Edmund Sapir famously remarked that blowing a candle produces a gesture and a sound that are identical to the gesture and sound made when pronouncing the (German) consonant W. The example illustrates the central claim that what makes a sound a consonant or a candle blowing is not in the sound or in the gesture but represents the contribution of the linguistic and the movement system in which the gesture and the sound of speaking a “W” or of candle blowing do in fact function. Theorists of motor behavior also must confront this issue and ask on what grounds motor behavior can convey meaning. Here we raise this issue by briefly addressing the question whether emotional body expressions belong to our competence for emotional body language. Communication signals used by higher organisms come in many shapes and forms and involve all sensory modalities. Among all these signals actions involving the whole body and expressing emotion (emotional body language, or EBL for short) by their saliency and frequency occupy a privileged position in many species. There is increasing behavioral and neurofunctional evidence for a specialized system of EBL perception in humans, which involves among others subcortical structures including the amygdala and cortical areas consisting mainly of premotor cortex, superior temporal sulcus, and inferior frontal gyrus. With application of neuroscientific methods to this relatively underexplored domain, at least in human emotion research, a major challenge now is to outline what the specific questions are for a biological theory of EBL (de Gelder 2006). Our ability to communicate through EBL traditionally counts as a nonverbal skill and therefore, much as the expression “body language” is entrenched in popular parlance, it is most often assumed that EBL is not really a language in any systematic and serious sense of the term. Yet EBL may share some fundamental properties with language, and exploring what these would have to be may bring into focus what issues a biological theory of EBL is up against. Among the principal characteristics EBL shares with language, we figure the facts that the signal is complex and consists of coarticulated movements of different body parts, that it is perceived automatically, and that acquisition in ontogeny is rapid and effortless. Another salient characteristic is recursiveness. Just as the fluent speaker is capable of an almost limitless number of speech utterances, the presence of a normal ability for perception and production of EBL means that we are capable of an almost limitless number of body movements to express emotions with. On the perception side, the observer represents the visual input provided by observing EBL in a way that ultimately gives him access to the emotional meaning and intention. On the production side, the agent encodes his emotional meaning in a behavioral intention and ultimately into motor output. As is the case in linguistic theory, a critical question is how the perception and production side are linked. Different alternatives for the internal representation of language continue to be vigorously discussed; it is fair to say that many mysteries remain as to the relation between perception and the internal representation on the one hand and the internal representation and production side on the other. While there is agreement that perception and production are two sides of the same coin, it is unclear how each is connected and whether each has its own underlying representation system. Perception and production each take as their input a very different set of signals. The latter maps from linguistic intentions to movements of the articulators, the former from sounds patterns to thoughts. This problem of the so-called dual code has haunted linguists for a very long time. Theories of motor behavior face issues very similar to the ones debated in linguistics; for example, to develop models of how the brain perceives action in movement, how it distinguishes biological movement from motion and noise, whether to capture a particular human action in a single complex model or to make extensive use of semantic knowledge and a