In this paper, I argue that metonymy, metaphor and personification have to be described as reinterpretation operations that are used if the standard interpretation of a predicate-argument structure either violates semantic restrictions or is contextually implausible. These processes differ with respect to their focuss on three different concepts. For metonymy, the interpretation conflict is resolved by introducing a second argument, namely a definite individual standing in a real world relation to the original argument. Metaphoric reinterpretation adds a second predicate that has a broader meaning than the original predicate and thus avoids the conflict. Personification introduces a second world, more precisely a fictional text world, where the original conflict does not occur. The three processes have in common that, firstly, they are non-destructive, i.e. they do not change the given lexical items, but add a specific component to enable a suitable interpretation, and, secondly, they can apply freely whenever a semantic or contextual mismatch occurs, i.e. they need not to be licensed by specific lexical items.
