Scrambling and Surprising Constituents in Dependent-Marking Languages
Youngju Choi
UIUC Linguistics
4PM, Thursday, March 16
Lucy Ellis Lounge, FLB
The main idea of the Principle and Parameter theory is that a limited set of universal principles can explain all the languages with a small bit of parameter setting which is responsible for the differences between languages. It looked very promising when you only consider Indo-European languages. Howerver, as Baker (1996) criticized, the parameters become smaller and smaller, resulting in a microparameter for each construction. When it comes to languages which are very different from Indo-European ones, the problem becomes much more serious. The same configurational approach to nonconfigurational languages creates an infinite number of microparameters. If UG is relaxed, as claimed in Bouchard (1997, 2001), to allow various possibilities human sensory motor capacity can accomodate, the small parameters will be merged into several macroparameters.
I propose a macroparameter for dependent-marking languages like Korean and Japanese as shown in (1).
(1) Dependent-Marking Parameter : Argument Centeredness
a. Case markers change arguments into functors.
b. Arguments as functors look for appropriate types of predicates
c. Arguments combine first.
The Argument-centered property can explain various recalcitrant phenomena in dependent –marking languages, including head-finality, free word order, argument cluster constituency, pro-drop, and multiple identical case constructions. Among them, I am going to talk about free word order that is called scrambling ((2)) and argument cluster constituents, which is called surprising constituents ((3)).
(2) a. Cheli-lul Yenghi-ka manna-ss-ta
C-Acc Y-Nom meet-Pst-Dcl
‘Yenghi met Cheli’ (Local Scrambling)
b. Cheli-lul Yenghi-ka Swunhi-ka manna-ss-ta-ko sayngkakhay-ess-ta
C-Acc Y-Nom S-Nom meet-Pst-Dcl-Comp think-Pst-Dcl
‘Yenghi thought that Swunghi met Cheli’ (Long Scrambling)
(3) Cheli-nun [[ecey Tongswu-eykey kongchayk twu-kwuen]-kwa
C-Top yesterday T-from notebook two-Class-Conj
[onul Yenghi-eykey chayk sey-kwuen]]-lul cwu-ess-ta
today Y-from book three-Class-Acc give-Pst-Dcl
‘Cheli gave two notebooks to Tongswu yesterday
and three books to Yenghi today’
To implement the dependent-marking parameter, i) left-to-right incremental structure building is assumed (Phillips 2003), ii) radical lexicalism applies to bound morphemes; case markers have information that when they attach to their hosts (NPs), they let their hosts expect verbs to come, and iii) thus, bare nominals enter the computation as root and get their categorial status, VP/V, only when it is combined with case markers (flexible categorization of Borer (2003)).
Since any case-marked NP is VP/V, when it meets another case-marked NP, they combine with each other, looking for an appropriate predicate. Thus, their order can be switched (local scrambling). Due to the constraint that the NPs with the same case marker do not combine, when an argument cluster hit the second Nom-marked NP, the cluster takes the first Nom-marked NP out and takes the second one, satisfying the predicate requirement first (long scrambling). The surprising constituent is explained by competition between a case marker and a conjunction marker. When a case marker is added, a predicate is expected and when a conjunction marker is added, another NP is expected.