A secret agent in the middle?

This paper is concerned with agentivity in middle constructions in English. More specifically, it investigates what the reported “agentive flavour” is. The paper argues that there is at least some evidence for an Agent argument being structurally present in some way in the middle. If that is the case, however, it has to be explained why some of the tests show negative results for the presence of an Agent. Other possible ways to analyse middles are also discussed, one of which claims that there is no Agent in middles, at any level. The agentive flavour is instead due to an association between certain verbs and the participants they normally take when used to refer to an event. However, as is discussed in the end of the paper, there is structural evidence that such an analysis is untenable.