THE EFFECT GF svmcncAL STRUCTURE on WORD mzmmcmon BY Kmaeaeaarm CHsLnREN Dissertation for the Degree of Ph. D. , MECHIGAN STATE UNEVERSETY DOROTHY GRANSKOG 1974. This is to certify that the thesis entitled The Effect of Syntactical Structure on Word Identification by Kindergarten Children presented by Dorothy R. Granskog has been accepted towards fulfillment of the requirements for Ph.D. degree in Elementary Education flirt % MZA/ Major professor Date November 30, 1973 0-7639 (4'4 *‘ ammo av ' . HOAG & SUN? 800K BINDERY INC. LIBRARY BI NDERS ”MICHEL HIGHER] .Qr / ABSTRACT THE EFFECT OF SYNTACTICAL STRUCTURE ON WORD IDENTIFICATION BY KINDERGARTEN CHILDREN BY Dorothy Granskog This study began with the postulate that reading begins with the basic processes of speech and that these processes are carried over into reading through the use of phrase structure rules. This supposition was oper- ationalized by hypothesizing that nonreading kinder- garteners would learn to identify words as wholes if given an opportunity to attend to the syntactical structure of the sentence, preferably when representa- tive of uncommon usage. Ideas for methodology were developed with E. J. Gibson's theory of perceptual learning in mind. An experimental descriptive study was designed. There were two groups of subjects, twenty in each group. One group had discrimination practice upon uncommon syntactical structure. The other group had discrimination practice upon the same sentences repre- sentative of common usage. Both groups were tested for word identification upon new sentences containing (:7 8”715L\C> Dorothy Granskog previously presented words. Uncommon usage syntactical practice yielded superior word identification. Stu- dent's E, two-tailed test for mean difference, was sig- nificant at the .05 per cent level. Three additional hypotheses of a descriptive nature were also tested. Three unanticipated factors emerged. These were the sex of the subject, the age of the subject, and the sentence administered first in discrimination practice. THE EFFECT OF SYNTACTICAL STRUCTURE ON WORD IDENTIFICATION BY KINDERGARTEN CHILDREN BY 0”“ Dorothy'Granskog A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree DOCTOR OF PHILOSOPHY Department of Elementary and Special Education 1974 ACKNOWLEDGMENTS I wish to express my sincere thanks to my adviser, Professor William V. Hicks, for his interest, encourage- ment, and advice on this study. Thanks also to Dr. Lois Bader, Dr. Gerald Duffy, and Dr. James Snoddy for their suggestions in preparing the proposal and this manuscript. Thanks are extended to Professor Byron Van Roekel for the fine background that I received, and to Dr. Andrew C. Porter and Susan Thrash of the Office of Research Con- sultation for their suggestions with respect to the design and data. Most of all I am grateful for the fine support of my husband, Walfred, and our children--David and Susan, Alice and Alvin, Carol and Dick, Jane, Lois, and young Andrew, who by virtue of his birth, got his mother started on a Ph.D. in the first place. ii TABLE OF CONTENTS Chapter Page I. ANALYSIS OF THE PROBLEM . . . . . . . 1 Need for Study . . . . . . . . . 3 Problem Analysis . . . . . . . . 5 meaning 0 O O O O O O O O O 5 Universals of Language Structure . . 7 Statement of the Problem. . . . . . 10 sumary O O O O O I O O O O 0 13 II. REVIEW OF RESEARCH, THEORETICAL FRAME- WORK, AND DEFINITIONS . . . . . . . 15 Linguistic Consciousness for Gram- matical Units. . . . . . . . . 15 Karpova's Study. . . . . . . . l7 Huttenlocher's Study . . . . . . 20 The Baldwin and Baum Study . . . . 21 Comments . . . . . . . . . . 22 Theoretical Formulations. . . . . . 25 Distinctive Features . . . . . . 26 Invariants . . . . . . . . . 28 Amodal Perception . . . . . . . 30 Definitions . . . . . . . . . . 31 Adult Grammar . . . . . . . . 31 Incorporating Adult and Child Grammars . . . . . . . . . 33 Terms for Psychological Tasks . . . 37 Summary . . . . . . . . . . . 38 iii Chapter III. HYPOTHESES . . . . . . The Issues . . . . . Hypotheses . . . . . Hypothesis (1). Hypothesis (2). Hypothesis (3). Hypothesis (4). Summary . . . . . . IV. METHODOLOGY AND PROCEDURES. Methodology. . . . . Sentences . . . . Design . . . . . Procedures . . . . . Subjects. . . . . Materials . . . . Experimental Session. Pretest . . . . . Discrimination Practice. Trials . . . . . WOrd Identification . Limitations. . . . . Concluding Remarks . . O O O O V. PRESENTATION AND ANALYSIS OF DATA Hypothesis (l) Findings. sex 0 O O O O 0 Age 0 O O O I O Hypothesis (2) Findings. Hypothesis (3) Findings. Hypothesis (4) Findings. Summary . . . . . . VI. SUMMARY AND CONCLUSIONS Summary . . . . . . Conclusions. . . . . iv Page 39 40 42 43 44 45 46 49 50 51 52 56 56 57 59 59 61 62 63 64 66 67 68 70 7O 71 76 78 84 86 88 Chapter Page The Problem . . . . . . . . . 88 Methodology and Hypotheses . . . . 90 mplications. O O O O O O O O O 91 Chapter Summary. . . . . . . . . 95 SELECTED BIBLIOGRAPHY . . . . . . . . . . 97 Table LIST OF TABLES Karpova's data: Three stages . . . . . . Sentences: Discrimination and identification. Scores and means for word identification . . Percentages of subjects identifying words by age . . . . . . . . . . . . . WOrd identifications and word recognitions. . Responding by subjects . . . . . . . . Means for word responses. . . . . . . . Frequencies for word identifications. . . . Frequencies for subjects. . . . . . . . vi Page 19 60 69 72 73 74 74 76 77 LIST OF FIGURES Figure Page 1. Hierarchical organization of the sentence . . 33 2. Generalized tree diagram. . . . . . . . 34 3. Uncommon usage . . . . . . . . . . . 36 4. Interaction with sex . . . . . . . . . 70 5. Amodal perception . . . . . . . . . . 79 6. Amodal perception . . . . . . . . . . 80 vii CHAPTER I ANALYSIS OF THE PROBLEM When a child starts to school he already has considerable command of language. Language exists in speech, and as a result, the proposition that reading begins in speech has been more or less taken for granted in reading readiness programs.1 Only recently has this postulate been of concern. The reason for this state of affairs has been explained by Eleanor J. Gibson as follows: Despite decades of concern on the part of educators, parents, and proponents of homespun wisdom, we seem to know little more about how to teach reading than our great grandparents did. In fact we do not even know why it has to be taught. Why doesn't it just grow, like language? No one teaches a child to speak. we do not know much about how a child acquires Speech either, but in recent years studies of the developmental process have been very instruc- tive. I think the reason for this is that we have begun to look at the process as a piece of natural history somewhat as the ethologist looks at behavior. Observation followed by a careful 1Marion Monroe, Growigginto Readin (Chicago: Scott Foresman and Company,’1 1), pp. 2071 1. analysis may be the essential preliminary to a good theory. Perhaps we have not really tried it with reading.2 Eleanor J. Gibson is actively engaged in research on reading and the developmental process and has been so engaged for some time. As a result she has been able to lay out certain principles of perceptual learning and development.3 Through her research experiences, she states what the relationship between reading and speech is in the following: The origin of reading in speech is obvious. Long before the child goes to school he has learned to segment a sequential stream of acoustical infor- mation; to divide it into valid units of structure; to discriminate these units by means of an economi- cal set of distinctive features; to assign symbolic meanings to units of an appropriate size; to infer the rules that structure the units in permissible ways; and even to recombine units in these rulelike ways so as to produce original messages. Surely this massive achievement must transfer in some way to the perception of written speech, which is also processed sequentially. It, too, must be segmented, discriminated, assigned symbolic meaning, and its combination rules mastered. That there is a carry- over is clear from a comparison of hearing children, who must do without this headstart.4 In these remarks, Gibson gives the basic processes of speech that are pertinent for the origins of reading 7 2Eleanor J. Gibson, "The Ontogeny of Reading," American Psychologist, XXV (1970), l36. 3Eleanor J. Gibson, Principles of Perceptual Learning_and Development (New York: .AppIeton-Century- Crofts, Educational Division, Meredith Corporation, l969), 537 pp. 41bid., p. 136. in speech. She makes no mention of words whatsoever. In another source, she says words "are the constituents of higher order units defined by the rules of grammar."5 With her reference to rules she means phrase structure rules.6 These are the rules that demonstrate how the sentence divides into grammatical units-~or basic sen- tence relations--such as the subject of the sentence, the predicate of the sentence, and the object of the verb.7 WOrds, then, are the end products of a division of the sentence into its grammatical units.8 Need for Study A major portion of the primary grades program of instruction is devoted to beginning reading. As Gibson has already indicated9 no method of instruction has really taken into account how reading begins in speech. Gibson's analysis of the basic processes of speech for the origins of reading has never been put to test. The SIbid., p. 428. 51bid., p. 441. 7Noam Chomsky, Syntactic Structures (The Hague: Mouton, 1957), pp. 26-27} 8Gibson, Principles of Perceptual Learning, p. 432. 9Gibson, "Ontogeny of Reading," p. 136. usual method of instruction sees reading as "talk wrote down"10 in such a way that a word, not the sentence or the grammatical unit, to which the word belongs, is taken as a unit: . . . he--the child--must understand how a printed word is related to a spoken word--that it has the same meaning as a spoken word, only we see the row of printed letters that make up a printed word and we hear the series of sounds that blend into a spoKen word. Each printed word stands for only one spoken word and means just what the Spoken word means.11 With the word as the unit, sentences in beginning reading materials may be too short to permit the use of gram- matical units to induce words by way of the basic pro- cesses of speech. One consequence may be that efficient use of a child's basic processes of speech postulated by Gibson occurs too rarely for instruction to be really effective. The purpose of this study is to show that kinder- garten children learn to identify words as wholes by using phrase structure rules to segment the sentence inco grammatical units, a feat that the child accom- plishes through the use of basic sentence relations-- the subject of the sentence, the predicate of the sen- tence, and the object of the verb. In other words, this loMonroe, Growing in Reading, p. 207. 11Ibid., p. 209. study attempts to operationalize the relationship between speech and reading in terms of Gibson's analysis. However, because of the interpretation of Monroe's analysis, Operationalizing Gibson's analysis places such constraints upon the researcher that the procedures used may not be directly applied to the classroom without further research. Problem Analysis The ability to identify words as wholes seems to be predictive of success in beginning reading instruction. Chall investigated the research literature on "meaning" and "coding" approaches.12 Apparently, she took for granted that the origins of reading in speech had been adequately described by Monroe.l3 As a consequence, she found "coding" approaches superior to "meaning" approaches. This conclusion leaves unexplained why the ability to identify words as wholes should be predictive of success in beginning reading in the first place. Meaning Monroe's remarks in regard to "reading is talk wrote down” fail to indicate what is meant by the term 12Jeanne Chall, Learning to Read: The Great Debate (New York: McGraw-HiIl, Inc., 1967), 362 pp. 13Monroe, Growing in Reading, p. 209. ”meaning." Since her analysis of the.relationship between speech and reading hinges upon "meaning," this is a crucial omission. In the remaining portion of her chap- ter, she encourages teachers to elicit speech from chil- dren and to write down what they say.14 In order to be able to elicit speech, a child must first have something to talk about, a circumstance that often makes it neces- sary for teachers to provide common experiences so that the class does, in fact, have something to say. Such a procedure appears to be a roundabout way of tapping the origins of reading in speech. In her analysis of the origins of reading in speech, Gibson has enumerated the basic processes of speech involved. The children begin by segmenting a stream of acoustical information. Eventually, "symbolic" 15 "Symbolic" meaning is assigned to grammatical units. meaning is the kind of meaning that derives from the sentence taken as a whole. It is that which is assigned to grammatical units when a unit is perceived as a basic sentence relation, such as the subject of the sentence, predicate of the sentence, and object of the verb. Rules are used to rewrite the sentence into its gram- matical components. This division into grammatical 14Ibid., pp. 207-21. 15Gibson, "Ontogeny of Reading," p. 136. units is accomplished without recourse to the thought of the sentence. A term like noun phrase or subject of the sentence carries symbolic meaning because either can apply to a very Specific relation in all sentences-- an economical way of perceiving units. It is clear that this is the explanation for meaning that Gibson has in mind. In the same article she says "The writing-to- Speech code is not a simple matter of paired-associate 16 "It is not obvious that the word if auto- 17 learning." matically a unit for the child." "Simple segmentation of this sort . . . doesn't provide the rules for which 18 "There is carry-over to reading of 19 we are looking." the unit-forming principles of speech." Universals of Language Structure Eleanor J. Gibson has said that the question of how reading originates in speech should be seen as a piece of natural history.20 In recent years a search has been made for uniformities among languages with respect to structure and native language learning. Forty languages representing fourteen different language 16Ibid., p. 139. 17Ibid., p. 140. laIbid. 191bid. 20 Ibid., p. 136. families have been studied.21 These studies have demon- strated the existence of certain uniformities among languages for structure and native language learning. These uniformities are called universals. For example, all of the languages studied had a way of expressing what is meant in English by terms such as noun phrases, verb phrases, subject of the sentence, predicate of the sentence, object of the verb. They all had a way of con- verting a declarative active-voice transitive verb sen- tence into other forms such as the interrogative, imper- ative, passive, or relative clauses.22 For all of these languages, the syntactical structure organized itself into a hierarchy that Operated from the top down--from the largest unit, sentences, to phrases, to subphrases, to words, to word parts--a hierarchy that together with its breakdown into grammatical units is called phrase structure grammar.23 All of the children studied seemed to go through the same set of stages in acquiring their native language. They used word order even when word order was not pertinent in the grammar for the language 21Dan I. Slobin, “Cognitive Prerequisites for the Development of Grammar," in Studies of Child Language Development, ed. by Charles A. Ferguson and Dan Isaac Slobin (New York: Holt, Rinehart and Winston, Inc., 1973), pp. 175-77. 22Dan I. Slobin, Psycholinguistics (Glenview, 111.: Scott Foresman and Company,’197l), pp. 12-17. 23Ibid. being learned. All of the children acquired a set of function classes and also a set of content classes that are needed to carry out syntactical relations. The classes, though, that young children use in speech 24 All of the belonged to child, not adult, grammars. children were capable of saying things that they had never heard. They were not learning their native language by imitating adults. They were imitating in keeping with their abilities to perform. Imitation has been found to be a useful research technique in that it permits control over adult input so that child output can be compared with it.25 The uniformities among languages for language structure and native language learning have indicated that syntax is a connected and organized system for the way words appear in sentences and are combined into grammatical units. In all of the languages there was a surface structure, a form of the language as it is normally represented in print or heard. In all of the languages syntax related the surface structure to pho- nology, and at the same time, it could also break up the 24Ibid., p. 55. 25Dan I. Slobin and Charles A. welsh, "Elicited Imitation as a Research Tool in Develogmental Psycho- linguistics," in Studies of Child Language Development, ed. by Charles A.-Ferguson and Dan Isaac SlObin (New York: Holt, Rinehart and Winston, 1973), pp. 485-96. 10 surface structure into grammatical units so that the underlying meanings could be reached. Breaking the sentence into grammatical units is accomplished through the use of phrase structure rules, but these rules say nothing about the processes whereby children arrive at 26 Immediate memory processes seem to originate 27 meaning. in the syntactical_structure of sentences. In view of the uniformities reported for language structure and native language learning, it has been con- cluded that all languages have the same general defi- nition for form and function and that young children grasp what this definition is about as soon as they learn to speak. Young children handle language in sentences, and because they do, an economy of effort exists through the use of the hierarchical organization of grammatical units that characterizes all sentences.28 Statement of the Problem The uniformities of language structure and of native language learning have indicated that children have child, not adult, grammars. This being the case, it may not be possible to predict what children will do 26Slobin, Psycholinguistics, pp. 12-17. 27Ibid., p. 32. 28Slobin, "Cognitive Prerequisites," pp. 179-80. 11 with the grammar of the writing system from their Speak- ing habits. For example, words that are crucial in operating the phrase structure rules in English are frequently omitted in a child's speech. It may, how- ever, be possible to predict the use of rules even though the output of these rules for children and adults may not be exactly the same. What is needed is the identification and an analysis of the basic processes of speech that pertain to the origins of reading in Speech together with a theory for how perceptual learning takes place. Then a methodology for carrying out research and a way of explaining the results can be developed. A generalized set of rules that incorporates both adult and child grammars and the use of a task that permits control over adult input, so that child output can be compared with it as a child speaks and discriminates sentences in print are required also. This study meets these pre- requisites by using a generalized diagram to define the phrase structure rules in keeping with both child and adult grammars and by using imitation-of-oral reading as a way of providing discrimination practice. This task permits the basic processes of speech to transfer while the child discriminates along the printed sentence. E. J. Gibson's principles of perceptual learning29 were 29Gibson, Principles of Perceptual Learning, 537 pp. 12 used as a theoretical base and her analysis of how read- ing begins in speech.was used to specify the basic pro- cesses of speech that were pertinent for the transfer from speech to reading.30 A study of experimental, descriptive design was drawn up. The researcher prepared a pretest to eliminate children who might already be able to read, prepared a set of active-voice transitive verb sentences and manipu- lated these sentences by exchanging the subjects of the verbs with the objects of the verb so that the pre- dicates would be marked off, a procedure that made it possible to assign a control group to discrimination practice upon sentences representative of common usage and an experimental group to discrimination practice upon sentences representative of uncommon usage.31 For discrimination practice, an imitation-of—oral reading task was devised. Both groups were tested at the end for word identification upon a common set of new sen- tences containing words previously presented during discrimination practice. It was hypothesized that kindergarten children would learn to identify words as a result of their ability to segment sentences into grammatical units, 30Gibson, "Ontogeny of Reading," pp. 136-43. 31Slobin, "Cognitive Prerequisites," p. 201. 13 this ability being predicted to be superior for subjects receiving discrimination practice with sentences repre- sentative of uncommon usage. (In these sentences, a grammatical unit, the predicate, had been marked off by having the subject and the object of the verb exchange places.) Additional hypotheses were needed to explicate in further detail how reading begins in speech. These hypotheses were (1) that an economy of effort would be apparent from attending to the sentence in grammatical units: (2) that the particular words instrumental in using phrase structure rules would be the ones identified most often, and (3) that the perception that occurred during discrimination practice would reflect how the grammar had been attended to. Summer 2 It was noted that beginning reading instruction could be made more efficient if it were known how reading begins with speech. The usual explanation that reading is ”talk wrote down" was contrasted with the more sophis- ticated explanation of Eleanor J. Gibson. The origins of reading in speech were attributed to the use of unit- forming principles-—the phrase structure rules that divide sentences into grammatical units. The source of these rules was found in the uniformities noted for language structure and of native language learning. They are arrived at through the use of basic sentence l4 relations such as the subject of the sentence, the predi- cate of the sentence, and the object of the verb. The fact that child and adult grammars are not the same was seen as the problem. The problem was handled by using a general diagram that incorporated both adult and child phrase structure rules for sentence structure, by using imitation-of-oral reading as a discrimination practice task, and by taking Gibson's principles of perceptual learning and her analysis of those basic processes of speech that were deemed pertinent into account. A description of the design for the study together with the hypotheses were given. CHAPTER II REVIEW OF RESEARCH, THEORETICAL FRAMEWORK, AND DEFINITIONS The postulate is that reading originates with the basic processes of Speech. The processes that are capable of transfer from Speech to print are predicted to be the phrase structure rules that divide and rewrite sentences into grammatical units. This chapter is divided into three sections. The first deals with the research literature on the abilities of the young child to break sentences into smaller units. From this review, it should be apparent whether the sentence or the word is the apprOpriate unit with which to work. The second section presents theoretical formulations, and the third gives the formal definitions for the terms used in this study. Linguistic Consciousness for Grammatical Units Gibson's analysis of the relationship between speech and reading starts with the sentence as the unit. 15 16 Segmenting "a stream of acousticalinformation"l involves the ability to "localize a difference and to give explicit recognition" of boundaries for grammatical units and words.2 This ability to localize a difference and to give explicit recognition of boundaries has been desig- nated as "linguistic consciousness."3 Even though studies dealing with linguistic consciousness are rare,4 it is generally accepted that a young child has this consciousness for basic sentence relations--the subject of the sentence, the predicate, and the object of the verb. Children do not seem to undergo development in this respect. These relations seem to be uniformly present in native language learning.5 Behavioral evi- dence as to how a child breaks up sentences is needed. 1Gibson, "Ontogeny of Reading," p. 136. 2Charles A. Ferguson and Dan I. Slobin, "Segmen- tation," in Studies of Child Language Development, ed. by Charles A. Ferguson and Dan ISaac Slobin (New York: Holt, Rinehart and Winston, 1973): P. 138. 3Ibid. 41bid. 5Roger W. Brown, Courtney Cazden, and Ursula Bellugi, "The Child's Grammar from I to III," in Studies of Child Language Development, ed. by Charles A. Ferguson and Dan Isaac Slobin (New York: Holt, Rinehart and Winston, 1973), p. 307. 17 With such evidence, it should be possible to specify what is available for transfer from speech to print. Kagpova's Study Using Russian children ranging in age from three to seven years, Karpova tried to find out how children break sentences into words.6 He instructed children to repeat the sentence after him and then to tell him "How many words are here? Which is the first word? The second word? The third word, etc.?" Most children, once they understood the instruc- tions, had no difficulty repeating the sentence after him, but sometimes a child would summarize the meaning in one word instead of giving the exact repetition of the sentence as he had been asked to do: Karpova said, "Cold weather came." The child said, "Winter." Some- times the child expanded the sentence by adding an inter- pretation: Karpova said, "Vanya went home." The child gave, "Because it was bad weather.” Sometimes the child repeated a few important nouns from the sentence: Kar- pova stated, "They brought up a kitten and two puppies." The child gave, "Puppy and kitty." 6S. N. Karpova, "Osoznaniye Slovesnogo Sostava Rechi Rebenkom Doshkol'nogo Vozrasta," Vo r Psikhol., I, No. 4 (1955), 43-55, abstracted in D. I. SIoBin, "Abstracts of Soviet Studies of Child Language," in The Genesis of Language: A Psycholinguistic Approach, ed. by F. Smith and G. A. MiIler (Cambridge, Mass.: M.I.T. Press, 1966), pp. 370-71. 18 Three stages were observed in.breaking a sen- tence up into words. They were: 1. The youngest children took the sentence as a unit. Karpova gave the sentence, "Galya and Vova went walking." The child (four years, six months) repeated, "Gayla and Vova went walking." Karpova asked, "How many words are there?" Child, "Two." Karpova, "What's the first word?" Child, "Gayla went walking and Vova went walking." In another example, a seven-year-old child cor- rectly repeated "The boy is laughing" and then reported that there was only one word because "only one boy was laughing." At the next stage, older children began in this fashion, but with repeated questioning, their responses changed. For example, a child, after repeating "Gayla and Vova went walking" stated "There are two words. Vova is one word and Gayla is the other." Karpova interpreted this to be an isolation of nouns and as the first step in breaking the sentence into words. Upon further questioning, children at this level will break the sentence into the subject and the predicate. For example, "Misha ran quickly." "What is the first word?" Misha." "What is the second word?" "Ran quickly.” 19 3. A few of the older children were able to break the sentence into all of its separate words, with the exception of prepositions and con- junctions. But some of these children also broke words into their component syllables at the same time. Karpova took for granted that children had to be taught to break sentences into appropriate units. He presented the sentence orally and had the child move a plastic counter for each word in the sentence, a pro- cedure that is not explained. He organized his data by age. Table 1 presents the percentage of children to be found in each stage. TABLE 1.--Karpova's data: Three stages Percentage of Children in Each Stage EEEEE 3-6 to 5 years 5 to 6 years 6 to 7 years 1 ~74% 45% 20% 2 22% 32% 60% 3 4% 23% 20% Karpova's study demonstrated how a child breaks a sentence that is heard and imitated into words. The process involved increased differentiation within the sentence. Studies on the acquisition of grammar Show 20 that a child's utterances increase in length.7 DeSpite this evidence for additive processes rather than dif- ferentiation, Huttenlocher showed that the acquisition of grammar could not proceed by placing separate words together to make phrases and sentences longer. Huttenlocher's Study Huttenlocher8 assigned four- and five-year-old children to two groups: Those assigned to the first group were instructed to say the last word of a pair first and the first word of the pair last. Those assigned to the second group were told to say the last word of a pair, wait for a tap, and then to give the first word of the pair. Her word pairs fell into five categories: (1) Digits and letters; (2) Like parts of speech such as black-white; (3) Commonly encountered two-word sequences that are not grammatical when reversed such as red apple; (4) Commonly encountered sequences that are grammati- cal when reversed such as you are; and 7Brown, Cazden, and Bellugi, "Child's Grammar," p. 295. 8Janellen Huttenlocher, "Children's Language WOrd-Phrase Relationships," Science, CXLIII (1964), 264. 21 (5) Anomalous pairs that are not encountered in everyday Speech such as table goes. Because her subjects did not know where one word member of the pair ended and the other began, having subjects learn to wait for a tap before giving the second member of the reversed pair facilitated performance. Both groups had extreme difficulty telling where one word ended and the other began for highly practiced and fre- quently heard sequences such as you are. Since her sub- jects could not differentiate between multiple word utterances and single words readily, single words could not serve as separate vocabulary items or as the appro- priate unit for the acquisition of grammar. The Baldwin and Baum Study Baldwin and Baum wanted to delineate some of the underlying units in the language of nursery school children.9 They assumed that the underlying units would be less vulnerable to being split in the middle of vocalization than nonunits, and therefore tried to locate a child's underlying units by interrupting a child's Speech. Their subjects were instructed to repeat a number of sentences after the experimenter, 9Alfred L. Baldwin and Esther Baum, "The Inter- ruptability of Wbrds in the Speech of Nursery School Children," in A Basic Research Program on Reading, ed. by Harry Levin, et aIZ, CRP 639 (Ithaca, N.Y.: Cornell University, 1963). 22 and that when a lighted Santa Claus face went out, to stop speaking whatever they were saying, and to wait until Santa's face was lighted again before resuming the sentence from the precise point of interruption. Twenty- six sentences, consisting of word sequences known to be familiar to the children from previous observation of their daily speech were constructed. Points of inter- ruption were designated to_occur in the middle of the sentence in each case. The children tended to complete the entire phrase or sentence. As was true of Karpova's study, the younger children tended to treat the whole sentence as a unit. Comments These studies Show that in making a transfer from basic processes of Speech to print, the unit is not the word. A child does not seem able to perceive words in oral language unless he first perceives the larger unit, the sentence or phrase to which the words belong. All three of these studies involved imitation. Baldwin and Baum found that children had difficulty localizing word boundaries even though the sentences 10 had been formed from their own language. Karpova's subjects were reported to have had some difficulty loIbid. 23 grasping the instructions.11 Sentences convey thought as well as grammar. Karpova's subjects may have confused the two. Huttenlocher's subjects did better with anomalous sequences like table goes than they did with I O 12 meaningful ones like you are. Anomalous sequences may rule out the possibility of confusing grammar with thought. It is also possible that imitation, comprehension, and production of sentences represent three separate sets of task demands. Fraser, Bellugi, and Brown investigated the claim that an understanding of grammatical contrasts must precede production of them.13 Ten pairs of sen- tences containing grammatical contrasts that could be pictured were constructed. Imitation was operationalized as the correct repetition of the contrasts; comprehension was operationalized as the correct identification of the picture named by a particular grammatical contrast; and production was operationalized as the ability to give contrasting features in sentences for the picture in question. Imitation was easier than comprehension and llKarpova, "Osoznaniye Slovesnogo," p. 370. 12 p. 264. Huttenlocher, "WOrd-Phrase Relationships," 13Colin C. Fraser, Ursula Bellugi, and Roger W. Brown, "Control of Grammar in Imitation, Comprehension, and Production," Journal of Verbal Behavior and Verbal Learning, II (1963), 121-35. 24 comprehension was easier than production. Imitation was seen to be a perceptual-motor task that.relied largely upon control over a highly systematic Speech.system. The task demands of imitation, comprehension, and pro- duction could be differentiated. There are numerous examples that demonstrate how children imitate in keeping with their linguistic com- petence. "Syntactic structures take up space in.memory, and frequently content will be sacrificed to the retention of form in immediate, rote imitation."14 In her analysis of the basic processes of Speech that transfer from Speech to print, Gibson mentioned the use of "symbolic meaning."15 The fact that imitation tasks tap linguistic competence and that imitation demands can be differentiated from those of compre- hension lends support to the view that symbolic meaning might consist of a linguistic consciousness for the basic sentence relations-~the subject of the sentence, the predicate, and the object of the verb. On the other hand, in the studies cited, no support has been found for Monroe's account for the way that reading begins in 14Dan I. Slobin and Charles A. Welsh, "Elicited Imitation as a Research Tool in Developmental Psycholin- guistics," in Studies of Child Language Development, ed. by Charles A. Ferguson and Dan Isaac Slobin (New York: Holt, Rinehart and Winston, 1973), p. 496. 15Gibson, "Ontogeny of Reading," p. 136. 25 speech.16 The children in these studies were not attend- ing to individual words or to their individual vocabulary meanings. Theoretical Formulations E. J. Gibson's theory of perceptual learning states that environment is far richer in potential stimu- lation information than the organism is capable of registering. At first the organism's ability to extract information from environmental stimulation is very crude. With practice, the organism begins to respond to stimuli previously not responded to. The organism responds to differences, points where change can occur. A cor- respondence of increasing specificity between environ- mental stimulation and the organism's perception of it comes with practice. A change in what the organism is capable of responding to occurs. He uses the responses that he already has to do the extracting of information. There is no response Shaping. Reinforcement or knowledge of results are not needed. Because information must be discovered in stimulation, ways of enhancing the stimulus so that the desired information can be picked up are needed. But because the information in stimulation is already structured, pick up of it does not depend upon 6Marion Monroe, Growing Into Reading (Chicago: Scott Foresman and Company, 1951), p. 209. 26 complex thinking processes or the formation of associ- ations. The basic task in perceptual learning is dis- . 0 0 l7 crimination. Distinctive Features Distinctive features are properties of objects and events that differentiate them from one another. A system of distinctive features for distinguishing minimal units of Speech, the phonemes, has been worked out by Jakobson and Halle.18 Gibson has used this system as a "model for attempts to Specify distinguishing pro- perties of other sets of objects and events."19 A set of twelve distinctive features is sufficient to yield unique bundles of features for all of the phonemes in all languages. From this set of twelve, a language- makes a selection. English, for example, uses only nine distinctive feature oppositions to define the twenty-eight phonemes listed by Jakobson, Fant, and 20 Halle. Each distinctive feature opposition represents l7Gibson, Principles of Perceptual Learning, 537 pp. 18Roman Jakobson and Morris Halle, Fundamentals of Language (The Hague: Mouton and Company, 1956), 87 pp. 19Gibson, Principles of Perceptual Learning, p. 83. 20Roman Jakobson, C. Gunnar M. Fant, and Morris Halle, Preliminaries to Speech Analysis (Cambridge, Mass.: M.I.T. Press, sixth printing, 1965), p. 43. 27 two alternatives such as vocalic/nonvocalic; consonantal/ nonconsonantal; continuant/interrupted. Each phoneme has a unique patterning for its bundle of distinctive feature oppositions. All the perceiver really needs to perceive in order to detect a difference capable of dif- ferentiation is "otherness," "contrast," or "different than." Developmentally in the acquisition of these features, differentiation is progressively ordered in hierarchical fashion. There is a stratified splitting of categories. For example, it has been demonstrated that a child may begin with the consonant-vowel (open/ closed) distinction. A distinctive feature opposition for the consonant and another for the vowel may then be chosen. Each of these will split in half in predictable fashion. If nasal/oral is chosen for the consonant, then continuant/interrupted will become the next division.21 Quite apart from speech perception, E. J. Gibson has discussed research evidence with controlled stimuli that shows that the perceptual strategy for discrimination is most efficient when the least frequent distinctive feature is selected, whereas the strategy for identifi- cation is best when the distinctive feature attended to 21Ruth H. Weir, Language in the Crib (The Hague: Mouton, 1962), p. 43. 28 comes closest to splitting the set into a fifty-fifty distribution.22 The demands of discrimination and identification tasks are not the same. Invariants An invariant is a relation, a contrast, that stays the same under many kinds of change such as size or color. A critical dimension must be discriminated. The discovery of critical dimensions is believed to be 23 A relation or contrast becomes an abstraction. abstract from being confronted with many different cases for the contrast in question. Conscious search is not necessarily involved. The most primitive demon- stration of an invariant that is abstract is transpo- Sition; for example, if a pair of hens are trained to peck at grain on the darker of two gray squares, they will continue to peck at the darker even when a new stimulus pair is presented and the originally darker square is now the lighter. Gibson felt that the research literature on I O O O O O 24 tranSpOSition was relevant for phoneme discrimination. 22Gibson, Principles of Perceptual Learning, pp. 122-23. 23Ibid.. pp. 108-11. 24Ibid., p. 110. 29 It may also be relevant for the perception of phrase structure rules. For an invariant to exist, a dimension must be abstracted. This study is concerned with gram- matical syntax as a dimension of language. Phonology is another dimension, the one to which phoneme discrimi- nation pertains. Abstracting a dimension in print cannot occur unless the difference between the two dimensions, grammar and phonology, is discriminated. In Gibson's model for the development of cognitive processes,25 the abstraction of distinctive features and the abstraction of invariants follow parallel lines. Both lines of development feed into the formation of representations, sensory-motor, imaginal, and conceptual. Both arise from the same source--a differentiation of simple patterns and objects from background stimulation. It seems theoretically valid to assume that distinctive features and the grammar in syntax have the same origins also. In this study, it was assumed that the origins of discriminations for phonology and grammar would begin with the selection of a distinctive feature for the phoneme #. (Jakobson, Fant, and Halle have listed # as a phoneme.) It carries the distinctive features of 25Ibid., pp. 160-61. 30 26 It has potential 28 nonconsonantal, nonvocalic, and lax. 27 in grammar. It occurs between grammatical units. Amodal Perception A reader uses his senses to perceive grammatical units in sentences. The writing-to-speech code has evolved from linguistic characteristics of speech over time. Theoretically, there may be a kind of resonance in the nervous system, a tuning of the system to the organization of language sounds. It is known that the distinctive features of speech are such that they cannot be attributed to the domain of any one sensory modality.29 They are representative of several modalities. There is no transfer from one modality to another. What is held in one is latent in another at the same time. The sensory representations in the nervous system become correlated so that they can take over for one another when one group drOps out. Under appropriate stimulus 26Jakobson, Fant, and Halle, Speech Analysis, p. 43. 27Noam Chomsky, Syntactic Structures (The Hague: Mouton, 1957), pp. 38-40. 28Norman C. Stageberg, An Introductory En lish Grammar (New York: Holt, Rinehart and Winston, l , pp. 48-67. 29 Jakobson, Fant, and Halle, Speech Analysis, pp. 43-52. 31 enhancement, the nervous system gets itself attuned to the stimulus input of the writing-to-Speech code. This attuning needs no construct for memory. The information needed is picked up. Perceptual functioning and per— ceptual content, in some sense, begin to mirror one another. The perception involved is amodal, abstract, and relational.30 Definitions This study is experimental and descriptive. Three sets of definitions are given. These three sets are: (1) Those that pertain to adult grammar; (2) Those that demonstrate how adult and child grammars have been incorporated; and (3) Those that pertain to the psychological tasks of discrimination and identification. Adult Grammar 1. Surface structure is that form of language that is normally heard or read. 2. Syntactical and syntax refer to the combining of the subject of the sentence, the predicate 30Gibson, Principles of Perceptual Learning, pp. 218-190 32 of the sentence, and the object of the verb in such a way that a sentence results. Structure in this study is limited to the subject- predicate-object of the verb sentence. Syntacti- cal structure consists of a set of phrase structure rules for dividing the sentence into grammatical units such as Noun Phrase and Nerb Phrase. Phrase structure is a set of rules that is used to rewrite the sentence into its con- stituents. The rules are: gentence as an axiom is given. Pentence is rewritten as Noun Phrase and Yerb Phrase. Noun Phrase is rewritten as 2 (article) and Noun. yerb Phrase is rewritten as Kerb and Noun Phrase. P.is rewritten as 323 or g. Noun is rewritten as boy, girl, ball, etc. yerb is rewritten as hit, etc.31 Hierarchical organization of the sentence is that which puts these rules into tree diagram form. The tree for these rules is shown in Figure l. 31Chomsky, Syntactic Structures, pp. 26-27. 33 S ///§€\\ //////VP\\\\\\\ T N verb //NP\ T N the boy hit the ball Fig. 1.--Hierarchical organization of the sentence 5. A constituent is a unit that can be replaced by a single word without changing the whole structure of the sentence: the boy can be replaced by he and the subject-predicate-object of the verb structure still remains. 6. words are the end products of the hierarchy. They result from using the phrase structure rules. Incorporating Adult and Child Grammars sentence relations. The young child's grammar consists of basic 32 In this study, like the term 32Brown, Cazden, and Bellugi, "Child's Grammar," 306-07. 34 structure, basic sentence relations refer to subject, the predicate, and the object of the verb. The phrase structure rules rewrite the sentence into Noun Phrase and gerb Phrase. In this study, the young child is to arrive at this rewriting by using his awareness of basic sentence relations. A new tree that includes his basic sentence relations was drawn. This tree was called a "generalized tree diagram." It gives the three major grammatical divisions of the sentence that are to be discriminated during discrimination practice as follows: 5V Noun Phrase Verb Phrase Adverb Subject . Verb Noun Phrase Predicate Object of Verb Sentence the mongrel did have some rest the tutor thought matters were legal the chief is mending weapons now Fig. 2.--Generalized tree diagram 35 The sentences in Figure 2 represent common usage.33 These were the sentences used during discrimination practice for the control group. Uncommon usagg refers to the sentence that results when the words in the positions of the subject and the object of the verb exchange places. It was predicted that the child would put phrase structure rules to work during discrimination practice as a result of the pressure for preserving subject-verb-object of the verb sentence relations.34 The experimental subjects were given discrimination practice on sentences of uncommon usage. All sentences have syntactical structure. Because this is so, only an improvement in word identification can be expected as a result of discrimination practice. The control group had common usage sentences. The experimental group had uncommon usage sentences. The dichotomy in terms of usage is one that is external to both linguistic theory35 and psychological theory.36 33Dan I. Slobin, "Cognitive Prerequisites for the DevelOpment of Grammar," in Studies of Child Language Deve10pment, ed. by Charles A. Ferguson and Dan Isaac Slobin (New York: Holt, Rinehart and Winston, 1973), pp. 197-201. 34Ibid. 35Chomsky, Syntactic Structures, pp. 26-27. 36Gibson, Principles of Perceptual Learning, pp. 102-050 36 (In psychological theory, uncommon usage Should constitute enhancement because the switch of the subject with the object of the verb automatically marks off the verb so that this particular grammatical unit can be picked up.)37 In Figure 3 the sentences representative of uncommon usage which were used for discrimination practice with the experimental group are given. The grammatical units designated by the two generalized tree diagrams begin with certain words that commonly begin noun phrases and verb phrases. These words were called phrase structure indicators. From these sentences two words were singled out for use in testing word identification. These two words were the and did. Noun Phrase Verb Phrase Adverb Subject Sentence / Verb Noun Phrase Predicate Object of the verb the matters were thought tutor legal some rest did have the mongrel the weapon is mending chiefs now Fig. 3.--Uncommon usage 37Ibid. 37 Terms for Psychological Tasks It may be recalled that the design of this study consists of pretest, discrimination practice, and a test for word identification, and that discrimination and identification tasks do not involve the same psychologi- cal processes.38 This fact was taken into account by defining identification responses apart from discrimi- nation responses. Wbrd identification is the ability to select the word that has been previously presented and to give its correct name. Word recognition is the ability to select a word that has been previously pre- sented without the ability to give its correct name. (In this study, these two response measures do not overlap. A word is either an identification or a recognition, not both.) word identification and word recognition both pertain to the test phase of the design. Three response measures were taken during dis- crimination practice, which was given by way of an imitation-of-oral reading task. The three response measures were naming, pointing, and omitting. Nami g is the word said by the child when he imitates. Pointing is the word pointed to and that is to be pronounced for him by the researcher when he imitates. Omitting is a word that is neither named or pointed to. Amodal 38Gibson, Principles of Perceptual Learning, pp. 122-23. 38 perception should occur.39 Amodal perception refers to the pattern produced across the sentence by these three responses. Summary This chapter dealt with a review of research, theoretical formulations, and definitions for terms used in this study. The review of research indicated that young children perceive language in terms of sentences and grammatical units, but not words. The description of how speech transfers to print by Monroe4O was not supported by the research evidence. The relationship of Gibson's theory of differentiation in perceptual learning to language was given in some detail. Defi- nitions needed to carry out a study in keeping with research findings and theoretical framework were given. 391bid., pp. 218-19. 4oMonroe, Reading, p. 209. CHAPTER III HYPOTHESES Eleanor J. Gibson thought that the "unit-forming principles of speech" were surrogates for grammatical syntax in the writing-to-Speech code.1 It has been suggested that the phrase structure rules of a child's Speech might represent one such set of surrogates. The phrase structure rules are needed to segment the sentence and to rewrite the segments so that the hierarchical organization of the sentence is retained as the child perceives words. Because syntactical structure is always present in sentences, the child may be able to learn to identify words, even though he does not pick up syn- tactical structure in print very well. Many children have learned words even though they never had the oppor- tunity for apprehending a grammatical unit through practice upon sentences representative of uncommon usage. The hypotheses must be capable of explaining what happens in both cases: when children receive lGibson, "Ontogeny of Reading," pp. 139-40. 39 40 practice upon sentences representative of common usage and when children receive practice upon sentences representative of uncommon usage. Two different sorts of hypotheses are needed: a null hypothesis that demonstrates that either word identification does or does not occur and descriptive hypotheses that tell something about the responses being made and the sub- jects making them without confusing the two. The Issues It may be recalled that the acquisition of grammar had appeared to be additive whereas the per- ception of words in sentences appeared to result from a breakdown of the sentence as a whole into grammatical units. Because the child does not always use the words that are critical for identifying phrase structure rules, it was concluded that the child would have to avail him- self of these rules by way of his knowledge of basic sentence relations--the subject of the sentence, the predicate of the sentence, and the object of the verb.2 In other words it is easier to see how a young child operates with basic sentence relations than it is to see how he operates with units like Noun Phrase and Kerb Phrase of the phrase structure rules. (Child grammars are not like adult grammars.) A generalized 2Brown, Cazden, and Bellugi, "Child's Grammar," pp. 295-332. 41 tree was drawn so that the basic sentence relations would be included in the phrase structure rules, and so what the child was responding to during discrimi- nation practice--the perceptual content--could be defined. Syntactical structure was made uncommon by marking off the predicate through an exchange in the words held in the positions for the subject and the object of the verb. The advantage of marking off the predicate lies in the economy of effort involved. Since the child has grammatical units for the subject, the predicate, and the object of the verb already available from speech, once the sentence is divided and some gram- matical unit, like the predicate, is apprehended in print, phrase structure rules of speech can operate directly. The child should be able to pick out the predicate when it is automatically makred off by having the subject and the object of the verb change places. This being the case, improvement in discrimination should occur to the point where the child apprehends sentences in terms of their grammatical divisions even though the sentences are no longer representative of uncommon usage. This study gives practice upon uncommon syntactical structure for the experimental group and upon common syntactical structure for the control group. WOrd identification is then measured for both groups upon new sentences containing old words presented during 42 discrimination practice. In this way it should be possible to tell if word identification improves as a result of practice upon a manipulation of syntactical structure that draws attention to grammatical units even though it is also representative of uncommon usage. In addition to the issue of improvement of word identification as a result of practice upon uncommon sentences, the theoretical framework of this study requires the researcher to Show how economy of effort applies, whether phrase structure has been attended to, and how perceptual functioning and perceptual content correlate over discrimination practice. Hypotheses Three reSponse measures were taken on the word identification sentences, the new sentences containing old words presented during discrimination practice. These response measures were: (1) WCrd identification, which is the ability to give the word its name; (2) word recognition which is the ability to select a word as having been seen before; and (3) Phrase structure indicators which are words that designate the beginning of Noun Phrase and Predicate on the generalized tree diagram; 43 for the sentences used in word identification test, these words are the words the and did. The first hypothesis is concerned only with word identifi- cation. The second is concerned with both word identifi- cation and word recognition responses and the numbers of subjects giving these responses. The third hypothesis is concerned with phrase structure indicators and the number of subjects responding to these words. The fourth hypothesis is not concerned with word identifi- cation. It deals with the response measures taken on the discrimination practice sentences. Hypothesis (1) Because syntactical structure was manipulated so that divisions into grammatical units might be found by the experimental subjects, but not for the control subjects, the experimental subjects should give more identifications than subjects assigned to the control group. Null hypothesis: -There is no difference between the mean number of word identifications given by the experimental and control subjects. A E test for the significance of the difference between means applies. The statistic to be used is student's E. The level of significance to be reached is .05 per cent, two-tailed test. 44 Hypothesis (2) Perceiving sentences in terms of their constituent grammatical units, and eventually words as parts of these units, represents an economy of effort. When a child uses his phrase structure rules by way of his knowledge of basic sentence relations, words ought to be identified because there is the whole sentence together with its grammar to help him. It is hypothesized that the experi- mental subjects will make more word identifications pro- portionately when the number of word identifications to word recognitions is taken into account and that when a count of the number of subjects making responses, word identifications and recognitions combined, and the mean number to responses per subject is found, that the exper- imental and control groups will not differ. Separate counts for words identified and recog- nized will be taken for the experimental group and also for the control group. The proportion of word identifi- cations to word recognitions will be computed for each group. The proportion of identifications is predicted to be higher for the experimental group. A count of the subjects making word identifications and word recognition for the experimental and also for the control group will be made. The means for the number of word identifications and word recognitions together will be found for only those subjects capable of responding. 45 When this is done, it will be evident that there is no difference between these means. The higher proportion of word identifications to word recognitions for the experimental subjects will then be presumed to be due to the increased accuracy in "localizing a difference and giving explicit recognition of boundaries for grammatical units and words."3 In order to avoid distorting the form that the data take or initiating confusions, the data will be tabulated and presented without further statistical analysis. Hypothesis (3) It is necessary to show that phrase structure rules were being used. Attention was drawn to the phrase structure indicators in defining phrase structure with a generalized tree diagram. These are those words that occur at the beginning of Noun Phrase and Predicate. Considering the fact that young children do not always utter these words in their daily speech, getting them to identify these words in reading ought to constitute evidence that rules must have been used. A list of the words identified by subjects in both the experimental and control groups together with the frequency with 3Charles A. Ferguson and Dan I. Slobin, "Segmen- tation," in Studies of Child Language, ed. by Charles A. Ferguson and Dan Isaac Slobin (New York: Holt, Rine- hart and Winston, 1973), p. 138. ' 46 which each of the different words was identified by each of the two groups will be made. Upon inspection of these frequencies, it will be evident that phrase structure indicators were identified more often by the experimental subjects. When the subjects identifying phrase structure indicators are counted, it will be evident that a higher percentage of subjects in the experimental group iden- tified these words. (The researcher might be expected to offer an explanation whenever experimental subjects identified words without also identifying a phrase structure indicator.) When subjects recognizing (as opposed to identifying) phrase structure indicators are counted, it will be found that the number of subjects locating these words is the same for both the experi- mental and control groups. This information will be presented in tabular form. Hypothesis (4) According to Gibson, the perception that accounts for word identification develOps over discrimination practice. The processes of identification tasks and the processes of discrimination tasks are not the same.4 This being the case, it is not possible to conclude what happened during discrimination practice from the findings 4Gibson, Principles of Perceptual Learning, PP. 122-23. 47 on the word identification task alone. A separate analy- sis of what happened during discrimination practice is needed. This hypothesis deals with discrimination prac- tice with Gibson's distinction between discrbmination and identification in mind. Theoretically, practice upon syntactical structure, preferably when it is easy to apprehend the sentence in terms of grammatical units, results in a sensitization of a child's perceptual functioning to the pattern for the hierarchical organization of the sentence provided by the generalized tree diagram. Naming, point- ing, and omitting are not dependent upon sensations unique to a given sensory modality. Since they are the responses recorded during discrimination practice, they should illustrate what Gibson means when she says that perception is amodal.S The stimulus content being responded to is the hierarchical organization of the sentence described by the generalized tree diagram. Presumably, rules are applied by way of the child's knowledge of basic sentence relations--the subject of the sentence, the predicate, and the object of the verb. The use of rules in this way would make perception abstract. Because basic sentence relations are being seen in terms of the hierarchical organization for the 51bid., pp. 218-19. 48 sentence as a whole, perception is relational. Gibson cited the research literature on transposition to explain the terms "abstract" and "relational."6 Only the experimental subjects receive discrimi- nation practice upon uncommon sentences. For them, it should be easy to "localize a difference and to give explicit recognition of boundaries for grammatical units."7 For these subjects, it is hypothesized that naming, pointing, and omitting will produce a figure that shows that these responses correlate at certain points in the sentence. Naming, pointing, and omitting responses for individual words across the sentence, all three discrimination practice sentences being taken together, will be summed to produce this figure. The correlations of these three separate response measures will be seen as a form of perceptual functioning. It is also hypothesized that a pattern that mirrors the grammar, child grammar though it may be, will result at points where naming, pointing, and omitting correlate. In order for a pattern for a grammatical unit to be seen, all three responses should be taken together as a single 61bid., pp. 109-10. 7Ferguson and Slobin, "Segmentation," p. 138. 49 figure. In this way, the figure Should display an "increase in specificity of responding to a set of stimuli."8 Summary In demonstrating how children attend to syntac- tical structure when they learn to identify words, it is necessary to observe how many of them respond to words and also which words were given a response. So that these two different categories of response measures would not be confused or the data distorted, only one null hypothesis, needed to prove that word identification had in fact occurred, was used. The remaining hypotheses were descriptive. The relevant data were specified. These particular hypotheses must demonstrate that: (1) Economy of effort occurs with practice upon uncommon sentences: (2) That phrase structure rules were used; and (3) That efficient perception is amodal, abstract, and relational. 8Ibid., p. 77. CHAPTER IV METHODOLOGY AND PROCEDURES This study intends to Show that when kindergarten children learn to identify words, they do so by noting the syntactical organization of sentences. It was hypothesized: (l) (2) (3) (4) That word identification resulted from attending to syntactical structure and that identification of words would be even better syntactical structure was perceived in sentences represen- tative of uncommon usage; That an economy of effort would account for the efficiency of identification with uncommon sentences; That words instrumental in using phrase structure rules would be identified; and That the kind of perception required would reflect the syntactical organization being responded to with respect to symbolic gram- matical meaning. 50 51 These hypotheses assumed that phrase structure rules of the writing-to-Speech code had for their surrogates the same rules in speech. In this way, it was being said that reading began with the basic processes of Speech. In such a case, word identification becomes the outcome of a carefully managed discrimination process. Eleanor J. Gibson has given several suggestions for managing O 0 0 0 I 1 discrimination practice. Methodology In promoting efficient discrimination, Gibson has stated that neither reinforcement nor knowledge of results are necessary. Neither is it necessary to call into play complex processes like thinking or imagination (a circumstance that suggests that the illogic of chang- ing the order for the subject and the object of the verb in uncommon syntactical structure might not be an analogous illogic for the child). Further, for the greatest effectiveness, higher order studies must be designed so that they permit processing of the total pattern: the instructions, the materials, what the child and the researcher do must all be taken as a 2 o o o 0 whole. By exerCiSing care in preparing sentences, 1Gibson, Principles of Perceptual Learning, pp. 121-42. 2Ibid. 52 and by having an imiation-of—oral reading task, the con- ditions for the efficient management of discrimination practice should obtain. Sentences The design of this study necessitated a pretest to serve as a criterion for the acceptance of children as subjects, sentences for discrimination practice, and sentences for word identification. Since it is syn- tactical structure that must be discriminated, it was desirable to keep sentence length constant while varying the type of syntactical unit within the sentence. For discrimination practice, the sentences were limited to six words. The type of syntactical unit for the object of the verb was either a Noun Phrase or a Sentence in adult grammar. The transitive verb was seen in terms of a large comprehensive class rather than in terms of some subdivision within this particular class of verbs. The word thought was selected because it often takes a sen- tence as its object, and because it can also be a noun. The verb hgzg_was chosen because it can be either a helping or a main verb, and because it cannot take passive voice. The verb mend was selected because it takes both an inanimate object and passive voice and because of its nasal sounds. The verbs did_and is were chosen because of a suspected need for vowel variety, and because these words may also serve as the main verb 53 as is the case for the verb w§£g_also. The researcher surmised that if children identifying words as wholes from their syntactical units during discrimination practice, that they would not notice the individual letters within words. As a consequence, they might say that they had seen words that had never been presented. The words think and by were chosen for the word identifi- cation sentences to check out this possibility. (It might be recalled that phrase structure rules divide sentences into their constituents, and that constituents are units replaceable by a single word without destroying the subject-predicate-object of the verb structure of the sentence.) It was also assumed that the words used would present a comprehensive display of the distinctive features for phonemes, the minimal units of speech. The words legal, rest, matters, mongrel, tutor, weapons, and ghi§£_Were selected partly for this reason, and partly because rather anomalous sounding sentences could be made from them. (Children may be unduly inclined to rely upon their highly systematic motor control over speech disadvantageously in an imitation- type task if sentences are not somewhat anomalous.) New was included because it is a familiar adverb that fits into the sentence at several points: The chief is now mending weapons. Now the chief is mending weapons. 54 The chief new is mending weapons. Chief is mending 223 the weapons. If the child needs a word to use as a placeholder in segmenting the sentence, this word might serve such a purpose during discrimination practice. Both E222 and the were chosen to provide variety in forestalling the possible effects of fatigue that might develop during practice. The surface structure of sentences contains not only grammatical syntax, but also thought and phonology as well. It may not be possible to attribute word identification to the effects of syntactical structure unless the effects of thought and phonology are also taken into account. The influence of thought was pre- sumably controlled by using anomalous sentences. Pro- viding variety for the distinctive features of phonemes may not control the influence of phonology. Young chil- dren may have a propensity for responding to sounds rather than to syntax. Gibson thought that efficient discrimination proceeded by the detection of the least frequent distinctive feature that also had the potential of becoming an invariant which later on figures into abstracting processes needed for higher order rules.3 It was decided to take advantage of the child's pro- pensity for some sound element, and to use it to make syntactical structure more obvious. In English, the 3Ibid., pp. 122-23. 55 phoneme # carries the distinctive feature lax of the tense/lax distinctive feature Opposition.4 Presumably, if this feature is perceived in the right places, an invariant for the predicate can be detected potentially such that the child's basic sentence relations of Speech-- the subject of the sentence, predicate of the sentence, and object of the verb--become the subject of the verb, the predicate, and the object of the verb in print. Exchanging the words for the subject and the object of the verb might automatically create juncture points that mark off the beginning and the end of the predicate succinctly. It was believed that the child would select some least frequent distinctive feature (maybe lax) to help him gain control over his speech habits in such a way as to permit entry into syntactical structure when he looks at the sentence during the course of imitating the researcher's oral reading. This thinking explains the rationale behind the preparation of uncommon sen- tences--ones in which words held by the subject and the object of the verb switched places. It might be noted that it is the distinctive feature for a phoneme, not the phoneme itself necessarily, that gets the child from phonology into grammatical syntax. 4Jakobson, Fant, and Halle, Speech Analysis, p. 43. 56 Design Discriminating and identifying words are not alike.5 In discrimination, the least frequent feature has the greatest utility for differentiation. In identification, confusion is least when stimuli differ by a feature with 50 per cent frequency in the set. It is the discrimination of differences for dimensions that transfers from discrimination processes to identification processes. In this study, the discrimination of dif- ferences for dimensions may be those that pertain to the differences between phonology and syntactical structure. It may take time for the transfer of differences to take place. Assuming this to be the case, activities that were the same for all subjects, and that kept the sub- jects task-oriented, intervened between discrimination practice and the presentation of the word identification sentences. The design consisted of pretesting, dis- crimination practice, intervening activities, and presentation of the word identification sentences. Procedures This section describes how subjects were obtained, how materials were made, and how the experimental session was conducted. 5Gibson, Principles of Perceptual Learning, p. 123. 57 Subjects The public school authorities of a school system in a metropolitan area in the Great Lakes region made kindergarten children available to the researcher. With the exception of one child, all were white and all were reported to be representative of middle-class socio— economic home backgrounds. All of the subjects had the "J same two classroom teachers in a team teaching arrange- ment. The researcher originally constructed sentences if that were representative of common usage, uncommon usage, and of ungrammatical usage--all using the same words. A fourth set of materials was made by combining a common usage sentence, uncommon usage sentence, and an ungram- matical usage sentence to make an inconsistent discrimi- nation practice treatment. There were originally four treatments: one for common usage, one for uncommon but grammatical usage; one for ungrammatical usage; and one for inconsistent practice. A total of eighty-seven sub- jects were made available. Six were dropped because they could read on the pretest. One was dropped because he had a severe speech problem and could not make him- self understood. Without any identifying information with the exception of name and first initial of last name, children were arbitrarily assigned to four treat- ments so that there were twenty different children in each treatment. After the subjects were tested, the 58 data were examined. It became apparent that a problem common to all four treatments could not be found. A problem was identified for the common and uncommon usage treatments, and therefore these two groups were used for this study. The data for the remaining two treat— ments have been set aside for a separate problem analysis at another time. Testing of subjects began immediately rt after Christmas vacation. There was a morning and an : afternoon kindergarten session. Because the eXperimental :w session lasted about an hour, only two subjects could be tested on a given school day. (The researcher was observed by the school principal and by each of the two classroom teachers once. The presence of these observers during the experimental session for three subjects is not believed to have influenced the results in any manner.) This study has a control and an experimental group. The control group had discrimination practice upon sentences that were representative of common usage-- subject-predicate-object of the verb. The experimental group differs in that they had discrimination practice upon sentences representative of uncommon usage--object of the verb-predicate-subject. These sentences were made by exchanging the subjects with the objects of the verb of the common usage sentences. It was assumed that the experimental subjects would perceive the object of 59 the verb as the subject and the subject as the object of the verb, but that in doing so the predicate would automatically be marked off for them. Materials The materials consisted of a packet of pretest cards, a set of discrimination sentence cards, and a ”a set of word identification sentence cards. For the pre- test packet, each word used in the sentences was printed on a small white index card in black manuscript with a felt pen. Each card was covered with laminated plastic to prevent soiling. For discrimination practice, each sentence was printed in black manuscript about one and one-half inches high on oak tag strips 18' x 6.“ Each strip had a hinged cover so that exposure could be con- trolled. The sentence was covered with laminated plastic. The word identification sentence cards were made in the same manner. The sentences as they appeared on the sen- tence cards are given in Table 2. Experimental Session Children had been informed by their classroom teachers that each of them would have an opportunity to play the "reading game" with the researcher. The researcher escorted the subject from the classroom to a small room in the school set aside by the school authorities for the purpose of testing subjects 60 individually in this study. The child's cooperation was secured by exchanging a few remarks on the way to the experimental room. A pretest, discrimination practice, intervening activities, and word identification were administered in this order. TABLE 2.--Sentences: Discrimination and identification Discrimination Experimental Group some rest did have the mongrel the weapon is mending chiefs now the matters were legal thought tutor Control Group the mongrel did have some rest the chief is mending weapons now the tutor thought matters were legal Identification Both Groups the mongrel did mend legal matters the chiefs think the rest legal the weapons were mended by the tutor Pretest Only subjects incapable of identifying any of the words used in the sentences were accepted as sub- jects. The words were briefly exposed, one card at a time, and the child was asked if he knew what the word was. After exposure, each card was placed in a pile face down. Discrimination practice began immediately. 61 Discrimination Practice The researcher and the subject sat on a 4' x 6' cotton throw rug. The researcher sat facing the child so that the subject could see the researcher's mouth, and said, "You and I will take turns being the teacher. First, it will be my turn; then it will be your turn. When your turn comes, you should look at the card and ,qJ point the way I do. You should learn to say what I say when I point also. If you don't remember on your turn, you should look carefully when it is my turn again. Remember you must look, point, and say what I say, so that you learn the game. And you must keep on trying very hard. Now we'll practice." Then the child and the researcher positioned themselves so that they sat side by side, and so that the child could no longer see the researcher's mouth without turning his head deliber- ately to do so. The sentence card, cover closed, was placed in front of the child. The researcher said, "Ready?" She then looked to see if the child was look- ing at the card, opened the cover, and pointed to each word as she read the sentence without intonation and without letting her voice fall at the end of each word, at the rate of about one word per second. She closed the cover and said, "That was my turn." The cover was cpened and she said, "You show me where you start to point. What does it say? You point while I say the 62 words so that you know how to look and point." The child and the researcher went through the sentence where- upon the researcher said, ”You must put your finger under the one where you want help. We'll do this again." The child placed his finger under the words. (The researcher took the child's finger in hand to get him started when necessary.) The researcher read the sentence again while the child did his best to point. Then the researcher said, ”You must put your finger under the one where you want help." She closed the cover and said, "This time, you must try to remember what I say." She then looked at the child to get attention, cpened the cover, read the words, and left the cover Open; the child tried to point and to say the words. The child was told, "Do that again.” Whatever the child said and did by way of point- ing was accepted. The cover was closed. The researcher looked at the child and said, "Now we are ready to start the game. You must keep on trying very hard. Try to remember what I say and put your finger under the one where you want help." Trials were counted from this point on. Trials A trial consisted of placing a sentence card in front of the child. The researcher said, "Ready?" She cpened the cover, read the sentence, closed the cover, looked at the child to get his attention, opened the 63 cover, the child imitates, with the researcher recording his response to each word on a form prepared for this purpose. When the responses were recorded, the cover was closed. Discrimination practice consisted of nine trials for each of the three sentences. (A naming response was the word said. An omitted response was either a word skipped or something offered for which a place could not be located in the sentence. The pointing responses were words pointed to. Egggl and mongol were acceptable naming responses for legal and mongrel.) Wbrd Identification The word identification sentence card was placed before the child with the cover closed. The researcher looked at the child to get attention, opened the cover, and said, "We have had all of these," while pointing to each word in turn. "Where are the ones that you know? Is it here? Here?" until every single word had been looked at by the child. The child's pointing responses were recorded. The researcher asked, "What is it?" WOrds correctly named were scored as word identifications. Wbrds incorrectly named or simply pointed to were desig- nated as word recognition responses. Six trials of bmitation-of—oral reading were given for each of the three sentences in the same manner as used in discrimi- nation practice. Three sentences were presented. Order of presenting sentences varied for both discrimination 64 practice and word identification. Materials were then put in order. The child was permitted to relax, and was then escorted back to the classroom. Limitations Without further research, the procedures used in this study should not be extended to classroom instruction. It may be recalled that this study purports to show how “1 reading begins with the basic processes of speech as identified and analyzed by E. J. Gibson6 and as opposed to the analysis suggested by Marion Monroe.7 Gibson's analysis differed from Monroe's in two respects: 1. Gibson stated that the unit was the sentence and that it in turn was segmented into its grammatical constituents whereas Monroe stated that the unit was the word; and 2. Gibson talked of "symbolic meaning"--the kind that applies to units like Noun Phrase and Verb Phrase or basic sentence relations like the sub- ject, the predicate, and the object of the verb-- the kind that applies to sentences generally, whereas Monroe talked of meaning as a vocabulary item in which the child is called up to recount some experience. 6Gibson, "Ontogeny of Reading," p. 136. 7Monroe, Reading, pp. 207-21. 65 In order to demonstrate what happens when speech transfers to print in terms of Gibson's analysis, it was necessary to use an imitation-of-oral reading task. In this way a comparison of adult grammar with child grammar could be made and it was also possible for the processes of speech and the discrimination processes on print to occur simultaneously. It was necessary to rule out the ‘ ff influence of the king of meaning that Monroe talked about. This was accomplished by administering both *" discrimination practice and identification test in the same experimental session and by constructing rather anomalous sounding sentences for discrimination practice and test. Learning to identify words may have rendered more difficult than is usually the case, and as a result, both the number of words identified and the numbers of subjects capable of identifying them appear rather small. Using anomalous sentences hardly seems advisable in classroom practice. This study should be replicated upon larger numbers of subjects and with transitive verb sentences that are more representative of those used in the classroom. It would also be advisable to find out what the ramifications of imitation-of-oral reading may be. Four treatment groups of subjects were originally tested. The data for two groups were set aside because a problem common to all four could not be found. Analyzing the 66 problems for the data set aside and comparing them with the problem of this study might be one way of getting at the complexities of imitation-of—oral reading. The find- ings of this study together with the additional research suggested are needed before specific recommendations for improving classroom materials and instruction can be made. When this study was conducted, it was not possible for the researcher to randomize the assigning of subjects to the four experimental groups. Instead with the barest minimum of information, the researcher arbitrarily assigned subjects to the four different treatments. There may be some undetected margin of error because randomization was not followed. Concluding Remarks Methodology and procedures were develOped with Gibson's suggestions for efficient management of dis- crimination in mind. How the sentences had been con- structed was explained. Attention was drawn to the fact that this study purports to show how reading begins in speech in keeping with Gibson's identification and analysis of those basic processes that are pertinent in reading. Using Gibson's analysis made it necessary to design a study that would not yield directly the infor- mation needed to improve classroom instruction and materials. Suggestions for additional research were given. CHAPTER V PRESENTATION AND ANALYSIS OF DATA Word identification begins in speech. Because words are the end products of a hierarchical organization of the sentence, phrase structure rules must be used to effect a breakdown that starts with the sentence as a whole. It is the operations of these rules that accom- plishes a transfer from speech to print. The child arrives at these rules by way of his knowledge of basic sentence relations--the subject of the sentence, the predicate of the sentence, and the object of the verb. The meaning that is involved is of a symbolic nature, the kind that is inherent in grammar, and therefore the kind that applies to all sentences. Several hypotheses were formulated to demonstrate how reading begins in speech in this manner. These hypotheses predicted that children would identify words in keeping with the oppor- tunity to pick up syntactical structure; that efficient word identification would reflect an economy of effort; that phrase structure indicators would be identified; and that a pattern for efficient perceptual functioning 67 68 would emerge over discrimination practice. These pre- dictions are restated and the evidence for them is pre- sented herewith. Hypothesis (1) Findings This hypothesis predicted that there would be no difference between the mean number of word identifications given by the experimental and control groups. A 2 test for the significance of the difference between means was used to test this prediction. Table 3 gives the scores for subjects assigned to the experimental and control groups. The scores are the numbers of word identifi- cations given. The experimental group identified thirty-seven words for a mean of 1.85, and the control group identified six words for a mean of .30: student's E, two-tailed test is 12.18. (For thirty-eight degrees of freedom, Ir? must be at least 12.025 for the .05 per cent level of significance.) The null hypothesis of no mean dif- ference is rejected. The eXperimental subjects did identify more words than the control group did. Both groups had practice upon anomalous sentence with syn- tactical structure present. The superior performance was predicted to represent an improvement in discrimi- nation due to practice upon uncommon sentences. (In these sentences, the predicate had been marked off by exchanging the subjects with the objects of the verbs.) TABLE 3.--Scores and means for word identification 69 Experimental Group Control Group subjects score subjects score 1. Kathleen 3 Victor 0 2. Scott 0 Mary 2 3. Becky 0 Gary 0 4. Bobbi (female) 1 Chris (female) 0 5. Mike 2 Richard 0 6. Chuck 11 Mike 0 7. Annette 2 Scott 0 8. Joe 0 Roxanne 0 9. Steve 0 Kathryn 0 10. Donald 0 Eddie 0 11. Danny 1 Kate 2 12. Delores 0 Mike 0 13. Judy 0 Meredith (female) 0 14. Beth 0 James 0 15. Sue 1 Tom 0 16. Daphne 1 Linda 2 17. Crystal 3 Joe C. 0 18. Robert 10 Anne 0 19. Lisa 2 Joe 0 20. Kate 0 Julie 0 Total number of words 37 6 Mean number of words 1.85 .20 70 Because there were subjects in the control group who did identify words, improvement in discrimination appears to have occurred. §g§ An examination of the scores suggested a pos- sibility of an interaction of treatment with sex of the subject. Since the uniformities of native language learning had not suggested this interaction, this evi- dence was unexpected. The data are summarized in Figure 4. 3.00 3.00 65 X 1.08 8 1.00 . -75 Girls .00 .00 T1 T2 Experimental Control Fig. 4.--Interaction with sex Age If children are merely improving in discrimi- nation when they identify words, age would not be expected to be a factor. By the time a child starts to school, his syntax in speech seems to be quite well 71 mastered.1 The age range in this study is limited by the selection of only kindergarten subjects. Subjects in the experimental and control groups did not differ with respect to mean age. Their mean ages were 5.74 years and 5.78 years reSpectively. When the performance of the younger subjects, sixty-two through sixty-nine months, is compared with that of the older subjects, seventy through seventy-seven months, age seems to be of no consequence for the experimental subjects whereas only the older subjects identified words in the control group. The youngest control subject to identify words was seventy-four months. The youngest experimental subject to identify words was sixty-three months. The data are summarized in Table 4. Hypothesis (2) Findings A major limitation of this study derives from the fact that the meaning involved in a transfer from speech to reading is of a symbolic nature. When meaning is defined as being something specified by grammar, it is necessary to rule out the possible effects of extraneous experiences in order to show that this is so. Discrimi- nation practice and test were confined to a single session and the sentences used were deliberately made anomalous. 1Paula Menyuk, "A Descriptive Study of the Syn- tactic Structures in the Language of Children: Nursery School and First Grade" (unpublished Ph.D. dissertation, Boston University, 1961). 72 The transfer from Speech to reading took place during imitation-of-oral reading. As a consequence of this approach to meaning, the number of words identified and the number of subjects capable of responding were expected to be rather small. This expectation was upheld: of a possible 380 word recognitions and identifi- cations, only 73 were obtained for the better performing group. Of a possible 20 subjects, only 11 in the better performing group identified words. TABLE 4.--Percentages of subjects identifying words by age Younger Subjects Older Subjects (62-69 months) (70-77 months) Experi- Experi- mental Control mental Control No word identified 40% 100% 50% 63% One word identified 30% 0% 10% 0% Two words identified 10% 0% 20% 37% Three or more words 20% 0% 20% 0% 100% 100% 100% 100% Hypothesis (2) dealt with economy of effort. If a word is the end product of a breakdown for the hierarchical organization of the sentence, economy of effort should result from regarding words as parts of this type of organization. It was predicted that the 73 proportion of words identified to words recognized would be larger for the experimental subjects, that the number of subjects capable of responding meaningfully would be greater for the experimental group, but that the mean number of words recognized and identified per subject capable of responding meaningfully would be the same for the two groups. The relevant data are given in Tables 5, 6, and 7. TABLE 5.--Wbrd identifications and word recognitions Number of words Experimental Control WOrd Identifications 37 6 word Recognitions 36 , 32 Total 73 38 Proportion of Identifi- cations to Recognitions 37/36 = 100 6/32 = 19 Table 5 shows that the experimental subjects gave at least one word identification for every word recog- nition and that the control subjects gave only one word identification for every five recognitions. Table 6 shows that a larger percentage of subjects appeared to 'be responding meaningfully for discrimination practice with uncommon sentences: 85 per cent as opposed to 60 per cent. Table 7 shows that the mean number of 74 TABLE 6.--Responding by subjects Number of Subjects Experimental Group 17 of 20 subjects gave word identifi- cations and recognitions 17/20 = 85% responded meaningfully ll of thEEe 17 gave word identifi- cations 6 of 17 gave word recognitions Control Group 17 of 20 subjects responded 5 of 17 gave meaningless letter names 3 of 17 gave word identifications 9 of 17 gave word recognitions 12/20 = 62% responded meaningfully Ag. _ ‘4-“ .. . . .1 .‘ I TABLE 7.--Means for word responses Experimental Control Wbrd Identifications 37 6 word Recognitions 36 32 Total 73 38 Means 73/17 = 4.29 38/12 = 3.17 75 word identifications and word recognitions combined was 4.29 for the experimental subjects and 3.17 for the control subjects. The difference between these means hardly seems large enough to.account for the fact that six times as many words (37 to 6) were identified by the experimental subjects and that nearly four times as many subjects (11 to 3) were capable of identifying words in the experimental group. Two words not presented during discrimination practice were used in the word identification sentences. These words were thigk and Ex. This was done because it was thought that efficient attending to syntactical structure would preclude attending to letters within words. This may be the case: Chuck, the most efficient subject in word identification, was the only subject to report having seen the word by before. Four experimental and one control subject reported having seen think before. The word 22E appears in discrimination practice but not in word identification. It was used in discrimination practice because it was surmised that subjects might use this word as a placeholder, and therefore might learn to identify other words with it. One experimental subject, only, gave the word 22! as a word recognition in word identification. This same subject failed to identify words. In view of how the term "constituent" 76 was defined, there may be no need for a placeholder in segmenting the sentence during discrimination practice. Hypothesis (3) Findings This hypothesis predicted that if subjects were attending to syntactical structure of sentences, they would identify the particular words appearing as phrase structure indicators on the generalized tree diagram. The words so specified for the word identification sentences are 329 and gig. The data for word identifi- cations are given in Table 8 and the data for subjects giving these responses are given in Table 9. TABLE 8.--Frequencies for word identifications Number of WOrd Identifications Experimental Control 1. the 16 2 2. did 5 0 3. chief 7 2 4. legal 3 2 5. mongrel l 0 6. matters 1 0 7. weapons 1 0 8. rest 1 0 9. tutor 2 0 totals 37 6 77 TABLE 9. Frequencies for subjects Subjects Identifying Phrase Structure Indicators Experimental Group 11 gave word identifications 8 of these gave 20 identifications for the words the and did 77% of these subjEEEs ideHEIfied these words Control Group 3 subjects identified words 1 of these identified the twice 33% of these subjects idEEtified these words Subjects Recognizing Phrase Structure Indicators Experimental Group 6 subjects recognized words 3 of these located did and the 50% located did and the Control Group 9 subjects recognized words 5 of these located did or the 55% located did or tHe These data suggest that phrase structure indi- cators SEE and gig tended to be identified more often by a higher percentage of subjects in the experimental group (77% versus 33%), and that once these words were identified that several other identifications might also occur. But this thinking does not hold entirely. It does not explain why, for example, Danny and Bobbi of the experimental group gave as their only identifications the words matters and mongrel respectively. An exami- nation was made in terms of the sentence administered first during discrimination practice. In both groups, 78 subjects who had had the Eg£g£_sentence first, with but one exception, identified 522 or dig_or both. It may be that the sentence administered first exerts some influence upon the selection of the least frequent phoneme, and that this selection in turn enters into the abstracting processes of phrase structure rules. Interestingly, the list of words in Table 8 contains no words used as verbs with the exception of the word gig. The frequencies with which individual words were identified do suggest the possibility that a distinctive feature of some sound might be selected and used. Merely recognizing E22 and gig_did not seem to promote word identification in that the experimental and control subjects do not seem to differ in this reSpect. Hypothesis (4) Findings It was predicted that at the end of discrimination practice that patterns for perceptual functioning and for perceptual content will emerge for the experimental group. These patterns were to consist of naming, pgigt— Egg, and omitting responses summed separately for each word across the sentence. A pattern for perceptual con— tent, the grammatical units being responded to, was also predicted to emerge. This pattern was to be a composite for all three of these responses. The evidence for the experimental group is given in Figure 5 and for the control group in Figure 6. Total Number of words Given 400 350 300 250 200 150 100 U1 0 79 3'07 1! J32 \0 »._~__ 61 (Total number of words possible) 540 540 540 540 540 540 l 2 3 4 5 6 the matters were legal thought tutor the weapon is mending chiefs now some rest did have the mongrel Experimental Group Fig. 5.--Amodal perception 400 350 300 250 200 150 100 50 80 ttfl -._-.. (Total number of words possible) 540 540 540 540 540 540 1 2 3 4 5 6 the tutor thought ‘matters were legal the chief is mending weapons now the mongrel did have some rest Control Group Fig. 6.--Amodal perception 81 Figure 5, for the experimental subjects, shows that the predicate tends to be marked off as syntactical unit and that naming, pointing, and omitting responses are correlated at these points, that is, where naming drops off, pointing and omitting responses pick up. Naming, pointing, and omitting are amodal in that they are not sensations that are unique to some particular I "r sensory modality. They are forms of responding that .5 take place separately. They do not correlate in Figure 6 for the control subjects. In Figure 5, for the experimental subjects, if naming, pointing, and omitting are each taken separately, it is not necessarily obvious that a syntactical unit, such as the predicate, is being responded to consistently. All three must be taken together as a composite at their points of correlation before such a conclusion can be reached. Further, the Egtgg sentence was not divided as expected from the generalized tree diagram: thought was not perceived as the predicate. nggl_was perceived in this way instead. Maybe this performance reflects an unexplainable peculiarity of child grammars. However, it does appear from the points of correlation that some kind of abstracting may have gone on. In Figure 6, for the control subjects who had practice upon sentences representative of common usage, it is obvious that the control subjects actually did 82 more naming of phrase structure indicators £23 (435 to 391; 289 to 277) and gig_(3l4 to 221) than did the sub- jects receiving practice upon sentences representative of uncommon usage. An examination was made to see if the words named during discrimination practice were also the ones identified later on for experimental subjects. According to Gibson, discrimination and identification processes are not the same, and therefore the naming that occurs during discrimination practice need not be the same as the naming that occurs during word identifi- cation.2 Examination revealed the possibility of little relationship between naming during discrimination prac- tice and naming on word identification test. For example Lisa gave a total of eight naming responses of fifty-four possibilities for the tutor sentence during discrimination practice. All eight were given to the word lgggl. On word identification test, she selected the words dig_and ghigf and failed to identify the word lgggl. Perhaps some abstracting process accounts for the differences between discrimination and identifi- cation. Gibson used the research literature on trans- position to explain what was meant by the terms 2Gibson, Principles of Perceptual Learning, pp. 122-23. 83 "relational" and "abstract."3 Her interpretation of this literature.may apply to Figures 5 and 6. Gibson observed that transposition dealt with the discrimination of structure in particular. She reported that organisms, both human and subhuman, could be trained to give relational responses like "darker than," "heavier than," "different than" whenever two adjacent contrasting stimuli were presented. The information extracted was "contrast." Once "contrast" was extracted, this infor- mation functioned as invariant and transferred to other stimuli possessing the same contrast dimensions. Stimuli for "contrast" do not in themselves exist. What exists are the conditions that give rise to this perception. Perhaps the correlations in Figure 5 represent a detection of the contrast "different than" in such a way that an abstraction can be made. If this interpretation is plausible for locating the boundaries of grammatical units, then practice upon sentences representative of uncommon usage possess the stimulus conditions that give rise to the abstraction "contrast" whereas practice upon sentences representative of common usage may not. The control subjects, Figure 6, apparently did differentiate between words, and maybe parts of the sentence also. But there is no correlation among naming, pointing, and omitting; an abstraction may not have been 31bid., pp. 217-38; 283-95. 84 made. The failure to abstract efficiently may, in turn, have made it difficult for these subjects to grasp the real point of the instructions. In other words, they saw differences but didn't know what to do with the dif- ferences seen. Because the sentences were anomalous, one consequence may have been a tendency to revert to classroom experience. All of the subjects in this study were receiving instruction in giving the letter names in their classroom. This bit of experience may have been put to use inapproPriately with the result that five of these subjects gave meaningless letter names. Monroe4 tried to accomplish a transfer from speech to print by capitalizing upon the experiences that children have had. The evidence in Figure 6 suggests that it is possible that such a procedure may not necessarily be useful. Summary The data for four hypotheses were presented. Hypothesis (1) demonstrated that the experimental sub- jects did identify more words than did the control sub- jects. The mean difference was significant at the .05 per cent level, two-tailed test. The three remaining hypotheses were descriptive in nature. Hypothesis (2) indicated that more experimental subjects were capable 4Monroe, Reading, pp. 207-21. 85 of responding and that these subjects gave at least one word identification for every word recognition whereas the control subjects gave five recoqnitions for a single identification. Hypothesis (3) demonstrated that the experimental subjects identified the phrase structure indicators £23 and did_more often than was the case for the control subjects. Hypothesis (4) demonstrated that for the experimental subjects, naming, pointing, and omitting responses correlated with the beginning and ending points for the predicate. The evidence also suggested that three unpredicted variables may have influenced word identification. These variables were (1) sex of the subject; (2) age of the subject; and (3) the sentence administered first during discrimination practice. CHAPTER VI SUMMARY AND CONCLUSIONS Summary This study started with the postulate that read- ing begins with the basic processes of speech, and that these processes are carried over into reading through the use of phrase structure rules. To demonstrate that this is the case, an experimental descriptive study was designed. The supposition was operationalized by hypothesizing that kindergarten children would learn to identify words as wholes if given an opportunity to attend to syntactical structure of sentences, preferably of sentences representative of uncommon usage. Ideas for methodology and for interpreting the results were developed from Eleanor J. Gibson's principles of per- ceptual learning.l Children were tested in an experi- mental session that lasted an hour. They were first pretested, and only those kindergarten children who could not already read the words in this study were 1Gibson, Principles of Perceptual Learning, 537 pp. 86 F? a a." 87 accepted as subjects. For this study, there were two groups of subjects, twenty in each group. One group, the experimental group had discrimination practice upon sentences representative of uncommon usage. The other group, the control group, had discrimination practice upon the same sentence representative of common usage. Both groups were tested for word identification upon new sentences containing the same words previously pre- sented. Subjects were given discrimination practice through the use of an imitation-of-oral reading task. The syntactical structure for the sentences used in this task was formally defined as consisting of subject of the sentence, predicate of the sentence, and the object of the verb. While it was difficult to get kinder- garteners to learn to identify many words in a single experimental session, it was evident that children did identify words as wholes, that they did so by attending to the syntactical structure of the sentence, the uncommon sentences, those with the object of the verb, the predicate of the sentence, and the subject of the sentence yielding superior results as predicted. Stu- dent's E, two-tailed test, for no difference between means was significant at the .05 per cent level of significance. Three additional hypotheses of a descriptive nature were tested. For these hypotheses, the data were 88 tabulated and presented. It was necessary to avoid dis- torting the form that the data took. Care was taken to avoid possible sources of confusion that might arise when both the words being identified and the numbers of sub- jects identifying them must be examined. The data pre- sented demonstrated that there had been an economy of effort, that phrase structure indicators were being identified, and that the predicate of the sentence had 1* “ “““"" been marked off by naming, pointing, and omitting during discrimination practice for subjects having discrimi- nation practice upon uncommon sentences. Three unantici- pated factors also emerged. These were the possibility of an influence for the sex of the subject, for age, depending upon whether the sentences were uncommon or not, and for the sentence administered first during discrimination practice. Conclusions The Problem The problem identified was that adult and child grammars are not alike. Eleanor J. Gibson had identified and analyzed those particular processes of speech which were pertinent for a transfer from speech to print.2 2Gibson, "Ontogeny of Reading," p. 136. 89 Unlike Marion Monroe3 who saw the word as the unit and who interpreted the transfer from Speech to print in terms of experiences and thoughts that words brought to mind, Gibson stated that the transfer took place through the use of rules that started with the sentence as a unit and then divided it up into grammatical units. The meaning that transfers in such a case is symbolic and abstract. It belongs to the grammar, and therefore applies to all sentences in general. In order to Show that the transfer takes place as Gibson has stated, it was necessary to rule out extraneous influences. Dis- crimination practice and test for word identification took place in a single experimental session. Anomalous sentences were constructed. A generalized tree capable of incorporating the rules for both adult and child grammars was drawn up to describe the syntactical structure of sentences. An imitation-of-oral reading task was used to give discrimination practice. This task permitted control over adult input so that child output could be compared with it. It also permitted a transfer of basic Speech processes while discriminations in sentences were being made. These procedures were probably necessary. It may not be possible to find out 3Monroe, Reading, pp. 207-21. 90 how rules effect a transfer from speech to print with- out giving the procedures followed here much consideration. Methodology and Hypotheses Eleanor J. Gibson provided an identification of those Speech processes pertinent in a transfer from Speech to print; she also provided a theoretical frame- work for conducting the study and for interpreting the findings. AS a result, during discrimination practice, reinforcement was not given, complex processes like thinking were not called into play, and the study was designed to permit processing of the total pattern.4 Gibson's theory states that the transfer from speech to print represents only an improvement in discriminating syntactical structure from Speech. Insofar as this study had been concerned, Gibson's theory has been borne out. It may not be necessary to exclude a child who can already read as was done in this study. A null hypothesis was used to Show that word identification occurred Optimally with discrimination practice upon uncommon sentences--those in which the subjects and objects of the verbs had switched places, thereby marking off the predicates. The three remaining hypotheses were descriptive in terms of what might happen according to theory. Both the words--the stimuli 4Gibson, Principles of Perceptual Learning, 537 pp. 91 to be responded to--and the responses to be made to the stimuli were identified and defined. It seems likely that researchers may need to work in similar fashion in the future. It is difficult to visualize how economy of effort and amodal perception might be operationalized in beginning reading. It is also difficult to specify what a child is doing when he is using abstract rules without a theory. Implications Gibson looked at the process of learning to read as a piece of natural history.5 She assumes that chil- dren already have the appropriate responses from Speech. They need to discover how and where their responses apply in printed materials. Gibson believes that it is neces- sary to manipulate the stimulus material so that the subject has an Opportunity to discover what must be found. Confusions that have been attributed to the shortcomings of phrase structure (John is easy to please. and John is eager to please. being a case in point.)6 are not to be assigned to the ineptness of the pupil, but rather to the ineptness of the professional in providing apprOpriate instruction. The findings of 5Gibson, "Ontogeny of Reading," p. 136. 6Chomsky, Syntactic Structures. 92 this study suggest that the sentences used in beginning reading instruction Should be those that allow the child to apply phrase structure rules by way of his knowledge of basic sentence relations-—the subject of the sentence, the predicate of the sentence, and the object of the verb. But before guidelines can be given as to how sentences should be prepared, this study Should be replicated. It was not possible to randomize the assigning of subjects to groups. Although the researcher was a stranger to the children and to the school, it is not known how much error may exist because the researcher arbitrarily assigned subjects to different treaflment groups. Comparatively few subjects were involved in the analyses of data for some of the hypotheses. In the superior group, only eleven subjects succeeded in learning to identify words. Now that it has been demon- strated that the sentence and its grammatical sub- divisions are the appropriate units, not words as ‘Monroe thought, it is necessary to find out whether anomalous sentences are really crucial in effecting a transfer from Speech to print. A child may use phrase structure rules with thought-provoking sentences also. In this study, it was the uncommon sequence of object of the verb-predicate-subject that produced the superior results, apparently because the predicate may have been automatically marked off by switching the 93 subjects with the objects of the verb. Much more infor- mation is needed regarding the possibilities for manipu- lating the ordering of grammatical units in reading material so that some one unit can be perceived. Imitation has been found to be a useful research tool in studies of child language development.7 The uses of imitation in reading research or in the classroom seem to be unknown. This study originally started with four treatment groups. Data for two groups were laid aside because a problem common to all four treatments could not be found. One starting point in assessing the possible contributions of imitation in reading research would be to analyze these data. Additional development of Gibson's theory would be desirable. She has cited evidence that indicates that discrimination and identification tasks do not involve the same processes. Discrimination is most efficient with the detection of a least frequent dis- tinctive feature and that identification will be most efficient when stimuli differ by a feature with 50 per cent frequency in the set.8 Some selection of a 7Slobin and Welsh, "Elicited Imitation as a Research Tool in Developmental Psycholinguistics," in Studies of Child Lan uage Development, ed. by Charles A. Ferguson and Dan I. Slobin (New York: Holt, Rinehart and Winston, 1973), pp. 485-96. 8Gibson, Principles of Perceptual Learning, pp. 122-23. 94 distinctive feature must take place during discrimination, and during the course of abstracting through the use of phrase structure rules by way of an ability to handle basic sentence relations, the requirements for identifi- cation are met. Discrimination and identification were kept dichotomous in this study. But some confusion may exist; naming occurs in both discrimination and in identification. Observation of individual performances revealed that the words named during discrimination were not necessarily the same words identified. The fact that the same term applies to both processes together with the fact that it is not known how dis- tinctive features are discovered renders it impossible to specify the distinctive features that were used by subjects in this study. Keeping the order in which sen- tences are administered constant might help to clarify the different ways naming is being used. Both age and sex appeared to be factors that need further study. In the control group, the data suggest that the subject had to be a girl at least seventy-four months old in order to identify words. In the experimental group, both older and younger sub- jects of both sexes identified and failed to identify words. Also there were two boys, one sixty-four months and the other, seventy-three months, who gave a sub— stantial number of word identifications, ten in one case 95 and eleven in the other. Menyuk9 has not found sex differences in the acquisition of syntactical structures in speech and she found age to be a factor when the age range was much greater than it is in this study. For Gibson, greater specificity of the correlation between stimulation and discrimination, increasing differen- tiation, and the pick up of invariant relations occur with increases in experiences that age provides.10 Insofar as the experimental treatment is concerned, it would appear possible to use nursery school subjects as well as kindergarten children. Chapter Summary An abstract was provided in the opening summary. Conclusions were drawn with respect to the handling of the problem and contributions of Gibson's theory. Replication of the study upon a larger sample and with thought-provoking sentences was suggested. The need for information on ways of manipulating grammatical sequences in sentences, on the possible uses of imi- tation in reading research, and a need for further development of Gibson's theory were noted. Research 9Menyuk, "Descriptive Study of Syntactic Structures." 10Gibson, Principles of Perceptual Learning, pp. 450-70. 96 investigating the effects of age, sex, and the sentence administered first was also mentioned. SELECTED BIBLIOGRAPHY SELECTED BIBLIOGRAPHY Bellugi, Ursula, and Brown, Roger W. The Acquisition of Lan ua e. Monograph of the Society for ResearCh in CHiId Development. Serial No. 92, XXIX, No. l (1964). Brown, Roger W., and Berko, Jean. "WOrd Associations and the Acquisition of Grammar.“ Child DevelOp- Cazden, Courtney. "Environmental Assistance to the Child's Acquisition of Grammar." Unpublished Ed.D. thesis, Harvard University, 1965. Chall, Jeanne. Learning to Read: The Great Debate. New York: McGraw-HiIl Book Company, 1967. Chomsky, Carol. The Ac uisition of Syntax in Children from 5 to . Cambridge, Mass.: M.I.T. Press, 1969. Chomsky, Noam. Syntactic Structures. The Hague: Mouton, 1957. Chukovsky, Kornei. From Two to Five. Berkeley, Calif.: University of CaIifornia Press, 1971. Deese, James. "Influence of Inter—Item Associative Strength upon Immediate Recall." Psychological Representative, V (1959), 305-12. Entwistle, Doris. Egrd Associations of Young Children. Baltimore: John Hopkins Press, 1966. Epstein, William. "The Influence of Syntactical Structure in Learning." American Journal of Psychology, LXXIV (1961), 524-31. . "A Further Study on the Influence of Syntac- tical Structure in Learning." American Journal of Psychology, LXXV (1962), 121-26. 97 98 Epstein, William. "Temporal Schemata in Syntactically Structural Material.“ Journal of General Psychology, LXVIII (1963), 154-64. Erwin, Susan M. "Changes with Age in the Verbal Deter- minants of WOrd Associations.". American Journal of Psychology, LXXIV (1961), 361-72. Pagan, William. "An Investigation into the Relationship Between Reading Difficulty and Sentence Transfor- mations.” Unpublished Ph.D. dissertation, Uni- versity of Alberta, 1969. Ferguson, Charles A., and Slobin, Dan I. Studies of Child Language Development. New York: Holt, Rinehart and Winston, Inc., 1973. Fraser, C. C.; Bellugi, U.; and Brown, R. W. "Control of Grammar in Imitation, Comprehension, and Pro- duction.” Journal of Verb Behavior and Verb Learning, II (1963), 121-35. Gibson, Eleanor J. Principles of Perceptual Learning and Development. New York: Appleton-Century-Crofts, Educational Division, Meredith Corporation, 1969. . "The Ontogeny of Reading." American Psycho- logist, XXV (1970), 136-43. Gibson, James J. The Senses Considered as Perceptual Systems. Boston: Houghton Mifflin Company, 1966. Hoffman, Martin, and Hoffman, Lois, eds. Review of Child Development Research. II. Ann Arbor, Mich.: university of MiEhigan Press, 1966. Huey, Edmund Burke. The Psychology and Pedagogy of Reading. Cambridge, Mass.: M.I.T. Press, 1968. Huttenlocher, Janellen. "Children's Language WOrd-Phrase Relationships." Science, CXLIII (1964), 264. Jakobson, Roman. Child Language Aphasia and Phonological Universals. The Hague: Mouton, 1972. , and Halle, Morris. Fundamentals of Language. The Hague: Mouton, 1956. ; Fant, C.; Gunnar, M.; and Halle, Morris. Pre- liminaries to Speech Analysis. Cambridge, Mass.: M.I.T. Press, sixth printing, 1965. 99 Lenneberg, Eric H. New Directions in the Study of Lan- guage. Cambridge, Mass.: ’M.I.T. Press, 1964. Levin, Harry, and others, eds. A Basic Research Program on Reading. CRP-639. Ithaca, N.Y.: Cornell University, 1963. , and Williams, Joanna. Basic Studies on Reading. New York: Basic Books, 1970. Marks, L., and Miller, G. A. "The Role of Semantic and Syntactic Constraints in the Memorization of English Sentences." Journal of Verb Behavior and Verb Learning, III (1964), 1-5. Menyuk, Paula. "A Descriptive Study of the Syntactic Structures in the Language of Children: Nursery School and First Grade." Unpublished Ed.D. thesis, Boston University, 1961. . Sgntences Children Use. Cambridge, Mass.: M.I.T. Press, 1969. Miller, G. A., and Isard, S. "Some Perceptual Conse- quences of Linguistic Rules." Journal of Verb Behavior and Verb Learning, II (I963), 217-28. Monroe, Marion. Growing into Reading. Chicago: Scott Foresman and Company, 1951. Pick, Anne. "Improvement of Visual and Tactual Form Dis- crimination." Unpublished Ph.D. dissertation, Cornell University, 1963. Reiss, B. F. "Genetic Changes in Semantic Conditioning." Journal of Experimental Psychology, XXXVI (1946), 143-52. Schlesinger, I. M. Sentence Structure and the Reading Process. The Hague: Mouton, 1968. Slobin, Dan 1., ed. A Field Manual for Cross-Cultural Study of the Acquisition of Communicative Com- etence. Béfkeley, Califi} University of CaIifornia, ASUC Bookstore, 1967. . Psycholinguistics. Glenview, 111.: Scott Foresman and Company, 1971. . The Ontogenesis of Grammar. New York: Academic Press, 1971. 100 Smith, Frank. Psycholinguistics and Reading. New York: Holt, Rinehart and Winston, Inc., 1973. , and Miller, G. A., eds. The Genesis of Language: A PsycholinguistiC‘ApproaCh. Cam- bridge, Mass.: M.I.T. Press, 1966. Stageberg, Norman. An Introductory English Grammar. New York: HoIt, Rinehart and Winston, Inc., 1965. Stevenson, Harold, ed. Child Psychology. 62nd Yearbook of the National Society for the Study of Edu- cation. Chicago: University of Chicago Press, 1963. Turner, Elizabeth Ann. "Developmental Studies of Sen- tence Voice and Reversibility." Unpublished Ph.D. dissertation, Cornell University, 1966. Vygotsky, Lev Semenovich. Thought and Language. Cambridge, Mass.: M.I.T. Press, 1962. Weir, Ruth H. Language in the Crib. The Hague: Mouton,'I962. MITIHHIIWITHWIWIUIW Imus MN MW 3 1293 03061| 7272