ABSTRACT THE EFFECTS OF CUING AND REPETITION IN INDUCING AND MAINTAINING BIDIALECTALISM BY Mark Blaine Milkovich In this research Bandura's social learning theory was employed in an effort to induce and maintain functional bi— dialectalism in lower and middle socioeconomic status Chicano children. Uing the variables cuing and repetition, opera- tionalizations of Bandura's attentional and motoric reproduc- tion processes, four experimental conditions were generated: (1) non-cuing and nonimitation, (2) non-cuing and imitation, (3) cuing and nonimitation, and (4) cuing and imitation. Two dependent variables were employed: (1) learning—-the use of the appropriate syntactic structure (the one used by the model) in the presence of the experimental variables; and (2) retention--the use of the appropriate syntax in the absence of the experimental variables. Subjects participated in three treatment sessions. In the first two they were exposed to a model who alternately used Spanish syntactic structures and Phonology or English syntactic structures and regionally standard phonology while employing the English lexicon to describe pictures. Subjects were judged as having employed the appropriate syntactic structure if it matched that of the model. The number of Mark Blaine Milkovich these matches constituted the value of the learning variable. In the third treatment session the subject was judged to have employed the appropriate syntactic structure in the employed Spanish syntax in responding to the model's Spanish phonology or English syntax in response to the model's English phonol- ogy. In this third session the model employed syntactic structures which were the same in Spanish and English. Cuing increased learning but not retention. Imitation failed to increase either learning or retention. The results are dis- cussed in the light of a general pattern of linguistic rule learning. THE EFFECTS OF CUING AND REPETITION IN INDUCING AND MAINTAINING BIDIALECTALISM BY Mark Blaine Milkovich A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Communication 1976 Accepted by the faculty of the Department of Communication, College of Communication Arts, Michigan State University, in partial fulfillment of the require- ments for the Doctor of PhilOSOphy degree. £Wlfiyf$ (ll/6% Direcéor of Difsertation )~’ ‘ Guidance Committee: (8/4323(1§Luflif -1 , Chairman V 26.. g 2! W @ng K %/JM ACKNOWLEDGMENTS A number of peOple made this dissertation possible. My wife Carolyn and daughter Lori accepted my long stays at the office; I think they may have thrived on them. Drs. Katrina Simmons and Eileen Earhart worked on this project as members of my committee at the expense of other pressing obligations. Their contributions were substantial. Two groups of people merit special recognition not just for their help on this dissertation but for all my work at Michigan State and not just for myself but for my classmates as well. Ruth Langenbacher, Barb Haslem, Susie Pavick, and Judy Mead made all our lives easier by willingly doing tasks none of us had the right to expect or demand. If our label, "graduate students," and their label, "secretaries," imply were were not colleagues, there has been some semantic slippage. I only hope they won't object to our presumptious- ness in identifying with them. Professors Erv Bettinghaus and G. R. Miller were consistently fair, demanding, critical, and supportive. They engendered in their students a respect for scholastic quality, a need for productivity, and a dedi- cation to the field of Communication. If I could pay them a greater compliment I don't know what it could be. ii TABLE OF CONTENTS Chapter Page I INTRODUCTION . . . . . . . . 1 Overview . . . . . . . . 1 Background . . . . . . . 2 Strategies for Dialect Change . . . 6 Why Bidialectalism . . . . . 14 The Research Problem . . . . . 17 II METHODS AND PROCEDURES . . . . . 22 Subjects . . . . . . . . 22 Stimuli . . . . . . . . 22 Treatments . . . . . . . 25 Procedures . . . . . . . 27 III RESULTS . . . . . . . . . 31 Findings . . . . . . . . 31 Summary . . . . . . . . 41 IV CONCLUSIONS AND IMPLICATIONS . . . . 43 ”PENDIX O O O O O O O O O O O 5 2 Testing Procedures . . . . . . 52 REFERENCES . . . . . . . . . . 57 iii LI ST OF TABLES Table Page 1. The Effects of Cuing and Imitation on Learning . . . . . . . . . 33 2. The Effects of Cuing and Repetition on Retention . . . . . . . . . 34 3. Effects of Structural Difficulty, Grade, Sex, and 1.0., on Learning . . . . . 34 4. Effects of Structural Difficulty, Grade, Sex, and 1.0., on Retention . . . . . 35 iv CHAPTER I INTRODUCTION Overview This study tests a method for inducing and maintaining functional bidialectalism in children. Children from lower socioeconomic groups, especially those who are members of racial and ethnic minorities, are often faced with a diffi- cult problem. They need to know one version of the English language to perform well in the school and another to func- tion in their homes and neighborhoods. This chapter pre- sents the long range problem these children face and reviews the strategies which have been suggested for dealing with it. Considerable space is devoted to that discussion because any future policy decisions on teaching Standard English to children who now employ Nonstandard English will be greatly influenced by the social issues which surround the suggested strategies. The chapter closes by considering the implications of the two major strategies and discussing the variables which need to be examined prior to successfully implementing one of them, functional bidialectalism. Later chapters discuss the methods and procedures, results, and implications of the research. 1 Background Individuals acquire their speech styles from those around them. If someone sounds like a southerner, it is because he has spent more time interacting with southerners than northerners. These regional language variations first emerged because geographic barriers like mountains, rivers, and distance limited communication between peOple in dif- ferent parts of the country. Over time the language of any group changes and the changes themselves differ from group to group. The less people from different regions talk to one another, the less their language versions can influence one another, and the more dissimilar their speech styles or dialects will become. Regional dialect differences are more interesting than bothersome. Although they result from a lack of communica- tion, they seldom prevent it. Dialects develop not only from the isolation imposed by geographical factors, but also from the isolation imposed by social class, racial, ethnic, and cultural group membership. Mbst pe0ple spend relatively little time interacting in socioeconomic groups much differ- ent than their own. As a result, different versions of the language develop within different social groups. Social dialects not only result from isolation, they also enhance it. This could happen because the group members want it that way ("I'm proud of my heritage and the speech that goes with it. I wish to associate with those who share that 3 pride.") and/or because nonmembers want it that way ("You've got your speech and heritage and I'm not going to let you forget it. I'll let you know when I want to interact with people like you.”). From a purely linguistic perspective, differences among social dialects are superficial (cf. Houston, 1970; Cannon, 1973; Macaulay, 1973); no native speaker of English is linguistically retarded because he_uses one particular dialect instead of another. In discussing different dialects, Marckwardt (1971) said: "No linguist has ever called this into question, but no linguist in his right mind could ever say they have equal prestige" (p. 33). Although there is no single dialect which is the prestige dialect, there are regionally standard dialects which "the majority of educated speakers [in a geographical region] habitually use," (Allen, 1969) and they have the characteristic that speakers of one regionally standard dialect have little trouble understanding speakers of another regionally standard dialect. That is, the syntactic variations among the various standard dialects are minimal. However, prestige is a relative thing; one group's prestige dialect is a non-prestige dialect in another group's eyes. Thus, Negro Nonstandard English, whose existence as a single dialect is probably as mythical as that of a single Standard English dialect, confers status on its users in some quarters but not others. Nonstandard dialects are likely to be maintained because social 4 pressures quietly but firmly push the speaker to employ the prestige dialect of the group whether or not it is a stan- dard dialect. It does not much matter what we label the different dialects or, for the purposes of this paper, that different people speak different dialects; the focus here is on the consequences of those differences. Williams (1970) argues that the socioeconomically disadvantaged are caught up in a poverty cycle that includes developmental, educational, occupational, and economic disadvantages. He points out that while these groups share a number of common socioeconomic problems, a major distinguishing feature of a particular group is their language and the way they use it. Delia (1975) and other Communication scholars have shown that these individuals are employing more than just a dis- tinctive language style, they are employing a distinctive communication style involving-both verbal and nonverbal be- haviors. If the assumption that educational disadvantage leads to occupational and economic disadvantages, there is reason to believe that the communication styles of the socioeconomically disadvantaged may be a cause of their problems. Williams, Whitehead, and Miller (1972) presented video tapes of Mexican-American, Anglo, and Black children from ”middle and low status homes" to teachers and obtained the following results: (1) teachers tended to give global evaluations of language samples along dimensions identified as 5 confidence-eagerness and ethnicity-nonstandardness; and (2) teachers' expectations of children's per- formance in subject matters are partially predict- able upon the basis of language attitudes; the degree of prediction increases when the subject matter area is directly within the language arts (p. 275). Crowl and MacGinitie (1974) examined the direction of such predictions. They tape recorded identically worded answers to typical school questions using six white and six black ninth graders to generate the stimulus tapes. They found ”significantly higher grades were assigned by sixty-two ex- perienced white teachers to the recorded answers when spoken by white students than when spoken by black students" (p. 304). It was not possible to distinguish teacher's suscept- ability to dialect influence on the basis of their "sex, age, years of teaching experience, most frequently taught grade level, or the percentage of black students most frequently taught" (p. 304). Given Delia's (1975) recent evidence that the problem is communication rather than language based, Communication scholars will be required to help solve it. However there are advantages to initially restricting research to language rather than the whole of communication behavior. First, many of the background studies have been conducted for the needed research on verbal communication. And, second, language differences are more obvious and easier to work with than nonverbal differences. Strategies for Dialect Change The research conducted here focuses on the pedagogical rather than the social problems resulting from this discrim- ination. Since the two are so highly interdependent however, it is necessary to discuss both. As Shuy (1973) lists them, three strategies have been pr0posed for dealing with non- standard dialects and resultant discrimination. The first is to eliminate them and replace them with the regionally standard dialect. Even before we acknowledged the prestige factor in dialect differences, this was standard procedure in the classroom. Some argue that it is still the prevailing attitude. ”Unconventional speakers from Huckleberry Finn to the latest ghetto dr0pout have traditionally been treated as linguistic pariahs up with whom no standard speaker should have to put" (Eskey, 1974, p. 769). The second strategy is based on the premise that while it is important for the student to be able to use a standard dialect, there is also a need for the nonstandard dialect. Given this, the bidialectical child could switch back and forth between dialects as he thought the situation demanded. The third strategy advocates changing "the attitudes of the majority to accept linguistic and cultural pluralism" (Shuy, 1973, p. 302), accepting individuals for what they are instead of for their social and cultural group membership. This strategy is more attitude than action (no large scale attempts have been made to modify the majority's position on 7 linguistic pluralism) but it has been a sounding board for arguing against the idea that "the prejudices of middle- class whites cannot be changed but must be accepted and in- deed enforced on lesser breeds" (Sledd, 1969, p. 1309). The arguments for and against each of these strategies are ex- amined below., The original strategy for dealing with non-prestige dialects, eliminating them, is no longer advocated. The space is taken to discuss it here because an important issue in examining the third strategy will be the contention that the dialect elimination strategy has been reincarnated in the second (bidialectalism) strategy. The contentions of the first position were: (1) the language of the disadvantaged child is de- ficient [i.e. an incomplete language system]; (2) the disadvantaged child does not use words properly; (3) the language of the disadvantaged child does not provide him with an adequate basis for (abstract and other) thinking; and (4) to the disadvantaged child, language is dispensable: such children tend to com- municate nonverbally in preference to verbally (Houston, 1970, pp 952-53). These assumptions had been around a long time and they died hard. As late as 1964, Stewart used the term "substandard" in talking about dialects and Brottman (1968, cited in Hall and Turner, 1974) asked if the language of lower socio- economic groups was adequate for learning the concepts necessary for school. For the most part, however, "Miss Fidditch's dream of language intrinsically good" [read superior] (Sledd, 1969, p. 1308) finally succumbed, in the 8 mid 1960's, to the arguments advanced by a number of de- veloping social sciences. I Shuy describes the situation this way. Educational, psychological, and linguistic theory were all in need of a “real world" issue to flex their new theoretic muscles and the funding to make flexing possible. The study of dialect and the plight of Blacks in the ghetto had both needed in- gredients. The research efforts resulted in the overthrow of the dialect elimination strategy, largely on the basis of Chomsky's arguments that all dialects are created equal and, as complete language systems, nonstandard dialects are as completely capable of serving any communication needs of their users as any other linguistic system. Support for this position came from several sources. Labov gt_gl. (1968) demonstrated the equal complexity of the language used by lower socioeconomic class Blacks and middle-class Whites: Sinclair de Zwart (1969) showed how, according to Piagetian theory, language was neither a necessary nor suf- ficient condition for certain mental operations: and the bias built into standardized tests of language abilities was documented. The work of Labov and others in the area is important not because it prevented the elimination of nonstandard dialects, but because it provided the data which can be used to make informed decisions about dialect issues. The elimination strategy which not only sought to get rid of 9 the non-prestige dialect but also replace it with the pres- tige dialect of the region was doomed from the start. The only way people have ever been baptized into new languages is by immersion (Falk, 1973). As long as the dialect being taught in the schools does not meet the child's language needs in the home, there is no way he can give up his non- standard dialect. This point is consonant with what we know about foreign language teaching. As Falk (1973) points out the only way to make a fluent Spanish speaker out of an English speaker has been to embed him in a Spanish speaking community where he will not need or have any use for his English. The best situation the elimination strategy could hOpe for was some dialect leveling. That is, by being ex- posed to the standard dialect, the child incorporates some of its features in his own speech. When speakers of differ- ent versions of a language come in contact with one another, some leVeling occurs. The amount of leveling depends on the amount of contact. Out of the demonstrations of the adequacy of nonstandard dialects as communication systems came the arguments for functional bidialectalism. Functional bidialectalism would allow the child to maintain a nonstandard dialect for use in the home and community and provide a key to social and voca- tional mobility, a standard dialect. At least those are the hOped for results. Matison (1974) points out that teaching a standard English dialect is usually defended on 10 the following grounds: (1) "having control of a Standard English may provide the Opportunity to gain access to certain social groups"; and (2) ". . . learning to speak a standard dialect may be vocationally beneficial and economically ex- pediant" (p. 492). However, Cannon (1973) lists "four huge, presently unresolvable kinds of problems facing the teacher" (p. 385) which are associated with bidialectalism. First, we lack adequate descriptions of the relevant dialects. As it now stands, teachers cannot be sure whether they are ad- monishing the child for his bad Standard English or his good Standard English. Second, until dialect descriptions are obtained there can be no adequate teaching materials for teaching the child the syntactic rules of his nonstandard dialect. Third, until the dialect descriptions and teaching materials are obtained, teachers cannot receive the neces- sary training to make them bidialectical. And, fourth, once these language related problems are solved there are still communication problems which will arise when members of different cultures or subcultures interact. Although Hymes (1974) has dealt extensively with the communication rules followed in different subcultures such as who speaks when, what is apprOpriate to say, etc., these issues have received very little attention in education publications. Even when they are considered (cf. Kocher, 1974), they are treated as linguistic rules, part of the language itself. 11 These four problems are usually acknowledged by the proponents of bidialectalism but they temper criticisms by pointing to advances which have been made on all of them (cf. Shuy, 1973). Opponents of bidialectalism, who argue for the third strategy of leaving the language of the non- standard dialect user alone, offer more vitriolic criticisms: In the annals of free enterprise, the early sixties will be memorable for the invention of functional bidialectalism, a scheme best described by an elderly and unregenerate Southern dame as 'turning black trash into white trash' (Sledd, 1969, p. 1308); or Bidialectalism . . . is a less vague and haphazard continuation of earlier attempts, as old as popular education, to eradicate dialect. And it offers the lower class a traditional choice; covert so that the ongoing social game will be fairer to you. There is no offer to change the rules of the game or its name (O'Neil, 1972, p. 433). They suggest that instead of teaching standard dialects we ought to be persuading the public to accept linguistic pluralism. The foundations supporting such criticism are readily available. As already noted, there is no basis for contend- ing that one dialect is linguistically superior to another. In addition, support for the argument that student-teacher dialect differences cause comprehension problems in the school is withering. Weener (1969) found that middle not lower socioeconomic class children suffered significant com- prehension loss when messages were presented in dialects other than those used in their home. 12 Hall and Turner (1974) offered the following explana- tion for the opposite findings in earlier studies. Typical- ly these studies required children to repeat phrases or sentences and exact repetition is taken as evidence of com- prehension. In encoding the message which was just given, the nonstandard dialect user structures the message accord- ing to the syntactic rules of his own dialect. His repeti- tions are therefore systematically different from the original messages but as the result of encoding rather than decoding (comprehension) processes. Since the lower socio- economic class child has already learned to comprehend the standard dialect it seems more equitable to require non- standard dialect comprehension skills of the middle socio- economic class child than to require lower socioeconomic class children to also acquire standard dialect production skills. As the theory of bidialectalism can be attacked so can its practice. The bidialectalists simply cannot demonstrate that they are teaching anybody a standard dialect. At best, they are able, through such things as long sessions of pattern practice drills, to teach a particular syntactic rule of Standard English and show how it differs from the corresponding rule in the nonstandard dialect. In criticiz- ing such teaching techniques, Sledd (1969) comments: . . . Professor Troike can argue the success of his methods by showing that after six months of drills 13 a little black girl could repeat his hat after her teacher, instead of translating automatically to he hat. Unfortunately, tapes do not record the psychological damage, or compare the effectiveness of other ways of teaching, or show what might better have been learned in the same time instead of learn- ing to repeat his hat (p. 1312). Although no data support Sledd (1969), the idea of psycho- logical damage resulting from teaching a standard dialect is often cited (cf. Chisholm, 1974; Isenbarger and Smith, 1973) as a weakness of the bidialectalism position. The issue is not new: opponents of bilingual education in the 1920's suggested that "children who are instructed bilingually from an early age will suffer cognitive or intellectual retarda- tion in comparison with their monolingually instructed counterparts" (Tucker and D'Anglejan, 1971, p. 491). What seems to happen on both sides of the bidialectal- ism fence is that a data-based argument is pushed aside for a more emotion-packed issue about which there are no avail— able data. In large part it has resulted from the all out effort to win rather than resolve the social dialects con- frontation. As Shuy (1973) describes it: There has at no time been a willingness to accept the motives of the camp whose position was attacked as even potentially sincere. The issue has made enemies irrationally, probably because of the heavily charged emotions involved . . . . Otherwise respect- able scholars have resorted to tactics of name calling, inuendo, wrenching from context, doctored quotations, and selective reading in the attacks on presumably Opposing positions (p. 302-3). As a result of this persuasion campaign strategy: the two themes which permeate the arguments against teaching 14 Standard English are: (1) "it is morally wrong to teach a standard dialect, for it may alienate nonstandard speakers from their subcultures," and (2) "it may be educationally impossible or at least inefficient to teach a standard dia- lect" (Hess, 1974, p. 284). Why Bidialectalism To successfully argue for bidialectalism it is necessary to counter these two arguments as well as the "four huge, presently unresolvable kinds of problems facing the teacher" (Cannon, 1973). Educationally inefficient methods will not be tolerated. The arguable position is not that there can— not be efficient methods but that it is up to proponents of bidialectalism to find them if they are ever to be developed and used. The ability to understand and communicate with another will not guarantee acceptance of the other's world view. Sharing a social dialect is a necessary though not sufficient condition for inducing the acceptance of cultural pluralism. Although the communication rules do differ between subcul- tures as Hymes mentions, there is increasing evidence that all members of our culture may share more of a social inter- action rule system than we usually imagine. The research of Berger and his colleagues (Berger and Larimer, 1974; Calabrese, 1975; Berger g£_al., 1975; and Larimer and Berger, 1974) has shown that even across social classes, we all sequence infor- mation in initial interactions in much the same way. 15 We begin with the exchange of biographical-demographic information which can be used to judge the apprOpriateness of alternative messages for the other. Before long the con- versation shifts to the sharing of attitudes and opinions. As Miller and Steinberg (1975) point out, when we have and employ this sort of information, information about the other as an individual, we are no longer stereotyping. Since stereotyping is a prerequisite for prejudice, overcoming it is a step toward the acceptance of cultural pluralism. The major issue facing bidialectalists is the equity of teaching dialects. Although there are data that nonstandard dialect speakers can comprehend standard dialects, there are no data indicating standard dialect speakers can adequately understand the nonstandard dialect. The nonstandard dialect speaker is exposed to Standard English throughout the educa- tional system; the standard dialect speaker has not been continually exposed to any of the several forms of Nonstand— ard English. The solution suggested by Sledd is teaching Nonstandard English to Standard English speakers. One of the few certainties with respect to dialect change is that the process is truly democratic, the majority rules. Nonstandard dialects cannot be taught to Standard English speakers without teaching standard dialects to Non- standard English speakers. Not only does this idea have at least a chance of succeeding, it is also the most equitable. It appears almost impossible to persuade a white middle-class l6 majority to accept minority group dialects unless they have either learned that dialect or are developing interpersonal relationships with minority group members. In this case, group characteristics such as dialects cease to be the most important items. Either option includes teaching Standard English to Nonstandard English speaking students. If we can develop methods for teaching second dialects, it is likely that teachers will be the only members of the middle-class who will be required to learn Nonstandard English. Although it is not the most equitable solution, many will argue that it is the large economic, occupational, and educa- tional rewards the nonstandard dialect speaker obtains which justify teaching second dialects at all. Since standard dialect speakers will not obtain such large rewards, second dialect learning is, for them, not worth the time expenditure. Although not totally equitable, teaching standard dialects is not as inequitable as Sledd and O'Neil argue given the alter- native. In the past we have argued that teaching standard dialects is justifiable because it provides access to the re- wards allotted the middle-class. Let us shift and examine the harm which presents itself if standard dialects are not taught. Court ordered busing is forcing contact among members of different language groups. Increased contact will produce dialect leveling. The standard and nonstandard dialects will become more alike as a function of greater contact with one 17 another. The nonstandard dialect speaker will lose or modify the speech style we argue is so much a part of his social and cultural heritage and become the "white trash" Sledd's "unregenerate Southern dame" describes. There is evidence that such leveling is occurring now. Seitz (1975) studied the imitative ability of black children in integrat- ed and segregated schools and found that "those who attended an integrated school were both better on standard and poorer on nonstandard sentences than those who attended a segregated school" (p. 217). If the nonstandard dialect is as important to minority group children as we argue it is, we need effi- cient and effective means of teaching these children to dis- tinguish and maintain both standard and nonstandard dialects. The Research Problem The purpose of this study was to determine whether children could learn and retain two syntactic systems and employ whichever'is appr0priate for a particular social setting. The underlying proposition is that if the child uses cues from the social setting to determine the appropriate set of structures and if he can use either set, he will be functional- 1y bidialectical. The theoretic model for this study is Bandura's social learning theory. It posits that much of what we learn can be obtained by observing models perform the behavior. In a review of research on children's learning of rule governed behavior, Zimmerman and Rosenthal (1974) con- clude that "modeling procedures, both alone and in conjunction 18 with other variables, were effective in teaching children drawn from diverse populations to respond according to gen- eralized linguistic rules" (p. 32). A number of the studies they reviewed used Mexican-American children as subjects (e.g. Carroll, Rosenthal, and Brysh, 1972; Rosenthal and Whitebook, 1970; and Harris and Hassemer, 1972). The theory argues that social learning is governed by four processes: (1) attentional processes - awareness of the critical features of the modeled behavior, (2) motoric reproduction processes--imitating the critical features of the modeled behavior, (3) reinforcement processes--rewards offered for the successful imitation of the modeled behavior, and (4) retention processes--integrating and retaining the critical features of the modeled behavior to allow for cor- rect future performance. The function each process serves in the present study is given below. "A person cannot learn much by observation if he does not attend to, or recognize, the essential features of the model's behavior" (Bandura, 1971, p. 6). Until recently this common sense assertion did not receive direct support. How- ever, Yussen (1974) has demonstrated that attention levels influence recall. In his study there was a .79 correlation between frequency of attention and recall and a .55 correla- tion between duration of attention and recall. The correla- tion between frequency and duration of recall was .78. Yessen also found that instructions to remember what was modeled 19 also increased attention and recall. Studies of rule governed behaviors other than those associated with language have also found that the more explicit the information about critical features of modeled behavior the greater the learn- ing (of. Rosenthal and Zimmerman, 1972). On the basis of these studies it is hypothesized that: H1: Cuing (signaling) the critical features of the modeled behaViors will significantly increase retention. Cuing (signaling) the critical features of the modeled behaviors will significantly increase learning. N O. The Operational definitions Of learning and retention are made clear in Chapter III. For now it is important to note that learning refers to correct usage of linguistic struc- tures modeled during two treatment sessions, while retention refers to correct usage of those structures in the absence of modeled behaviors on a day following the second treatment session. The second process, motoric reproduction, has been found effective in teaching some syntactic structures. Whitehurst, Ironsmith, and Goldfein (1974) presented two groups of four and five year Old children with pictures. In the modeling condition the experimenter described twenty pictures using passive sentences and the children were asked to describe another twenty pictures. In the control con- dition the child was presented with twenty pictures and asked to describe them all. The experimenter did not 20 describe any pictures in this condition. All children in the modeling condition produced passives while none of the children in the control condition did so. A comprehension test indicated that the modeling procedure provided the greatest increments in comprehension Of that structure. More direct evidence of learning facilitation through motoric reproduction comes from a study by Whitehurst and Novak (1973). The study has a limited utility here, how- ever, because the pre-school aged subjects went through four training sessions a week "over a number of weeks." In the modeling condition the experimenter described training pictures using participial, prepositional, appositive, or infinitive phrases. The subjects then described a number of probe pictures. In the imitation condition the subjects performed an additional task; they were rewarded for exact repetitions of the experimenter's description Of the train- ing pictures. Imitation was effective in inducing production of modeled structures by all subjects for all phrase types. In the modeling only condition some subjects successfully produced some phrase types. Although no studies could be found which examined the relationship between imitation and retention, the studies which examined imitation and learning and Bandura's theoretic position on the relationship between retention and learning suggest the following hypotheses: H3: Imitation will significantly increase retention. 21 H4: Imitation will significantly increase learning. There are a number of other variables which may influ- ence the effects of the treatments. They include age (grade), sex, and intelligence. Because the study is de- signed to test a method which has to work for all children, there are no hypothesized relationships between these vari- ables and the two dependent variables. The following chapter describes the Operationalizations of the variables, data collection procedures, and data analysis techniques involved in testing the hypotheses. CHAPTER II METHODS AND PROCEDURES Subjects One hundred thirty-three first, second, and third graders, approximately 46 from each grade, acted as sub- jects for the study. They were drawn from schools in two Salt Lake City, Utah school districts. It was necessary to draw from two districts in order to assure a reasonable socioeconomic mix within the sample. This was the only pro- cedure which could assure such a mix since the school dis— tricts would not allow ascertainment of the child's socio- economic background. One district served an older section of town. The classrooms from.which these children were drawn were supported by Title I and Follow Through programs. The other school district serves a new suburban, middle-class pOpulation. Equal numbers of students were drawn from each district. All of the children were exposed to Spanish in the home and the vast majority were exposed to Spanish in the schools through either teachers' or teacher aides' usage. Stimuli During each of the first two days of the experiment sub- jects were presented with twenty training pictures of children 22 23 engaged in some action. These pictures were taken from children's coloring books and presented in notebooks. The presentation of these pictures was accompanied by a tape recorded description of the action using Spanish phonology and syntactic structures or the regionally standard English phonology and syntactic structures. All such descriptions were repeated a second time. The English lexicon was used for both types of descriptions. A probe picture depicting a similar but different activity by different children was presented to the subject following each training picture. Using the phonology employed in the description of the train- ing picture, a tape recorded message asked the child to de- scribe the action in the probe picture. The question used was ”What does he (she or they) do?" The syntactic structures used in the study were taken from research conducted by Politzer and Ramirez (1973). They have the following characteristics. First, the use of the Spanish syntax yields ungrammatical English. In other words, the Spanish and English syntactic rules produce dif- ferent surface structures for the same underlying structure. Second, the Spanish syntactic structures are Often used by Mexican-American children in lieu of Standard English struc- tures. The structures selected are presented below. 24 l. The violation of the mandatory subject- verb-Object order Of English when using Spanish syntax. Spanish - "Then he it broke." English - "Then he broke it." 2. The addition of redundant subject pronouns when using Spanish syntax. Spanish - ”My brother he went to Mexico." English - "My brother went to Mexico." 3. The omission of object pronouns when using Spanish syntax. Spanish - "I don't know in English." English - "I don't know it in English." 4. The substitution of the definite article for the possessive pronoun when using Spanish syntax. Spanish - "The apple fell down on the head." English - "The apple fell down on his head." The interference between Spanish and English does not account for all differences between the standard and nonstandard dialect. However, the child's deviations from the standard and nonstandard dialect may also be deviations from appro- priate usage of the nonstandard dialect. Differences based on language interference were chosen to avoid the possibility of teaching structures which are incorrect for both the standard and nonstandard dialects. In addition, these syn- tactic variations cover the four major Options for differences: (1) addition, (2) omission, (3) substitution, and (4) 25 rearrangement. Use of the English lexicon in conjunction with Spanish syntax and phonology constitutes the Operation- al definition of the Nonstandard Mexican-American dialect. The total of forty pictures were grouped into twenty pairs. For each of the four syntactic structures a pair was randomly assigned the Spanish or English syntax as its appro- priate description and the two pictures in the pair were ran- domly assigned the training or probe function. Each picture was used to teach as many structures as possible. The voice for all stimulus tapes is that of a twenty- six year old male Spanish-English bilingual. His native language is Spanish but he has been bilingual since he was four. Radio broadcast training enables him to employ region- ally standard phonology in his English. The data were collected by Caucasian females who were either school district speech pathologists or students in the Speech Pathology and Audiology Division of the Department of Communication at the University of Utah. Caucasian females were used because more women than men teach children this age and because Caucasian teachers are most likely to require a method which like this one is not of their own design. Treatments There were four experimental conditions in the study: (1) cuing and imitation, (2) cuing only, (3) imitation only, and (4) neither cuing nor imitation. Cuing involved 26 signaling both the phonological system to be used in the up- coming description and the phrase which includes the syntac- tic difference between the two languages. Subjects were taught to anticipate the phonological system using the follow- ing method. Subjects in the cuing and imitation or cuing only conditions were told: Sometimes Chicanos and Anglos talk differently. A Chicano would say "Good morning, how are you?" (Spanish phonology) while an Anglo would say "Good morning, how are you?" (English phonology). Let's try another example. A Chicano would say "You have a nice school." (Spanish phonology) while an Anglo would say "You have a nice school." (English phonology). Now I'm going to give you some sentences and you tell me if the person saying them is Chicano or Anglo. Subjects were then presented with nine Anglo and nine Chicano sentences. If they could correctly identify the type Of speakers in all instances they moved on to the next part of the training. If they made an error more example sentences were given and the subjects tested again. This procedure was repeated until the subjects correctly identified the type of speaker for all sentences. Once the subject demonstrated an ability to distinguish English and Spanish phonology, he was taught to associate the cue word "ready" with the former and the cue word "listo" with the latter. This association was formed by telling the subject: When you hear "listo," the man is going to talk like a Chicano; when you hear the word ”ready," the man is going to talk like an Anglo. 27 A series of example sentences were then presented. They were preceded by the cue word apprOpriate for the phonology used. A test sequence of cue words followed by sentences differing in phonology was then presented. The tape was stopped after the cue word had been given and started after his prediction about the phonology of the upcoming sentence. This allowed the subject to verify for himself the correct- ness or incorrectness of his responses. When the child suc- cessfully completed the eighteen sentence phonology identifi- cations this portion Of the training was considered complete. Cuing the syntactic differences was accomplished by raising the volume of the phrase containing the syntactic difference three decibels over the volume of that phrase in the imitation only and neither cuing nor imitation conditions. Training was done by instructing the subjects: Part of this sentence will be said loudly. Pay attention to the loud part so you can repeat it for me. The sentences presented employed both Spanish and English phonology and syntactic structures acceptable in either language. When the subject could repeat the emphasized phrases for the eighteen sentences, irrespective of whether he matched the syntax of the presented sentence, he was con- sidered trained. Procedures All training sessions for the subjects in the cuing con- ditions took place immediately preceding the presentation of 28 the test materials. This required approximately fifteen minutes on the first day of experimental treatment and approx- imately ten minutes on the second day. The subjects were then presented with two sentences de- scribing the training picture and asked, by the tape, to de- scribe the probe picture which was displayed by turning the page. In the cuing conditions the descriptive sentences were preceded by either "listo" or "ready." These tapes were simply dubbed and these two cuing words electronically re- moved for the non-cuing conditions. Electronic means were also used to raise the volume of the portion of the sentence containing the syntactic variation three decibels. In the imitation conditions the child shadowed the tape as the speaker uttered the descriptive sentence the second time. This sequence was followed for twenty training/probe picture pairs on each of two days. The pairs alternated between Spanish and English syntax. On the third day children in all conditions were exposed to the same stimulus. This was a series of questions asking the child to describe the action in forty different pictures each of which was presented twice. In one presentation the question was asked using Spanish phonology, in the other English phonology was employed. The sequence in which the pictures were displayed was determined using a table of ran- dom numbers. 29 Each Of the subjects was exposed to only one syntactic structure over the three days. Subjects were randomly as- signed to both structures and experimental conditions. The women administering the tapes were trained to follow the following procedures with respect to rewarding the child during the treatment sessions. They were to verbally reward the child if and only if the child began to show signs of fatigue or disinterest. Although reinforce- ment has been found effective in teaching linguistic struc- tures through modeling, its presence is not necessary (Zimmerman and Rosenthal, 1974). In this study it presented a complication. The purpose of the study was to train children to talk like the individual they were talking with. An alternative, therefore was to have the child rewarded with both "bueno" and "good." However that could be considered as a cuing stimulus and was therefore not acceptable for the non-cuing conditions. The other alternative, consistently using either ”bueno" or "good," could have reduced the ef- fectiveness of the language differences portrayed in the tapes. This seemed the better alternative but in order to minimize its potential effects the reward "good" was used sparingly and only under the conditions specified above. The data were collected by simultaneously tape record- ing both the stimulus materials and the child's responses to them. Coders later went through the tapes and noted whether the child employed the structure modeled for him and, in the 30 imitation condition, whether the child correctly mimicked the descriptive sentences given. The number of times the child employed the syntax of the model during the first two days is the operational definition of learning. The number of times the child used the syntax appropriately for the phonology of the model on the third day is the operational definition of retention. CHAPTER III RESULTS Findings This chapter outlines the results of this study with respect to two questions. The first question is: What are the effects of cuing and repetition on the learning and re- tention of bidialectalism? In this study learning is Opera— tionalized as the number of times the subject in describing probe pictures employed the syntactic structure matching that modeled in conjunction with the use of the appropriate phonology by the model in his description of the training pictures. That is, although the model employed either Spanish syntax and phonology or English syntax and phonology, the child's behavior was judged as a match if he employed the model's syntax alone. Retention was operationally de- fined as the number of times on the third day of the experi- ment the child's response to the question "What does he (she or they) do?" used the syntactic structure, Spanish or English, which matched the phonology of the model. The second general question is: If the method is suc- cessful, how general is its utility with respect to differ— ent types of children and linguistic structures? For 31 32 maximal utility of the method two conditions have to be met. First, the method must work irrespective of the child's age, sex, and intelligence. Second, the method must work for syntactic structures irrespective of their complexity. To answer the second question, the following data were collected: (1) the child's grade, (2) sex, and (3) teacher's evaluations of academic achievement. For the purposes of the first question all syntactic structures employed in the study were assumed to be Of equal difficulty in spite of the fact that the assumption was known to be false. For the purposes of the second question a measure of syntactic difficulty was obtained by counting the number of times subjects in the re- petition conditions failed, in the forty trials, to correctly mimic the model's descriptions of the training pictures. The mean of the score was taken as the structural difficulty for that structure. Although the first question is primarily theoretic and the second primarily pedagogical, the strong relationship be- tween the two makes it useful to present the analyses done to answer both questions before detailing the results of those analyses. Social learning theory predicts that correct performance of a behavior and the retention of that behavior for use in future situations are positively correlated. Whether or not they were correlated in this data was not only a theoretic but also a methodological issue. If they were correlated, 33 data analytic procedures for multiple dependent variables were required. If not, data analytic procedures for indi- vidual dependent variables were called for. A Pearson cor- relation computed between the learning and retention variable yielded an r = .08. For this reason two-way analyses of variance for single dependent variables were used to assess the effects of cuing and imitation on the learning and maintenance of Spanish and English syntactic structures. Hypotheses 2 and 4 predict that cuing and imitation respectively will significantly increase learning. The re- sults of the two-way analysis of variance used to test these hypotheses are presented in Table 1. Table l. The Effects of Cuing and Imitation on Learning Source of Variation S. S. d.f. M.S. F. Cuing 3.995 1 3.995 1.108 Repetition 13.918 1 13.918 3.859* Interaction .415 l .415 .115 Residual 465.307 129 3.607 *p < .05 The results of this analysis lead to the following conclu- sions: (1) cuing cannot be said to increase learning, and (2) repetition did increase learning. The results of the two-way analysis of variance used to test the retention hy- potheses are presented in Table 2. 34 Table 2. The Effects of Cuing and Repetition on Retention Source of Variation S. S. d.f. M.S. F. Cuing .045 1 .045 .004 Repetition 8.036 1 8.036 .735 Interaction 3.284 1 3.284 .300 Residual 1422.391 129 10.937 The results Of this analysis lead to the conclusion that neither cuing nor imitation can be said to increase retention. A regression analysis was then used to determine the effects of child's age, sex, intelligence, and difficulty of the structure on learning and retention. Here again the lack of correlation between the dependent measures required two analyses. Table 3 contains the contribution Of these predictor variables to the variability in learning. Table 3. Effects of Structural Difficulty, Grade, Sex, and 1.0., on Learning Variable Simple r Multiple R r square r2 Change Difficulty -.04963 .04963 .00246 .00246 Grade .08415 .10208 .01042 .00796 Sex .15241 .17973 .03230 .02188 1.0. -.01618 .17975 .03231 .00001 None of the simple correlations are significant and, with the exception of sex, none of the variables account for even 1% Of the variability in learning. In the overall re- gression the F value for these variables with 4 and 106 35 degrees of freedom was .88479. These degrees of freedom are lower than those in the previous analysis due to casewise missing data Options. Table 4 lists the contributions of structural difficulty, grade, sex, and 1.0. to variablility in retention. Table 4. Effects of Structural Difficulty, Grade, Sex, and I.Q..on Retention Variable Simple r Multiple R r square r2 Change Difficulty -.06432 .06432 .00414 .00414 Grade .09474 .11993 .01438 .01025 Sex .05669 .13041 .01701 .00262 1.0. -.ll3l9 .17346 .03009 .01308 Here again, none of the correlations are significant, and the predictor variables accounted for only minute propor- tions of the variability in retention (F = .82205, d.f. = 4 and 106). These analyses lead to the following three major con- clusions. First, cuing did not increase either learning or retention. Second, repetition increased learning but did not increase retention. Third, the method seemed to Operate independently of the child's grade, sex, 1.0., and the struc- ture's difficulty. These findings raise the questions: (1) why weren't learning and retention correlated?, (2) why did repetition have an effect on learning while cuing did not?, and (3) why did cuing and repetition fail to affect retention? These questions get at the heart of not only 36 what was going on in this study but also what are the inter- relationships among the processes which govern learning from the social learning theory perspective. The answers come from a direct examination of the raw data rather than from any statistical analyses. Let me begin by explaining why there were no retention findings of statistical significance. The first two points to consider are these: (1) the subjects were only exposed to the syntactic differences for ten minutes a day for two days, and (2) a full day elapsed between this exposure and the retention measure. Taken to- gether this means the subjects were each exposed to twenty sentences of each type of syntactic structure and experienced considerable delay before they were able to demonstrate re- tention. The mean score for retention was 40.0 with a small standard deviation of 3.3. On the surface of things this is exactly what is expected by chance so the conclusion that there were absolutely no effects seems reasonable. However, there is an important difference which emerges from a care- ful examination of the data. There are two ways subjects could Obtain the score expected by chance. First, they could consistently employ one structure. Second, they could alternate between the Spanish and English indiscriminately. In other words, they might be aware of a difference but unsure about the relationship between the model's phonology in asking the questions and the two syntactic structures. 37 TO get at this the structure with the least difference between the conditions on both the learning and retention scores was examined. Each of the subjects was dominant in one language or the other as evidenced by their tendency to use one language's syntactic structures more than the other. I then divided the number of matches on the retention task into two groups, those where the child matched using his dominant language's syntax and those where he matched using the second language's syntax. The former was then divided by the latter. This produces a ratio where the larger the ratio the greater the child's reliance on one language in responding to the questions, and the smaller that ratio the smaller the reliance on any one language. These ratios are: Non-Cuing Non-Repetition 5.3 Non-Cuing Repetition 5.2 Cuing Non-Repetition 4.6 Cuing Repetition 4.3 This indicates that in situations where no learning was evidenced there was still a difference in flexibility Of language usage attributable to cuing. To see if this trend held in situations where there was learning, these procedures were applied to the structure where there was the greatest amount of learning on the first two days of the experiment. The ratios for each of the conditions are: 38 Non-Cuing Non-Repetition 2.1 Non-Cuing Repetition 1.5 Cuing Non-Repetition 1.9 Cuing Repetition 3.0 This pattern is much different than the one obtained in the low learning situation but for an unexpected reason. Twenty-nine percent of the subjects in the cuing conditions actually shifted their dominant syntactic structures be- tween the first two days and the third. That is, if they primarily relied on English structures on the first two days, they shifted to Spanish on the third day and vice versa. There is a general pattern in adopting linguistic structures where the child begins by using a structure spar- ingly, he then begins to employ it indiscriminantly, and he cuts back on its usage gradually to finally employ it only in appropriate situations. This argument is supported by the fact that the patterns Of means for the four treatment groups for each of these structures was the same. Based strictly on the number of matches, the cuing groups matched more than the non-cuing groups and the repetition groups matched more than the non-repetition groups. The structures where intermediate amounts of learning were evidenced were then examined. The subjects exposed to these structures relied solely on their dominant language to provide the retention data. This will not make sense until another factor involved in the study is considered. 39 The finding, however, is consistent with the one obtained from examining the two sets of extreme scores. That is, although repetition did not affect retention, cuing did. That effect depended on learning which in turn depended on repetition. This coincides with a develOpmental pattern typically followed by children when they are adopting linguistic structures. First, they employ the new structure hesitantly. Second, this hesitation is lost and they begin to use the new structure indiscriminantly in all situations. Third, they gradually cut back on the usage of the new structure until it is only used in appropriate situations. It appears that for the low learning structure the children got to the first of these steps. In the high learning situation the subjects were performing in concert with the second step. And, none of the conditions was able to bring the subject's behavior to the third stage. To explain the behavior of the subjects exposed to the structures in which there was intermediate learning it is necessary to answer the questions posed earlier: (1) why did repetition effect learning while cuing did not, and (2) why weren't learning and retention correlated? During the first two days of the experiment the repe- tition actually served a cuing function. When shadowing the model it became immediately apparent to the subjects in the repetition conditions that the model was using two 40 different syntactic structures. In the non-repetition con- ditions however no subject had a sequence of four consecu- tive matches before reaching the final six items on either of the first two days. In other words, in the non-repetition conditions even the subjects who knew there were to be dif- ferences in the sentence's phonology and did apparently recognize syntactic differences as evidenced by their use of both earlier on during these two days, did not link phonology and syntax until 70% of the trials on each day had been used up. Shadowing, on the other hand, forced the subjects to make that connection and note the difference they hadn't been told about from the start of the experimental treatments on the first two days. As a result, the subjects in the repe- tition conditions had the automatic advantage of a cuing procedure. The fact that it was cuing and not repetition which in- fluenced retention is reasonable. The accidental cuing associated with repetition during the first two days of the experiment was situation bound. It was the presence of the model's sentences which provided that information. When the third day arrived the repetition subjects showed no evidence of having abstracted and integrated that relation- ship between phonology and syntax. The subjects in the ex- perimentally induced cuing conditions had that abstraction done for them in the form of the cuing training tapes used before the first two day's experimental treatments. They 41 were in a position to apply that knowledge during the third day and apparently did so as evidenced by their increased flexibility in moving back and forth between Spanish and English syntactic structures. Given this, the subjects in the moderate learning situ- ations were faced with the following. Unlike the subjects in the low learning situation, they knew that cuing informa- tion alone was not enough, thus they couldn't just be flex- ible in their selection of structures without knowing that wasn't all there was to it. unlike the subjects in the high learning situation they didn't learn enough from repe- tition to know that their syntax was to match the phonology. It is not unreasonable that they would then place their re- liance on the structures of their dominant language. If it hadn't been the most useful tool in the past it would not be their dominant language. Summary The primary purpose of this study was to test four hy- potheses which focused on how cuing and repetition influ- enced the learning and retention of bidialectalism. Hy- pothesis 1 stated that cuing the critical features of the modeled behaviors will significantly increase retention; the null was not rejected. Hypothesis 2 stated that cuing the critical features of the modeled behaviors will signif- icantly increase learning; the null was rejected. Hypoth- esis 3 stated that imitation will significantly increase 42 retention; the null was not rejected. Hypothesis 4 stated that imitation will significantly increase learning; the null was not rejected. The secondary purpose of the study was to determine how child's grade, sex, and 1.0. as well as the difficulty of learning a particular linguistic structure would influ- ence the learning and retention processes. The results indicated that these variables played extremely small or no roles in the learning and retention process. Given the results of the hypothesis testing, post hoc analyses were used to attempt to understand why no signif- icant differences were found for the hypothesis which focused on retention. That analysis suggested that cuing did play a role in the subjects' retention although this role was not the one it was expected to play. Instead the cuing appeared to induce the linguistic flexibility which can be construed as a logical prerequisite for retention as it was Operation- ally defined. The effects Of cuing were confounded by learning. In the low learning, cuing situation subjects evidenced the greatest amount of flexibility. In the mod- erate learning, cuing situation subjects demonstrated the greatest amount of rigidity. In the high learning, cuing situation subjects were linguistically rigid but employed the linguistic structure associated with the language used least during the two days of experimental treatments. CHAPTER IV CONCLUSIONS AND IMPLICATIONS This chapter discusses the results in terms of their implications for social learning theory, teaching bidialec- talism, and future research in the area. It begins however by considering the conclusions which appear reasonable in the light of the results Obtained in testing the hypotheses. The primary research question in this study was: what effects do repetition and cuing have on the learning and retention of bidialectalism? This question was broken into four parts each of which assumed the form of a research hypothesis. One hypothesis predicted that learning would be increased significantly as a result of repetition. Past studies had shown that repetition would lead subjects to produce modeled linguistic structures. However, none Of the past studies required subjects to produce two structures equally capable Of conveying a single denotative meaning in accordance with extralinguistic rules. The implicit extralinguistic rule in this case consisted of the selection of structures in accordance with the ethnic background of the other. The null hypothesis was rejected in this study. 43 44 The rejection of the null hypothesis is important. It indicates that in using social learning theory to produce bidialectalism, the theory is at least as capable of pro- ducing linguistic changes as it has been in single dialect studies. The evaluations