不同具体性词汇的识别：效价和唤醒的贡献

【作者】 姚昭；

【导师】 王振宏；

【作者基本信息】 陕西师范大学 ， 基础心理学， 2014， 博士

【摘要】 词汇的情绪维度(效价和唤醒)和具体性都是影响词汇识别的重要变量,也是情绪与语言关系研究领域中的热点话题。大量研究考察了词汇情绪维度或具体性水平对词汇识别情绪效应和情绪启动效应的影响。就情绪维度而言,效价和唤醒是如何发挥作用的目前还没有取得一致的结论。情绪的环形模型(Rusell,1980)认为情绪的相关效应来自效价或者唤醒的独立贡献,因为二者是分离的、独立的维度。情绪的动机模型(Bradley&Lang,2001)认为情绪的相关效应来自效价和唤醒维度的共同作用,因为二者是相关的,效价越高的刺激,唤醒水平也就越高；就具体性而言,在仔细匹配影响词汇识别的其它变量后,具体词的优势加工消失或者变为抽象词的优势加工。作为解释具体性效应的经典理论,双重编码理论(Paivio,1991)和语境有效性模型(Schwanenflugel,1991)都不能很好地解释这一现象,因此研究者从具身认知的角度出发,提出了语义表征的具身理论,该理论认为抽象词比具体词包含了更多的情绪体验信息,强调了词汇的具体性水平与情绪维度之间有不可忽视的关系。近期的研究发现词汇的效价和具体性能够交互地影响词汇识别的情绪效应和情绪启动效应。然而,作为情绪的另一个重要维度,唤醒与具体性之间有怎样的关系?效价和唤醒的关系目前还有争议(独立的vs.相关的),不同具体性词汇的情绪体验信息差异量是否会影响效价和唤醒的关系?如果对词汇的情绪维度(效价和唤醒)和具体性进行正交操作,词汇的效价、唤醒、具体性如何影响词汇识别的情绪效应和情绪启动效应?综合以往研究的结果,本研究假设情绪维度(效价和唤醒)和具体性在词汇识别过程中可能不是完全独立的因素,效价和唤醒对不同具体性词汇识别的情绪效应及情绪启动效应的贡献可能有所不同。本研究的目的是较为深入地揭示词汇情绪维度和具体性之间的内部关系,这不仅有利于扩展对情绪与语言关系的认识,还有助于解释目前不一致的结论,既具有非常迫切的现实意义也具有十分重要的理论意义。研究一通过对大量汉语双字词的语义学变量和词汇学变量进行评定或查询,通过回归分析初步推断效价、唤醒和具体性之间可能存在的关系。研究二和研究三为ERP冲实验,对不同具体性词汇的效价(正、负)和唤醒(高、低)进行正交操作,采用词汇判定任务探讨词汇的效价、唤醒和具体性如何影响词汇识别的情绪效应和情绪启动效应,以及这个影响发生的认知加工阶段。基于以上研究,本论文得到以下结论：(1)效价和唤醒的评定得分均能显著地预测词汇的具体性水平,表明词汇的情绪维度和具体性水平有直接的、内在的联系。(2)效价和唤醒的关系在不同具体性词汇中的表现方式不同,具体词的效价和唤醒的评定得分之间没有明显的趋势,表明二者是独立的、分离的维度,与情绪的环形模型一致；抽象词的效价和唤醒的评定得分成U型分布,表明二者是相关的、不可分离的维度,与情绪的动机模型一致。(3)具体词的效价和唤醒独立地影响词汇识别的情绪效应,在具体词的高、低唤醒水平上,正性具体词均比负性具体词的反应时更短,诱发了减小的N400和增强的LPC。抽象词的效价和唤醒交互地影响词汇识别的情绪效应,体现在反应时和LPC成分上。正性的低唤醒抽象词比高唤醒抽象词的反应时更短,诱发的LPC更小,而负性的高唤醒抽象词比低唤醒抽象词的反应时更短,但在LPC成分上没有差异。(4)不同情绪维度的具体启动词没有产生明显的情绪启动效应,启-靶具体词对在情绪一致条件和不一致条件下的反应时、N400和LPC均没有显著的差异。然而,抽象词的效价和唤醒交互地影响词汇识别的情绪启动效应,正性的高、低唤醒抽象词产生了明显的启动效应,情绪一致条件条比不一致条件的反应时更短,LPC波幅更小；负性低唤醒抽象词产生了零启动效应,情绪一致和不一致条件下的反应时和LPC均没有显著的差异,而负性高唤醒抽象词产生了反转的启动效应,一致条件的反应时长于不一致条件,且诱发更大的LPC。总的来说,本研究发现效价和唤醒可以预测词汇具体性的评定,词汇的具体性水平影响效价和唤醒的关系；具体词的情绪效应主要来自效价的贡献,抽象词的情绪效应来自效价和唤醒的共同贡献；除效价、唤醒之外,启动刺激的具体性水平也会影响情绪启动效应。更多还原

【Abstract】 Emotional dimensions (valence and arousal) and concreteness are the important variables in the processing of words recognition. Many studies from behavior, electrophysiology and brain imaging have explored the influence of emotional dimensions or word’s concreteness on emotion effect and affective priming effect. With regard to emotional dimensions, there is no consistent conclusion about how valence and arousal play roles in the processing of emotional words. The affective circumplex model of Russell (1980) depicts valence and arousal as orthogonal, bipolar, and independent dimensions. In contrast, the motivated emotion model by Bradley and Lang (2001) predicts that emotion effects can only arise from a combination of valence and arousal (as stimuli of high emotional valence by definition also have high arousal values). With regard to word’s concreteness, once a large number of lexical and semantic variables are controlled, the advantaged processing of concrete words will disappear or change to the advantage of abstract words. This phenomenon was not explained by the Dual-Coding Theory (Paivio,1991) or the Context-Availability Model, though they are the classic theory to explain concreteness effect. Thus, Vigliocco et al.(2009) came up with the embodied theory of semantic representation, which proposed abstract words may have more emotional experiential information than do the concrete words, and emphasized a considerable relationship between emotion and concreteness.Recently studies have reported that valence and concreteness interactively impacted on emotion effect and affective priming effect. However, as another important dimension, what is the relationship between arousal and concreteness? Due to the difference in the amount of emotional experiential information, whether concreteness influences on the relationship between valence and arousal? If we orthogonally manipulated the emotional dimensions (positive and negative valence, high-and low-arousal) and word’s concreteness (concrete, abstract) in the processing of word recognition, how emotional dimensions and word’s concreteness influence emotion effect and affective priming effect. Colligating the results of previous studies, the present study hypothesized that emotional dimensions and word’s concreteness were not completely independent variables and interactively influenced on emotion effect and affective priming effect. The aim of the present study is to explore the internal relations between emotional dimensions and concreteness, which makes for extending to knowledge of the relationship between emotion and language, and also for clarifying the inconsistent conclusions.The first study collected a large number of Chinese two-character words, and then rated or inquired words’ lexical and semantic features inferring the possible relationship between emotional dimensions and word’s concreteness via regression analysis. Using event-related potential (ERP) measures, the aim of the second and third studies is to orthogonally manipulate valence and arousal of concrete words or of abstract words during a lexical decision task, to investigate whether and how the relationship between valence and arousal are influenced by concreteness, to identify which ERP components are modulated by these variables and whether these variables have independent or interactive effects on brain activity.According to the studies above, the main findings are as follows:(1) Emotional valence ratings and arousal ratings significantly predicted concreteness ratings. More valenced words tend to be more abstract, whereas neutral words tend to be more concrete, indicating that there has a direct and internal relationship between emotionality and concreteness.(2) The internal structure of "emotion"(the relationship of valence and arousal) was influenced by words concreteness. For concrete words, there is no clear trend between valence ratings and arousal ratings, but for abstract words, emotional valence and arousal ratings are strongly related via a U-shaped distribution function within affective space.(3) Valence and arousal of concrete words independently influenced on the emotion effect. For high-and low-arousal concrete words, positive words were recognized faster than negative words, and elicited the decreased N400and the enhanced LPC. Valence and arousal of abstract words interactively influenced on the emotion effect. Positive low-arousal abstract words had the shorter reaction times (RTs) and the decreased LPC than positive high-arousal ones. RTs were shorter for negative high-arousal words compared negative low-arousal words, but no difference in LPC.(4) Concrete prime words did not produce the obvious affective priming effect, independent of its valence and arousal. However, valence and arousal of abstract words interactively impacted on the affective priming effect. The processing of positive high-and low abstract targets was facilitated by a previous exposure to a congruent prime, as reflected by the short RTs and the reduced LPC; No RTs or LPC differences were found between negative low-arousal abstract congruent pairs and incongruent pairs; And negative high-arousal abstract primes produced reversed priming effect, as compared to incongruent-abstract trials, the processing of congruent-abstract trials was associated with longer RTs and enhanced amplitude of a LPC.To sum up, these findings of the present study suggested that valence ratings and arousal ratings significantly predicted concreteness ratings, and the relationship of valence and arousal was influenced by word’s concreteness; Concrete words’ valence dimension independently contributed to emotion effect, while emotion effect of abstract words came from the conjunct contribution of valence and arousal; Primed concreteness had a differential contribution of valence and arousal to affective priming effect.更多还原