This paper shows how the phasal movement hypothesis (1) is supported by six criteria of EF-movement and thereby contributes to the optimality of grammar:
(1) Phasal Movement Hypothesis (Chomsky 2005)
All A’-movements are solely induced by EF (=edge feature) of a phase head, i.e., EF-
We propose and motivate six criteria of EF-movement: 1. Lack of Agree; 2. EF-property; 3. Optionality; 4. Discourse Effects; 5. Lack of Covert Movement; and 6. Lack of Agree-related Movement Constraints.
Criterion 4 of Discourse Effects is motivated in depth with respect to syntactic freezing: it is shown that the D-Effect Freezing Principle is better motivated than the Criterial Freezing Principle (Rizzi 2004) and the Inactivation Freezing Principle (Chomsky 2005).
EF-movement or pure IM is a null hypothesis for movement that approaches SMT, since it “comes free” just like EM; hence, motivation of EF-movement in natural language contributes to the optimality of grammar as follows: (a) It contributes to the optimality of movement theory through motivation of optimal movement, i.e., EF-movement. (b) It contributes to optimal characterization of other overt movements, i.e., Agree-movement and head-movement. (c) It contributes to optimal characterization of CI interface conditions necessary for the grammar of natural language.
The extension of EF-movement, or EF-movement being induced by a non-phase head, is explored and motivated for phenomena like clause-internal scrambling.
Take a concrete example (2a, b) from Korean, which can be accounted for by our six criteria of EF-movement under the assumption that Wh-movement is EF-movement:
(2) a. *[TP kui-uy chinkwu-ka nwukwui-lul ttayli-ess-ni]?
he-GEN friend-NOM who-ACC hit-PAST-Q
'Hisi friend hit whoi?'
b. [CP nwukwui-lulj [TP tj [TP kui-uy chinkwu-ka tj ttayli-ess-ni]]]?
According to criteria 2, 3, 4, and 5, (2a) is out violating Weak Crossover whereas (2b) avoids Weak Crossover in terms of the parallel movements of clause-internal A-scrambling (=extended EF-movement) and A’-scrambling (=non-extended EF-movement) of the wh-phrase. You might immediately object, saying “Wh-movement in English is different!” Well, I will show you that the difference between Korean and English is not due to wh-phrase proper but due to something else.
Take another example from Korean, which can be accounted for by our six criteria of EF-movement under the assumption that Scrambling is EF-movement and that Case particle necessarily supports focus effects:
'Only John everyone loves.' (every>only;*only>every)
b. [CP/TP John-mani [TP ti [TP motun salam-i ti salangha-n-ta]]].
J.-only every man-NOM love-ASP-DEC
‘Only John everyone loves.' (every>only; only>every)
According to criteria 2, 3, and 4, the object John-man ‘John-only’ has to move to Spec-C due to the focus particle -man ‘only’ and the accusative Case particle –ul, obligatorily inducing reconstruction, in (3a) whereas it does not have to in (3b) since it may move to Spec-T or Spec-C due to lack of the Case-particle; hence, the difference in scope interpretation between (3a, b).