Language is not a module. Well, at least, it is not a feedforward encapsulated domain-specific perceptual input system in the way that Fodor (1983) imagined. To be sure, there are regions of cortex that are conspicuously specialized for language-like processes (e.g., Gazzaniga, 2000; Kuperberg, Holcomb, Sitnikova, Greve, Dale, & Caplan, 2003; Ojemann, 1983), but when cognitive neuroscientists refer to these cortical areas as “modules,” they certainly do not imply solely feedforward synaptic projections or encapsulation from neighboring cortical areas. The vast and recurrent interconnectedness between anatomically and functionally segregated cortical areas (e.g., Douglas, Koch, Mahowald, Martin, & Suarez, 1995; Haxby, Gobbini, Furey, Ishai, Schouten, & Pietrini, 2001; Van Orden, Jansen op de Haar, & Bosman, 1997) unavoidably compromises any assumptions of information encapsulation, and can even wind up blurring the distinction between feedback and feedforward signals.
What this means is that we should expect language processes to function in concert with other perceptual, cognitive, and motor processes, not independently of them. For example, McGurk's famous and compelling demonstration of visual perception of mouth shape influencing the immediate percept of a spoken phoneme (McGurk & MacDonald, 1976) is emblematic of the intense degree to which speech perception and visual perception pay close attention to one another. More recently, visual perception has also been shown to play a strong role in spoken word recognition, syntactic processing, and reference resolution (Tanenhaus, Spivey-Knowlton, Eberhard, & Sedivy, 1995).