The present study inquires into word-image interaction in the translation process of illustrated technical texts. The method employed for this purpose is Choice Network Analysis, which compares the translations of the same source text by multiple translators in order to empirically derive the options, the set of possible solutions, that are available when translating each verbal item. The data of the study consists of eight translations of an illustrated technical text, produced by a group of Master’s level translations students. The study sets out to assess if the options offered by the multimodal source text are based entirely on verbal information or on a negotiation of meaning from two different modes. The analysis implies that visual information could modify verbal information; in the most extreme cases, visual information could cause verbal information being disregarded altogether. This indicates that images can reattribute the meaning of verbal items in translation.