During the 20th century, the term “naturalistic” within the language’s sciences corresponded with a physicalist tendency, directing the so-called “linguistic turn” to a clearly synchronic, analytic-deductive and logical-formal paradigma. In the first period of cognitive science this tendency took the form of a theoretical approach to language in which the dualism between the mechanical-morphological components and the psychic ones fully arose. Within the computational hypotheses this theoretical solution has proved to fit completely to the distinction between hardware and software. As is well-known, these models revealed even the simplest semantic uses of language as unsuitable to explain. One of the most negative aspects of this real epistemological defeat of the most ambitious linguistic philosophy of the 20th century is that it entailed the progressive theoretical decrease of human language role within the new dominant paradigm of cognitive science. In this paper we support a different theoretical position that can place language contribution at the core of the scientific debate, while remaining within the boundaries of cognitive science’s epistemology. This can be possible if we use a different naturalistic philosophy of language, based entirely on evolutionary-developmental biology and on the fundamental concept of morphological constraints, rather than centered on physicalist stances. Moreover, we believe that our position can open to cognitive science further possibilities of application that are currently clouded by the neuroscience’s primacy and the totalizing approach of experimental methodologies