When the study of relationships focuses on the dyadic level of analysi
s, it must also take account of the social context in which relationsh
ips come into being. That context is strongly bordered by talk and the
rhetorical basis of normal (and scientific) discourse. Discourse and
other actions typically reproduce the relationship in its own image on
successive occasions, though they do not have to do so. The appearanc
e of stability in relationships is thus the result of perpetually cons
tructive activity of partners, not the result of 'the nature' of the r
elationship itself.