Heidi Harley
University of Arizona
The "bundling" hypothesis and the disparate functions of little v
I first consider several case studies that suggest that Pylkkänen’s 2002, 2008Voice-bundling parameter is on the right track. There are bundling languages where it appears that both v and Voice functions are tightly correlated, appearing and disappearing together; and there are splitting languages where the functions are distributed across two distinct projections and can be manipulated independently. The bundling languages I will consider are Chol and Persian, while for splitting languages we look at Hiaki and Chemehuevi. We apply the predictions of the splitting/bundling parameter to the interaction of passive and light verb constructions in Italian, suggesting that it is Voice-bundling.
I then focus narrowly on v, reviewing arguments from Key (2013) and Jung (2015) about productive causatives, applicatives, and passive, looking at Key’s treatment of Turkish causatives and Jung’s discussion of Korean and Hiaki applicatives and causatives. These patterns indicate that productive causatives are not a recursive v, as assumed in Harley (1995, 2013), but instead realize a pure “Caus” category. The verbalizing v and the causativizing Caus are categorically and morphosyntactically distinct. Finally, a sketchy, possibly cartographic picture of the hierarchy of derivational verbal projections begins to emerge.
Dear Heidi,
I find the reasoning to be very interesting. You examined my PhD thesis way back and I learnt a lot from your comments and I am still learning from your work! I work on Berber (especially, Tarifit Berber), which also resists passivization. Berber does have a passive morpheme but it can be found only with a very small number of verbs (3 or 4). Typical transitive-agentive verbs realise their passive using inchoative or middle passive kind of morphology similar to what you discussed from Persian. Some other verbs resist all kinds passivization and can only be used in the active voice. Berber has also a causative (agentive) morpheme but never appears with unaccusative verbs. This causative morpheme can also combine with a lexical root to derive an intransitive-unergative verb with a subject-causer. Under the ‘bundling’ hypothesis, it can be concluded that Berber is a Voice-bundling language. I was always puzzled by the issue of transitive verbs resisting the passive in my language. However, I have two questions about the analysis and I would be most grateful if you could spare some time to clarify them:
(1) In the Italian examples in (21), you show that fare ‘make’ can be passivized only when it is used as the main verb. But the verb resists passivization when used as a light verb. But why is it that fare ‘make’ can still be passivized as the main verb? my understanding is that any verb that can be passivized has a Voice or (agentive) little-v in the active sentence. There must be a reason in your reasoning but somehow I cannot see it (sorry!).
(2) The second point has to do with English. Pylkkänen (2002, 2008) observes that Japanese can causativise unergative verbs (takes 2 arguments) which she takes as evidence that it is a Voice-splitting language, but this option is not available to English and is therefore Voice-bundling. However, Voice-bundling languages are expected to resist passivization but English does not. Thanks very much in advance.