The role of peer feedback in learning to write explanatory texts: why the tutors learn the most

Abstract

French students in grades 4 and 5 in schools around Paris wrote explanatory texts in L1 following a lesson in the Life Sciences, four times over the course of the school year. Each session included written correspondence with another student; half the students (Group G1) made suggestions about ways to improve the drafts of the other half of the students (Group G2). All students then revised their text. The results show: (1) an improvement in the quality of the texts produced over the year, even at the first draft stage; (2) greater progress for the students who gave suggestions compared with those who received them; and (3) a fragility of the performance of the last writing session. The results are interpreted in light of the concepts of knowledge transformation strategy and dialogism. The ‘stepping back’ encouraged by adopting the role of tutor to a fellow student may have allowed Group G1 students to build knowledge about explanatory texts and better ‘orchestrate’ the different voices present in their texts. The difficulty of the last writing session, which required a conceptual transformation and a reorganisation of the information provided by the lesson, could explain the poorer performance observed in this session.

Keywords:

• collaborative learning
• interaction
• literacy practices
• metalinguistic awareness
• writing

The authors thank the Institut Universitaire de Formation des Maîtres (IUFM) of Créteil (Université de Paris Est Créteil) for supporting this research, as well as everyone who contributed to it: Patrick Avel, particularly for his work in developing the science sessions, the teachers of the participating classes – Isabelle Bezat, Virginie Collin, Fabrice Elbaz, and Jean-Louis Marc – and Fatiha Amouri, Annick Cautela, Denis Legros, Alain Maillard, and Evelyn Rosset.

Notes

1. Reading comprehension assessments (ordering information and relative importance-judging tests) were used to establish levels. RL1 = weaker readers, RL2 = stronger readers. Reading comprehension is taken as an indicator of a student's overall academic level and notably as an indicator of ‘writing proficiency’ (Shanahan, 1984). Reader level almost always coincides with the writing level in our sample (see the ‘Results’ section). Of the original 110 participants, Group G2 had more strong readers. To obtain an equivalent number of strong and weak readers in each group, we had to remove several. We also had to discard students who were absent for one or more sessions.

2. This sequence allowed introducing students to the criteria of a successful explanatory text. This was done through a series of exercises that current space limitations prevent us from describing in further detail. Suffice it to say that the exercise was identical for both groups, and the criteria were coherent with those used to code the written texts.

3. The following instructions were given for each respective writing sequence:

• Sequence 1: Explain why a bubbler is necessary for fish living in a small aquarium.

• Sequence 2: Explain why some islands and areas of the world are at risk of being underwater in the future.

• Sequence 3: Explain how the heart accomplishes such a complicated and crucial job.

• Sequence 4: Explain why certain animals are visible during the winter in France while others disappear.

4. This first check by an adult makes poorly spelled texts more readable and prevents the students from focusing on superficial linguistic revisions as young students often do (Chanquoy, 2001).

5. The instructions were the following: Each of you will suggest comments and revisions for three other students’ writing samples, to help them improve their texts. If you make comments and suggestions, it is because the other students’ texts are not clear, not specific enough or slightly incorrect. Don't forget that these texts are meant to be read on the Internet website.

6. Coding was done independently by two specifically trained members of the research team. Any discrepancy between the two coders was resolved by a third independent coder.

7. Students within the same group had the same writing level at the beginning of the year. In the first Sequence, students from both Group G1 and Group G2 produced almost the same amount of relevant information (with means of 36.59 and 40.53, respectively), and the same amount of irrelevant information (20.24 and 17.64, respectively). In addition, as expected, the amount of relevant information differed as a function of reader level and the factor Reader Level was significant.

8. Comments and criticisms from Group G1 students showed quite a range, including:

• – comments about the inclusion or correctness of information: ‘Really good but you forgot the animals that migrate’. ‘Your text is really good, but it's not: it empties and inflates, it's: it empties and fills up’.

• – corrections about scientific terms: ‘replace the word door by valve’.

• – suggestions about wording: ‘I'm going to rewrite your sentence: animals that disappear, die, migrate, or hide (snails, lizards)’

• – comments about general organisation: ‘There are two reasons for the ice melting. Explain them’. ‘You could emphasise what is most important about putting oxygen in the water in an aquarium’.

• – comments about what is expected in a scientific explanatory text: ‘You can include a conclusion’. ‘Your text does not explain enough’.

• These comments and their results were analysed in a previous publication. See Crinon, Marin, and Cautela (2008).

9. Examples chosen as illustrations here are always found in several other students’ texts.

10. There may be other explanations, not explored in the present paper, for the greater progress made by the Group G1 students. For example, their motivation and feelings of self-efficacy (Pajares, 2003) may have been increased by assuming the role of tutor, and thus the role of a competent writer able to make helpful comments.

11. We have translated the students’ writing from French into English. See the Appendix for original texts in French.

12. One could imagine that the example and comments of an ‘expert’ capable of using the knowledge transformation strategy would be helpful here.

13. In this part of the Discussion section, the texts are analysed differently than described above. Here they are examined in terms of their ‘enunciative characteristics’. The examples provided can be considered typical insofar as the relevant characteristics are found throughout the texts of students in the same group.

14. Olson (1994) shows that a fundamental difference between the oral and the written text consists in the necessity of the latter to explain, most notably by lexical means, how one wishes one's utterance to be interpreted.