Why do we think polysemy exists? Revisiting traditional arguments in light of recent scepticism

Nicholas Allott (University of Oslo)

It’s traditional to distinguish between monosemy, homonymy and polysemy, where a polysemous lexical item is one with more than one related sense. Polysemy is poorly understood but has received considerable theoretical and experimental attention in recent years, including at least one paper arguing that it doesn’t exist (Brody & Feiman, 2024). 

I focus on two traditional motivations for taking words to be polysemous. The argument from translation claims polysemy is evidenced by some (not all) cases where a word in one language requires two distinct lexical items to translate it in another. The argument from meaning change claims that historical changes of sense rely on a stage where the lexical item is polysemous. I try to evaluate the reach and force of these arguments, in the process articulating some of the assumptions they rely on and shedding light (I hope) on a number of live discussions, including whether regular and irregular polysemy are distinct kinds, whether polysemous words have univocal lexical representations, and when polysemy claims are about lexical representations and when they are about components of a proposition expressed by the speaker of an utterance. In passing, I suggest that Borer’s view of lexical semantics implies that one of Brody & Feiman’s arguments has a broader reach than they have claimed.