The effects of verb bias, context and tasks on mandarin Chinese reflexives

Mandarin Reflexive;Reflexive Resolution;Verb Biases, Sentence Processing;Event-Related Potentials;Context;Anaphor;Anterior Negativity;Nref;Task Demands;Antecedent selection

Lu, Hsin-Yi

关键词: Mandarin Reflexive;    Reflexive Resolution;    Verb Biases, Sentence Processing;    Event-Related Potentials;    Context;    Anaphor;    Anterior Negativity;    Nref;    Task Demands;    Antecedent selection;   

Others  :  https://www.ideals.illinois.edu/bitstream/handle/2142/24513/LU_HSINYI.pdf?sequence=1&isAllowed=y

美国|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

The interpretation of Mandarin reflexives can be ambiguous, especially if there are multiple possible antecedents mentioned in the context prior to the reflexives. Misunderstanding between speakers and listeners can happen if listeners misinterpret the referent of the reflexives that speakers intended. For example, in English sentence “John knew that Peter likes himself”, the antecedent for ‘himself’ must be Peter. In the Mandarin equivalent, the antecedent of the reflexive could be either Peter or John. Thus, there is more ambiguity about the antecedents of reflexives in Mandarin than in English. Theoretical frameworks have been devoted to explain how multiple antecedents is allowed in Mandarin; however, it is not clear how people interpret Mandarin reflexives from a processing perspective. This study investigated the issue from such perspective, exploring people’s preferences for interpreting Mandarin reflexives. The study focused on how factors like verb subcategorization (i.e. verb biases, properties that a verb tends to take a direct object more than a sentential complement or vice versa) or context information influence such interpretation by using off-line and on-line reading times and event-related potential measurements. Time courses of when such decision was made were also investigated.Two self-paced reading experiments used sentences containing two possible antecedents for Mandarin reflexives ziji and taziji to examine whether verb bias properties influence the antecedent selection. It is found that whether there was any antecedent preference depended on the structural bias of the verb preceding the reflexive. For bare reflexive ziji, the distant antecedent was preferred after a verb that usually takes direct objects, but after a verb that usually takes embedded clauses, no preference was observed. For complex reflexive taziji which is thought to co-refer more to the local antecedent, it is found that local antecedents were indeed preferred, but only for verbs that take embedded clauses. After verbs that usually take direct objects, the distant antecedent was still preferred. These results were interpreted to mean that readers expected sentences to end soon after DO-bias verbs so they chose the most prominent available antecedent, which was the distant main clause subject. In contrast, after Clause-bias verbs they expected a longer sentence so they waited for further disambiguating information, causing prominence to have less impact. Consistent with this interpretation, in another self-paced reading study with a context sentence preceding the same target sentences, there was again an antecedent preference only after Clause-bias verbs, but the preference was for whichever of the possible antecedents was made more prominent by mentioning it in the context sentence.Two off-line completion tasks and forced-choice tasks were used to investigate people’s antecedent preferences for both Mandarin bare and complex reflexive interpretations given no time constraint. The results of the completion task for bare reflexive were consistent with native speakers’ intuition as described in the literature, whereas the forced-choice task results were similar to what was found in the reading time experiments. Both off-line results for complex reflexive showed that there was a strong tendency for complex reflexive to co-refer to local antecedents; however, this tendency was only observed in sentences with SC-bias verbs. After DO-bias verbs, people were affected greatly by the presence of long distance antecedents, suggesting that long distance nouns attracted people’s selections even for complex reflexive taziji, which is usually thought to be locally bound in the literatures. The findings of on-line and off-line studies were taken together to show that verb biases and context have a great influence on the antecedent selection and the time courses of the reflexive resolution. Readers might choose to resolve or not to resolve the reflexive ambiguity on-line, depending on the on-line and off-line task demands.ERPs evoked by the disambiguating verb showed more negativity in a 300-500ms window over frontal scalp regions when sentences were disambiguated toward local antecedents than when they were disambiguated toward distant antecedents. This was taken to suggest that people might in general have more difficulty integrating disambiguating verbs consistent with local antecedents during reflexive resolution. If people take distant antecedents as the default interpretation of the reflexive, when the meaning of the disambiguating verb suggests otherwise, they have to switch to the local antecedent, resulting in more frontal negativity from integration difficulty and increased ambiguity about the antecedent of the reflexive.