Abstract
In this paper, I discuss some arguments presented in the previous research against the deletion-based account (i.e., the Reduction Analysis) of the Mandarin b? comparative, which include the lack of the subcomparative, the lack of the embedded standard, and the distribution of the quantificational adverb d?u in this construction. Working with an Agree-based theory of selection and a featural account of PF-deletion, I show that a deletion-based account does capture these facts. The proposal is built on the assumptions that i) deletion may rescue some illegitimate representations caused by uninterpretable features at PF (Wurmbrand 2014; see also Merchant 2009; Bo?kovi? 2011; and others) and ii) the b?-constituent contains a small clause (i.e., vP) and lacks all the higher functional clausal heads (e.g., T, Asp, Mod, etc.) (cf. Pancheva 2006).