The Acquisition Of Split Intransitivity By Chinese Learners Of English

L2 learners have been noted to apply passive morphology to some intransitive verbs like happen and die at times, and thus generate such ungrammatical sentences as"What was happened"and"She was died", which is called overpassivization. Interestingly, however, the passive forms may not be found in the target language input, nor can they be directly transferred from learners'L1. Then what are the underlying triggers behind this phenomenon? A preliminary reading of the relevant literature reveals that the majority of these passivized errors are confined to a particular subclass of English intransitive verbs, i.e. unaccusative verbs. According to the Unaccusative Hypothesis, the traditional intransitive verbs split into two classes: unaccusatives and unergatives, which have distinct syntactic properties. The sole argument of an unaccusative verb is an object at D-structure which may move to the subject position at S-structure, whereas the sole argument of an unergative verb is a subject at all levels of representation. Questions arise as to why the overpassivization mainly occurs to the unaccusative verbs? Are the unaccusative verbs represented differently from the unergative verbs in L2 learners'mental grammars? Is the overpassivization phenomenon attributable to the functions of UG?To this end, the present study investigates Chinese learners'knowledge of English split intransitivity within the theoretical framework of generative grammar. It explores how the knowledge of English split intransitivity is represented in L2 learners'mental grammars, and how the knowledge evolves over time. The study further scrutinizes the extent to which the UG mechanisms are involved in the L2 acquisition of argument structure. Previous research on the mental representation of the two kinds of intransitive verbs heavily relies on the data of overpassivization and is restricted to the binary distinction between the unaccusatives and the unergatives. The present study extends to investigating the acquisition of unaccusative diagnostics, and to detecting L2 learners' sensitivity to the semantic elements so as to further test the psychological reality of the Split Intransitivity Hierarchy that divides the unaccusatives and the unergatives into the core and the peripheral.The study is cross-sectional with a total of 170 subjects of three different learning stages: low, intermediate and advanced. Two types of tasks are applied: a written elicited production task (WEPT), and a grammatical judgment task which consists of two sets: GJTA and GJTB. WEPT and GJTB test learners'production and rejection of the passivized intransitives, and GJTA tests if learners are aware of the compatibility or incompatibility of the unaccusative diagnostics with the unaccusatives or the unergatives. The study has yielded the following results:First, the unaccusative-unergative distinction is correctly represented in Chinese EFL learners'mental grammars. The learners are found more liable to apply inappropriate passive morphology to the unaccusative verbs than to the unergative verbs, though the two types of verbs appear in the same surface sentence structures of the subject-verb order in the target language input. Furthermore, the learners realize that English resultative and there-insertion constructions are more compatible with the unaccusatives, whereas pseudo-passive constructions are more compatible with the unergatives.Second, the learners are sensitive to the semantic constraints that divide the intransitive verbs into the core and the peripheral unaccusatives and unergatives. The closer the verb type is to the unaccusative side, the more likely these verbs are incorrectly passivized. Conversely, the more unergative-like the verbs are, the more accurate the learners are with them. Unexpected findings are that learners, especially those of relatively low proficiency, perform worse with the peripheral unergatives than with the unaccusative verbs. These findings justify the necessity to make finer semantic classifications of the intransitive verbs, and lend further empirical support to Sorace's Split Intransitivity Hierarchy. Third, the learners'knowledge of split intransitivity generally develops with their English proficiency. Learners at all levels tend to discriminate between the unaccusatives and the unergatives, and the more advanced the learners are, the more obvious distinctions are made. The findings disconfirm the Unaccusative Trap Hypothesis, which claims that low-proficiency learners do not make an unaccusative-unergative distinction.The developmental patterns vary for different verb types in the acquisition of the subject-verb order. The proficiency effect is the most evident on the acquisition of the unaccusative verbs. The knowledge development of the unaccusative verbs takes on a rising linear slope. In the case of the unergative verbs, the advanced learners consistently perform better than the low and the intermediate learners, whereas the low and the intermediate learners do not manifest much difference. The development of the unergative verbs can be said to take the shape of a plow. The knowledge developmental patterns also differ between the core and the peripheral verbs. The core unaccusatives generally present a rising linear slope of development, i.e. the learners'performance correlates with their language proficiency. A plow-shape development route is found with both the peripheral unaccusatives and the peripheral unergatives, i.e. the learners'knowledge does not develop remarkably from the low to the intermediate stage. The knowledge development of the core unergative verbs takes the shape of a flat line, for learners at all levels of proficiency have a good command of the core unergatives, and consequently the knowledge does not develop markedly from the low to the advanced stage.Finally, the present study suggests that UG operates in the L2 acquisition of English split intransitivity. The Chinese learners of English have been found to display the knowledge of a linguistic universal -- the unaccusative-unergative distinction, and to observe the universal principle of the Uniformity of Theta Assignment Hypothesis. The target language input and the learners'L1 also have an effect on the acquisition. The study suggests that multiple factors might be responsible for the overpassivization phenomenon, and that distinct factors might function for learners of different developmental stages.The present study provides suggestive evidence for the generative model of grammar incorporating more than one level of linguistic representation. It has added additional evidence for the functions of UG in the L2 acquisition of argument structure. The findings have largely validated the psychological reality of the Unaccusative Hypothesis and the Split Intransitivity Hierarchy in SLA. The findings are also of significant implications for language teaching.