Medical educators develop student selection criteria and design curricula based on underlying assumptions about who is best suited to the profession and how these learners should be taught. Often these assumptions are not made explicit but instead are embedded in the words and phrases used to describe trainees and curricula. They may then be considered inevitable, rather than being seen as particular social constructs. Using Foucauldian critical discourse analysis methodology, the authors examined a major shift in language in the late 1950s in North American medical education texts. The discourse of the good doctor as a man of character, which had been present since the 1910 Flexner Report, was replaced by a new discourse of characteristics. Analysis of this sudden discursive shift shows a change in thinking about the medical trainee and learning environment from a personal journey of discovery to a dissectible set of component parts that could be individually measured and manipulated. Understanding the discursive effects of language that we use will allow medical educators greater insight into the implications and consequences of different constructions of important issues in medical education.