Many previous studies have compared automated analyses to writing experts' assessments, and less is known about the relationship between such features and content instructors' assessments of writing. The data analysed in this study is part of a larger study investigating growth in academic writing skills during undergraduate study. In this exploratory study, a corpus of 117 assignments written by 14 New Zealand first-year students with English L1 was collected and analysed to identify textual features that predict content instructors' assessments of written assignments. Analysis of the whole corpus found five linguistic variables that predicted grades assigned. Subsequent analysis of two separate subcorpora found three language features specific to humanities and four specific to the social sciences. The results suggest that although understanding of course content is the primary concern in written assignments at university, the linguistic quality of written assignments also contributes to academic success during the first year.