A Distributed Morphological Study Of Compounding In English And Chinese

In this dissertation we adopt Distributed Morphology (DM) to analyze compounding in English and Chinese and reduplication in Chinese (which can be treated as compounding too), and to highlight the conceptual and analytical advantages of DM over lexicalism. Theoretically, our reformulation of head movement and linearization algorithm based on the LCA will optimize the current theoretical model(s) of DM. Empirically, we use our optimized DM to recapitulate the compounding (including primary and synthetic compounding) in the two languages and the morphology and phonology in Chinese reduplication. Specifically, primary compounds are classified into some subtypes which are structurally reanalyzed one by one; the morpheme-order alternation in synthetic compounds and the morphology/syntax-phonlogy interaction in reduplications are accounted for. Generally speaking, we hold that DM can provide a unified analysis of English and Chinese compounding as well as Chinese reduplication. And we draw some specific main conclusions as given below.We begin with the organization of grammar. This dissertation adheres to the basic DM hypothesis that there is an independent morphological module which is generally called MS (morphological structure) in DM literature. We contend that setting up MS is not contradictory to DM’s "Single-engine" Hypothesis that words are constructed by/in syntax. Rather, as the "bridge" between syntax (i.e., the generative system) and phonology/PF (i.e., the interpretative system), MS (and its operations) can access to various types of syntactic and phonological information they need. This not only accords with SMT, but also facilitates analyses of many particular empirical problems. For instance, different versions of head movements in both languages, morphological doubling (reduplication) and serial verb construction formation in Chinese all belong to the readjusting processes at MS; they are compromises syntax makes in response to the specific interface/PF requirements and individual morphemes’(moprho) phonological idiosyncrasies in particular languages. All in all, such grammatical design of DM facilitates the recapitulations of various linguistic facts in a certain language and many cross-linguistic variations with respect to compounding. For example, unlike English, Chinese has (ⅰ) morpheme-order alternation related to syllable counts,(ⅱ) highly productive CRD compounding and (ⅲ) reduplication, etc. All these differences in Chinese and English compounding may have their reasons found at MS.In this dissertation, Primary compounding is defined as a process wherein the uncategorized roots Pair-Merge first and the thus formed (?) continues to merge with a categorizing x. We do not regard word/compound as a primitive linguistic constituent for the following reasons.(ⅰ) Word does not fall into any specific structural level (all so-called words are phrases by analysis which we may term aP, vP, nP, etc.).(ⅱ) Word does not constitute domain of special phonology and semantics which are rigidly distinct from the syntactic domain. The boundary between word and phrase and between lexicon and syntax is not as clear-cut as lexicalists have assumed. Our work aims at recapitulating the phonological and semantic differences between word and phrase by means of a theory that does not recognize word. Analytically speaking, according to DM’s Word-internal Phase Theory, single-phased structure corresponds basically to word/compound while multi-phased structure should be analyzed as phrase. This assumption is confirmed by the fact that single-phased compound tends to exhibit non-compositional semantics and special phonology such as neutral tone or fore-stress. After forsaking the lexicalist notions of "generative lexicon" and "word", we solve the issue of compound vs. phrase distinction that has been inflicting lexicalism. This reflects the advantages of DM.We also classify compounds in English and Chinese on the basis of the hypotheses on (primary) compounding we just mentioned and Bisetto, Scalise and Guevara’s compound classification model. In English, the SUB vs. ATR distinction just instantiates the word vs. phrase differentiation, which is cogently evidenced by the different stress patterns in English SUBs and ATRs. The SUBs, with argument-head structure, assume the phase1phonology of fore-stress, while ATRs, with modifier (non-argument)-head structure, normally take the phase2end-stress (i.e., phrasal stress in our analysis)(roughly the same is true for Chinese). As to the unexpected fore-stressed ATR and end-stress SBU, we give an analysis which crucially relies on the distinction between [Attr] and [Attrdef] features. Feature-based analysis is also preliminarily extended to the issue why the phonological/metrical manifestation of the ATR vs. SUB distinction in Chinese is lost.This work proposes a new model of linearization, a new notion of head movement and the Split VP Shell or functional hierarchy depicted below.The relations among the linearization model, the new notion of head movement and the proposed functional hierarchy is as follow. The root at the bottom of the hierarchy will successively raise/head-move to merge with various functional heads which are occupied by various light verbs whose distribution is constrained in the way that is illustrated in the diagram above. Similarly, the merger of root and light verbs (the head movement) are restricted in the way that the light verbs to be incorporated make sure that the resultant composite predicate (label) is appropriately interpreted. The composite predicate formed by successive head-movement should be fully specified with the necessary and sufficient information for the semantic interpretation. Restricted in this way, the head will move to the proper position where it is Spelled-out and its linear relation with other nodes on the tree is directly worked out by virtue of our LCA-based linearization algorithm.This satisfactorily explains why the synthetic compounds in English and Chinese have OV order while phrases in the two languages have VO order. Compounds do not encode event-aspectual meaning so root is not required (and allowed) to move to the Inner Aspect head where telicity is checked. As a result, root is linearly behind the internal argument noun.We view reduplication as a word formation, more precisely, structure building process that occurs at PF. Being the secondary exponence of certain morpho-syntactic or semantic features (i.e., the primary exponence), reduplication can be categorized as a type of MS readjusting operation. For example, verb reduplication is analyzed as the secondary exponence of the feature(s)[Tentative] and/or [Frequentative] on the inner Asp node. Different from what Prosodic Morphology and OT Correspondence Theory, reduplication is not phonologically driven, but rather a morpho-semantically motivated process of morpheme-doubling (very similar to morphological doubling posited by Inkelas and Zoll (2005)). Returning to Chinese, noun and type2-adjective reduplications are derivations where a diminutive suffix (i.e., the reduplicant) is added to a root (i.e. the base). But the majority of Chinese reduplications, i.e., verb, type1-adjective and classifier reduplications are in essence "root plus root" process, which is similar to but not identifiable with compounding. As to the reduplication phonology, we criticize some lexical phonological analyses of it, specifically the practice of segmenting lexicon into sequential strata and mandating the application order of tone sandhi, stress assignment and tone reduction on different strata. Then we propose that the null cyclic functional head v triggers the readjustment rule of tone reduction, whose application domain is in turn defined by the phase structure as DM Phase Theory assumes. Therefore, we offer a better account for the tonal phonological difference between adjective and verb reduplications.Finally, we propose that unlike that in English, Chinese unaccusative vP (TRANSP as we define) is a phase. This proposition is supported by morpheme-order alternation related to syllable counts attested in Chinese synthetic compounds and the optional tone reduction in the second syllable of object-taking reduplicated verbs in Chinese. But this is an issue calling for more in-depth further research.