Display Full Text | Download Full Text
2.04 MB PDF file

Degree

Doctor of Philosophy, Ohio State University, Psychology, 2007.

Abstract

The connectionist approach to reading aloud has been a serious challenge to the traditional dual-route theory, but several issues concerning the theoretical distinctions of the connectionist approach from the dual-route theory remain unresolved. First, through what kind of internal structure a single-route connectionist model represents the two seemingly distinct kinds of ability to process regularities and exceptions without relying on dual-route structure, has yet to be fully answered. Second, the question of whether the single-route connectionist model is indeed functionally a single mechanism is yet to be convincingly demonstrated. Third, whether a single-route model can simulate surface dylexia, of which the dual-route theory has been the only traditional interpretation, has not been thoroughly investigated. By taking a model from Plaut et al. (1996) and examining it closely, the present study attempts to resolve these issues. Various forms of network analysis demonstrate that the representational system in hidden unit space is structured in the same way regardless of learning regularities or exceptions. Further analyses about the effect of the reading network's exception learning upon its nonword reading reveal a proper viewpoint on the relationship between its regularity and exception learning. That is, ‘exception learning’ in connectionist modeling of reading aloud does affect the model's nonword reading performance just as ‘noise capturing’ in statistical modeling does the model's generalization. In reality, however, the severity of “ordinary exceptions” in normal word reading happens to be not high enough to ruin the network's nonword reading, as “noise” does in statistical modeling. These findings are also used to disprove the idea that the reading network may have developed a functional dual route. A careful interpretation of the findings shows not only that a hidden representation corresponding to the correct pronunciation of an exception word develops even in the baseline network, but also that it is by no means distinguishable from that in the network that learned exception words additionally. Lastly, it is clearly demonstrated that surface dyslexia can be simulated by a single-mechanism connectionist reading model with damage of a particular type, which is motivated by one of the main findings of this study.

Keywords

Quasiregular mappings; Word naming; Reading aloud; Connectionist model; Hidden unit representation

Advisor

Jay I Myung

Pages

174p.

Document number: osu1171994549
Permalink: http://rave.ohiolink.edu/etdc/view?acc_num=osu1171994549

This ETD has been downloaded 405 times (through March 2013)