The study we present in this paper is part of a wider project which includes the analysis of the positive aspects and limitations of feedback and the evolution of the written commentary process (Bruno, 2006). It was realized in a collaborative context within the area of Physics and Chemistry and was guided by the following research questions: (1) Which feedback characteristics help students learn? (2) What are the feedback limitations and its degree of applicability? Our objective is to contribute to the construction of a wider body of knowledge that may help teachers implement assessment for learning by giving appropriate feedback to their students. Formative assessment is a kind of assessment that has been mentioned for some years now in several Portuguese laws. However assessment models are generally used to give marks but are insufficiently active in helping students to improve their learning process (Black, Harrison, Lee, Marshall & Wiliam, 2003; Gibbs & Simpson, 2003; Fernandes, 2005). Nowadays the importance that formative assessment can have in helping students’ learning, even in raising standards, is unquestionable: “Is there evidence that improving formative assessment raises standards? The answer was an unequivocal yes (…) There have been few initiatives in education with such a strong body of evidence to support a claim to raise standards” (Black, Harrison, Lee, Marshall &Wiliam, 2002, p. 1). One way to monitor students’ learning is by using feedback. The teacher should help students understand what their real level of knowledge is (their state in relation to learning objectives) and, simultaneously, give information or clues to help them fill the gap between their actual level and the desired one. One way of giving feedback is through written comments. The principal advantage of this kind of feedback is the fact that teachers have more time to analyze students’ work more and comments can result from a deeper reflection (Crespo, 2000). But it is still difficult to interpret students’ work and know what to do next. Each situation is a different case, specific and singular. That is why “it is difficult to know what line of questioning or which feedback is more adequate in each moment” (Santos, 2004, p. 165). Shepard (2000) adds that “being able to ask the right questions at the right time, anticipate conceptual pitfalls, and have ready a repertoire of tasks that will help students take the next steps requires deep knowledge of subject matter” (p. 12). Some studies point out that feedback in the form of comments can help students identify the strong and weak points of their work and learn from the revision process (Black & Wiliam, 1998). However, feedback will not solve any problem if it is not well thought, structured and adequately integrated in the students’ learning process (Gibbs & Simpson, 2003; Gipss, 1999). Usually it tends to be “insufficiently explained, badly distributed, and thus it barely contributes to help students learning” (Fernandes, 2005, p. 143). Feedback quality matters more than quantity (Gibbs & Simpson, 2003; Sadler, 1998). Therefore, this research was undertaken in order to understand which characteristics promote more and better learning.

This research is of a qualitative and interpretative nature (Silverman, 2004) since it deals with the questions posed about the potential and limitations of feedback. This study was a collaborative effort between one of the authors (IB) and another middle school Physics and Chemistry teacher. We expected that the diversity of experiences, skills and perspectives would help drafting commentary feedback and contribute to the success of this assessment method (Boavida & Ponte, 2002). Written comments were made on students’ assignments from two eighth-grade (13-years-old) classrooms. Because it would take too long to write down comments for all assignments in a collaborative way, we selected eight students (four from each class) that had shown a weak improvement between the two versions of the first assessment tool. During the work sessions, the two teachers wrote comments on the assignments of these eight students. The first versions were not marked. The study took place in a school in the south of Portugal during the 2005-2006 school year. Five tools of learning/assessment were researched: one small test, three worksheets (two of them including experimental work) and one experimental report. The first version of the assignments was done inside the classroom except for the experimental report. After the feedback comments were provided, the second version of all assignments was done individually, some inside the classroom and some outside. In Class A, the researcher (IB) made participant observation (Cohen, Manion, & Morrison, 2003) during the lessons in which the learning/assessment tools were implemented. Moreover, the lessons during which the students rewrote their assignments were audio taped (and then totally transcribed). The two teachers realized 19 work sessions in which they: i) conceived the five learning/assessment instruments; ii) wrote down the comments on the assignments of eight students, and iii) analyzed the written comments based on the evolution between the first and the second version. During the work sessions, the data was collected using participant observation with audiotape and later transcription in full. The documental analysis of the first and second versions of the assignments (including the written comments) was the raw material considered in this research study. The researcher also carried out semi-structured interviews (Burns, 2000), which were taped and transcribed in full, with the eight students with the aim of getting a better understanding of the given feedback´s potential and limitations. The interviews were used to collect descriptive data with the participants’ words to help us to identify the comments’ characteristics that helped and those that did not help students to improve their work, and understand why the second version had not always shown improvement as we were expecting. The data analysis was a systematic and recurrent content analysis using categories related to the form and content of the comments.

The results of this study point out that comments often helped students improve their work. Interviewed students said comments helped them: i) understand worksheets questions; ii) identify their mistakes, and; iii) improve their work. These results are consistent with other studies (Black & Wiliam, 1998; Wiliam, 1999) which conclude that when feedback focus on what is necessary to do to improve the work and when clues are given about how to do that, they lead to better performances. Although the majority of comments were formative, some of them were not. So, feedback is not necessarily synonymous of pedagogical monitoring, as Black and Wiliam (1998) and Santos and Dias (2006) say. We could also corroborate the conclusions of the last authors when we realized that the same comment can be efficient for some students but not for others. As far as the comment’s form is concerned, it is essential that the teacher has a good handwriting. This study, as well as Goldstein’s (2004), shows that it is better to write the comment next to the part of the student’s work that must be reviewed rather than anywhere else. The number of tasks asked also seems to influence students’ performance. The majority of items for which comments had been given asking to accomplish two or more tasks, were not changed the way we were expecting. In most cases, there was no progression between the two versions. We have evidence that requesting one task for each different topic helps students distinguish and understand the different tasks they have to do. However, and according to Veslin and Veslin (1992), we should not give back an assignment full of comments. The experimental report which was the tool with more written comments was also the one with the greatest percentage of items with no progression in the second version. It is understandable that in these situations students feel unmotivated because all their work is being challenged. In agreement with Shepard (2000) and Veslin and Veslin (1992) it is important to ignore some mistakes in order to maintain students´ motivation and self-confidence. The comments’ syntactic form was another analyzed category. Contrary to Goldstein (1999, in Goldstein, 2004) results, this study shows that syntactic form can influence students’ performance. One of the students, instead of improving her initial answer, used the comment to directly answer some questions that were simply intended to help her identify her mistakes. However, when we applied the imperative verb tense she usually corrected her initial answer. Regarding comments’ content, it is important to use a simple vocabulary and familiar expressions. This is one aspect that has already been referred to by several authors, like Santos (2004), Suffolk County Council (2001) and Veslin and Veslin (1992). Simple and direct comments seem to be able to help students if they know how to improve their work, but when the reformulation involves less developed or more complex skills, the indication of what is necessary to do seems to be insufficient. Santos and Dias (2006) have also found that not all students succeed in using clues, even if they are simple. However, and according to Conrad and Goldstein (1999, in Goldstein, 2004), the comments which indicate strategies of resolution are particularly helpful. This kind of clues seemed to help students work autonomously. The clarification of what was asked in the sheet, in other words, the distinction of what should or should not be included helped students, in some cases. The comments which give part of the answer showed some limitations. These kind of comments prevent students from correcting/improving the item by themselves and the information that is given conditions their answers. Students, frequently, did not go much further from the information that already had been given in the comment. Feedback’s success depends on the teacher’s knowledge of the difficulties and skills of each student regarding a particular situation. This is the reason why it is so difficult to know what is appropriate to write in each case (Menino, 2004; Santos, 2004). Furthermore, it is important to give feedback as soon as possible. The sooner we give feedback, the more probable students find it useful and are able to use it because they can remember what they have done in the first place and why they did it that way (Gibbs & Simpson, 2003; Wiggins, 1993). It is also important that this process occurs systematically so that students can understand these dynamics. More importantly, students and teachers can understand each other more easily in the course of time. This conclusion was evident in this study and is in accordance with Black and Wiliam’s review (1998). Writing comments is demanding, complex and time-consuming. The collaborative work can be a helpful context to develop this task. We must add though that even if we were able to follow all these suggestions we could never be sure that our feedback would help students to learn and improve their work. There is a great number of factors that can influence their success, like motivation and available time.