ABSTRACT

The relationship between the distribution of nominalizers in Lhasa Tibetan and the argument/adjunct property of relevant syntactic elements is approached from a generative perspective. The distribution of nominalizers in Lhasa Tibetan demonstrates a regular pattern. Some nominalizers are bi-functional in that they can mark both participant and event nominalizations while others are uni-functional in that they can only mark participant nominalizations. It is found that the difference between the two types of nominalizers correlates to whether the nominalizer (NML) is argument-associated or adjunct-associated. An account of the correlation is developed in the theoretical framework of generative grammar. It is argued that the syntactic derivation of an NML-phrase gives rise to a binding relationship between the nominalizer and the suppressed element in the source constituent Aspect Phrase (AspP) or the AspP itself, leading to a condition on its semantic interpretation. The condition is satisfied in a participant NML-phrase headed by a nominalizer of either type and in an event NML-phrase headed by a bi-functional nominalizer. It is not in an event NML-phrase headed by a uni-functional nominalizer for the reason that in the calculation of event semantics arguments align with events while adjuncts align with predicates. Specifically, a bi-functional nominalizer, being argument-associated, semantically matches both a suppressed argument in a participant NML-phrase and the source constituent AspP, whereas a uni-functional one, being adjunct-associated, semantically matches a suppressed adjunct in a participant NML-phrase but not the source constituent AspP. Consequently, no event NML-phrase headed by an adjunct-associated nominalizer is found in this language. The findings of this study have implications for both analyzing the distribution of nominalizers in other Tibeto-Burman languages and the syntactic and semantic mechanisms that constrain them, and for classifying the argument/adjunct asymmetry, which is fundamental in most current linguistic frameworks as well as research on human sentence processing.

摘要

从生成的角度探讨了拉萨藏语中名词化词的分布与相关句法元素的论元/附属属性之间的关系。拉萨藏语名词化词的分布呈现出规律。一些名词化器是双功能的，因为它们可以标记参与者和事件的名词化，而另一些是单功能的，因为它们只能标记参与者的名词化。发现两种类型的名词化器之间的差异与名词化器（NML）是与参数相关还是与附属物相关。在生成语法的理论框架中发展了相关性的说明。有人认为，NML 短语的句法推导导致了名词化器与源成分方面短语 (AspP) 或 AspP 本身中的被抑制元素之间的绑定关系，从而导致了对其语义解释的条件。该条件在以任一类型的名词化器为首的参与者 NML 短语中以及在由双功能名词化器为首的事件 NML 短语中得到满足。它不在以单功能名词化器为首的事件 NML 短语中，因为在计算事件语义时，参数与事件对齐，而附属词与谓词对齐。具体来说，与参数相关的双功能名词化器在语义上匹配参与者 NML 短语中的抑制参数和源成分 AspP，而单功能名词化器，与辅助相关，语义上匹配参与者 NML 短语中的抑制辅助，但不匹配源成分 AspP。因此，在这种语言中找不到以附加关联名词化器为首的事件 NML 短语。这项研究的结果对分析其他藏缅语中名词化词的分布以及限制它们的句法和语义机制以及对论元/辅助不对称性进行分类（这是大多数当前语言框架的基础）以及人类句子处理研究。