A Typological Analysis of Wh-Questions and the Scope of Wh-Words

This paper argues that the wh-scope is marked via a pure AGREE relation that holds between a probe on C and a goal on wh-phrase; the wh-scope in the languages with syntactic wh-movements such as English, German, Bulgarian, and Polish is marked via AGREE between their [wh]-features, whereas the wh-scope in the languages with Q-morphemes such as Korean and Japanese is marked via AGREE between their [Q]-features. This is based on two assumptions. One is that the Phase Impenetrability Condition (PIC) is only sensitive to EPP properties. The other is that an EPP property triggering wh-Move invariably resides in the probe u[Wh] on C alone. The former is to account for why AGREE for wh-scope is immune from the PIC, but overt syntactic wh-Move cannot be free from the PIC. This is the reason why overt wh-extraction out of the Islands cannot be free from the Island violation in English, but wh-in-situ in the Islands is immune from the Island violation in Korean and Japanese. The latter assumption is to account for why Korean and Japanese cannot have syntactic wh-Move to Spec,CP, I further propose that an intermediate C may have a defective u[Wh]EPP, distinct from a complete u[Wh]Q,EPP, only when the CP is selected by a lexical head. This proposal guarantees a wh-phrase to undergo successive cyclic wh-movement in the languages with overt syntactic wh-Move on the basis of AGREE theory. The above assumptions are supported by ‘that-t’ effects, Island phenomena, higher wh-effects, Superiority Effects, intervention effects on the scope-bearing elements, and wh- expletive constructions, along with a typological distinction between Bulgarian-type languages and Polish-type languages. I argue that Superiority Effects are triggered in English only when the probe u[Wh]Q on an interrogative C is [+multiple], but the EPP property becomes a 1-spec property. However, Korean and Japanese allow Multiple AGREE with respect to the u[Q]wh probe with a [+multiple] property, but they cannot have the EPP property to induce overt wh-Move because the EPP property to induce overt wh-Move invariably resides in the probe u[Wh] on C alone. This accounts for why Korean and Japanese don’t exhibit Superiority Effects in multiple wh-questions. The typological difference between the Bulgarian-type languages and the Polish-type languages, allowing Multiple AGREE and Multiple Move at a time, is ascribed to the presence or absence of a tax-paying for Multiple wh-Move. I argue that the Bulgarian-type languages, unlike the Polish-type languages, pay a tax for Multiple Move, so that they ought to obey Superiority in multiple wh-questions. I propose that an intermediate C optionally allows the defective u[Wh]EPP, aside from the complete u[Wh]Q,EPP, in the Bulgarian-type languages, but not in the Polish-type languages, so as to ensure successive cyclic wh-fronting via the intermediate Spec,CP. This accounts for why the Bulgarian-type languages, unlike the Polish-type languages, are free from wh-island constraint. In wh-expletive constructions that appear in German, Iraqi Arabic, and Hungarian, a wh-expletive is assumed to have dual properties to function as a probe and a goal.