口语理解中词汇语音的预激活及其认知机制
文献类型：学位论文
| 作者 | 赵梓彤 |
| 答辩日期 | 2022-06 |
| 文献子类 | 硕士 |
| 授予单位 | 中国科学院大学 |
| 授予地点 | 中国科学院心理研究所 |
| 其他责任者 | 李晓庆 |
| 关键词 | 口语理解 语音预激活 策略性加工 视觉情境范式 眼动 |
| 学位名称 | 理学硕士 |
| 学位专业 | 基础心理学 |
| 其他题名 | The Anticipatory Processing of Lexical Phonology in Speech Comprehension and its Underlying Cognitive Mechanisms |
| 中文摘要 | During the process of speech comprehension, listeners can make prediction about the upcoming content based on personal prior knowledge and available contextual information, and these pre-activated representations are potentially able to facilitate the perception and comprehension of incoming speech input, hence being beneficial to inter-person speech communication. The presence of predictive processing in speech comprehension has been consistently admitted by researchers. The unsettled questions, however, are that whether listeners pre-activate the phonology representation/features of incoming words during online speech comprehension, and what is the cognitive mechanism underlying this phonological prediction. The current research aimed to investigate the possibility of phonological prediction during spoken sentence comprehension by using the visual world paradigm, and to further explore the cognitive mechanism that supports this phonological prediction. Experiment 1a and 1b explored whether the phonological representation of an incoming word can be pre-activated during online speech comprehension. Participants listened to spoken sentences for comprehension, while reading two single-character words presented visually on screen (with one word being Critical Word and another one being Distractor). Compared to the target word carried in the speech signal, the Critical Word presented visually was exactly the same word (target word), its homophone (homophone word), only with the same lexical tone (tonal word), or unrelated. Participants in Experiment 1a were required to judge whether one of the two words on the screen has been presented in the spoken sentence (target word judgement), and those in Experiment 1b were asked to judge whether one of the two visual words has the same pronunciation as one of the words heard in sentence (pronunciation judgement). The results of 1a showed that, prior to the acoustic onset of the target words, compared to the unrelated words, 1) participants showed fixation preference to the target words, indicating the presence of lexical prediction; 2) participants showed fixation preference to the homophone and tonal words within only one time window of 100 ms. The fixation preference to homophone and tonal words (compared to the unrelated words) only in one time window might be caused by random statistical error, hence being not able to provide solid evidence for the presence of phonological prediction. Another group of participants was recruited into Experiment 1b, being required to perform “pronunciation judgement” task and to make response as quickly as possible. The results of 1b showed that: prior to the acoustic onset of the spoken target words, 1) participants showed fixation preference to target words, 2) participants made anticipatory fixation to homophones in a continuous three time windows of 100 ms (namely, within 300 ms), which provided evidence for the pre-activation of phonological representation prior to the actual presence of the spoken target word. Experiment 2 further investigated the mechanism that supports phonological prediction during online speech comprehension. The same group of participants were asked to perform two different tasks (target-word judgement or pronunciation judgement), with stimuli being counterbalanced across the two tasks. The experimental results show that, prior to the acoustic onset of the target words, participants demonstrated fixation preference to homophones during the “pronunciation judgement” task, but not during the “target-word judgement” task, indicating the flexibility of phonological prediction. In sum, the present results provide experimental evidence for the existence, at least under some circumstances, of phonological prediction during online speech comprehension, and also support a flexible and strategic processing mechanism of phonological prediction. These findings can function as a guide for the further investigation of the neural basis of phonological prediction flexibility and the further development of corresponding intervention method. |
| 英文摘要 | 在口语理解的过程中，听者会通过信息输入和个人的先验知识对即将出现的内容做出假设，并提前激活词汇的表征信息，从而实现快速理解并与说话者进行后续交流。例如，基于语义和词汇知识的自上而下语音预期，是帮助听者高效地感知理解口语信号的一个重要加工机制，这在噪声等特殊场景中尤为重要。预期在言语理解中的地位已经被大家普遍认同，但有些方面依旧存在争议。其中，目前研究者争论的一个热点问题是，听者能否在句子理解的过程中预先激活即将出现的词汇的语音信息，以及词汇的语音为什么会被激活？具体而言，口语理解中的语音预激活（如果存在的话）是自动产生的、还是被策略性、灵活性调控的？ 本研究采用眼动技术，利用视觉情境范式的变式探讨了口语理解过程中词汇语音的预激活及支持其预激活的认知机制。实验 1a、1b 探讨口语理解中是否存在语音信息的预激活。被试在听句子的同时，在电脑屏幕上会看到两个字（一个是“关键字”，另一个是“干扰字”）；屏幕上的“关键字”与句中的目标字完全一样（目标字）、只有声调相同（声调相同字）、只有读音相同（同音字）、或者音形义都无关（无关字）；“干扰字”与句中目标字的音形义都无关。实验 1a 的任务是判断屏幕上的“字”在刚刚听到的句子中是否出现过（目标判断任务）；实验1b 的任务是判断屏幕上的“字的读音”在刚刚听到的句子中是否出现过（读音判断任务）。实验 1a 的结果发现：在听觉目标字出现前，1）相比于无关字，被试对目标字产生注视偏好，表明对目标字产生了预期；2）被试有 100 ms 时间窗对同音字和声调相同字产生注视偏好，但由于此效应只出现在单个统计分析时间窗、不能排除是随机效应，因此实验 1a 的结果不能为语音预期提供确定的实验支持。采取与实验 1a 完全相同的实验材料，我们招募另一批被试参与了实验 1b，实验任务由目标判断变为读音判断、同时对判断速度提出要求。实验 1b 的结果发现：1）重复了以往研究和实验 1a 中的词汇预期效应； 2）在听觉目标出现前，被试在连续 300 ms 的时间窗内（连续三个统计分析时间窗）对同音字产生预期性注视，证明词汇的预期可以到达语音表征的水平。实验 2 的目的是进一步采取更严格的实验设计，探讨口语理解过程中语音预激活的认知机制，即语音表征的预激活是否具有灵活性。实验中同一批被试执行两种不同的任务，包括目标判断任务和读音判断任务。实验 2 结果发现，词汇语音的预激活是灵活、策略性的，可以根据不同的任务需求进行调节；在听觉目标出现前，被试只有在执行读音判断任务时，才会对同音字产生注视偏好。结果支持语音预激活的策略性加工机制。 总之，本研究结果表明，在口语理解过程中，确实在有些任务场景中会出现语音表征的预激活；但是，语音表征的预激活并不是完全自动的、而是具有策略性和灵活性。这一研究结果不仅有助于我们更深刻地理解口语理解的认知机理，而且为我们将来进一步探索语音预期灵活性的神经基础和干预提升技术奠定了基础。 |
| 语种 | 中文 |
| 源URL | [http://ir.psych.ac.cn/handle/311026/43146]  |
| 专题 | 心理研究所_认知与发展心理学研究室 |
| 推荐引用方式 GB/T 7714 | 赵梓彤. 口语理解中词汇语音的预激活及其认知机制[D]. 中国科学院心理研究所. 中国科学院大学. 2022. |
入库方式： OAI收割
来源：心理研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。