The modular approach to language in its career of 30 years had alternating and rivaling views regarding the place of pragmatics. A first approach basically is the one outlined by Fodor (1983) that would pack pragmatic aspects of language use under the rubric of the mushy General Problem Solver component of the architecture, thus extracting it from considerations of modularity altogether. The rival Massive Modular approaches such as Dan Sperber’s would be willing to treat pragmatic aspects as one crucial module as part of a general architecture with modularity all over the place. The paper after summarizing the theoretical interpretations calls for a less dedicated distributed processing and representation system where modularity rather than a simple starting point might be seen as the result of a process of modularization. Three types of empirical data are surveyed. First, studies that seem to support a specialized pragmatic module are discussed, namely from right hemisphere damaged populations and brain imaging data that imply a strong involvement of right hemisphere in a variety of pragmatic aspects from emotional stress to understanding non-literal language. A second line of data comes from developmental neuroscience considerations. Studies with autistic and other cognitively challenged populations suffering from a presupposed overall architectural deficit indicate the crucial role of a Theory of Mind not only in tasks of second order representation, in attributing a sophisticated Belief-Intention system to others, but in language processing as well. One interpretation of these data is to postulate a module of social and psychological cognition, that would be a driving source of language use. The primacy of language use should be left open as an option. It is possible that language use itself and thus language pragmatics in different intentional contexts is partly responsible for the development of the seemingly encapsulated system of mentalization. This would correspond to the general idea of early prepared systems being modified during an interface buildup process in development that roughly corresponds to an overall use of language for metacognitive purposes. A third line evidence calling for a balanced treatment of the modularity issue comes from theories of Paleobiology. Theories like the ones proposed by Donald, Mithen, Wilkins and Wakefield should also be considered in this regard. According to these theories human language may actually have resulted from a loosening of boundaries between encapsulated modules, rather then from strengthening them. This line of reasoning is especially interesting since it supports the idea that human language emerges by necessity as the result of an interaction between different “intelligences’“, together with elementary societal organization and a social mind.