|其他题名||The Interplay between LemmaSelection and Phonological Encoding in Chinese Speech Production|
|关键词||语言产生 词汇选择 音韵编码 图画命名 电生理|
|其他摘要||Speech production involves the timely coordination of various information types such as semantic/syntactic, orthographic and phonological information. How these forms of information relate to each other remains a controversial issue: the relationship between lemma selection and phonological encoding is hotly debated in speech production theories. In the speech production system, lemma selection involves the retrieval of semantic/syntactic information from mental lexicon whereas phonological encoding involves the corresponding phonological processing. Previous studies concerning on the information transmission are largely based on Indo-European languages, which lack empirical evidence from nonalphabetic languages (e.g., Chinese). Chinese, as a logographic script, has distinct linguistic properties, it cannot be assumed the results from alphabetical scripts generalize across all languages. For instance, orthography and phonology are unavoidably confounded in alphabetic languages. However, in Chinese “pure” phonology could be manipulated in which orthographic overlap is avoided. In the present thesis, we investigated the interplay between lemma selection and phonological encoding by exploring (i) the direction of information transmission (one vs. two way); (ii) the continuum of information transmission (serial or cascaded).|
In Study 1, the timecourse of semantic and phonological information and the corresponding interaction were investigated using picture-word interference task both in electrophysiological and behavioural measurements. Previous studies have generally shown non-additive effects of semantic and form overlap between pictures and words, which may indicate underlying non-discrete processing stages in lexical retrieval. Unlike in existing studies in alphabetic languages where phonology is confounded with orthography, “pure” phonological distractors were used in which orthographic overlap was avoided. Results showed (i) clear semantic and phonological effects both in behavioral and electrophysiological measurements (250-450ms after picture onset for the semantic effect; 450-600ms for the phonological effect); (ii) no effect was found in naming latencies when distractor and target were semantically and phonologically related, suggesting that both effects cancelled each other out; however, ERP responses were modulated separately in both time windows; (iii) most importantly, a clear dissociation between semantic and phonological effects, with no temporal overlap and no statistical interaction. Overall, the results suggest that in Mandarin spoken production, information is transmitted from semantic to phonological levels in a discrete fashion.
In Study 2, the continuum of information transmission (whether or not there is activation of multiple phonological codes) was investigated using multidisciplinary approaches both in electrophysiological and behavioural measurements. The results of Experiment 4 and 5 showed that multiple phonological nodes were activated mandatorily when the semantic relationship of target and non-target words was enhanced in English whereas no such activation no matter the semantic relatedness between semantic competitors and targets had been boosted or not. Those results tend to indicate that the information flows in different fashion in English and Chinese: Cascaded in English while Discrete in Chinese. Word association task was conducted in Experiment 6 and the result replicated the null effect in phonologically related condition in Chinese. To further investigate the information transmission, Experiment 7-9 explored the phonological activation (in first language) of ignored picture in bilingual context. Results showed (i) clear semantic effects both in behavioral and electrophysiological measurements. Naming latencies were faster in the semantically related condition than in the unrelated condition, ERP responses were modulated in 250-600ms after picture onset for the semantic effect. (ii) no effect was found in naming latencies when context pictures and targets were phonologically related whereas wide scalp activity showed that phonologically related vs. unrelated condition was diverged in 250-600ms after picture onset.
In all, those results in Chinese reveal that there is no interaction between lemma selection and phonological encoding; under normal circumstance, no activation of multiple phonological nodes; semantic and phonological information could be activated in parallel in certain tasks. Based on the existing frameworks and those results, we attempt to postulate a model of Chinese speech production to help with a common theoretical framework across languages.
|朱雪冰. 汉语口语产生中词汇选择与音韵编码的关系[D]. 北京. 中国科学院研究生院,2014.|
|朱雪冰 - 博士学位论文.pdf（2497KB）||学位论文||开放获取||CC BY-NC-SA||请求全文|