22 November 2017
Time: 1:00 - 2:00pm
Venue: Scape 1.04
As part of our LingLunch series, Michelle Sheehan (Anglia Ruskin) will present some of her recent research, partly based on joint work with Sonia Cyrino.
No escape hatch for A-movement: evidence from causative/perception verbs
In this talk, I defend the position that A-movement does not have access to phase-edge escape hatches. The crucial context which illustrates this is the clausal complements of causative/perception verbs. Whereas passives of ECM are generally possible, this is not the case where the complement is a bare verb in English (Higginbotham 1983, Felser 1999). This is because such complements are phases (as defined by independent diagnostics such as VP-ellipsis and VP-fronting – Harwood 2015):
(1) a. *Kimi was made/had/let seen/heard/witnessed [ti sing]
- Kimi was made/seen/heard [ ti to sing].
- Kim was seen/heard/witnessed [ ti singing].
- Sami was made [ ti angry] by the news.
I show that if we adopt PIC2 (Chomsky 2001) and Legate’s (2003) claim that all vPs are phases, we can explain these contrasts as a direct effect of phase theory: the lower vP is transferred before the matrix T probes. This is avoided in (1b-c) because a T-related projection is present and so an EPP helps the causee escape spell-out. In (1d), the small clause complement is non-verbal, hence non-phasal. As expected, such effects are not limited to English; they are widely attested in other languages with ECM complements (Brazilian Portuguese, German, Dutch, French, Italian, Spanish, Danish, Swedish). The same restriction extends, moreover, to languages which have more monoclausal causative constructions, where the causee receives dative case in transitive contexts (French, Italian, European Portuguese, Korean – Kayne 1975, Folli & Harley 2007, Gocalves 1999, Jung 2014). This has important implications for phase theory. I argue that the ‘monoclausality’ in languages like Italian and French is actually mono-phasality: where a light verb selects a vP, they form a single phase. I also discuss morphological causatives. As predicted, ‘lexical’ causatives (Pylkkänen’s 2002, 2008, root-selecting causatives) generally permit passivisation, as they are simply monoclausal. What is more surprising, however, is the fact that syntactic causatives (Pylkkänen’s 2002, 2008 voice-selecting causatives) also generally permit passivisation (Japanese, Haiki, Hindi, Turkish, Zulu, Sotho – Harley 2017, Ramchand 2009, Key 2013, Buell 2005, Machobane 1991), even though these complements also seem to be vPs. I propose that the crucial difference between Italian/French/Korean and Japanese/Hindi/Turkish is that in the latter kind of language the voice + cause + vP forms a single phase, whereas in the former both voice and the do verb both count as phase heads.