In the present paper we aim at contributing to the discussion on whether the use of exploratory software faciclitates pupils in mistakes’ reflection and negotiation. Towards this aim we designed and implemented two educational activities which have inbuilt the characteristics of experimentation, exploration, collaboration and communication on construcions. We used a version of Logo integrated in the E-Slate component environment (http://e-slate.cti.gr/). Drawing on the concept of “distributed constructionism” (Resnick, 1996), we took advantage of the fact that Logo-code is a shareable artifact that even though it carries personal meanings it also allows for extensions and changes (Harel & Papert, 1991; Hoyles & Sutherland, 1992) and we encouraged pupils to exchange and try out one another’s Logo constructions (Trouki, 2003; Yiannoutsou, 2005). Our main research questions were the following: • How pupils talk about their mistakes during intra-group and inter-group discussions? • How the specific educational activities promote mistakes’ negotiation during inter-group discussions? Taking into account that thought and learning is influenced by the characteristics of the symbolic tools mediating the construction of representations (based on the idea that thought is not just expressed through a symbolic system –e.g. language- but it comes into existence through it, Vygotksy, 1986) we focus on a) the characteristics of the feedback provided by the software and b) the dynamic link between different representations (i.e. between the symbolic expression and its graphical output). The feedback offered by an exploratory software such as the one used for the research we are reporting here, has the following characteristics: a) it merely presents the results of the pupils’ actions on the software without an evaluation of these actions nor hints for the correct answer and b) it is based on pupils’ previous knowledge (Arcavi & Hadas, 2000) as well as on the pupils’ knowledge of what should be the expected result (i.e. what their construction/model should look like). When programming is involved the issue is not about knowing or not knowing the answer. A program that works is not an answer you know, it is a process of exploring which includes planning, trying out things and going back to change the initial program. In this process discovering and correcting the mistakes hold an important part because they are important for making the program work (Papert, 1980).

The data were collected during two studies that took place in real school setting. In the first study pupils of 6th grade worked in groups of three using Logo to construct the model of a bridge and Outlook Express to exchange e-mail messages with pupils from another school who worked on the same project. In the second study 14 year old pupils were engaged in a cartography activity; they had to construct the buildings and their places on campus in order to participate in a computerized treasure hunt game. Pupils worked in pairs to construct a different half of the map and then exchanged codes with another pair in order to have as end product the whole map of the campus. Acting as participant observers we collected the following data: • Field notes • Tape-recordings of pupils’ intra-group discussions • Pupils’ interviews • Pupils’ e-mail messages, Logo codes and models

Our main findings are: • Pupils tent to focus on the final succeeded outcome instead of the trial-and-error process. • The exchange of Logo-codes brought into the surface the process of construction and served as ground for trying out other groups’ codes and talking about mistakes. • It was not the teacher who pointed out pupils’ mistakes at the end of the activity in the form of final evaluation; pupils themselves examined Logo-codes and this was part of the process. • Pupils’ comments on mistakes served as point of departure for further experimentation as they spent time and effort to explore alternative strategies. • To negotiate their mistakes pupils were based on the specifics of their construction and used the characteristics of the computer environment. Our findings suggest that the long-held classroom practice doesn’t provide pupils the culture of dealing with constructions; so, implications for teaching interventions are also discussed.