文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > A COMPARISON OF NATURAL AND MACHINE LEARNING OF STRESS
侵权投诉

A COMPARISON OF NATURAL AND MACHINE LEARNING OF STRESS

In the language acquisition literature children are reported to make typical errors in assigning stress to words. These errors are taken to be unequivocal evidence that children learn rules for stress assignment as opposed to learning on a word-by-word bas

0.Introduction

How do children learn to assign stress to the words of their native language? Oneapproach holds that they abstract the rules for stress assignment from the language theyhear. This view is commonly advocated in the psycholinguistic literature for theacquisition of metrical phenomena, but also for acquisition in other linguistic domainssuch as morphology or syntax. Alternatively, children may well memorize individualwords together with their appropriate stress pattern. After all, a language user - child oradult - needs a lexicon in which at least the morphologically simplex words of thelanguage are stored. The information available for each item in the lexicon may thusinclude segmental information as well as the word's stress pattern (Dell, 1986; Levelt &Wheeldon, 1994). The main advantage of the former approach is that it relieves thelexicon from much redundancy: rules explicitly embody the general regularities thatunderlie the system of stress assignment. The latter approach in which all wordsincluding their stress pattern are stored in memory is less economical in this respect, butit captures in a natural way the fact that most (if not all) words that children use, theyhave heard correctly stressed in the input. Learning is drastically different in bothapproaches: in the first approach, children have to abstract rules in some way from theinput, which is not necessary when they simply memorize words and their stress pattern.In this approach rule abstraction is not part of the learning process. Stressing ofpreviously unseen words is possible in a lexical approach by incorporating a form oflearning (i.e. lazy learning) which works on the basis of the stored examples directlyinstead of on rules extracted from the examples. This possibility is not considered in thelanguage acquisition literature.

In the language acquisition literature the main focus of attention is on children's learningof stress rules. Various languages have been studied, including Spanish (Klein, 1984;Hochberg, 1988a, 1988b), Dutch (Fikkert, 1993, 1994; Nouveau, 1993, 1994; Wijnen,Krikhaar & Den Os, 1994) and English (Gerken, in press; Echols, 1993; Echols &Newport 1992). These investigations set out to show how children learn the rules ofstress assignment in spite of the paradox which is probably most clearly articulated byHochberg (1988b, p. 684): on the one hand "... there is no need, other than economy, forthem [i.e., children] to generalize stress RULES." On the other hand children do learnstress rules although rule learning is not necessary: "... a finding of stress rule learning inSpanish could be seen as an indication that children's propensity to hypothesize linguistic

上一页第3页

你可能喜欢

热门文档

相关文档

站点地图 | 文档上传 | 侵权投诉 | 手机版
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.guandang.net All Rights Reserved.  -  email