The study investigated achievement status as a factor determining the use of language-based activities for learning science. A total of 154 eighth-grade students were randomly assigned to four groups, all stratified for gender and achievement level. The treatments involved various combinations of talk and writing, and descriptive and explanatory tasks. The dependent measures included scores on multiple choice tests obtained at three times during the study. Records of student talk and writing were also analyzed to identify patterns of differences between groups of achievers. The findings suggested that low achievers complete more problems, and develop better understanding and comprehension of ecology concepts when they have engaged in peer discussions of explanatory tasks. In comparison, high achievers benefit more from writing than talking, and writing explanations enhances comprehension more than restricted writing activities.