This study addresses aspects of the phenomenon of suppletion which are relatively rarely addressed: how can semantically distinct lexemes come to acquire the status of synonymous paradigmatic alternants, and why do those alternants assume their particular paradigmatic distribution? I am especially concerned with semantically 'canonical' suppletions, where there is no possibility that what is involved is a kind of lexicalization of a morphosyntactic distinction having particularly high 'relevance' to the suppleting lexeme, since there is simply no difference of meaning between the alternants. A survey of such suppletions in the verb across the Romance languages yields two main findings: (i) that the source of the synonymy which diachronically feeds suppletion is either linguistic borrowing, or lexemes which are in a relationship of hyponymy-hyperonymy, where the semantic distinction is inherently tenuous; (ii) that in Romance suppletions have paradigmatic distributions that almost never 'make sense' in terms of being aligned with any kind of morphosyntactic meaning - and that even when they appear to,make sense', there are reasons to suspect that this fact is fortuitous. Rather, suppletions show a strong tendency to become aligned with recurrent, but profoundly idiosyncratic, patterns of paradigmatic allomorphy largely resulting, historically, from sound change. I suggest that one of a number of possible reactions to synonymy (something generally disfavoured in the lexicon) is to map semantically vacuous lexical distinctions onto semantically vacuous paradigmatic distinctions. In conclusion, I propose that studies of suppletion in all languages might benefit from asking more often why suppletions have the paradigmatic distributions they have, and being prepared to accept the possibility that an answer may lie in synchronically abstract and highly idiosyncratic aspects of paradigmatic structure.
