Mathematical literacy, viewed as a set of ideas involving applications of mathematics to real-world contexts, has recently featured in curricular discussions about the aims for mathematics education. This article explores the effect that differences in the way that a mathematical task is contextualised can have on students' mathematical arguments and, therefore, on their perceived levels of mathematical literacy. Seventy-two students' responses to three similar measurement tasks are described according to Kaiser and Willander's levels of mathematical literacy. The arguments used for assigning each level of mathematical literacy are then investigated for the presence of specific macro- and micro-linguistic features. The context of the task affects what students perceive to be most relevant approaches to use, which are reflected in the arguments they give; this, in turn, affects external judgements of their level of mathematical literacy.