One classic collection of essays by anthropologists on the definition of religion which was required reading in the course at King’s College, London on anthropology of religion was Anthropological Approaches to the Study of Religion (1966) edited by E. M. Banton. Though this is now an old book it contains interesting and influential essays by Victor Turner, Clifford Geertz, Melford Spiro and others which are still frequently referred to, especially the one by Geertz. However their overall effect has been, I would argue, to reinscribe and validate ‘religion’ into the general academic discourse on which they have had considerable influence. While raising and discussing many of the problems of applying a Europhone category in the context of radically different languages and cultures, these essays did not interrogate the ideological power dynamics behind the discourse itself. The category ‘religion’ and its demarcation from the social or secular was not systematically questioned; only the best way to define religion for research purposes.
For example, Geertz famously defined religion as “ a system of symbols which acts to  establish powerful, pervasive, and long-lasting moods and motivations…by  formulating conceptions of a general order of existence and  clothing these conceptions with such an aura of factuality that  the moods and motivations seem uniquely realistic”. But this definition arguably straddles all dominant ideologies, and does not tell us how a religious ideology differs from a non-religious one. Nor does it sufficiently draw attention to the power of dominant institutions (such as preaching, courts, persuasive theories by educated elites, advertising or the media) to protect these symbols, and to promote the sense of their inescapable reality. A powerful analogy may be from feminist analysis of the way dominant gender categories become transformed into inescapable facts of biological nature, disguising the power relations inherent in the representations. The assumption that there is some essential distinction between religious and non-religious domains – which is still today a globalizing discourse – is an ideological construct which takes on an appearance of naturalness and inevitability.
Spiro’s definition was a sophisticated reworking, in the context of his own interesting ethnography of Burmese Buddhism, of E. B. Tylor’s definition as belief in gods or superhuman agents. However, one of the problems with a definition in terms of gods or the supernatural or the superhuman is that these terms themselves are difficult to translate into many non-European languages. Even within European Christendom the meaning of God has been policed and contested by powerful theological agencies, and it is not at all clear that the Trinitarian God of the Aristotelian-Thomist synthesis is equivalent to what Calvin understood by God. The stretch may be even further to the conceptions of Unitarianism or Deism. Muslim theologians who believe in Allah have held that the Christian Trinitarian God is itself a form of idolatry.
What anthropologists and others now sometimes refer to as ‘gods’ has been used historically by Christians in the sense of false idols, pagan heresies, demons and devil worship. These theological misrepresentations of other people’s concepts do not engender confidence in their use as neutral descriptive and analytical concepts. This point is strengthened by the fact that, even today, some evangelical missionaries still hold these beliefs and still use this kind of language. For example, a Protestant mission in Mexico was motivated by the desire to save people from their pagan village economies and “raise the rate of return on conversions”.
To take just two examples of non-European languages, Sanskrit and Japanese: it is problematic to claim that gods provides a neutral translation for Indian categories such as Brahman, deva, devata or Bodhisattva; or into Japanese categories such as kami, hotoke, or bosatsu. It is equally problematic to attribute belief in the ‘supernatural’ and its supposed distinction from the ‘natural’ to non-European languages and cultures around the world. Some writers have substituted the term ‘superhuman’ as a way to resolve this problem of the ‘supernatural’ while retaining the term ‘religion’ as a distinct form of life. But if the term superhuman has any advantages, it tends to erode a distinction between ‘religious’ and ‘non-religious’ domains. In some Indian conceptions there is no ultimate distinction between the human and the superhuman, as the practice of kissing the feet of enlightened gurus and powerful politicians suggests. Many sadhus are believed to be ‘living gods’ in the sense that they have become one with the divine reality which permeates what we illusorily experience as a mundane world. This is not a pedantic distinction; the veneration given to a sadhu or a living bodhisattva is part of a total system of representations that defines the identity of billions of people.
It is astonishing that experts in International Relations believe they can classify these complex ideologies without any real knowledge in simplistic English categories and then advocate foreign policy decisions on their basis. In Japan the Emperor was ikigami (usually but perhaps misleadingly translated as ‘living god’) at a time when the Meiji Constitution of 1889 constituted State Shinto as the Japanese equivalent of the secular State. In 1946 the US Occupation forces rewrote the Constitution which declared that State Shinto cannot legally exist and that shinto is really a religion and should be classified as such; and that the Emperor is no longer Ikigami but something more like a British Constitutional Monarch. Here it is clear that power decides what gets classified as a religious belief and what gets classified as a secular one.