The original proposition of language analysis set forth by analytical philosophy stemmed from the assumption of the existence of a logically perfect language, which would mirror the logical form of the facts. This ideal language would clearly and correctly reveal the logical structure of the world, avoiding the 'traps' of daily language. The philosophy developed on the second phase of Wittgenstein's work breaks apart this notion of a unitary language in a multiplicity of language games, based upon particular forms of life. Grammar, or the set of rules that govern a language, becomes autonomous, since it does not account for an assumed essence or form of reality, but acquires its meaning in the use of the expressions it regulates. This autonomy of grammar makes room for the existence of different systems endowed with meaning and, therefore, allows us to speak of an otherness of forms of representation. This dissertation intends to point out this 'opening' introduced by Wittgenstein, which was partly foreshadowed on his critique of the works of the anthropologist J. G. Frazer. It also presents some discussions that it raised inside and outside of analytical philosophy. Finally, the objective is to sketch a method of conceptual analysis, derived from the encounter between anthropology and philosophy, as an alternative approach to the analytical train.