This article argues that the otherwise praiseworthy liberalism of target-text oriented descriptive Translation Studies makes it difficult to justify any distinction we might like to draw for e.g. pedagogical or criticism-related reasons between choice and error in translation. It proposes that such a distinction may be drawn on the basis of a (linguistically oriented) distinction between semantic patterning in the case of the former and formal (graphemic/phonemic) patterning in the case of the latter. Since it is necessary to pre-establish the distinction between error and choice in translation on non-linguistic grounds if we are to argue for informative concomitance between it and the two types of linguistic patterning, the linguistic distinction is illustrated with reference to a set of texts in the case of which a strong argument for distinguishing choice from error can be made on (non-linguistic) historico-sociological grounds.