It is widely argued that engaging in communicative language tasks helps a learner develop in an L2 in several ways. Tasks provide an opportunity not only to produce the target language, but also, through conversational adjustments, to manipulate and modify it. Checking and clarifying problem utterances ('negotiating for meaning') ensures that task participants receive comprehensive input and generate comprehensible output, both of which have been claimed as crucial to second language acquisition (SLA). Task type is considered significant, with those tasks requiring an exchange of information most likely to prompt negotiations for meaning. This paper reports a classroom observation of the language produced by intermediate EFL students engaged in required and optional information exchange tasks in both dyads and small groups. The results show no clear overall effect for task type or grouping, though there was a discernible trend for dyads doing a two-way task to produce more negotiated interaction. However, it was noticeable that many students in the small groups did not speak at all, many more in both dyads and small groups did not initiate any negotiated interaction, and very few students in either setting produced any modified utterances. Such positive results as were obtained seemed to be due to the disproportionate influence of a small number of the students, and so were not typical of the group as a whole. The setting of the study within a classroom, as opposed to a venue especially arranged for data collecting, is suggested as a significant variable, with important implications for group work research methodology. It is also suggested, contrary to much SL theorizing, that 'negotiating for meaning' is not a strategy that language learners are predisposed to employ when they encounter gaps in their understanding.