| dc.description.abstract | This thesis explores the noun-verb categories of disyllabic compounds in Mandarin under the theoretical frameworks of Distributed Morphology (Halle and Marantz 1993, Harley and Noyer 1999) and Exo-skeletal Model (Borer 2005a, 2005b, 2013, 2014). This study focuses on two main research questions. First, are there default categories or specific headedness rules for Mandarin compounds? Second, are there different degrees of malleability among different types of compounds? To investigate these questions, I conducted an experiment with newly-composed Mandarin compounds in NN, NV, VN, and VV patterns. Test 1 was a classifier matching test, where participants were asked to choose either the nominal classifier or the verbal classifier to match the neologism. This corresponds the first research question by examining only the bare forms on these neologisms. Test 2 was an acceptability test, asking participants to rate the degree of acceptability for neologisms under nominal and verbal functional structures, which corresponds to the second research question concerning malleability. Based on the results, which revealed different preferences and probabilities in Test 1, I proposed a language perception model with probabilities. In this model, syntax generates all possible competing structures and probabilities were assigned before they are sent to Logical Form (LF) for further meaning processing. I also propose that probabilities for Mandarin compounds are assigned based on a new Headedness Model with weighted constraints. The head of a compound is determined by probabilistic reasoning under various morphological constraints. The nominal preference of VN compounds with unaccusative monovalent verbs is also explained by the lack of VoiceP. In Test 2, VN compounds demonstrated greater malleability than VV and NN compounds due to their richer potential in syntactic structure generation. The non-preference for NV compounds is attributed to their non-productivity in Mandarin. | |
