2000) and Darhower (2002) examined the text-based chat interactions of L2 learners in a discussion task. They indicate that the learners in online groups work collaboratively by providing guidance to each other and strategize ways in which to achieve intersubjectivity. Studies in telecollaboration (Belz & Kinginger, 2002; Kinginger, 2000) also suggest that the pragmatic competence of French and German L2 learners develops through collaborative e-mail and chat exchanges with their French/German partners. ACMC provides opportunities for collaborative learning to some extent; however, the potential of collaborative learning in this context is more complex, given the time interval between messages. Time Interval as a Controversial Factor in L2 Learning The time interval between the interactions in ACMC is a controversial aspect in L2 learning. It prevents learners from receiving immediate feedback, which is a key element in collaborative learning. Studies on novice-expert dialogue describe how experts guide novices in task completion by adjusting the task difficulty (Radziszewska & Rogoff, 1991; Wertsch, Minick, & Arns, 1984). Rogoff and Gardner (1984) state that scaffolded assistance enables learners to grasp new task components that novices would be unable to complete without assistance. From this perspective, in order to address the needs of novices, feedback should be provided through dialogue. Describing the procedure of effective assistance in the Zone of Proximal Development (ZPD), Aljaafreh and Lantolf (1994) state, "First, intervention should be graduated. Help provided by a more experienced member in a joint activity is designed to discover the novice's ZPD in order to offer the appropriate level of assistance and to encourage the learner to function at his or her potential level of ability… Second, help should be contingent, meaning that it should be offered only when it is needed, and withdrawn as soon as the novice shows signs of self-control and the ability to function independently" (p. 468). Unlike synchronous interaction, exchanges in ACMC often have significant time delays between messages, reducing the opportunity of providing adjusted assistance. Kitade (2006) suggests that half of the initiation moves (i.e., requests for solving linguistic problems) in e-mail exchanges between learners of the Japanese language and Japanese students are ignored. Stockwell (2004) indicates that in L2 contexts, learners of Japanese in Australia rarely surmount conversational breakdown with their online Japanese partners. Lamy and Goodfellow (1999) and Kitade (2006) suggest that the time intervals between messages in asynchronous conferences and e-mail exchanges may decrease the coherence of the discourse and lessen the pressure on participants to negotiate the meaning of written communication. On the other hand, the positive aspect of ACMC is that its asynchronous nature offers abundant time, which amplifies the abovementioned advantages of text-mediation. Lapadat (2002) emphasizes the similarities between the benefits of ACMC and those of conventional writing by stating that "online participants can and do take time to think, to polish what they say, and edit. Participants in asynchronous conferences produce less in total quantity (e.g., number of words), but their contributions to the discussion tend to be carefully crafted, adapted to the audience, dense with meaning, coherent, and complete" (p. 8). In order to assess the status of the interlocutors' knowledge and to frame their messages, participants in ACMC need to consider the perspectives and metalinguistic sensibilities of others. Lamy and Goodfellow (1999) propose that asynchronous conferences are particularly appropriate for "reflective conversations," in which the learners discuss metalinguistic and L2 learning issues, because of the time flexibility and access to previous texts. In sum, the asynchronous nature of interaction may reduce opportunities for scaffolding in the context of collaborative learning; however, it may enhance the reflective process. An and Frick (2006) examine college students' perceptions of ACMC and report that its biggest advantage in the L2 learning context is Language Learning & Technology 65 Keiko Kitade The Role of Offline Metalanguage Talk in Asyncrhonous CMC the ample time available to the participants to reflect upon and develop ideas. At the same time, they also regard as a shortcoming the lack of immediate feedback. The Collaborative Context in ACMC In-Class Activity In order to amplify the benefits of ACMC and to compensate for its shortcomings, it is necessary to examine an ACMC in-class activity, paying close attention to the learners' total engagement, rather than limiting the attention to their online interaction. Learners engaging in ACMC in-class activities can undertake two types of activities: online interactions and offline interactions with peers, referring to the online texts they are attempting to write or comprehend. According to Wells (1999), combining the advantages of spoken and text-based communication helps expedite a child's learning process. The collaborative activity of "talk about text," where speech and text function interdependently within an activity, enables learners not only to engage in reflective thinking with text-based communication but also to receive assistance from their partners in the collaborative context. By observing children's talk in activities involving reading and composing texts, Wells (1999) discovers that the talk about text activity is successful in providing direct assistance to students when they are restricted by their individual knowledge. The offline peer interaction during ACMC activities emerges in the writing process and may have some functions in common with those in the collaborative talkabout-text activity. Studies on L2 writing activities also explore the potential of collaborative talk about written text during the following activities: revision (de Guerrero & Villamil, 2000; Villamil & de Guerrero, 1998), joint writing of a story (Swain & Lapkin, 1998), and joint reflection with native speakers regarding the revised text (Swain & Lapkin, 2002; Tocalli-Beller & Swain, 2005). Peer collaboration during writing or revision has been recognized as an effective technique for enhancing the writing skills of L2 learners (Cumming, in press; Villamil & de Guerrero, 1998). Several studies discussed below claim that peer dialogue plays a crucial role in L2 learning, particularly when it involves metalanguage talk during writing activities. Collaborative Dialogue as a Medium for Observing Learning in an ACMC Activity As discussed above, dialogic interactions can play a significant role in student learning. Expanding on Vygotsky's original claim about expert-novice dialogic interactions, some studies examine the scaffolding behavior in peer interactions and illustrate how learners are capable of assisting each other in bridging their ZPDs (Brooks & Donato, 1994; Donato, 1994; Ohta, 1995; Platt & Brooks, 1994). These studies employ descriptive analyses to illustrate the learners' behaviors on a moment-to-moment basis and the changes that take place during collaborative dialogue. Vygotsky (1978) perceives learners' mental processes to be dynamic phenomena. Underlining Vygotsky's claim, Wertsch (1985, 1991) suggests that a microgenetic analysis is required to observe the development of such dynamic phenomena. De Guerrero and Villamil (2000) describe a microgenetic approach as "one in which moment-to-moment changes in participants' behavior were noted and examined" (p. 54). In their investigation of peer talk among intermediate ESL learners during the revision of writing, they show that learners provide each other with knowledge about language, and that the opportunity to exteriorize their thoughts allows students to reinforce and reconstruct their knowledge of the target language. Studies by Swain and others (Swain, 2000; Swain, 2006; Swain, Brooks, & Tacalli-Beller, 2002; Swain & Lapkin, 1998) propose that the observation of peer dialogue reveals learners' mental processes. Swain and Lapkin (1998) suggest that verbalization in a collaborative context not only enacts the thoughts constituting the mental process but also makes them observable, since "in a joint problem-solving activity, what normally remains hidden in individually internalized thought may manifest itself in dialogue" (Swain & Lapkin, 1998, p. 321). Language Learning & Technology 66 Keiko Kitade The Role of Offline Metalanguage Talk in Asyncrhonous CMC In order to address the question of how learners solve their linguistic problems and the extent to which scaffolding may impact the knowledge of individual learners, Swain and Lapkin (1995, 1998) highlight the importance of metalinguistic episodes in dialogues. They refer to the episodes as "language related episodes (LREs), which are parts of a dialogue where the students talk about the language they are producing, question their language use, or correct themselves or others" (Swain, 1998, p. 70). Provided below is an example of a metalanguage episode from the offline data used in this study. B is looking for the word ataeru to state Warui eikyo: o ataeru 'have a bad influence' in B1. Then, B and W search for the word by listing candidates: agaru, ageru, ataeru, ataeteageru. B1: Warui eikyo: ga agaru? W2: Ageru? B3: Ageru? Ataeru? W4: Ataeteageru? B5: Ataeru. Ataeru. W6: Ataeru. B7: Un. By identifying the episodes that are related to linguistic aspects and included in a tailor-made test, the dyad-specific posttest may measure how the linguistic issues discussed were dialogically retained by the learners for at least a short period of time. The dyad-specific posttests are created on the basis of the metalanguage episodes, as determined by the audio recordings of the peer dialogue during the performance of the collaborative tasks. Swain (1998, 2000) states that in joint dictogloss tasks, the learners were able to remember the solutions they arrived at with respect to 70–80% of the items in the LREs on the posttest, which was held 7–10 days later. The high scores in the posttest suggest that metalanguage talk in collaborative peer dialogue may be important for L2 learning. In sum, the findings of the analysis of collaborative dialogue that occur during the writing and revising activities indicate the potential of offline interaction in ACMC to serve as a learning opportunity. Moreover, an analysis of the offline interaction may provide a verbal protocol that demonstrates a learner's status on a moment-to-moment basis. However, offline interaction in ACMC may differ from dialogues that emerge during other writing activities. In an ACMC activity, learners are required to comprehend the received online messages and compose text messages that are framed specially for their partners. In addition, they have two types of interlocutors who can provide assistance for both taskoriented and linguistic needs: online partner(s) and offline peers. The incorporation of a descriptive approach should be effective in revealing the learners' actual behaviors beyond the domain of online interaction and the developmental process of learning that occurs in this context. Many previous studies on ACMC have examined only online interactions (e.g., Kinginger, 2000; Kitade, 2006; Lamy & Goodfellow, 1999; Schwienhorst, 2003; Stockwell & Levy, 2001) without addressing the role of offline interactions or the learners' engagement in combined online and offline interactions. In order to fully understand how learners implement a task in the ACMC context and the potential of this task with regard to L2 learning, this study incorporates a sociocultural perspective and examines both online and offline interactions to reveal how each type of interaction—online, offline, or combined interactions—can provide learners with opportunities for collaborative learning. The study investigates learners' metalinguistic talk in online and offline interactions in order to identify the types of knowledge used and to show how they are co-constructed from the two types of interaction. A posttest is also employed to investigate the extent to which learners retain the co-constructed knowledge, at least in the short term. The research questions are as follows: Language Learning & Technology 67 Keiko Kitade The Role of Offline Metalanguage Talk in Asyncrhonous CMC RQ1: What kind of discourse structure do the learners engage in to construct metalinguistic episodes in the ACMC activity with respect to the following: (a) online versus offline interaction and (b) receptive versus productive modes? RQ2: What kinds of metalinguistic episodes are discussed by the learners to convey their intentions to their native key pals in terms of the linguistic focus (lexical-, syntactic-, phonological-, or discoursebased)? RQ3: To what extent do learners retain the knowledge co-constructed with their peers in the metalinguistic episodes ? METHOD Participants In order to examine learners' interactions in a classroom environment rather than in an experimental context, this study was conducted in two content-based Japanese study classes held in two terms: Term 1 (Fall, 2005) and Term 2 (Spring, 2006). In these classes, the learners and their classmates studied and discussed social problems and cultural aspects related to Japan through in-class discussions or using the bulletin board system (BBS). The participants comprised 36 exchange students (Term 1: 8 students; Term 2: 28 students) studying Japanese in half- or one-year language programs at a Japanese college. They were enrolled in the advanced-low level Japanese course, comprising eight classes per week, the contentbased class being held once every week. During the data collection process, the students stayed in Japan for one month, immediately after learning Japanese in their own countries: Korea, China, Taiwan, France, Germany, Italy, Sweden, Denmark, England, the Philippines, Australia, Canada, and the U.S.A. They indicated in the questionnaire that they regularly wrote e-mails in Japanese and had no difficulty typing in Japanese. The 32 Japanese volunteers were all undergraduate students; during the data collection, 10 of them were attending the Japanese language teaching seminar class. Task and Procedures The participants were randomly paired with their classmates and engaged in a decision-making task with one or two Japanese partners; they could interact with their classmates only through the BBS. The participants engaged in the task during 60 minutes of a 90-minute class, which was held once a week for four weeks. To accomplish the task, the Japanese partners were also instructed to hold discussions with the participants through the BBS. During the first week of each term, the participants were given instructions on the use of the BBS and were introduced to the available online dictionaries. The students in both Terms 1 and 2 began comparing educational or job-hunting system in Japan with those of their own countries. In the subsequent three weeks, the participants in Term 1 were instructed to discuss their ideal school with the instruction: "If you were to start a school, what kind of school would you want to establish?" They had to answer this question with respect to the educational objectives, educational system, content covered, educational environment and facilities, and name of the school. The participants in Term 2 discussed their ideal job. They were provided with the following instruction: "When you look for a job, what kinds of conditions do you need?" They had to answer this question with respect to salary, holidays, working hours, interests, stability, and human relations. After discussing and arriving at an agreement in four weeks, all the participants were instructed to write a summary of their responses in Japanese. The Japanese learners were also required to submit a handwritten report about the linguistic and cultural aspects they had learned through the activity. Language Learning & Technology 68 Keiko Kitade The Role of Offline Metalanguage Talk in Asyncrhonous CMC Data Data were obtained from three sources. The first was the interactions conducted during the task with two types of traceable interactions: text-based online interaction through the BBS and audio-recorded offline verbal interaction. A total of 16 recordings, each lasting an average of 59.0 minutes for Term 1, and 39 recordings, each lasting an average of 51.5 minutes for Term 2, were transcribed in order to create the posttest and identify the metalinguistic episodes. The online BBS messages included 60 messages for Term 1 and 186 messages for Term 2. The type of BBS employed for the ACMC activity, termed "zoops," enabled registered participants to engage in discussions using a group thread. Secondly, two research assistants took notes on their observations in the classroom during all sessions in order to capture the learners' nonverbal behavior, including the use of dictionaries, which could not be captured by audio-recorded data. Finally, interviews were conducted and questionnaires were distributed for the purpose of documenting the behavior and perceptions of the Japanese-as-a-second-language participants. The questions focused primarily on three aspects: (a) the learners' behavior while they were reading and writing online messages (i.e., if and in what order they paid attention to the organization, content, or form of the messages); (b) their opinions of the pair work with their classmates; and (c) their impression of the ACMC interactions for language learning, compared to the other modes of interaction. In both terms, a research assistant conducted audio-recorded interviews with students; these lasted for an average of 13 minutes each. The interviews included more open-ended questions that were designed to extract more detailed answers. The audio-recorded offline data were transcribed and the metalinguistic episodes were identified based on the definition of LREs provided by Swain and Lapkin (1998). The discourse structure of the metalinguistic episodes in the ACMC activities was identified and then, the preferred types of metalinguistic episodes were determined with respect to linguistic focus on the metalinguistic episodes and their corresponding solutions. At the beginning, the interrater reliability of two raters—obtained through the identification and categorization of metalinguistic episodes—was found to be 90.6%. However, following the discussion of the items that differentiated the assessments (between the two categories), the disagreements were resolved and the interrater reliability reached 100%. Based on the identified metalinguistic episodes in the online and offline interactions, test items were individually developed for each L2 learning participant for the posttest; this test was administered during the sixth week. Similar to the dyad-specific posttest by Swain and Lapkin (1998), the format of the questions used in this study varied depending on how the test items were originally discussed in the metalinguistic episodes. Some examples of the posttest are provided in Appendix A. FINDINGS Discourse Structure of Metalinguistic Episodes One of the most salient structural features of the metalinguistic episodes was that they were conducted through a combination of online and offline interactions. Figure 1 illustrates the dual interactions in the ACMC activities: the online interaction between a learner and a Japanese student and the offline interaction between Learners A and B, peers. All the metalinguistic episodes were triggered either by a linguistic item in the Japanese partner's online message or a linguistic item that Learner A attempted to write in the online message to his/her Japanese partner. The metalinguistic episodes took place in both online and offline modes. After the learners discussed and agreed upon a certain linguistic form in the offline mode, as shown by the rectangles in Figure 1 (off1, off2, and off3), Learner A replied in an online message to his/her Japanese partner, using the agreed upon linguistic item. The time interval between the online messages enabled the learners to engage in offline peer interaction, while communicating with their online Japanese partners. Language Learning & Technology 69 Keiko Kitade The Role of Offline Metalanguage Talk in Asyncrhonous CMC The metalinguistic episodes in the offline interaction are illustrated in a simplified structure in Figure 1 (offline1, offline2, and offline 3), but the actual exchanges in the data are more complicated and varied, depending on each episode. In one of the more complicated structures, the learners asked the instructor or the teaching assistant to provide assistance when they were unable to solve the problem through discussions with their partners. J On1 L A On5 On3 On2 Off 1 On4 Off 3 Off 1 Off 3 L B Off 2 2 Off 2 Figure 1. Combination of online and offline interactions in the ACMC activity. J: Learner A's Japanese partner; LA: learner A; LB: Learner B (Learner A's peer partner); On: online interaction; Off: offline interaction. The unique functions of repetition in the collaborative dialogue and the evidence of written repetition constitute another significant feature of the metalinguistic episodes. From a sociocultural perspective, Dicamilla and Anton (1997), which examines L2 peer dialogue in a joint writing task, indicates the extensive use of repetition (32% of the total utterances) and demonstrates that repetition plays an essential role in establishing and maintaining intersubjectivity (Rommetveit, 1985) among peers. Repetition enables learners to indicate and maintain a mental space wherein they can confirm their agreement with what has been constructed thus far and add new information to it. The availability of written repetition also shapes the offline dialogue in a manner that differs from regular face-to-face interactions. After or while solving the linguistic problem in the peer dialogue, the learners returned to the online message (on2 in Figure 1) and replied to their online partners using the decided linguistic item. Therefore, the online texts frequently show traceable evidence of not only the learners' transferred knowledge but also the shared information obtained through peer collaboration. Moreover, the written repetition enabled the learners to establish and maintain intersubjectivity, as mentioned above, particularly in contexts in which the learners sat side-by-side and viewed text that was typed by another learner on his/her computer screen. Excerpt 1 illustrates how, when writing a response to their Japanese partners, spoken repetitions were used to collaboratively construct the learners' knowledge in the offline peer dialogue. Excerpt 1 (J & E, Session 3, 16:21–18:35): J and E are peers summarizing the group discussion on a young Japanese individual (Furi:ta:) who makes a living by working a part-time job. In the excerpt, they Language Learning & Technology 70 Keiko Kitade The Role of Offline Metalanguage Talk in Asyncrhonous CMC are discussing and writing the reason why they do not support Furi:ta:. During the discussion, E is typing. In Excerpt 1, Learners J and E are negotiating the choice of the most appropriate particle (case marker) for the sentence they are creating. In Line 3, J is misusing o, an object marker particle; in Line 4, E suggests another particle, ni, a dative case marker particle. However, the verb haratta 'paid' sounds awkward, and in Line 7, J suggests the use of another particle, no. Finally, E suggests ni with the verb kakatta 'cost' instead of haratta. In Line 15, J repeats the complete sentence he uttered in line 1 with the appropriate particle (dative case marker), ni, and the verb, kakatta. The following (partial) notation system was used in the transcripts: (.), (..): pauses [brackets]: The contents within brackets are the transcriber's comments. *asterisk: The words/phrases marked with an asterisk are incorrect. Boldface: Boldface is used to highlight the grammatical aspects under discussion. 1J: Nazeka to yu: to ano (..) kyo:iku kyo: The reason is. Well (..) Education Edu 2E: ((typing)) Kyo:iku ((typing)) ((typing)) Education ((typing)) 3J: Kyo:iku *o haratta okane no imi wa nai (.) kana? The money paid for the education [with the wrong usage of the object marker particle o] would be meaningless (.) I wonder? 4E: Kyo:iku ni ka. For the education [with the correct usage of the dative marker particle ni], is it? 5J: Kyo:iku ni haratta Paid for the education [with the correct usage of the dative marker particle ni] 6E: A a = Ah a = 7J:= No tame no kane da no imi da. tabun no. It means money for the sake of it. It is probably no [with the particle no] 8E: Kyo:iku ni For the education [with the correct usage of the dative marker particle ni] 9J: (Kyo:iku) (Education) 10E: Ah, kyo:iku NI: kakatta okane? Ah, the cost of education? [Emphasis with the correct usage of the dative marker particle ni] Language Learning & Technology 71 Keiko Kitade The Role of Offline Metalanguage Talk in Asyncrhonous CMC 11J: A un kyo:iku Ah, yeah. Education. 12E: Kakatta okane? The cost? 13J: Un, kakatta okane. Yes, the cost. 14E: ((typing)) okane ((typing)) The money 15J: Kyo:iku ni kakatta okane no imi wa nai. The cost for education would be meaningless. [with the correct usage of the dative marker particle ni] 16E: ((typing)) This demonstrates that the learners pay attention to linguistic accuracy as well as the content of the message, and they co-construct the knowledge to produce the most grammatically appropriate sentence. Interestingly, the word kyo:iku 'education' first appears in Line 1 and is repeated in Lines 2, 3, 4, 5, 8, 9, 10, 11, and 15. As J and E search for the appropriate particle following kyo:iku, they repeat the word to indicate the point of agreement, that is, the point at which their knowledge is shared to add new information. In other words, "kyo:iku strategically facilitates the scaffolding by indicating the momentary mental space and producing the correct particle. Repetition also functions as a confirmation check, as is the case in Line 12, and an acceptance, as in Line 13. After collaboratively solving the linguistic problems, J repeats the completed sentence to reconfirm its modified, completed version (Line 15), and E types the sentence. The BBS message typed by E is also confirmed to be identical to sentence J, which is uttered in Line 15. BBS messages, such as those composed during this discussion, provide noticeable written evidence of the knowledge gained through peer dialogue. However, in one case, the learners acquired non-target knowledge through the dialogue. Such an instance implies that learners can co-construct the knowledge gained and reproduce this knowledge in the subsequent text; however, the acquired knowledge may be non-target and, therefore, may require confirmation by experts. Preferred Modes for Metalinguistic Episodes With respect to the preferred mode of metalinguistic episodes, participants clearly selected the offline mode: As shown in Table 1, most of the metalinguistic episodes (Term 1: all episodes; Term 2: all episodes, except two online instances) were discussed during offline verbal interactions rather than online interactions. There are two possible explanations for this finding. As previous studies suggest, learners are reluctant to ask their online partners for help with linguistic matters, due to the less frequent exchanges and lack of instant responses (Kitade, 2006). It is difficult to obtain extensive exchanges with repetitions in asynchronous interactions. Further, the act of soliciting linguistic help from online native partners may be threatening. However, the opportunity for offline collaborative dialogue through synchronous peer dialogue, where learners feel less threatened to ask for linguistic assistance, may avoid these disadvantages of ACMC. The other question regarding the preferred types of metalinguistic episodes is the manner in which these episodes are triggered. In comprehending their online partners' messages and in producing their own messages, learners may face linguistic problems. The data from both Terms 1 and 2 indicate that 91.7% Language Learning & Technology 72 Keiko Kitade The Role of Offline Metalanguage Talk in Asyncrhonous CMC and 93.5% of the metalinguistic episodes, respectively, took place during the productive mode, where the learners were trying to compose messages. More complex cognitive processes are required during the productive mode than during the receptive mode and these processes may promote more metalinguistic episodes. In addition, the use of online dictionaries may reduce the burden of comprehending messages with unfamiliar words. Apart from the metalinguistic episodes in the audio-recorded data, the class observations and learners' interviews suggested that there were more instances in which the learners faced linguistic problems. In these instances, the learners solved the linguistic challenges by consulting online or electronic dictionaries. More possible explanations will be discussed in the following section, which will deal with the learners' perceptions. Table 1. Frequency of Metalinguistic Episodes in Different Modes Online metalinguistic episodes Offline metalinguistic episodes Productive mode Receptive mode Term 1 0 85 78 7 Term 2 2 203 190 10+3* Total (%) 2 (0.6) 288 (99.3) 268(93.0) 20(6.9) Note. Three metalinguistic episodes were triggered when the learners attempted to write a response, but the resources are originally from the online partners' messages. The other noteworthy finding is that the number of metalinguistic episodes among the pairs indicate a variation, as shown in Tables 1-a and 1-b. For instance, pairs 2 and 6 were able to engage in more than 30 metalinguistic episodes; this accounts for five times the number of episodes engaged in by pairs 4 and 15, that is, 6. Thus, the factors affecting the number of metalinguistic episodes in pairs should be investigated using a larger population. Table 1-a. Frequency of Metalinguistic episodes (offline) among pairs -Term1Pair (Gender) Pair 1 (M-F) Pair 2 (F-F) Pair 3 (M-F) Pair 4 (M-F) Total Metalinguistic episodes (%) 22 (25.8) 33 (38.8) 24 (28.2) 6 (7.0) 85 (100) Table 1-b. Frequency of Metalinguistic episodes (offline) among pairs –Term2Pair (Gender) Pair 5 (M-F) Pair 6 (F-F) Pair 7 (M-F) Pair 8 (F-F) Pair 9 (M-F) Pair 10 (F-F) Pair 11 (M-F) Pair 12 (M-F) Pair 13 (M-F) Pair 14 (M-F) Pair 15 (M-F) Pair 16 (M-F) Pair 17 (M-M) Pair 18 (M-M) Total Metalinguistic episodes (%) 15 (7.3) 32 (15.7) 9 (4.4) 13 (6.4) 8 (3.9) 8 (3.9) 14 (6.8) 8 (3.9) 26 (12.8) 29 (14.2) 6 (2.9) 8 (3.9) 15 (7.3) 12 (5.9) 203 (100) Language Learning & Technology 73 Keiko Kitade The Role of Offline Metalanguage Talk in Asyncrhonous CMC Linguistic Focus of Metalinguistic Episodes in the ACMC Activity All the metalinguistic episodes are categorized as lexis-based, form-based, discourse-based, phonologicalor orthographic-based, a combination of lexical and syntactic, or a combination of phonological and lexisbased. The categorization is modified from the classification suggested by Swain and Lapkin (1998, 2002) because some aspects, such as the phonological and orthographic focus, are salient in CMC. In lexis-based metalinguistic episodes, the learners search for and confirm or select the appropriate vocabulary from alternative Japanese vocabulary items. In form-based metalinguistic episodes, the learners address one aspect of Japanese syntax or morphology. The orthography (i.e., spelling and Kanji) and phonological focus (e.g., voiced or voiceless sounds) are categorized in the independent group. However, in discourse-based metalinguistic episodes, the learners focus on aspects such as discourse markers, logical sequencing, stylistics including the degree of politeness, or text structure. Some metalinguistic episodes pertain to more than one linguistic focus and are categorized in the combined groups (see Appendix B for examples.) As shown in Table 2, the results of both Terms 1 and 2 were similar with regard to the linguistic focus of the metalinguistic episodes, although the different topics of the task may have affected the number and types of metalinguistic episodes that occurred. Out of a total of 288 metalinguistic episodes, 142 (49.3%) were lexis-based, 74 (25.6%) were form-based, 36 (12.5%) were phonological and orthographic-based, and 16 (5.5%) were discourse-based. The metalinguistic episodes involving a combination of the lexical and syntactic and the lexical and phonological focus account for less than 5% each. Table 2. Linguistic Focus of Metalinguistic Episodes Lexis Form Phonological & orthographic Discourse Lexical & form Phonological & lexical Total Term1 (%) 43 (50.5) 27 (31.7) 8 (9.4) 5 (5.8) 0 (0) 2 (2.3) 85 (100) Term 2 (%) 99 (48.7) 47 (23.1) 28 (13.7) 11 (5.4) 13 (6.4) 5 (2.4) 203 (100) Total (%) 142 (49.3) 74 (25.6) 36 (12.5) 16 (5.5) 13 (4.5) 7 (2.4) 288(100) Approximately half the metalinguistic episodes had a lexical focus, but the percentage of form-related episodes (approximately 31% in Term 1 and Term 2 with the combination of lexical and form aspects, 6.4%) was quite significant. Due to the availability of both online and electronic dictionaries, many metalinguistic episodes involved more than just vocabulary searching. Most metalinguistic episodes were classified into three conditions. The first is when learners lack confidence about their knowledge or hypothesis and request quick verbal assurance from their peers or the instructor. The second is when learners are unable to choose the appropriate item from those known to them or in the list suggested in the dictionary. The last condition is when the problem encountered by learners is beyond the scope of the dictionary. For example, when a learner seeks an expression to describe a highly abstract concept, he/she would be unable to find a suitable expression even in his/her L1. On these occasions, learners are unable to solve the problem using dictionaries and need to ask their partners or instructors for further assistance. The last two conditions, in particular, often lead learners to engage in more complex metalinguistic explanations (i.e., why one is more appropriate/inappropriate than the other) where learners need to verbalize their moment-to-moment state of knowledge. Excerpt 2 illustrates the metalinguistic episodes in which pair Y and D engages in a dialogue to coconstruct grammatical knowledge. In order to formulate the educational objective suggested by Y and D for their ideal school, Y suggests the use of the expression they have just learned in the other class, A to Language Learning & Technology 74 Keiko Kitade The Role of Offline Metalanguage Talk in Asyncrhonous CMC yu: no wa B no koto de aru [Noun A is to Noun B], as seen in Lines 3, 11, and 13. However, E explains to Y that the use of the expression is inappropriate in the sentence due to the following reasons: (a) the structure suggested by Y contains a verb, but the expression requires a noun, as observed in Lines 14–18 and (b) the expression is suitable for a more general definition, when in reality, they are attempting to pen their opinions, as in Lines 18–22. Y understands D's explanations and suggests a different expression, Verb koto da to omoimasu [we think it is to Verb] in Line 23. Excerpt 2 (Y & D, Session 2): After the discussion on the educational objectives of their ideal school, Y and D begin noting down their ideas. Based on their discussion, D is typing the message. 1Y: Tabun kyo:iku mokuhyo: to yu: no wa (.) The educational objective is to probably (.) 2D: U:n. Kyo:iku mokuhyo: ((typing)) nn. Yeah:. The educational objective ((typing)) mm. 3Y: Tabun to yu: no wa naninani no koto de aru toka kakeba i: ka? (.) un etto: Would it be okay to write something like to yu: no wa naninani no koto de aru?? [a structure to express it is to such and such] (.) mnn well: 4D: Etto: Well: 5Y: Ki-ho-n-te-ki-na The basic 6D: Kihonteki na? ((typing)) The basic (.) un (.) things ((typing)) 7Y: Kihonteki na (.) un. (.) koto-o The basic (.) un (.) things 8D: Koto-o ((typing)) Things [with the objective case marker o'] ((typing)) 9Y: Benkyo: saseru, benkyo: suru? benkyo: saseru? ...saseru kana? sase Benkyo: saseru [Let them study, using causative form for study], benkyo: suru [study]? ...Benkyo: saseru, I wonder? (.) sase. [the use of the causative form (study or let them study)] 10D: Benkyo:suru (benkyo: suru)[study without the causative form] 11Y: *Saseru no koto de aru* (.) *no koto* ((typing)) *Saseru no koto de aru [*it is to let them study]* (.) *no koto ((typing)) [the wrong usage of the structure] 12D: (Demo) (But) 13Y: Demo nanka to yu: no wa naninani no koto de aru desho? kono hyo:gen dakara: Language Learning & Technology 75 Keiko Kitade The Role of Offline Metalanguage Talk in Asyncrhonous CMC But nanka to yu: no wa naninani no koto de aru [a structure something is to such and such], isn't it? Because of this expression: 14D: mnn. Ano:: (..) no no nai ho: ga i: to omo: kedo mnn. well:: (.) but I think it would be better without no [a nominalizer] 15Y: Demo nanka= But somehow= 16D: =Demo saseru wa do:shi de meishi ja nai =But saseru [the causative verb suffix] is a verb, but not a noun. 17Y: Demo etto getsuyo:bi kana? kono iroiro na hyo:gen yatta desho? renshu: shita dakara sono hyo:gen wa naninani to yu: nowa naninani no koto de aru kara. But I wonder if it was Monday? We learned these kinds of expressions, didn't we? We practiced it, and the expression was naninani to yu: nowa naninani no koto de aru [something is to such and such.] 18D: Tabun me:shi, me:shi no ho: ga tekito: to omo: n da ne. sore wa hyakkajiten ja nakute= I think probably a noun; a noun is more appropriate. This is not an encyclopedia, but= 19Y: =Un, so so hai wakatta nanka (.) minna to= =Yeah. Right right. Yes, I understand. Some (.) Everyone and= 20D: =Jijitsu no yo:na =Like a fact 21Y: Un un wakatta hai= Yeah. I got it. Yes = 22D: =Watashitachi no iken =Our opinion. 23Y: Un. so so so ka. da to omoimasu tte kaita ho: ga i: desho? ((typing)) ja saseru koto da to omoimasu. Yeah. Right right right. It would be better to write, da to omoimasu [we think it is], wouldn't it? ((typing)) Well, saseru koto da to omoimasu [we think it is to let them study]. 24D: Ano: ((unintelligible)) un. Well: ((unintelligible)) yeah The feedback provided by D contains a metalinguistic explanation addressing what Y had overlooked. Metalinguistic feedback is claimed to promote a particular type of learners' repair that engages the learners in deeper cognitive processing (Lyster & Ranta, 1997). Furthermore, D's feedback matches Y's requirement because D and Y share the same knowledge of the expression learnt in the same class. Such instances of metalinguistic episodes indicate that if the learners conduct the ACMC task by themselves at home or individually, they will not be able to solve many of the challenges they will encounter. They may miss the opportunities for metalinguistic episodes where learning may take place. Language Learning & Technology 76 Keiko Kitade The Role of Offline Metalanguage Talk in Asyncrhonous CMC Results of the Individually Tailored Posttest The individually tailored posttest was developed on the basis of the metalinguistic episodes identified in the audio-recorded offline dialogue in Term 2. The multiple-choice questions were created to include the choices of both the correct and incorrect items discussed (and not discussed) in the metalinguistic episodes (see Appendix A for examples). The posttest was administered six weeks after the last ACMC session; the learners were told not to use any assistance during this test and that the result would not affect their grades. From the 180 questions developed from the metalinguistic episodes, 132 (73.3%) were correctly resolved during the posttest, as shown in Table 3. This finding confirms the results of dyad-specific posttests by LaPierre (1994) and Swain (1998), which were conducted one week to ten days after the task session and indicate a 70-80% correspondence. Although this study does not ignore the possibility of any change in the effect of language learning subsequent to the metalinguistic episodes and prior to the posttests, it suggests that a high rate of linguistic knowledge constructed through dialogue can remain in memory for a minimum of 6 weeks. The self-reported lexical items that the learners indicated they had learned through the ACMC activities were tested in the same exam sheet; 28 (49.1%) out of 57 were correctly resolved. Compared to the results for the self-reported lexical items, the lexical items in the metalinguistic episodes indicate a higher rate of resolution (68.8%). Interestingly, most newly learned lexical items reported by the learners were originally from the online partner's messages and not the item discussed in the metalinguistic episodes. However, the posttest result demonstrates that the lexical items discussed in the peer dialogue had a higher rate of resolution than those that the learners believed they had learned. The other significant finding in the posttest is the high rate of resolution (79.5%) of syntactic items. Selecting the correct syntactic items in the posttest may not necessarily imply that the learners completely understand the syntactic aspect and are capable of applying it to any given context. However, the data demonstrate that the learners were at least able to choose the correct syntactic item from the alternatives they listed in the metalinguistic episodes and could do so by themselves—something they were unable to accomplish before. Table 3. Posttest Result Lexical Phonological & orthographic Form Discourse Lexical & form Phonological & lexical Total Total number of items in posttest 90 27 44 2a 12 5 180b Items answered correctly 62 20 35 2 8 5 132 Resolution rate (%) 68.8 74.0 79.5 100.0 66.6 100.0 73.3 DISCUSSION Significance of Offline Metalanguage Talk The analysis of the metalinguistic episodes during the offline talk demonstrated the unique discourse structure of such episodes and suggested the possibility of high retention (73.3%) of linguistic knowledge when it was discussed among peers. The high scores on the posttest imply that the opportunities to discuss linguistic aspects with peers and instructors not only reflect the linguistic challenges encountered by the learners but also have the potential to promote longer maintenance of the item in their individual Language Learning & Technology 77 Keiko Kitade The Role of Offline Metalanguage Talk in Asyncrhonous CMC memories. The unique structural features of the metalinguistic episodes also demonstrate opportunities for the learners to enhance their knowledge. One major finding is that the learners' metalanguage talk did not occur in the online interaction with the Japanese speakers but rather in the offline verbal interaction. The retrospective interviews and questionnaires indicate that the learners are highly motivated to interact with online Japanese partners, but preferred the offline mode for metalanguage talk because of the availability of prompt and comprehensible responses. The offline context—in which extensive exchanges with repetitions are available—helps learners to establish and maintain intersubjectivity and obtain graduated and contingent assistance. Unlike offline peer dialogues, asynchronous online interactions lack the exchanges that are needed to create and maintain such discourse. The other explanation for the preference of offline interactions is related to the interlocutors' effect. In offline modes, the assistance provided by peers who share similar background knowledge is more comprehensible. Further, it is less face-threatening to request linguistic help from offline peers than to request assistance from online Japanese partners that the learners have never met. The other crucial feature of the structure of metalinguistic episodes is the written and spoken repetition discussed by the peers. The learners had the opportunity to incorporate the linguistic solution discussed in the offline interaction into the online messages they subsequently wrote to their Japanese partners. The learners' written repetition of what was already discussed in offline metalinguistic dialogues functions not only as a message to the online Japanese partners but also as visualized evidence indicating the intersubjectivity agreed upon by the peers in their offline interaction. By viewing the repeated written words/phrases on the shared computer screen and listening to the spoken repetition, the learners are able to indicate their stance to one another and be acknowledged for it. Further studies examining the role and effect of written repetition may explore the potential of the distinguishable discourse structures of metalinguistic episodes during an ACMC activity. Although ACMC activities are frequently conducted as outside-the-classroom assignments, the findings in this study indicate the significance of the in-class ACMC activity, since this entails the beneficial aspects of offline talk. Although reference to online dictionaries is useful, the learners' retrospective interviews suggest that there are limitations in the scope of these dictionaries. Unlike the receptive mode (reading), which requires only comprehension, the productive mode requires the selection of the correct linguistic knowledge and awareness of how to apply that knowledge in a particular context. Collaborative peer context is able to meet such complicated demands that cannot be solved using dictionaries. Factors Affecting Opportunities for Learning: Pair Work and Task Activities Although most learners indicated in the questionnaires that they took advantage of the allotted time and peers' help between the online asynchronous messages to write more complex and accurate texts, some learners are more self-directed and hesitate to ask for assistance frequently. Such individual differences are also apparent in the learners' perception of pair work. Previous studies based on the sociocultural perspective, particularly in classroom-based research (Foster, 1993; Swain & Lapkin, 1998), suggest that the manner in which the learners perceive, interact, and conduct pair work varies depending on the pair. The responses to the question regarding the perception of pair work indicate that 61% of learners considered it to be helpful, 30.5% perceived no difference between pair and individual work, and 8% experienced difficulties working in a pair. As indicated by previous studies examining pair interactions (Storch, 2002; Storch & Wigglesworth, in press; Swain & Lapkin, 1998), the amount and pattern of metalinguistic episodes observed in each pair and the manner in which the tasks were approached varied. The learners' perceptions of pair work may be related to the congeniality of the two learners and may affect the amount and pattern of metalinguistic episodes in the pair. Learners' reasons for disliking pair work were a preference for an independent learning style and an inability to get along with their partners. The pairs who had more metalinguistic Language Learning & Technology 78 Keiko Kitade The Role of Offline Metalanguage Talk in Asyncrhonous CMC episodes indicated that they enjoyed pair work and gained significant knowledge from their peers in terms of the language and content of the discussion. On the other hand, the peers who had fewer metalinguistic episodes tended to perceive offline peer interactions as an ineffective context for language learning. As some studies suggest (Berg, 1999; Swain, Brooks, & Tocalli-Beller, 2002; Tang & Tithecott, 1999), providing explicit instructions about the rationale of employing peer collaboration and training on collaboration may promote a positive perception and increase collaborative work. The type, complexity, and operationalization of the task also moderate the benefits of offline interaction. The learners were instructed to collaboratively write more than one online message in each session, but how they collaborated to carry out the task (write the messages) varied between the pairs. The metalinguistic episodes seemed to consistently take place in the coauthoring (i.e., joint writing) context, where the pair, using the same computer screen, collaboratively discussed and decided how and with what content they should respond to their online Japanese partners. However, the pairs who discussed ideas and decided who would write what, and then individually wrote a message, tended to have relatively fewer metalinguistic episodes. The nature of the task may vary depending on how the learners construct the task up to the final outcome (Coughlan & Duff, 1994), and a more qualitative analysis examining the operationalization of the writing task in each process (e.g., prewriting discussion, composition, and revision) and the pattern of collaboration (Storch, 2002) should be incorporated in order to address the tasks involving collaborative writing in CMC. In particular, the effect of coauthoring, in which both learners are equally responsible for online messages, may be one area of investigation for future research. Another factor that may reduce the opportunity for learning through peer collaboration is the restrictive nature of peer assistance. In one episode, a pair reached a non-target solution during the metalinguistic talk. In another episode, a pair was unsuccessful in finding the correct grammatical form and instead used an easier, alternative word in its place. This pair was attempting to find a subjunctive form of the word yokereba 'good' (yokereba is conjugated rather uniquely in Japanese). After listing the incorrect forms, the pair agreed instead to employ the more well-known word ok. Most of the episodes in the data indicate that the learners solicited the instructor or the assistant for help when they were unsure about their linguistic solution or unable to arrive at one. Such instances suggest that the availability of assistance from an instructor or expert is crucial during peer collaboration. Methodological Suggestions for Future Studies Unlike other studies that focus on the effect of planning time in experimental settings, this study incorporated a microgenetic approach to examine the learners' actual behaviors in an offline setting in which there is a time stipulation and where peer collaboration occurs naturally. Although offline behaviors are not stored in the scripts and are not as easy to observe as online behaviors, the analysis of offline data reveals some of the learners' actual behavior while executing ACMC tasks, such as collaborative knowledge-buildings during the asynchronous exchanges. The incorporation of audio and visual recordings may capture the non-verbal cues and demonstrate further details of the peer collaborative process. The findings from the observation of the learners' offline behaviors suggest that the planning time of ACMC is not an independent factor. On the other hand, the availability of external resources during the time interval between the messages (dictionaries and metalanguage talk with peers and the instructor) is advantageous for L2 learning. While an experimental study is useful in addressing the general and statistical significance of the effect of planning time on the quality of production, a more naturalistic and microgenetic approach that takes into account the availability of external resources, learners' actual behaviors in executing tasks, and their long-term development should also be considered when studying the effective practical application of ACMC in L2 contexts. Besides the variety in the amount and quality of pair work and the actual activity carried out in a task, it is necessary to improve the methods of measuring the transfer of knowledge obtained in peer collaborative dialogue. This study employed a tailor-made test based on items that were resolved during peer-to-peer Language Learning & Technology 79 Keiko Kitade The Role of Offline Metalanguage Talk in Asyncrhonous CMC dialogue; therefore, it addressed the extent to which the knowledge obtained through collaborative dialogue was individually transferred and maintained. The items in the posttest posed questions in a new context; however, each item was asked in varied contexts in order to examine whether the learner had truly acquired the encountered knowledge item in question and was capable of applying it. A follow-up posttest to study long-term effects should also be useful. Although the tailor-made posttest is a relatively new method that requires some modification and is difficult to apply to larger data, Swain (2000) indicates that this method directly demonstrates the dual role of a language, suggested in the Vygotskian perspective: language as a mediation tool of cognition ("saying as a cognitive activity") and the construction of knowledge that reflects itself ("what they said becomes an outcome of that activity.") At the same time, traditional pre-experimental and postexperimental studies examining the quality of the production (i.e., accuracy, complexity, and fluency) with statistical evidence using a larger data sample should also contribute to the exploration of the effect of collaboration and planning time. A study with a combination of methodologies and varied approaches should suggest the effective pedagogical application of ACMC where learners engage in different modes of interaction. CONCLUSION This study examined the benefits of offline dialogue in an ACMC activity. Since the size of the sample was small, further research considering factors such as the proficiency levels of learners and online partners, the use of the L1 (in the context of foreign language learning), and the goal of the activity is necessary for generalizing the findings. Addressing the preferred type of metalinguistic episodes, the points at which they occur, and their relationship with language learning in an ACMC activity, this study demonstrates how ACMC tasks can be structured to allow learners to take advantage of text-mediated reflective processes that are amplified with sufficient time stipulations and peer collaboration. More so than SCMC, ACMC provides greater access to real interactions with expert speakers (i.e., without the difficulty posed by time differences), particularly for learning intercultural communication and pragmatic competence. However, the asynchronous nature of ACMC can be perceived as unfavorable for L2 learning because it reduces the opportunity for instant and tailored feedback. This study suggests that offline dialogue may compensate for this shortcoming and serve as an occasion for L2 learning and knowledge building. The offline verbal peer dialogue data demonstrates that knowledge of the target language may be collaboratively constructed, and the tailored posttests suggest that learners retain this knowledge for at least a short period of time. The findings related to the role of offline peer dialogue in the ACMC activity suggest a need for the reexamination of CMC and the alternative pedagogical application of ACMC activities. The methodological implication raised for future studies on CMC is the incorporation of a more detailed analysis examining the learners' actual behaviors in carrying out CMC activities. Previous studies on CMC have paid less attention to the role of offline interaction in language learning; however, the potential of offline interactions to create a collaborative context, not only among online interlocutors but also among offline peers, should be investigated in future studies. The collaborative peer relationship enables learners to engage in interactions whereby they deepen their knowledge not only in terms of the content but also linguistic aspects and at a level higher than they would have achieved individually. To effectively incorporate collaborative learning in a CMC context, more pedagogical techniques (e.g., encouraging coauthoring activities, taking careful consideration when matching pairs, guiding the learners in pairing activities) should be carefully considered. Language Learning & Technology 80 Keiko Kitade The Role of Offline Metalanguage Talk in Asyncrhonous CMC APPENDICES Appendix A Examples of Posttest Questions { }の中から一番適切なものを選んでください。Please choose the most appropriate one from { }. Lexical: 仕事を選ぶ時、どんな国や町にあるかという会社の{位置・職場・立場}が大事だ。 Form: 関西弁を使うと、日本人から{わらわせる・わらわれる・わらわられる}かな。 Appendix B Examples of Each Metalinguistic episodes Category <Lexis>: To choose one from two Japanese words for "every": daigaku "goto-ni" vs. daigaku "*tabi-ni" To choose one from two Japanese words for "position": daigaku no "*tachiba" vs. daigaku no "ichi" (for location) <Form>: The potential verb form for "tsukuru (to make)":"*tsuku-rareru" vs. "tsukur-eru" The use of the causative form: Kodomo ni "asonde-hoshii" vs. "asobasete-hoshii" <Discourse>: Inserting "tatoeba" (for example) to create cohesion. "-yo: dearu (sentence final expression)" is too formal and change it to "yo:-desu" <Lexis and form> Searching for both the lexical item(s) and structure together. Discussing to find the expression "he got fired" in Japanese. <Phonological and lexical> To choose one from two phonologically similar words: "seikaku (personality)"or "seikatsu (life)" <Phonological & orthographic> The spelling of "message" in Japanese is "messe:gi," but the learner wrote as "*mesegi." Language Learning & Technology 81 Keiko Kitade The Role of Offline Metalanguage Talk in Asyncrhonous CMC ACKNOWLEDGMENTS This research is supported in part by a Grant-in-Aid for Scientific Research of the Ministry of Education, Culture, Sports, Science and Technology. I would like to acknowledge and thank Maiko Ikeda for transcribing the data, as well as the reviewers and editors of LLT for the helpful comments and feedback. ABOUT THE AUTHOR Keiko Kitade (Ph.D., University of Hawaii at Manoa) is Associate Professor of Japanese at Ritsumeikan University, Japan. She teaches Japanese language, Japanese linguistics, and Japanese language pedagogy. Her research interests are second language (L2) learning, computer-mediated communication (CMC), discourse analysis, and technology and learning. Email: [email protected] REFERENCES Aljaafreh, A., & Lantolf, J. P. (1994). Negative feedback as regulation and second language learning in the zone of proximal development. The Modern Language Journal, 78, 465-487. An, Y.-J., & Frick, T. (2006). Student perceptions of asynchronous computer-mediated communication in face-to-face courses. Journal of Computer-Mediated Communication, 11(2), 5. Retrieved December 27, 2007, from http://jcmc.indiana.edu/vol11/issue2/an.html. Belz, A. J., & Kinginger, C. (2002). The cross-linguistic development of address form use in telecollaborative language learning: Two case studies. The Canadian Modern Language Review, 59(2), 189-214. Berg, E. C. (1999). The effects of trained peer response on ESL students' revision types and writing quality. Journal of Second Language Writing, 8(3) 215-241. Brooks, F. B., & Donato, R. (1994). Vygotskian approaches to understanding foreign language learner discourse during communicative tasks. Hispania, 77, 261-274. Chang-Wells, G.L., & Wells, G. (1992). Constructing knowledge together. Portmouth, NH: Heinemann. Coughlan, P., & Duff, P. (1994). Same task, different activities: Analysis of a second language acquisition task from an activity theory perspective. In J. P. Lantolf and G. Appel (Eds.): Vygotskian approaches to second language research (pp. 183-193). Norwood, NJ: Ablex. Cumming, A. (in press). Writing in the L2 classroom: Issues in research and pedagogy. In R. Manchon (Ed.), International Journal of English Studies, 1, 2 [special issue]. Darhower, M. (2002). Interactional features of synchronous computer-mediated communication in the intermediate L2 class: A sociocultural case study. CALICO Journal, 19(2), 249-277. de Guerrero, M. & Villamil, O. S. (2000). Activating the ZPD: Mutual scaffolding in L2 peer revision. The Modern Language Journal, 84(1), 51-68. Dicamilla, F. J., & Anton, M. (1997). Repetition in the collaborative discourse of L2 learners: A Vygotskian perspective. The Canadian Modern Language Review, 53, 609-633. Donato, R. (1994). Collective scaffolding in second language learning. In J. P. Lantolf & G. Appel (Eds.), Vygotskian approaches to second language research (pp. 33-56). Norwood, NJ: Ablex. Language Learning & Technology 82 Keiko Kitade The Role of Offline Metalanguage Talk in Asyncrhonous CMC Foster, P. (1993). Discoursal outcomes of small group work in an EFL classroom: A look at the interaction of non-native speakers. Thames Valley University Working Papers in English Language Teaching, 2, 1-30. Kinginger, C. (2000). Learning the pragmatics of solidarity in the networked foreign classroom. In J. K. Hall & L. S. Verplaetse (Eds.), Second and foreign language learning through classroom interaction (pp. 23-46). Mahwah, NJ: Lawrence Erlbaum. Kitade, K. (2000). L2 learners' discourse and SLA theories in CMC: Collaborative interaction in internet chat. Computer Assisted Language Learning, 13(2), 143-166. Kitade, K. (2006). The negotiation model in asynchronous computer-mediated communication. Computer Assisted Language Instruction Consortium (CALICO) Journal, 23(2), 319-348 Lamy, M. & Goodfellow, R. (1999). "Reflective conversation" in the virtual language classroom. Language Learning & Technology, 2(2), 43-61. Lantolf, P. J. (Ed.) (2000). Sociocultural theory and second language learning. Oxford: Oxford University Press. Lantolf, J., & Appel, G. (1994). Vygotskian approaches to second language research. Norwood, NJ: Ablex. Lapadat, J. C. (2002). Written interaction: A key component in online learning. Journal of Computer Mediated Communication, 7(4). Retrieved January 31, 2008, from http://jcmc.indiana.edu/vol7/issue4/lapadat.html. LaPierre, D. (1994). Language output in a cooperative learning setting: Determining its effects on second language learning. M.A. thesis, University of Toronto. Lotman, Y. M. (1998). Text within a text. Soviet Psychology, 26(3), 32-51. Lyster, R. & Ranta, L. (1997). Corrective feedback and learner uptake: Negotiation of form in communicative classrooms. Studies in Second Language Acquisition, 19, 37-66. Ohta, A. (1995). Applying sociocultural theory to an analysis of learner discourse: Learner-learner collaborative interaction in the zone of proximal development. Issues in Applied Linguistics, 6(2), 93-121. Platt, E. & Brooks, F. B. (1994). The ‘acquisition rich environment' revised. The Modern Language Journal, 78 (4), 497-511. Radziszewska, B., & Rogoff, B. (1991). Children's guided participation in planning imaginary errands with skilled adult or peer partners. Developmental Psychology, 27, 381-397. Rogoff, B., & Gardner, W. (1984). Adult guidance of cognitive development. In B. Rogoff & J. Lave (Eds.), Everyday cognition: Its development in social context (pp. 95-116). Cambridge, MA: Harvard University Press. Rommetveit, R. (1985). Language acquisition as increasing linguistic structuring of experience and symbolic behavior control. In J. V. Wertch (Ed.), Culture, communication, and cognition (pp. 183-204). Cambridge: Cambridge University Press. Schwienhorst, K. (2003). Learner autonomy and tandem learning: Putting principles into practice in synchronous and asynchronous telecommunications environments. Computer Assisted Language Learning, 16(5), 427-443. Stockwell, G. (2004). Communication breakdown in asynchronous computer-mediated communication (CMC). Australian Language and Literacy Matters, 1(3), 7-31. Language Learning & Technology 83 Keiko Kitade The Role of Offline Metalanguage Talk in Asyncrhonous CMC Stockwell, G., & Levy, M. (2001). Sustainability of e-mail interaction between native speaker and nonnative speakers. Computer Assisted Language Learning, 14(5), 419-442. Storch, N. (2002). Patterns of interaction in ESL pair work. Language Learning, 52(1), 119-158. Storch, N. & G. Wigglesworth. In press. Writing tasks: comparing individual and collaborative writing. In Mar del Pilar Garcia-Mayo (Ed.), Investigating tasks in formal language settings. Clevedon, UK: Multilingual Matters. Swain, M. (1998). Focus on form through conscious reflection. In C. Doughty & J. Williams (Eds.), Focus on form in classroom second language acquisition. Cambridge: Cambridge University Press. Swain, M. (2000). The output hypothesis and beyond: Mediating acquisition through collaborative dialogue. In J. Lantolf (Ed.), Sociocultural theory and second language learning (pp. 97-114). Oxford: Oxford University Press. Swain, M. (2006). Verbal protocols: What does it mean for research to use speaking as a data collection tool? In M. Chalhoub-Deville, C. Chapelle, & P. Duff (Eds.), Inference and generalizability in applied linguistics: Multiple perspectives (pp. 97-113). Amsterdam: John Benjamin Publishing Company. Swain, M., Brooks, L., and Tocalli-Beller, A. (2002). Peer-peer dialogue as a means of second language learning. Annual Review of Applied Linguistics, 22, 171-185. Swain, M., & Lapkin, S. (1995). Problems in output and the cognitive processes they generate: A step towards second language learning. Applied Linguistics, 16(3), 371-391. Swain, M., & Lapkin, S. (1998). Interaction and second language learning: Two adolescent French immersion students working together. The Modern Language Journal, 82(3), 320-337. Swain, M. & Lapkin, S. (2002). Talking it through: Two French immersion learners' response to reformulation. International Journal of Educational Research, 37(3-4), 285-304. Tang, G. M., & Tithecott, J. (1999). Peer response in ESL writing. TESL Canada Journal, 16(2), 20-38. Tocalli-Beller, A. & Swain, M. (2005). Reformation: The cognitive conflict and L2 learning it generates. International Journal of Applied Linguistics, 15(1), 5-28. Villamil, O. S., & de Guerrero, M. C. M. (1998). Assessing the impact of peer revision on L2 writing. Applied Linguistics, 19(4), 491-514. Vygotsky, L. S. (1978). Mind in society: The development of higher psychological processes. Cambridge, MA: Harvard University Press. Warschauer, M. (1997). Computer-mediated collaborative learning: Theory and practice. The Modern Language Journal, 81(3), 470-481. Wells, G. (1999). Dialogic inquiry: Toward a sociocultural practice and theory of education. Cambridge: Cambridge University Press. Wertsch, J. (1985). Vygotsky and the social formation of mind. Cambridge, MA: Harvard University Press. Wertsch, J. (1991). Voices of the mind: A sociocultural approach to mediated action. Cambridge, MA: Harvard University Press. Wertsch, J. V., Minick, N., & Arns, F. J. (1984). The creation of context in joint problem-solving. In B. Rogoff & J. Lave (Eds.), Everyday cognition: Its development in social context (pp. 151-171). Cambridge, MA: Harvard University Press. Language Learning & Technology 84 Language Learning & Technology http://llt.msu.edu/vol12num1/smith/ February 2008, Volume 12, Number 1 pp. 85-103 METHODOLOGICAL HURDLES IN CAPTURING CMC DATA: THE CASE OF THE MISSING SELF-REPAIR Bryan Smith Arizona State University This paper reports on a study of the use of self-repair among learners of German in a taskbased CMC environment. The purpose of the study was two-fold. The first goal sought to establish how potential interpretations of CMC data may be very different depending on the method of data collection and evaluation employed. The second goal was to explicitly examine the nature of CMC self-repair in the task-based foreign language CALL classroom. Paired participants (n=46) engaged in six jigsaw tasks over the course of one university semester via the chat function in Blackboard. Chat data were evaluated first by using only the chat log file and second by examining a video file of the screen capture of the entire interaction. Results show a fundamental difference in the interpretation of the chat interaction which varies as a function of the data collection and evaluation methods employed. The findings also suggest a possible difference in the nature of self-repair across face-to-face and SCMC environments. In view of the results, this paper calls for CALL researchers to abandon the reliance on printed chat log files when attempting to interpret SCMC interactional data. INTRODUCTION Though the field of computer-assisted language learning (CALL) is a rapidly emerging area within applied linguistics, the amount of CALL research on many current SLA topics is relatively modest. Though computer-mediated communication is fundamentally different than face-to-face communication, CALL/SLA studies often do not adequately ground their inquiry in existing "traditional" SLA research. Further, many CALL studies do not make use of existing technology in their data collection and analysis methods, which can severely limit the impact and relevance of their findings. This is unfortunate because CALL can be a powerful vehicle for exploring many of the core elements of current SLA theory. One of these core areas is the role of self-repair in L2 development. The present study explores self-repair in a task-based synchronous computer-mediated communicative (SCMC) environment and employs video screen capture software to evaluate the amount and nature of self-initiated self-repair (SISR) in an SCMC context. In this paper I will first establish the relevance of self-repair to SLA in both traditional and CMC environments. Then, through the discussion of a classroom-based empirical L2 study I will point out how using video screen capture technology yields a markedly different and more precise picture of the nature of SISR in an SCMC environment. Self-repair in SLA Learner self-repair or self-correction has been explored in a variety of educational contexts from various theoretical perspectives and with a focus on both native speakers and second/foreign language learners1. Self-repairs are seen as important from an SLA perspective because they provide us insights into a learner's interlanguage (IL) development. Indeed, self-repair is viewed by many as evidence of noticing an observable behavior from which we can infer that a learner has engaged in some monitoring strategy or has noticed a production error (Kormos, 1999). Self-repair occurs when speakers detect that their output is faulty or inappropriate in some way. The speech flow is halted and a self-correction is executed. Foster & Ohta (2005) define self-correction as "self-initiated, self-repair, [which] occurs when a learner corrects his or her own utterance without being prompted to do so by another person" (p.420). Wouk (2005) makes the distinction between same turn Copyright © 2008, ISSN 1094-3501 85 Bryan Smith Methodological Hurdles in Capturing CMC Data self-repair in which a speaker, in the process of producing an utterance, stops that utterance before completion and continues it in some way that involves alteration of the syntactic structure that is being produced. The speaker may abort the utterance in progress and begin a completely new structure or change the syntactic framework of the utterance, utilizing lexical elements of the old syntax in a new syntactic frame. Buckwalter (2001) notes that the SLA literature normally equates repair with correction. This view precludes any investigation of difficulty that occurs in the absence of observable error such as a learner's preemptive action when anticipating difficulty. Kormos (2000) labels such occurrences as covert self-repair, whereby a learner notices an error prior to articulation and repairs it. Most of the SLA research on self-repair involves the overt variety since the phenomenon of covert self-repair can only be explored in highly controlled experimental settings or through the use of verbal reports or stimulated recall (Gass & Mackey, 2000). Output Self-repair is a type of modified output first argued by Swain (1985) to be key in SLA and is now considered to be a fundamental construct in current SLA theory (Izumi, 2003; Shehadeh, 1999, 2002; Swain & Lapkin, 1995). Modified or "pushed" output refers to corrections or rephrasings that are elicited by the L2 learner's interlocutor. Swain's (1985; 2005) comprehensible output hypothesis argues for the importance of learner output in terms of enhancing the noticing of one's own errors and states that the role of output is, at a minimum, "to provide opportunities for contextualized, meaningful use, to test out hypotheses about the target language, and to move the learner from a purely semantic analysis of the language to a syntactic analysis of it" (p. 252). Swain & Lapkin (1995) argue that when learners produce the target language, external or internal feedback leads them to notice a gap in their existing (IL) knowledge. This noticing pushes them to consciously reprocess their utterances to produce modified output. Research also suggests that learning depends partly on learners' ability to focus on form when they notice such a gap in their IL and also on the extent to which noticing is learner-initiated (Doughty & Williams, 1998; Long & Robinson, 1998). The benefits of pushed output have also been discussed in terms of learner collaboration, which results in language related episodes "where students reflect consciously on the language they are producing" (Swain 2001:53). Though much of the interactionist research to date suggest that language related episodes are often triggered by lexis, there is also evidence that a great deal of learner collaboration is related to form (Ohta, 2001; Swain, 2001). Though technically different from pushed output, self-initiated self-repairs are functionally similar to pushed output in that they serve to test hypotheses about the target language, trigger creative solutions to problems, and expand the learner's existing resources (Kormos, 1999). Self-repair, therefore, occupies an important position in SLA theory. Research on Self–repair Self-correction data have been collected through a variety of means in L2 research, including picture description, spatial description, interviews, storytelling, open narration, and information gap activities (Camps, 2003; Fathman, 1980; Kormos, 2000; Lennon, 1990; van Hest, 1996; Verhoeven, 1989). Generally speaking, the research to date on self-repair suggests that language learners tend to prefer selfover other-repair (Buckwalter, 2001; Foster & Ohta, 2005; van Lier, 1988); that L2 speakers self-repair more often than native speakers (Kormos, 2000; van Hest, 1996); and that self-repair more often leads to modified output than does other-initiated repair. Researchers have also explored various types of selfrepairs such as repairs of the message conveyed, repairs in the manner of expression (or appropriateness repairs), and error repairs, which have included lexical, phonetic, and grammatical repairs. Out of this work emerge trends in the type of SISR that occurs during learner interaction as well as several key variables that affect the nature and amount of self-repair, including learner preferences, developmental factors, and task type. Language Learning & Technology 86 Bryan Smith Methodological Hurdles in Capturing CMC Data Types of self-repair In most of the SLA research on self-repair, researchers have generally concentrated on the focus and structure of self-repair. Kormos (2000) found that lexical errors were repaired considerably more frequently than grammatical errors in the L2. This is in line with a general assumption by researchers of L2 production that L2 learners pay considerably more attention to lexical appropriateness than to grammatical accuracy. Kormos also found that the distribution of self-repairs shows no considerable difference between L1 and L2 data in the frequency of the various types of self-repairs. In contrast, Lennon (1990) found relatively little self-correction among his university level L1 German L2 English speakers, though when they did self-correct they most often focused on lexical items. Fathman (1980) found that 50% of all repairs were lexical in nature. Van Hest (1996) discusses self-repair structure in terms of three components: 1) a reparandum, which is either an error or an inappropriate expression; 2) an editing phase, which occurs immediately following the interruption of the flow of speech; and 3) a reparatum, which is the actual correction or change of the problematic item. Much of this work is based on research from L1 psycholinguistics and Levelt's (1983) repair classification system (Kormos, 2000; van Hest, 1996). This work makes a distinction between overt and covert repair. From this perspective, covert repairs (those made before articulation) proceed the same way as overt repairs. One must infer covert repair through indirect evidence such as word or phrase repetitions, syllabic repetition, silent pauses, etc. (Postma & Kolk, 1992). Van Hest proposes the model below for classifying overt self-repair. The present study adapts this model for classifying instances of SISR. The reasons for choosing the model proposed in van Hest (1996) for the current study were that this model is based largely on Levelt (1983), which has been applied widely in the in the literature on self-repair (Kormos, 2000). Further, this model is quite systematic and reliable, emerging out of a corpus of almost 5,000 self-repairs produced by Dutch speakers in their L1 and L2 across various task types. Finally, adapting such a strong existing taxonomy allows for more powerful comparisons of results across related studies. Overt Self-Repairs • Error repair (E-repair): Those repairs made because the speaker has made an error. • Appropriateness repair (A-repair): Those repairs made because the speaker thinks the original message is inappropriate in some way. For example, a message may be perceived as not having been specific enough. • Different repair (D-repair): Those repairs in which the speaker interrupts his current message to introduce a new, totally different topic. • Rest repair (R-repair): All other types of overt self-repair. Covert Self-Repairs • Those cases whereby the speaker discovers imminent trouble in his/her message and “interrupts” him/herself before the troublesome item is uttered. In this work van Hest (1996) found that appropriateness repairs accounted for 39.7% of self-repairs, with error repairs making up 22.4%, different repairs 10.1%, with 12.3% of self-repairs remaining unclassified. Covert repairs made up 15.5% of all L2 self-corrections. Interestingly, Levelt's model does not explain where errors of morphology should go, though some researchers have collapsed syntactic and morphological errors together in a broader category of "grammatical" errors (Lai & Zhao, 2006). Developmental factors in self-repair Both age and L2 proficiency level seem to affect the amount and nature of self-repair (Camps, 2003; Fathman 1980; Kormos, 1999; van Hest, 1996; Verhoeven, 1989). Of these factors, proficiency-related variables in self-repair are particularly relevant to the current study. Camps (2003) suggests that learners who make a large number of errors possess a more limited knowledge of the target language, and Language Learning & Technology 87 Bryan Smith Methodological Hurdles in Capturing CMC Data therefore are not as well prepared to notice errors and correct them. They may be unaware of errors because either they do not know what a correct form would look like or they are too busy attending to other elements in their production (like finding suitable lexical items to express their ideas). In contrast, van Hest (1996) found that advanced learners correct themselves less frequently than lower level learners. Additionally, beginning and intermediate L2 speakers produced significantly more selfrepairs of lexical errors and significantly fewer repairs of lexical appropriateness than the advanced L2 speakers. Likewise, Kormos (1999) found that participants at a higher level of proficiency self-corrected linguistic errors significantly less frequently than learners at the pre-intermediate level, whereas they repaired the appropriateness of informational content more frequently than pre-intermediate students did. Task type as a variable in self-repair L2 self-repair research also reveals that the frequency of repairs concerning the information content of the message varies across tasks (Poulisse, 1997; van Hest, 1996). Van Hest, for example, suggests an effect for task type, concluding that tasks requiring more precise expression will result in more appropriateness repairs than in tasks with less-rigid structure. Kormos (2000) also suggests that the frequency of appropriateness repairs both in L1 and L2 is affected by task characteristics and the situational variables of the interaction. Self-Repair in a CALL Context Self-repair has been accepted by many as evidence of noticing (Lai & Zhao, 2006), which has been argued to be fundamental to the SLA process (Schmidt, 1993). Text-based chat has been argued to be a good venue for exploring self-repair, since it seems to provide an increase in processing time and opportunity for learners to focus on form (Pellettieri, 1999; Shehadeh, 2001; Smith, 2004), which may lead to a heightened potential for noticing one's own errors. Indeed, Yuan (2003) suggests that the nature of SCMC requires learners to attend to both linguistic forms as well as the meaning of their communication. The printed text may also add to the salience of input and output in general and the noticing of non-targetlike input and output in particular (Izumi, 2002; Salaberry, 2000; Smith, 2004). Smith has also argued that a heightened saliency of linguistic input and output is a favorable byproduct of the CMC interface, with increased saliency due largely to the permanence of the message. This notion of permanence has also been used to explain the lack of learner uptake in a synchronous computer-mediated communication (SCMC) environment (Smith, 2005). Kitade (2000) suggests that internet chat provides opportunities for learners to self correct both grammatical and pragmatic errors in their own linguistic output for essentially two reasons: first, there is no turn-taking competition and, second, there is more time for things like selfmonitoring. Also, there are few paralinguistic cues available in text-based chat, which might reduce the sense of urgency to respond, and this, in turn, might facilitate learners' ability to monitor their language output more closely. SCMC texts are not ephemeral like oral/aural input and learners can scroll up/down to access an earlier message quite easily. Whether they in fact do this is an empirical question. Taken together, these features may positively influence a learner's ability to notice and subsequently correct nontarget like language (Lai & Zhao, 2006; Smith & Gorsuch, 2004). Recent studies on CMC self repair Studies of self-repair in a CMC context are few. Those that do exist tend not to build on the existing work on self-repair (from the non-CMC applied linguistics literature) and take a limited view of the types of self-repair investigated. These studies are often stifled by a failure to employ existing technology in the data collection and evaluation phases of the research. Jepson (2005) found that though both voice and text chats contained various types of repair moves, selfcorrection was not among them. He suggests that self-correction in an SCMC context may be rare Language Learning & Technology 88 Bryan Smith Methodological Hurdles in Capturing CMC Data because speakers do not notice their errors or because the non-native speaker – non-native speaker SCMC context is not conducive to self-correction. Lee (2002) considered self-corrections in the framework of a broader study on the nature of modification devices in a CMC environment and categorized them as belonging to one of two categories, lexical or grammatical. Though no statistical data were provided, Lee reports that most of the self-corrections made by her intermediate-level L2 learners of Spanish were made on the "concordance of gender and number" (p. 284), and only occasionally was incorrect usage of lexical items recognized and self-corrected. She suggests that learner keyboarding skills, language proficiency, and attention to linguistic aspects might contribute to a high number of errors in (written) production. Yuan (2003) found that of 512 errors in his CMC data, only 44 or 8.59% of them were self-repaired. Over 43% of the self-repairs were grammatical in nature (sentence structure, agreement, noun/article, and preposition), though errors of nouns/articles and prepositions were only very rarely self-corrected. Errors of verb tense, modals, and adjective-noun sequences were never self-corrected. Lexical self-repairs accounted for almost 30% of the self-repairs. Yuan also counted self-repairs of spelling errors (25%). Most notable is the fact that although learners made 57 verb tense errors (exactly as many as sentence structure errors), none of these errors were self-corrected. Yuan argues that evaluating such CMC chat logs allows one the advantage of seeing certain processes that the learner undergoes while trying to construct meaning in their L2, arguing that they provide "real, recorded examples of errors (repaired or unrepaired) learners made while trying to achieve certain communicative goals" (p. 204). One of the very few studies to employ screen capture technology in an SLA/CALL study is Lai & Zhao's (2006) study which examines the capacity of text-based chat to promote learners' noticing of their own problematic production. In this study instances of learner self-correction are viewed as evidence of noticing. Lai & Zhao found that online chat was superior to face-to-face interaction for promoting noticing of mistakes even after controlling for differences in the amount of language output produced in each condition. As we can see from these few studies, using printed chat logs in the evaluation of SCMC data is the methodological "industry standard" (with the notable exception of Lai & Zhao). Though chat logs certainly do have value for interpreting SCMC interaction, they fail to capture a significant portion of the data. It is precisely these "missing data" that may provide the most insight into the potential roles of monitoring, attention and noticing, and pushed output in interlanguage development within a CMC context. In order to gain a more complete view of learner CMC interaction, especially that which involves learner self-repair, use of a dynamic screen capture record is required. Relying on a static artifact to make claims about a dynamic process requires an uncomfortably wide and unnecessary leap of faith. For example, Jepson (2005) reports that there were absolutely no self-correction moves in his data, claiming that the SCMC environment is perhaps not conducive to self-correction. This seems unlikely since there is ample research that strongly suggests a heightened degree of attention to form in a CMC context. More likely is the possibility that significant self-repair did occur, but the data collection methodology employed was not sensitive enough to detect it. Jepson (2005) suggests employing technology that records each keystroke in an effort to uncover what he calls "hidden" self-correction. He comments that in his study it was not possible to observe if participants edited their own messages before they sent them and acknowledges that some self-correction repair moves may not have been measurable. This is an important point of which some CALL researchers have taken note (see, for example, Pellettieri's, 1999, use of YTalk). However, this approach is not only cumbersome when it comes to data analysis; it also obscures other potentially interesting elements of online interaction such as scrolling as a strategy to "recapture" previous content. Due to this limitation, Smith & Gorsuch (2004) suggest that claims about the occurrence of certain interactional moves and strategies largely require one to infer too much in those studies that use only a hard copy transcript of the interaction. Language Learning & Technology 89 Bryan Smith Methodological Hurdles in Capturing CMC Data It seems, then, that if there are instances of self-repair that appear on the chat logs, there must be many more that are attempted, but edited out before the message is sent to the interlocutor. It is important to note that from an interactionist perspective on SLA, the potential value of this output in the form of selfrepairs is not diminished by the fact that they may be subsequently edited out by the learner. CMCovert repair In an SCMC context a unique type of self-repair is possible that may be considered a CMC-specific form of self-repair. In this case a message is typed, but a self-initiated correction or rephrasing is executed before the message is sent to the interlocutor. The self-repair is certainly overt from a psychological perspective, but essentially "covert" from the interlocutor's perspective in that there is never any evidence of there having been a self-repair or rephrasing. Further, such repair is often not immediate and regularly contains several "embedded" self-repairs in the same evolving message. The proposed methodology allows an examination of this type of self-repair, heretofore lost to methodological limitations of the research design. This type of self-repair is different from true covert repair since, in the latter, we may not expect the same output-related benefit that may only be present upon actually producing the target language. Of course, some of the argued benefits of pushed output for the speaker are obtained only once the interlocutor reacts to a speaker's productive output. In an SCMC environment, these conditions will be the same, as will possible benefits achieved when one engages in truly covert self-repair2. However, what I term here as "CMCovert" self-repair is an interesting and largely unexplored phenomenon that may provide us valuable insights into the nature of such self-repair as well as the effect of CMCovert self-repair on SLA. Given the pervasiveness of such repairs I argue that any examination of the occurrence and possible effects of SCMC self-repair from the repairer’s point of view simply must include this CMCovert self-repair. To do this one must abandon the practice of simply using printed chat logs to analyze CMC interactional data and employ more dynamic means such as screen capture technology. This does not only apply to investigations of self-repair, but to CMC learner interaction in general as suggested by Smith & Gorsuch (2004). THE CURRENT STUDY The purpose of the current study is two-fold. First, I wish to establish empirically just how misguided it would be for CALL researchers to continue to rely on printed chat logs alone when making SLA-related claims about SCMC interaction. This inadequacy is revealed by comparing two types of data from the same task-based interaction sessions: the printed chat log and the Camtasia screen capture video record of the same session for the same participants (hereafter chatscript). The amount of SISR, then, is the dependent variable in this study with the learners in this study serving as their own control group. Data collection/evaluation methodology is the independent variable. Second, the nature of learner self-repair in this SCMC context will be explored. Though there is considerable research on face-to-face self-repair, there is very little CALL work on this topic. Research Questions This study explores what Buckwalter (2001) has described as self-initiated self-repair, albeit in a SCMC environment. Following Lai & Zhao (2006), self-repairs were defined as episodes where the participants immediately corrected their own production without prompts from their interlocutors. Some of these episodes are visible on the final chat logs of each session and, as we will see, some are not. All SISR episodes, however, are visible on the video file of the relevant chat session. The research questions are as follows: Language Learning & Technology 90 Bryan Smith • • Methodological Hurdles in Capturing CMC Data Does the hard copy transcript of chat interactions differ from that which is available using a screen capture program in terms of the amount of SISR that is evident? What is the nature of CMCovert self-repair? In order to answer these questions, hard copies of all the chat logs for all participants across all tasks were analyzed and coded for instances of SISR. Hypothesis 1 predicts that evidence of a significantly higher amount of SISR will be found in the screen capture condition. This prediction is based largely on Smith & Gorsuch (2004), who found that a similar method of data collection captured a much richer picture of the CMC interaction of the participants in their study. Drawing from previous research on self-correction as well as CALL, a substantial amount of CMCovert self-repair is expected (Hypothesis 2a). More pronounced attention to form manifested in a higher number of grammatical self-repairs than lexical selfrepairs are expected due to the increased planning/monitoring, and processing time afforded by the CMC medium (compared to a face-to-face setting). Also, because of the relatively low proficiency level of the participants (Novice-High), more self-correction of grammar-related problems is expected (Hypothesis 2b). It is also expected that few appropriateness (A) self-repairs relative to the number of error (E) selfrepairs will be found also due mostly to the relatively low proficiency level of the participants (Hypothesis 2c). A summary of the four directional hypotheses follows below: • • • • Hypothesis 1: A significantly higher amount of SISR will be found in the screen capture condition than in the hard copy transcript condition. Hypothesis 2a: A substantial amount of CMCovert self-repair is expected. Hypothesis 2b: More SISR of grammar-related problems than lexical ones are expected. Hypothesis 2c: Few appropriateness (A) SISRs relative to the number of error (E) SISRs will be found. Methodology Participants Forty-six beginner-high proficiency level students participated in this study as part of their regularly scheduled German language course at a major southwestern university in the United States. Students were required to meet once every other week in the foreign language micro-computing lab. Six CMC sessions were scheduled over the course of the semester. All students were either sophomore or junior undergraduates and all were native speakers of English. None were German majors. Their proficiency level and placement in the German sequence was determined by an in-house online placement test. All participants in the present study were characterized by the instructor of record as roughly at the ACTFL Novice-High proficiency level and were familiar with the chat function in Blackboard. All participants did complete one training session prior to data collection to ensure they were familiar with the general task and procedures since they were not necessarily accustomed to performing similar task-based CMC activities in their German class. Though there was some evidence of target cultural materials and short samples of authentic literature, the core textbook and instruction were largely organized along a grammatical syllabus. Materials Paired participants completed one jigsaw task per session over the course of the semester, which resulted in a potential total of six tasks per student (assuming perfect attendance). This task type was chosen because of its structural requirement of two-way information exchange by participants who are striving to reach a convergent goal. Pairs were not necessarily matched from week to week. Though each task was slightly different, they all follow Pica, Kanagy, and Falondon's (1993) task features for jigsaw tasks. Four of the six tasks were video-based, whereby one learner (learner A) would view a two-minute dramatic video clip that corresponded to the week's assigned course content. The other learner (learner B) would not view this clip but would study a series of eight stills from the same video clip, randomly arranged. Language Learning & Technology 91 Bryan Smith Methodological Hurdles in Capturing CMC Data The stills were such that a logical order was not discernable simply by examining the photos alone but quickly and easily sequenced upon viewing the clip from which they were taken. The remaining two tasks were standard sequential ordering tasks, where learners each held three out-of-sequence pictures which when put together made up a logical story sequence. The video-based tasks were directly tied to the core content and textbook of the course and came from the ancillary DVD and workbook accompanying the main course textbook. Participants interacted with one another via the chat function in Blackboard and were assigned to one of various paired "groups" under Blackboard's Communication tool, Virtual Classroom. Following Smith & Gorsuch (2004) and Lai & Zhao (2006), the dynamic screen capture software Camtasia was employed to record exactly what appeared on each participant's computer screen in real time. Camtasia has the capability of recording and creating a movie file of each participant's computer screen, allowing one to play back the chat session in its entirety. Procedures Participants were required to attend and participate in each session since these were built into the syllabus of the course. Participation in the study was voluntary and followed the university's prescribed IRB protocol. The six CALL tasks described above were completed every other week during the middle part of the semester (over 12 weeks). Each class lasted approximately one hour. The average amount of time it took pairs to complete each task was just over 25.5 minutes as measured by the time stamp on the chat logs of the interactions. No specific time limit was placed on students once they began the task; however, given the length of the class, participants realistically had about 40 minutes to complete each task. All students worked collaboratively online with a partner. Each participant was given task sheet A or B. All of those holding task sheet A were grouped together and separate from those students in group B. This was done in order to reduce the chance that any participant would gain visual access to their interlocutor’s (partner’s) task sheet/video clip. For the video-based tasks, the clip for that session was made available to group A in each dyad. These students viewed the clip with headphones on while their partner studied the still images of various scenes from the same video clip (group B). Once the video clip had played through one time, participants were directed to interact with their assigned chat partner and decide the proper order of the pictures held by student B. In order to successfully complete this task, learner A had to describe in detail the events in the short video clip (in the target language) while learner B attempted to place the pictures in order based on these descriptions. Likewise, learner B was told to describe each picture to learner A in order to facilitate this ordering. Learners were instructed to interact using the target language with the goal of agreeing on a likely order to the still images held by learner B. Once this order was agreed on learners were to declare the task completed by typing the proposed order of the pictures and writing "finished". Upon task completion, Camtasia was stopped and the video record of the interaction saved to a removable storage drive. The chat logs of these interactions were saved automatically in Blackboard. Data analysis Following Kormos (2000), truly covert repairs were not considered in the data since one can only infer these occurrences. Rather, CMCovert self-repairs (those that are recorded, but which do not appear on the chat logs), as well as those which are truly overt (appearing on the chat logs), were considered. Following Lai & Zhao (2006) and Lee (2002), spelling mistakes/corrections were not counted in the data. Errors in the obligatory capitalization of German nouns and subsequent self-corrections were counted and coded as lexical errors (EL). Since the first research question sought to compare the amount of self-repair evident when employing two data collection methodologies in a paired groups fashion, only those participants whose chat records showed evidence of self-repair across both of these methodologies were candidates for inclusion in these data. To this end, all printed chat logs (n=94) of pair interaction were evaluated and coded for instances of self correction (method A) using the coding scheme below (see Table 1), which is adapted from van Hest (1996). Table 2 shows examples of each of these categories from the data. Language Learning & Technology 92 Bryan Smith Methodological Hurdles in Capturing CMC Data Table 1. Description of Types of Self-Repair. Type of self-repair Description Lexical (EL) Error Morphological (EM) Syntactic (ES) Lexical (AL) Appropriateness Syntactic (AS) Insertion (AI) Different (D) Rest (R) The learner has selected the wrong word and substitutes the correct one for it. The learner corrects a morphological error. The learner produces a grammatical construction which cannot be finished without violating the grammar of the target language. The learner replaces one term with another, usually more precise, term which better fits the concept s/he wishes to express. The learner replaces the original syntactic construction with a construction which, in his/her opinion, is more appropriate. The speaker repeats part of the original utterance and inserts one or more words to specify his/her message. The speaker interrupts his/her current message to introduce a new, totally different topic. Abandonments. All other self-repairs that do not fit cleanly into any of the other categories. Note: Other categories discussed in van Hest (1996) were either not appropriate for the CMC context, for example, phonological errors, or were not present in the data and, therefore, are not included in these tables. This resulted in a total of eight chatlogs. That is, out of the transcripts of the twenty-eight students completing the tasks described above, there was evidence of self-repair on only eight of the 94 chatlog transcripts, or 8.5%. This resulted in a total of 9 instances of self-repair (see Table 3). Next, the corresponding Camtasia file was viewed in its entirety (method B) for each corresponding transcript (n=8). Each video file was coded for instances of SISR according to the same coding scheme (Table 1). Table 2. Examples of Self-repair Types Type of self-repair Error Lexical (EL) Morphological (EM) Syntactic (ES) Lexical (AL) Appropriateness Syntactic (AS) Insertion (AI) Different (D) Camtasia Chatlog …ist hat auch… …welche Buchstaben ist… …weil er dieas maädchen… …bitte Wurst und S Schnitzel. …wann er mit sein freundin… Drei Leute sind [im Morgen] im Restaurant. [+] die männer sind look▌ wie sagt man looking to look at? …hat auch… …welche Buchstab ist … …weil er das mädchen… …bitte Schnitzel. …wann er sein freundin… Drei Leaute sind im Morgen im Restaurant. wie sagt man to look at? Note: A coding scheme adapted from Smith & Gorsuch (2004) is shown in the Appendix. Results The chat data for the target learners yielded a total of 1,464 words. The number of words produced by these learners on the tasks considered ranged from 101 to 398 (M=183, SD=93). As can be seen from Language Learning & Technology 93 Bryan Smith Methodological Hurdles in Capturing CMC Data Table 3 below, there were many more instances of self-repair evident when using method B. A Wilcoxon Signed Ranks Test shows that significantly more instances of learner self-repair were captured using method B than method A (z = 2.53, p =. 01, r = .79; see Table 4). To allow for a clearer comparison to previous studies, "standardized scores" for instances of self-correction relative to the quantity of discourse produced by each participant were calculated. As can be seen from Table 5, the Wilcoxon Signed Ranks Test yielded very similar results (z = 2.52, p = .01, r = .88). Viewed another way, learners who engaged in task-based CMC interaction with a partner actually self-repaired about six times per one hundred words (of printed transcript), whereas they appear to have self-repaired less than once per one hundred words based on the hard copies of the chat logs alone3. Table 3. Self-repairs by Type, Across Method of Data Collection Type of Self-Repair Lexical (EL) Morphological (EM) Syntactic (ES) Total (E) Lexical (AL) Syntactic (AS) Insertion (AI) Total (A) D R Error Appropriateness Different Rest Chat log Method Camtasia Totals 3 23 26 2 4 9 0 0 0 0 0 0 29 31 16 68 5 9 3 17 1 0 20 77 5 9 3 17 1 0 Table 4. Comparison of Self-repairs in chat logs vs. Camtasia Method A (Chat log) B (Camtasia) N 8 8 M 1.13 10.75 SD .35 5.31 z p effect-size r 2.53 .01 .79 Figure 1 below shows two transcript versions of the same chat interaction. The left column is the "Camtasia-enhanced" chatscript, which was transcribed while viewing the video screen capture file. The column on the right shows a traditional chat log of the same segment without the benefit of the Camtasia file. In column B only one instance of SISR is evident (lines 6b-7b) whereas column A shows at least three non-spelling self-repairs (lines 5a, 6a-7a, and 8a). The SISR in lines 5a and 8a go undetected when relying on printed chat logs alone. The appendix shows the coding scheme employed in evaluating the Camtasia data. Table 6 shows the data realigned to collapse the categories of morphological and syntactic selfcorrections together into the broader category of "grammatical" self-repair. This will allow for an easier comparison with data reported in previous studies discussed earlier. Although there were almost twice as many grammatical self-repairs as lexical self-repairs, this difference was not statistically significant. Finally, when we compare the differences in self-corrections of errors with self-corrections of appropriateness issues, we find that learners self-repaired significantly more errors (z = 2.53, p =. 01, r = .68; see Table 7). In order to see if there were any differences in which types of self-repairs learners engaged in within the error (EL, EM, ES) and appropriateness (AL, AS, AI) categories (see Table 3) a Friedman test was performed. Differences within each category were not statistically significant. Language Learning & Technology 94 Bryan Smith Methodological Hurdles in Capturing CMC Data SCMC Chatscript Column A 1a. Pierre: er hat seine geschaft aufgeraumt 12:59:43 2a. Pierre: bildung C er geht nach hause und seine garge ist nicht sauber 1:00:08 3a. Daniel: D-Der man ging ins [2a] garage und es ist sehr schmustivg 1:00:29 4a. Daniel: E-er gi heht raus deines zimmem er r 1:00:54 5a. Daniel: F-ErDas Garage ist sauber und er hat eine hacke im hand 1:01:15 6a. Daniel: **in deiner hand 1:01:26 7a. Daniel: seiner* 1:01:31 8a. Daniel: meine ist def seine geschaftrft? 1:02:19 9a. Daniel: ich habe kaein geschaft nur ein garage und ein Bettzimmer. 1:02:36 Hard copy of transcript Column B 1b. Pierre: er hat seine geschaft aufgeraumt 12:59:43 2b. Pierre: bildung C er geht nach hause und seine garge ist nicht sauber 1:00:08 3b. Daniel: D-Der man ging ins garage und es ist sehr schmustig 1:00:29 4b. Daniel: E-er geht raus deines zimmer 1:00:54 5b. Daniel: F-Das Garage ist sauber und er hat eine hacke im hand 1:01:15 6b. Daniel: **in deiner hand 1:01:26 7b. Daniel: seiner* 1:01:31 8b. Daniel: seine geschaft? 1:02:19 9b. Daniel: ich habe kein geschaft nur ein garage und ein Bettzimmer. 1:02:36 Figure 1. Camtasia enabled chatscript and printed chat log comparison. Click the following link to view this actual segment from the data (link to Camtasia Example Flash file) Table 5. Self-Repairs Per Word Across the Two Data Collection Methods Method M .007 SD .003 z P effect-size r A (Chat log) N 8 B (Camtasia) 8 .061 .020 2.52 .01 .88 Table 6. Percentage of the Total Number of Self-Repairs (Grammatical vs. Lexical) Method Grammatical self-repairs Lexical self-repairs Other self-repairs (AI & D) A (Chat log) 6 (67%) 3 (33%) 0 (0%) B (Camtasia) 54 (63%) 28 (33%) 4 (5%) Note: Morphological and syntactic categories were collapsed together into a larger grammatical category Table 7. Method B Error and Appropriateness Self-Repairs Self-repair type M 8.50 SD 4.63 z P effect-size r Errors N 8 Appropriateness 8 2.13 1.36 2.53 .01 .68 In sum, the data suggest that the Camtasia screen capture method (method B) of data collection and analysis yields significantly more evidence of SISR than does the chat log method (method A). There are also significantly more SISRs of errors than appropriateness points. In all cases the effects were of a strong magnitude as indicated by the high effect-size r measure. No differences across sub-types of error and appropriateness SISRs were found. Language Learning & Technology 95 Bryan Smith Methodological Hurdles in Capturing CMC Data The degree of support for the specific hypotheses presented in the previous section is listed below: • Hypothesis 1: A significantly higher amount of SISR will be found in the screen capture condition than in the hard copy transcript condition. Result: Strongly supported • Hypothesis 2a: A substantial amount of CMCovert self-repair is expected. Result: Strongly supported • Hypothesis 2b: More SISR of grammar-related problems than lexical ones are expected. Result: Partially supported • Hypothesis 2c: Few appropriateness (A) SISRs relative to the number of error (E) SISRs will be found. Result: Strongly supported DISCUSSION The data suggest that relying on printed transcripts alone may create the impression that learners do not self-correct very often in an SCMC environment – clearly a faulty conclusion. Indeed, these numbers show that evaluating instances of self-correction on the basis of printed chat logs alone leads to an underestimation by over eight-fold of the amount of self correction that actually occurred. In fact, the present data may be a conservative estimate of this difference since Lai & Zhao's (2006) data show an even higher rate of self-correction in an SCMC setting (29 self-corrections per one hundred words). Such a pronounced and fundamental mis-characterization of the nature of SCMC interaction can help explain conclusions such as those in Jepson (2005) who, finding little self-repair in his CMC data, stated that though both voice and text chats contained various types of repair moves, self-correction was not among them. Jepson also suggests that self-correction in an SCMC context may be rare because speakers do not notice their errors, and thus would not need to correct them. Alternatively, he suggests that since selfcorrection is very dependent on the social context of the interaction (Kormos, 1999), it may be that the non-native speaker – non-native speaker SCMC context is somehow not conducive to self-correction. For example, learners may not see the need for accuracy or may perceive self-correction as face threatening. This rationale, however, contradicts much of the recent CMC interactionist research which suggests that the CMC environment most likely enhances noticing (Lai & Zhao, 2006; Smith, 2004). The results from the current study are similar in some ways to previous findings and quite different in other ways. Yuan (2003) found that learners corrected under 9% of their errors in a CMC environment. His data, which includes self-repair of spelling errors, suggests that the CMC environment does not always make errors more salient to learners, at least not verb tense and modal errors, which were never self-repaired. In order to make Yuan's data more comparable to this study, the spelling self-corrections in his data need to be removed. This results in a new number of 33 self-repaired errors (down from 44) of which 20 were grammatical (61%) and 13 were lexical (39%). These numbers are quite similar to those found in the present study (63% and 33% respectively). It is also interesting to note that the method of data collection does not seem to influence the relative amount of each type of self-correction recorded (Table 6). This point is reinforced when we consider that Yuan's data were based on printed transcripts of the chat interaction. In terms of which types of SISR learners engage in, it seems the results are largely in conflict with the existing self-repair literature, which is largely limited to face-to-face studies. Van Hest (1996), for example, found that almost 40% of SISR were appropriateness repairs compared with 22% of error SISRs. Kormos (2000) reports that almost 39% of all SISRs were error repairs compared with almost 23% appropriateness SISRs. This study found a relatively high percentage of "different" repairs (almost 22%) Language Learning & Technology 96 Bryan Smith Methodological Hurdles in Capturing CMC Data as well. Kormos also found little difference in grammatical and lexical error self-repairs (16.9% and 14.2% respectively), which is similar in some ways to the findings in the present study. The little data available that specifically focuses on learner self-correction in a CMC environment suggests that learners do self-correct in an SCMC task-based setting, perhaps due to a heightened degree of noticing, which is fostered by the SCMC environment itself. Second, learners tend to focus on errors rather than appropriateness issues when engaged in SISR and virtually no "different" and "rest" points. Third, learners seem to correct grammatical points more often than lexical points, though this difference was not statistically significant. Though it is normally of little value to compare CALL with face-to-face studies, it is interesting to note the pronounced differences in what learners seem to self-correct across the two environments. Taken together, the face-to-face self-repair literature seems to show a clear preference for lexical self-repairs over grammatical self-repairs, precisely the opposite of that found in most of the SCMC studies. The question, then, is why this may be so. Written communication normally affords more opportunity for attention to form, whereas spoken language often occurs under more time pressure to achieve fluency (Chapelle, 2003). The SCMC interaction allows for more processing time, which is conducive to focusing on form. The visual saliency of the text as well as the permanency of the written word, which enables one to review the previous "utterances," allows learners to focus their attention on the formal aspects of their output without disrupting the flow of communication. Context Influencing the Nature of Self-Correction Linguistic context It is not surprising that self-repair of grammar-related problems is so common in the data given the relatively high percentage of morphological errors made by the learners. Aside from the noted potential for the CMC environment to enhance a focus on form, it may also be that errors that do not require a major restructuring of the utterance, but rather merely need to apply simple rules of grammar (as do morphological errors) are more likely to be self-corrected by L2 speakers (Kormos, 2000). In the SCMC environment this condition may be enhanced as these relatively "minor" morphological errors are rendered more salient due largely to their "permanence" on the screen. This may help explain why Table 3 shows nearly a 2:1 ratio of the number of morphological errors self-repaired to the number of syntactic errors repaired in the Camtasia condition. Classroom Context The nature of the language instruction that the learners are used to may influence what they choose to self-correct. For example, if learners receive instruction that stresses the importance of grammatical accuracy in successful communication and students regard grammatical errors as serious flaws in their performance, they may make an effort to correct their grammatical errors. This notion is supported by Bardovi-Harlig and Dörnyei (1998), who found that grammatical errors were more salient for L2 speakers in a foreign language setting than for L2 learners in a naturalistic environment. The students in the present study fall clearly into the former category. There may also be a complex combination of L1 and L2 linguistic, classroom, and cultural influences that come together to influence the nature of SISR. Shonerd (1994) notes the seeming selectivity of learner self-repair and suggests the nature of self-repair may be culturally bound. His Japanese L1 speakers made more morphological and syntactic self-repairs of their English usage and fewer lexical and pronunciation repairs than did other L1 groups. In terms of the present data, there were numerous SISRs that involved the capitalization of German nouns. As mentioned, these SISRs were coded as lexical errors. The Language Learning & Technology 97 Bryan Smith Methodological Hurdles in Capturing CMC Data interaction of German and English linguistic factors and the classroom cultural context may have come together to influence the nature of SISRs in this study. Limitations of the Current Study Perhaps the clearest limitation of the current findings is the small sample size (n=8) from which the data are drawn. Thus, it is hard to generalize from these data. However, the choice was made from the outset to include only those chat records which contained hard copy evidence of SISR. This resulted in a low pool of participants (much lower than actually took part in the study) from which the self-repair data was drawn. Indeed, this is the only way to really make the comparisons needed to answer research question 1, which asked whether the hard copy transcript of chat interactions would differ significantly from a Camtasia-enabled transcript. Perhaps another limitation was that the interaction was not anonymous. The role of anonymity in CMC interaction is well established (Zhao, 1998). It could be that knowing who one's interlocutor is has some effect on the amount and nature of one's SISR. For example, it could be that knowing the identity of one's interlocutor may affect whether or not one self corrects depending on the learner's relationship to that interlocutor. This is an empirical question that could be taken up in future research. Another limitation is that the tasks learners completed each week, while conforming broadly to Pica, Kanagy, and Falondon's (1993) jigsaw task type, were slightly different, which may have influenced the nature of the SISRs. In addition to the large body of literature on the effect of task type on learner interaction, the self-repair literature also suggests that the frequency and type of self-repair is affected by the task structure (see, for example, van Lier, 1988). It would be interesting to test van Hest's (1996) assertion that those tasks requiring more precise expression will result in more A-repair. A first step in this work would be to establish what exactly draws out "more precise" expression in an SCMC context that goes beyond the "precision" elicited by the CMC medium itself. CONCLUSION AND FUTURE DIRECTIONS The results of this study suggest that relying on printed chat logs alone when analyzing SCMC data is a very tenuous undertaking. The recommendation here is to abandon this practice in CMC research in favor of one similar to that presented in this study, at least when data salient to the specific area of inquiry may be lost. Future inquiry using this methodology may include exploring the influence of classroom context on the type of self-correction. It would be interesting to see if, for example, learners accustomed to a communicative classroom context engage in a different type of SISR than those learners who are used to more structure-oriented contexts. Second, because there seems to be a trend toward learners self-repairing grammatical points rather than lexical ones in a SCMC environment, it would be interesting to explore this notion further with larger numbers of students perhaps across various target languages. As it stands, there is a theoretical rationale to explain this occurrence as well as some empirical evidence to support it, but the data are far from conclusive. Third, the SCMC medium may itself influence the nature of SISR. One artifact of this medium is the ability learners have of scrolling back in the chat text to review previous messages. Indeed, Lai & Zhao (2006) base some of their predictions on this assumption. Though such occurrences in an SCMC setting are well established, the influence of this feature of SCMC interaction has not specifically been shown to directly impact the nature or amount of SISR. Future research may wish to explore this idea more explicitly. Finally, it seems that existing models of SISR are insufficient to account for what occurs in an SCMC context. A model of SISR specific to the SCMC context would be helpful for future inquiry into this important area of applied linguistics. Language Learning & Technology 98 Bryan Smith Methodological Hurdles in Capturing CMC Data NOTES 1. Self-repair in this paper is considered synonymous with self-initiated self-completed correction. Self repair and self-correction are used interchangeably. 2. As one reviewer points out, in true covert repair learners may engage in hypothesis testing in their minds as evidenced in some think aloud protocols. However, I argue that CMCovert self-repair is indeed unique in that in an SCMC environment we often see many lengthy and embedded self-repairs, whereas in a face-to-face environment these are often shorter, more direct, and immediate. 3. The number of words produced by each individual was used when calculating this figure and not the total words produced by the dyad. Language Learning & Technology 99 Bryan Smith Methodological Hurdles in Capturing CMC Data APPENDIX SCMC Coding Scheme Coding symbol strikethrough Black bar + underline ▌ [post-hoc inserted text] [post-hoc deleted text] [+] [-] [+/-] [line number] Example: [3] Meaning Indicates text that has been typed and subsequently deleted before message is sent. Indicates text with embedded deletion has been deleted. Black vertical bar indicates where deletion begins. Deleted text is underlined to the left of the bar. Indicates that the text within the brackets has been inserted at a later point in the message. Explanation Strikethrough shows messages or parts of messages that a learner has typed but deleted before sending the final message. None of the strikethrough text appears on the screen of the interlocutor nor on the hard copy of the chat log. Black bar + underline is used when a message or part of a message which already has some deleted sections is subsequently deleted in its entirety. This coding allows the acknowledgment of deletions of text with embedded deletions. Note: A second, third, etc. occurrence of post-hoc inserted text is signified with double/triple brackets respectively [[text]], [[[text]]]. Indicates that the deleted text within the brackets was deleted at a later point in the message. Indicates the point in the message at which the [posthoc inserted text] was inserted. Indicates the point in the message at which the [posthoc deleted text] was deleted. Note: A second, third, etc. occurrence of post-hoc deleted text is signified with double/triple brackets respectively [[text]], [[[text]]]. Indicates the point in the message at which a correction was made. This code is used for short one or two character corrections such as for capitalization, spelling, typos, etc. This code eliminates the need for using the more lengthy [-][+] in sequence. Indicates the point in text currently being typed but not yet sent where a new line from the interlocutor appears on the screen of the target participant. Often a line from the interlocutor will appear on the screen mid-way through a message which is in the process of being typed. This code indicates both the point at which this new message appears and its line number on the chat log/chatscript. For example in the chat text below Jordan's line 3 appears while Katarina is typing line 6. Specifically, it occurs immediately after Katarina types the word "nimmt". Language Learning & Technology Note: The point in the message at which a second, third, etc. post-hoc insertion is made is signified with double/triple brackets respectively [[+]], [[[+]]]. Note: The point in the message at which a second, third, etc. post-hoc deletion is made is signified with double/triple brackets respectively [[-]], [[[-]]]. 100 Bryan Smith Methodological Hurdles in Capturing CMC Data ACKNOWLEDGMENTS The author wishes to thank the following people (in alphabetical order). Shana Bell, Jamison Gray, Peter Lafford, and, of course, the participants in the study. Also, many thanks go out to the reviewers and those at LLT for such insightful suggestions and close reading of the manuscript. ABOUT THE AUTHOR Bryan Smith is Assistant Professor of Educational Linguistics at Arizona State University. His research interests include language learner interaction and computer-mediated communication in the second/foreign language classroom. Email: [email protected] REFERENCES Bardovi-Harlig, K., & Dornyei, Z. (1998). Do language learners recognize pragmatic violations? Pragmatic versus grammatical awareness in instructed L2 learning. TESOL Quarterly, 32, 233-262. Buckwalter, P. (2001). Repair sequences in Spanish L2 dyadic discourse: A descriptive study. The Modern Language Journal, 85, 380-397. Camps, J. (2003). The analysis of oral self-correction as a window into the development of past time reference in Spanish. Foreign Language Annals, 36, 233-242. Chapelle, C. (2003). English language learning and technology: Lectures on applied linguistics in the age of information and communication technology. Philadelphia: John Benjamins. Doughty, C., & J. Williams. (Eds.). (1998). Focus on form in classroom second language acquisition. Cambridge: Cambridge University Press. Fathman, A. (1980). Repetition and correction as an indication of speech planning and execution processes among second language learners. In H. Dechert, & M. Raupach (Eds.), Towards a crosslinguistic assessment of speech production (pp. 77-85). Frankfurt: Lang. Foster, P., & Ohta, A. S. (2005). Negotiation for meaning and peer assistance in second language classrooms. Applied Linguistics, 26, 402-430. Gass, S. M., & Mackey, A. (2000). Stimulated recall methodology in second language research. Mahwah, NJ: Erlbaum Associates. Izumi, S. (2003). Comprehension and production processes in second language learning: In search of the psycholinguistic rationale of the output hypothesis. Applied Linguistics, 24, 168-196. Jepson, K. (2005). Conversations and negotiated interaction in text and voice chatrooms. Language Learning & Technology, 9(3), 79-98. Kormos, J. (1999). Monitoring and self-repair in L2. Language Learning, 49, 303-342. Kormos, J. (2000). The role of attention in monitoring second language speech production. Language Learning, 50, 343-384. Lai, C., & Zhao, Y. (2006). Noticing and text-based chat. Language Learning & Technology, 10(3), 102120. Language Learning & Technology 101 Bryan Smith Methodological Hurdles in Capturing CMC Data Lee, L. (2002). Synchronous online exchanges: A study of modification devices on non-native discourse. System, 30, 275-288. Lennon, P. (1990). Investigating fluency in EFL: A quantitative approach. Language Learning, 40, 387417. Levelt, W. (1983). Monitoring and self-repair in speech. Cognition, 14, 41-104. Long, M.H., & Robinson, P. (1998). Focus on form: Theory, research, and practice. In C. Doughty & J. Williams (Eds.), Focus on form in classroom SLA (pp. 15-41). New York: Cambridge University Press. Ohta, A. S. (2001). Second language acquisition processes in the classroom setting. Learning Japanese. Mahweh, NJ: Erlbaum Associates. Pellettieri, J. (1999). Why-talk? Investigating the role of task-based interaction through synchronous network-based communication among classroom learners of Spanish. Unpublished doctoral dissertation, University of California, Davis. Pica, T., Kanagy, R., & Falodun, J. (1993). Choosing and using communication tasks for second language research and instruction. In G. Crookes & S. M. Gass (Eds.), Tasks and second language learning (pp. 9– 34). Clevedon, UK: Multilingual Matters. Postma, A., & Kolk, H. (1992). The effects of noise masking and required accuracy on speech errors, disfluencies, and self-repairs. Journal of Speech and Hearing Research, 35, 537-544. Poulisse, N. (1997). Compensatory strategies and the principles of clarity and economy. In G. Kasper & E. Kellerman, Communication strategies: Psycholinguistic and sociolinguistic perspectives (pp. 49-64). London: Longman. Schmidt, R. (1993). Awareness and second language acquisition. Annual Review of Applied Linguistics, 13, 206–226. Shehadeh, A. (2002). Comprehensible output, from occurrence to acquisition: An agenda for acquisitional research. Language Learning, 52, 597-647. Shehadeh, A. (2001). Self-and other-initiated modified output during task-based interaction. TESOL Quarterly, 35, 433-457. Shehadeh, A. (1999). Non-native speakers' production of modified comprehensible output and second language learning. Language Learning, 49, 627-675. Shonerd, H. (1994). Repair in spontaneous speech: A window on second language development. In V. John-Steiner, C. Panofsky & L. W. Smith, (Eds.) Sociocultural approaches to language and literacy: An interactionist perspective (pp. 82-108). Cambridge: Cambridge University Press. Smith, B. (2004). Computer-mediated negotiated interaction and lexical acquisition. Studies in Second Language Acquisition, 26, 365-398. Smith, B. (2005). The relationship between negotiated interaction, learner uptake, and lexical acquisition in task-based computer-mediated communication. TESOL Quarterly, 39, 33-58. Smith, B., & Gorsuch, G. J. (2004). Synchronous computer mediated communication captured by usability lab technologies: New interpretations. System, 32, 553-575. Swain, M. K. (1985). Communicative competence: Some roles of comprehensible input and comprehensible output in its development. In S. M. Gass & C. G. Madden (Eds.), Input in second language acquisition (pp. 235–253). Rowley, MA: Newbury House. Language Learning & Technology 102 Bryan Smith Methodological Hurdles in Capturing CMC Data Swain, M. (2001). Integrating language and content teaching through collaborative tasks. Canadian Modern Language Review, 58, 44-63. Swain, M. (2005). The output hypothesis: Theory and research. In E. Hinkel (Ed.), Handbook on research in second language teaching and learning (pp. 471-484). Mahwah, NJ: Lawrence Erlbaum. Swain, M., & Lapkin, S. (1995). Problems in output and the cognitive processes they generate: A step towards second language learning. Applied Linguistics, 16, 371-391. van Hest, E. (1996). Self-repair in L1 and L2 production. Tilburg: Tilburg University Press. van Lier, L. (1988). The classroom and the language learner. New York: Longman. Verhoeven, L. T. (1989). Monitoring in children's second language speech. Second Language Research, 5, 141-155. Wouk, F. (2005). The syntax of repair in Indonesian. Discourse Studies, 7, 237-258. Zhao, Y. (1998). The effects of anonymity on peer review. International Journal of Educational Telecommunication, 4, 311-346. Language Learning & Technology 103 Language Learning & Technology http://llt.msu.edu/vol12num1/mcquillan February 2008, Volume 12, Number 1 pp. 104-108 COMMENTARY: CAN FREE READING TAKE YOU ALL THE WAY? A RESPONSE TO COBB (2007) Jeff McQuillan Center for Educational Development Stephen D. Krashen University of Southern California Cobb (2007) argues that free reading cannot provide L2 readers with sufficient opportunities for acquiring vocabulary in order to reach an adequate level of reading comprehension of English texts. In this paper, we argue that (1) Cobb severely underestimates the amount of reading even a very modest reading habit would afford L2 readers, and therefore underestimates the impact of free reading on L2 vocabulary development; and (2) Cobb’s data show that free reading is in fact a very powerful tool in vocabulary acquisition. COBB'S CLAIM: FREE READING IS NOT SUFFICIENT Krashen (1989, 2004) and others have argued that free reading is a major contributor to vocabulary development among both first and second language readers (see also McQuillan, 1998). Free and extensive reading advocates have claimed that such reading can and does provide acquirers with sufficient resources to reach a high level of literacy development. Cobb (2007) claims, however, that free reading cannot possibly provide sufficient opportunities for L2 readers to reach high level of vocabulary acquisition, of going "all the way" to the state of a fluent adult L2 reader. Cobb cites evidence showing that vocabulary acquisition requires a minimum of six to ten exposures to a word family, and that the minimal number of word families required for comprehension of non-specialist materials in English is 3000 to 5000, depending on which estimate is used (2007, p. 41). For this study, Cobb used the low end of these estimates (six exposures to a word family, 3000 word family level). Cobb analyzed how frequently vocabulary occurred in three subsets of a corpus of academic, fiction, and newspaper texts, each subset containing between 163,000 and 179,000 words, in order to determine if words occur in sufficient frequency for acquisition (see Table 1). (Cobb explains that the newspaper sample is about 100 pages of newspaper reading, the academic sample about 17 scientific papers, and the fiction sample about six stories the size of Alice in Wonderland.) Cobb estimated that in a "year or two" of language study, a student could read the equivalent of one of these three subsets, or roughly 175,000 words (p. 41). He considered this to be an "optimistic" estimate. Table 1. Number of Words in Each Sample Subset Press Academic Fiction Total Words in Sample 179,000 163,000 175.000 517,000 Cobb then randomly selected ten word families each from of the 1000, 2000, and 3000 most frequently appearing word families in English and determined how many times those families appear at each level for each of the three genres of reading material. Using corpus analysis, he found that while the frequency of recurrence for the 10 word families would probably be sufficient at the 1000 word level for any of the Copyright © 2008, ISSN 1094-3501 104 Jeff McQuillan and Stephen D. Krashen A Response to Cobb (2007) three genres, free reading would be insufficient to attain the 2000 and 3000 word level. As illustrated in Table 2 (from Cobb, Table 1), one word out of the sample of ten does not appear often enough in newspapers, two out of ten in academic writing, and three out of ten in fiction. Table 2. Results for 2000 Word Frequency Word Families: Frequency of Occurrence (From Cobb, 2007, Table 1) persua' grade' technolog' moon' wire' maintain' analy' drama' depress' sue' Threshold Press 17 14 9 6 3 16 12 40 14 8 9 Academic 3 25 8 27 5 49 129 14 7 7 8 Fiction 7 8 0 31 20 6 4 8 9 1 7 Total 27 47 17 64 28 71 145 62 30 16 10 The situation is even more serious at the 3000 word level, with six out of ten failing to make the minimum threshold of six occurrences in the press corpus, eight out of ten in the academic corpus, and five out of ten in the fiction corpus (Table 3). Table 3. Results for 3000 Word Frequency Word Families: Frequency of Occurrence (From Cobb, 2007, Table 1) irritat' millimeter’ urgen' transmi' chew' naked’ civiliz' contest’ charm’ prompt' Threshold Press 3 0 7 5 0 2 5 15 10 6 4 Academic 0 0 1 9 0 1 12 1 1 4 3 Fiction 6 0 7 1 3 18 12 4 12 5 5 Total 9 0 15 15 3 21 29 20 23 15 8 Cobb thus concluded that "even the largest plausible amounts of free reading will not take the learner very far in the 3000-family zone" (p. 44). CRITIQUE AND REINTERPRETATION OF COBB'S DATA Cobb’s analysis suffers from two major problems. First, the amount of reading that Cobb proposes as "optimistic" is, in fact, pessimistic in the extreme. The number of words read is a product of time spent reading and reading rate. Table 4 summarizes the results from 11 studies that have reported L2 reading rates with readers from a variety of L1 backgrounds in both EFL and ESL settings. Fraser (2007) summarizes the results of several studies included in Table 4 (Cushing-Weigle & Jensen, 1996; Haynes & Language Learning & Technology 105 Jeff McQuillan and Stephen D. Krashen A Response to Cobb (2007) Carr, 1990; Nassaji & Geva, 1999; Oller & Tullius, 1973; Taguchi, 1997), and data reported from these studies are taken directly from her Appendix A. For one study (National Institute for Literacy, 2003), oral reading rates were used. Studies are ordered by average reading rate in words-per-minute. L2 reading proficiency is based on the researcher’s own classification of the students’ levels. Table 4. Average Reading Rates of L2 Readers Study Population Taguchi, Takayasu-Maass, & Gorsuch (2004) Haynes & Carr (1990) 1st year college EFL Hirai (1999)1 National Institute for Literacy (2003)2 Taguchi & Gorsuch (2002) Taguchi (1997) Fraser (2007)3 – China Group 1st – 3rd year college EFL Adult ESL Fraser (2007)3 – Canada Group 1st-4th year undergraduate ESL Undergraduate EFL Graduate ESL Undergraduate & graduate ESL/EFL Cushing-Weigle & Jensen (1996) Nassaji & Geva (1999) Oller & Tullius (1973) 1 Undergraduate EFL 1st year college EFL Undergraduate EFL 3rd year undergraduate EFL L2 Reading Proficiency Beginning Average L2 Reading Rate 83 wpm Intermediate/ Advanced Various Beginning 86 wpm Beginning Beginning Intermediate/ Advanced Intermediate/ Advanced Advanced Advanced Intermediate/ Advanced 115 wpm 127 wpm 135.5 wpm 87.5 wpm 102 wpm (oral) 140.4 wpm 158 wpm 179 wpm 206 wpm Table 1, whole group score Cohort 11 of all ESL readers 3 Table 1, Task 4 – Learning 2 It should be noted that these studies probably underestimate reading rates achieved during free reading. The texts used to determine reading rate in all cases were selected by the researcher, and thus may have been too difficult for the reader or on a topic about which the reader lacked sufficient background knowledge. It seems likely that students engaged in free reading, where the text is self-selected and thus probably a closer fit for the reader’s proficiency and background knowledge, would read at a faster rate. It is clear from Table 4 that L2 reading rates vary widely, with more proficient readers reading faster than less proficient ones. We conservatively choose 100 wpm as an average reading rate for our analysis, which is slightly below the average rate for readers at a beginning level of L2 reading proficiency for the studies included here (106.8 wpm). An adult L2 reader reading at a speed of 100 words-per-minute would take 1,750 minutes to go through 175,000 words of text. That is the equivalent of 29.2 hours of reading which, over the course of two years of language study, would amount to a mere 2.4 minutes of free reading per day. What Cobb is actually demonstrating is that a very small amount of reading over a period of 12 to 24 months would not be sufficient to make one a fluent L2 reader. Second, an examination of the "total" columns in Tables 2 and 3 reveals that a reader who read newspaper, academic, and fiction texts (all three subsets), for a total of about 517,000 words, would easily pass even the more demanding criterion of ten encounters for all of the words at the 2000-family level, and for eight of the ten at the 3000-family level. An L2 acquirer reading 100 words-per-minute would be able to accomplish this in a little more than 86 hours, or at the rate of one hour per day, the equivalent of a single academic quarter (approximately 13 Language Learning & Technology 106 Jeff McQuillan and Stephen D. Krashen A Response to Cobb (2007) weeks). Free reading across a variety of genres can indeed give you the necessary vocabulary for adultlevel fluency. The contrast between this estimate and Cobb’s is presented in Table 5. Table 5. Two Estimates of Amount of Reading (at 100 words per minute) Estimate Cobb McQuillan & Krashen Words 175,000 517,000 Hours 29.2 86.2 Daily 2.4 minutes over 2 years 60 minutes over 1 academic quarter FREE READING: MORE POWERFUL THAN WE THOUGHT? What is surprising about Cobb’s data is just how powerful free reading really is, even at the minimal levels he used. Even if a reader stuck to one genre, and read as little as Cobb suggests, a lot would be accomplished. With just 100 pages of newspaper text alone, for example, one can make significant progress toward the 2000-family level. Cobb’s analysis shows that you would have sufficient encounters for acquisition of nine of the ten sample word families. Similar progress could be made by reading the equivalent of six books the length of Alice in Wonderland, which, while perhaps insufficient for academic purposes, is very impressive. A reader who dedicated a modest 20 minutes per day to free reading would, over Cobb’s hypothetical two years of language study, would encounter 1,460,000 words – a substantial number, more than eight times the number of words in Cobb’s estimate. It seems likely, then, that this amount would allow one to reach even the 5000-word level. Free reading may not be sufficient to meet the needs of all demanding academic or specialized texts, although nothing in Cobb’s analysis would preclude that possibility. Further research should take into account a more realistic estimate of the volume of reading by the typical L2 acquirer. Cobb has shown us, however, that for a modest investment in time, free reading does appear to be more than adequate to reach the vocabulary levels that he argues are necessary for a fluent L2 reader. ABOUT THE AUTHORS Jeff McQuillan is a Senior Research Associate at the Center for Educational Development in Los Angeles, California. Email: [email protected] Stephen D. Krashen is Emeritus Professor of Education at the University of Southern California in Los Angeles, California. Email: [email protected] REFERENCES Cobb, T. (2007). Computing the vocabulary demands of L2 reading. Language Learning & Technology, 11(3), 38-63. Retrieved October 7, 2007 from http://llt.msu.edu/vol11num3/cobb. Cushing-Weigle, S., & Jensen, L. (1996). Reading rate improvements in university ESL classes. CATESOL Journal, 9, 55-71. Language Learning & Technology 107 Jeff McQuillan and Stephen D. Krashen A Response to Cobb (2007) Fraser, C. (2007). Reading rate in L1 Mandarin Chinese and L2 English across five reading tasks. The Modern Language Journal, 91(3), 372-394. Haynes, M., & Carr, T.H. (1990). Writing system background and second language reading: A component skills analysis of English reading by native speakers of Chinese. In T.H. Carr & B.A. Levy (Eds.), Reading and its development: Component skills approaches (pp. 375-418). San Diego, CA: Academic Press. Hirai, A. (1999). The relationship between listening and reading rates of Japanese EFL learners. The Modern Language Journal, 83(3), 367-384. Krashen, S.D. (1989). We acquire vocabulary and spelling by reading: Additional evidence for the Input Hypothesis. The Modern Language Journal, 73, 440-464. Krashen S.D. (2004). The power of reading, 2nd edition. Portsmouth, NH: Heinemann. McQuillan, J. (1998). The literacy crisis: False claims, real solutions. Portsmouth, NH: Heinemann. Nassaji, H., & Geva, E. (1999). The contribution of phonological and orthographic processing skills to adult ESL reading: Evidence from native speakers of Farsi. Applied Psycholinguistics, 20, 241-267. National Institute for Literacy (NIFL). (2004). Adult reading components study. Washington, DC. Retrieved October 7, 2007, from http://www.nifl.gov/readingprofiles/FT_Browse11.htm. Oller, J. W., & Tullius, J. R. (1973). Reading skills of non-native speakers of English. International Review of Applied Linguistics, 11, 69–79. Taguchi, E. (1997). The effects of repeated readings on the development of lower identification skills of FL readers. Reading in a Foreign Language, 11, 97-119. Taguchi, E., & Gorsuch, G. (2002). Transfer effects of repeated EFL reading on reading new passages: A preliminary investigation. Reading in a Foreign Language, 14(1), 43-65. Taguchi, E., Takayasu-Maass, M., & Gorsuch, G. (2004). Developing reading fluency in EFL: How assisted repeated reading and extensive reading affect fluency development. Reading in a Foreign Language, 16(2), 70-96. Language Learning & Technology 108 Language Learning & Technology http://llt.msu.edu/vol12num1/cobb February 2008, Volume 12, Number 1 pp. 109-114 COMMENTARY: RESPONSE TO MCQUILLAN AND KRASHEN (2008) Tom Cobb Université du Québec à Montreal I was glad to receive a response from Jeff McQuillan and Stephen Krashen to my piece "Computing the Demands of Vocabulary Acquisition from Reading" (Language Learning & Technology, October, 2007), because drafting a reply forces me to be even more clear about what I am saying. I was initially surprised to see the lead responder was an expert in first language (L1) reading rather than second (L2) but on second thought McQuillan’s (The Literacy Crisis, 1998) participation makes sense. Position Review I argued that building an adequate functional L2 lexicon for reading from reading alone (Krashen’s longstanding position) cannot be done by the majority of learners in the normal time frame of instructed L2 learning. An example of such a time frame would be the year or two of ESL preparation granted to foreign students on arrival in a North American university. A minimal functional lexicon is 3,000 word families, which provides about 90% known-word coverage of average texts. But lexicon building from reading alone will stall shortly after 2,000 families. This happens for the demonstrable reason that 3,000level words (and other less frequent words) do not appear often enough in the amount of reading of natural texts that such learners are likely to accomplish. Research has shown that words need to appear minimally six times for learning to take place. As proof I offered three samples of natural text at what I proposed was the outer limit of such an amount, namely any of the journalism, academic, or literary sub-corpora of the Brown corpus. Each of these amounts to about 175,000 words, of which 10% are words beyond the 2,000-most-frequent level (minus proper nouns). Through elementary corpus analysis, I showed that a learner who managed to read any one of these collections would meet no more than half of the third thousand word families six times apiece. A similar analysis of the collected works of a major author (300,000+ words) and another of an entire set of graded readers (375,000+ words) pointed to the same conclusion: reading these texts in their entirety cannot provide enough repeated exposures to enough 3,000-level vocabulary to support the acquisition of a minimal functional lexicon. Critics’ Response I would have expected a critique of this analysis to focus on the assumption that most of the words in a text need to be known for reading to proceed successfully, given the abiding belief that learners can easily expand their vocabularies by guessing new word meanings from context, as was assumed but never shown in many classic accounts of the reading process. So I was surprised that the critics’ actual problem was with the claim that L2 learners would have trouble reading 175,000 words of fairly difficult natural text in a year or two. Doing the math, McQuillan and Krashen propose that even "reading relatively slowly at a speed of 100 words-per-minute," (p. 106) L2 learners should be able to read 175,000 words in 1,750 minutes or 29.2 hours, which, spread over two years, "amounts to only 2.4 minutes of free reading per day" (p. 106). Such readers would make light work of any of the Brown sub-corpora, or indeed all three of them. As a teacher and coordinator of many L2 reading courses and programs, I wondered if we were talking about the same world. In my experience, even strong ESL readers find small amounts of unsimplified text fairly hard going, particularly if the text type is expository rather than narrative. I myself after decades of working in and around French cannot get through Le Monde before the next edition is on the doorstep. Indeed, expository academic or journalistic text has always been the stuff of the intensive reading course, Copyright © 2008, ISSN 1094-3501 109 Tom Cobb Response to McQuillan and Krashen (2008) wherein a good deal of scaffolding is provided and no great volume of text actually gets read. Were my critics and I talking about the same kind of learner and the same kind of reading? The Reading Rate Research To support their position, McQuillan and Krashen cite a half dozen reading rate studies, conveniently gathered in the literature review of a recent study by Fraser (2007). In this research and Fraser’s own study, L2 reading rates of 100 words-per-minute (wpm) and even somewhat higher appear to be the norm. But a little digging below the numbers raises questions about their applicability to the matter at hand. The first thing to note about the Fraser (2007) study is that while my critics use its results to establish how much L2 readers can read, the researcher herself interprets the data to show how slow and arduous L2 reading is even for experienced readers, and how it remains so for long periods, even for those living and studying in the L2 culture. The second point is the nature of the participant groups in Fraser’s study, neither of which greatly resembles a group of learners who are at the point of taking on the third thousand words as a prelude to academic reading. One consists of students who have specialized in English language and literature (English majors in a Chinese university), while the other consists of learners well into their studies in a Canadian university; some of them had lived in Canada for as long as 12 years (2007, p. 378). Third and most important is the nature of the material Fraser’s subjects read at the cited rates of 135 and 140 wpm. Fraser reports that in terms of grade equivalent level, the two experimental texts were found to be suitable for use with Grade 9 or 10 high school students. Her analysis using Vocabprofile confirms their non-university character; the frequency profiling revealed large proportions of first 2,000 level lexis -- 83% in one text and 86.8% in the other (p. 394). These proportions of basic lexis are substantially higher than those consistently found in more typical university-level texts, as represented by, say, the academic section of the Brown corpus. Table 1 shows randomly chosen profiles from segments of this corpus. The mean coverage of the first 2,000 words in the Brown texts is only 78.53%, with a very small standard deviation (3.01%). The differences between the Brown mean and the means of Fraser’s two experimental texts (5% and 8%) may seem minor, but from the L2 reader’s perspective, the added load of 5% more non-basic vocabulary means one more ‘hard’ word per two lines of text. An added 8% means one more in almost every line. Table 1. Lexical Frequency Profiles across Disciplines (coverage percentages). Brown Segment Discipline No. of words 1000 2,000 1000 + 2,000 AWL J32 J29 J26 J25 J22 J12 J11 MEAN SD Linguistics Sociology History Social Psychology Development Medicine (anatomy) Zoology 2031 2084 2036 2059 2023 2024 2026 73.51 74.23 69.3 73.63 76.42 71.05 75.12 73.32 2.42 8.37 4.75 5.7 3.11 4.55 3.80 6.17 5.21 1.74 81.88 78.98 75.00 76.74 80.97 74.85 81.29 78.53 3.01 12.60 13.44 14.49 14.38 12.26 6.72 7.31 11.60 3.24 1K + 2K + AWL 94.48 92.42 89.49 91.12 93.23 81.57 88.60 90.13 4.30 Notes: (1) Table from Cobb & Horst, 2004. (2) Segments from the Brown corpus are described in the Brown University website accessible from the Compleat Lexical Tutor at http://www.lextutor.ca. (3) AWL stands for Academic Word List. Language Learning & Technology 110 Tom Cobb Response to McQuillan and Krashen (2008) To summarize, the experimental materials were easy texts for these learners. As we will see, the other reading rate studies from Fraser (2007) that McQuillan and Krashen cite are similarly inapplicable to the question under discussion. Either the readers were too advanced to be considered typical classroom ESL learners, or the reading materials were much simpler than those specified in my original paper, or both. But before examining these studies more closely, it may be useful to remind ourselves of the kind of learners and texts this discussion is about. At issue are the many ESL and EFL learners worldwide who are trying to move beyond a basic 2,000-word lexicon, by reading texts that contain significant amounts of post-2,000 lexis. Such learners and texts are not hypothetical but typical. In the case of learners, Laufer’s (2000) review of seven vocabulary size studies showed that university entry level ESL/ EFL learners in three Asian, two Middle Eastern, and three European countries were working with an average 2100 known word families (SD 977). In the case of texts, those bearing post-basic vocabulary are common if not the norm in academic and professional contexts; as mentioned, the Brown academic corpus bears an average 10% (SD 3%), that is, one post-2,000 word, on average, in every line. As I have already shown, Fraser’s (2007) reading rates are simply not applicable to the learners and texts in question. Nor are the four other main sources of rate evidence reviewed in her study and cited by McQuillan and Krashen. First, Nassanji and Geva’s (1999) study of 60 Farsi-speakers hardly pertains to ESL learners at the beginning of their academic studies. They were in fact "graduate students at a major Canadian university… who had been living in Canada for 3 to 6 years" (p. 246). Their mean comprehension scores on the Nelson-Denny Reading Test, a measure designed for L1 readers, were over 50% (p. 251). The several Taguchi studies cited (e.g., Taguchi, Takayasu-Maass, & Gorsuch, 2004) all involve repeated readings of simplified texts from the Heinemann New Wave series of graded readers, with an upper limit of 2200 word families (Hill, 1997). Hirai’s (1999) study involves an experimental reading task that measures not reading rate but rather "rauding" rate, following Carver’s (1990) notion that reading rate is best measured using texts that the readers find easy to read. Similarly, Haynes and Carr’s (1990) study used a text chosen specifically because of its familiar topic and the fact that it "contained numerous 'lexical familiarizations'… definitions, examples, stipulations, synonyms, paraphrases, illustrations, etc., which the author had provided to clarify the meaning of new terms introduced in the text" (p. 396). It goes without saying that not all authentic academic texts would be so lexically familiarized. Thus the L2 reading rate research cited is not applicable to academic reading. Even if it were, simple multiplication of reading rate times hours and days would only tell us what learners might be able to read in principle. This would still have to be confirmed in studies of what some particular learners had read in fact. Therefore I returned to some of the research literature where I thought I remembered specific amounts of L2 reading having been documented. Was any of it (a) of the type we are talking about and (b) as much as McQuillan and Krashen claimed should be possible? The Amount-of-Reading Research Not all extensive reading studies produce such useful specifics as numbers of words or pages read, but there are some. The largest amounts of L2 reading on record seem to hail from Japanese contexts. One is Rosszell’s (2007) doctoral study, in which university learners read an average 40 pages per week, which at 300 words per page, 12,000 words per week, 40 weeks per year, amounts to almost half a million words in a year. This is in the order of the rate McQuillan and Krashen hypothesize. However, the texts that these learners were reading were graded readers from Oxford’s Bookworm series, levels 4 to 6 (6 is highest). As mentioned in my original paper, this type of text does not include adequate inputs from the third thousand level and indeed makes no claim to. An even larger amount of reading was recorded by Beniko Mason (2004). Her study involved 18-year old English majors at a junior college in Japan reading fully 1000 pages, or 250,000 words, per semester (although not all participants were able to meet these targets). Extrapolating this reading to four terms or Language Learning & Technology 111 Tom Cobb Response to McQuillan and Krashen (2008) two years, the amount of reading would indeed seem to be about 1 million words. This is the kind of figure McQuillan and Krashen are talking about, and it is equivalent to the size of the entire Brown corpus. But in fact, it was not the Brown corpus or anything resembling it in lexical composition that the learners were reading. Again, all this reading is of graded or simplified texts – much of it at the very elementary 600 word level. Certainly, there is nothing wrong with reading simplified texts! But learners reading large amounts of such text do not make the point that McQuillan and Krashen wish to make. In fact, reading simplified texts is a very good thing for language learners to do, for many reasons including but not limited to increasing vocabulary (Horst 2005; Pigada & Schmitt, 2006). Indeed, the second part of my LLT piece described ways of using technology to expand the library of graded materials that are accessible to ESL teachers and learners. Text computing can help us expand the range (to include more expository material) and vocabulary level (to provide a smooth rise up to a vocabulary size of 3,000+ word families) of available graded materials. At present, there is no smooth rise. Rather, "there exists a wide gap between the highest level of graded readers and the vocabulary demands of academic text and unsimplified novels" (Nation, forthcoming, p. 1). Even the best graded reader series, e.g., Oxford’s Bookworm series, make no claims beyond 2,500 words. The Longman Bridge Series (1945) was a systematic grading of materials up to 8,000 words, but it is long out of print. The new Penguin/Longman Active Reading series may claim successor status to Bridge with its 3,000 word-family target, but none of the studies I located had used this series. In other words, the large amounts of reading reported in some of the published research is reading of a type that by definition cannot be the route to an adequate functional reading lexicon. But are there no studies involving the reading of unsimplified texts? Given the number of learners worldwide who are trying to improve their reading ability for advanced study through English, it is surprising how little research addresses this common objective. An exception is work by Parry, who looked at academic ESL learners reading large amounts of unsimplified texts in credit courses at U.S. universities. A preliminary goal of these studies was to estimate how many words the learners had managed to read. In a case study of two learners, one read as few as 7,500 words of an assigned anthropology text over a complete term, while another read as many as 72,000 words from the same textbook (Parry, 1997). At first glance, the second reader’s rate lines up nicely with McQuillan and Krashen’s estimate. If 72,000 words of academic text can be read in a term, and two years is four terms, and the learner is taking four such courses at a time, then he is reading over one million (72,000 x 4 x 4 = 1,152,000) words of natural academic text in two years. This is the size of the whole Brown corpus and then some, and its lexical composition is probably similar. So can we conclude that some ESL learners can read at the rate McQuillan and Krashen propose, and presumably experience the vocabulary growth that goes with it? Not exactly. At the start of the experiment, Parry asked her two readers to write down all the words they thought were new or difficult while reading, along with the page number. Then at the end of the academic session the readers were asked to provide the meanings of words they had noted, first out of context and then in the page context where the word was first noticed. Out of context, the 72,000-word reader could remember having seen only 5% of the words he had originally noted, and with the help of the context could provide correct or partly correct meanings (in his L1) for only 28%. The 7,500-word reader, on the other hand, could remember seeing 29% of her words and could give correct or partly correct meanings for 63%. In other words, the fast reader was reading a rather large amount but with very little vocabulary growth, while the slow reader was not getting her reading done but was learning some of the new words in the small amounts she did manage to complete. Language Learning & Technology 112 Tom Cobb Response to McQuillan and Krashen (2008) These and a number of similar Parry studies are small but nonetheless, they ring true for anyone who has taught academic ESL reading in North America or elsewhere. Learners in such courses typically struggle to get through a few pages - or else read quickly with low comprehension and almost no vocabulary growth. Conclusion: Not Reading Alone Parry’s case studies suggest that most academic ESL learners cannot read their way to an adequate functional second lexicon. The scale of her research is not big enough to be conclusive, but there is no evidence I know of to contradict it. Also, Parry’s findings are congruent with some other well established evidence. Replicated research shows that reading becomes arduous and comprehension suffers when unknown word densities exceed 5% (e.g. Laufer, 1989). Nation (2006) sets the ideal criterion as low as 2%. But academic readers with knowledge of about 2,000 words are reading texts bearing at least 10% unknown items, and as their ESL teachers can attest, their pain is all too real. If McQuillan and Krashen have relevant counter-evidence to any of this, I welcome it. Then the discussion can proceed on a different basis. Until then, the adequacy of free reading is an idea with high credibility in the time frame of L1 acquisition, and some credibility in an extended time frame of L2 acquisition under conditions of exceptional motivation. But carried into the typical time frame of instructed L2 acquisition, it is an idea that grossly misrepresents the problems faced by L2 readers who need to read to learn in their second languages. For these learners, an adequate second lexicon will not happen by itself; it will be provisioned through well-designed instruction including but not limited to reading. ABOUT THE AUTHOR Tom Cobb began his career in English literature but soon moved toward language study and instruction. He has taught and coordinated reading and writing courses at a wide range of levels in the United Kingdom, Saudi Arabia, Oman, Hong Kong, Australia, New Zealand, Japan, Mexico – and, of course, Canada. He currently trains TESL trainees in the uses of computing in language learning at a Frenchlanguage university in Montreal. Email: [email protected] REFERENCES Carver, R. (1990). Reading rate: A review of research and theory. San Diego, CA: Academic Press. Cobb, T. & Horst, M. (2004). Is there room for an AWL in French? In P. Bogaards & B. Laufer (Eds.), Vocabulary in a second language: Selection, acquisition, and testing (pp. 15-38). Amsterdam: John Benjamins. Fraser, C. (2007). Reading rate in L1 Mandarin Chinese and L2 English across five reading tasks. The Modern Language Journal, 91(3), 372-394. Haynes, M., & Carr, T.H. (1990). Writing system background and second language reading: A component skills analysis of English reading by native speakers of Chinese. In T.H. Carr & B.A. Levy (Eds.), Reading and its development: Component skills approaches (pp. 375-418). San Diego, CA: Academic Press. Hill, D. (1997). Graded (basal) readers -- Choosing the best. The Language Teacher Online, 21(5). Retrieved January 12, 2008, from http://www.jalt-publications.org/tlt/files/97/may/choosing.html. Language Learning & Technology 113 Tom Cobb Response to McQuillan and Krashen (2008) Hirai, A. (1999). The relationship between listening and reading rates of Japanese EFL learners. The Modern Language Journal, 83(3), 367-384. Horst, M. (2005). Learning L2 vocabulary through extensive reading: A measurement study. Canadian Modern Language Review, 61(3), 355-382. Laufer, B. (1989). What percentage of text-lexis is essential for comprehension? In C. Lauren & M. Nordman (Eds.), Special language: From humans thinking to thinking machines (pp. 316-323). Clevedon, UK: Multilingual Matters. Laufer, B. (2000). Task effect on instructed vocabulary learning: The hypothesis of 'involvement.' Selected Papers from AILA ’99 Tokyo (pp. 47-62). Tokyo: Waseda University Press. Mason, B. (2004). The effect of adding supplementary writing to an extensive reading program. International Journal of Foreign Language Teaching, 1(1), 2-16. McQuillan, J. (1998). Literacy crisis: False claims, real solutions. Portsmouth NH: Heinemann. Nagy, W.E., & Anderson, R.C. (1984). How many words are there in printed school English? Reading Research Quarterly, 19(3), 304-330. Nassaji, H., & Geva, E. (1999). The contribution of phonological and orthographic processing skills to adult ESL reading: Evidence from native speakers of Farsi. Applied Psycholinguistics, 20(2), 241-267. Nation, P. (forthcoming) New roles for L2 vocabulary? In L. Wei & V. Cook (Eds.) Language teaching and learning. Contemporary Applied Linguistics Series. London: Continuum International. Nation, P. (2006). How large a vocabulary is needed for reading and listening? In M. Horst and T. Cobb (Eds.), [Special Issue on Second Language Vocabulary Acquisition]. Canadian Modern Language Review, 63(1), 59-81. Parry, K. (1997). Vocabulary and comprehension: Two portraits. In J. Coady & T. Huckin (Eds.) Second language vocabulary acquisition (pp. 55-68). Cambridge, UK: Cambridge University Press. Pigada, M., & Schmitt, N. (2006). Vocabulary acquisition from extensive reading: A case study. Reading in a Foreign Language, 18(1), 1-28. Rosszell, R. (2007). Extensive reading and intensive vocabulary study in a Japanese university. Unpublished doctoral dissertation, Temple University, Japan. Taguchi, E., Takayasu-Maass, M., & Gorsuch, G. (2004). Developing reading fluency in EFL: How assisted repeated reading and extensive reading affect fluency development. Reading in a Foreign Language, 16(2), 70-96. Language Learning & Technology 114