This paper analyses student teachers' meta-narratives associated with 'doing' mathematics. The quality of the vocabulary associated with the doing of the mathematical objects of a sample of 56 multilingual third-year mathematics methodology students preparing to be secondary teachers was explored and assessed. Sfard's theoretical concept of 'pronounced foci' and her 'critical categories' for determining if a narrative was mathematical were used to analyse the data. Meta-narratives refer to the written mathematical metacognitive monologues that manifested students' thinking while they were doing the mathematical objects. Students were required to write, in language suitable for teaching, a description of the procedure which led to a solution. The procedures and their associated meta-narratives provided data for a mixed-method analysis in this case study. The absence of a mathematical vocabulary presented difficulties when students engaged in meta-narratives. The data also show that inexact mathematical vocabularies are not culturally specific since the 11 English mother tongue students demonstrated a similar lack of command of mathematical language as their peers. Meta-narratives revealed mathematical misconceptions that may not have been detected without them. For two of the three mathematical objects there is a significant but moderate positive correlation between correct vocabulary and correct procedure. The findings support the need for mathematical vocabulary (and language) to be a teaching focus when preparing teachers for their careers and that the assessment of tasks is not completely achieved by evaluating only the procedural 'doing' of mathematics.