Doctoral Dissertation Research: Telicity and Event Structure in Language Comprehension

博士论文研究：语言理解中的Telicity和事件结构摘要在美国国家科学基金会的支持下，Erin O 'Bryan女士将在Tom贝弗教授的指导下进行语言学博士论文的研究。 词汇语义学领域的最新研究表明，语义特征的传递性（事件是否具有特定的时间端点）与论元结构（动词要求或允许的论元）有关（货车Hout 2000，Jackendoff 1996，Sanz 1996）。最近在句子处理领域的研究调查了理解花园小径句子（结构上被误解并需要重新分析的句子）的难度差异，例如The horse raced past the barn fell。 贝弗和桑兹预测，含有集类动词（没有特定时间端点的动词，如race）的简化关系从句花园小径句应该比含有目的动词（有特定端点的动词，如arrive或melt）的简化关系从句花园小径句更难。这一预测是基于这样一个事实，即本质上目的动词需要一个动词短语-内部宾语，因此将有助于正确分析缩减关系句。吉布森（Gibson，1991）和普里切特（Pritchett，1992）曾预言，带有可选择不及物动词（无直接宾语）的及物动词的缩减关系分句应该比带有强制及物动词的缩减关系分句更难。 这一预测来自于及物动词需要一个显性直接宾语的事实，因此也将有助于正确分析缩减关系句。这项研究的主旨是确定这些解释中哪一个是正确的，或者两者都是正确的。 贝弗和桑兹（准备中）、吉布森（1991）和普里切特（1992）的预测是基于语言学和心理语言学理论、语言学的可接受性判断和计算模型。拟议的研究将测试这些预测与在线句子理解实验使用的扬声器变化onitoring任务（汤森贝弗1991）。这项任务已被证明反映了本地处理的复杂性，并具有可用于实验组的重要优点，因此可以包括大量的参与者。 在拟议的实验中，参与者将听取类似于（a）到（d）中的缩减关系从句句子。实验将测试是否telicity和次范畴化的影响程度的花园路径，以及这些因素是否有一个比其他更直接的效果。第一个假设是，与集类动词（如（a）和（B））有关的简化关系比与目的动词（如（c）和（d））有关的简化关系会导致更大的加工难度。 第二个假设是，与可能不及物动词（a）和（c）的约简关系相比，与强制及物动词（B）和（d）的约简关系会导致更大的加工困难。 a.消防队员在警长的带领下向人群挥手致意。B.消防队员在警长的陪同下向人群挥手致意。C.消防队员受到元帅的敬礼，向人群挥手致意。D.元帅派来的消防员向人群挥手致意。 处理难度将被测量为遗漏检测副短语中的说话者变化音节，例如marshal，以及主句动词，例如waved。说话者改变是由不同的说话者说出的音节，而不是要求参与者监控的句子的其余部分。 这两个说话人的变化位置允许调查的问题，是否有一个因素对理解有更直接的影响。预期的是，无论是Telicity或次范畴化将有一个相对较大的影响，早期的探测位置。如果Telicity有一个更直接的效果，它表明，一个动词的词汇语义，句法学家认为是结构的，虽然它也携带意义，有一个更早的，也许是特权的作用，在理解。如果次范畴化具有更直接的效果，这表明动词的分布特性具有更早的，也许是更优先的作用。 该研究将阐明语义变量在理解中的作用，而这些变量在之前的研究中尚未得到控制。这也将增加我们对语义和分布动词属性的在线使用的理解。此外，本研究还将揭示主位角色的真实性质、主位角色与次范畴化框架的关系以及主位角色与事件结构的关系。

Doctoral dissertation research: Telicity and event structure in language comprehensionABSTRACTWith NSF support, Ms. Erin O'Bryan will conduct research under the supervision of Professor Tom Bever for a doctoral dissertation in linguistics. Recent research in the area of lexical semantics has shown that the semantic feature telicity (whether or not an event has a specified endpoint in time) is related to argument structure (the arguments that a verb requires or allows) (van Hout 2000, Jackendoff 1996, Sanz 1996). Recent research in the area of sentence processing has investigated differences in the degree of difficulty in understanding garden path sentences (sentences that are structurally misinterpreted and require reanalysis), such as The horse raced past the barn fell. Bever and Sanz have predicted that reduced relative clause garden path sentences with atelic verbs (verbs without a specified endpoint in time, such as race) should be more difficult than those with telic verbs (verb with a specified endpoint, such as arrive or melt). This prediction follows from the fact that intrinsically telic verbs require a verb phrase-internal object, and hence will facilitate the correct analysis of reduced relative sentences. Gibson (1991) and Pritchett (1992) have predicted that reduced relative clause sentences with transitive verbs that can optionally occur intransitively (without a direct object) should be more difficult than those with obligatorily transitive verbs. This prediction follows from the fact that transitive verbs require an overt direct object, and hence will also facilitate the correct analysis of reduced relative sentences. The thrust of the proposed research is to determine which of these explanations is correct, or if they both are. The predictions made by Bever and Sanz (in preparation), Gibson (1991), and Pritchett (1992) are based on linguistic and psycholinguistic theory, linguistic judgements of acceptability, and computational models. The proposed research will test these predictions with an online sentence comprehension experiment using the speaker change onitoring task (Townsend & Bever 1991). This task has been shown to reflect local processing complexity and has the important virtue of being usable with experimental groups, so that large numbers of participants can be included. In the proposed experiment, participants will listen to reduced relative clause sentences like those in (a) through (d). The experiment will test whether telicity and subcategorization affect the extent of garden pathing and whether either of these factors has a more immediate effect than the other. The first hypothesis is that reduced relatives with atelic verbs, such as (a) and (b), will result in more processing difficulty than reduced relatives with telic verbs, such as (c) and (d). The second hypothesis is that reduced relatives with potentially intransitive verbs, such as (a) and (c), will result in more processing difficulty than those with obligatorily transitive verbs, such as (b) and (d). a. The firefighter led by the marshal waved to the crowd. b. The firefighter escorted by the marshal waved to the crowd. c. The firefighter saluted by the marshal waved to the crowd. d. The firefighter sent by the marshal waved to the crowd. Processing difficulty will be measured as missed detections of speaker-change syllables in the by-phrase, e.g. on marshal, and on the main clause verb, e.g. waved. The speaker-change is a syllable spoken by a different speaker than the rest of the sentence that participants are asked to monitor for. The two speaker-change locations allow the investigation of the question of whether one of the factors has a more immediate effect on comprehension. The expectation is that either telicity or subcategorization will have a relatively larger effect on the early probe position. If telicity has a more immediate effect, it suggests that a verb's lexical semantics, which syntacticians consider to be structural though it also carries meaning, has an earlier and perhaps privileged role in comprehension. If subcategorization has a more immediate effect, it suggests that distributional properties of the verb have an earlier and perhaps more privileged role. The research will clarify the role of semantic variables in comprehension that have not been controlled for in prior research. It will also increase our understanding of the online use of semantic and distributional verb properties. Additionally, the research will reveal information about the true nature of thematic roles, their relationship with subcategorization frames, and their relationship with event structure.