Chapter 1. Introduction and Background
Table 1.1: Frequencies with which four-letter words are fixated at different fixation posi-
tions during reading. In the example, ’-’ indicates the fixation position. The data is taken
from McDonald and Shillcock (submitted).
that the optimal viewing position slightly left of centre allows each hemisphere to receive the same amount of information about the word.
The finding that there is a preferred viewing position is interesting for the split-fovea connectionist model. What will happen to the model if this fact is incorporated into the network and will this network behave differently from a normal control network? The network presented in the next chapter integrates the findings about the fixation positions during reading with the split-fovea connectionist network. Table 1.1 shows the data used for frequency of fixations at different positions for four-letter words. The frequency data is taken from McDonald and Shillcock (submitted) and the frequency of the fixations was evaluated with data from the EMBRA corpus which has eye move- ment data from people reading newspaper articles. Note that in table 1.1, the words are fixated slightly right of centre, which is contrary to the claim made before about the preferred viewing location. This is due to the fact that the data in this table is col- lected during reading and not with single word presentations. However, this is only the case for short words. For longer words the highest frequency of fixations will again be slightly left of centre (McDonald and Shillcock, submitted). One explanation for this behaviour is that while reading, the next word (the one after the currently fixated word) will already be in parafoveal vision. Because of this, the first few letters of the