東華大學圖書館 |

語系: 繁體中文

說明(常見問題)

回圖書館首頁

手機版館藏查詢

登入

回首頁

切換: 標籤 | MARC模式 | ISBD

Are rules a thing of the past? A si...

Carnegie Mellon University.

FindBook

Google Book

Amazon

博客來

Are rules a thing of the past? A single mechanism account of English past tense acquisition and processing.

紀錄類型:	書目-語言資料,印刷品 : Monograph/item
正題名/作者:	Are rules a thing of the past? A single mechanism account of English past tense acquisition and processing./
作者:	Hoeffner, James H.
面頁冊數:	335 p.
附註:	Chair: James McClelland.
Contained By:	Dissertation Abstracts International60-11B.
標題:	Language, Linguistics. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=9950010
ISBN:	9780599519954

Are rules a thing of the past? A single mechanism account of English past tense acquisition and processing.
Hoeffner, James H.

Are rules a thing of the past? A single mechanism account of English past tense acquisition and processing. - 335 p.

Chair: James McClelland.

Thesis (Ph.D.)--Carnegie Mellon University, 1996.

We also considered the ability of networks to explain universal constraints on morphological operations and found that the network's consistency bias led to a preference for the stem+affix inflectional systems that predominate in natural languages.

ISBN: 9780599519954Subjects--Topical Terms:

1018079
Language, Linguistics.

Are rules a thing of the past? A single mechanism account of English past tense acquisition and processing.
LDR:03359nam 2200349 a 45 001 852294
005 20100629
008 100629s1996 ||||||||||||||||| ||eng d
020 $a 9780599519954
035 $a (UMI)AAI9950010
035 $a AAI9950010
040 $a UMI $c UMI
100 1 $a Hoeffner, James H. $3 1018097
245 1 0 $a Are rules a thing of the past? A single mechanism account of English past tense acquisition and processing.
300 $a 335 p.
500 $a Chair: James McClelland.
500 $a Source: Dissertation Abstracts International, Volume: 60-11, Section: B, page: 5816.
502 $a Thesis (Ph.D.)--Carnegie Mellon University, 1996.
520 $a We also considered the ability of networks to explain universal constraints on morphological operations and found that the network's consistency bias led to a preference for the stem+affix inflectional systems that predominate in natural languages.
520 $a In this thesis, we investigated competing models of the acquisition and processing of the English past tense. We first considered the ability of the dual mechanism model advocated by Pinker and colleagues (Pinker, 1991; Marcus et al, 1992; Prasada & Pinker, 1993) to explain the developmental data. We showed that, contrary to Marcus et al, past tense overregularization is not triggered by the sudden acquisition of an all or none regular tense marking rule. Several other claims of the dual mechanism model were also called into question by our analysis of the developmental data.
520 $a We then investigated the ability of a basic backpropagation network to simulate the developmental data and found several deficiencies in the network's performance. We investigated the reasons for the model's deficiencies, and demonstrated how the model's performance could be improved. A refined developmental model making use of these improvements was able to fit many aspects of the developmental data, including the 'modal' rate of overregularization, the protracted acquisition of regular tense marking, the early advantage of irregular marking over regular, the effects of frequency on overregularization errors, and the differential effects of regular and irregular 'gangs' on the tendency to overregularize irregular verbs.
520 $a Another series of simulations addressed data from adult studies. We showed that our networks could replicate the asymmetrical generalization gradients, the frequency by regularity interaction, and other effects of frequency and consistency on the processing of past tense verbs by adult subjects (Prasada & Pinker, 1993; Seidenberg, 1992).
520 $a We conclude that currently there are no compelling reasons to believe that a specialized, genetically programmed rule module is necessary to explain the acquisition or processing of regular inflectional morphology. On the contrary, we find strong evidence of a powerful and flexible general purpose learning system that is capable of subsuming tasks previously assigned to multiple mechanisms within the confines of a single network.
590 $a School code: 0041.
650 4 $a Language, Linguistics. $3 1018079
650 4 $a Language, Modern. $3 1018098
650 4 $a Psychology, Developmental. $3 1017557
650 4 $a Psychology, Experimental. $3 517106
690 $a 0290
690 $a 0291
690 $a 0620
690 $a 0623
710 2 $a Carnegie Mellon University. $3 1018096
773 0 $t Dissertation Abstracts International $g 60-11B.
790 $a 0041
790 1 0 $a McClelland, James, $e advisor
791 $a Ph.D.
792 $a 1996
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=9950010