It's a two-headed monster!

While working on my thesis, I came across the following interesting sentence today:

Taro-wa [[[dare-ga neko-o turete kita] -no] -ga nigedasita ka] sitte iru.
Taro-TOP who-NOM cat-ACC brought along NM NOM ran away Q know.
Lit. "Taro knows who(x) brought along a cat and that the cat that x brought along ran away."

Of particular interest to me is the nesting of relative clauses here. We have our most embedded clause, "dare-ga neko-o turete kita." This clause then merges with a matrix clause, "... -ga nigedasita ka." In this matrix verb, neko is interpreted as the subject of "nigedasita." However, this is further embedded in another matrix clause, "Taro wa ... sitte iru." Oddly enough in this one, dare "who" is interpreted as the object of "sitte iru!" The innermost embedded clause provides two separate heads for two successive matrix clauses!

Has anyone ever seen data like this from any other language? This is particularly complicated by the fact that the above example is a head-internal relative, and not an English-style external relative (I can't even think of an equivalent English example). I'm quite curious as to what is going on here.

Thanks everyone!

As one might guess from Oliver's last post, the 2nd Annual Cornell Undergraduate Linguistics Colloquium has come and gone. Thank you to all of our participants for braving the Ithaca weather to come and make it a wonderful weekend!

We are hoping to put out a conference proceedings by the end of May - most likely in a widely accessible online format. It will contain both the abstracts and papers that were presented at the conference.

Colloquium, this weekend

At long last, the 2nd Annual Cornell Undergraduate Linguistics Colloquium is almost here. Please feel free to join us in Morrill 106A. You can find the nearly final schedule here.

Back in business

At today's UnderLings meeting, we decided to revive the undergraduate blog without renewing the domain, which means we're back to good old Blogger. As the current Events Coordinator and Field Marshal of Keepin' It Real, I think this is a good idea; the current seniors in particular have got some really interesting research going on, and we ought to be sharing it. Also, an informal blog like this will be a great way to get new people interested in Cornell UnderLings.

I believe I'll start keepin' it real by calling on my fellow UnderLings to start posting again. Let's hear about some of these upcoming honors theses. If you're not doing a thesis, don't let that stop you–post about what you're learning in your ling classes, or about something you heard on the street that got you thinking. I wouldn't mind getting some responses to the article in the Sun about Cornellian slang– any takers?

And as long as I'm giving out ideas, how about a shameless plug for the 2008 Cornell Undergraduate Linguistics Colloquium? Abstracts are due in a week (Feb. 8th), so it's not too late, but it will be soon. Mark your calendars for March 8-9th if you're on campus, and be sure to show up even if you aren't presenting, because it's sure to be a great event. Here's the call for papers.

Stay tuned for some tidbits about what I'm up to with the semantics of Japanese honorifics. To any visitors, welcome, and to the contributors, welcome back.

New Blog!

Fortunately (or unfortunately) LA is moving off blogger!

We've harnessed the power of Ruby On Rails via Typo to move to our new address:

http://www.linguisticsanonymous.org

So I'll leave this blog up for a bit, but I''ll come down eventually. If you're a contributor and didn't get an email from me, let me know.

Thoughts on (Structural) Case

So I meant to get this up while I was writing the paper, but I'm sure we're all aware how much extra time one has during finals. Anyway, this post is an attempted answer to the pretheoretical problem I discussed last time.

The main issue with the standard agreement/case theory is that it can't necessarily make any claims about the uniqueness of the Spec-Head relationship. However, this is exactly the relationship that seems to be needed to explain stronger agreement patterns in asymmetrically agreeing languages. More specifically, we have the following generalization to account for:

Let H be a head and K be a DP with phi-features which agrees with H. If H manifests a feature F morphologically when H c-commands K, then H will manifest that feature when K is in [Spec, H].

Obviously this generalization says nothing about the converse, and it is meant to capture increased agreement in the Spec-Head configuration.

This kind of agreement pattern is evidenced in Arabic, as we have already seen. However, there is an analogous asymmetry in Berber involving case. The standard word order in Berber is VSO, with SVO as a free alternate. In the VSO configuration, the subject DP is not in the nominative case, but instead in a form known as the Construct State (CS - not to be too closely confused Semitic construct state genitives). In the SVO configuration, however, the nominal appears in the nominative. Typically, these two phenomena are treated separately - but what generalizations can be drawn from viewing them as two sides of the same coin?

A very important one - that structural case features may not be necessary in the narrow syntax, and indeed may even be problematic.

Recall from the previous post that a phi-defective head cannot assign case to a DP. If we assume that, in languages like Berber and Arabic, the phi-defective head T can exist in matrix clauses, we come much closer to explaining this asymmetry. Specifically, it can explain why there is weak phi-feature agreement in VSO configurations in Arabic, and the appearance of a (presumably) inherent or oblique case on Berber VSO subjects. We could then say that the phi-complete T is responsible for agreeing with the subject DP, valuing its case nominative, and raising the subject to [Spec, T] and producing an SVO configuration. In these cases, we would expect full agreement and nominative case because of the feature-completeness of T.

However, this does not get us all the way. My next post will explore the need to revise the notions of EPP and the definition of nominative case, as well as discuss why case features such as [NOM], [ACC], and [uCase] may not be necessary.

distance learning phonetics anyone?

A heads up to anyone interested in learning (or refreshing their) phonetics: University College London is running a pilot distance-learning course this summer from 12 June to 04 August. Since it is a pilot version, the course is completely free-of-charge. The course focuses on English phonetics, and they estimate 2-3 hours/wk of online material. The profs running it will be looking for feedback during the whole process too, so if you want to be part of the pilot you can get more info and apply at: www.phon.ucl.ac.uk/home/phonline/

Finals, and a Theoretical Problem

Okay, so it's finals time here, which means, at least for me, copious amounts of paper writing. In an effort to organize my thoughts, I'll be posting probably the better part of all the major arguments from my papers over the next few days. At the end I'll put up references and organize the posts, much the way Jedediah did. The first of these is the pretheoretical problem for my syntax paper, which is on the theoretical status of structural case.

Chomsky (2000) notes that "...[W]e assume that only a probe with a full complement of phi-features is capable of deleting the feature that activates the matched goal." This is used to keep participles from deleting structural case and freezing their subjects in embedded clauses. Continuing, Chomsky goes on to say that the Spec-Head relationship is obtained by EPP (Extended Projection Principle), which in recent works (Chomsky 2000, 2001) is assumed to be a reflex of phi-feature checking. This means that the Spec-Head relationship should, presumably, have no special theoretical significance than relationships established under c-command specifically, or more generally, than relationships established by Agree(H, K) where H is a head/probe and K the matching DP/goal. Since Move and Agree are separate operations, Agree(H, K) does not imply Move(K, Spec-H).

This is an interesting conclusion in and of itself, that the Spec-Head relationship should not, in some way, be "special" as was assumed under Government and Binding. However, when we begin to consider structural case. Case is presumed to be assigned under an Agree relation between a case assigning head H and a goal K which agrees with H. So regardless of whether or not there are such abstract features as [NOM] and [ACC], they can be checked on a head H iff Agree(H, K). Therefore we are back to the first point from this post: a head H with incomplete phi-features cannot assign (structural) case.

This is an issue, however, as it appears to be patently false in Arabic and possibly Kabyle Berber. More specifically, in Arabic Spec-Head subjects appear to be special since they can license full agreement whereas c-commanded subjects cannot. Moreover, Berber postverbal subjects appear in a different case known in the Berber literature as the Construct State of the nominal. Inter alia, Ouhalla (1995) has analyzed this as genitive case, whereas preverbal subjects are nominative.

But how can this be? The Spec-Head relationship is not supposed to be "special"...

More on this later.