Sažetak The problem of specifying the errors committed in the process of written translation is a moot point. In this paper, the levels-of-processing approach is proposed as a viable toolkit for tracking, specifying, and more importantly, analyzing those errors when translating from English into Arabic. The levels suggested are the lexico-semantic level, the co-textual level, and the contextual level. The paradigm in this research is to take the output, i.e. the translations, as evidence for the presence or absence of certain levels that directly or obliquely have a bearing on written translation errors, and to record the highly recurrent or operative and the least observable ones.
