A Vocabulary Study of Children in a Foreign Industrial Community
- Author:
Alice M. Jones, Ph.D., Clinical Psychologist
Beaver County Child Study Bureau
The Problem
This study was undertaken in the attempt to help solve a practical clinical problem. The writer is the director of a child study clinic and has referred to her for examination and study large numbers of children from foreign, industrial communities. It is her custom to include the Stanford Binet in a minimum examination schedule. In dealing with the child of foreign nativity or foreign parentage it is clearly necessary to evaluate very carefully mental age scores based on tests involving language. The particular problem approached by this study is the determination of what can be expected of the normal child of foreign parentage and industrial background when the Stanford Binet is used.
The obvious approach to this problem would be the testing of fairly large nationality groups with the complete, individual Stanford examination. Time for research through this clinic is limited, and such a large project was viewed as impractical for the present. Since in dealing with the foreign groups the chief variable factor is presumably the language factor, it seemed likely that some light might be thrown on the problem of expected performance on the Stanford Revision by the use of the vocabulary test. It was therefore decided to approach the larger problem by giving the oral, individual vocabulary test to the children of an entire elementary school system located in a largely foreign industrial community, to see what were the tentative norms obtained by such an investigation.1
The Group
The school district selected was that of Aliquippa, Pennsylvania. In Aliquippa is located a part of the Aliquippa plant of the Jones and Laughlin Steel Corporation. The vast majority of the workers of the town are employed in the steel mills.
1 That the use of the vocabulary test is justifiable for some predication of general success on the Stanford Binet is indicated in Doctor Terman’s article, “The Vocabulary Test as a Measure of Intelligence,” J. of Ed. Psychol. 1918, 9, 452-456. The implications of this article will be stressed in the discussion of the results of this study. The group under consideration consists of all the public school children in Aliquippa up to the eighth grade, who were eight years of age and over at the time of the testing. This does not include all the school children of the community; there is a flourishing parochial school. (Judging from her experience in testing parochial school children in this district, however, the writer is convinced that a similar study in the parochial school would result in lowering rather than raising norms found for foreign children in the public schools.)
The children of the group range in age from eight to sixteen years and in grade from first to eighth. In nationality they distribute as follows:
Nationality Number of cases Italian 178 Slavish 172 Mixed American 56 Serbian 35 Grannish 20 Kroatian 20 Hungarian 16 Jewish 13 Russian 11 Slovanian 10 Polish 6 Lithuanian 5 German and Austrian 4 ‘1 Kreiner ” 2 “Onvalt” 2 Roumanian 1 Tyrolean 1 Total 562 This group is polyglot in the extreme. Approximately 90 per cent of the children are of foreign parentage. Of this nonAmerican group, 70 per cent are distributed about equally between the Italian and Slavish races.
The economic situation in the town is, in general, good. Wages are adequate. Foreign living standards prevail for the most part. The church and school are the only available cultural sources. There is no attempt at formal adult Americanization work in the town itself.
Procedure The examination consisted of an oral, individual vocabulary test, using the first list of words in the Stanford Revision, standard procedure. Two examiners2 worked simultaneously during most of the study. Both examiners were Ph.D.’s in clinical psychology, and both were accustomed to the use of this test with foreign children. An analysis of the vocabulary scores was made, by age, grade and nationality. A further analysis of the successes and failures on each word of the list, also according to the age and nationality, is in preparation. All distributions include the small American group.
Results
The first distribution made was unselective as to sex and nationality, and based on the chronological age of the pupils. This table shows the score range, the quartiles, median, and probable error at each age level from eight to fifteen years. The number of cases at each age level is, of course, regrettably small but it is large enough to show tendencies.
At the right of Table I appear three other measures, made by the writer because of her interest and curiosity concerning the situation revealed. The first column in this group shows the number of eases at each age level at or exceeding the expected norm for their age (scores 20 or more at year 8, 25 or more at year 9, 30 or more at year 10, etc.). Column two in this group shows the number of cases at each age level scoring 19 or less. This score was selected since 20 is the minimum score considered in the standardization. Column three of this group shows the number of children at each age level who make a score only half their expected score or less (10 or less at year 8, 12 or less at year 9, 15 or less at year 10, etc.).
To persons accustomed to using the Stanford Binet with foreign children, a rather close scrutiny of these results is worth-while. The range of scores is at all levels very wide and peculiar in that the minimum scores attained are so extremely low. The medians are, in all cases, also, very low. At the ten year level, for instance, the median is 16, with an expected median of 30. This is the most striking case, but there is a consistent wide discrepancy 2 The author is indebted to Doctor Eebecea Learning Fieger, former director of the Beaver County Child Study Bureau, for her assistance in conducting these tests. between the expected score and the actual score. In no case does the upper quartile even approach the expected median score, while the lower quartiles are fantastically low.
The total number of children in this school who score at norm or better for their age, is eighty, or 14.9 per cent.3 This includes
Table I AGE DISTRIBUTION? UNSELECTIVE Age 8 yrs. 9 yrs. 10 yrs. 11 yrs. 12 yrs. 13 yrs. 14 yrs. 15 yrs. 16 yrs. Number Cases 95 96 71 71 78 60 47 27 4 Score Range 2-34 2-35 6-40 8-48 6-62 18-62 10-62 16-60 Upper Quartile 14 20 21 31 36 42 44 46 Lower Q10 10 12 17 22 25 28 28 Median 12 16 16 23 28.5 30 34.5 36 P. E. 4.5 7 7 8.5 Not included No. Cases at norm or better 12 15 6 14 13 12 5 3 0 No. Cases score 19 or less 83 64 45 23 14 3 6 1 0 No. Cases 50% or less expected score 33 34 29 19 14 8 9 2 1
the American group. Within the American group, thirty, or 53.5 per cent reach the expected score. Removing the American group, of the foreign group, fifty, or 10 per cent, of the children reach or exceed the expected score. Removing the American and Jewish group, leaves forty-one cases of the mixed foreign group (8.5 per cent) who reach or exceed the expected score.
Two hundred and thirty-nine cases score 19 or less. This includes children (presumably normal children) at every age level. Forty-two per cent of the entire group have a vocabulary score of less than eight years. No child in the group is less than eight years of age, and only seventeen per cent of the group is under nine years of age.
One hundred and forty-nine children, or 27 per cent of the whole group, make only half the expected score or less. * Since so small a percentage of the cases were able to achieve the expected norm, the nationalities represented were distributed. That distribution is as follows: American 30, Slavish 16, Italian 13, Jewish 9, Russian 3, Grannish 2, Kroatian 2, Hungarian, Slovanian, Lithuanian, German, each 1 case.
Table II shows the distribution of these cases according to grade in school. The score range, quartiles, median and probable Table II SCORE DISTRIBUTION?GRADE Grade No. cases Min. Max. Med. Mode 25% 75% P. E. Cases at norm. or better Cases below norm, median Age grade 1st 2nd 3rd 4th 5th 6th 7th 8th 12 48 119 133 42 73 63 62 2 4 2 6 18 10 21 28 16 21 34 35 50 48 54 62 12 14 18 28 31 31 42 10 12 None 16 None 38 None 42 14 25.5 23 27 37 x 14 16 23 32 36 36 50 x 3 4 4.5 3.24 7.5 4.5 6.5 0 1 21 3 5 18 12 47 98 130 34 65 58 44 Total 522 64
488 error are given for each grade. The median age for the grade as distributed is given for comparison with the median scores obtained. The number of cases at or exceeding the norm for the
Table III TO SHOW RELATIVE SUCCESS OF VARIOUS NATIONALITY GROUPS Nationality Group No. cases No. cases at norm or better No. cases 50% or lower Italian Slavish Mixed American Serbian Grannish Croatian Hungarian Jewish Russian Slovanian 178 172 56 35 20 20 16 13 11 10 13 16 30 1 2 2 1 9 3 1 57 42 1 14 7 10 5 0 4 1
median age for their grade is reported and also the number of cases scoring at a point 50 per cent of their expected score or less.
There appears to be very little of contradictory significance in Table II. As much vocabulary handicap is encountered (relatively) grade by grade as age by age. Table III is inserted to show the relative success of the various nationality groups as measured in terms of the number of children able to reach the expected age norm, and the number who score half or less of the expected score for their ages. Only nationalities where ten or more children were recruited are included in this table. The most superior group (although the number is so small as to make generalization dangerous) appears to be the Jewish. The poorest group is the Serbian, with only 2.8 per cent of the cases scoring at norm or better, and 40 per cent giving a score of half the expected score or less. There seems to be a slight advantage in favor of the Slavish as compared with the Italian group, although both these larger groups show an extreme retardation with reference to both measures.
Discussion
It becomes at once apparent from these tentative results that if we are to use this vocabulary test intelligently in the diagnosis of foreign children, we must build up separate norms for foreign children.
These children constitute a normal group of the type which may be found in any industrial town of the same general nationality constitution. A recent survey of the school by the Illinois tests does not show the school to be retarded with reference to the established norms on that test. The writer has worked in and through the Aliquippa schools for three years. In spite of the language handicap, these children make satisfactory progress with a course of study planned without especial regard to their special needs to meet state and county requirements. There is nothing whatever to support the contention that we are dealing here with a group of subnormal children. That they progress under their handicap might be interpreted as evidence of superior rather than inferior intelligence.
It might also be suggested, with reference to the group, that the economic and social status of the children in this community are such as to limit their vocabulary scores. The inclusion of the American group in the same school, living under similar social and economic conditions, seems to have invalidated this criticism. American and Jewish children within the group show satisfactory scores for their ages and grades.
Anyone working with the foreign child problem realizes, of course, that we must carefully weigh mental age scores in the light of the language handicap. The relation of vocabulary mastery and self-expression to intelligence ratings has been discussed at length upon many occasions. But clinicians on the job everywhere do continue to use the Stanford Revision and cannot help but be somewhat impressed by low mental ages and I.Q.’s as at least a part of the composite picture presented by a given case.
There is surprisingly little investigation of the Stanford vocabulary test alone reported in the literature to date. Dr Terman4 in 1908, has an article in justification of the vocabulary test as a test of intelligence. In this article, he meets certain objections advanced by his critics, to the use of a vocabulary test as a test of intelligence. His article has an extremely interesting and direct bearing on the present study.
His method of demonstrating the validity of the vocabulary test consisted in correlating mental age and vocabulary scores. For a miscellaneous group of six hundred thirty-one school children the coefficient of correlation between mental age and vocabulary score was plus 0.91. Doctor Terman further states in the same article, that “the probable error of a mental age based on the vocabulary test alone, is only 9.6 months in the case of school children, and that the chances are approximately six to one that such a mental age will be in error more than a year and a half.”
Correlation between vocabulary score and mental age score would, of course, justify the vocabulary test as an intelligence measure if one admits the validity of the mental age as an intelligence measure. On the other hand, from the point of view of this investigation, the high correlation which Doctor Terman discovered merely indicates the close interdependence between vocabulary and mental age, in other words, the vital importance played by vocabulary in a mental age rating on the Stanford Revision. Doctor Terman states in the same article that “Portuguese and Italian children from homes where a foreign language is spoken, are, for the first two years of school life, considerably below the median for American children of the same mental age. The difference, however, entirely disappears by the time the child has attained the mental age twelve years. (It is obviously true that the child cannot attain the mental age 12 years without an adequate vocabulary, in view of the verbal nature of the tests at year 10 and 12, especially with a correlation of pins 0.91 between mental age and vocabulary score.)’’
The author does not here wish to comment upon the value of correlation of this type in showing the validity of a part of a test as an intelligence measure.
In the light of Doctor Terman’s statements concerning the high correlation between mental age and vocabulary, it becomes inevitable that a vast number of these children who appear in the present study are foredoomed to a low mental age rating and a low I.Q. Every week the writer has occasion to examine children from one or the other of these nationality groups with findings somewhat uniformly depressing. Test ratings on these cases, rigidly interpreted, would result in diagnosis of Borderline or Moron for children who probably are not even dull, as compared with their own nationality group.
If 27 per cent of unselected mixed foreign children do not do better than half as well on vocabulary as American children of their chronological age, then we must view with extreme caution findings in terms of mental age on children recruited from foreign groups. Especially, caution should be observed about “labeling” as borderline, doubtful, or possibly defective, children who may represent an average sampling from their own linguistic group. The writer has had enough clinic experience and contacts to realize the extent to which such ratings are prevalent. Pintner,5 in an article on the comparison of American and foreign groups, shows the relative standing of foreign children and American children on the National Intelligence Tests, and on a non-language test. His results show the foreign groups uniformly inferior on the National Intelligence Tests, but the total foreign group as equal to the American group, and the Polish and German groups as superior to the American on the non-language involving tests. He concludes with a wise note that “Caution is needed in drawing conclusions as to the intelligence of foreign children when tested solely by means of tests which pre-suppose the knowledge or reading of the English language.” Just how much caution is needed is, I believe, indicated, at least, by these very significant vocabulary findings. It is to be feared that many cases have been rated as dull, borderline, and defective, and consistently treated as such in school, and by social agencies, where the controlling factor in the mental age score was the vocabulary.
Summary
Tests of 552 children in a foreign, industrial community, entire public school population, eight years of age and over, with vocabulary list I, Stanford Revision Binet Tests, oral, individual presentation shows:
1. Very wide range and extremely low medians at each age and grade level. Medians in some cases scarcely more than half expected score. (Tables 1, 2, and 3.) 2. Only 14.6 per cent of school population attain the expected score for their age or better. Of the mixed foreign groups (excluding American and Jewish cases) only 8.5 per cent reach the expected norm or better. Of the included small American group, living under identical economic and similar social conditions, 53.5 per cent attain the expected norms for their age. 3. 40 per cent Serbian, 32 per cent Italian, 24 per cent Slavish children tested attain only half the expected score or less. Of the entire group, 27 per cent fail to reach a score of more than half the expected score for their age. 4. 42 per cent of the entire school have a vocabulary score of less than eight years. No child in the group is less than eight years old, and only 17 per cent are under nine years of age. Conclusions 1. Since, in justifying the use of a vocabulary test as a test of intelligence, Dr Terman showed a correlation of plus 0.91 between vocabulary score and mental age score on the Stanford Binet, it becomes obvious that a large number of the children in this study are fore-doomed to a low mental age score. 2. The study emphasizes in a striking way the care which must be taken in the evaluation of mental age scores on children of foreign parentage. Extreme injustice may be done a foreign child in classifying him as dull, borderline, and defective, when he is, as a matter of fact, giving a performance average or better for his race and economic group. 3. If we continue the use of the Stanford Binet in a minimum psychological examination, it is imperative that we build up special norms for each foreign-language group.
Disclaimer
The historical material in this project falls into one of three categories for clearances and permissions:
Material currently under copyright, made available with a Creative Commons license chosen by the publisher.
Material that is in the public domain
Material identified by the Welcome Trust as an Orphan Work, made available with a Creative Commons Attribution-NonCommercial 4.0 International License.
While we are in the process of adding metadata to the articles, please check the article at its original source for specific copyrights.