Although a number of descriptive and predictive factors have been identified as significant in relation to language shift, a straight-forward search for the causes of language shift and maintenance has been unsuccessful. Several scholars have developed theoretical or empirical models to describe language maintenance and shift and have more or less presented these models as applicable universally. In the light of data from a language shift situation, this paper will explore the validity of two models namely, ‘Perceived Benefit Model’ and ‘Network Analysis’. The paper will demonstrate  that because of the diversity of the range of variables and their relative values in different social and cultural contexts, it is not possible to develop a mathematical formula which has the predictive power for something as complex as language maintenance and viability. The paper will conclude that in the absence of a single universally accepted paradigm for the consideration of language maintenance and language shift, the researchers should combine different models to suit the requirements of the language situation that they intend to study.