Abstract
In this article, I discuss the implementation of neural network models for reading English words aloud. Since 1989, there has been existing a debate about the models of reading English words aloud. One is the Dual Route Cascaded (DRC) model. The other is the Triangle model, whose original version was developed in 1989. Because there are arbitrary variables of both models, we did not decide which model gives better accounts for the numerous data given by psychological experiments and neuropsychological evidence. Therefore, in order to give a solution of this debate, an attempt to integrate both models was made. Introducing the Mixture of Experts Network, an elegant solution to overcome the arbitrariness of both models could be given. The Mixture of Expert Network can include both the models as a special case of this Network. From the Mixture of Expert Network's point of view, the difference between the Dual Route Cascaded model and the Triangle model would be able to describe as the quantitative difference of the dispersion parameters.
