【摘要】：抽象是生物體認識世界和感知世界所必須的一項高級的認知功能,即從具體的,復雜的事物中總結(jié)內(nèi)在的規(guī)律,從而抽取出一個普遍存在的共性。對于動物來說,這一能力是它們生存的需要,從而可以從復雜多變的環(huán)境中抽取規(guī)律以保證生存。同時,這一能力對于人類日常的語言交流是非常關(guān)鍵的,因為在日常生活中,我們聽到的語音信號具有很大的可變性,如聲音的強度,頻率和時間特性,并且還經(jīng)常夾雜著環(huán)境噪音。為了保證語音的交流,我們的聽覺皮層必須對這種語言信號的可變性作出應變,即聽覺皮層不僅需要處理聲音信號本身,同時還要處理各種聲音信號之間的關(guān)系,在復雜變化的語音信息流中獲取并且形成有效的抽象的客體概念,比如說話人的識別或者話語的內(nèi)容。這種語音信號之間的關(guān)系即語言中的聽覺規(guī)則是不變的和抽象的,盡管語音信號本身具有很大的可變性。當前對于這一課題的研究已經(jīng)取得了很大的進展,如使用fMRI的方法研究語言中聽覺信號的規(guī)律的提取反映了在注意階段大腦聽覺皮層具有對于語音信號進行抽象的能力。在注意前階段,對于聲音信號中不變的聽覺規(guī)律的提取的研究主要集中在非語言(non-speech)的背景下,即使用一些純音(pure tone)作為聲音刺激材料,在不同的純音對(pure tone pair)頻率不斷改變的情況下,其隱含的頻率上升或者下降的規(guī)律可以在注意前階段就被提取出來。盡管這提示我們?nèi)祟惖倪@一在注意前階段處理聲音頻率之間的規(guī)律的能力對于語言的感知和音樂的欣賞是非常關(guān)鍵的,而如果直接在復雜變化的語言信號(speech)中抽取聲音頻率的規(guī)律,同時這一頻率的規(guī)律又直接影響語義(semantics)的理解,這樣的研究較之用非語言材料更能直接的反映對于人類語言感知所非常關(guān)鍵的變化聲音信號中規(guī)律的提取能力。本論文采用漢語聲調(diào)作為聲音材料,通過人工合成上百個漢語元音(a, e, i, u)并且將這些元音賦予聲調(diào)(一聲調(diào),二聲調(diào)和四聲調(diào)),在這些聲音中,每一個元音都具有3個不用的聲音強度,并且都具有3種不同的聲調(diào),而每一種聲調(diào)的基頻又是不斷變化的,即合成的一聲調(diào)具有10個不同的基頻梯度,合成的二聲調(diào)和四聲調(diào)具有3個不同的基頻梯度,通過將這些人工合成的上百個漢語元音隨即播放而形成一個復雜的語音聽覺流,而在這一聽覺流中嵌入一個聲音基頻特征的抽象的規(guī)律,即盡管聲音的共振峰,聲音強度和基頻本身都在不斷變化,但其中大部分聲音都可以感知為一聲調(diào)(flat tone),在偶爾的情況下,這一規(guī)律會被二聲調(diào)(rising tone)或者四聲調(diào)(falling tone)的隨機出現(xiàn)所打破。本研究的實驗結(jié)果表明,在復雜語音聽覺信息流中隱含的抽象的聲調(diào)信息的規(guī)律可以被母語受試者在注意前階段抽取出來,即誘發(fā)穩(wěn)定的失匹配負波成分(mismatch negativity),而在實驗記錄中受試者并不會發(fā)現(xiàn)也不會察覺這一規(guī)律。本研究反映了語音感知過程中存在的一種感覺智能(sensory intelligence),而這一感覺智能的存在提示了人類在注意前階段就可以在復雜和不斷變化的聽覺環(huán)境中提取抽象的和有效的聽覺規(guī)律,并且這一感覺智能的存在可以大大節(jié)省人們有限的意識資源。
[Abstract]:Abstract: Abstract: It is a high-level cognitive function that organisms recognize the world and perceive the world, that is to summarize the inherent laws from concrete and complex things, thus drawing out a common general character. For animals, this ability is the need to survive so that laws can be drawn from a complex and complex environment to ensure survival. At the same time, this capability is critical to human daily verbal communication, because in everyday life, we hear voice signals that have great variability, such as the intensity, frequency and time characteristics of sound, and are often mixed with environmental noise. in order to guarantee the exchange of speech, our auditory cortex must strain the variability of the speech signal, i. e. the auditory cortex requires not only the sound signal itself but also the relationship between the various sound signals, An effective abstract object concept, such as a speaker's recognition or utterance, is acquired and formed in complex varying voice information streams. The relationship between such speech signals is that the auditory rules in the language are invariant and abstract, although the speech signal itself has great variability. The current research on this subject has made great progress, such as the extraction of auditory signals in speech using the method of fmri reflects the ability of the auditory cortex of the brain to abstract the speech signal during the attention phase. In the pre-attention stage, the study of the extraction of the same auditory law in the sound signal is mainly focused on non-speech, i.e. using a certain tone as the sound stimulating material, in the case where the frequency of a different acoustic wave is constantly changing, the regularity of its implicit frequency rise or fall can be extracted at the pre-notice stage. although this suggests that the ability of our humans to process the regular between sound frequencies at the pre-stage stage is critical to the perception of language and appreciation of the music, and if the regularity of the sound frequency is extracted directly in a complex varying speech signal, At the same time, the regularity of this frequency directly affects the understanding of semantics, which is more direct than non-linguistic materials to reflect the regular extraction capability of the very critical changes in human language perception. In this paper, the Chinese tone is used as a sound material, and a hundred Chinese vowels (a, e, i, u) are artificially synthesized and the vowels are endowed with tones (a sound tone, two tones and four tones), in which each vowel has three unused sound intensities, and all have three different tones, and the fundamental frequency of each tone is constantly changing, that is, the synthesized one tone has 10 different fundamental frequency gradients, the synthesized two tones and four tones have three different fundamental frequency gradients, by playing the artificially synthesized hundreds of Chinese vowels to form a complex voice stream, an abstract pattern of sound fundamental frequency characteristics is embedded in this auditory stream, i.e., although the resonance peaks, the sound intensity and the fundamental frequency of the sound are constantly changing, But most of the sounds can be perceived as flat tone and, occasionally, this rule is broken by the random appearance of two tones or four tones. The experimental results of this study show that the law of abstract tone information implied in complex voice auditory information flow can be extracted by native subjects at the pre-attention stage, that is, to induce stable distortion-matched negative wave components. The subject was not found to be unaware of this rule in the experimental record. this study reflects a sensory intelligence that is present during speech perception, and the presence of this sensory intelligence suggests that humans can extract abstract and effective auditory rules in complex and constantly changing auditory environments prior to the attention of humans, and the existence of this sensory intelligence can greatly save people's limited awareness resources.
【學位授予單位】：中國科學技術(shù)大學
【學位級別】：博士
【學位授予年份】：2012
【分類號】：R767.92

【參考文獻】

相關(guān)期刊論文 前2條

1 羅躍嘉,魏景漢;Mismatch negativity of ERP in cross-modal attention[J];Science in China(Series C:Life Sciences);1997年06期

2 胡凈,胡潔;基于正則化方法的加權(quán)最小模估計在腦磁源成像中的應用[J];生物醫(yī)學工程學雜志;2003年01期

，

本文編號：2254326