The first part of this paper discusses the synecdochial device, a key concept of a universal nature within discourse, & thus translation. I suggest linking Lederer's view of the synecdoche (1976, 1981, 1994) - according to which the linguistic formulation is but a part used to designate the whole, i.e., the sense of an utterance - to the cognitive theory of metaphor & metonymy (CTMM) (Lakoff & Johnsen, 1980; Langacker, 1987; Lakoff & Turner, 1989; Gibbs, 1994; Panther & Radder, 1999; Barcelona, 2000), where words are considered as cues for cognitive constructions. The synecdoches are the overt linguistic results of an individual's ability to construct a cognitive configuration. Due to the fact that a notion or a situation is generally conceptualized differently in two different languages, the synecdoches used to convey the sense of an utterance in discourse will consequently not be the same in the two languages. The second part of this paper sets out to examine to what extent professional translators are aware of the need to resort to new synecdoches in their target text in order to convey the cognitive configurations they have in mind after having grasped the sense of both novel & entrenched linguistic expressions in the source text. As mental processes cannot be observed directly, they need to be inferred from the translator's behavioral data. Drawing on experiential cases where the professional translators' behavior is observed on the basis of data gathered from two process-oriented on-line approaches - think-aloud-protocols & Translog keyboard logging - I attempt to infer the reasons for their choice of translation shifts vs. pre-assigned correspondences. I finally suggest classifying their transferring procedures into three categories: automatic, reflected, or explorative. (LLBA, Adapted from the source document, Accession Number 200505095, (c) CSA [2004]. All rights reserved)