Since `possible translation' is defined in terms of `having the same syntactic meaning', the relation is symmetric. That is, for each pair of sentences and , translates to if and only if translates to . Because we preserve the syntactic structure only to a certain degree (wherever possible), we suspect that the relation is not transitive.
Given the assumption that linguistically possible translation is a symmetric relation, we have defined the computational model in such a way that it is reversible (see below).