Eye movements have been extensively investigated as an indicator of visual information processing and cognition (Inhoff, Pollatsek, Posner, & Rayner, 1989; Kowler, 1990; Noton & Stark, 1971; O’Regan, 1981, 1990; Rayner, 1978, 1983, 1992, 1995; Rayner & Pollatsek, 1987; Stark & Ellis, 1981; Yarbus, 1967), since Javal (1878) first observed par saccades in reading. It was discovered that, during the reading process, the eyes do not scan smoothly along text lines. There are two types of eye behavior in reading: saccade and fixation. In saccades, eyes change gaze directions (or visual axes) rapidly from one position to another, the main function of which is to bring a new part of text into the field of foveal vision. On the other hand, the fixation pauses are necessary for extracting the visual information from the text. It was found from many vision studies that the visual threshold for object detecting is increased during saccadic eye movement. Therefore, the vision system is nearly “turned off” and no visual input signal is sent to the brain during this rapid eye movement. This phenomenon is called saccadic suppression (Hubel, 1988; Latour, 1962; Stevenson, Volkmann, Kelly, & Riggs, 1986; Volkmann, 1962, 1976, 19861). Experimental data from reading investigations also demonstrated that the! acquisition of visual information from text exists only in fixation pauses (Wolverton & Zola, 1979, 1983). There are microsaccades, slow drifts, and high-frequency tremors in a fixation pause; however, their individual function in reading is still controversial (Ditchburn, 1980; Kowler & Steinman, 1980). The main eye-movement parameters,

employed to rep-resent the temporal and spatial reading behavior, are the fixation duration and saccade length (including both forward and regressive saccade) or recognition span (Adler-Grinberg & Stark, 1978; McConkie, 1983; McConkie & Rayner, 1975; Morris, Rayner, & Pollatsek, 1990; O’Regan, 1990; Rayner, 1983, 1995; Rayner & Inhoff, 1981; Rayner & Pollatsek, 1987; Sun, Morita, & Stark, 1985). In this chapter, we often refer to saccade length as the reading recognition span because this term can be used to delineate spatial processing even when no eye movements are recorded, as occurs in the Rapid Serial Visual Presentation (RSVP) technique (Adler-Grinberg & Stark, 1978; Fang & Sun, 1993; Sun, Morita, & Stark, 1985). For the sake of practical experimental data processing and recording, the fundamental parameters of reading eye movements could be summed up as the fixation position and the fixation duration because the saccade length or recognition span can be easily deduced from fixation positions (Sun & Chen, 1995).