Comparing Components Of Reading Difficulties Between L1 And L2 Children

Reading difficulties may hinder not only children's academic development but also their development during the whole life.Many English-as-a-second-language(ESL)children may have reading difficulties.However,studies on this field were not sufficient.Many studies on reading difficulties always concerned the first language(L1)and mainly focused on one or two perspectives with a small sample size and few variables.Moreover,previous studies were frequent to discuss the relationship between factors and reading development,but they seldom profiled the characteristics of poor readers as a cohort.In order to fill in the research gap,the present study investigated the characteristics of second language(L2,referring to ESL in this thesis)poor readers by comparing the components of reading difficulties between L1 and L2 poor readers under the framework of the componential model of reading(CMR)with a large sample size and multi-variables in a holistic view.In the meanwhile,the study also intended to exam the explanatory power of the CMR model in L2 reading difficulties.The present study established L1 and L2 datasets by extracting data from the 2016 Progress in International Reading Literacy Study(PIRLS 2016).A total of 16,968 L1 speakers including 2,583 poor readers and 14,385 non-poor readers;and 8,972 L2 speakers including 3,410 poor readers and 5,562 non-poor readers were involved.A total of 349 context variables were analyzed.The L1/L2 optimal sets of variables differentiating poor and non-poor readers were found out,based on the support vector machine(SVM)and the SVM-combined recursive feature elimination(SVM-RFE)model.The independent-samples t-test and the chi-square test described the discrepancy in each selected variable between poor and non-poor readers of L1 and L2,respectively.The explanatory power of the CMR in L2 reading proved by analyzing the L2 optimal sets of variables.And after comparing the L1 and L2 optimal sets of variables which differentiated poor readers,unique characteristics of L2 poor readers were demonstrated.The major findings are listed as follows.First,the explanatory power of CMR in L2 reading difficulties was verified with supplementary ecological components about school and information and communications technology(ICT)added in and additional associators of physical condition detected.Second,compared with the L1 optimal set of variables,differentiating L2 poor readers from non-poor readers needs more variables to make a collective effect.Third,much more ecological components(including family,school and classroom environments)with smaller discrepancy between poor and non-poor readers were included in L2 than in L1.This study is a fresh attempt to employ a machine learning technique of classification in the field of reading difficulties,which demonstrates its efficiency in differentiating poor readers and uncovering the collective effect of optimal effects on L2 reading difficulties.Practical suggestions for future research and English L2 learning were also identified.