The classical study of ritual centres on the question of the connections between features of ritual action, on the one hand, and various religious representations, on the other. The connections are supposed to go in both directions. On the one hand, rituals are described as contexts in which particular religious models are ‘expressed’, displayed, made manifest in actions and statements (see e.g. Beattie 1970). Conversely, it is generally accepted that participants in ritual make use of various shared cultural models to give ‘meaning’ to their actions. Both assumptions, however, are generally couched in such vague terms that they do not seem to have any practical implications for the study of religious ritual. The first hypothesis, which could be called the ‘expressive’ understanding of ritual, seems to be mostly a projection of the anthropologists’ own methodologies. Because anthropologists consider rituals as a primary source of information about the religious representations of a given group, they spontaneously tend to consider that that is the way in which the participants too represent their rituals. This, however, is not really supported by the evidence. The hypothesis is all the more difficult to maintain, as the particular aspects of ritual contexts make communication and therefore ‘expression’ particularly awkward. As Bloch pointed out (1974 passim; 1985: 19–45), religious ritual makes constant use of behavioural modes (dance, special languages, stereotyped gestures) which, if anything, reduce the range of information that can be conveyed by their performance. In other words, rituals are remarkable in that they are particularly nonexpressive, as other authors have stressed (Rappaport 1974; Staal 1979).