Abstract:
Objectives: The primary objective was to evaluate the Plaut, McClelland, Seidenberg & Patterson (1996) model of reading against developmental data. The model was evaluated, at two points during training, against data from 23 children in the first term (Time 1) and third term (Time 2) of their reception year. Two main findings emerged:
Two adaptations were made to the training of the network, to bring it closer to the learning environment of a child. An incremental training regime was adopted, as was training on grapheme-phoneme correspondences (GPCs).
Design: All combinations of the two adaptations, incremental training and training on GPCs, resulted in four networks, each of which was evaluated against the children's data.
Methods: Two points were identified in each network's training where word reading performance matched the children's at Times1 and 2. Non-word reading was then assessed, as were the types of errors made.
Results: Non-word reading sharply improved relative to word reading. However, each network still produced more non-lexical than lexical errors.
Conclusions: The combination of incremental training and training on GPCs improved generalisation (non-word reading), bringing performance closer to the children's. However, the type of errors made by each network remained qualitatively different to the children's.
Reference: Plaut, D. C., McClelland, J.L., Seidenberg, M.S. & Patterson, K. (1996). Understanding normal and impaired word reading: Computational principles in quasi-regular domains. Psychological Review, 103, 56-115.
Copyright Notice: The documents distributed here have been provided as a means to ensure timely dissemination of scholarly and technical work on a noncommercial basis. Copyright and all rights therein are maintained by the authors or by other copyright holders, notwithstanding that they have offered their works here electronically. It is understood that all persons copying this information will adhere to the terms and constraints invoked by each author's copyright. These works may not be reposted without the explicit permission of the copyright holder.