1There has been much debate about whether and how to correct students’ written productions. The context of this study is that of academic writing and writing for professional and research purposes. The goal is for researchers to achieve an acceptable draft even if it is not native-like. At the same time, it seems to us that accuracy is important. Sentence-level errors may distract the reader and give a poor impression, while more global errors may hinder comprehension and lead to rejection of a research manuscript on the grounds of “poor English”. As Master claims,
[...] local errors such as articles, prepositions and subject-verb agreement, even though they do not affect understanding, can have a negative impact on the reader, especially non ESL faculty, as they have the impression that the writer has inadequate control of the language (1995: 184).
2In writing courses, work can be conducted on the process of writing: drafting, revising, reformulating. Peer feedback may be used so that the students themselves identify the errors and edit the text collaboratively, leading to discussion of the clarity of the draft. In the classroom situation, the teacher’s written feedback is often indirect consisting of underlining or using a code to highlight errors, so that students will revise their draft once their attention has been drawn to certain elements. However, specific writing courses do not exist in all institutions and there may be little class-time available for this revision process. The public we are dealing with, doctoral students and researchers, have very little time to devote to the learning of English and assistance is required directly as they draft their article or letter. In this paper we describe how non-native speaker (NNS) students’ and researchers’ initial drafts are used as part of an on-line tool to facilitate the writing process.
3From our experience of teaching writing courses and re-reading letters and articles for students and researchers at our institution, it has become apparent that both novice writers and experienced researchers make typical language errors. There would seem to be a common core of errors of verb form, article use, prepositions, vocabulary, for example, present in different authors’ drafts and which may recur over time. Birch’s findings (Birch 1996) concerned French speakers and Webber hierarchised errors found in Italian writers’ drafts (Webber 1993) but this may be true for several national language groups. Lennon defines an error as
a linguistic form or combination of forms which, in the same context and under similar conditions of production, would, in all likelihood, not be produced by the speaker’s native speaker counterparts. (1991: 182)
4A tool to aid novice writers should therefore focus on local errors and grammatical acceptability while at the same time illustrating more appropriate lexical and grammatical choices and reformulations. Students’ or researchers’ initial drafts can thus be used as pedagogical input and the errors can then be seen as
Something normal and important for learning to occur, a ‘window’ on the learner’s internal grammar, a learning strategy perhaps necessary to promote second language acquisition. (Selinker 1992: 144)
5Attention can be drawn to specific L1 interference or developmental errors and as well as providing the correct form, explanations and comments may be added.
6The tool described here, TYOS® (“Type Your Own Script”), will be launched in 2009. Since it will be for free use online, the authors hope that users will recommend it to other users. No special form of promotion has been envisaged. The tool is based upon a corpus of NNS drafts covering various genres of scientific writing: cover letters, letters of request, letters of complaint, replies to reviewers, abstracts, research articles and case reports. All the texts were written by authors from our institution who requested editorial assistance with the work. The fields covered include medicine, biology, dentistry, psychology, wine science and biochemistry. Although some characteristics of scientific writing are discipline-specific, the aim is to highlight structural and linguistic features likely to occur in discourse from as wide a range of disciplines as possible. The drafts have been analysed and processed didactically on several levels as models, which can then be imitated by both accomplished and novice researchers, and by ESP students either as a stand-alone application or in a classroom setting. While it has been designed on the basis of a corpus produced by Francophone authors, its generic nature makes it usable by any language group. A full description has been given elsewhere (Cooke & Birch 2008).
7Several versions of each corrected text are available so that the learner can focus on different elements. For example, the learner may choose to focus on the rhetorical organisation of the letter or text, so the functions of various key sentences are highlighted to explain the structure of information. In the other versions, attention is drawn to interesting linguistic features. The user scrolls down and may choose to study for example “useful phrases and expressions” in the complete context of a corrected draft. These typical expressions are particularly important in letter writing and the ‘drag and drop’ option means that the user may begin writing his own draft by imitating certain features of the model texts and re-using certain expressions. “Verb forms” are also highlighted and commented upon as this has been found to be problematic in student drafts. The other versions illustrate “interesting grammar points”, “discourse linkers” and “vocabulary points”, all elements which from our re-reading and revising of texts have been found to pose problem for ESP students.
8A basic function of TYOS® is that the user also has access to the NNS author’s initial draft and the version corrected by an NS corrector. Typical errors can therefore be seen in context and the learner can focus on the corrections, reformulations and improvements made to the draft and the actual writing process.
9An example produced by a doctoral student and contained in TYOS® is given here. While clearly incorporating successfully a number of structural features learned from the previous lesson of a scientific writing course, this student’s written assignment contains a number of local errors that, when corrected, provide further food for thought for him. Of course, this particular draft is already fairly well written but the methodology for processing texts in TYOS is generic, so an original or poorer quality would still provide appropriate input for the user. In the remarks below, we set aside the interesting but largely subjective debate about what may be considered as a more or less serious error.
Figure 1 : Original version
10[Image non convertie]
Figure 2: Version corrected or modified with further suggestions
11[Image non convertie]
12In the demonstration given here, the actual comments are not presented as they are on the active screen in TYPS®. In the latter, the comments float and can be copied and pasted to a word processing application.
13Comment 1 – the lack of capital letter – concerns a very basic rule that, even if broken, in no way hinders effective communication, even though it may overall provide an impression of incorrectness and inadequate mastery of writing conventions.
14On the other hand, comment 2 focuses on an issue that poses considerable problems for NNS students and researchers: establishing whether adjectives can be compounded in front of nouns and, if so, in which order. In this case, it would seem that the author of this letter was aware of the problem but preferred to err on the side of caution. Nevertheless, the result is problematic since it denotes to the native speaker a lack of awareness of a linguistic convention in science. This in turn may generate the feeling for the native speaker that the author somehow does not quite belong fully to the community of which he would clearly like to be a part. Hence, NNS writers need to be helped to pay special attention to such a feature in English.
15As for comment 3, it highlights a typical interlanguage difficulty for Francophones: the problem of rendering the notion of “lequel” when used as an adjective. Discussed collectively in class, this difficulty could provide much useful feedback for the author and others in the group.
16Comment 4 can be rated somewhere near comment 1 in terms of the minor hindrance that it represents to the native speaker. Yet a point like this is easily teachable and retainable for the learner. If the latter succeeds in seeing the difference between the ways in which the idea of “afin de” is expressed at the beginning of a sentence and within it, then a significant piece of “micro-information” will have been learnt.
17Finally, comment 5 simply represents a suggestion for smoother and more pertinent expression, and as such is more anecdotal.
18Even if the author of such a letter were to know how to circumvent these difficulties, his written production might still not qualify as quasi-native speaker discourse. Yet the objective of achieving the latter perhaps misses the real point: the real need is for NNS authors to be able to control their discourse in such a way that it may be deemed acceptable to the members of the community for whom it is written. That definition offers scope for the production of defective but effective discourse.
19As mentioned above, there has been much debate recently about the efficacy of error correction in the learning process. Truscott (1996; 2007) has controversially claimed that correction is ineffective and even harmful and in a recent study asserts that “successful error reduction during revision is not a predictor of learning” (Truscott & Hsu 2008: 292). However other authors such as Ferris, while recognising the need for controlled and longitudinal studies to evaluate the efficacy of correction, have responded that...
recent second language acquisition (SLA) research on focus and form (in both written and spoken language) strongly suggests that adult second language acquirers in particular need their errors made salient and explicit to them so that they can avoid fossilization and continue developing linguistic competence (2004: 54).
20Ferris continues by saying that “students who receive feedback on their written errors will be more likely to self-correct them during revision than those who receive no feedback” (2004: 56). This point of view is also supported by Bitchener (2008) who found in his study that written corrective feedback had a significant effect on improving accuracy concerning the use of articles, and that this effect was retained two months later. In an earlier study, Bitchener et al. (2005) found that explicit written feedback and conference feedback led to greater accuracy with the past simple tense and the definite article in new pieces of writing. It is pointed out however that...
significant variations in accuracy across the 4 pieces of writing support earlier SLA discoveries that L2 learners, in the process of acquiring new linguistic forms, may perform them with accuracy on one occasion but fail to do so on other similar occasions. (Bitchener et al 2005: 191)
21In conclusion, an important part of our writing courses is to spend time on correction of drafts and peer feedback on texts to raise awareness of typical errors and to draw attention to certain features of texts. TYOS® aims to do the same by highlighting problems in initial drafts and the corrector’s changes and revisions. It is hoped that the various features of the tool will give users more autonomy in writing and learning scientific English. Following the launch of the tool, further studies will be needed to investigate how learners actually use the tool: how they browse through the different versions of a text, the sequences used, the information they discover. It will also remain to be seen how efficient the tool is in terms of successful written submissions and more specifically the efficacy of highlighting errors and corrections in a draft. In particular, it will be interesting to investigate how users interact with the screen shots in TYOS® and to what extent their writing is improved by it.