Since the results of the three empirical studies，including the eye tracking experiment，the memory recognition test and the evaluation task，are obtained with the same group of participants，we can conclude that there is some degree of correspondence among the perception，memory and appreciation in the light of coherence construction.The results of the three studies reflect different cognitive and psychological processes in people's mind when they are reading/viewing，processing and comprehending the multimodal English learning advertisements.

As mentioned in the eye-tracking experiment reported in Section 7.1，global coherence mainly affects the overall looking time of stimuli and the local coherence affects the gaze switch between AOIs which is meant to find semantic relations to arrive at comprehension.Cohesive devices in the layout，for example salience，work in the bottom-up processes in reading through attracting the direction and path of the attention.These findings are in accordance with the results in the memory recognition test introduced in detail in Section 7.2 and this accordance can shed some light on the cognitive processes in the reading.That is，the overall allocation of fixation in the whole time course of viewing/reading，reflected in the reaction time and looking time，is largely decided by the reader's expectation of the content which is attributed to global coherence.The comprehension-related processes，such as building semantic relations among component elements and the correction rate in memory recognition test，are decided by the semantic connection—local coherence.The effects of layout cohesive factors such as salience occur in a more haphazard fashion，such as attraction of initial attention and scan path.However，due to their capability of attracting attention，they have a significant effect on the recognition memory.That is，salient elements are better recalled in short-term memory than less salient ones.In the evaluation task，the pattern is similar.Most of the reasons people gave for their like or dislike of a leaflet are related to the content which is related to global coherence and local coherence.The reasons based on layout factors given by the participants are much fewer.

The consistency in the findings across the three empirical studies supports the view in many empirical studies in visual cognition that the reader's interest and semantic factors are higher-level factors whereas the surface salience factors such as size and color are low-level factors.For example，Nyström and Holmqvist（2008）report in their eyetracking experiments that semantic factors can easily override low-level factors in fixation selection.The empirical evidence can further verify the theoretical assumption in Chapter 3，that is，the global coherence which is related to the reader's expectation and local coherence which is semantic in nature have higher status than the cohesive devices in layout.Therefore the theoretical assumptions about the relationships among the three levels of coherence construction stated in Chapter 3 are supported.

Although the empirical studies in this chapter have some limitations，for example，due to the lack of control in the test material，only correlation but no causal relationships can be obtained between the coherence resources and the cognitive dimensions of measurement，the results we've obtained are encouraging.More research in this direction will hopefully be conducted in the future.