UNPACKING L2 WRITING RESPONSES: A CORPUS-BASED STUDY ON TEACHER FEEDBACK TO STUDENT WRITING

Over the past decades, corpus linguistics has becom widespread in studies dealing with applied linguistics. Teachers have become acquainte d with corpus lingustic methods and are using computer technology in their professional pra ctice. A well-known example is the learner corpora with which researchers managed to attain in valuable results concerning various aspects of learner language. This article, however, present s a new field within corpus linguistics: the teacher corpus. A corpus on teacher language (inste ad of learner language) has a lot to offer in terms of methodology and pedagogy. Introduction Computer corpora have gained a foothold in the fiel d of language teaching and other language related fields so much so that they have permeated all walks of language educational life. This means that, for instance, good dictionaries are inc onceivable without relying on a renowned collection of spoken and written texts (e.g. Hornby , 1997; Sinclair, 1987a; 1987b). Corpusbased interlanguage L2 studies have resulted in eff ective pedagogical approaches (Kaszubski, 1998; Ringbom, 1998). The world of corpora can prom ote effective self-directed learning (Szirmai [2001], for example, presents an account o n the corpus-based software Contexts). On the other hand, speakers of a foreign language can an lyse their own language both in and outside classroom in order to get authentic data ab out what they or others are doing well and what needs to be improved. These are but a few poss ibilities that corpus linguistics has to offer. Some teachers of ESL writing began to use corpus li nguistics, not surprisingly, as they recognized the potential benefits of storing and us ing language for pedagogical purposes. The field of Second Language Writing (SLW) is drenched with corpus-related methodology touching upon every field such as English for Acade mic Purposes (e.g. Coxhead & Byrd, 2007), feedback-giving (e.g. Horváth, 2002), data-d riven learning (e.g. Granger & Tribble, 1998) and several others. The present paper deals w ith the teacher corpus: a collection of language data produced by teachers of EFL. The corp us is a collection of teacher comments that can be found at the bottom of the student pape rs, usually in red. The paper is organised as follows: the background s ection will shed some light on the rationale of teaching second language writing: the implications for classroom use and the aims teachers and students may have. Evaluating stu dent papers and genre-pedagogy will be in our focus as these have influential role from ou r corpus’s point of view. Then, after the brief consideration of the background issues on whi ch this corpus is based, the presentation and practical application of the PAC corpus will fo llow. Finally, the discussion of the results and the conclusion will end the paper. Background: objectives and implications of second l anguage writing Teachers often have the responsibility of having to prepare their students for externally validated exams such as the school leaving examinat ion. This kind of examination has its own requirements concerning writing – at both advanced and ordinary levels. The requirements are explicitly stated and are no secret, therefore the recipe seems to be easy: practice, practice and practice. This might be true, but one might encount er the potential pitfall of plunging into excessive writing practice without proper preparati on on the teacher’s part. Ferris (2007) argues that there are some options for teachers wis hing to apply effective responses. The first is that they “consult the course rubric or grading criteria if such instruments exist” (p. 170). The school leaving exam, obviously, has this gradin g scheme, and it is available for teachers and students alike (see Appendix A for the writing assessment scheme of the Hungarian school leaving exam evaluation sheet). It would be reasonable to say that if teachers see th aspects they will evaluate (the correctness of grammar, vocabulary, style, etc.) an d the weight each aspect carries compared to other aspects, they might organise their work ac cordingly. If the school leaving exam requires students to write a letter to their friend s in which they turn down an invitation, or a formal letter which enquires about summer courses i n Br tain, one might anticipate that the essential points such as formality and informality, paragraphing, vocabulary or grammar will be covered as well. However, it seems that the focu s f attention has shifted to the importance of grammar (e.g. Zamel, 1985). A contrastive rhetor ical study between L1 Hungarian and L2 English student writers showed that the observed Hu ngarian teachers of English, when it comes to evaluating writing, often resort to gramma tical correctness as their main focus (Kiszely, 2003). This way, surface level correction can carry more weight in the teacher feedback compared to other aspects such as style, a nd the writing evaluation practice can become lopsided. It is easier to correct deviations of hard-and-fast rules than correct elusive aspects such as style. Evidently, every teacher would immediatel y draw a red pen when seeing the sentence She didn’t give no kisses to him , because they know that formal rules do not allow the speakers of English to use two negatives in one sen tence. However, there are no exact rules for style, even, for ‘the impression on reader’ (wh ich is one of the school leaving exam categories). Therefore, teachers rarely put out fro m the safe shores of grammar. According to Ferris (2007), teachers have another t ool to rely on besides grade rubrics: the specific requirements of a given task. Obviousl y, letter writing requires other strategies than writing an argumentative essay. In this respec t, Hungarian teachers are doing what they are expected to do: make the students practise the genre they are expected to write in the school leaving exam. This is what Second Language A cquisition calls washback effect (Ellis, 1997). The rule of it is simple: if a is required, but b is not, then as a teacher I will make my students practice a, because I want them to know that, and no one will enquire about whether they know b. For instance, argumentative essay is virtually un known in the Hungarian school leaving exam as only letter writing is required. Th is is not to say that no information is disseminated about argumentative essays in classes, but dauntingly less compared to letter writing. Thirdly, “the teacher might also provide feedback t ailored to the needs and progress of individual students” (Ferris, 2007, p. 170). This i s what we call ‘differentiated pedagogy’ in Hungary, which is a panacea if one has time for it, as not only needs-based annotation is required, but needs-based writing lessons are also necessary. The introduction of genre-based pedagogies would re spond to some of the concerns of those teachers that see beyond the sway of letter w riting and experiment with teaching various genres. Identifying these genres that may be benefi cial or our students empowers teachers to teach them effectively. Knowing what to teach has c ons iousness-raising function, as the teacher’s responsibility here is to “illuminate the g nres that matter to students” (Hyland, 2007, p. 160), as well as to take into consideratio n the following methodology-related principles: Writing as a social activity : writing is a purposeful activity where intentions , purposes and audience are equally important. Writing is needs-oriented : teachers should keep in mind what kind of writing is needed in the students’ future situations. Explicit outcomes and expectations are needed : teachers should be explicit about what is expected from students and what the intended way of achieving that during the course. Learning to write is a social activity : teaching is a sequence of “scaffolded developmental steps in which teachers and peers pla y a major role” (Hyland, 2007, p. 153, the principles are based on Hyland, [2007, p. 152]). Teaching writing with the help of models is a commo n practice. The course books are filled with authentic writing material from which students can infer the characteristics of a genre. With appropriate teacher instruction, students can render these features into their own writing practice. Thus, it would be logical to think that L 1 materials as models are the main facilitators of effective writing instruction. That is, using these should enable students to create texts that are similar to the model texts in terms of characteristics and quality. The situation is not as simple as that, however. Gilqui n et al. (2007), when talking about the relationship between L1 corpora and English for Spe cific Purposes (ESP) as well as English for General Purposes material design, enunciated th at there are several problems with it, namely that (a) they are hardly corpus-informed, (b ) the corpora that are used in practice are native corpora, which are different in terms of err ors: native novice writers experience different difficulties than novice non-native write rs. A corpus can provide students with lots of authentic data, but differentiating between the ne ds of L1-L2 as well as novice and expert student writers is a key when using corpora as mode ls. The division between learner and native corpora is not a new idea. The International Corpus of Learner English (started by Granger, 1993 ; 1 94; 1996) is probably the best known example of a collection of learner language. Teache rs and students alike that are part of a corpus research project can familiarize themselves with the writing processes, the features of various genres and contribute to the enhancement of the texts stored through the synthesis of their previous work and the knowledge they inferred from corpus analyses.