汉字模糊度、字频和语义启动在N400上的交互作用：N400的IA模型的证据

通过两个实验, 采用事件相关电位技术, 记录大学生被试在同一字判断任务中的脑电波形。两个实验涉及三个因素, 包括模糊度、字频和语义, 语义和字频为被试内设计, 以启动字-靶字(模糊)-探测字的呈现序列为一个测试(trial), 要求被试判断靶字和探测字是否是同一个字。两个实验除了靶字的模糊度不同外, 其他材料完全相同。结果发现, 实验1字频和语义启动间存在交互作用, 其中低频条件下无语义启动比语义启动的N400波幅更偏负, 高频条件则无这种差异; 实验2中只存在语义启动效应, 字频与语义启动间不存在交互作用; 综合两个实验的ERP结果, 模糊度、字频和语义启动三者间存在交互作用, 且高频而非低频条件下模糊度与语义间也有交互作用, 无启动条件下字频和模糊度间也存在交互作用。本研究因此支持N400的IA模型(交互激活模型), 不支持核证模型和词汇后加工理论。

Interactions of Blurredness, Frequency, and Semantic Priming on N400 in Chinese Characters: Evidence for the IA Model of N400

Several theories, such as the post-lexical processing theory and prelexical processing theory, have been proposed to explain the mechanisms of N400. However, there were certain deficits in these theories and in the current paradigm of priming study. Therefore, a new approach and new theories to investigate N400 effects are required. Verification model and interactive activation models were employed in this study to address these problems. The objective was to ascertain which of these theories provides a much better explanation for the mechanisms of N400. The interaction study approach was adopted. This study adopted a delayed character-matching task to investigate how semantic priming, frequency, and blurredness interact on the N400 components of the Event-Related Potentials (ERPs), which were recorded while participants performed the task. Three Chinese characters were presented in succession in each trial, representing the prime, the target, and the probe stimuli. The prime and the probe were intact characters, whereas the target was blurred one. The materials and procedures in the two experiments were the same, except for the blurredness level of the targets: The targets were slightly blurred in Experiment 1 and were more blurred in Experiment 2. The priming and frequency were manipulated in each experiment with four treatment conditions, namely, primed high frequency, primed low frequency, unprimed high frequency, and unprimed low frequency conditions. Blureedness was manipulated across experiments. The subjects’ task was to press one keyboard if the target and the probe were the same character, and to press another otherwise. The results of Experiment 1 showed larger N400 amplitudes in unprimed condition than in primed condition for low-frequency target characters, but not for high-frequency target characters. By contrast, Experiment 2 showed N400 priming effects, but no N400 frequency effects. Combining the date from the two experiments, both a three-way interaction among the blurredness, frequency, and priming, and an interaction between blurredness and frequency in unprimed condition were found. An interaction between blurredness and priming under high frequency was observed, whereas interaction between blurredness and priming under low-frequency condition did not occur. These interactions are interpreted to indicate that frequency, blurredness, and priming are involved in the same stage of lexical processing. In conclusion, the results of the present study are partly expected from the lexical verification explanation constructed based on the verification model, completely expected from the interactive activation model, but completely unexpected from the postlexical theory.