Eye Movements in Reading

A reader’s eyes move along a line of text in a sequence of fixations separated by jumps called saccades. Reading, therefore, takes the form of a series of “snapshots” during which text information is acquired. During one such snapshot the recognition of a single English word of average length takes about 100 ms (Rayner and Pollatsek 1989). From the pioneering research of McConkie et al. (1988), it emerged that the effective target of a given eye movement is the centre of the to-be-fixated word. The eyes aim to attain an “optimal viewing position”, which, if fixated, facilitates rapid word recognition (O'Regan 1990; Rayner 1979). In reality, fixation locations are normally distributed with means somewhere between the beginning and the centre of the word. This latter position in the word is referred to as the preferred viewing location (PVL). The eye frequently undershoots or overshoots the optimal position or even the word boundary with the consequence that extra fixations and/or movements back to previously read words may be made (Nuthmann, Engbert and Kliegl 2005).

An important issue in eye movement control in reading is what factors determine the target of a new fixation. The results of computer modelling by Reilly and O'Regan (1998) suggested that the decision is a rather low level one. Simulation results suggested that the default target for the eye is the next most visually salient word to the right of the current fixation and within a limited range of words. Higher-level processes appear to intervene only if there are comprehension difficulties.

Computational models such as that of Reilly and O’Regan (1998) have played an important role in exploring the dynamics of eye movement control in reading. The principles and algorithms that different computational models embody, reflect the theoretical commitments of their authors. In effect, the models are computational instantiations of aspects of their theories. In the case of E-Z Reader (Reichle et al. 1998), sequential lexical processing is proposed as the obligatory trigger for the generation of all eye movements made in normal reading. An alternative view is that the positioning of fixations in reading can be achieved by using a set of rather simple eye movement patterns that operate independently of the type or meaning of the words being read

(e.g., Reilly and Radach 2006 and 2002; Engbert et al. 2005; Reilly and O'Regan 1998). Both perspectives have their merits and can account for important aspects of eye movement behaviour during reading. On the other hand, all of the approaches have their limitations. Therefore, the question of interest is not whether eye movements are determined by oculomotor factors (i.e., relating to low-level vision and eye muscle control) or higher-level lexical and linguistic ones, but to what degree these two groups of factors are involved and how they interact.

One way of deciding among competing accounts is to observe which models can best explain reading behaviour for a range of writing systems, not just the Roman system, which has dominated research to date. While there has been a scattering of studies on Chinese reading even going back as far as 1927 (Yang and McConkie 1999; Shen 1927), writing systems other than alphabetic ones have generally been neglected. The study by Yang and McConkie (1999) represented the renewal of interest in non-Roman writing systems, an interest that has been steadily growing (Reilly and Radach 2012). The specific focus of this chapter is on the word-targeting process and the concomitant landing site distributions, since this is likely to be the first point in the informationprocessing pathway where differences in writing systems start to have an impact (see Figure 1 for clarification of the relevant eye movement terminology).

Figure 1. Schematic representation of the main parameters used in the study of eye movements in reading.

A fixation is the point at which the eye remains still during reading and it typically lasts for about a quarter of a second. Eye movements involve discrete ballistic jumps of the eye, referred to as saccades, which typically progress from left to right and have an average extent of eight characters for English texts. Their length will vary across writing system. The term “landing site” refers to the end location of a saccade. Note that saccades can sometimes skip words if they are short.