Evidence for DP in Chinese from Reduplicative Classifiers and DP-Internal Information Structural Phenomena

Abstract

This dissertation argues in favor of the Universal DP Hypothesis, which claims that all languages have DPs which contain extended functional projections above the NP, as opposed to the Parameterized DP Hypothesis, which argues that only languages with overt articles project DPs, and languages without overt articles have NPs as nominal arguments. I mainly focus on three kinds of data: classifier reduplication, topic and focus within DP, and modifying phrasal constituents. The first type of evidence is reduplicative classifiers. In order to make a case for functional projections above NP and the Universal DP Hypothesis, I adopt Travis’ (2001, 2003) framework to account for two types of classifier reduplication within the Chinese nominal. First, plural reduplication moves a Cl to NUM and creates a copy of it at NUM head. Second, the “Each/Every” type of reduplication in Cantonese can occur without a preceding sentential topic or adverbial because of CL-to-D movement. The moved classifier takes on the function of a determiner and becomes an outer layer of restriction for the quantifier following it. This also corroborates my thesis that the DP layer exists in Chinese. The reduplication facts shown in this chapter also demonstrate that extended functional layers exist above the NP. This suggests that the Universal DP Hypothesis is correct. The second type of evidence is topic and focus within DP. I mainly study the non-canonical NP-Num-CL order (NP Inversion) and NP Ellipsis. For NP Inversion, I followed Simpson (2005), Lin (2010), Hsu (2012) in claiming that the non-canonical NP-Num-CL order is derived through movement of NP to the left periphery of DP. I showed that the competing analysis presented by Tang (1996), which argues that the NP and [Num-CL] are in a predication relation, is problematic. In particular, that analysis predicts that [NP Num-CL] sequences only appear in clause-final position. I showed that this prediction is not borne out. For NP Ellipsis, I adopted Ntelitheos’ (2003) (see also Corver and van Koppen 2009) proposal and treats nominal ellipsis as DP-internal topicalization followed by movement of the focused remnant XP. This proposal resolves the shortcomings of empty category approaches to ellipsis like Lobeck (1995), while making the case that nominal ellipsis involves discourse-related projections just like verbal ellipsis (Johnson 2001). I showed that NP Ellipsis in Chinese is also PF deletion of [spec, DTopP] at the left edge of DP. However, since Lin (2010) has shown that NP can transformationally move to [spec, CP] from inside the DP, it was also important to show that the PF deletion of NP happens at the left edge of DP, and not CP. Consequently, I showed that NP-Ellipsis happens at the left edge of DP by demonstrating that movement of NP to the left periphery is subject to the Complex NP Constraint. The above arguments offer clear evidence that there are information structure related positions in DP. Finally, I argue that by assuming a DP-left periphery, we can not only account for the high and low positions for modifying constituents in the Chinese DP, and also relative order of APs and RCs that modify nouns. I follow den Dikken & Singhapreecha’s (2004) Predicate Inversion approach to modifying constituents in Chinese. It’s been found that grounding relative clauses or APs must appear before their non-grounding counterparts. I argue that in Mandarin, DE is the spell out of n, DFoc, and DTop. The two DEs have different functions. DE in DTop expresses specificity, definiteness, or genericity. DE in n carries a nominalizing meaning. Phrasal constituents are base-generated low as small clause predicates contained inside nP. They move to [spec, nP] to check [Nom]. If information structure is involved, they will move further to [spec, DTopP] (for grounding phrasal constituents) or [spec, DFocP] (for non-grounding phrasal constituents), Therefore, even though Chinese does not have overt determiners, an unpronounced D layer can still be detected by observing the effects they have on modifying constituents. I therefore conclude that Chinese nominals exhibit the same properties as languages with a determiner.