Abstract: Long et al. (2006) found that the construction of situation model was conscious while the construction of textbase was unconscious through two paradigms (IRK and PDP)in text reading. With a careful examination to the experiment materials, it is found that the materials at two levels of coherence (low and high) are all globally coherent. Previous research on text-reading established a wide acknowledgement that the local coherence condition connects with the construction of textbase while the global coherence condition connects with the construction of situation model. However, the evidence given by Long is insufficient to prove the fact that the construction of textbase is unconscious because the conclusion is purely based on materials that are globally coherent. In this study, two experiments were designed to explore the influence of different coherence levels on the consciousness of text representation with the two paradigms (IRK and PDP).
A self-paced, line-by-line reading paradigm was used in this study. In experiment 1, 30 university students were asked to read 16 narrative passages. One independent variable is the relationship between the elaboration on a characteristic of the protagonist and the subsequent target action carried out by the protagonist. The independent variable had two levels: high coherence and low coherence. The other independent variable is the coherence in the passage, which also had two levels: global coherence and local coherence. The dependent variable was the rate of “remember” or “know” judgment. Participants were randomly assigned to one of the four material sets and asked to read the texts, followed by a recognition test. They were instructed to press the key labeled new if they believed the sentence did not appear in one of the texts they had read. Otherwise, press the key labeled old, then they were asked to decide whether they had a vivid and clear consciousness of the sentence; if so, they were asked to press the key labeled R (for remember). If they did not have a vivid and clear consciousness of the sentence but still believed the sentence had appeared in one of the texts, they were asked to press the key labeled K (for know). In experiment 2, 40 University students were asked to read 16 groups of narrative passages. The two independent variables were the same as those in experiment 1. The dependent variable was the hit rate of “Old” judgment and the false alarm rate of “New” judgment. Participants were randomly assigned to a material set. Each participant read sixteen blocks of texts, and finished a recognition test in between each block. Upon finishing reading a block, they were given instruction (either inclusion or exclusion). In inclusion test, participants were asked to respond old if the sentence appeared in either Story A or Story B and were asked to respond new to any new item. In exclusion test, participants were asked to respond old only if the sentence appeared in Story B and were asked to respond new if the sentence was from Story A or if the item was new.
According to the results from the two experiments, we find that in the global coherence condition, coherence level (high and low) only affects conscious processing but not unconscious processing; however, in the local coherence condition, coherence level affects both conscious and unconscious processing. To sum up, coherence level can provide different influence on text representation process in different coherence condition.

Key words: text-reading, coherence, conscious processing, unconscious processing