Each language is a unique tool for analyzing and synthesizing the world, incorporating the knowledge and values of a speech community. According to Sapir (1931), linguistic "categories [including] number, gender, case, tense, mode, voice, "aspect", and a host of others ... are not so much discovered in experience as imposed upon it." Thus to lose such a tool is to "forget" a way of constructing reality, to blot out a perspective evolved over many generations. Who can say whether a concept that evolved in one language would never have evolved in another? The extreme version of the Sapir-Whorf hypothesis--that perception and cognition are determined by the structure of whatever language one happens to speak--has been demolished by Chomskyan linguistics (see Pinker 1994: 59-63). But the central arguments of the thesis is still strongly visible in the current approaches of conceptual semantics developed by Jackendoff and others which can be fruitfully used to understand the conceptual loss of indigenous knowledge system in the speakers of any endangered language over a period of time. The recent developments in the area of acquisition and analysis of endangered language data have fruitfully accommodated both the radical version of Whorfian thought and mentalist approaches of generative grammar. What is most important and theoretically provocative is to look for conceptual continuity in the available data from a very small number of speakers using any one of the frameworks. What has remained and in still in circulation as the basic primitive concepts of an endangered speech community are diluted in a scenario of imposed multilingualism and other bio-sociological reasons that may not be linguistic per se. Is it possible to recover these primitives from the very detailed analysis of the lexicographic structures? In this paper, what the author would like to argue for is the hypothesis that conceptual structures of a language that has long lost its proficient users can never be relocated completely and satisfactorily by eliciting linguistic data how much vigorous and technically sound that activity may be. To support the claim data from speakers Great Andamanese vocabulary will be provided and tentative analysis will be put forward as still the work is in progress. (Contains 3 tables.)