IR
T16
انگلیسی
IR
Learning English Pronominals from Input And General-Purpose Learning Mechanis (Simulated by Artificial Neural Networks)
[Thesis]
/محمدرضایوسفی حلوایی
Tabriz: tabriz university
, 1388
86 P.
مکانیزم های آموزش
داده
شبکه های عصبی مصنوعی
Print
Bibiography
M.A.
, English Language Teaching
1388/07/05
For the last half of twentieth century Generative Linguists following Chomsky have argued that grammar is innate, rule-governed and exists in brain as domain-specific module. One of the areas of language that attracted their attention the most is the complex system of referential elements in any language. This complexity is often discussed and construed as an indication of the innate predispositions for language learning. This study is done on the learnability of some of these pronouns by input and general-purpose learning mechanisms.the possibility of learning English pronominal elements in connectionist networks was investigated. The selected pronouns were he, she, him, her, himself and herself. These were tested in two different connectionist networks. First a feed forward back propagation network and second an Elman recurrent network. What the obtained results wholly indicate is the possibility of learning English pronouns by artificial neural network. But there were some inconsistencies between the results obtained from these two networks. First in almost all the simulations, the results obtained from the Elman network were closer to desired values. Then there were some cases where both a pronoun and its referent, the proper name, were activated. Also the networks' power for generalization to novel names was not as good as the encountered names. The reason for these behaviors of the networks is discussed in the related chapter. The findings of this study contribute to the emergentist and associationist accounts of language acquisition. It has the theoretical implication that the role of innate language-specific predispositions for language acquisition is not as much important as it is asserted in Generativism and there is the possibility of acquiring pronominal elements in English by data and general-purpose learning mechanisms..
یادگیری ضمائر انگلیسی با داده و مکانیزم های عمومی آموزش (شبیه سازی با شبکه های عصبی مصنوعی)
Artificial Neural Networks
Connectionism
Input
Learning Mechanisms
Yousefi Halvaei, Mohammad Reza
Ansarin, Ali Akbar, Supervisor
Feizi Derakhshi, Mohammad, Co-supervisor
ایران
20211010
پایان نامهPE,1127,.Y6L3,1388
یادگیری ضمائر انگلیسی با داده و مکانیزم های عمومی آموزش ) شبیه سازی با شبکه های عصبی مصنوعی(
فوق سری
عادی
E,yosefi.doc
1037312
کاربر
متن
0
T 16
سیاه و سفید
Location: Central library
old catalog
p
TL
276903
1
a
Y
