Recursion is assumed to be central in syntax to help us make “infinite use of finite means” (Wilhelm von Humboldt), while phonology is often claimed to lack recursion. Neeleman & van de Koot (2006) present a detailed argument for fundamental differences between syntax and phonology; recursion being one of them. In a similar vein, Jackendoff (2007: 39) takes phonological structures to be “not recursive [since] they cannot be embedded indefinitely deeply in other structures of the same type. […] For example, a rhyme cannot be subordinate to a syllable that is in turn subordinate to another rhyme.” Note here the equation of recursion with self-embedding (“same type”) and the assumption that common notions like syllables and rhymes are adequate phonological objects.
I will look at both issues in the first part of my talk and their consequences for Government Phonology (GP) 2.0, a successor of “classical” GP. According to GP 2.0, the objects that earlier phonological theories have been dealing with are too coarse. The level of magnification needs to be increased, as a result of which notions such as segment, syllable and rhyme become epiphenomenal at best. The ultimate objects that phonological structures are made up of are (subsegmental) positions, which are combined in a way similar to how minimalist merge (Chomsky 1995) combines objects. Merge is a recursive operation joining two objects in a set, with the labelling of that set being an independent issue. Self-embedding is thus only one possible outcome of recursion, but not the defining one. Under such a definition, and with household concepts of phonological constituency gone, it also makes sense to speak of recursion in phonology.
In the second part I will look at reasons why we should think about phonology as parallel to syntax. Neeleman & van de Koot (2006) point out (correctly, I believe) that any argument in favour of parallels between the two modules must not only show that phonology can be done with hierarchical structures, but why it must be. Syntax needs hierarchical structures to express asymmetries, and similar asymmetries in phonology can be found between elements, i.e., the successors of phonological features. Asymmetries between the elements I and U (Pöchtrager 2009, 2015) seem to be particularly wide-spread, and we will see how a syntax-inspired phonological model can make sense of them in Mandarin, Japanese, English, Finnish, Turkish etc.