Research perspectives on the objects-early debate

Abstract

In March 2004, SIGCSE members contributed to a mailing list discussion on the question of whether programming should be taught objects first or imperative first. We analyse that discussion, exploring how the CS community debates the issue and whether contributors' positions are supported by the research literature on novice programmers. We applied four distinct research methods to the discussion: cognitive science, rhetorical analysis in the critical tradition, phenomenography and biography. We identify the cognitive claims made in the email discussion and find there is not a consensus in the research literature as to whether the objects first approach or the imperative approach is harder to learn. From the rhetorical analysis, we find that the discussion was not so much a debate between OO-first versus imperative-first, but instead was more for and against OO-first. Our phenomenographic analysis identified and categorized the underlying complexity of the discussion. We also applied a biographical method to explore the extent to which the participants' views are shaped by their own prior experience. The paper concludes with some reflections upon paradigms, and the manner in which the CS discipline community defines itself.