Interpreting the Learning-Mismatch between L1 (Bangla) in Learning L2 (English): An I.R.T. Approach
affiliation not provided to SSRN
Indian Statistical Institute
Surja Shankar Ray
Indian Statistical Institute, Kolkata
March 7, 1997
Since the geniuses of the two languages, viz. Bangla, i.e., L1 and English, L2 are in many ways different externally and because English is important in the Indian context –mainly as a language for national discourse, higher studies, business, administration – the present study was considered worthwhile as it aims to gauge the extent to which L1, i.e. mother-tongue or Bangla interferes with the learning of English in the context of West Bengal (India). When English is taught in West Bengal’s Bangla medium schools as L2, the students confront some major problems and this hinders their learning of L2. Therefore, it is necessary to identify the linguistic zones where the interferences occur and also determine the extent of encroachment/cohabitation of these two languages. It is assumed that questions or items loaded with selected components (a mismatch between L1 and L2) are expected to be relatively difficult for a learner. In the present study, questions with differential loadings or mismatch have been identified and nature of the difficulties of the items faced by students have been proved. The aim while constructing the questionnaire was to locate the areas of supposed mismatch between the epiphenomenal parameters of LI and L2 and to find out the nature of actual classroom interaction: The following suppositions were made:Agreement, Word order, Passivization, Tense, Tautology, Selectional Restrictions, Suppletion or internal change, Yes-No Questions and WH-Questions. Keeping in mind areas of mismatch or difficulty and components stated before, a suitable psychometric model (Item Response Theory) was adopted to measure the learning achievement and component effect. For the purpose of our study the Rasch model was adopted. Items were dichotomously scored; 1 for correct response and 0 for wrong response or no response. To put it mathematically, response to an item i by an examinee j is denoted as uij. Learning achievement is considered as a continuous variable θ. According to IRT, the probability of giving a correct response to i by an examinee with ability θj is defined as Item Characteristic Function (ICF). The following inferences can be drawn from the results mentioned above: 1.From the results obtained, it is certain that where the idiomatic use of English is demanded, the difficulty level is the highest. This only confirms the fact that Idiomatic usages are culture-specific(context-sensitive) and greater attention should be paid to them when pursuing the syllabus. 2. Question structures, passsivizations, word-order, tense and selectional restrictions: these components have presented moderate difficulty. One reason could be the mismatch between L1 and L2. Here the interference of L1 is obvious and hardly needs explanation. 3.Simple tenses and suppletion, internal changes have proved easy Items for examinees. The possible reasons are (a) these tenses are without any riders attached. In other words they are not determined by auxiliary markers. But, in cases where such markers are demanded, the problem of tense has proved difficult. (b) in the case of affixations, the uniqueness of words must have determined correct responses This is obviously because of the lack of familiarity with the word and because of distracters. 4. It may further be noted that the multiplicity of components in an item is not the determining factor for a correct response. This is perhaps because of our gestalt way of learning a language. A learner perceives language as a whole and not as something fragmented as made out by grammarians. 5.Moreover, it is observable that items on affixations have proved easy. This Is perhaps because greater emphasis is put on learning words rather than on learning sentence structures. In conclusion it should be noted that the performance of English of the Bangla Medium school on the whole is poor, 22.37 being the average score.
Number of Pages in PDF File: 8
Keywords: Item Response Theory, Language Learningworking papers series
Date posted: March 14, 2012
© 2013 Social Science Electronic Publishing, Inc. All Rights Reserved.
This page was processed by apollo7 in 0.437 seconds