Abstract
If signs of type X regularily cooccur with signs of type Y, this can be explained in two ways: (i) X is required by the valency of Y, or (ii) there is a complex construction Z that requires both X and Y. I argue that, when deciding between (i) and (ii), grammarians should consider, among other things, whether the cooccurence of X and Y resembles prototypical instances of (i). This recommendation is based on the fact that the properties of the prototype of (i) show that valency is a morphosyntactic feature of word forms (like plural). Therefore, if the cooccurence of X and Y does not resemble the prototype of (i), treating it as an instance of (i) would establish a wrong parallel to morphosyntactic phenomena. And, on the other hand, if it does resemble this prototype, treating it as an instance of (ii) would miss this parallel. – In the perspective of this proposal, I discuss some phenomena whose status between (i) and (ii) is controversial.
