DATA ANALYSIS OF THE GAP TEST
6.15 General comments on the results
Two hundred variables were identified for analysis. Of these, seventeen variables recorded insufficient data for analyses to be made. These included the unacceptable self-corrections for the less and more able readers, the non-responses for the more able readers and for the group of bilingual girls and boys, the total number of reversals for the more able readers and for the bilingual group, and the acceptable multiple miscues for the bilingual group. Insufficient data was obtained for two reasons. In some cases a large number of pairs recorded the same number of errors and therefore could not be given a "d" ranking. This resulted in insufficient pairs from which to obtain data. Secondly, the number of miscues generated was sometimes small. This occurred with unacceptable self- corrections, non-responses and reversals. The number of unacceptable self-corrections was only forty for all the matched pairs, and the number of non-responses only twenty-eight for the more able readers and thirty-seven for the bilingual group of boys and girls. The total number of reversals was only twenty- five. Obtaining data in such cases was not possible.
Eleven variables in the miscue analysis of reading, including the substitutions, recorded data of statistical significance. Among the less able readers, the data showed statistically significant less unacceptable stress and intonation errors for the less able monolingual readers (.05 level) and less of the non-responses (.01 level) for the same group. The less able bilingual readers recorded statistically significant less of the acceptable multiple miscues (.01 level) and of the acceptable insertions (.05 level). The older monolingual readers recorded statistically significant less repetitions (.02 level). This would suggest the less able bilingual readers still in the process of acquiring oral language skills in English, which they could then transfer to their reading. As the data for the more able readers showed no statistical significance, then this might suggest that the bilingual children were achieving parity with their monolingual peers.
In the analysis of the substitutions, the monolingual children in the older age group recorded a statistically significant smaller number of substitutions (.05 level). The monolingual children in the younger and older age groups and also among the less able readers recorded less acceptable graphophonic substitions (.02, .05 and .05 levels respectively) than the bilingual readers. This could indicate bilingual children either having a greater control of phonic skills or having a greater dependency on phonics for the interpretation of the text compared with monolinguals. Monolingual children also recorded statistically significant less of the unacceptable semantic substitutions in both the older age group (.02 level) and the group of less able readers (.05 level). This might indicate the
need for the extension of oral language support for certain
groups of bilingual pupils and for continued help with conceptual development in a second language. It could also indicate the value that community language teaching might have in helping with concepts.
The analysis of the data for both the retelling of the story and the comprehension tests indicated that younger bilingual readers scored statistically significant less marks when recalling the story plot (.05 level) and main events of a story (.05 level) and on questions on word meanings (.01 level) and literal comprehension (.02 level) and inferential comprehension (.05 level). Less able bilingual readers scored statistically significant less marks on the questions on word meanings (.01 level), literal comprehension (.01 level) and inferential comprehension (.05 level). There would seem to be a tendency among the younger and less able bilingual readers for reduced conceptualisation of what had been read compared with mono lingual readers. This is hardly surprising as they are reading in a second language. However, no differences of statistical significance were recorded for the group of more able readers except for misconceptions.
One variable that gave cause for concern was that of the mis conceptions made during the retelling of the story. In all groups the monolingual pupils recorded statistically significant less misconceptions than bilingual pupils (.01 level). An analysis has been made of these misconceptions and could provide a useful reference for future teaching.
The findings must be seen as being tentative and there is a need for the study to be replicated with other groups of bilingual children. However the findings do seem to replicate the work of some authorities. Goodman (1978), in his work on the miscue variables of bilingual children, found that "there is a slightly greater tendency for second language readers to come closer graphically and phonetically when they miscue." However he also identified progression in reading stating that "if they speak English fluently they will act like native speakers in reading English" and that "as they become bilingual, readers will show this in their reading as they do in their speech." A similar progression has been indicated in this study. The more able bilingual readers, for example, recorded only one variable, the misconceptions made when retelling the story, where the data recorded statistically significant differences.
Ellis (1980), analysing the miscues in the reading of twelve Asian children aged 7.2 to 11.2 years with a reading age of 7.4 to 8.5 years, also identified "the more able readers" as being able to "decode at the semantic level", whereas "the less able readers" tried to identify words from graphic information. This also was supported by the present research, where bilingual subjects in both age groups and among the less able readers recorded statistically significant more acceptable graphophonic substitutions than the monolingual subjects. However, there were no differentials between the more able bilingual and monolingual readers at the semantic level.
Some research and statements by authorities were not always replicated in this study. Robinson’s research (1980), which
analysed the social background of ethnic minority children with respect to reading performance, suggested that parents of Pakistani children in the north of England, who might be of low socio-economic status and speak little English, might have children underperforming in schools. This was supported for some of the differentials already listed, as some of the data for both the chronological age groups and for the less able bilingual readers indicated that the bilingual children had not reached parity with their monolingual peers. However, with the more able readers, any differences in performance were only apparent in the misconceptions made when retelling the story. As literacy skills are gained in a second language, it would seem as if parity between bilingual and monolingual readers is reached. Authorities have suggested that there might be differentials in attitudes between bilingual and monolingual children and between Muslim boys and girls (1.4). However when the attitude to reading variables were examined, no differentials were found.
CHAPTER 7