In a previous study (Patterson & Weideman, 2013), we discussed the importance of acknowledging the typicality of academic discourse as a starting point for critically engaging with constructs of academic literacy. In this article, various attempts at identifying the typical features of academic discourse are surveyed and critiqued. The preliminary conclusion is that the uniqueness of academic discourse lies in the analytical or logical language that characterises it (see Patterson & Weideman, 2013 for an extended explanation). Using this characteristic feature as a criterion allows us to sift through the various opinions on what constitutes both academic discourse and academic literacy in a way that is potentially productive. It suggests on a number of points ways in which one might add components to the current definition of academic literacy that forms the test construct of academic literacy tests such as TALL, TAG, and TALPS. The article concludes by suggesting some modifications and additions to the design of current test task types in tests of academic literacy. These tentative suggestions may allow theoretically defensible modifications to be made to the construct of a number of tests of academic literacy. TALL, TAG, the relevant part of the NBTs, and TALPS are generally high stakes tests that are widely used in South Africa. Since no critical examination of their construct, which is now more than a decade old, has so far been undertaken, we hope that these proposals do not only come at an appropriate moment, but may also be useful to those responsible for developing further versions of these tests.