자유 관계절에 대한 소고

On Free Relatives

With a view to clarify the paradoxical nature of free relatives (i.e. DP-like distribution vs. CP-like form), this paper critically reviews some recent approaches to free relatives, and proposes an alternative that may eschew the problematic issues thereof. On the observation of the similarities and differences between free relatives and indirect wh-questions, this paper suggests that these two clauses are base generated as CPs, their differences stemming from the different types of Merge. Adopting and adapting Richards (2009), it is assumed that the wh-phrase in indirect wh-questions undergoes set Merge, accompanying Agree and results in the projection of the target, whereas free relatives involve Agree-less pair Merge of the CP to the copy of the wh-phrase, yielding an ordered pair {<D, D>, {DP, CP}}. An analysis on this line naturally explains that free relatives, being adjuncts, do not allow extraction. The labeling paradox that lexical items can provide the label of their projection though they are not heads is also clarified: They are not heads, contra Chomsky (2005) and Cecchetto and Donati (2007), but phrases in conformity to the Uniformity Condition. Thus the examples of free relatives that involve movement of DP, PP, AP and AdvP, etc are no longer exceptional to the labeling algorithm. Besides, the proposed alternative can explain the matching phenomena with no ad hoc stipulation that requires a null D head dominating CP. The issue of &#981;-criterion violation does not arise, either, since the wh-phrase is &#981;-marked by the predicate of the free relatives, while the ordered pair is &#981;- marked by the matrix verb.