Skip to main content

Full text of "ERIC ED066058: On the Relationship of Language Dominance and the Effects of Viewing CARRASCOLENDAS."

See other formats


ED 066 058 

EM 010 287 





Williams, Frederick; Van Wart, Geraldine 

On the Relationship of Language Dominance and the 

Effects of Viewing CARRASCOLENDAS. 

Texas Univ. , Austin. Center for Communication 

Sep 72 




MF-S0.65 UC-S3.29 

Bilingual Education; ♦Bilingual Students; Elementary 
School Students; ♦Instructional Television; *Mexican 
Americans; Spanish Speaking 


A study was made of the relationship between the 
language dominance of a child and the effects of viewing a bilingual 
television program called Carrascolendas. A previous study showed 
that the program did have an effect on average knowledge gains among 
viewers. In order to ascertain whether these gains were in some way 
related to the language dominance of the child, an index was 
constructed to determine the child's language dominance*— Spanish, 
English, or bilingual. When this index was correlated with gains made 
as a result of viewing the program, no significant evidence was found 
that the effects of viewing Carrascolendas were related to, or 
dependent upon, the child's language dominance. (JY) 

E/AOIO £8 7 ED 06605b 


1 1 





an occasional publication of the 


School of Communication / The University of Texas at Austin / Austin, Texas 78712 (512/471-1095) 


FALL 1972 





Frederick Williams 
Geraldine Van Wart 

The main focus of this report is to answer the question: What is 

the relationship between the language dominance of a child and the ef- 
fects of viewing CARRASCOLENDAS? In the report of the second year eval 
uation of the television series (Williams, McRae, and Van Wart, 1972) 
the overall effects of viewing the program were gauged in terms of a 
pre and post series testing technique where the test items reflected 
a sampling of behavioral objectives of the program. The tests were ad- 
ministered to three groups: (1) children who viewed the program and 

who engaged in supplementary class activities, (2) children who viewed 
the program but had no supplementary class activities, and (3) a group 
of comparable children who did not view the program, nor have the ac- 
tivities. Overall gains — that is, differences between post and pre 
test scores — were indexed in two ways. One score reflected frequencies 
of items that were correct or partly correct as compared with items 
which were not correct. A second scoring procedure differentiated re- 
sponses by means of a four step scale ranging from totally incorrect 
("1"), partly correct ("2"), totally correct but prompted ("3” ), and 
altogether perfect ("4"). Analyses of both frequency of correct items 
and scale measures indicated that the viewer groups generally made 
greater gains on test items than did the nonviewer groups. These gains 
were greater in English test items than they were for the Spanish ones . 
Further, particularly on English items, viewers who had supplementary 
activities experienced greater gains than viewers who did not have ac- 
tivities, although both groups were in most cases greater than groups 
of children who were nonviewers. In more general terms, these results 
were interpreted as indicating that CARRASCOLENDAS did have an effect 
upon average knowledge gains and that supplementary activities some- 
what facilitated these gains. Further details on these results and 
analyses of subtests may be found in the main report (Williams, et al. , 
1972) , 


- 2 - 

A practical question, and one not researched in the main report, 
was whether these gains were related to, or affected in some way by, the 
language dominance of the children in the experiment. In answering 
this question, it should be recalled at the outset that the children 
were, as much as possible, randomly assigned to the three different 
groups in the experiment. Thus it was not likely that any differences 
among the control and viewer groups would be due to one group having 
more children dominant in one language or another. As was anticipated 
in the design of the experiment, the children in the three groups are 
comparable with the exception of the treatment — viz., viewing, view- 
ing plus activities, not viewing. 

We, of course, must have some index in order to consider language 
dominance. One such index was the teacher's rating of a child as 
either being English dominant, Spanish dominant, or bilingual. Another 
index of language dominance was based upon f ieldworkers ' ratings of the 
children's fluency in responding to questions which required continuous 
responses. As described in more detail in the main report, some ques- 
tions were meant to elicit a continuous response from the child, that 
is, the question could not be answered v;ith one word or a short phrase. 

On these, the fieldworker scored the child on a one-to-four fluency scale. 

If these fluency ratings had a useful degree of validity, we would 
expect that they would agree to a reasonable degree with the teachers' 
division of the children into different language dominant groups. Table 
1 presents the summary comparison on this point. It is clear from this 
summary, that children who were classified as being Spanish dominant 
had a higher average fluency score in Spanish items (2.00) as compared 
to their fluency scores when responding to English items (1.45). As 
would be expected, the reverse is true for the children who are rated 
as English dominant by the teachers. Their fluency score on English 
items (2.26) exceeds on the average their fluency score on Spanish items 
(0.72). It would be anticipated that children who classified as bilin- 
gual could be more nearly equal in terms of average fluency in both lan- 
guages, and the results in Table 1 point to this. The conclusion here, 
then, is that the 'fluency scores have a reasonable degree of validity 

1-Based on the senior researcher’s considerable experience with 
speech ratings (Williams, Whitehead, and Hiller, 1971) it is not diffi- 
cult to assume that gross ratings of fluency, particularly on only a 
four-step scale, can be reasonably obtained. Although it is not for- 
mally described in the report, the fieldworkers were practiced in the 
use of the scale and ratings were compared between them as well as 
with other members of the research staff. In short,, these ratings 
are assumed to be reliable. Their validity is subsequently discussed 
in this report. 


Table 1 

Average Fluency Ratings in 
Responses to Spanish and English 
Items as Compared by Teachers * 
Classification of the Students 


Spanish items 
English items 

Children* s 


1. 90a* 



English Spanish 

0.72c 2.00a 

2.26a 1.45b 

*Means with common subscripts 
are not significantly (£<.05) 
different from one another. 


as paired with the teacher's classification of the children into the 
different groups. In subsequent analyses aimed at answering the ques- 
tion of the relationship of language dominance to the effects of 
CARRASCO LENDAS , both of these indexes of language dominance will be 
employed . 2 

Since the present question is one of relationship, the most ap- 
plicable statistical model is one of correlation. This will index 
the degree to which the gain scores in either the Spanish or English 
items are related to measures of language dominance. At the same 
time, this approach will also allow us to inquire how these relations 
compare with other relevant factors, such as whether or not a child 
viewed the program or whether or not he had supplementary activities. 

Table 2 presents a series of correlation coefficients which re- 
presents indexes of the relationship of the language dominance mea- 
sures with gain scores (1-4 scale, on the Spanish and English test 
items.). The table also includes similar indexes of relationships 
of whether or not the child viewed a program and whether or not he 
engaged in supplementary activities. 

Let us first consider items having a relationship with gains in 
the Spanish testing area. There is only one item which stands out as 
having anything more than a trivial relationship with gains in the 
Spanish testing area, and this is the variable of whether the child 
did or did not view the program. A correlation of .19 in this case 
would only be expected to occur by chance in 1 out of 10 cases of 
random sampling. We can have some confidence that it is not just a 
random occurrence although it is a very small correlation. At best 
the conclusion here is that the gains in the Spanish test area (which 
were small in the first place) are somewhat related in a correlational 
sense to whether or not the child viewed the program. At the same 
time, we can note that there is no relationship between gains in the 
Spanish test area and either of the language measures or the activity 
variable . 

The most encompassing generalization here is that if gains in the 
Spanish area were to be predicted from anything, it would be upon the 
basis of viewing or not viewing the program. This generalization can 
be supported in a more definitive manner by use of a prediction equa- 
tion which indexes the degree to which variables of viewing, activi- 

20f course, it would be possible to employ more detailed language 
measures in future research, but such measures require time and expense, 
and should not begin to mask the main purpose of this research. 


ties, and language dominance jointly predict gain scores in the Span- 
ish testing area. When the equation was calculated, it indicated that 
the index of multiple correlation of all these variables upon gains in 
the Spanish area was +.2341. Roughly 86% of the prediction in this 
equation was due to the variable of whether or not the child watched 
the program. It is very clear, then, that gains here are independent 
of language dominance measures. 

Results of similar analyses for the gains on the English test 
items can be interpreted using the same strategies. First, as can be 
noted in Table 2, there are several marked correlations of the various 
variables with gains in the English test items. There are interpret- 
able correlations, for example, of whether or not the child viewed the 
program (.49) and whether or not he also engaged in supplementary ac- 
tivity (.63), both with gains in the English scoring area. Neither 
of these correlations has over a 1 in a 100 chance of occuring ran- 
domly. They are indexes of relationship in which we can have a sub- 
stantial degree of confidence. At the same time, however, these are 
the only two variables which have any major observable relationship 
with gains in the English area. 

It can be noted in passing that there is a small negative corre- 
lation in the teachers' rating of a child as not being English domi- 
nant and gains in the English items. Similarly there is a very small 
positive correlation between a child's average ratings in Spanish flu- 
ency and gains on the English test items. Both of these correlations 
are too small in magnitude to have anything more than a passing inter- 
est in. But they do suggest that children who may be less adept in 
English than others, may be among ;he ones who gain in the English 
items as a result of viewing CARRASCOLENDAS . 

As in the analysis discussed earlier, it was further possible to 
see how the variables of language dominance, viewing, and activities, 
jointly predict gains in the English test items. An appropriate equa- 
tion was calculated. It indicated that there was a multiple correla- 
tion of these variables with English test items of a magnitude of +.687. 
This is a correlation of substantial magnitude in this type of situation, 
and it indicated scores in the English test items are substantially pre- 
dictable. One basis for prediction is whether a child viewed the pro- 
gram, and this contributed to about 28% of the predictability in the 
equation. The second is the effects of viewing combined with activities 
which contributes to about 68% of the prediction of gain scores. None 
of the measures of language dominance contributes significantly to this 
equation. This is strong evidence that the gains in the English lan- 
guage area, particularly as predicted by the two variables just dis- 
cussed, are independent of language dominance. 

The overall conclusion, then, of this series of analyses is that 
based upon the measures incorporated in this study, no salient evidence 

Table 2 

Relations of Viewing, Activities 
and Language Measures* with 
Gain Scores in Spanish 
and English Items 

Gain Scores 





Viewing : 






Teacher classification 




English dominant: 



Spanish dominant: 










•Variables such 

as group and teacher 


fication are entered in a 

binary code (l=yes, 0 

=no) . 


can be found that effects of viewing CARRASC0LENDA3 , either in the area 
of Spanish test items or English test items, are related to, or depen- 
dent upon, the child's language dominance. Of course, this does not 
preclude situations where such an effect c ould be the case. It only 
says that in a school involving children of various degrees of English 
as against Spanish dominance, as_ well as bilingual capabilities , effects 
of the program do not seem to interact in any important way with lan- 
guage dominance. Since this is considered to be an important as well 
as practical finding of the study, attempts will be made in the larger 
field study conducted in 1972-1973 to include indexes of the children’s 
language capabilities, and to assess the relationship of these with 
gains due to viewing CARRASCOLENDAS . 



Williams, F. , Van Wart, G. , and McRae, S. CARRASCOLENDAS : 
Effects of a Spanish/English television series for 
primary school children. Grant No. OEG-09-530094- 
4239 (280). U.S. Department of Health, Education 

and Welfare, Office of Education, 1972. 

Williams, F. , Whitehead, J. L. , and Miller, L. Attitudinal 
correlates of children's speech characteristics. 
Project No. 0-0336, Grant No. OEG-0-70-7868 (508). 

U.S. Department of Health, Education, and Welfare, 
Office of Education, Bureau of Research, 1971.