CHAPTER IV
Discussion
Introduction
The methodology utilized in this study has not been limited to one statistical design, nor has it relied on one single measure of linguistic proficiency. Several methods and instruments have been integrated to allow the collection of as much relevant data as may be conjectured to influence the learning of a second language. Cummins’ framework of linguistic interdependence has set the parameters for data interpretation. It has served as the hypothesised criterion in determining the bilingual profile of a language minority population of Hispanic children undergoing a primary school experience in a bilingual school.
However, within this framework a specific question has been given priority: whether language interdependence is also developmental, increasing over time? The question asked here concerns the problem of how the first language may act as a contributing factor vis a vis second language development in the school, from one grade to the next. Therefore, yearly academic language development becomes the focus of concern.
The steps followed in Chapter III were logically sequenced to allow for an explanation and a prediction of second language behaviour in each of the two dimensions previously describe. The following discussion will therefore continue with this sequence.
The t Test
Results of this test show two distinct populations of children in terms of their linguistic profile: students whose bilingualism allows for incremental proficiency in L 2, and students whose bilingualism seem to provide no contributory effect on L 2. Because the high Spanish proficiency group was in fact the group which contributed to high English proficiency, Spanish proficiency appears to be developmentally essential to second language development.
A poorly developed (or developing) first language system may have consequently delayed progress in the second language. The paradigm of bilingual development as a psychosocial process contingent upon first language status (affective and social) in the context of this specific socioeconomic setting may be reflected in this outcome.
Results indicate, on the other hand, decreasing English proficiency in both groups, the high and the low Spanish proficient group, in each consecutive year, as well as a diminishing of the range of the means of the grouping variable year by year. A similar pattern was observed for the Spanish proficiency scores. Table 12 details statistics showing a yearly downward trend on mean scores for each one of the linguistic variables considered in the study.
A diminishing standard deviation in English proficiency scores, not observed in the Spanish variables, may result for several reasons. One, the children tested were LEP students (these students yield poor results on tests which measure the level English proficiency) allowing for less variability to exist in these children’s English linguistic repertoire. There is therefore a smaller range of knowledge of English allowing for mean changes in score to be greater depending on the lack or the existence of further learning. Secondly, the first grade English language proficiency test (MAC) included mostly material related to basic interpersonal communicative skills minimizing the use of the academic aspects of the language. This allowed for the tapping of a considerable range of skills, depending on the previous experience of the child with language use in ordinary day by day communication. This may explain the high variability of scores in first grade.
However, this was not the case in second and third grade. A greater use of the academic knowledge of English was demanded by the second and third grade English proficiency tests,thus reflecting the formal learning of the language by the student on each of these successive years. Poorly consolidated academic first language skills consequent to the lack of attention to the developmental interdependence of these children’s bilingualism may account for the downward trend in variability and the mean scores of these students’ English proficiencies.
Possible Sampling Bias
Limiting sampling procedures may possibly explain the results observed above, which show a significant downward trend on the language proficiency yield of the full sample. This would mean that the sample does not represent the full bilingual population, but only that portion whose bilingualism is not additive. If this were the case, the study sample would serve only to explain the profile of that part of the population of bilingual children who failed to learn English adequately. It would destroy the purpose of the study, that is, the analysis of additive bilingualism or effective linguistic interdependence.
The sample used in this study was arbitrarily limited to children who had remained in the bilingual program for at least three years. Children not selected in the sample were presumably similar to the sample in their bilingual, socioeconomic and family backgrounds. However, the focus of concern was that the children not selected may differ linguistically from those selected in the sample.
These not selected exited from the city-wide system or the program after the first or second year of school. Some may have moved out of town. High mobility is common in this area among migrant families which mostly make up this specific population.
Others may have been mainstreamed into the regular curriculum on account of having made significant gains in English proficiency and/or other reasons. Some may have
transferred to private schools.
No statistics are available to describe the number of children leaving the system and the number of children being mainstreamed into a regular monolingual program. Estimates
given by the principals and teachers of the three schools which hosted the study indicate that about fifty percent of the children never stay in the same school longer than two
or three years and about ten percent are mainstreamed yearly during the first three years, from first to third grade.
The exclusion of children from the sample due to higher geographic mobility would not likely explain the downward trend in proficiency scores on the sample group. Yet
children not included in the study because they had been mainstreamed into the regular program and out of the bilingual program (these did not qualify for the administration of the CTBS Espanol during their second and third grade years) may at first hand appear to provide an explanation for the trend. These children may have presumably been mainstreamed mainly because they had acquired a high threshold level of English proficiency. If this is true, the better part of the linguistically proficient children may have been excluded from the study.
However, the exclusion of the mainstreamed population from the sample does not explain the trend for the following reason: the study sample revealed a high variability on the
normal curve equivalent scores of the MAC-1982 similar to that generated by the full population of bilingual children (Maculaitis, J. (1982 p. 356).
The figures in Table 12 (Chapter III) illustrate normal curve equivalent mean values for the sample. A normal curve equivalent score of 50 is set for the mean. The standard
deviation of this type of normalized standard score is 21.06. Figures for 1982 in both languages (as shown in Table 12) demonstrate that these children yielded mean normal curve equivalent scores similar to or better than the normed data in 1982. The Standard deviation score for English-82 is also similar to that of the normal curve equivalent standard deviation (see Table 12).
One further point to consider relates to the characteristics of the mainstreamed population. Criteria other than high scores on English proficiency are ordinarily used in the Newark district for mainstreaming. Parents and/or teachers request that their children and students be mainstreamed to allow what they believe to be a more rapid learning of English after the discontinuation of Spanish. Still others are kept in the bilingual program because their parents wish them to stay in a bilingual program for reasons of Spanish language maintenance in spite of having reached threshold or above-threshold levels of proficiency in English. These decisions about who may be mainstreamed significantly change the linguistic profile of the mainstreamed population and lend support to the assertion that the sample utilized for this study is representative of the full bilingual population of children in the bilingual program.
The question which still remains to be treated is whether other variables may have intervened in explaining why children of the sample which yielded initially high
scores on both Spanish and English proficiency tests, on the average, failed to maintain similar or improved levels of proficiency from year to year. To answer to this question,
two important variables must be taken into account: the prevalent transitional bilingual education curriculum and the kind of relationship existing between the school and the
community it serves.
1. The curriculum may have contributed significantly to the decreasing Spanish and English proficiency main effects. The following comments (ex post facto, to be sure) may well
explain why curricular procedures may partially account for the problem. During the course of the two years which follow the first test (second and third grade) the child is exposed
to an ambivalent bilingual education curriculum. In the first place the curriculum is not sensitive to the differences in the linguistic profile of two distinct groups of children already observed to exist, and therefore, not geared to provide a differential methodology for the
teaching of English.
In addition, during observations of third grade classes and after several informal interviews with five of the teachers of the children of the sample, it was obvious that Spanish, in practice, could not be taught effectively as a formal learning discipline. The highly mixed student-language composition of the classrooms prevented any real attempt to teach Spanish effectively. Spanish served primarily as a medium of instruction; however the teachers frequently confronted difficulties when teaching Spanish asa subject in the same manner as it is conventionally taught to regular monolingual Spanish speaking students in Spanish speaking countries. To give one example, one of the textbooks used in third grade contained readings about life in Puerto Rico some twenty years ago. Students were confounded because the stories they read were very much unrelated to their everyday experiences. This distance between their old country traditions and their current experiences seemed too far apart to allow for the building of cultural-linguistic identity, thus making the learning of Spanish more difficult.
On the other hand, English, which was in fact the target language of the school, was not treated as an academic language in the same way as it was for children of the regular program. It served basically two purposes: as a code switching strategy during the Spanish language curriculum, on one hand, and as a single subject course on the other.
English was rarely used as an alternate medium language in the bilingual curriculum. For example, course material was not regularly taught in English. On the other hand, Spanish was never given the status of a long term target language, therefore fostering poor motivation and discouraging attitudes towards engaging seriously in its study.
Garcia (1983) has indicated that the quality of language learning is seriously hampered when the teacher mixes languages, addresses children in both languages indiscriminately and teaches the same subject by means of two different language codes. This unsystematic approach to the teaching of languages was observed to be similar for the bilingual students used in this specific study. Garcia (1983) furthermore insists that language selection within the classroom must be analogous to the structure offithe speech community. She, therefore, supports a partial maintenance program, also suggested by Fishman and Lovas, (1972). Such a curriculum calls for the careful and rigid compartmentalization of language uses in the classroom. Each teacher must, then, be aware of the individual linguistic weaknesses and strengths of the students,allowing those students with strong vernacular backgrounds to maintain and foster that language and those with a poorly developed first language to understand the importance oflearning it well. Both languages should be, therefore, allowed equal status as academic and oral languages (Fishman1980), although functionally compartmentalized from each other.
2. The second variable addresses the status of the school-community relationships, that is, the bilingual school and the language minorities it serves. The following comments regarding bilingual education extend beyond Cummins’ linguistic interdependence hypothesis. However, these may relate considerably to long term educational outcomes involving the student population being served. Bilingual schools in the United States are public institutions so structured as to serve a specific purpose, that of preventing language minority children with little no understanding of English from experiencing insufficient academic progress in an all-English school environment. The recommended remedy is teaching course material in the vernacular of the minority language child until he or she is
minimally capable of undergoing schooling in English. The results of this specific pedagogy have already been partially observed in this and other research findings. It is well established that the end product, for children undergoing bilingual education, is frequently school drop-out and/or a minimally prepared individual.
These schools are not managed or controlled by the minority language community they serve, but by local school administrations, which answer to local and state school
boards, and Education Departments controlled by the white majority population. So communities which are currently served by the bilingual schools in their area remain powerless educationally, economically and politically to initiate action on behalf of the ethnic student population and decide what kind of education they want for their children.
It appears necessary for the wider community of Hispanics to organize and set their own educational goals in conjunction with the Hispanic population directly engaged in the education of Hispanic children. It seems that otherwise, bilingual schools will remain institutionalized to serve the good of the majority, which is frequently different from that of the language minority populations. Furthermore it may not be until parents create theirown ethnic-mother-tongue schools that long lasting and authentic support of the vernacular will ever be effected. Fishman (1980), Fishman and Milan (1980), and Garcia (1983) argue for the social organization of Hispanics as an ethnolinguistic group, the Spanish-mother-tongue school as asocial institution and community oriented planning of a
curriculum to teach both English and Spanish in schools controlled by Hispanics.
Yet there is still another crucial factor which must be considered when planning ethnic-controlled ultimate goal of any educational enterprise regarding creating competitive educational excellence for their children (Fishman 1980; Garcia, 1983).
Analysis of Covariance
The analysis of covariance was included in the statistical analysis of the results primarily to observe the effect of intelligence on linguistic development. If registered values on the IQ variable significantly covaried with L2 variables, then intelligence may have been considered an underlying contributing factor of English proficiency. Results show that this was not the case. Intelligence did appear to be higher in the high proficiency group than in the low proficiency group, but the difference was not significant enough to change the results of the analysis.
Intelligence
One reason for the minimal effect of intelligence on linguistic behavior is the relative homogeneity of the IQ scores of the sample. For example, the coefficient of variation of verbal and performance IQ are .13 and .10 respectively. On the other hand the coefficients of
variation of the language variables range from .35 to .25, both comparatively higher than the IQ coefficient of variation values.
Verbal IQ for the full population is lower than average (89), quite a bit smaller than the average IQ score for the normed sample. Performance IQ, for the sample, on the
other hand, was 103, somewhat higher than it would be expected. This piece of information supports data about the decreasing linguistic profile on these children.
In addition, what has in fact emerged from this specific profile is the possibility of a compensatory effect performance IQ as a coping strategy for the academic verbal deficit which seems to be developing in these children. Thurstone’s principle of parsimony (1959) which explains the diversity of factors or traits observed in individuals by reverting to an underlying simpler unifying” ability may serve to understand this phenomenon. These children appear to be resolving their academic verbal delay by creating a balance between performance and verbal intelligence. A cognitive strategy regarding the need for environmental adaptation seems to be in progress. It underlies the general abilities of children in these levels of progress.
The use of this strategy may be reinforced by the specific characteristics of the school and neighborhood. These require that the child make use of fine motor coordination as well as visual motor and socially adaptive skills in handling day to day social transactions calling
for a diminished preocupation with academic endeavours.
Interaction
Analysis of variance results followed along the lines of the t Test. There were significant main effects across groups and from year to year on the proficiency levels of English. However, one further piece of information was provided by this analysis. There was an interaction effect between the grouping variable and time. This means that the L1 high group differentially decreased in L2 proficiency from year to year as compared to the L1 low group.
What reason can be attributed to this effect? If Cummins’ hypothesis is to be applied here, the following explanation seems to be compatible. The high Spanish proficient group did not maintain or improve its Spanish proficiency level from one year to another. In fact, it diminished in competence.
Consequently, English proficiency would be developmentally arrested under such circumstances, finding little support from the stronger linguistic system. The interaction between these two variables (group and time) is further explained (and this is possibly crucial to it) by the fact that the low group did not diminish its competence as intensely as did the high group. The low growth observed poor proficiency in Spanish and English from the beginning. Therefore, there was not much competence at the beginning to allow for significant decreasing changes in the mean values of the group. On the other hand, the high group did have sufficient levels of competence to allow for significant loss in mean proficiency.
Factor Analysis
Three significant factors were generated by the analysis. These are the following:
l. Regressive Spanish-English Interdependence
This means that Spanish and English academic language scores loaded on one single factor (the strongest factor) reflecting a bilingual profile of inter linguistic developmentally related skills (CALP). However, it was also observed that, these childrens’ language scores decreased in value from year to year. For this reason, a regressive interdependence has developed between the languages, one apparently leading the other downward.
2. Regressive English 0ral-Academic Proficiency
This factor describes factor loadings on English proficiency variables as well as time lived on the mainland versus the homeland. It combines two aspects of language proficiency, academic and oral. The correlation between longer time lived on the mainland and academic English seems reasonable. Children who lived longer in the mainland learned more English than children who lived in a non English environment.
It also seems compatible with Cummins' hypothesis that these children evidenced a gradual decreasing trend in academic English proficiency, as it was in fact observed. It confirms Cummins' assertion about the need for maintaining cognitive and academic skills in the first language in order for second language skills to evidence positive development. Longer residence on the mainland does allow for better communicative skills in English to develop, but it may also explain a lack of development of deep level cognitive and academic skills in Spanish.
Some of these children showed good communicative skill in English and also reflected these abilities in first grad when the MAC was administered. This test, given to first graders, relies mostly on oral skills to determine English proficiency because they are not yet sufficiently exposed to academic English. However, having good communicative interpersonal skills in English is not sufficient to allow for the development of cognitive academic skills in English for children whose first language is not English. Regressive English oral-academic proficiency additionally indicates how communicative skills in English assumed to be superior in children who have lived longer in the mainland may interact with curriculum variables to generate a negatively skewed development of English. Cummins (1979) pointed out a distinction between oral and literate linguistic skills reporting them to be separate abilities involved in classroom learning.
A teacher may assume, for example, that a Puerto Rican child who speaks English without an "accent" may be as good as a native speaker of English on tasks which require elaborate thinking in that language. This misconception may lead to decisions which possibly delay his or her linguistic progress. An erroneous strategy applied to these children may be to deemphasize their need of progress in Spanish because a successful transition into English has been presumably accomplished.
3. Positive Father-Child relationship and School Grades This factor combines two apparently unrelated areas of academic development: the father’s presence in the home and school grades. Since neither variable refers to language development as such, they will not be treated in depth here. However, they do bear on two very important areas of the child's progress: home discipline and emotional stability.
Most fathers seemed to have reported absent from the house because of a previous history of conflict and separation. This may explain the children's difficulty in adjusting to a school environment. The home environment may not have been conducive to the child's academic development.
The first of the two factors clearly demonstrates a linguistic interdependence pattern very much like that explained by Cummins’ hypothesis. L1 and L2 are shown to be intercorrelated in a single cluster of linguistic variables including within language and between language correlations, within and across the dimension of time for two consecutive years. The second factor revealed how these children’s improved pragmatic proficiencies in English only, correlated with academic proficiency. However, academic English proficiencies deteriorated throughout the span of three years, and Spanish academic proficiencies did not seem to have consolidated on these children. This, then confirms Cummins’ assertion that surface skills in English are unrelated to the development of deep cognitive-linguistic skills. The lack of support by the vernacular language would ultimately result in the global diminishing of bilingual proficiency. The regressive bilingual development of these children during the first three years of school, from first to third grade, is of great concern. This information may be instrumental in a thorough evaluation of the bilingual curriculum currently in use in the school system.
Multiple Regression Analysis
English proficiency test scores obtained from these children in 1982 and 1983 are good predictors of these childrens’ proficiency in English in 1984. It has already been observed in the correlation matrix (see Table 15, Chapter III). The correlation coefficient for English 1982-84 is .62; for English 1983-84 it is .56.
Spanish proficiency, on the other hand, seems not to play a crucial role in the development of English proficiency at the term of three years. The correlation coefficient (Spanish and English 84) for Spanish-1982 is .18; for Spanish-1983, .21. These values are much lower than would be expected. The correlation between English andSpanish variables had been much higher in previous years (see Table 15, Chapter III). For example, the correlations between English-83 and Spanish-82 and 83 are .46 and .45 respectively. However, both languages show decreasing trends in proficiency at a yearly pace. The decreasing proficiency values in progressive yearly measures may provide a different bilingual longitudinal profile than predicted by Cummins’ hypothesis. This interlinguistic pattern between English 1984 and Spanish 1982-83 may be interpreted to mean that as both languages deteriorated in terms of academic competence their interdependence is likewise canceled. Each language apparently maintains its own reverse developmental course.
This may be called regressive developmental bilingualism. Some questions still remain to be answered. Why did the remaining variables not included in the stepwise regression analysis not contribute to English variance in 1984. These are (1) time lived by the child in he household, (2) academic grades, (3) oral exposure to English and Spanish, (4) education level of the parents, (5) presence of the father in the household, and (6) performance intelligence.
The effect of academic grades is significantly reduced by the inclusion of English-82 and school behaviour grades in the regression analysis. Academic grades correlate highly with English-82 and school behaviour grades the making their inclusion in the regression redundant. In fact these variables, together with both Spanish variables, belong to the first factor (the Factor Analysis).
The effect of the subjects’ time lived in the homeland has also been absorbed by English-82 and school behaviour grades. Therefore, its contribution has been eliminated from the regression.
The remaining variables, oral exposure to English and Spanish, the education level of parents, the presence of the father in the household, and performance intelligence
exhibit very small partial correlation values with the dependent variable, English-84. These are then eliminated from the regression as being irrelevant to English development (1984) in and of themselves.
A final word about stepwise (forward and backward stepping) regression. Cummins hypothesis asserts that Spanish and English are interdependent variables whenever verified in a first and second language paradigm within the context of school. Additive bilingualism would exist if the first language is maintained at competence levels.
However classrooms in bilingual programs frequently contain students of diverse backgrounds. Half of them may be in the school in transit to other districts within the city, state or nation. Some display a better command of Spanish than of English (or vice versa). Others may be fair in both, English and Spanish, while still they are poor in both. Some may have good academic skills and interpersonal communicative skills, while some have only one and not the other. These children all remain and share the same classroom. They all receive the same treatment. Frequently, the teacher may switch (for only brief periods) to the English language in the all Spanish curriculum for the benefit of those who are dominant in English, or are unable to follow instructions easily in Spanish. This means that there are children who speak English (basic communicative skills) better than Spanish, though still semilingual in both languages from the academic point of view. Others (and not a small number) speak no English, but their academic Spanish is so poor that they are not able to understand the teacher, when he or she attempts to speak in academic Spanish.
During these primary years of school there is a large diversity (to say the least) in the following terms:
1. Longitudinal stability of placement of students. Children in large numbers leave and join a group or classroom.
2. BICS and CALP skills. A diversity of language skills at the communicative and academic levels is observed within each classroom. Furthermore, the curriculum has no specific academic linguistic philosophy and a clear distribution of linguistic priorities to support a long term incremental maintenance bilingual program.
No hay comentarios:
Publicar un comentario