Processes of unbounded spreading are often claimed to be myopic (e.g. Wilson 2003, McCarthy 2009): the ability of some feature [F] to spread from some segment z to some segment y does not depend on its ability to spread from y to x. Recent work (e.g. Walker 2010, 2014; Jardine 2016) has however cast doubt on the universality of this claim. This paper contributes to the discussion on (non-)myopia on by suggesting that a kind of non-myopic process, trigger deletion, is attested in Gurindji (Pama–Nyungan, McConvell 1988): when the spreading domain contains a certain kind of blocking segment, the spreading trigger deletes. In order to capture this pattern, as well as the extant typology of non-myopic processes, I argue that any successful analysis of unbounded spreading must allow surface candidates to be globally evaluated.
