Coordination in natural language is often syntactically explained by the rule schema X longrightarrow X′ and X", where X′ and X" refer to expressions of a like syntactic type. While this schema does explain a lot of instances of coordination, it is well known that this schema also gives rise to serious structural ambiguity, leading not only to numerous unintended interpretations but also to a high degree of unnecessary computational complexity. In order to identify intended interpretations, we need to take into account a number of clues at all the levels of morphology, syntax, semantics, and pragmatics/discourse. In this paper, we propose, with an accompanying data analysis, principles of symmetry and coherency for an effective and reasonable identification of intended interpretations from coordination-bearing sentences.