This paper argues, based on novel evidence and data from the literature, that neither theories of weak familiarity (Arkoh and Matthewson 2013) nor situational uniqueness (Bombi 2018) adequately capture the empirical terrain of definiteness in Akan. I show that the distribution of the Akan definiteness marker nó spans the strong and weak article definite contexts described in Schwarz (2009). The only context where nó is not allowed is when the noun is semantically unique. To account for this distribution, I argue that nó carries a weak familiarity presupposition in line with Arkoh and Matthewson (2013), but it also carries an antiuniqueness presupposition as proposed by Robinson (2005), Dayal and Jiang (2022) for demonstratives. It takes a property and returns the unique familiar entity that satisfies that property on the condition that there is more than one satisfier of that property in the world. Regarding the definite readings of bare nouns in Akan, I contend that, in addition to nó, there is also a covert iota type-shift. The fundamental assumptions underlying this analysis are that type-shifters and their rankings are universal and that the blocking principle is activated if the lexical item and the type-shifter have the same meaning. The distribution of nó and bare definites, along with the demonstrative saa...nó, suggests a distinct determiner system that differs from languages with two definite determiners, like German (Schwarz 2009), and languages with bare nouns, like Mandarin (Jenks 2015).
