University of Southern California Dissertations and Theses

Predicting Success In The Study Of Descriptive Linguistics

(USC Thesis Other)

Predicting Success In The Study Of Descriptive Linguistics

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content This dissertation has been , . . . 66-11,583 m icrofilm ed exactly as received RENFER, Mary Emma Fewell, 1914- PREDICTING SUCCESS IN THE STUDY OF DESCRIPTIVE LINGUISTICS, University of Southern California, Ph.D., 1966 Education, psychology University Microfilms, Inc., Ann Arbor, Michigan (©Copyright hy Mary Emma Pewell Renfer 1966 PREDICTING SUCCESS IN THE STUDY OF DESCRIPTIVE LINGUISTICS by Mary Emma Fewell Renfer A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY {Educational Psychology) June 1966 UNIVERSITY OF SOUTHERN CALIFORNIA TH E GRADUATE SC H O O L U N IV ER SITY PARK LO S A N G ELES. C A L IFO R N IA 9 0 0 0 7 This dissertation, written by ........ under the direction of h.ft7?...Dissertation Com mittee, and approved by all its members, has been presented to and accepted by the Graduate School, in partial fulfillment of requirements for the degree of D O C T O R OF P H I L O S O P H Y Dean Date J y n e . 1 2 6 6 DISSERTATION COM M ITTEE j L l . . V TABLE OF CONTENTS Chapter Page LIST OF T A BLES.............................. iv I. THE PROBLEM................................... 1 Introduction Statement of the Problem Importance of the Study Nature of the Design Development of Hypotheses Statement of the Hypotheses Scope and Limitations Organization of the Study II. REVIEW OF THE LITERATURE................... 28 Introduction Cognitive Elements of Success The Modern Language Aptitude Test Noncognitive Elements of Success Summary III. METHOD OF PROCEDURE.......................... 48 Introduction Sample Predictor Variables Criterion Measures Procedure Summary IV. ANALYSIS OF DATA AND FINDINGS............... 73 Simple Data Description Intercorrelations of the Independent Variables Intercorrelations of the Dependent Variables Validity Coefficients Regression Analysis Summary ii Chapter V. SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS . Introduction Summary Summary of Procedures Summary of Findings Relative to Hypotheses Conclusions and Recommendations Page 112 BIBLIOGRAPHY 141 51 75 76 78 83 85 86 88 90 91 92 95 LIST OF TABLES Sex, Age, and Previous Language Training for the Four Schools and Total Sample ......... Measures of Central Tendency and Variability of the Independent Variables ............... Means and Standard Deviations on the MLAT for Two Previous Researches and the Present Study Correlations Between the Biographical Items and the Other Independent Variables .... Correlations Between Kuder Interests and Cognitive Variables ........................ Intercorrelation Matrix for the Kuder Interest Variables ................................... Intercorrelational Matrix for the Cognitive Independent Variables ...................... Intercorrelation Matrix for the Dependent Variables .................................. Correlation Coefficients Obtained Between the Criteria and the Biographical Items .... Correlation Coefficients Obtained Between Kuder Interest Scores and the Criteria . . . Correlation Coefficients Obtained Between Cognitive Variables and the Criteria .... Correlation Coefficients Obtained Between the MLAT Short Form, MLAT Total, and Henmon- Nelson Total and the Two Major Criteria . . Standard Score Regression Coefficients, and t Ratios Indicating Their Significance, for Each Independent Variable Used in Predicting iv Table the Two Major Criterion Variables 14. Multiple Correlation Coefficients and the Corresponding Coefficients of Determination for Predicting the Two Major Criterion Vari ables by Means of Selected Independent Variables ................................... 15. Standard Score Regression Coefficients, and t Ratios Indicating Their Significance, for Each Independent Variable in the Complete Battery Which Attained Significance Level When Predicting the Two Major Criterion Variables from the Complete Battery .... 16. Standard Score Regression Coefficients, and t Ratios Indicating Their Significance, for Each Variable Attaining Significance in Regression Studies Employing the Kuder, the MLAT Total, and the Henmon-Nelson Total to Predict the Two Major Criteria ............. 17. Correlation Coefficients and Variance Statis tics Obtained When Predicting Individual Course Grades from the MLAT, the MLAT Short Form, and from the Complete Battery .... 18. Standard Score Regression Coefficients, and t Ratios Indicating Their Significance, Obtained When Predicting Individual Course Grades from the Complete Battery ........... Page 97 98 101 103 105 107 v CHAPTER I THE PROBLEM Introduction In this relatively literate twentieth century there exist hundreds of thousands of preliterate people speaking tribal languages for which no written form exists. The problem of communication generated by this fact is of grave concern to the organizations and nations involved with these peoples. Through the years many techniques for the study of languages have evolved, but only recently have attention and interest been focused on a method of approach that tackles the problem in terms of internal language structure (39). Descriptive linguistics is the discipline that studies the various aspects of human speech, thus providing a broader perspective for all language programs and a basis for a more efficient and scientific approach to a specific language. A student in this discipline acquires the lin guistic skills necessary for reducing unwritten languages to writing, which, when applied, eventuate in the possibil ity that people of preliterate tribes have the opportunity of becoming literate. The modern world with its irrepressible demands for effective communication has pressed the science of descrip tive linguistics to the forefront for the solution of its problem. The enormity and expanse of the frontiers of the world's linguistic task may be seen in a brief citation of statistics. Welmers (102), an authority on African lan guages, has written of the plight of the continent, which has witnessed the birth of more than forty new nations in recent years. He stated that, besides the innumerable dialects spoken in Africa, there are about one thousand actual languages— mutually unintelligible speech forms. The great majority of these languages demand linguistic study. A number of the tribal groups have populations num bering in the hundreds of thousands. Nigeria, one of the most progressive of the countries, has about 250 languages, most of which need study. One of India's many problems is national solidar ity. It is a country faced with a complex of ethnic groups and languages which present probably the world's most stag gering single national linguistic problem. The 1951 census 3 enumerated a total of 84 5 languages or dialects. Included were 742 Indian languages, each spoken by more than 100,000 people, and 63 non-Indian languages (107). Estimates as late as 1957 brought the language figure for New Guinea (Australian New Guinea and West Irian) alone to the neighborhood of 700. Although a large number of the languages are spoken by but a handful of people, the world's second largest island has assumed stra tegic importance in the world of linguistics as well as politics (107). Recent studies for Latin America suggest the need for linguists by noting the existence of more than six hundred languages (11). In neighboring, progressive Mexico approximately 38 per cent of the population is illiterate (6), and 10 per cent of the total population, or roughly 3,5 00,000 people, speak an estimated one hundred languages (11) . Why the various nations concerned do not dispatch teams of teachers specifically trained to teach their pre literate peoples their national language is a question that might reasonably be raised. The answer is perhaps best expressed by Theodore Andersson in writing for the Texas Foreign Language Association Bulletin. He stated that in learning a language, "understanding precedes speaking . . . audiolingual skills precede the literate skills of reading and writing, and that the learning of the mother tongue precedes the learning of a second language" (3). It appears that basic to teaching a national language to pre literate people is the establishment of literate communica tion through the medium of their own language. Implied in the process through which the mother tongue becomes the cultural bridge to the national language is language reduc tion, including linguistic and grammatical analysis, the construction of primers and other linguistic materials, and, finally, the actual instruction of the people in reading and writing their own language. The rising tide of nationalism has highlighted the demand for specialists in descriptive linguistics who are trained in the techniques required for engaging in a lin- guistic-literacy program of this nature. Educational institutions and organizations are involved in the task of training students in this highly technical and scientific field, but few, if any, proven guidelines exist that may be employed for selection, placement, guidance, or diag nostic help for those interested in the study of descrip tive linguistics. Increased efficiency in the employment of time and manpower could possibly be effected were more known about the nature of the factors that contribute to success in this type of study. Statement of the Problem The problem investigated in this research project was that of determining from certain selected independent variables those which are most highly predictive of success in the study of descriptive linguistics, the discipline that undertakes the study of language in terms of its internal structure. It may be virtually tautological to assert that certain basic similarities underlie all phases of linguistic study so that what is pertinent to modern foreign language learning should also have bearing upon the study of descriptive linguistics. Earlier in the present century many educational psychologists viewed foreign language aptitude as a func tion of general intelligence measurable by ordinary apti tude tests rather than as a specific ability amenable to unique assessment (52). Since World War II, however, and due in large part to the careful research of John B. Carroll of Harvard University with Stanley M. Sapon of Ohio State University (21), several more or less independent 6 factors in the realm of language learning have been identi fied through use of factor analysis. Thus, rather than seeing language aptitude largely as a function of general intelligence, Carroll considers it "a fairly specialized talent (or group of talents), relatively independent of those traits ordinarily included under 'intelligence'" (24). He also set forth the proposition "that a relatively small fraction of the general population seems to have enough of this talent to be worth subjecting to the rigor ous, intensive, expensive training programs in foreign language operated by military and governmental organiza tions, or by such private organizations as missionary societies, businesses, and industries engaged in overseas operations" (24). As a means of verifying the hypothesis that lan guage aptitude is a "fairly specialized talent," several validation studies for predicting success in learning modern foreign languages have been conducted (26). It has not yet been determined, however, to what extent the dimen sions relevant to success with written languages are similar to those factors underlying success in descriptive linguistics. Institutions and organizations interested in train 7 ing and assigning personnel for this type work are cur rently using tests which have been shown to possess some value for indicating aptitude for learning a modern foreign language without the assurance that the instruments are of like value for "pioneer" linguistic work (14,35). To date, no published materials are available to serve as guidelines in this respect. With the hope of supplying some of these guidelines, this investigation was undertaken in 1964 and 1965 at the schools conducted by the Summer Institute of Linguistics in affiliation with the Universities of Okla homa, Washington, and North Dakota. Importance of the Study A measure of the importance of this study may be derived from the nature and purpose of the organization with the cooperation of whose offices this investigation has been undertaken. The Summer Institute of Linguistics is a scientific and educational corporation which has in progress the broadest network of linguistic research being undertaken by any private organization in the world today (89). D. H. Hymes of Harvard University has described the organization as making "perhaps the single most important contribution 8 to the needs of linguistic science today" (57). The Institute currently has over eighteen hundred members drawn from the United States, Canada, Great Britain, Germany, Switzerland, Australia, and New Zealand. Presently, it is engaged in the linguistic investigation of nearly four hundred of the world's hitherto unwritten lan guages. Among these are thirty-four in Peru, thirty-five in Brazil, eighteen in Colombia, seven in Ecuador, fourteen in Bolivia, fourteen in Central America, eighty-seven in Mexico, thirty-three in North America, seventeen in West Africa, seventeen in Viet Nam, forty-one in the Philip pines, seven in Australia, and sixty-six in New Guinea. In addition to its extensive field work, the Insti tute conducts a program of academic instruction in the field of descriptive linguistics within the framework of seven linguistic institutes around the world. Courses are offered in descriptive linguistics, anthropology, literacy, and related subjects. The schools abroad are located in England, Germany, New Zealand, and Australia. The three Institutes in the United States are affiliated with the Universities of Oklahoma, Washington, and North Dakota. Over five hundred students yearly receive linguistic train ing in courses for which university credit is offered. In some countries, affiliation exists apart from the operation of linguistic institutes as such. For example, cooperative linguistic research agreements are maintained with the University of Manila; the University of Nigeria, Enugu Campus; the University of Brasilia; and Daccan College of Poona, India. Within the context of this relationship, the Uni versity of Nigeria recently gave the Institute an eleven- acre site on its Enugu Campus for the erection of the Linguistic Research Center specializing in West African languages. A contract with Daccan College, a graduate uni versity, signed as recently as 1965, marked the beginning of a program of collaboration with India's scholars, assisting them with that country's language problem. Based upon its association with the University of Brasilia, Institute personnel have formed the teaching core as professors for that university's full-scale linguistics program leading to the Ph.D. degree in the field. The program arose as a part of the planning of Brazil's Federal Council on Education which called for "intensive courses in applied linguistics for Brazilian professors who may be called upon to hold chairs of linguistics in the eighty-odd teachers colleges scattered all over the country" (40). 10 The Summer Institute of Linguistics maintains quality control of both its field research and linguistic instruction through a hard professional core of some thirty Ph.D.'s in linguistics and related subjects. The instruc tors and professors listed in the Institute's annual catalogs come directly from their field assignments to teach the summer courses. The fruit of thirty years of the Institute's research is available in a published Bibliog raphy , which contains nearly 2,000 entries authored by its members (89). Among the listings are 101 technical lin guistic volumes and monographs, 1,008 educational items, over 500 professional articles, and a volume of sociologi cal and anthropological articles. The latest edition, February 1964, reflects the greatly accelerated production of recent years. The bibliographic entries for the past five years equal 50 per cent of the total publications of the first twenty-five years of the organization's history. The Summer Institute of Linguistics enters a country only upon government invitation and generally works under contract with the respective departments of educa tion, culture, Indian affairs, or universities. Since its linguistics investigation is directed toward preliterate cultures, the Institute most frequently penetrates into 11 tribal groups caught in the backwash of modern national civilizations. Within this operational context, the organization seeks to bring a full spectrum of civilizing influences and services. Kenneth L. Pike, president of this scientific and educational corporation, insists that the Institute's purposes press beyond the confines of technical knowledge to embrace a concept of the total man— intellectual, physical, and spiritual. As Pike has noted, the Institute: . . . cooperates with various government agencies in small-scale agricultural projects, for example, to contribute toward the economic welfare of these peoples. On the moral front, it tries to bring to bear its technical competencies by cooperating with the Wycliffe Bible Translators, who are working toward giving the vernacular groups parts of the Bible in their own language. It is believed that nonsectarian versions of this Book are the single most potent force which can be made available to give these folk access to values and power for liv ing which have infused our own western civilization for good. (89) In pursuit of its scientific and ancillary objec tives the Institute and the work of its personnel have come under scrutiny from a number of quarters. Several repre sentative comments from these sources will, in a measure, serve as an assessment of the contemporary relevance of the scientific, educational, and cultural impact of the Insti tute' s work. 12 Working under the Australian Minister of Terri tories, G. T. Roscoe, Director of Education in Papua and New Guinea, recorded his evaluation of the work of the Institute: It is my considered opinion not only that the Summer Institute of Linguistics has . . . already rendered valuable service to this Administration in its special field, but also that its potential usefulness is much greater than can at present be foreseen. It is a great encouragement to me to know that there are scientifically trained workers in the Territory to whom I can turn for expert advice on matters connected with the very difficult subject of vernacular languages. I am charged by the Minister of Territories with the responsibility of promoting Universal Primary Education and Uni versal Literacy in English, and I regard the tech nical advice of the S.I.L. as a very valuable help in fulfilling this responsibility. . . . It is the policy of the organization to cooperate closely with the Government of every country in which it works and this is the case in the Territory of Papua and New Guinea. (84) The heads of state of several countries have com mented on the contribution made by this organization. Diosdada Macapagal, the former President of the Philip pines, has written of the comprehensive nature of the service rendered to his country: We should be thankful to the Summer Institute of Linguistics for the great and important service - it has been rendering in the integration of the minority groups into the mainstream of the country's cultural life. . . . The Institute is now working in 40 dialect groups extending from the Ivatans of the far north to the Tausugs of Jolo. The 13 introduction of literature in indigenous dialects naturally stimulates these people to learn to read and write, and as the process continues, it is expected that there will result an instinctive de sire on the part of these minority groups to know more of their own people . . . thus hastening their integration into the national life of the country. What is more praiseworthy, however, in the Institute is that each member is required to learn the dialect of the region to which he is assigned. Indeed, this is the only way they really can help understand the natives and hasten their integration, for the spoken language is the fastest means of communication and therefore of understanding and friendship. (65) Jose Maria Velasco Ibarra, while President of the Republic of Ecuador, evaluated the work of the Institute in his country both in terms of its "civilizing" influence and its contribution to international good will. I earnestly hope you may have complete success in your civilizing labors. Yours is the way to bring about closer relationships between the United States and Latin America. Your efforts to get peoples to understand each other, drawing them to gether culturally, will develop indestructible ties on this continent. (97) Outside observers also have noted the contribution of the Summer Institute of Linguistics in providing pre literate tribal peoples with a basis of literate self- respect with which to meet the onslaught of modern civili zation, thus enabling them less traumatically to enter as useful citizens into the national life of their respective countries. 14 After a visit to the Peruvian selva, where in the past twenty years the Institute has reduced thirty-two tribal languages to writing and, in cooperation with the Ministry of Education, conducts its most extensive pioneer program of bilingual education, the renowned British his torian Arnold Toynbee commented with respect to the trans- cultural features of the work: Thanks to your work for the Indians, the terrific impact of modern material civilization upon their life is being eased, so that their encounter with the modern world may perhaps have a happier ending. (95) Kenneth F. Weaver, a senior staff member of the National Geographic Magazine, in writing of "The Five Worlds of Peru," came to a similar conclusion. In treating the Amazonian Jungle "world" of that land, he wrote: One major influence— a nondenominational U.S. missionary group known as the Summer Institute of Linguistics— is helping prepare the Indians for the severe adjustment they face as roads and planes threaten their isolation. (99) In Farewell to Eden, an anthropological monograph based upon an investigation of the culture of the Amahuaca tribe of the Peruvian Amazon, Matthew Huxley ventured a comparative evaluation of approaches to the problem con fronting isolated, preliterate tribal groups as they face the inexorable impact of modern civilization, carrying with 15 it a potential for their cultural extinction or their suc cessful national integration. The conclusion of his thesis expressed the conviction that the "survival— certainly cultural survival— of primitive people will depend largely upon the efforts of groups like the S. I. L. whose motivat ing force drives them to inhospitable and frequently unrewarding areas of the world" (56). Comments from observers such as Huxley and Toynbee reflect the strategic character of the organization with whose cooperation the present project has been undertaken. From this association the study may be viewed as deriving a measure of its importance as it may facilitate more effec tive assessment and screening of personnel for the pro jected expansion of both the academic and field operations of the Institute. An added dimension of the value of the study arises from the fact that no predictive study in the field of descriptive linguistics has previously been under taken that would permit inferences to be made with respect to linguistic aptitude in general. Nature of the Design The objective of this project was to ascertain which, among the various factors measured, contributed most 16 to success in the study of descriptive linguistics. The study thus resolved itself into a predictive evaluation of certain basic information derived from a battery of tests and several biographical items. Three standardized batteries and three biographical items provided the independent variables, while staff evaluations and course grades served as the criterion variables. Cognitive factors were measured by the Henmon- Nelson Intelligence Test and the Modern Language Aptitude Test. The noncognitive elements were tapped by the biographical items and the Kuder Vocational Interest Pref erence Record. A definitive study of the relationship of each variable to each of the others was accomplished by means of intercorrelational tables. The optimal predictive capacity of the entire battery was derived from regression studies. Multiple correlation coefficients were also obtained from a reduced number of predictors used in vari ous combinations for predicting the criteria. The validity of assuming that predictive variables reflect reliable information on qualities that make for success in a particular discipline rests upon the basic concepts embraced by modern symbolic logicians. In his search for meaning the logician sets forth predictive 17 evaluation as one fundamental criterion of truth (76). This philosophical concept provides a foundation for assum ing that if certain variables possess predictive validity relative to linguistic studies, then an understanding of these variables might throw light on the nature of linguis tic aptitude. Development of Hypotheses A consideration of the variables involved in the present design gave rise to certain fundamental questions for which answers were sought. Correlational study. For the current project the Henmon-Nelson Intelligence Test, the MLAT, the Kuder Voca tional Interest Preference Record, and the biographical data yielded twenty-two independent variables for study against the six dependent variables of staff evaluations and course grades. An intelligence test such as the Henmon-Nelson provided a measure of verbal knowledge and reasoning ability, two qualities assumed to be critical to success in any academic field. Abilities perhaps unique to success in learning a foreign language were tapped by the MLAT. The cognitive factors assessed by these two bat teries could be expected to bear significant weight in 18 predicting success in the field of descriptive linguistics. The rationale supporting the inclusion of an inter est inventory in a predictive battery lay not so much in the intrinsic capacity of such a measure to predict success as it did in its "additive" qualities. Numerous studies have shown that an interest test included in a predictive battery has produced significantly higher coefficients than it was possible to obtain with cognitive measures alone (43,82,7). Besides the cognitive and noncognitive variables mentioned, certain factors of a biographical nature pre sented themselves as possible determinants of differential functioning in the study of linguistics. Age, sex, and previous language study were included as predictor vari ables. Considering the three types of variables, the ques tion first to be answered by this study was: Can the Henmon-Nelson, the MLAT, the Kuder, and selected biograph ical data effectively predict success in the study of descriptive linguistics? Language aptitude. For a number of years general intelligence tests seemed to provide as efficient a basis for predicting success in foreign languages as did the 19 specially designed prognosis tests. In 1931 Kaulfers (59) reported that in the previous thirty years only fifty-one studies had been conducted along this vein and that he himself believed that no distinct language talent could be differentiated from any other. He expressed his opinion of language aptitude testing by declaring that the picture was "totally muddled." However, after World Waf II the press ing need for linguists led Carroll and Sapon in a concerted effort to learn more about the nature of foreign language aptitude. The research, based upon factor analysis stud ies, eventuated in the construction of a new type foreign language prognosis test. This test, the MLAT, published in 1960, has been a fundamental factor in the improved effi ciency of the screening program serving the Defense Lan guage School and the Foreign Service Institute (24). The primary consideration in these schools is the intensive training of students in speaking and understanding modern foreign languages. Descriptive linguistics, on the other hand, stresses basic internal language structure; but implicit in this is the ability also to hear, understand, and reproduce words and expressions totally foreign. The differences in emphasis between the two aspects of linguis tics work gave rise to the question: Does the MLAT 20 contribute uniquely to the prediction of success in descriptive linguistic studies? Quantitative score. Some studies have shown the scores on the Numerical section of an aptitude test to be better predictors of marks in foreign language courses than the scores on the Verbal section (50). Brown found the Quantitative scores on the College Board test to be more predictive of achievement in languages than of success in any other discipline (9). These findings suggested another factor to be examined in this project: Is the Quantitative score on the Henmon-Nelson intelligence test more predic tive of achievement in descriptive linguistics than the Verbal score? Interest inventory. Cronbach (29) has stated that an interest test such as the Kuder or the Strong measures a "factor of undeniable significance in school success." The suggestion was also made by Anastasi (2) that some measure of interest should be included in every predictive battery. Super (92) apparently embraced the same belief for he set forth the rationale for such an inclusion when he related interest to motivation. Motivation is unques tionably a major factor in academic success, but there are no instruments proven valid for its assessment. In lieu of 21 such a measure, an interest test has much to offer by tap ping one aspect of that many-faceted phenomenon. This fact, viewed in the light of the knowledge that the present research involved a sample having a fairly definite voca tional orientation, provided the basis for the following question: Does the Kuder Preference Record— Vocational make a significant contribution to prediction in this study? Age. Age has often been viewed as being particu larly relevant to achievement in foreign languages. No research, however, has established such a relationship. Most of the studies pertaining to this subject have been conducted with samples taken from an ordinary school set ting in which age differences were not of sufficient magni tude validly to test hypotheses relating to the function of age upon performance in foreign language learning. The sample involved in the current project differed in this regard; its wide variability in age afforded the opportun ity to study the effect of age upon the learning of descriptive linguistics. The question considered was: Is age a significant factor to success in the study of descriptive linguistics? Sex. When any aspect of speech development or 22 language learning is being examined, the question of sex differences in performance is generally raised. In verbal linguistic functioning, females have tended to evidence superiority over males throughout their life span, but this has been more with respect to verbal fluency than with verbal reasoning (18). Grades in foreign language courses have also favored the female (1). Since, however, descrip tive linguistics places more emphasis on reasoning, both verbal and quantitative, than on fluency itself, sex dif ferences in performance might be minimized. This line of reasoning led to the formulation of the following question for study: Is sex a significant factor in achievement in descriptive linguistics study? Previous language training. The effect of previous foreign language training upon the learning of a new lan guage is a study in itself and has given rise to some difficult questions. A correlation between success with a new language and previous language training might simply be the result of self-selection. The individual who is interested in foreign languages and successful in tackling them is likely to continue that which for him has been a rewarding experience; whereas, the one who has not been so successful will likely avoid that which is not for him a 23 pleasant experience. The correlation between previous language training and success with another may be, on the other hand, more in the nature of a function of transfer effect. The learning of a successful method of approach to one language may be of help in studying another. Rele vant to the present study was the possible relationship between amount of previous experience with foreign lan guages and success in descriptive linguistics— whether due to transfer effect, natural selective forces, or other factors. The question raised for examination was: Does the amount of previous foreign language training figure significantly in the degiree of achievement attained in the study of descriptive linguistics? Statement of the Hypotheses The brief consideration given to the seven ques tions just cited affords an introduction to the rationale out of which grew the hypotheses which were tested in this research study. 1. A battery comprised of biographical items, a vocational interest inventory, an instrument for measuring foreign language aptitude, and an intelligence test will produce significantly high multiple correlation coefficients when used to predict success in descriptive lin guistics, measured by either course grades or staff ratings. The MLAT will contribute more predictive power to the regression formula than will any other single battery. There will be no difference in the predictive power of the V score and the Q score of the Henmon-Nelson Intelligence Test. The ten scales of the Kuder Preference Record— Vocational will not contribute to the predictive power of the battery. Age will not correlate with achievement in descriptive linguistics. Sex will not correlate with success in linguis tic study. Amount of previous language training will not correlate with achievement in the study of descriptive linguistics. Scope and Limitations Scope. This investigation involved only the first 25 year linguistic students enrolled for the intensive summer courses offered by the Summer Institute of Linguistics in affiliation with the Universities of Oklahoma, Washington, and North Dakota. All of the students enrolled on the campuses during the summer of 1965 were included in the initial sample. In addition, the 1964 enrollees on the Norman, Oklahoma, campus were included. For the present project the objective was to assemble a battery which will be efficient in predicting success in the study of descriptive linguistics, the dis cipline which approaches the study of languages from their internal structure. Projected plans are to extend the present undertaking to embrace a longitudinal study in order to ascertain the effectiveness of the present battery in predicting success in the actual employment of the ac quired linguistics skills in field situations involving unwritten languages. Limitations. Certain limitations are implicit in any research design. Of particular note in the present study was the element of time limitation. Because of the crowded schedule of the normal Summer Institute registra tion routine, some of the tests that might have been included in the battery had to be omitted. During the 26 siommer of 1964 several of Guilford's experimental tests, presumed to measure some facet of inductive reasoning, were administered. In 1965, however, these were eliminated from the battery in order not to exceed the allotted three-hour test period. Organization of the Study Chapter II contains a review of the literature relevant to achievement in the study of descriptive lin guistics. Following a cursory survey of the literature dealing with both cognitive and noncognitive elements in academic success, a more definitive review of the research pertaining to foreign language aptitude is undertaken. Chapter III, which is a discussion of the design and methodology of the current project, defines the nature of the sample and presents a detailed analysis of the research instruments as well as of the criterion variables. Also included is a description of the statistical methodol ogy employed. Chapter IV presents the analysis of the findings resultant from the study. An examination of the simple data description, intercorrelations of the variables, validity coefficients, and multiple correlation coeffi cients, and multiple correlation coefficients is undertaken. 27 Chapter V is concerned with a summary of the project and a discussion of the findings with particular reference to the hypotheses set forth in Chapter I. Also presented are the conclusions that might logically be drawn from the results of this research. Recommendations are outlined, and suggestions for further research, which have been highlighted by the outcome of the present study, are made. CHAPTER II REVIEW OF THE LITERATURE Introduction In a consideration of the literature related to the present research problem, it must be pointed out that the discipline known as descriptive linguistics is still in its youth. Only in the past decade or so have world conditions forced upon psychologists, sociologists, anthropologists, communications engineers, political figures, and mission organization heads the recognition of the need for a science that will facilitate more efficient communication with the multiplied cultures that are forging into promi nence on the international scene (15,68,39). It can, with out doubt, be attributed to the youthfulness of the dis cipline that no literature is available which pertains directly to this project. It may reasonably be assumed, however, that certain components contributing to success in traditional foreign language learning might also be funda mental to success in the scientific study of language 29 structure. Based upon this assumption, the literature reviewed related largely to predictive studies in foreign language aptitude, to the research which resulted ulti mately in the development of the Modern Language Aptitude Test, and to investigations involving the correlation of selected noncognitive elements with academic success. Cognitive Elements of Success The major instruments employed in the earliest attempts to predict achievement in academic endeavors were measures of intelligence or ability. Numerous studies utilizing different tests and combinations of tests with other cognitive variables produced a variety of results, but rarely would more than 25 per cent of the total varia bility be accounted for by such measures. Weitz (101) placed the usually attained correlation figures of scholas tic aptitude and achievement between .30 and .70. Bernard (9) reviewed the whole scene and concluded that the valid ity coefficient for the same two variables "appears to have reached an asymptote in the area of about .50." Prognosis of success in foreign language study made on the basis of specially designed foreign language apti tude tests fared little better prior to World War II. The 30 tests were generally of the paper-and-pencil variety which in large degree tested achievement in English. Emphasis was upon vocabulary and grammar. Included in some of the tests were "work-samples" constructed from an artificial language, but always based upon the Indo-European structure (52,21). The best known of these tests were the Iowa Foreign Language Aptitude Examination, The Luria-Orleans Modern Language Prognosis Test, the George Washington Uni versity Language Aptitude Test, and the Symonds Foreign Language Prognosis Test (19). These aptitude tests generally showed high correla tions with intelligence measures, and some researchers even found that intelligence tests were superior in predicting achievement in the foreign language field. Bertine (10) and Douglas and Broom (32) found the Thorndike Intelligence test to yield the best index of foreign language success. In 1931 Kaulfers (59) reported that in the previous thirty years only fifty-one prognostic studies in foreign language aptitude had been conducted, with the results indicating to him that no distinct language talent could be differen tiated from any other. A few years later he reinforced his earlier statement by declaring that he felt that foreign language aptitude tests used for predictive purposes were 31 of little practical value (59). Kaulfer's pessimistic view of foreign language aptitude tests was not shared by all researchers. In a report prepared in 1929 for the Modern Foreign Language Study and the Canadian Committee on Modern Languages, Henmon reviewed several such aptitude tests and concluded that "special prognosis tests are more effective instru ments than general intelligence tests and that these, together with objective measurements in a trial period, furnish in the present state of our knowledge the best bases of prediction and classification" (52). With the advent of the 1930's, interest in testing for foreign language aptitude dwindled almost completely, if one is to judge by the dearth of such entries in the professional journals. Whether this lack of activity was due to the Depression or to Kaulfer's baleful pronounce ment can only be left to conjecture, but the fact remains that relatively no research along this line was in evidence until after World War II. This world conflict, however, with all its polit ical overtones, forced Americans to the realization that an insular outlook upon the world would no longer suffice. It was imperative that the vehicles of communication be at our 32 disposal. An upsurge of interest in foreign languages and foreign language testing was almost immediately apparent (24). Approximately twenty-five years after Henmon had concluded that special aptitude tests were more effective instruments than were general intelligence tests and Kaulfers had countered by indicating that he saw no evi dence to support the belief that a distinct language talent differentiated from other talents, Salomon (85) surveyed the research in the field for that quarter century to find that special aptitude tests used for prediction yielded correlation coefficients around .60. She added that almost every language teacher felt that a distinct language apti tude existed, apart from that resident in the term "general intelligence." But it remained for future research to deal scientifically with several still unanswered questions: 1) Is there a general intelligence level below which suc cess is improbable? 2) Can special aptitudes be determined and measured in a manner that permits valid prognosis? Experimentation with instruments designed to do a more effective job of predicting success in language study took different tacks. One of the earliest studies of the postwar period, conducted at the University of Wisconsin by Bottke and Milligan (13), signaled a new trend in for eign language aptitude testing. These researchers did not restrict themselves by experimenting only with those vari ables measurable by paper and pencil; they sought to assess the contribution of other variables such as speaking abil ity, hearing ability, and aptness in comprehension to success in foreign languages. Out of the eight tests designed, they found the following four to be significantly correlated with achievement: 1) ability to understand a foreign sentence that was reasonably similar to English, 2) ability to read English prose, 3) the audiometer test, and 4) the ability to mimic the sound of words. These findings were reinforced by the 1955 study conducted by Schneiderman (88) which revealed a meaningful relationship between articulatory ability and language ability. In accord also with Bottke and Milligan's research was the statement made by Eugene Nida, a pioneer in descriptive linguistics, who declared that "mimicry is the key to lan guage learning" (76). In 1952 a large-scale research study was undertaken at the Army Language School in Monterey, California (24). Two sets of variables were investigated. One set consisted of the data obtained from the scores on the AGCT, three 34 subtests of the Seashore Musical Aptitude Test (tonal mem ory, timbre, and pitch), and the Army Language Aptitude Test WPQ-1, a test previously used for screening candidates for the United States Military Academy at West Point. It was this last-named test only that showed significant cor relation with success in foreign language study, and that only with the grades obtained at the end of the first two- week period of study. The second set of variables was derived from a specially constructed ten-test battery designed to measure abilities in the verbal, perceptual, and auditory realms. None of these variables increased significantly the predic tability of the battery. A memory span for digits and an ability to detect the number of English words to be found in a scrambled sequence of letters illustrate the type of behavioral acts assessed by the battery. Tests designed to measure such aptitudes perhaps tapped factors that are no more than incidental to language learning. About the same time that the project at Monterey was in progress, Sapon (86) was experimenting with work- sample tests for foreign language prognosis. Seeking to eliminate the salient weaknesses of previously designed work-sample tests, he structured one which would minimize 35 English structure and reflect a broad spectrum of the lan guage patterns of the world. The twenty-six minute test sought at the same time to measure as many dimensions of language aptitude as possible: phonology, morphology, syntax, lexicon. The instructions for learning Temtem, the artificial language of the test, were audio-visual, given by pre-recorded tape and projected lantern slides. No English was used except in the introductory remarks and incidental instructions. The test was given to a class of airmen enrolled in an intensive course in Mandarin Chinese. The correlation obtained from the total test score and the academic grades was .53. However, the reliability of the test was only .68. The researcher indicated that the fairly low reliability coefficient was probably due to the small number of test items and to several ambiguous items. It is of import to note the primary disiderata set forth as the basis for the design of this instrument (86). They were: 1. Verisimilitude of linguistic structure, reflect ing the linguistic phenomena found in different basic types of languages of the world. 2. Minimization of structural similitude to English. 3. Pedagogical verisimilitude. Wolfe sets forth 36 the principle that "practice materials should vary in as many dimensions and over approxi mately as wide a range as will the situations to be encountered when the learning is to be applied" (106). Sapon interprets Wolfe's principle in training technique to his setting in the following manner: "In terms of a prog nosis problem where one is concerned not only with the aptitude of the individual, but also with the nature of his response to a trial program, the learning and the measurement techniques applied in the work sample should approximate by these standards those likely to be found in actual course work. Herein should also be considered the role of various factors in language learning" (86). 4. Practical considerations of complexity of material and testing time. This work-sample test is of special significance to this study since it was the forerunner and served as a prelim inary draft for portions of one of the major batteries employed in this study— the Modern Language Aptitude Test. 37 The Modern Language Aptitude Test A review of the research carried on at the McKay Laboratory at Harvard by Carroll and Sapon (21) in the area of foreign language aptitude is necessary for an under standing of the rationale underlying the construction of the Modern Language Aptitude Test. The main objective of their study was to find out more about the nature of for eign language aptitude. The methodology involved the factor analytical study of two predictive batteries. The project was undertaken in connection with the USAF's trial course in Mandarin Chinese. "Trial courses" were used by the Air Force as the final step in the screen ing process for selection for intensive training in a foreign language. Initial screening had taken place on a four-fold basis: 1) a score of 70 or better on the Armed Forces Qualification Test, 2) a rating of 8 or more on the Technician Specialty aptitude index, 3) a high school diploma or its equivalent, and 4) a desire to study lan guages. The tests assembled for the two batteries were designed to tap the widest possible variety of factors that might relate to success with foreign languages. Each bat tery was composed of twenty subtests which yielded a total 38 of 29 predictor variables. (Some of the tests were repeated in each battery.) The subtests were planned to measure individual differences along three broad lines: 1) ability with the English language that might reflect itself in the learning of a second language, 2) special abilities that might be unique to foreign language acquisi tion, and 3) work-sample tests that might be predictive of specific aptitudes in criterion performance. The criterion variables were academic grades and selection or nonselec tion for the intensive language course. The test batteries, each requiring approximately four hours, were administered to two groups of 80 and 88 course candidates, respectively. Validity coefficients ranged as high as .84. The two sets of data which yielded the highest multiple R with the criteria were subjected to factor analysis in order to disclose the factorial struc ture of the tests, thus affording a basis for making inferences about the nature of foreign language aptitude. Carroll was cautious with respect to the cross-identifica tion of factors, admitting that "it was not possible to cross-identify factors without several somewhat arbitrary decisions; factors have been matched," he stated, "in what the writer believes to be the optimal way, but other 39 solutions could have been argued for" (21). Seven factors were tentatively identified (21): 1. Factor A was clearly the verbal knowledge fac tor. With one exception every test involved the extent to which the subject knew the lexicon and idioms of his own language. 2. Factor B was difficult to interpret, and there were no pure tests in either sample. The author indicated that he would "entertain the hypothesis that B represented specific motiva tion, interest, or facility with respect to unusual linguistic material." For convenience, the factor was named "Linguistic Interest." The statement was made that "whatever the nature of the factor, it is probably one of the most important components of foreign language apti tude." 3. Factor C was the associative memory factor. The author was perceptive in observing that "undoubtedly, too, the criterion variable . . . includes much memory work." 4. Factor D at first glance appeared as the word fluency factor, but it was not fluency without quality, or that type of speed production of words governed by mechanical rules, usually with respect to initial letters used. Carroll, rather, called this the Sound-Symbol associa tion ability. He speculated that early reading training might be related to this ability. Factor E represented the factor largely specific to Tem-Tem learning and was actually involved in the extended learning period. (Tem-Tem was the artificial language constructed by Sapon for a work-sample test.) This factor was viewed as the inductive language learning factor. The Tem-Tem test presented not grammatical descrip tions and explanations but simply the linguistic forms and referents, suitably arranged to prompt the learner to arrive at the grammatical rules and constancies for himself. Factor F appeared to be the syntactical fluency or grammatical sensitivity factor. The G Factor was not valid in predicting foreign language success. Seemingly, it was the speed of association factor. While not being of par ticular relevance in the early stages of language 41 learning, this factor might be of significance in the ability to switch rapidly from one language to another. Further validation studies eventuated in the selec tion of five tests to comprise a semi-final battery known as the Psi-Lambda Foreign Language Aptitude Battery. This battery was used by the Air Force in its continued search for better devices for screening candidates for intensive foreign language training. Harding, at Lackland Air Force Base, active in this research, noted the effectiveness of the instrument in reducing the attrition among candidates (49). When selection was based almost solely on general verbal ability as measured by the Army classification tests attrition was as high as 50 per cent. Using the trial course as an additional screening device reduced the attri tion rate to 12 per cent. Harding's study showed the Psi- Lambda battery to be as effective for screening purposes as the trial course. At present, the plan implemented by the Air Force entails the use of the final form of the MLAT for screening for entrance into the trial course, a plan which has reduced waste of time and effort, increasing signifi cantly the efficiency of selection for intensive language study. 42 Noncognitive Elements of Success Numerous studies have been undertaken in the attempt to relate various noncognitive variables to aca demic success, but the task has been complicated by the lack of consensus on definitions of the different aspects of personality. Few would question the contribution such variables might make to a regression formula were they subject to reliable identification and measurement. Kaulfers (59) in 1931 prophesied that "objective measures of industry, per sistence, interest, and dynamic will-to-learn," when available, would, in combination with a reliable aptitude test, furnish the best basis for prognostication. Krug (61) found that personality inventories accounted for the unpredictive residue in a battery almost as well as high school standing. This is understandable since the person ality factors operative in the college academic context are relatively the same as those present during high school days. Dowd (33) declared that the factors which depress achievement antecede college years. Bendig (8) found that the need tc achieve, as measured by the Edwards Personal Preference Schedule, con tributed significantly to course achievement. Uhlinger and 43 Stephens' (96) study showed expectancy, or minimum goal, to be of more importance than achievement need. Middleton and Guthrie (70) found that autonomy, aggression, and achieve ment need correlated positively with achievement. Holland, who has been active in studying personal ity syndromes of National Merit Scholars, indicated that measures of superego, persistence, and deferred gratifica tion were useful in prediction. He stated that in predic tive studies done "at a high level of scholastic aptitude personality variables may yield validity coefficients which are from two to almost three times as great as those ob tained using aptitude measures alone" (53) . One noncognitive factor on which there is a degree of agreement with respect to definition and for which reliable measuring instruments are available is that of interest. Many predictive studies of achievement have included this factor as a variable. It may be that arriv ing at some estimate of interest is a back-door method of assessing one aspect of motivation, and it seems virtually tautological to refer to the relationship of motivation to achievement (36). Researchers attempting to predict scholastic achievement on the basis of expressed interest have met 44 with varying degrees of success. Detchen (31) at the University of Chicago noted the importance of interest fac tors on performance in the social sciences by stating that "student performance is a combination of more than ability and past performance in a field of study . . . These addi tional factors are not as intangible from the standpoint of measurement as they have sometimes been considered." She supported her declaration by isolating the 109 items on the Kuder that pertained to interest in the field of social science and correlated them with scores made on the Social Sciences Placement Test. The obtained correlation coeffi cient was approximately .70. Kendall and Hahn (60) recommended the inclusion of the Kuder in a battery for selecting medical students because of the significant correlation between the scien tific scale and grades obtained in medical school. At Brigham Young University, Romney (83) found low, though significant, correlation between scores on the literaty scale and achievement in college English. In a research project conducted by Crosby (30), a highly significant dif ference in chemistry grades was noted between those stu dents with expressed high interest on the scientific scale and those with low. The same pattern was evident in the 45 biological sciences. Significant correlation was also observed between computational interest and grades in accounting. Hake and Ruedisili (43), although less successful in finding a statistically significant relation ship between scores on the Kuder and freshman grades at the University of Wisconsin, declared that an interest test should be included in a predictive battery because of its significant contribution to a regression formula in the presence of other variables. Fairly recently, Barrilleaux (7) undertook a predictive study employing interest scores in conjunction with aptitude scores. The results reported in expectancy per cents were impressive. Cronbach very aptly reported what most researchers find— that the Strong and the Kuder keys produce a low correlation with grades but that "they measure a factor of undeniable significance in school success" (29). Yum (110) related interest to motivation by stating that interest scales "are not measures of ability but are meant to give some index of the extent to which an individ ual will be motivated in various areas of human activities, along with the ability at his disposal." Crosby (30) ex pressed a similar opinion by saying that "some measure of motivating factors such as interest is most essential to 46 adequate prediction of achievement in academic work." Summary A review of the literature revealed no research that pertained to the validity of a battery of tests in predicting success in the study of descriptive linguistics. Assuming, however, that there are certain factors common to success in any academic field, a brief overview of the literature pertaining to scholastic achievement in general was presented. Both cognitive and noncognitive variables were included. Research studies dealing with foreign lan guage aptitude received particular attention because of the natural relationship existing between the two aspects of linguistics as a discipline; the one emphasizing the abil ity to speak and read a foreign language, the other, the ability to analyze and describe the basic internal struc ture of a language whether in written or unwritten form. The language aptitude battery and the interest inventory used in the present project were singled out for special note. The historical background of the newest of the devices for assessing foreign language aptitude, the MLAT, throws light on the nature of the instrument itself and affords an indication of the rationale behind its 47 inclusion. The studies relating to the Kuder interest inventory were cited to point out the possibly unique fac tors being tapped by such a tool. A reliable measure of whatever degree of motivation is inherent in expressed interest is viewed as giving breadth to a predictive battery. CHAPTER III METHOD OF PROCEDURE Introduction The purpose of this research was to assemble a predictive battery that might provide the groundwork for the establishment of guidelines for use in any selection process that might be employed in the field of descriptive linguistics. In order to accomplish this, a battery of tests was administered to a sample of first-year linguistic students at the Summer Institute of Linguistics. Three standardized test batteries, measuring both cognitive and noncognitive elements of success, and three biographical items provided the twenty-two predictor variables. Course grades and staff evaluations supplied the six criterion variables. By means of a correlational study a multiple coefficient for the entire battery was obtained. Other correlational studies employing a reduced number of pre dictors against the same criteria were also included. Measures of central tendency, variability, intercorrela tions, validity coefficients, multiple R's, and other items 48 49 of statistical information were determined through the use of the IBM 7094 computer. A consideration of these four components, sample, predictor variables, criterion meas ures , and statistical methodology form the basis of the present chapter. Sample The Summer Institute of Linguistics (SIL), the largest linguistics training institute in the world, pro vided the sample for this study. The three North American summer institutes, affiliated with the Universities of Oklahoma, Washington, and North Dakota, offer classes in descriptive linguistics which are accredited through the universities as a part of their regular summer course offerings. The basic requirement for admission to the schools is three years of higher education or special permission. The initial summer course of study is usually eleven weeks in length, and the training offered is equivalent to that of a university major in linguistics. The academic product is described as a "linguistic technician." In the summers of 1964 and 1965, data were collect ed on first-year SIL students. In 1964 only students at 50 the Oklahoma branch were tested. The following year the battery was administered to the three branches. The final sample for whom a complete set of data were available was comprised of 36 men and women from the 1964 school at Oklahoma, 74 from the 1965 school, 27 from North Dakota, and 32 from the University of Washington, producing a total of 167 students. The sample ranged in age from 17 to 57 years, with an average being 26. A wide range in amount of previous foreign language study was also observable. The average number of years in this field was 4.6. Some enrollees had never engaged in foreign language learning while others had had as many as 23 years of such training. Table 1 presents by schools the biographical data obtained for the current project. Test data and biographical items were collected on the 167 students. Criterion data, which consisted of instructor-assigned grades and staff evaluations, were also obtained. Information of this category was complete for all but 18 students. These 18 were not enrolled for Field Problems, the one course not specifically linguistic in nature. TABLE 1 SEX, AGE, AND PREVIOUS LANGUAGE TRAINING FOR THE FOUR SCHOOLS AND TOTAL SAMPLE Amount of pre vious language Number Number training Chronological age School Men Women Total Years S.D. Mean Median S.D. Range Oklahoma 1964 11 23 34 4.2 3.8 26.2 25 5.7 17-43 Oklahoma 1965 26 48 74 5.1 4.3 25.7 25 5.4 17-55 North Dakota 1965 12 15 27 5.3 4.6 25.4 24 7.2 18-57 Washington 1965 13 19 32 3.4 2.4 26.2 26 4.9 19-40 Total 62 105 167 4.6 4.0 26.0 25 5.7 17.57 52 Predictor Variables Predictor variables employed for this project were derived from three standardized test batteries and selected biographical data. The batteries included an intelligence test, an aptitude measure, and an interest inventory. The rationale underlying the choice of instruments for assessing these facets of personality was that each measure would in varying degrees tap possibly unique quali ties which contribute to academic success. The inclusion of both cognitive and noncognitive variables provided for a more comprehensive study of the components of that illu sive quality making for achievement. A measurement of temperament was not included because of poor history such measures have had in lending significant weight to predic tive batteries. The three biographical items included in the study are sex, age, and number of years of previous language study. These were chosen since both in literature and in popular belief they are viewed as factors related to language learning. A total of twenty-two independent variables were obtained from the three batteries and the biographical items. The two batteries measuring cognitive factors, the Henmon-Nelson Intelligence Test and the Modern Language 53 Aptitude Test, yielded nine variables; the Kuder inventory and the biographical data supplied thirteen variables. Of the twenty-two variables, two represented composite scores, the MLAT and Henmon-Nelson totals. When the separate parts of these two batteries were employed in the computations, the composite variables were deleted. Henmon-Nelson Intelligence Test In a research design pertaining to academic achievement an intelligence test is basic, for though it may be conceded that the scores on such a test are not to the same degree predictive of success in all areas of life, they definitely make significant contribution in the realm of the scholastic. Nor would it be possible to ascertain the possible uniqueness of other tests apart from some measure of "general" intelligence. The instrument employed in the current project was the Henmon-Nelson Test of Mental Ability, which is a forty-minute, well designed, group intelligence test producing verbal, quantitative, and total scores. The Henmon-Nelson test, published by Houghton Mifflin Company, was constructed and revised by M. J. Nelson, Tom A. Lamke, and Paul C. Kelso. Earlier forms of the test, which took thirty minutes for administration, 54 yielded only the single composite score based upon ninety items, but the 1961 revision increased the number of items to one hundred, the testing time to forty minutes, and produced quantitative and verbal scores in addition to the previous one total score. The one hundred items, which are arranged in a "spiral omnibus" pattern, are suffi ciently varied to hold interest high. Included are ques tions of information, reasoning, figure analogies, verbal and concept analogies, and proverb interpretation. Two types of answer sheets are available, the IBM sheet for machine or hand scoring, and a self-marking answer sheet. The IBM form was used in this project. A major consideration in the selection of the Henmon-Nelson was the fact that it offered in a brief form perhaps as reliable a group measure of intelligence as could be found. The authors were commended by all the reviewers in Buros' Fifth Mental Measurements Yearbook with respect to the manner in which the test was constructed, validated, and standardized. D. Welty Lefever (19) de clared that one of the best indices for the validity of the test was the series of correlations between the Henmon- Nelson I.Q.'s and achievement test scores. Another evi dence was the "care with which the standardization was 55 conducted." He concluded that it was a satisfactory in strument for the many purposes for which a short intelli gence score is needed.- Leona E. Tyler (19) indicated that the authors were "eminently successful" in producing a short, self-administering, easily scored instrument which would have maximum predictive validity in school situa tions. She added that the Henmon-Nelson is a "carefully planned, carefully constructed, readily interpretable instrument designed to do one thing and do it well." Laurence F. Shaffer (19) concurred by asserting that the standardization "is a model of precise methodology . . . The revised Henmon-Nelson impresses the reviewer as a scholarly example of the best in test construction and is a remarkably efficient instrument for its length." One weak ness pointed out by John 0. Crites (20) was that the manual for use with the college and adult level test omitted tables of normative data for all groups other than college freshmen. In the light of the nature of the present sample this was regrettable for a comparison of this group with a normative was not possible. Modern Language Aptitude Test Following World War II, the intensive interest of 56 the Armed Services and the State Department in the rapid training of men to understand and speak other languages led to an intense study of the factors contributing to success in this field. It was recognized that not all of the men who qualified on the AGCT made good linguists, thus pointing up the need for more refined instruments for use in screening. John B. Carroll of Harvard University and Stanley M. Sapon (21) of Ohio State University were called upon to study the problem. This assignment resulted in their collaboration in a five-year research project de signed to identify to some degree the factors inherent in foreign language aptitude. The first battery constructed was not a randomized collection of tests, but, rather, twenty carefully chosen separate tests which the researchers hoped would throw light on the nature of the aptitude associated with second language learning. The tests fell into four broad cate gories: verbal ability, grammatical sensitivity, phonetic discrimination, and work-sample. Verbal ability, which had long been established as a major "factor" in language learning, was tapped by several of the tests. Grammatical sensitivity, viewed as critical in the acquisition of a foreign language, was measured by instruments designed to 57 test the examinee's ability to recognize the grammatical function of words in sentences without knowing the grammat ical terms. Aptitude for phonetic discrimination was found to be noncrucial because the ability to recognize sounds appeared to be more a matter of learning than actual audi tory discrimination. In the development process three work-sample tests, largely the work of Sapon (86) , were designed for study: 1) an artificial language in which the examinee was re quired to learn the names of a foreign language number system and then produce them from dictation, 2) an auto mated language test, simulating modern oral language learning, based bn the artificial language of "Tem-Tem," 3) a grammar-translation type, following the lines of tra ditional language instruction, based on "Perdaseb," another artificial language. These proved to be of value in delin eating more clearly the nature of foreign language apti tude. They were, however, deleted from the final battery because of their complexity and the time required for administration. The experimental tests were administered to approx imately 5,000 persons, 1,300 of them being enrolled in intensive foreign language courses conducted under the 58 auspices of governmental or military agencies. A factor analysis study based upon the foundation of the early batteries resulted in the identification of four variables, or complex of variables, underlying lan guage aptitude (24): Phonetic Coding. One of the most important abilities required in learning a foreign language is the ability to "code” auditory phonetic material in such a way that this material can be recognized, identified, and remembered over something longer than a few seconds. Phonetic coding refers not merely to the aptitude for mimicking combinations of sounds but also to the abil ity to "code" them in such a manner that retrieval is possible even after a lapse of time. This aptitude is chiefly measured by the Phonetic Script Test, which requires the examinee to learn how a series of sounds is represented by alphabetic signs. Paired-associates tests, artificial language numbers tests, and spelling clues tests may involve this ability. Grammatical sensitivity, which is measured by the Words in Sentences subtest of the MLAT, is described in the following manner: Grammatical Sensitivity. A second important variable in language aptitude is the ability to handle 1 1 grammar," i.e., the forms of language and their arrangements in natural utterances. This 59 implies that the individual is sensitive to the functions of words in a variety of contexts. A third ability is: Rote Memory for Foreign Language Materials. A third important variable is that of rote mem orization ability for foreign language materials. This ability is not the same as phonetic coding but refers to the "capacity to learn a large number of these associa tions in a relatively short time." The Paired Test is particularly efficient in assessing this trait, which is also tapped by the Number Learning Test. The fourth discernible factor in language learning was, unfortunately, not measured appreciably by any of the subtests of the MLAT. It is described by Carroll as follows: Inductive Language Learning Ability. A fourth variable is what may be called "inductive language learning ability." This is the ability to infer linguistic forms, rules, and patterns from new linguistic content itself with a minimum of super vision or guidance. (24) The background history of the MLAT and the major factors in foreign language aptitude, as conceived by Carroll and Sapon as a result of their carefully designed and executed research, provides the basis for an examina tion of the perfected instrument. The stated purpose of the MLAT, according to the 60 manual, is "chiefly to provide an indication of an indi vidual's probable degree of success in learning a foreign language." The MLAT is useful for prediction of success in learning not only to speak and understand a foreign lan guage, but also to read, write, and translate one. The applicability of this instrument is not confined to modern languages but is pertinent to the classical as well. No published research is available, however, which indicates its efficiency in predicting success in descriptive lin guistics. The present study proposed to examine this question. The MLAT is available in two forms: the Short Form of three parts, which takes thirty minutes of testing time; and the complete test of five parts, which involves the use of a pre-recorded magnetic tape for presenting the instruc tions and the stimuli and requires seventy minutes for administration. The complete form of the test was used in the current study. Based upon the five-year study of Carroll and Sapon, five subtests were selected for inclusion in a final battery which became known as the Modern Language Aptitude Test. The traits measured by the several parts are described in the following manner: 61 Part I. Number Learning* This seems to measure one aspect of the memory component of foreign lan guage aptitude, but the part also has a fairly large specific variance, which one might guess to be a special "auditory alertness" factor which would play a role in auditory comprehension of a foreign lan guage . Part II. Phonetic Script. This appears to measure what we have called sound-symbol association ability, that is, the ability to learn correspond ences between speech sounds and orthographic symbols. It may also measure a sort of memory for speech sounds, and it tends to correlate highly with the ability to mimic speech sounds and sound combina tions in foreign languages. Part III. Spelling Clues. Scores on this part depend to some extent on the student's English vocabulary knowledge. This subtest also measures the same kind of sound-symbol association ability as measured by Part II, Phonetic Script, but to a lesser extent. It is highly speeded. Part IV. Words in Sentences. This part is thought to measure sensitivity to grammatical struc ture, and may be expected to have particular rele vance to the student's ability to handle the gram matical aspects of a foreign language. As yet, it is not known how much scores on this part are a reflection of formal training in grammar; at any rate, no grammatical terminology is involved, so that the scores do not depend upon specific memory for grammatical terminology. Part V. Paired Associates. This part measures the rote memory aspect of the learning of foreign languages. (26) Validity and reliability coefficients that are given in detail in the manual for both the Short Form and the complete test are sufficiently high to justify the use 62 of the instrument with confidence. Tables of validity coefficients obtained in different settings and with dif ferent samples are presented, but the ones that are of particular significance for this study are those relating to achievement in "intensive" language courses. The students enrolled for these courses were more nearly char acteristic of the present sample in age, number of hours a day devoted to language study, motivation, and pressure. The criteria generally employed were grades assigned by instructors. Validity coefficients reported for the Army Language School and the Foreign Service Institute, which, in the main, have sponsored the "intensive" courses, ranged from .46 to .73. Odd-Even reliability coefficients for the complete test ranged from .92 to .96 (26). Test reviewers have reported favorably on both the uniqueness of the MLAT as a measuring instrument and its quality construction. Laurence Siegal (20) termed it a "highly successful" battery which was a "more valid pre- * dictor of success in learning a foreign language than various intelligence tests." Edward Bordin (20) declared that the reliability and validity data were of such a nature as to establish confidence in its usefulness as a diagnostic tool. 63 Kuder Preference Record— Vocational The Kuder Preference Record, published by Science Research Associates, is a vocational interest inventory yielding scores in ten broad areas. G. Frederick Kuder employed item analysis in the construction of this instru ment in order to select the items which best measured interest in the area for which they were designed. Form C, which was used in this project, consists of 780 items grouped into ten interest "clusters" and a verification index. The verification index, or "V" score, was an inno vation made in Form C for the purpose of identifying those examinees answering the questions carelessly or without an adequate understanding of the directions. The following broad fields of vocational interests with their attendant descriptions are represented: 1. Outdoor interest: indicating a preference for outdoor work, dealing largely with animals and growing things. 2. Mechanical interest: pointing up a liking for work with machines and tools. 3. Computational: indicating an interest in work ing with numbers. 4. Scientific: showing a preference for vocations 64 involving the discovery of new facts and the solution of problems. 5. Persuasive: implying a liking for meeting and dealing with people and promoting projects or things to sell. 6. Artistic: indicating an interest in doing crea tive work with one's hands— particularly as involving visual aesthetic senses. 7. Literary: showing a preference for occupations involving much reading and writing. 8. Musical: pointing out interest in vocations having to do with music. 9. Social service: indicating interest in work primarily aimed at helping people. 10. Clerical: implying an interest in jobs requir ing precision and accuracy— perhaps with a certain routine quality. Biographical Items Sex. Sex differences in performance in language courses have been noted by several researchers (1,28,23), the difference generally being in favor of the females. * Carroll, however, indicated in the MLAT manual that almost 65 no data had been collected that would permit a comparative validity study of the MLAT used with adult men and women. The present research provided an opportunity for studying the relative achievement of the sexes in the study of descriptive linguistics. The sample of 167 was composed of 62 men and 105 women. Age. The relation of the age factor to foreign language learning is a moot question. The age in years of each of the 167 subjects taking part in this study was considered a critical variable since very little previous research in the field of foreign language had been under taken with a group that was characterized by such a wide age range as the present one. The youngest student enrolled was 17; the oldest, 57. This presented a range of 40 years. Amount of previous foreign language study. The number of years of previous foreign language study was calculated in school-year units. Those who were bilingual were assigned credit for four years of formal training. No distinction was made between ancient and modern languages. Criterion Measures According to Guilford a good criterion measure is 66 the key to a validation study (42). For this reason two types of criterion measures were employed in this project. Grades received in course work and the opinion of qualified staff personnel as to the individual student's probable degree of future success in linguistics formed the basis for evaluating achievement in the study of descriptive linguistics. The traditional grading system, which is the gen erally accepted method of recording level of scholastic performance, is recognized as having inherent weaknesses. However, until educational institutions supply more reli able criteria, course grades still provide the best imme diate differentiation between levels of academic success. For the present study grades were obtained for courses in Phonetics, Phonemics, Morphology and Syntax, and Field Problems, a course not distinctively linguistic in charac ter but, rather, one comprised of elements of anthropology, literacy aids, and translation techniques. A linguistic G.P.A., which was derived by averaging the grades for the three strictly linguistic courses (Phonetics, Phonemics, and Morphology and Syntax) served as a major criterion measure. The rationale supporting the utilization of this composite score lay in the fact that language acquisition 67 is not a segmented affair but is a unitary function which embraces sounds, sound systems, syntax, and semantics in an orderly fashion. The courses at the Summer Institute of Linguistics for which students were assigned grades are described in the 1965 catalog of the Institute: Phonetics: (1) Theory of the formation of sound types, both "regular" and "exotic." (2) Drill in (a) analyzing, (b) recording with pho netic symbols, and (c) reproducing these sounds. (3) Theory of tone, intonation, stress, and quan tity. (4) Drill in reading English intonation; drill in recording and reproducing the sounds of the languages of various foreign speakers on the campus. (6) Stimulation of the students1 ability to mimic sounds. Phonemics: (1) Background theory of systems of sounds. (2) Procedures for determining such systems under field conditions. (3) Practice with hypothetical and actual language material (a) to instill useful analytical habits, {b) to build up linguistic judgment (based on samples showing the most frequently met types of linguistic phenomena and (c) to teach methods of describing linguistic materials. (4) Methods for forming practical orthographies for literary purposes (i.e., reducing a language to writing). Morphology and Syntax: Presentation of funda mental techniques employed in grammatical analysis of a language: analysis and description of types and variations of morphemes, their groupings in sequence and arrangement into classes; relation be tween morphemes, words, and sentences; categories of meaning; morphological and syntactic construc tions. Numerous graded problems based on actual languages introduce the student to language types from many parts of the world. By handling 68 progressively more complex problems, the student gains sufficient facility to be able to analyze the intricacies of real languages. During the final weeks of these introductory courses the student works with a native speaker of a language unknown to him. In this field problem opportunity is given for the student to begin the analysis of the phonology and grammar of an actual language under the supervision of the course in structors. At this time the normal classroom sched ule is abandoned for more personalized help in the language analysis. Field Problems: Anthropology: A few introduc tory lectures designed to acquaint the prospective missionary-linguist with problems which he will meet in attempting to live with and understand aboriginal peoples. An attempt to build a sympa thetic approach to the culture, social organization, and religious and ethical concepts of native peoples. Literacy Aids: Lectures designed to prepare the student to construct and teach reading materials, once the analysis of the phonemes of the particular language has been made. Translation Techniques: Brief consideration of language techniques applied to translation. Prob lems in the theory of meaning as seen through trans lation methodology. Methods and helps which the American Bible Society suggests for insuring accu racy and consistency in translation. (91) The instructors employed a letter-grading system to designate the level of achievement attained by the individ ual student. For computational purposes the letter grades were assigned numerical values, beginning with a possible A+ equaling 12 to F equaling zero. In addition to the course grades, evaluations of 69 the probable degree of future success of each of the stu dents in descriptive linguistics were obtained from two staff members independently of each other. The instruc tors, highly trained and experienced linguists, were requested to make evaluations according to five categories. The categories "poor," "below average," "average," "above average," and "superior" were each divided into two parts to permit a ten-point total gradation in the assessment. Securing the appraisal of these judges had the value of providing an estimate of the student's motivation, deter mination, and persistence, along with other factors rele vant to success in this field. Ratings received for each student were averaged to yield the criterion measure termed "staff evaluation." A product-moment coefficient was computed for the staff ratings in order to determine the extent to which the appraisal of the two judges was correlated. The obtained coefficient of .75 indicated a fair degree of reliability for the staff evaluation criterion measure. The total number of criteria employed in the cur rent research was six: four course grades, a composite linguistic G.P.A. derived from averaging the grades assigned in the three distinctly linguistic courses, and 70 staff evaluation. Linguistic G.P.A. and staff evaluation were considered the major dependent variables; hence, the discussion and conclusions found in Chapter V will be based largely on these criteria. Procedure The data obtained from the biographical material and three standardized test batteries were submitted to the IBM 7094 computer at the Health Science Computing Facility at the University of California, Los Angeles. The statis tical computations yielded a simple description of the data, intercorrelations of the independent and dependent variables, validity coefficients, and multiple correlation coefficients and regression studies for the total battery as well as for several combinations of a reduced number of the predictors. The simple data description, which included meas ures of central tendency and variability for each variable, provided opportunity to study the sample involved and their performance with respect to the battery and to achievement in the study of descriptive linguistics. An examination of the intercorrelations of both independent and dependent variables was undertaken for the purpose of determining any 71 existing relationships that might be particularly relevant to this project. Also, the extent to which any two sep arate variables overlapped gave indication of the relative value of those variables to prediction. The validity coefficients between predictors and criteria were estab lished in order to study the prognostic quality of each independent variable against each of the criterion measures. Multiple correlation studies, however, were the major consideration of the study since the validity of the entire battery for predicting success in descriptive lin guistics was hereby determined. Not only was the entire battery thus studied, but multiple R's for a reduced number of selected independent variables in various combinations were obtained. The variables attaining a significant t value in the several regression studies were singled out for special note. (The t value represented the ratio be tween the regression coefficient and its standard error.) Beta weights for the predictors in the regression studies were computed by means of the following formula: SD n I pred. pred. = Ppred. SDcrit Summary The present chapter was designed to outline the procedure employed in this research to assemble a battery which would be predictive of success in the study of descriptive linguistics. A description of the sample, predictor variables, criterion measures, and procedures was undertaken. The sample was composed of 167 men and women en rolled for the intensive course in descriptive linguistics offered by the Summer Institute of Linguistics. To them was administered the battery of twenty-two variables derived from the Kuder Preference Record— Vocational, the Modern Language Aptitude Test, the Henmon-Nelson Intelli gence Test, and three biographical items. The criteria against which these independent variables were examined were measures of the student's success in the study of descriptive linguistics as assessed by instructor-assigned course grades and staff evaluations. It was pointed out that the average of the grades obtained in the purely lin guistic courses served as one of the major criteria and staff evaluation as the other. The procedures largely involved correlational studies. The statistical computa tions were accomplished through use of the IBM 7094 computer. CHAPTER IV ANALYSIS OF DATA AND FINDINGS The present chapter undertakes an analysis of the data and of the findings in the present research. It is divided into five parts. The first section is concerned with simple data description, including the measures of central tendency and variability for each of the variables. The second and third sections examine the inter correlations of the independent variables and of the dependent variables. The fourth part pertains to the validity coefficients, while the fifth undertakes an analy sis of the multiple correlation coefficients obtained from the complete battery and from selected combinations of variables. Simple Data Description A statistical description of the data is provided in this chapter to permit a study of the performance of the present sample with respect to the variables employed. 73 74 Table 2 summarizes the mean, standard deviation, standard error of the mean, the maximum and minimum scores, and the range for each variable. The range and standard deviation of the variables provide an indication of the character of the sampling distribution. Age, amount of previous language study, and performance on the foreign language aptitude battery have been singled out for special consideration. The widespread distribution in age afforded an opportunity to observe possible effects of this variable upon success in learning descriptive linguistics. The mean age in years of the total group was 26; the standard devia tion, 5.7. The amount of previous foreign language study also revealed a broad range, from 0 to 23 years, with a mean of 4.6 and a standard deviation of 4. The extent of this range provided an adequate basis for examining the rela tionship between previous language training and success in descriptive linguistic study. From the data published in the MLAT manual (p. 14), it was possible to compare the performance of the present sample on the foreign language aptitude battery with that of two other groups of comparable age. Table 3 outlines 75 TABLE 2 MEASURES OF CENTRAL TENDENCY AND VARIABILITY OF THE INDEPENDENT VARIABLES Variable Mean S.D. s .d .m Maximum score Minimum score Ran< Age 26. 0 5.7 .44 57 17 40 Language Study 4.6 4.0 .31 23 0 23 Outdoor 48.6 13.5 1.04 77 11 66 Mechanical 30.1 12.4 . 96 61 4 57 Computational 21.6 9.0 . 70 45 2 43 Scientific 37.6 10.5 .81 61 13 48 Persuasive 28. 4 11.6 . 90 64 7 57 Artistic 26.2 9.1 .71 48 3 45 Literary 20.9 7.7 .59 63 7 56 Musical 15. 6 6.1 .47 36 0 36 Social Service 59.4 11.7 . 91 80 21 59 Clerical 41.1 14.1 1. 09 79 14 65 MLAT Part I 33.4 8.6 .67 43 12 31 MLAT Part II 26.2 4.1 .32 31 14 17 MLAT Part III 19. 0 9.4 .73 49 1 48 MLAT Part IV 28.7 7.7 . 59 43 8 35 MLAT Part V 18.4 5.1 .39 24 5 19 MLAT Total 126.1 25.8 2. 00 182 63 119 H-N Q 20.6 7.8 . 60 39 4 35 H-N V 43.4 10.3 .79 59 12 47 H-N Total 64.1 15. 9 1. 23 94 24 70 76 the means and standard deviations of the MLAT scores obtained by the Army Language School, the State Depart ment' s Foreign Language Institute, and the students at the Summer Institute of Linguistics. A higher level of apti tude and a relatively narrower range was shown for the S.I.L. enrollees over the other two groups. TABLE 3 MEANS AND STANDARD DEVIATIONS ON THE MLAT FOR TWO PREVIOUS RESEARCHES AND THE PRESENT STUDY School Mean S.D. Army Language School (1956) 107.5 31.4 Foreign Service Institute (1956) 114.2 32.8 Summer Institute of Linguistics (1964, 126.1 25.8 1965) Intercorrelations of the Independent Variables The analysis of the intercorrelations of the inde pendent variables falls into three categories: biographical items, interest variables, and cognitive elements. The biographical and interest items did not generally produce significant correlations with the other independent vari ables. Amount of previous language study was the exception. It revealed a high degree of relationship to most of the 77 aptitude and intelligence measures. The cognitive vari ables showed high intercorrelations with other such vari ables. Biographical Items Sex. A sex difference in performance on the instruments employed in this project was more evident in the biographical and noncognitive realm than in the cogni tive (see Table 4). Scores from the Kuder revealed that men were significantly more interested than women in Out door, Mechanical, Computational, and Social Service voca tions. Women, on the other hand, scored higher on the Scientific, Artistic, Musical, and Clerical scales. Analysis of the data pertaining to the amount of previous foreign language training disclosed that the men in the sample had engaged in more years of language study than had the women. An explanatory factor might be that some of the men had completed several years of theological seminary training prior to enrolling for the linguistic course. Such training traditionally includes the study of Greek, Hebrew, and, at times, Latin. Significant correlations between sex and cognitive elements were found with only two variables, Parts IV and 78 TABLE 4 CORRELATIONS BETWEEN THE BIOGRAPHICAL ITEMS AND THE OTHER INDEPENDENT VARIABLES Variables Sex Age Language Study Sex Age Lang. Study 1.00 -.08 -.241 -.08 1.00 . 24* -.24 .24] 1.00 Kuder Interest Outdoor Mechanical Computational Scientific Persuasive Artistic Literary Musical Social Service Clerical -. 21 -. 51b -. 19a -.31b -.08 . 23b -.01 . 18a . 23b . 22b .07 .14 .12 .03 .05 ‘.12 .07 .10 .07 .11 .03 .06 . 00 . 06 .04 .07 . 07 .06 . 04 ■.06 MLAT I MLAT II MLAT III MLAT IV MLAT V MLAT Total .05 .09 . 12 . 19a . 22b .14 ■.15 .07 .14 .12 . 31b . 20a . 17a . 30b . 18a . 20a .05 . 24b Henmon-Nelson Q Henmon-Nelson V Henmon-Nelson Total .13 ■.12 •.14 .07 .04 .07 .15 .35* .301 aSignificant at .05 bSignificant at . 01 cFemales were given the higher score 79 V of the MLAT. The women performed significantly better than the men on both these variables, grammatical sensi tivity and memory facility. With respect to grammatical sensitivity, McCarthy in 1954 noted somewhat the same trend when she found females slightly superior to males in composition skills and other aspects of language (114). In a similar context, Anastasi stated that girls do rela tively better "in tasks involving mastery of the mechanics of language" (1). Part V of the MLAT, entitled Paired Associates, was designed to measure the memory aspect of foreign language learning. The correlation of sex with this variable was significant at the .01 level. This finding, too, was in harmony with previous studies. A sex difference is usually noted in performance on memory sections of group intelli gence tests in which women tend to make higher scores than men (1). Anastasi, in an attempt to explain this phenome non, suggested that the female's degree of superiority in verbal functions and sensory imagery, as they facilitated retention, could contribute to her excelling in logical memory tasks. The manual for the MLAT stated that girls tended to make higher scores on the aptitude measure than did boys 80 but that data were not available for comparing the per formance of adult men and women. The present study, which included similar data, did not show sex to be a significant factor in the MLAT total score. Age. Table 4 indicates the relationships found between age and the other predictor variables. A fairly consistent negative trend was noted, although an acceptable significance level was obtained with only three of the variables: amount of previous language study, Part V of the MLAT, and the MLAT total. It appears relatively clear that the age factor does not show an appreciable overlap with other predictors. The relatively high degree of relationship between age and language study may be traced to the fact that some of the younger students not only had had no graduate work but probably had not yet completed their undergraduate studies at the time of this investigation. An understanding of the relationship revealed be tween age and the two aptitude variables could rest in the nature of Part V of the MLAT. This test was designed to measure the rote memory ability employed in foreign lan guage study by tapping the capacity of the individual for learning a large number of associations in a relatively 81 short period of time. The negative correlation between age and performance on a test demanding memory acuity fol lows previous research findings along this line. Memory tests have been found to favor the younger subjects over the older (1). Studies employing the Wechsler Adult Intelligence Scale found scores on the Digit Spans subtest, a timed measure of memory span, to decline rapidly with age (100). The significant correlation of age with the total score of the MLAT could be associated with the same factor since Part V was so highly correlated (see Table 7). Further consideration of age and foreign language aptitude will be discussed in a later section as they relate to the criterion measures. Amount of previous foreign language study. The number of years of previous language study was signifi cantly related to all the cognitive measures of the battery with the exception of the MLAT V and the Henmon-Nelson Q (see Table 4). Performance on Part V of the MLAT, designed to measure the rote memory aspect of foreign language apti tude, may have been adversely affected by the interference experienced when a previously learned family of foreign language symbols competes for dominance with a new set of symbols. 82 Part II of the MLAT and the Henmon-Nelson Verbal produced the highest Pearson r's with amount of language training. Part II was constructed to measure the ability to learn correspondences between speech sounds and written symbols, a factor closely associated with the ability to mimic. Whether the relationship between this ability and the amount of previous foreign language study was the result of natural selective forces or of learned responses cannot be determined here. Previous experience with a foreign language could make a student more alert to varia tions in sound as well as train him in the art of mimicry. On the other hand, ability of this nature might merely have been reinforced through earlier successful experiences with foreign languages. The correlation between verbal intelligence and language training likewise defies definitive explanation. Foreign language study probably increases a person's word awareness, but a natural selective factor may also be operative. Interest Variables Table 5 reveals that the Kuder interest variables were not generally highly correlated with the cognitive TABLE 5 CORRELATIONS BETWEEN KUDER INTERESTS AND COGNITIVE VARIABLES Interest Variable C o g n i t i v e V a r iable MLAT Henmon-NeIson I II III IV V Total Q V Total Outdoor -.1 1 i • o £> . -.13 - . 25b CM r— 1 • 1 - . 18a .05 r" o • I -.0 7 Mechanical .01 .09 -.0 8 - . 23b -.15 -.1 1 .05 -.0 2 .01 Computational .07 -.0 1 -.0 2 .03 00 o . 1 .15 . 39b i . o to . 17a Scientific -.0 6 .03 -.0 7 -.12 -.1 1 -.09 .14 .06 .11 Persuasive -.03 -.0 9 -.07 -.03 0 • 1 i • o - . 21b .00 -.1 0 Artistic .11 .07 .12 .06 GO O • .12 -.0 4 .01 i —1 O • 1 Literary . 16a -.0 3 .05 . 21b .18 .15 .06 .15 .13 Musical .04 .14 .15 . 25b .14 . 19a .09 .09 .11 Social Service -.0 2 -.0 8 -.0 7 .02 .02 -.0 3 -.1 5 -.1 4 - . 16a Clerical -.1 0 -.1 1 .04 .00 -.0 2 0 . 1 i— 1 0 • 1 -.1 0 -.0 7 Significant at .05 Significant at .01 QO CO 84 elements of the battery. The quantitative measure on the intelligence test was, however, significantly correlated with two interest scales. With Computational interest the Pearson r was .39; with Persuasive interest, -.21. Both coefficients were significant at the .01 or better. A number of the interest scales revealed high intercorrelations with other interest variables. Table 6 sets forth these intercorrelations and the significance level attained. Cognitive Variables The intercorrelations of the cognitive variables with each other were all significant at the .01 level or better (see Table 7). The magnitude of the correlation coefficient, .71, obtained between the MLAT total and the Henmon-Nelson total might have been expected with a trun cated sample such as the present one. The coefficient of determination (.50) revealed the fact that there was a relatively large amount of overlap between the two bat teries when used to predict success in linguistic studies. Hence, the predictive efficiency of the entire battery was reduced. TABLE 6 INTERCORRELATION MATRIX FOR THE KUDER INTEREST VARIABLES Variable lc 2 3 4 5 6 7 8 9 10 Outdoor (1) 1.00 Mechanical (2) .45b 1.00 Computational (3) -.05 -.03 1.00 Scientific (4) . 42b . 51b . 27b 1.00 Persuasive (5) -. 46b -. 27b -. 21b -. 35b 1. 00 Artistic (6) -.04 -. 21b -. 17a -. 34b -.14 1.00 Literary (7) -.15 -. 22b -. 19a -. 39b .13 -.02 1.00 Musical (8) -. 18a -. 25b -.10 -. 16a .11 .01 -.03 1.00 Social Service (9) -.08 -. 26b -. 19a -.11 -.02 -.10 -.06 -.04 1.00 Clerical (10) -. 44b -. 27b . 41b -. 21b -.03 -.09 -.08 -.09 -.34b 1.00 Significant at .05 ^Significant at .01 S cNumbers in column headings represent items bearing corresponding numbers in stub. 86 TABLE 7 INTERCORRELATIONAL MATRIX FOR THE COGNITIVE INDEPENDENT VARIABLES3 Variables lb 2 3 4 5 6 7 8 9 MLAT I (1) 1. 00 II (2) . 38 1.00 III (3) .34 .51 1. 00 IV (4) .41 .51 .46 1. 00 V (5) .50 .39 .41 .45 1. 00 Total (6) .72 .70 .77 .77 .70 Henmon-Nelson Q (7) .40 .47 .47 .47 . 37 . 59 1.00 V (8) .32 .54 . 58 . 53 .38 .64 .53 1.00 Total (9) .41 . 58 . 60 .57 .43 .71 .83 .91 1.00 aAll coefficients were significant at .01 level or better. ^Numbers in column headings represent items bearing corres ponding (parenthetical) numbers in stub. Intercorrelations of the Dependent Variables The criteria used in the present project to measure success in the study of descriptive linguistics were the grades assigned by instructors and the staff evaluations. The courses in which grades were obtained were Phonetics, Phonemics, Morphology and Syntax, and a course entitled Field Problems. Field Problems is not distinctively a linguistic course but embraces elements of anthropology and literacy as they relate to the application of linguis tics in the field situation. The major criterion measures were the linguistics G.P.A. (obtained by averaging the grades received in Phonetics, Phonemics, and Morphology and Syntax) and the staff evaluations. According to the statistics presented in Table 8, the highest intercorrelations between the dependent vari ables were found between linguistic G.P.A. and the individ ual linguistic courses. This was to be expected since the G.P.A. represented a composite score for the three linguis tic courses. In order partially to offset these spuriously high correlations, a correction formula was used to obtain estimated correlations between each linguistic course grade and the total of the other grades in linguistic courses TABLE 8 INTERCORRELATION MATRIX FOR THE DEPENDENT VARIABLES Dependent Variables la Phonetics (1) 1.00 .76 .58 .24 .87(.71)b .59 Phonemics (2) 1.00 .73 .36 .93(.83)b .73 Morphology and Syntax (3) 1.00 .37 .86(.70)b .71 Field Prob- (4) 1.00 .36 .36 lems Linguistic (5) 1.00 .76 G.P.A. Staff Evalu- (6) 1.00 ations aNumbers in column headings represent items bearing cor responding (parenthetical) numbers in stub. Coefficients in the parentheses represent estimated cor relations between each linguistic course grade and the total of the grades in the other linguistic courses. Correction Formula (41): r = rtp * t ~ d p . __ _______ 7^ < r z2 + <r2 ■ 2 r (T <r r vt up tp t p p = part score t = total score q = t - p (see Table 8). The smallest intercorrelations were noted for Field Problems and the linguistic courses. Field Problems, it must be remembered, is not a linguistic course per se but may be considered in the general classification of anthro pology and similar studies. It would not, therefore, produce correlations with the other courses as high as those found between one type of linguistic course and another. Validity Coefficients The correlation coefficients obtained between the independent and the dependent variables indicate the degree of relationship existing between the individual predictors and each criterion measure. These coefficients are pre sented in Tables 9, 10, and 11. The relationships existing between the independent and dependent variables were exam ined according to the grouping of the independent vari ables: biographical items, Kuder Preference Record scores, and cognitive measures. Biographical Items Two of the three biographical items included in the present study showed significant relationship to the 90 TABLE 9 CORRELATION COEFFICIENTS OBTAINED BETWEEN THE CRITERIA AND THE BIOGRAPHICAL ITEMS Significant at .05 Significant at .01 Language Criteria Sex Age Study Phonetics .00 -.24 .20 Phonemics -.03 -.23b . 23b Morphology and Syntax .02 -.25b .18a Field Problems .00 .02 .12 Linguistic -.00 -.27b . 23b G.P.A. Staff Evalua- -.08 -.33b .18b tion TABLE 10 CORRELATION COEFFICIENTS OBTAINED BETWEEN KUDER INTEREST SCORES AND THE CRITERIA Interests (Kuder) C r i t e r i a Phonetics Phonemics Morphology & Syntax Field Problems Linguistics G.P.A. Staff Evaluation Outdoor .01 .01 -.12 -.03 -.04 -.06 Mechanical -.00 .00 -.04 .00 -.02 -.01 Computational .00 .13 .13 .05 .10 • o VO Scientific .06 .15 .01 . 03 .09 .09 Persuasive -.12 -. 16a -.13 -.12 -.15 -.14 Artistic .15 .09 .05 .01 .11 .04 Literary .01 .08 .15 .07 00 o • . 19a Musical . 17a .04 .13 .00 .12 .12 Social Service -.10 -.15 -.02 .00 -.07 . 02 Clerical -.10 -.07 0 . 1 .06 r • o 00 00 0 • 1 Significant at . 05 VO TABLE 11 CORRELATION COEFFICIENTS OBTAINED BETWEEN COGNITIVE VARIABLES AND THE CRITERIA Cognitive Variables Criteria Phonetics Phonemics Morphology & Syntax Field Problems Linguistics G.P.A. Staff Evaluation MLAT Part I Part II Part III Part IV Part V Total 34 61 42 41 43 57 42 46 37 49 45 59 .49 .44 .32 .57 .49 .62 .18' . 21 .22 . 19‘ .21 .26 .47 .57 .41 .55 .51 .66 34 37 34 45 40 51 Henmon-Nelson Q V Total 45 43 50 52 48 57 57 47 58 30 22 29 .57 .52 .62 51 45 54 Significant at .05; all the other coefficients were significant at .01. VO ro 93 dependent variables. Age was negatively correlated with the criteria, while previous years of language study was positively correlated. The sex variable produced a close to zero correlation with all six criterion measures (see Table 9). Discussion of this finding is taken up in the following chapter. The age variable showed the highest correlation with the criterion variables. The validity coefficients, which attained a .01 level in each case, with the exception of Field Problems, were negative, indicating that age appeared here as an adverse factor in the study of linguis tics. The unfavorable influence of age, according to the statistics obtained, was more pronounced for staff evalua tion (-.33) than it was for the linguistic G.P.A. (-.27). The correlations obtained between the amount of previous foreign language study and the criterion measures indicated that training of this nature was an advantage when undertaking courses in descriptive linguistics. As noted in Table 9, the coefficients all reached the .05 level of significance or better. This is consonant with Dunkel's findings of 1948. After researching the litera ture, he concluded that a year or more of study in one foreign language was an asset in learning another (34). 94 Kuder Preference Record Of the sixty validity coefficients obtained from the Kuder interest inventory, only three were sufficiently high to be significant at the .05 level (see Table 10). Persuasive interest and grades in Phonemics were negatively correlated; Musical interest was correlated with grades in Phonetics in a positive direction; and Literary interest was positively correlated with staff evaluation criterion measure. Possible explanations of these correlations could only be conjectured, but attention might be called to the fact that in Phonetics a student becomes involved with learning to record and reproduce different sounds. Thus, whether because of natural sensitivity, training and expe rience with sound patterns, or other factors, the relation ship between Phonetics and Musical interest hardly appeared as a startling find. The sizable difference between the correlations obtained between Literary interest and the two separate criterion measures of success in linguistic study also presented an interesting question upon which to speculate. 95 Cognitive Factors The cognitive measures, which included the parts and totals of the MLAT and the Henmon-Nelson intelligence test, all produced coefficients that, with the exception of Parts I and IV correlated with Field Problems, attained a significance level of .01 or better. These data are summarized in Table 11. A comparative study of the relationships found be tween the cognitive batteries and the two major criteria suggests the relative validity of the two instruments, used separately and alone, for predicting success in descriptive linguistics (see Table 12). When G.P.A. was the criterion of success, the language aptitude battery was revealed as being somewhat more efficient for predictive purposes. With staff evaluation as the criterion, however, the intel ligence measure appeared to be the better instrument. TABLE 12 CORRELATION COEFFICIENTS OBTAINED BETWEEN THE MLAT SHORT FORM, MLAT TOTAL, AND HENMON-NELSON TOTAL AND THE TWO MAJOR CRITERIA ________________Criteria_______________ Linguistic G.P.A. Staff Evaluation Predictors_______________ r r2 r r^ MLAT Short Form .63 .40 .51 .26 MLAT Total .66 .44 .51 .26 Henmon-Nelson Total .62 .38 .54 .29 96 Regression Analysis The examination of the multiple correlation coeffi cients obtained in this research is divided into three sections. First considered are the coefficients derived from the entire battery and from selected combinations of variables for predicting the linguistic G.P.A. Following that, an analysis of the same variables to indicate prob able success in linguistics, as judged by staff evalua tions, is undertaken. Finally, the coefficients for the individual courses in linguistics are examined. Linguistic G.P.A. A regression equation was developed for the total group of independent variables (omitting the total scores of the MLAT and the Henmon-Nelson since they were composite scores) and the linguistic G.P.A. (see Table 13). This computation resulted in a multiple correlation coefficient of .77 and a coefficient of determination of .59. A second regression equation was developed to pre dict linguistic G.P.A. by employing the ten Kuder scores and the totals of the MLAT and Henmon-Nelson as the predic tor variables. As shown in Table 14, the multiple R ob tained was .71; the coefficient of determination, .51. 97 TABLE 13 STANDARD SCORE REGRESSION COEFFICIENTS, AND t RATIOS INDICATING THEIR SIGNIFICANCE, FOR EACH INDEPEND ENT VARIABLE USED IN PREDICTING THE TWO MAJOR CRITERION VARIABLES G.P.A. Criterion Staff Evaluation Criterion Variable ____ t ratio t ratioc Biographical Items Sex .02 .22 -.08 -.93 Age -.17 -2.85b -.03 -4.02b Language Study . 10 1.67 .12 1.74 Kuder Interests Outdoor .04 .46 .07 .13 Mechanical .04 .44 .15 1.43 Computational .01 .19 .02 .17 Scientific .17 1. 93 . 23 2.32a Persuasive .02 .29 .04 .34 Artistic .14 1. 88 .20 2.17a —Literary .08 1.13 . 26 3.04b Musical .05 .75 .14 1.92 Social Service .07 .77 .26 2. 52a Clerical .11 .94 .22 1.72 Cognitive MLAT I . 09 1.29 -.02 -.22 MLAT II .23 3. 06b .02 . 28 MLAT III -.10 -1.29 -.05 -.68 MLAT IV .15 1. 98a .12 1.43 MLAT V .17 2.39a .13 1.69 Henmon-Nelson Q .24 2. 88a .29 3. llb Henmon-Nelson V .09 1.14 .13 1.41 Significant at .05 level ■"significant at .05 level 3t ratio: ratio of regression coefficient to standard error of regression coefficient 98 TABLE 14 MULTIPLE CORRELATION COEFFICIENTS AND THE CORRES PONDING COEFFICIENTS OF DETERMINATION FOR PREDICTING THE TWO MAJOR CRITERION VARIABLES BY MEANS OF SELECTED INDEPENDENT VARIABLES Predictors Linguistic G.P.A. Staff Evaluation R R^ R r2 Total Battery (Bio graphical items, Kuder, MLAT, Henmon- Nelson 77 ,59 70 49 Kuder, MLAT Total, Henmon-Nelson Total 71 . 51 63 .40 MLAT and Henmon-Nelson Totals 70 .48 57 .33 Kuder and MLAT Total MLAT Short Form 69 63 .48 .40 59 51 .35 . 26 99 Only the cognitive measures attained an acceptable signif icance level according to their t ratios. The other combinations of selected variables yielded multiple correlation coefficients of similar value. The MLAT and Henmon-Nelson total scores were predictive of G.P.A. to an extent indicated by a multiple coefficient of .70. The Kuder and the MLAT total together produced a multiple R of .69. The Short Form of the MLAT, composed of Parts III, IV, and V, produced a validity coefficient of .63 against the G.P.A. criterion. Although this form of the MLAT was less predictive of G.P.A. than were the selected variables already mentioned, the coefficient was only slightly lower than the .66 figure yielded by the total MLAT battery. These results were comparable to those reported in the manual, which stated that the Short Form appeared to have equal validity with the total battery for use with college populations but was somewhat less valid for intensive courses. A summary of the above statistics is given in Table 14. Staff Evaluation The same combinations of variables serving as pre dictors of linguistic G.P.A. were employed in a similar 100 manner with staff evaluation as the criterion. In every case the validity coefficients were smaller than those obtained in predicting the linguistic G.P.A. The largest multiple R was again derived from the employment of the total number of variables. The multiple correlation coef ficient was .70; the coefficient of determination .49 (see Table 14). An interesting difference between the regression studies for these two criteria, linguistic G.P.A. and staff evaluation, was the comparative contributions of cer tain of the variables. Table 15 presents the variables, singled out from the entire battery, which in the regres sion studies for the two separate criteria were of signif icant value in prediction. The divisions defined in Table 15 afford an imme diate appraisal of the major factors contributing to the multiple R. Age appeared as a handicap with both criterion measures but more so with staff evaluation than with G.P.A. Certain interest areas were of significance with staff evaluation as the criterion, but not with the linguistic G.P.A. The foreign language aptitude battery made a valu able contribution to G.P.A. prediction but not to staff evaluation. The Henmon-Nelson Q score was significant with 101 TABLE 15 STANDARD SCORE REGRESSION COEFFICIENTS, AND t RATIOS INDICATING THEIR SIGNIFICANCE, FOR EACH INDEPENDENT VARIABLE IN THE COMPLETE BATTERY WHICH ATTAINED SIGNIFICANCE LEVEL WHEN PREDICTING THE TWO MAJOR CRITERION VARIABLES FROM THE COMPLETE BATTERY Variables Linguistic G.P.A. j Q t ratios Staff A Evaluation t ratios Biographical Items Age -.17 -2.85b -.03 -4.02b Interest Scales Scientific .17 1.93 .23 2.32a Artistic .14 1.88 .20 2.17a Literary . 08 1.13 . 26 3.04b Social Service .07 .77 . 26 2.52a Cognitive MLAT Part II (Sound- . 23 3.06b .02 .28 symbol association) MLAT Part IV (Grammat■ .15 1. 98a .12 1.43 ical sensitivity) MLAT Part V (Rote .17 2.39 .13 1.69 memory) Henmon-Nelson Q .24 2.88b .29 3. llb Significant at .05 level ^Significant at .01 level 102 both measures of success in linguistic study. The same trend was manifested in the regression study employing the Kuder, the MLAT total, and the Henmon- Nelson total as the predictors. When the staff ratings were the criterion measure, vocational interest variables appeared of more value than the language aptitude vari ables (see Table 16). The measure of "general" intelli gence and the scores obtained on the Literary and Scien tific scales of the Kuder were more indicative of probable future success in descriptive linguistics, judged by staff personnel, than was the foreign language special aptitude test. The correlation coefficient derived from this com bination of variables for predicting the staff evaluation criterion was .63. Other regression studies employed a reduced number of selected variables to predict staff evaluation. They produced the coefficients set forth in Table 14. The MLAT Short Form, comprised of Parts III, IV, and V, correlated with staff evaluation to the extent of .51. The total MLAT (five parts), correlated with the same criterion, produced a coefficient of the same magnitude. When success in descriptive linguistics was judged by staff evaluations, the complete MLAT reflected no apparent prognostic 103 TABLE 16 STANDARD SCORE REGRESSION COEFFICIENTS, AND t RATIOS INDICATING THEIR SIGNIFICANCE, FOR EACH VARIABLE ATTAINING SIGNIFICANCE IN REGRESSION STUDIES EMPLOYING THE KUDER, THE MLAT TOTAL, AND THE HENMON-NELSON TOTAL TO PREDICT THE TWO MAJOR CRITERIA Linguistic G.P.A. Staff Evaluation Criterion Criterion Variables & t ratio t ratio Scientific Interest / .15 1. 63 .25 2.39a Literary Interest .05 .73 . 28 3.18b Social Service Interest .03 .29 .23 2.17a MLAT Total .46 5. 37b .22 2.3la Henmon-Nelson Total .24 2.96b .35 3. 61b Significant at .05 level ^Significant at .01 level advantage over the Short Form. Table 14 serves to illustrate the consistent dif ference found between the sizes of the multiple R's obtained when predicting the two major criterion measures, linguistic G.P.A. and staff evaluation, from the several combinations of independent variables employed in this project. This disparity could have been a reflection of higher reliability in the G.P.A. criterion. Courses in descriptive linguistics are essentially scientific in character and, as such, are subject to a fairly high degree of objective measurement. On the other hand, staff person nel might have deemed some interest variables to reveal in an almost intangible manner the extent to which the indi vidual might be motivated to persistent effort in pursuing the study of descriptive linguistics. Linguistic Courses Coefficients for predicting success in the separate linguistic courses of Phonetics, Phonemics, and Morphology and Syntax were derived from the total number of variables and from the Short Form of the MLAT. The statistics are set forth in Table 17. Phonetics. Fifty per cent of the variance in 105 TABLE 17 CORRELATION COEFFICIENTS AND VARIANCE STATISTICS OBTAINED WHEN PREDICTING INDIVIDUAL COURSE GRADES FROM THE MLAT, THE MLAT SHORT FORM, AND FROM THE COMPLETE BATTERY MLAT MLAT Short Linguistic (Total Test) Form Total Battery Courses r r2 r r2 r r^ Phonetics .57 .32 .53 .28 .71 .50 Phonemics .59 .35 . 56 .32 .71 . 50 Morphology . 62 .38 .63 .40 .74 .55 and Syntax 106 predicting success in Phonetics was accounted for in the battery selected for this study. Of interest was the one variable out of the twenty that produced the most signifi cant t ratio (see Table 18). Part II of the MLAT, designed to measure sound-symbol association ability, appeared to have made the greatest contribution to success in Phonet ics, a course which involves specifically the study of the theory of the formation of sound types. Phonemics. A correlational pattern similar to that seen with Phonetics was evident when attempting to predict success in Phonemics. The multiple R was the same, .71. The difference lay in the individual variables which were significant for the two courses. Table 18 outlines these variables. Scientific and Artistic interests were found to be of major significance when predicting grades in Phone mics, the course in which, among other things, the theory of systems of sound is presented. This is considered the most exacting of the linguistic courses. Morphology and Syntax. The multiple coefficient obtained when predicting grades in Morphology and Syntax from the complete battery was higher than those derived for Phonetics and Phonemics. The most significant contributor to the multiple R (.74) was the Henmon-Nelson Q. TABLE 18 STANDARD SCORE REGRESSION COEFFICIENTS, AND t RATIOS INDICATING THEIR SIGNIFICANCE, OBTAINED WHEN PREDICTING INDIVIDUAL COURSE GRADES FROM THE COMPLETE BATTERY Phonetics______ Phonemics______ Morphology and Syntax Variables A t ratio t ratio A t ratio Age / -.15 -2.30a -.14 -2.19a r -.15 -2.45a Scientific .15 1.51 .30 3.00b .02 .21 Artistic .15 1.67 .24 2.64b .04 .51 Part II .43 5.13b .13 1.58 .05 .62 Part III .02 .22 -.06 -.74 -.20 -2.55a Part IV .00 .08 .18 2.07a .23 2. 84b Part V .12 1.57 .18 2. 26a .16 2. lla Henmon-Nelson Q .18 1.93 .19 2.07a .29 3.31b Significant at .05 ^Significant at .01 107 108 Apparently quantitative ability is of special value in learning the techniques used in the grammatical analysis of a language. Summary. The total battery employed in this project appeared to be an effective instrument for indicat ing probable success in linguistic courses. The multiple correlation coefficients for predicting the grades in Phonetics, Phonemics, and Morphology and Syntax were .71, .71, and .74, respectively. The use of the Short Form of the MLAT alone, however, reduced the validity coefficients to a considerable degree. The coefficients dropped to .53, .56, and .63 for the three courses. The total battery and the Short Form were more efficient in predicting achieve ment in Morphology and Syntax than in Phonetics and Pho nemics. Chapter Summary The burden of the present chapter has been an analysis of the data and of the findings of the present research. For the sake of clarity, the data have been arranged and presented in five distinct sections. The first part was a simple description of the data. This sketched a statistical picture of the variables employed and certain aspects of the sample involved. Wide 1 09 ranges in vocational interests, age, and previous years of language study were evident; whereas, the cognitive ele ments of the battery showed more restrictive ranges. The second section examined the intercorrelations of the independent variables. As would be expected in a battery containing biographical items, noncognitive, and cognitive, factors, the highest intercorrelations were found between the cognitive variables. Most were signifi cant at the .01 level. The intercorrelations of the criterion variables were presented in the third division. Field Problems had the lowest intercorrelations with the other measures. This was the one criterion which was not specifically a measure of performance in descriptive linguistics. Linguistic G.P.A. showed the highest intercorrelations with course grades, but it was this criterion that was a composite of the grades in the other linguistic courses. The correlations obtained between the predictor variables and each criterion measure were considered in the fourth section of this chapter. Age, number of years of previous language study, and all the cognitive variables were significantly related to the criteria. Age and suc cess in linguistic study, as judged by the criteria, were 110 consistently negatively correlated. The fifth division of the chapter undertook an analysis of the multiple correlation coefficients obtained for the whole battery and for selected combinations of the variables (see Table 14). The multiple R's for predicting linguistic G.P.A., staff evaluations, and individual course grades were considered separately. The complete battery yielded a multiple R with linguistic G.P.A. of .77; with staff evaluation, .70. The Kuder and the MLAT and Henmon- Nelson totals produced multiple coefficients of .71 and .63 with the same two criteria. The total scores of the two cognitive batteries, the MLAT and the Henmon-Nelson, yielded coefficients only slightly lower, .70, for G.P.A., but varying more for staff evaluation, .57. The Kuder and the MLAT total produced another near-equal multiple R of .69 to predict G.P.A., but a not-so-large .59 in predicting the staff evaluation criterion. From the Short Form of the MLAT were derived coefficients of .63 and .51 for G.P.A. and staff evaluation, respectively. The entire battery, as well as the selected combinations of variables, was more predictive for linguistic G.P.A. than it was for staff evaluations. In predicting individual course grades, the pattern was the same. Ill A discussion of the findings, which have been out lined in this chapter, and their relationship to the hypotheses formulated earlier is the subject of the follow ing chapter. CHAPTER V SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS Introduction The final chapter for this project concerns itself with a general summary of the research, conclusions that may logically be drawn from the study, and recommendations that may be made with respect to the practical employment of the findings. Suggestions for further research are also set forth. The summary presents a cursory review of the research design employed, including a brief sketch of the sample, predictors, criteria, and procedures. Primary emphasis, however, is accorded the recapitulation of the findings with particular reference to the hypotheses. Summary The purpose underlying the present research was to assemble a battery that will be predictive of success in the study of descriptive linguistics. A valid instrument 112 113 could be used to increase efficiency in the selection and guidance of students for work in this strategic discipline as well as to shed further light on the nature of linguis tic aptitude. The importance of the study is derived in part from the nature of the Summer Institute of Linguistics, under whose auspices the data were gathered, and requests re ceived from other institutions for reliable information to be employed for personnel screening purposes (14,35). Within the past thirty years approximately six thousand individuals have received training at one of the schools of the Summer Institute of Linguistics. Those trained have, for the most part, subsequently employed their skills in the academic world and in service organizations. The Wycliffe Bible Translators, with whom over 1,800 of these students are currently affiliated, is engaged in linguistic research and literacy work in nearly four hundred languages in sixteen different countries. Because of the vast ramifications of such work, there appeared the need of some scientific evidences upon which to counsel prospective descriptive linguists. In order to ascertain some of the fundamental elements which contribute to success in this discipline and thus, by 114 inference, to learn more about the nature of the aptitude involved, a battery of three standardized instruments and three biographical items was assembled. The battery, com prised of the MLAT, the Kuder Vocational Preference Record, the Henmon-Nelson Intelligence Test, was administered to 167 enrollees at the Summer Institute of Linguistics. These independent variables, together with biographical data relating to sex, age, and amount of previous foreign language study, were correlated with the criterion measures of success, instructor-assigned course grades, and staff evaluations. The intercorrelations, validity coefficients, and multiple correlation coefficients were set forth and analyzed in the previous chapter. Summary of Procedures Sample The sample employed in this research consisted of the students enrolled for the intensive course in descrip tive linguistics offered by the Summer Institute of Linguistics during the summers of 1964 and 1965. In 1964 only the S.I.L. students on the campus of the University of Oklahoma were tested. In 1965 the enrollees of the three schools of the Institute, affiliated with the Universities 115 of Oklahoma, Washington, and North Dakota, took part in the study. The group was composed of 62 men and 105 women, yielding a total of 167 students. The average age in years was 26, and the mean number of years of previous foreign language study was 4.6. Predictors The predictor variables were drawn from biograph ical data, an interest inventory, a foreign language apti tude battery, and an intelligence measure. Age, sex, and amount of previous language study; ten interest areas, as assessed by the Kuder Vocational Interest Record; six scores derived from the Modern Language Aptitude Test; and three measures yielded by the Henmon-Nelson Intelligence Test produced a total of twenty-two independent variables. Two of these, the MLAT and Henmon-Nelson totals, were com posite scores and were deleted when the part scores func tioned as predictors. Criteria The twenty-two independent variables listed above were correlated with six criterion measures. Course grades were obtained for each student in Phonetics, Phonemics, Morphology and Syntax, and Field Problems. In addition, 116 evaluations of the individual's probable degree of future success in the field of linguistics were made by staff per sonnel. The average of the several judgments served as one of the major criterion variables. The other major criterion was the linguistic G.P.A., which was derived by averaging the grades received in the strictly linguistic courses— omitting the Field Problems grade. Procedure The data obtained from administering the battery of twenty-two variables to the 167 students enrolled for the intensive course in descriptive linguistics conducted by the Summer Institute of Linguistics were submitted to the IBM 7 094 computer at the Health Science Computing Facility at the University of California, Los Angeles. The resulting simple data description, intercorrelations, validity coefficients, and multiple correlation coeffi cients were analyzed and are reported in the previous chap ter. The following section of the present chapter is devoted to tracing the relevance of the findings to the hypotheses proposed in Chapter I. 117 Summary of Findings Relative to Hypotheses Hypothesis 1 A battery comprised of biographical items, a vocational interest inventory, an instrument for measuring foreign language aptitude, and an intel ligence test will produce significantly high multiple correlation coefficients when used to predict success in descriptive linguistics, meas ured by either course grades or staff ratings. The first hypothesis set forth in this study was that a battery composed of selected biographical data, the Kuder Vocational Interest Record, the Modern Language Aptitude Test, and the Henmon-Nelson Intelligence Test would be predictive of achievement in the study of descrip tive linguistics against two types of criteria, grades and ratings. Instructor-assigned course grades were received in Phonetics, Phonemics, Morphology and Syntax, and Field Problems; but when a linguistics G.P.A. was computed for use as a major criterion measure, the Field Problems grade was omitted. The course was not distinctively linguistic in character. Each student was also rated by two judges on the probability of his future success in the field of lin guistics. The evaluations were made by staff linguists 118 with field experience. The two ratings were averaged to produce the second major criterion, staff evaluations. The complete battery of twenty independent vari ables (the composite scores omitted), derived from cogni tive and noncognitive measures, appeared from the results obtained to be rather highly effective in predicting suc cess in linguistic achievement, judged by either staff evaluations or course grades. The multiple correlation coefficients (.77 and .70) and multiple coefficients of determination (.59 and .49) shown in Table 14 considerably exceed the validity coefficients generally obtained when predicting academic success from intelligence tests, high school grades, aptitude tests, or achievement batteries. Rarely do coefficients of this nature exceed .50 (9,101). Based upon the results obtained with the present sample and the instruments employed, the evidence points to the effectiveness of this battery in predicting success in lin guistic studies. The first hypothesis was accepted. The differential between the multiple R's obtained with the two major criterion measures was not overly large. The reduced efficiency of the battery in predicting suc cess, as judged by staff personnel, might be partially explained by a possible difference in reliability between 119 the two criterion measures. The element of subjectivity in ratings could result in a diminution of reliability. A further distinction between the two major crite rion measures was observed in the degree to which the various independent variables contributed to predictions. The pattern of significant contributors revealed that interests were of more significance than cognitive measures when predicting staff evaluation (see Table 13). Staff evaluations, rather than being a relatively precisely measured course grade, represented the judgment of field experienced instructors in linguistics. The evaluations were unrestricted with respect to the basis for assessment. Thus, they reflected the observed ability of the student to marshal his newly acquired linguistic skills in a manner which might augur well for success on the field. Conceiv ably, Scientific and Literary interest factors might in some way have contributed to the student's approach to linguistic problems. Any attempt to explain the significant contribution made by Social Service interest (see Table 13) entails a review of the nature of the linguistic school in which the present research was conducted. As a service organization, the Summer Institute of Linguistics is primarily interested 120 in training men and women for "pioneer" linguistic work, which in most cases takes the linguist into the remotest villages of the world or to nomadic tribal jungle locations to live with "uncivilized" peoples in order to reduce their unwritten languages to writing, give them literacy mate rials, and train them in the employment of these "civiliz ing" tools. The degree to which the individual was dedicated to such goals could conceivably have been regarded as a vital factor when a staff member was called upon to assess that student's level of probable success in this field. A different measure of success was defined when achievement was judged on the basis of instructor-assigned grades. Courses in descriptive linguistics lend themselves to relatively objective measurement. This fact may pos sibly undergird the contribution to G.P.A. made by the standardized foreign language aptitude battery. When interpreting the value of the correlation coefficients found in this study, the ranges in language aptitude and in intelligence need to be kept in mind. In the course of normal academic screening that takes place as one progresses in school, a gradual restricting of range in ability occurs. The effect of group homogeneity on 121 prediction is generally noted in a smaller validity coeffi cient than when the range is broader. This fact lent added import to the magnitude of the multiple R's of .77 and .70 obtained against G.P.A. and staff evaluations, respec tively. Hypothesis 2 The Modern Language Aptitude Test will con tribute more predictive power to the regression formula than will any other single battery. Almost since the advent of the foreign language aptitude tests, the question as to the relative value of such a specialized battery has been debated. Some researchers have insisted that an instrument for measuring intelligence was as effective in predicting success in a foreign language as was the specialized aptitude measure itself (32). Others have defended the opposite viewpoint. Henmon was among those who believed the special aptitude battery to be the better predictor, but— along with "objec tive measurements in a trial period" (52). The MLAT, based upon a factor-analytical study of the nature of foreign language aptitude, appeared in 1960, offering a partial answer to the need expressed by Henmon. 122 The complete battery, requiring sixty to seventy minutes to administer, provides a brief "trial period" with an artificial language heard over a tape recorder. A Short Form, taking approximately thirty minutes, is also avail able, but it is a paper-pencil form of aptitude test, not involving auditory perception. The complete form was employed in this study in an attempt to ascertain the effectiveness of a foreign lan guage aptitude battery in predicting success in an inten sive course in descriptive linguistics. Computations involving the Short Form were also studied and compared with the complete battery. Shaycoft criticized the MLAT manual for not including such a comparison of data (20). The rationale for employing the longer test when a shorter form is available can only be defended if the validity coefficients are sufficiently higher with the long form and if the deleted parts are not needed for diagnostic purposes. An examination of Tables 9, 10, and 11 indicated that the validity coefficients obtained between the MLAT total and linguistic G.P.A. were the highest attained by any of the variables. The .66 coefficient, significant at better than the .01, indicated that the MLAT total was more predictive of success, as judged by G.P.A., than was any 123 other variable. The MLAT total did not reach the same level of predictive power when the criterion was staff evaluation. Instead, the Henmon-Nelson total produced the highest validity coefficient. With this criterion, the more gen eralized intelligence measure was a better predictor than the special aptitude battery. In the regression studies employing the total bat tery to predict the two major criteria (see Table 13) the t ratios of the standard score regression coefficients indicated the relative power of the several batteries and biographical items in the regression formula. The MLAT contributed the major portion of the weight when predict ing success according to the linguistic G.P.A. criterion. When, however, the criterion measure was staff evaluation, beta weights for the MLAT did not attain significance level. The Kuder and the Henmon-Nelson were the more pre dictive batteries. The hypothesis stating that the MLAT would contrib ute more predictive power to the regression coefficient than would any other battery was rejected at the .01 level when the criterion was staff evaluation. When the depend ent variable, however, was linguistic G.P.A., the 124 hypothesis was accepted. The Short Form of the MLAT was slightly less effec tive than the total five-part battery in predicting success against the G.P.A. criterion {see Table 12). The validity coefficient dropped from .66 to .63. When staff evaluation was the criterion, the coefficient obtained with the Short Form was the same as that yielded by the total MLAT (.51). No significant advantage accrued from the use of the longer five-part battery rather than the shorter form. From the standpoint of economy in time, the Short Form could be substituted for the total. No predictability would be lost with staff evaluation as the criterion and only a negli gible amount with the G.P.A. criterion. Hypothesis 3 There will be no difference in the predictive power of the V score and the Q score of the Henmon- Nelson Intelligence Test. In an extensive research project on foreign lan guage aptitude, conducted in White Plains, New York, the quantitative measures on achievement tests were found to be more predictive of success than were the verbal measures (50). Descriptive linguistics, perhaps even more than 125 traditional foreign language learning, could be related to quantitative ability. As a science of communication, descriptive linguistics employs scientific symbols to refer to other symbols (71). In this there appears to be a natural kinship with the field of science and mathematics. In the first chapter of this work, it was hypothe sized that there would be no difference in the predictive power of the V score and the Q score of the Henmon-Nelson Intelligence Test. Results dictated the rejection of this hypothesis at the .01 level. In regression studies with the two major criteria, the Q score of the intelligence test produced t ratios which were significant at better than the .01 level (Table 13). The V score did not attain significance with either of the criterion measures. Quan titative ability appeared as a critical factor, whether success were judged by G.P.A. or staff evaluations. Hypothesis 4 The ten scales of the Kuder Preference Record— Vocational will not contribute to the predictive power of the battery. The Kuder inventory, an instrument for assessing vocational interest, was included in this study in order to 126 have some measure of the interest aspect of motivation. Interest indicates to a certain extent the goals the indi vidual sets for himself and the degree to which he will expend his energies to achieve them. Gardner and Lambert studied motivational variables as they functioned in second language acquisition and found orientation (direction and intensity) to be of prime importance. The two factors which they defined by factor analysis as the most important in second language learning were linguistic aptitude and motivation, motivation being described in terms of intensity and orientation, or the desire to be like or be able to communicate with those speaking the foreign language (38). Hypothesis four anticipated that the Kuder would not contribute to the predictive power of the total bat tery. The data substantiated this hypothesis when the criterion was linguistic G.P.A. When, however, staff eval uation served as the measure of success, the hypothesis was rejected, for, as may be noted from Table 13, four of the ten Kuder scales attained significance in the multiple R at the .05 level or better. Literary interest was sig nificant at the .01 level; Scientific, Artistic, and Social Service at the .05 level. 127 A more definitive view of the function of the Kuder in this research was observed when the biographical data and separate parts of the cognitive measures were deleted and only the Kuder, MLAT total, and Henmon-Nelson total served as the predictors. Table 16 presents a comparative picture of the variables attaining significance in regres sion studies employing these selected variables to predict the two major criterion measures. It appears rather evident that the Kuder is of value when predicting the student's probable degree of future success in descriptive linguistics as estimated by staff personnel. The contribution made by the Kuder to prediction of course grades, however, seems relatively inconsequential. In research studies cited earlier, interest fac tors, as measured by the Kuder, played a significant supporting role in predicting academic grades. Pimsleur, Mosberg, and Morrison, in 1962, estimated that motivation (of which interest is a major factor) might account for 15 per cent of the variance in foreign language achievement (115). This was not the case in the present study. A clue to the explanation might be found in the nature of the Summer Institute's course of study. It is highly intensive, 128 demanding the student's full attention and time. Carroll found this characteristic to have direct implications with respect to motivation (24) , consequently reducing the amount of variance generally accounted for by this impal pable variable. Crites, through a factor analysis study of voca tional interests, adds another dimension to the possible explanation. He concluded that the effect of interest on motivation was one of kind rather than degree (113) . A student enrolled for an intensive course in a foreign language or descriptive linguistics has, by implication, indicated an interest of that nature. Hypothesis 5 Age will not correlate with achievement in descriptive linguistics. Age has commonly been regarded as a factor in lan guage learning. Younger subjects were considered to have the advantage in learning a foreign language. The small amount of research that has been done in this field has not supported this notion. Carroll cited studies indicating that adults, when they found it necessary, could learn a foreign language as easily as they might have in earlier 129 years (23). The manual for the MLAT makes almost the same statement (26), yet some of Carroll's later studies show a slightly negative correlation between age and success in foreign language study (24). Consonant with the views expressed in most of the studies on the subject, the fifth hypothesis stated that age would not be a significant factor in prediction. The statistical findings presented in Chapter IV indicated the rejection of this hypothesis. The validity coefficients obtained with age and each dependent variable, with the exception of Field Problems, were negatively correlated at the .01 level or better. This variable also made signifi cant contribution to the multiple R's. Age consistently appeared as a greater hindrance to success in linguistics when the criterion was staff evaluation rather than G.P.A. Added import is given to the findings of this research with respect to the function of age in the study of linguistics in that age was shown not to be a deterrent to success in the nonlinguistic Field Problems course. The correlations of age with the cognitive vari ables displayed a uniformly negative trend although signif icance was reached only with the MLAT Part V (.01) and the total (.05) (see Table 4). The speed element present in 130 the MLAT and Henmon-Nelson no doubt handicapped the older persons. Age decrements are particularly found in timed tests (1). The time factor may also be related to the grades obtained in course work. Memory and perceptual aspects of performance# which are tapped in varying degrees by the several parts of the MLAT# are likewise subject to age impairment (1). Both of these abilities are inherently necessary for work in descriptive linguistics. The discrepancy noted between the findings of this research and other studies cited with respect to the age factor could be due to the age level of the groups studied. Most predictive studies in foreign language learning have been conducted in school situations in which age differ ences are relatively small; and few, if any, subjects are over twenty years of age. The present sample, however, afforded an opportunity to study this factor in relation to success in linguistic study over a broad age spectrum. The mean age of the 167 men and women was 26 years, with a standard deviation of 5.7. This tentative explanation of the difference between previous studies and the current research finds reinforcement in the 1962 project conducted by Carroll. His sample was not the "average” school group; 131 the mean age was 34 years with a standard deviation of 5. In this study Carroll found age slightly negatively corre lated with success in foreign language learning. In the present study, age was more than "slightly" correlated with success in descriptive linguistics; it was negatively cor related at the .01 level or better with all the linguistic criteria. Hypothesis 6 Sex will not correlate with success in lin guistic study. Numerous studies have considered the relationship of sex to the various communication skills. In verbal fluency females have evidenced superiority (1). The same is true for foreign language study (28) . Carroll found, too, that girls made higher scores on the MLAT (26). In verbal reasoning, however, males have been shown to excel slightly (111). The several types of communicative skills required in the study of descriptive linguistics and the inadequacy of the literature with respect to the sex factor formed the basis for hypothesizing that sex would not be a significant factor in predicting success in linguistic study. The 132 statistical findings relevant to this variable substanti ated the hypothesis. Sex was not significant in predicting any one of the criterion measures and did not add any appreciable amount of value to the multiple R's. Hypothesis 7 Amount of previous language training will not correlate with achievement in the study of descrip tive linguistics. Amount of previous foreign language training has been shown in other research to be of positive value in learning another tongue (34,26). The cause is not clear. The selective process, in which one continues choosing those courses in which he has had previous success and en joyment, is a possible factor. The transfer effect of applying in a new situation previously acquired efficient methods of learning a language may be another. The seventh hypothesis set forth in the present project stated that the amount of previous foreign language would not be of significance in the study of descriptive linguistics. The validity coefficients obtained between previous language study and G.P.A. was .23; staff evalua tion, .18 (see Table 9). The hypothesis was rejected at 133 the .01 level when the criterion was linguistic G.P.A. and at the .05 level when staff evaluation was the criterion. Although the findings of this research still do not give indication as to the cause of the demonstrated posi tive effects of previous language learning, an interesting relationship was noted between this variable and all but two of the cognitive variables. With the MLAT Part II and total and the Henmon-Nelson V score and total, the Pearson r's were significant at the .01 level; with the MLAT Parts I, III, and IV, at the .05 (Table 4). The correlation be tween previous language study and language aptitude may be interpreted according to the tentative suggestions outlined at the beginning of this section, but the meaning of the correlations between this variable and the other cognitive measures is not as readily discernible. Could the selec tive process with reference to foreign language study be to some extent a function of intelligence level? Amount of previous language study did not attain significance level in the regression studies predicting either of the major criteria. Used alone, this variable was of value in predicting success; but in the presence of other predictors, it was relatively inconsequential. The contributory factor in language study no doubt functioned 134 in other predictor variables, in all likelihood, the cogni tive measures. Summary Hypothesis 1. A battery composed of biograph- 1. ical items, an intelligence test, an interest inventory, and a foreign language aptitude measure will be effective in predicting success in descrip tive linguistic study against the criteria of course grades and staff ratings. Disposition Accepted 2. The MLAT will be more predictive 2, of success in linguistics than will any of the other batteries or biographical items. Linguistic G.P.A. criterion: accept ed. Staff Evalua tion: rejected at .05 level. 3. There will be no difference in 3. Rejected at .01. the predictive power of the V score and the Q score of the Henmon-Nelson. 4. The ten scales of the Kuder will not contribute to the predictive power of the bat tery. 4. G.P.A. criterion: accepted for all scales. Staff evaluation crite rion: rejected at .01 for Literary interest; at .05 for Scientific, Artistic, and So cial Service. 5. Age will not correlate with achievement in descriptive linguistics. 5. Rejected at the .01 level. 135 Hypothesis Disposition 6. Sex will not correlate with 6. Accepted success in descriptive linguis tics . 7. Amount of previous language study will not correlate with achievement in descriptive linguistics. 7. G.P.A. criterion: rejected at .01 level. Staff evaluation: re jected at .05 level. Conclusions and Recommendations Based upon the findings here set forth, the follow ing conclusions and recommendations are made. Conclusions 1. Success in the study of descriptive linguistics can be predicted to a relatively high degree by means of the battery selected for this project. Approximately 5 0 per cent of the variance in each criterion variable can be accounted for by the instruments employed. apparently is not as effective in predicting success in descriptive linguistics as it is for use with foreign lan guages (20,26). The prognostic value of the intelligence test was almost as high as that for the special aptitude test. 2. The foreign language special aptitude battery 136 3. The quantitative score on an intelligence test is more indicative of success in descriptive linguistics than is the verbal score. 4. The Kuder interest inventory, which has added power to numerous predictive studies, possibly because it taps one aspect of motivation, seems to be of little value in a battery used to predict success in an intensive course in descriptive linguistics where motivation is generally high. 5. A more advanced chronological age appears to be somewhat of a handicap to success in linguistics courses, but the handicap evidences specificity. It does not necessarily generalize to other academic disciplines, as exemplified in the Field Problems course. 6. Sex is not a significantly contributing factor to success in the study of descriptive linguistics. 7. Some previous foreign language training is an asset in learning descriptive linguistics. However, what ever factor is operative to make it so is apparently measurable by other variables included in the battery, per haps the MLAT or the Henmon-Nelson. Recommendations X. Further study is recommended to determine the chronological age above which there appears little likeli hood of success in descriptive linguistics. 2. Age should be considered a significant factor when interviewing prospective students or for later place ment. 3. The Short Form of the MLAT, rather than the total, might well be employed in the testing program with out a serious loss of predictability. 4. It is recommended that a future study employ the Short Form of the MLAT to allow time for some measure of inductive reasoning, a factor Carroll considered of major importance in foreign language learning but which was not assessed to any appreciable degree by the MLAT. 5. It is recommended that a similar study be done with a sample composed of majors in descriptive linguistics enrolled in universities for semester-length, nonintensive linguistic courses. Motivational factors might be found to have more significance in such a setting. 6. A five-year follow-up study should be done to ascertain the validity of the present battery for predict ing long-range success in the actual application of 138 principles learned in descriptive linguistics as a requi site tool for "pioneer" translation work. 7. A five-year follow-up study should be conducted to establish the comparative value of instructor-assigned grades and staff evaluations as adequate criterion measures. 8. A study of a representative number of individ uals of the present sample after an interval of approxi mately five years should be made to obtain an indication of the relative value of the Kuder interest inventory in pre dicting success in the employment of linguistic tools. Interest areas might reveal the extent to which the indi vidual will persevere in various aspects of linguistic research and work. 9. Further research should be conducted to yield additional data on possible cutting scores for the foreign language aptitude test and the intelligence test. 10. It is recommended that similar studies be done in other institutions offering majors in descriptive lin guistics in order to set up local norms. Generalizations drawn from the present study to apply in another setting must be made with caution because of the inherently unique features of the present sample and the intensive nature of 139 the course of study. 11. It should be borne in mind that, while low scores obtained on aptitude and ability measures do not necessarily mean an individual could not learn descriptive linguistics or could not perform acceptably in bilingual education, literacy programs, and other essential tasks which are less scientifically exacting than language reduc tion, they are indicative of the fact that he might never attain a high degree of proficiency in the employment of such skills. It is recommended, therefore, that consider ation be given to such scores, as well as to age, when assignments are made. B I B L I O G R A P H Y 140 BIBLIOGRAPHY 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. Anastasi, Anne. Differential Psychology. New York: The Macmillan Company, 1958. ________ . Psychological Testing. New York: The Mac millan Company, 1954. Andersson, Theodore. "Why I Speack Spanich," Texas Foreign Language Association Bulletin, V (1963), 1-3. Angiolillo, Paul F. Armed Forces' Language Teaching. New York: S. F. Vanni Publishers, 1947. Anisfeld, Moshe and Lambert, Wallace E. "Social and Psychological Variables in Learning Hebrew," Journal of Abnormal Psychology, LXIII (1961), 524- 29. Ashdown, Helen. "Mexico Indigenous Statistics." Memorandum. Mexico City, D.F.: Instituto Linguis- tico de Verano, Feb. 19, 1966. Barrilleaux, Louis E. "High School Science Achieve ment as Related to Interest and I.Q.," Educational and Psychological Measurement, II (1961) , 929-36. Bendig, A. W. "Comparative Validity of Objective and Projective Measures of Need Achievement in Pre dicting Students' Achievement in Introductory Psychology," Journal of General Psychology, LX (1959), 237-43. Bernard, Borislow. "Self-Evaluation and Academic Achievement," Journal of Clinical Psychology, IX (1962), 246-54. Bertine, Florence M. "Means of Predicting Success in First Year College Foreign Language Work," Modern Language Forum, XIII (1928), 10-12. 11. Bible Society Record. Facts and figures about Latin 141 142 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. America. November, 1965. Bligh, Harold F. "Trends in the Measurement of Edu cational Achievement," Review of Educational Research, XXXV (1965), 34-52. Bottke, K. G. and Milligan, E. E. "Tests of Aural and Oral Aptitudes for Foreign Language Study," Modern Language Journal, XXIX (1945), 705-09. Breimeier, K. H. Dean of Students, Concordia Theo logical Seminary. Letter to R. A. Renfer. St. Louis, Missouri, July 8, 1965. Brown, Roger. Words and Things. Glencoe, Illinois: The Free Press, 1958. Buros, Oscar K. Tests in Print. Highland Park, N.J.: Gryphon Press, 1961. ________ (ed.). The Third Mental Measurements Year book. New Brunswick, N.J.: Rutgers University Press, 1949. ________ . The Fourth Mental Measurements Yearbook. Highland Park, N.J.: Gryphon Press, 1953. ________ . The Fifth Mental Measurements Yearbook. Highland Park, N.J.: Gryphon Press, 1959. ________ . The Sixth Mental Measurements Yearbook. Highland Park, N.J.: Gryphon Press, 1965. Carroll, John B. "A Factor Analysis of Two Foreign Language Aptitude Batteries," Journal of General Psychology, LIX (1958), 3-19. ________ . Language and Thought. Englewood Cliffs, N.J.: Prentice-Hall, Inc., 1964. ________ . "Research on Teaching Foreign Languages," in Handbook of Research on Teaching, edited by N. L. Gage. Chicago: Rand McNally and Company, 1963. 143 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. Carroll, John G. "The Prediction of Success in In tensive Foreign Language Training," Reprint from Training Research and Education. Pittsburgh: University of Pittsburgh Press, 1962. ________ . "Use of Modern Language Aptitude Test in Secondary Schools," National Council on Measure ments Used in Education. Sixteenth Yearbook, 1959, pp. 155-59. Carroll, John B. and Sapon, Stanley M. Manual, Modern Language Aptitude Test, MLAT. New York: The Psychological Corporation, 1959. Cherry, Colin. On Human Communication. New York: Science Editions, Inc., 1957. Cronbach, Lee J. Educational Psychology. New York: Harcourt, Brace and World, Inc., 1963. ________ . Essentials of Psychological Testing. New York: Harper and Brothers, 1960. Crosby, Richard C. "Scholastic Achievement and Measured Interests," Journal of Applied Psychol ogy, XXVII (1943), 101-03. Detchen, Lily. "The Effect of a Measure of Interest Factors on the Prediction of Performances in a College Social Science Comprehension Examination," Journal of Educational Psychology, XXXVII (1946), 45-52. Douglas, Josephine and Broom, M. Eustace. "Predict ing Success in the Study of Modern Languages in College," Modern Language Forum, XIV (1929), 9-11. Dowd, Robert J. "With the Technicians," Journal of Higher Education, XXIII (1952) , 327-30. Dunkel, H. B. Second Language Learning. Boston: Ginn, 1948. 35. Fetzer, Arthur G. Foreign Missions Secretary, Bap tist Mid-Missions. Letter to R. A. Renfer. Cleveland, Ohio. July 7, 1965. 144 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. Frandsen, Arden. "Interest and General Educational Development," Journal of Applied Psychology, XXXI (1947), 57-66. Gaier, Eugene L. "Trends in the Measurement of Per sonality," Review of Educational Research, XXXV (35), 63-81. Gardner, Robert C. and Lambert, Wallace E. "Motiva tional Variables in Second Language Acquisition," Canadian Journal of Psychology, XIII (1959), 266- 72. Gleason Jr., H. A. An Introduction to Descriptive Linguistics. New York: Henry Holt and Company, 1955. Gomes de Matos, Francisco. "Linguistics in Brazilian University Curriculum," The Linguistic Reporter, V (1963), 4. Guilford, J. P. Fundamental Statistics in Psychology and Education. 3d ed. New York: McGraw-Hill, 1956. ________ . Psychometric Methods. New York: McGraw- Hill, 1954. Hake, D. T. and Ruedisili, C. H. "Predicting Subject Grades of Liberal Arts Freshmen with Kuder Pref erence Record," Journal of Applied Psychology, XXXIII (1949), 553-59. Hall, Clarence W. "Two Thousand Tongues to Go," Reader's Digest, August, 1959. Harding, Francis D. "Language Aptitude Tests as Predictors of Success in a Six-Month Russian Course," Development Report (AFPTRC-TN-57-86). Lackland Air Force Base, Texas: Air Force Person nel and Training Research Center, 1957. ________ . "Language Aptitude Tests as Predictors of Success in a Trial Mandarin Chinese Course," Technical Memorandum (PL-TM-56-8). Lackland Air 145 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. Force Personnel and Training Research Center, 1956. Harding, Francis D. "Language Aptitude Tests as Predictors of Success in a Trial Russian Course," Technical Memorandum (PRL-TM-56-5). Air Force Personnel and Testing Research Center, 1956. ________ . "Prediction of Performance in Foreign Language Training Courses," Technical Memorandum - (WWROP-TN-60-24). Lackland Air Force Base, Texas: Personnel Laboratory, Air Research and Development Command, 1960. ________ . "Tests as Selectors of Language Students," The Modern Language Journal, XLII (1954), 12-122. Hascall, Edward 0. "Predicting Success in High School Foreign Language Study," Personnel and Guidance Journal, XL (1961), 361-67. Haugen, Einar. "Linguistics and the Wartime Program of Language Teaching," Modern Language Journal, XXXIX (1955), 243-45. Henmon, V. A. C. , et a^L. Prognosis Test in the Mod ern Foreign Languages. New York: The Macmillan Company, 1929. Holland, John L. "The Prediction of College Grades from Personality and Aptitude Variables," Journal of Educational Psychology, LI (1960), 245-54. ________ . "The Predicting of College Grades from the California Psychological Inventory and the Scho lastic Aptitude Test," Journal of Educational Psychology, L (1959), 135-42. Hutchinson, Mark E. "The Wartime Language Program as Related to Postwar Language Teaching," School and Society, LX (1944), 33-36. Huxley, Matthew and Capa, Cornell. Farewell to Eden. New York: Harper and Row Publishers, 1964. 146 57. Hymes, Dell H. Professor of Social Anthropology and Linguistics, Harvard University. Memorandum to Benjamin F. Elson, Director, Summer Institute of Linguistics, University of Washington. Cambridge, Mass., n.d. 58. Indians in the Hemisphere Today: Indianist Yearbook. Mexico, D.F.: Inter-American Indian Institute, 1962, XXII. 59. Kaulfers, W. V. "Present Status of Prognosis in Foreign Languages," School Review, XXXIX (1931), 585-96. 60. Kendall, William E. and Hahn, Milton E. "The Use of Tests in the Selection of Medical Students by the College of Medicine of Syracuse University," (Abstract), The American Psychologist, II (1947), 297. 61. Krug, Robert E. "Over- and Under-achievement and the Edwards Personal Preference Schedule," Journal of Applied Psychology, XLIII (1959) , 133-36. 62. Kuder, Frederic G. Administrator's Manual, Kuder Preference Record, Vocational Form C. Chicago: Science Research Associates, Inc., 1960. 63. Lambert, Wallace E. "Developmental Aspects of Second Language Acquisition," Journal of Social Psychol ogy, XLIII (1956), 83-104. 64. Lum, Mabel N. M. "Underachieving and Overachieving College Students," Journal of Educational Psychol ogy, LI (1960), 109-14. 65. Macapagal, Diosdado. Serving (Introduction). Manila, The Philippines: Summer Institute of Lin guistics, 1965. 66. MacNaughton, Jacquelin and Altenheim, Margarete R. "An Investigation of Prognosis in German," Modern Language Journal, XXXIV (1950), 553-60. 67. McCall, John N. "Trends in the Measurement of 147 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. Vocational Interest," Review of Educational Research, XXXV (1965), 53-62. McDavid Jr., Raven I. "Mencken Revisited," Harvard Educational Review, XXXIV (1964) , 211-25. Michael, William B. "A Short Evaluation of the Research Reviewed in Educational and Psychological Testing," Review of Educational Research, XXXV (1965), 92-99. Middleton Jr., George, and Guthrie, George M. "Per sonality Syndromes and Academic Achievement," Journal of Educational Psychology, L (1959), 66- 69. Miller, George A. Language and Communication. New York: McGraw-Hill Book Company, Inc., 1951. ________ . "Speech and Language," Handbook of Experi mental Psychology, edited by S. S. Stevens. New York: John Wiley and Sons, Inc., 1951. Nelson, M. J., Lamke, Tom A., and Kelso, Paul C. Examiner's Manual: The Henmon-Nelson Test of Mental Ability, College Level. Rev. ed. Boston: Houghton Mifflin Company, 1961. Nida, Eugene A. Learning a Foreign Language. New York: Division of Foreign Missions, National Council of the Churches of Christ in the U.S.A., 1950. ________ . Linguistic Interludes. Santa Ana, Calif.: Summer Institute of Linguistics, 1947. ________ . Toward a Science of Translating. Leiden, Netherlands: E. J. Brill, 1964. Pike, Kenneth L. Language and Life. Santa Ana, Calif.: Summer Institute of Linguistics, 1958. ________ . "Our Own Tongue Wherein We Were Born," Offprint from The Bible Translator, X (1959), 3- 17 (Revised and updated in statistics to January, 1965). 148 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. Pike, Kenneth L. and Pike, Eunice V. Live Issues in Descriptive Linguistics. 3d ed. Santa Ana, Calif.: Summer Institute of Linguistics, 1964. Quesada C., Francisco Miro. Taruka (Preface, Book 1, Primer, Quechua de Ayacucho con traducci<5n al espa'nol) . Republic of Peru: Ministry of Public Education. Pablo AlctJcer H. en colaboracic^n con el Instituto Linguistico de Verano, 1964. Ramsey, Philip P. "Prognosticating Achievement in the Foreign Languages." Unpublished Masters thesis, School of Education, University of Southern Cali fornia, 1949. Renfer, Mary Emma F. "Psychological Tests as Pre dictors of Achievement in a Theological Seminary." Unpublished Masters thesis, Southern Methodist University, Dallas, Texas, 1959. Romney, A. Kimball. "The Kuder Literary Scale as Related to Achievement in College English," Journal of Applied Psychology, 1950, XXXIV:40-41. Roscoe, G. T. Statement by Director of Education for the Australian Government in Papua and New Guinea. Quoted by W. Cameron Townsend, Director, Wycliffe Bible Translators, in general letter. Santa Ana, California, May 6, 1960. Salomon, Ellen. "A Generation of Prognosis Testing," Modern Language Journal, XXXVIII (1954), 299-303. Sapon, S. M. "Work Sample Tests for Foreign Language Prognosis," Journal of Psychology, XXXIX (1955), 97-104. Sawyer, Jesse 0. "Foreign Language Instruction," Review of Educational Research, XXXIV (1964) , 203- 10. Schneiderman, Norma. "A Study of the Relationship Between Articulatory Ability and Language Ability," Journal of Speech and Hearing Disorders, XX (1955), 359-63. 149 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. Summer Institute of Linguistics. Bibliography. Santa Ana, Calif.: Summer Institute of Linguis tics, 1964. ________ . "India Briefing." Grand Forks, North Dakota: Summer Institute of Linguistics, 1965. Mimeographed. ________ . SIL Course Bulletin. Santa Ana, Calif.: Summer Institute of Linguistics, 1965. Super, Donald E. Appraising Vocational Fitness. New York: Harper and Brothers, 1949. Time. Letter, Editorial offices to R. A. Renfer, December 13, 1965. Townsend, Cameron. "The Psychophonemic Method of Teaching to Read, America Indlgena, 1956, XVI: 123-32. Toynbee, Arnold. Letter to Harold Goodall, Director, Instituto Linguistico de Verano, Yarinacocha, Peru, n.d. Uhlinger, Carolyn A. and Stephens, Mark W. "Relation of Achievement Motivation to Academic Achievement in Students of Superior Ability," Journal of Edu cational Psychology, LI (1960), 259-66. Velasco Ibarra, Jose Maria, President of Ecuador. Letter to Director, Instituto Linguistico de Verano, Limoncocha, Ecuador, April 28, 1961. Wallis, Ethel E. and Bennett, Mary A. Two Thousand Tongues to Go. New York: Harper and Brothers Publishers, 1959. Weaver, Kenneth F. "The Five Worlds of Peru," National Geographic Magazine, CXXV (1964), 212-66. Wechsler, David. The Measurement of Adult Intelli gence. Baltimore: The Williams and Wilkins Com pany, 1944. 150 101. 102. 103. 104. 105. 106. 107. 108. 109. 110. 111. Weitz, Henry and Wilkinson, H. Jean. "The Relation ship Between Certain Nonlntellective Factors and Academic Success in College," Journal of Clinical Psychology, IV (1957), 54-60. Welmers, William E. "A Background in African Lan guages." University of California, Los Angeles, 1963-64. Mimeographed. ________ . Professor of Near Eastern and African Languages, University of California, Los Angeles. Letter to Mary Emma Renfer, November 22, 1965. Whorf, John B. Language, Thought, and Reality. Selected writings; John B. Carroll, Editor. New York: John Wiley and Sons, Inc., 1956. Willibrand, W. A. "Oklahoma Indians and the 'Summer Institute of Linguistics,1" The Chronicles of Oklahoma, XXXI:352-59. Wolfle, Dael. "Training," in Handbook of Experimen tal Psychology, edited by S. S. Stevens. New York: John Wiley and Sons, Inc., 1951. Worldwide Evangelization Crusade. World Survey. London: Worldwide Evangelization Crusade, 1961, Section 1. Worrell, Leonard. "Levels of Aspiration and Academic Success," Journal of Educational Psychology, L (1959), 47-54. Wycliffe Bible Translators, Inc. Ethnologue. 6th ed. Santa Ana, Calif.: Wycliffe Bible Transla tors, Inc., 1965. Yum, K. S. "Student Preferences in Divisional Stud ies and Their Preferential Activities," Journal of Psychology, XIII (1942), 193-200. Addendum Bennett, G. K.; Seashore, H. G.; and Wesman, A. G. Differential Aptitude Tests, Manual. New York: 151 112. 113. 114. 115. 116. Psychological Corporation, 1947. Carroll, John B. "Linguistics and the Psychology of Language," Review of Educational Research, XXXIV (1964), 119-26. Crites, John 0. "Vocational Interest in Relation to Vocational Motivation," Journal of Educational Psychology, LIV (1963), 277-85. McCarthy, Dorothea. "Language Development in Chil dren," in Manual of Child Psychology, edited by L. Carmichael. 2d ed. New York: Wiley, 1954. Pimsleur, Paul; Mosberg, Ludwig; and Morrison, Andrew L. "Student Factors in Foreign Language Teaching," Modern Language Journal, XLVI (1962), 160-70. Sawyer, Jesse 0. "Foreign Language Instruction," Review of Educational Research, XXXIV (1964), 203-10.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

A Problem-Centered Approach To Introductory Psychology

PDF

An Experimental Study Of The Effect Of Anxiety And Stress On The Group Intelligence Test Performance Of Elementary School Children

PDF

A Study Of The Effectiveness Of Teaching Methods Of Study To Selected High School Freshmen

PDF

Prediction Of Success In Training Among Electronics Technicians

PDF

Prediction Of Therapeutic And Intellectual Potential In Mentally Retardedchildren

PDF

A Study Of The Relationship Of Temperament Variables To The Ability To Make Certain Judgments Of Emotional Behavior

PDF

Validity Concomitants Of Various Scoring Procedures Which Attenuate The Effects Of Response Sets And Chance

PDF

Immediate Memory And Its Correlates With School Achievement

PDF

Classroom Climate And Pupil Characteristics In Special Classes For The Educationally Handicapped

PDF

The effect of noise on intellectual performance as related to personality and social factors in upper division high school students

PDF

A Factor Analysis Of Some Hypothesized Abilities Of Six-Year-Old Children

PDF

A Study To Compare The Effectiveness Of Individual And Group Counseling Approaches With Able Underachievers When Counselor Time Is Held Constant

PDF

A Study Of Affective Sets: The Effects Of Family And Non-Family Verbal Contexts On Word-Need Stimuli In A Word Association Experiment With Reference To Pleasant And Emotional Tones Of Associated...

PDF

Behavioral Seriousness And Impulse-Control Balance In Delinquency

PDF

An Analysis Of The Role Of The School Psychologist In The State Of California

PDF

The Relationship Of Creative Thinking Abilities To School Achievement

PDF

Anxiety Level And The Repression-Sensitization Dimension In Desensitization Therapies

PDF

An Evaluation Of Orientation Programs In Seven Selected Small Liberal Arts Colleges, Using Student Satisfaction Criteria

PDF

A Projective Look At The Educational History Of Adult Mental Patients

PDF

'Cost' And 'Utility' In The Prediction Of The Successful Junior College Student

Asset Metadata

Creator Renfer, Mary Emma Fewell (author)

Core Title Predicting Success In The Study Of Descriptive Linguistics

Degree Doctor of Philosophy

Degree Program Educational Psychology

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag education, educational psychology,OAI-PMH Harvest

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Lefever, David Welty (committee chair), Jacobs, Alfred (committee member), Meyers, Charles Edward (committee member), Wagner, Elmer E. (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c18-229237

Unique identifier UC11361033

Identifier 6611583.pdf (filename),usctheses-c18-229237 (legacy record id)

Legacy Identifier 6611583.pdf

Dmrecord 229237

Document Type Dissertation

Rights Renfer, Mary Emma Fewell

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA