Chapter 1. Introduction and Background
hand involves using real fixation data obtained by people reading newspaper articles. The frequency with which words are fixated at each position is entered into the second network. The aim of the paper is to establish any differences occurring in the fixation net due to the different fixation positions.
The next section will give a brief overview of both the Dual-Route model of reading and some connectionist models. In particular it will concentrate on the Seidenberg and McClelland (1989) and Harm and Seidenberg (1999) models since the former was a very influential model in the domain and the latter is an improvement of it. This will be followed by a brief description of the split-fovea model (Shillcock and Monaghan, 2003). Another important part of this paper is the reading behaviour of people, namely at which position they fixate words while reading. This is described in the last section of this chapter.
The next chapter will then present the architecture of the networks used for the current work. Additionally it will explain the training regimes used before going on to give the results of the networks after training. Finally, in chapter 3, there will be a brief discussion of the the results and a presentation of possible future work.
A lot of research has been done on visual word recognition (for an overview see for ex- ample Seidenberg (1995)). There are two main types of models of visual word recog- nition and reading. The more traditional Dual Route model (Coltheart et al., 1993; Weekes and Coltheart, 1996) and the Connectionist models (Seidenberg and McClel- land, 1989; Harm and Seidenberg, 1999; Plaut, 1999; Shillcock and Monaghan, 2003). Both of these types of models are concerned with the processes by which the visual