David Wilkins, in discussion with Kendon, Kita, Levinson and others in the Gesture Project of the Max-Planck Institute for Psycholinguistics, distinguished various possible senses of "convention" as applied to systems of gestures. One distinction is between an inventory of conventional form/motion complexes of the sort normally termed "emblems" and a wider set of conventions (e.g., of movement, of general hand shape [see Wilkins (1994b)], or of cardinal orientation of the sort I describe for GY) that may apply over a whole series of possibly even nonce gestures. One can also imagine conventions of gestural usage (limiting them to certain situations, or genres, for example) and function (evaluative metacommentary, for example, as is argued for "metaphoric gestures" by McNeill and his associates, or a kind of gestural parsing of discourse, as for example with "beats".)
Convention can also span such distinguishable arenas of gesture; thus, for example, conventions of politeness that govern the use of different hand shapes for showing the heights of children (as opposed to, say, farm animals), have been described by Poyatos (1983).
Brown and Levinson (1993) describe the cognate "uphill" and "downhill" system in neighboring Tenejapa Tzeltal, where the denotations are rotated 90 degrees ("uphill" denotes south). Although such terms are easily elicited in semi-experimental tasks where geographic orientation is the prime concern, there is nothing in Brown and Levinson's study to indicate that such directional terms are any more frequent in ordinary talk in Tenejapa than they are in Zinacantan.