The purpose of this article is to describe the way I tried to improve my method of correcting and commenting upon students' written work, and I sincerely hope that it will provoke thought and discussion around the issue of written feedback.

Before going any further though, let me briefly describe my own method of feedback, which consists of sentence-level corrections using a correction code and end of composition comments relating to higher-order concerns such as content and structuring, and let me also say that I consider a desirable approach to writing is one that stresses its recursive nature, involving constant planning, reviewing and editing.

In attempting to improve my method of providing written feedback, I chose a method called 'action research', involving critical reflection of my existing practice, finding out students' views via means of questionnaires and diary studies, and implementing changes in the light of these observations.

To summarise, some of my initial concerns relating to my method of I providing feedback were as follows:

• it took a long time to correct and annotate students' texts, but frequently the same mistakes were repeated in subsequent compositions.
• excessive importance was given to grammar, spelling and vocabulary at the expense of higher order concerns such as content and structuring:
• although I tried to correct and comment upon all aspects of students' writing, this probably amounted to overload in many cases and could have been alarming for students to see their work covered in annotations;
• instead of encouraging students to re-write their compositions, first drafts were treated as the final product;
• due to the time which elapsed between successive writing tasks, it was probably difficult for students to relate or apply corrections and comments from one composition to another;
• without first ensuring that students were equipped with a range of strategies for dealing with corrections and responding to comments it was unfair to expect them to improve their writing;
• at times I expected students to produce high quality pieces of work without adequately preparing them for writing tasks;
• expecting students to self-correct is unreasonable when they lack the necessary language knowledge and instead of using the correction code it would be better to supply the correct language/form in such cases;
• uncertainty as to students' intended meaning probably often resulted in feedback which was inaccurate or inappropriate;
• instead of discussing and negotiating approaches to writing, conventions and intended meanings, feedback was one-way and teacher-centred;
• finally, the grading system used offered little scope to indicate students' progress (in my teaching context it was only possible to award three different grades, e.g. 6-/6/6+), and this could negatively affect their motivation.

Subsequent questionnaires and diary entries revealed that students shared several of these concerns and were also critical of several aspects of the correction system, i.e. the fact that it merely consisted of highlighting mistakes but did not explain how they could improve their writing. Furthermore, it was suggested that instead of using the correction code it would be more useful for the teacher to provide the correct language, especially as valuable time was spent deciphering the code.

Generally speaking, although students appeared to use only a limited number of basic strategies for dealing with corrections and responding to comments, they had firm ideas about the type of feedback they wanted and preferred, namely the provision of honest opinions about their performance and practical tips for improving their language and writing skills.

Having gathered together a list of my concerns and observations from students, the next stage was to consider how the existing feedback system could be adapted and modified to make it more efficient and responsive to students' needs. In part this was not a difficult task, as there were numerous small changes which were easily implemented. However, there were other issues which were much more fundamental and implied a substantial alteration in my approach to teaching writing and providing feedback.

Without doubt, one of the most important of these issues was the need to emphasise the process approach to writing, and the importance of revising and re-writing such that students became accustomed to submitting first, second and third drafts, thus actively processing feedback instead of passively absorbing it. Secondly, to reduce the teacher-centredness of feedback and to encourage a more co-operative approach to writing, it was necessary to encourage self-and peer-correction, as well as group- and pair-writing activities. Finally, in order to provide students with essential skills and strategies for improving their writing, it was necessary to have regular input and practice sessions.

In short, for these more far-reaching changes to be implemented, time had to be found from within the schedule for this purpose, but at least it ensured that all students practised and improved their writing, which is a worthy compensation, especially as homework is not always forthcoming!

To end, the final stage of the action research consists of reviewing the appropriateness and effectiveness of the implemented changes, effectively taking us back to the beginning of the process, i.e. critical reflection of our methods, and so the cycle repeats itself. The English-Speaking Union of the Commonwealth now has active representation in 45 countries. Most of these countries run programmes to encourage the effective use of English as a means of international communication, as well as a variety of current affairs and cultural activities aimed at improving international understanding.