We are pleased to welcome Laura Kalin to our department. She earned her Ph.D. from UCLA in 2014 and is now a postdoctoral scholar at the University of Connecticut working under the supervision of Jonathan Bobaljik. Her research interests include syntax, morphology, syntactic variation, case and agreement, aspect, Aramaic, Malagasy, Hixkaryana, and Indo-Iranian languages.
She will be giving a talk on Friday, January 15, at 3 PM in SS 560A: "Phi-features as derivational time bombs: A new model of nominal licensing." A reception will follow in the department lounge.
The prevailing model of nominal licensing since Chomsky 2000, 2001 involves uninterpretable/unvalued phi-features entering into Agree with the closest c-commended nominal that has uninterpretable/unvalued Case. In this talk, I turn (most of) this paradigm on its head. Based on novel observations about parallels between the Person Case Constraint, (i), and Differential Object Marking, (ii), I propose that nominal licensing is driven by the needs of (certain) interpretable phi-features, rather than by abstract Case.
(i) The Person Case Constraint (PCC; canonical strong version, Bonet 1991): In a combination of a weak direct object and a weak indirect object, the direct object has to be third person.
(ii) Differential Object Marking (DOM; Combie 1979, Bossong 1991, Aissen 2003, inter alia): Objects high in definiteness/animacy are more likely to be overtly marked via case or agreement than objects low in definiteness/animacy.
The relative crosslinguistic stability of the PCC tells us that the interpretable feature [PARTICIPANT] is universally a derivational time bomb (in the sense of Preminger 2011), requiring licensing (via Agree). The crosslinguistic variability of DOM tells us that the choice of which other nominal features need licensing is a language-specific choice. In addition to derivational time bombs on nominal features, the proposed system is driven by a distinction between obligatory and secondary licensing loci (Levin and Massam 1985, Bobalijk 1993, Rezac 2001, inter alia), which secondary licensers activated only when a feature that is a derivational time bomb would otherwise go unlicensed.
This proposal has a number of benefits. First, two licensing phenomena that are typically considered to be non-canonical - but are in fact extremely common crosslinguistically - are reanalyzed as core instantiations of licensing. Second, while the PCC and DOM are unified under a consistent analysis, the system also correctly predicts where PCC effects will emerge as compared to DOM effects. Third, this account allows us to generalize over all nominals, without attributing special properties/features to subjects and indirect objects in opposition to direct objects, a weak spot in most previous accounts of the PCC and DOM. Finally, the proposal captures definiteness and animacy hierarchies without requiring such hierarchies to be primitives of the grammar. To conclude the talk, I speculate about why it should be that interpretable phi-features need licensing, and suggest that this reflects the need for certain phi-features to be anchored to the speech act (in the spirit of Ritter and Wiltschko 2014) in order to be correctly contextually interpreted.