Department of Linguistics
Monday, November 12th, 2012
McLennan Library Building Rm. M5-37A
Aspects of Argument Licensing
This thesis reexamines the distribution of arguments in different languages, proposing that Case itself is not universal, but a realization of an underlying principle of Argument Licensing. I claim that arguments must be syntactically licensed (i.e. by Agree) but that the features by which arguments are licensed can vary between languages. I discuss in detail the realization of Person Licensing in Ojibwe in the absence of Case, as well as in other languages showing Person Restrictions. Argument Licensing facilitates a deeper understanding of the principles that underlie arguments and their behaviour in the syntactic derivation, without being restricted by the exact realization of Case alone.
I first present a detailed analysis of Ojibwe verbal morphology, focusing on the infamous Algonquian Inverse System and verbal theme-sign suffixes that I claim encode the person (π) features of multiple arguments (see also Bloomfield 1957; Rhodes 1994; McGinnis 1999; Bruening 2001, 2009). I adopt the notion of Cyclic Agree of a π-probe on v with multiple arguments from Béjar & Rezac (2009), and posit morphosyntactic π-features (e.g. Harley & Ritter 2002) organized by entailment relations (see Cowper 2005). I revise the mechanics of Cyclic Agree to fully account for the Ojibwe data, including apparent intransitive morphology on a transitive verbal complex and the conjunct morphology found in embedded clauses. I claim that Person Restrictions (i.e. the Strong Person-Case Constraint, Bonet 1991) in Ojibwe ditransitives are accounted for by Cyclic Agree and the requirement of Person Licensing (i.e. π-features on arguments must enter an Agree relation). I argue that Person Restrictions are generally derived by a need for Person Licensing, and that similar restrictions in French, Spanish, Basque, Icelandic and other languages directly relate to the derivation of the Ojibwe Inverse System by Cyclic Agree with v (related approaches found in Anagnostopoulou 2003, 2005; Béjar & Rezac 2003; Adger & Harbour 2007; Heck & Richards 2010).
I further claim that Ojibwe is a caseless language, indicated by a lack of Case phenomena (see Ritter & Rosen 2005; see also related work in Ritter & Wiltschko 2004, 2007, 2010), and that Ojibwe arguments are subject to Person Licensing in the absence of Case. I claim that Case and Person Licensing are distinct realizations of the underlying principle of Argument Licensing (a generalized version of the Case Filter), and that Person Licensing is identified by a bundle of properties relating to the checking of π-features that are not shared with a standard view of Case. My view of licensing allows for a three-way typology where a language may license arguments by Case, Person, or by both as in Romance languages that have Case phenomena as well as Person Restrictions. I discuss a range of data that fall under my view of Person Licensing centered on v, including the Person Case Constraint, psych verb constructions and split-ergativity by person.
My main proposal is that Case itself is not a universal of human language but is instead a robust realization of an underlying universal of Argument Licensing. I claim that my approach to the distribution of arguments achieves better coverage of the data by allowing for greater cross-linguistic variation that is unified at a deeper level by the principle of Argument Licensing.
followed by a reception in the lounge
All are welcome!