Since 1970 the idea of text comprehension has become more or less established within cognitive science. Zwaan's cognitive control system, Johnson-Laird's mental models and Kintsch and van Dijk's text comprehension have explained a lot about the cognitive operation in text comprehension. Some of their ideas are relevant to the gestalt perception of linguistic structures. For instance, the situational model by Kintsch and van Dijk (1978) is closely linked to unifying organization. It means to build and assemble meaning out of a certain situation or context. Johnson-Laird in his "Mental models" also emphasizes the mental operation in text coherence. He argues that a text can be comprehended as coherent when referential continuity is applied in working memory. Gestalt perception of linguistic structure is the psychological operation which is also not so much to make cognitive analysis of the logic structure. It helps the reader to grasp the whole without his recognition of the analytic parts. In processing a text, readers perform several basic operations. For example, they assign meaning to words, parse the syntactic structure of a sentence, construct a theme for the text. In this process to construct the mental representation of the text, readers use their linguistic knowledge (knowledge of words, of grammar) and their world knowledge (knowledge of what is possible in reality, cultural knowledge, knowledge of the theme). Meanwhile perception also serves as an instrument for the registration of a particular feature.

The gestalt perception in general is to sensitize the subject to the holistic aspect in the text and to instigate him to its building of images. It may back up other cognitive processes or item-processing when they fail to create a coherent mental representation of the image. For instance, the imaginational impression can direct the attention of readers and help them to decide which information is relevant for the image-gestalt and must be activated.

The image or the scene is linguistically sequential; yet its mental actualization is three-dimensional. Just as de Beaugrande claimed, a text appears as a linear sequence of components, which might be represented by a simple formula such as:

T=C1 C2 C3 C4 C5 С6...{...}

Then→the→man→ in→white→said  {...}

                                                  (1978:31)

but in reading, information is redistributed among components for the purpose of comprehension. The mind retains not a level sequence, but a topography with some information stored higher, that is, perceptually more prominent, than others. This is the basic way of grouping-processing. It is simi lar to gestalt perception of patterns in that the reader is to capture the whole as an innate capactiy. He would sort the information accordingly, to build items into a unified whole, a three-dimensional picture.