The processing of morpheme position information has a very important role in the processing of the whole word. The recognition and processing of the whole word is closely related to the processing of morpheme identity information and position information. This article uses behavioral experiments and eye movement experiments to investigate whether morpheme position information processing is affected by vocabulary types, is whether lexical type effects occur during morpheme position information processing, and whether this effect is subject to word frequency adjustment. The experimental materials in this study selected three words, and two types of vocabulary were selected: the first three words, the second and third words of which formed a word, such as: nature, this type is called A+ BC type. The second three-word word, its one or two words form a word, such as: waiter, this type is called AB + C type. In Experiment 1, a start-up paradigm was adopted. The starting conditions were (for example, nature): the original word (nature), transposition non-word (substantially self), and non-negative non-word (grass trail), while the high-frequency low-frequency words were manipulated. Record the time when the target word is true or false. It was found that lexical type effects and word frequency effects were found to be significant in the processing of morpheme position information, and there were significant differences among several conditions under the high frequency conditions of A+BC type. Under low-frequency conditions, there is no significant difference in the starting effect between the original word condition and the transposition condition. For AB+C type, there are significant differences among several conditions under high and low frequency conditions. Experiment 2 uses the paradigm of paradigm, uses the three words in the experimental material to compose standardized sentences, divides the three-word region into regions of interest, sets a gaze boundary in front of the region of interest, and looks at the region before the boundary, the region of interest The start words for the three conditions are displayed: the original word (nature), the non-word of transposition (big self), and the non-negative non-word (grass trail). When the fixation point crosses the boundary, the area of interest becomes the original word. (nature). The index of eye movement in the area of interest under different conditions was recorded. It was found that both the high frequency and low frequency conditions of the A+BC type were the difference between the first and second conditions in terms of the first fixation time and fixation time index. Significant. The AB+C type has significant differences among several conditions under high-frequency conditions, but under low-frequency conditions, the difference between the original word and the transposition word is not significant, and the two are significantly different from the unrelated words. The experimental results show that the processing of Chinese three-character words is more flexible. When presented alone, word frequency and word segmentation have a significant effect on the processing of three-character position information. In sentence reading, word segmentation in the early case has a greater effect on the processing of position information than word frequency, and with the processing In-depth, after reading the entire sentence, the word frequency effect has a greater impact on the location information processing than the different segmentation of words. The results of this experiment support the interactive activation model and do not support the dual-path processing model.