A central question in second language (L2) acquisition is: What is easy and what is hard in L2 acquisition (DeKeyser, 2005)? Various answers and accounts have been provided, including the Interface Hypothesis (IH; Sorace, 2011), the Feature Reassembly Hypothesis (FRH; Lardiere, 2009), and the Cue-Based Retrieval Models (CBRMs; Cunnings, 2017). This thesis aims to examine the central question by testing L2 representations and processing of Chinese nominal phrases, an area that has received scant attention in existing research. We propose that L2 acquisition outcomes are contingent on first language (L1) transfer, L2 input, and the interface. We explore the L2 representations of several properties related to Chinese nominal phrases: The distinction between classifiers and measure words, the incompatibility of numeral classifiers with the collective marker -men, discourse and stylistic constraints on dropping yi (one) in Chinese number expressions, and the pragmatic constraint on generic yi-CL-N. We also investigate the L2 processing of classifier-noun (dis)agreement. A self-paced reading task and an acceptability judgment task were administered in this study. The L2 participants were from different L1 backgrounds (i.e., English: a non-classifier language; Korean: a classifier language). The results of this research do not support the IH, positing that some of the external interfaces are acquirable but some are not. Moreover, we find that L1 transfer profoundly affects L2 acquisition, with the reassembly of transferred features causing difficulties even for advanced learners, corroborating the FRH. The findings also lend strong support to Full Lexical Transfer models (Sprouse, 2006). In line with CBRMs, we observed an increased susceptibility to similarity-based interference among L2 learners. However, we discovered that the weighting of semantic cues from classifiers in L1 significantly impacts learners’ sensitivity to classifier-noun agreement and disagreement in L2 processing. We therefore contend that the presence or absence of classifiers in L1 can substantially influence L2 learners’ capacity to retrieve cues and their sensitivity to classifier-noun agreement violations.