In this article we investigate Halliday’s proposal to collapse cognate and locative complements into one semantic role, which he divides into a quality and quantity subtype. As empirical data we use sets of examples with cognate and locative complements extracted from the COBUILD corpus. The examples have been analysed and quantified in terms of the determiner and modifier types of the complements, which will be shown to be central to the research questions. In the first section we argue that Halliday’s semantic quality-quantity distinction should be replaced by the pragmatic-semantic distinction telic-atelic. In the second section we show that cognate and locative complements have a very different semantic relation to the verb, which does not warrant the claim that they fulfil the same semantic role.