This paper suggests that a linguistic analysis of TAPs elicited by authentic-like translation briefs might reveal differences in existing or potential translational proficiency. A distinction is made between addressee-oriented and addressee-free verbalisations, and in focus are evaluative statements and expressives. Two tentative hypotheses emerge, namely that the proportion and specificity of evaluations of the target text increases with translational proficiency, and that expressives reveal attitudinal differences attributable to different levels of proficiency.