Journal o f Psycholinguistic Research, Vol. 2, No. 2, 1973

Syntactic Complexity and Information Transmission in First-Graders: A Cross-Cultural Study Susan H. H o u s t o n I Received April 24, 1972 Some differences between child and adult communication are due to general developmental immaturity and some to language-specific factors. The Piagetan concept of syncretism exemplifies an hypothecated universal o f psycholinguistic development, and it results in child texts characterized by minimally structured temporal, causal, and logical connections. A different sort o f problem is that, within a specific language, certain syntactic items may fail to mature in children at the expected rate, because o f structural oddities o f the constructions. English has several such constructions, generally peculiar in their conflict between surface- and deep-structure subject o f the main verb. It is predictable that a syncretistic child attempting to deal with such items in a text will analyze them according to the surface-structure analogic method which defines verbal syncretism, and so will fail to make sense of them. Finally, cutting across these child language features are others peculiar to the language o f specific groups o f children, such as the putative communication impairment of the disadvantaged black child, whose language has been described by some researchers as less efficient and slower to mature than that o f others. Clearly, if this is so, then black children's language should at any young age show more evidence o f syncretistic communicative immaturity as well as slower development o f adult syntactic patterns. The present paper, then, investigates on a cross-cultural basis the dual hypotheses o f syncretism and faulty mastery of difficult syntax, by means o f an experimental story-repeating format first used by Piaget. Hypotheses are examined for the cause o f children's distinctive communication technique, and the whole question o f the significance of black~white differences in communicative style and verbal maturity is discussed.

INTRODUCTION Researchers in early child language have noted a n u m b e r of striking differences between the communication of children and that of adults. Those 1Program in Behavioral Sciences, S.U.N.Y., College of Arts and Sciences, Plattsburgh, New York. 99 9 1973 Plenum Publishing Corporation, 227 West 17th Street, New York, N.Y. 10011.

100

Hou~on

working within a Piagetan framework have ascribed some of these differences to egocentrism, or the lack of fixed ego boundary leading to the child's unawareness that others do not share his viewpoint automatically, nor vice versa; and to syncretism, the resultant idiosyncratic communication technique manifested by the child's apparent confusion of causal, logical, spatial, and temporal connections in his discourses. Syncretism in 6-8-year-olds was studied some 35 years ago by Piaget (cf. Piaget, 1955, p. 93 ff.), in an experiment, the design for which was adapted for the research reported in the present paper. Piaget worked with the children in pairs, telling one child from each pair a story, which that child was instructed to tell to the second who subsequently repeated it back to Piaget. Although the results indicated that children of 7 or younger understood about 70% of what they were told and expressed about 82% of that in their narratives, Piaget noted at the time his belief that even these numbers fail to convey the total extent of syncretism in children. One goal of the present research was to retest syncretism in young children using the Piagetan storytelling framework, employing, however, materials of greater interest to 5-6-year-olds and less lengthy and complex than those in the original study. Syncretism and its resultant failure at information transmission is but one item in the spectrum of distinguishing child language features. Child language is in a state of flux, and in even relatively advanced child speech certain constructions may fail to mature at the expected rate, owing not so much to infrequent appearance of these constructions in the adult model as to their syntactic characteristics. The psycholinguist Carol Chomsky recently studied the degree to which several structurally odd constructions were mastered by a group of children from 5 to 10 years of age (Chomsky, 1969). Of the items studied by Chornsky, two, which we shall refer to as "easy to see/hard to see" and "promise," were reinvestigated by the present research. According to Chomsky, a child given the sentence "John is eager to see" will correctly understand that John is the subject of the sentence and that the adjective eager is predicted of John, i.e., that John is doing the seeing. This is the standard predicate-adjective grammatical order in English-an order which is violated, however, in the sentence "John is easy to see," where John is not seeing but being seen. The problem, roughly a conflict between deep- and surface-structure subject, obtains similarly in the item "John promised Bill to leave" as contrasted to the normally patterned "John persuaded/ordered/told/ caused Bill to leave~ Chomsky tested the posited difficulty in interpreting these sentences by techniques such as the following: for the "easy/hard" construction, her child subjects were shown a blindfolded doll and asked whether it was easy or hard to see, and then to make the doll hard/easy to see (the adjective not selected);

Syntactic Complexity and Information Transmissionin First-Graders

10t

the "promise" sentence was tested bY giving the child two dolls to manipulate, representing Donald Duck and Bozo, along with such instructions as "Donald wants/tells[asks/promises Bozo to [perform some action such as a somersault]-make him do it" with the him deliberately ambiguous as to referent. Around 68% of Chomsky's subjects below the age of 7 misinterpreted the "easy/hard" construction and around 62% below 789 misinterpreted the "promise" construction. Although these results were suggestive, the deliberate use of misleading and tricky test items as well as the constrained and artificial nature of the study seemed to indicate the need for further examination of the topic in a different format. The Chomsky work gives insight into children's languageanalytical processes when presented with a linguistic puzzle; but a more interesting and relevant topic might be how children understand difficult constructions in a natural context such as a discourse or text, and in fact whether context might not usually disambiguate the constructions so as to moot the problem. Syncretism and unequal mastery of syntax are hypothecated problems of all child language; equally as compelling, however, are many problems peculiar to the language of specific groups of children, such as the currently controversial but little-researched putative communication impairment of the low-SES black child. The psychologist Bernstein has proposed (e.g., 1961) that Iow-SES children (white, in his studies) tend to learn a so-called "restricted code" or language of limited wider utility instead of the "eiaborated code" needed for dealing with schoolwork and other information, thus in effect limiting these children to immature and perhaps syncretistic communication forms longer than non-low-SES children. Other researchers (e.g., Bereiter and Englemann, 1966; Blank and Solomon, 1968) suggest that disadvantaged black children lack proper means to express causality or spatial-temporal sequences, and cannot communicate abstract concepts, given the limited capacity of their native language system. It could be predicted from these posttdates that low-SES btack children should perform even worse than nondisadvantaged children on experiments such as those by Piaget and Chomsky discussed above. The present study, then, was designed to investigate on a cross-cultural basis the dual hypotheses of syncretism and faulty mastery of the abovementioned syntactic constructions; in other words, to examine a relatively broad spectrum of postulated difficulties of child language development. Since in the constrained environment of many experiments, probably including that by Chomsky, disadvantaged black (and undoubtedly most other) children are Liable to show the highly atypical language and behavior I term the School register (cf. Houston 1969, i970, 1972) and thus achieve poor scores

102

Houston

nonrepresentative of their total behavioral/linguistic competence, I chose the relatively informal Piagetan storytelling framework for my study. Like Piaget, I worked with blocks of text rather than individual and more easily scorable linguistic items, and for this reason I also share with that psychologist a method of reporting my results dependent partly on statistics and partly on more empirical anecdotal presentation of the children's performance and how it may be interpreted. As with all studies of this sort, Piaget's of course included, some results which I consider highly significant can only be described by reporting my observations, unsupported by extensive numerical data-in some instances, such as those concerning paralinguistics, because psycholinguistic technology still lacks suitable analysis methods for these phenomena. Also included below is a somewhat detailed discussion of experimental procedures and difficulties encountered therewith, which I hope may serve as a guide to those planning future child-language research similar to that reported here.

SUBJECTS AND MATERIALS

The target population consisted of about 172 first-grade children (i.e., 5-6 years old), drawn from five schools in the Berkeley-Oakland schools systems. Preliminary analysis of scores from these schools treated the populations as consisting of lower, middle, and upper SES (socioeconomic status) levels. However, this analysis indicated that in fact the population consisted of two groups rather than three, with the dividing line falling, as might be predicted, just around middle-middle class. I have termed the two SES levels into which my subject population falls "M-I" and "M-2" rather than "lower" and "upper," to get away from traditional and often misleading SES labels; t h u s my design was in effect a 2 • 2 comparison of M-1 white, M-2 white, M-1 black, and M-2 black (although as it developed, the M-2 black population consisted of only three children, too small a sample to figure in the statistical analysis). The M-1 level ranges from around upper-lower class to just below middle-middle; M-2, from about middle-middle upwards. I add my tentative hypothesis that these two SES's, rather than three or more, are the most viable on a sizeable number of verbal and scholastic measures. The experimental materials consisted of two stories, originally containing 9 sentence units and 109 words each, a length within the memory span of the typical first-grade child. Included in the stories were several examples of the target syntactic items, "easy to see/hard to see" and "promise," as well as a number of other constructions difficult for this age group, such as passives, locatives, and temporal-sequence indicators. The children were worked with in

Syntactic Complexity and Information Transmission in First-Graders

103

pairs as in the Piagetan test, by my telling child A one story and having him tell it to child B who would subsequently retell it to me; each child from a pair would get to tell one story and retell one, with the order of stories alternated with each pair of children. The stories used in this study are as follows: (a) "C": Once there was a magic shell which could make wishes come true. The shell was found on the beach by a girl named Carol, who said, "I wish I had a puppy." Then she looked around, but she didn't see any puppies. Since the beach is flat, if there had been a puppy it would have been easy to see. So Carol went home sadly. When she got home her mother said, "I have a surprise for y o u - l o o k what I found in the back yard." Her mother gave her a little brown puppy. Carol was very happy, and she promised her mother to take good care of the dog. (b) "JP": This is a story about a little boy named Jimmy. Jimmy promised his mother to take his sister Penny to the park. In the park, Penny saw a man selling ice cream. She asked her brother to buy her an ice cream cone. But Penny wanted to play a trick on Jimmy. So while he was getting ice cream, she ran and hid under a bush. Jimmy looked everywhere for his sister, but she was hiding so she was hard to see. Finally Penny jumped out laughing and said, "I fooled you." Then Jimmy gave her the ice cream cone, and made her promise never to fool him like that again. These stories were constructed to be approximately equal in length and difficulty. Unfortunately this did not work out in practice, owing to two problems concerning the story titled "C." First, a characteristic of children at the syncretism stage is to omit semantically marginally significant material from their retelling of narratives. For this reason most children observed omitted the fourth sentence of "C," that beginning "Since the beach . . . . " thus shortening this story by one sentence in effect and so presumably simplifying it relative to "JP." On the other hand, it also developed that the concept of a magic shell which makes wishes come true and grants a little girl's wish for a puppy in a rather oblique way is too subtle for first-graders. Again in typical syncretistic fashion, first-grade children either omit the "shell" portion of the story entirely, or alternatively (and gratifyingly to the Piagetan psychologist) mention the existence of the shell without tying it to the remainder of the story or questioning the purpose of its presence, as in,

104

Houston

"There was this shell, and it was magic, and a little girl wanted a puppy, and she couldn't find one so she went home . . . . " etc. Analysis of the proportion of children performing in this manner appears below.

PROCEDURAL NOTES As noted, the subjects of this study were approximately 172 first-grade children from five schools in the East Bay Area of California; the tests were conducted in the schools themselves. The children were prepared by being told by their teachers that they would have a chance to .play a storytelling game. The teachers of each first-grade class selected pairs of children for the study randomly, making sure only to mix the children so as to avoid all same-sex or all opposite-sex pairs. For each Segment of the study children were chosen from the same classroom, so that members of a pak were presumably known to each other though were not necessarily best friends. When the pair arrived at the site of the experiment (most often the school nurse's office or the auditorium) I explained the procedures to them in a standard fashion, for instance: "We are going to play a storytelling game. One of you will wait over there [or outside the room, etc.] and Iql tell the other one a little story. Then we'll call back the person who was waiting, and the one I told the story to will tell it to him, and he'll tell it back to me." After making certain the instructions were comprehended by both children, as they invariably were, I let the children themselves decide who was to take the first turn at hearing and retelling a story. Once the first participant had been chosen, the other child was sent out of hearing range, the first child was reinstructed and urged to remember the story he would hear so he could repeat it "just the way I tell it," and I proceeded with the story. After the child had heard the complete story he would often protest that he would not be able to remember the whole thing; since these protests did not correlate in any way with the child's subsequent performance, they were either ignored or answered by simple reassurances. Then child B was called back, child A told him the story and was recorded doing so (on a Uher 4000 Report-L, on the desk in view of the children), and child B retold the story and was likewise recorded. It is worth noting that although this type of task is not one at which the children had had previous practice, they all accepted it without question or protest, and all children who delivered a performance on the task appeared to make an effort to perform to some personal criterion (cf. below). Since a central purpose of this study was to elicit natural language and behavior from children, avoiding the nonrepresentative School-register language often observable in experimental situations (cf. Houston, op. cit.), I feel it

Syntactic Complexity and Infoxmation Transmission in First-Gradezs

t05

would not be extraneous to this discussion to include a few words about child-interview techniques. My observation of teachers, student experimenters, and others over the years has led me to posit the existence of a register or at 1east style of language used by adults dealing with children, especially rely young children. This register may be characterized among other things by shortened sentences, slowed speech rate, unusually high numbers of intonation and stress peaks, and often simplified content as well. Gestural and postural abnormalities may appear as well. The effect of the register is often to elicit in the child being addressed a correspondingly marked register of language and behavior, that which I have referred to above as the School register (cf. Houston, 1972, for a full description of this type of language). The elicitation of linguistic components of Nonschool-register or "natural" language from children can be accomplished by eliciting the behavioral components of this register, by creating for the child the type of setting in which Nonschool-register behavior patterns would be appropriate. Since the experimental or observational setting with the presence of an unknown adult is a priori a School-register-eliciting situation, this is somewhat difficult for the experimenter. On the other hand, it should be noted that children will use Nonschool-register language and behavior with adults whom they know well, like, or trust; note too that such adults do not themselves typically use the peculiar types of child-addressed language and behavior described above. The key, then, would appear to be for the experimenter to behave as though he were in the role of an adult with whom the child would use Nonschool-register patterns. I have taught student experimenters to approach this by behaving toward the child subject, immediately upon meeting him, as though he were a person of their own age whom they have known for about 15 years. The purpose of the latter instruction is to get the experimenter over the initial stages of acquaintanceship with the child instantaneously, b y ignoring their existence, and to put the experimenter into the role of a Nonschool-register-eliciting person who is not addressing the child in an artificial or stereotyped manner. One might mention that insofar as it is possible to elicit natural language and behavior from a given child, this technique should succeed; but an occasional child cannot be induced to participate by any means. About six children from my sample were nonperformers and had to be excused, although a somewhat larger number finally agreed to participate after I showed patience with their initial shyness. SCORING The scoring of the material was carried out as follows: For each of the two stories used, text points or content items and syntactic items were scored

106

Houston

separately on both Tell (child A's recitation) and Retell (child B's recitation). Four possible scores were given for each item: if a child reproduced verbatim an item as given by me, he scored an R. If he produced essentially the same content (for text items) or a transformed or paraphrased version of the construction regardless of content (for syntax items) he scored a T. If the child omitted the item, he received an O; and if he attempted the item but erred he received an X. Addition of R + T scores thus yields a figure approximately equivalent to "percent of items correct," while addition of O + X scores = "percent of items incorrect." The initial hypotheses concerning these scores were that the main score differences would occur between M-1 and M-2 regardless of race, and that race would be a significant variable within M-1 but not within M-2 (note, however, that the M-2 black population was excluded from the numerical analysis because of its extremely small size relative to the other populations). The hypotheses were based on previous study of the language and behavior of children from each population. M-2 scores were expected to exceed those of M-1 for all measures, as is typical of such studies, and this was predicted to be especially noticeable for syntax-it was hoped that this study might provide some clues as to causes for this divergence. Both groups were expected to be about equally competent in retelling stories. In Table I are the means for three groups on all the main measures, i.e., mean percent of Rs and Ts achieved on Telling and Retelling by each group (M-1 black, M-1 white, M-2 white) for both text and syntax items; the three Table I a

R-tell Text T-tell Text R+T-tell Text R-tell Syntax T-tell Syntax R+T-tell Syntax R-retell Text T-ietell Text R+T-retell Text R-retell Syntax T-retell Syntax R+T-retell Syntax Tell-Re, Text Tell-Re, Syntax II

M-1 B

M-1W

M-2W

46 8 54 22 22 44 30 10 40 13 16 29 14 15

50 8 58 26 27 53 33 8 41 13 17 30 17 23

58 9 67 31 36 67 40 10 50 31 24 55 17 12 I

aApproximate n's: M-1 B, 59; M-1 W, 79; M-2 W, 31. Note that figures are given in percents.

Syntactic Complexity and Information Transmission in First-Graders

t07

black children constituting the M-2 black population were scored separately and are not included on this chart. Since the total number of possible points differed between the two stories, the scores have been converted to percents and combined in Table I. As indicated above, %R + %17 = total mean percent correct; 100% minus this figure = total mean percent incorrect or omitted (cf. below). A Tukey analysis for multiple comparison with unequal n's performed on this data revealed significant differences between M-1 and M-2, at the 0.05 level, on the following measures: R-tell, syntax; R-retell, syntax; R-tell, text; R-retell, text; T-tell, syntax; and T-retell, syntax-in other words, all main measures except T-tell and T-retell text, for which all populations scored nearly identically. Although M-1 black/M-1 white comparisons were not statistically significant, the scores were consistently different, with M-1 white scoring somewhat higher than M-1 black. It will be recalled that the original hypothesis proposed that M-1 black and M-1 white are different populations, a conclusion supported by the trend of these scores along with much corroborative data from largely nonquantifiable parameters of the study, as discussed below. An informal examination of the scores (and other behavior) of the three subjects in the M-2 black population indicated no difference between their performance and that of the M-2 white population, as predicted. The column in Table I marked "Tell-Re" gives for each population the difference between T+T-tell and R+T-retell, an approximate indication of how much worse children were at retelling the stories than at telling them, or otherwise stated, how much of the original content and syntax of the stories was lost between the recitation of child A and that of B. The lower the difference score, the better the group as a whole did. The M-1 white group was on the whole the least successful at preservation of content and syntax in retelling. Table II shows additional scores on several measures of varying significance. The item marked "Names" was included because of my observation that the proper names in the stories gave the children considerable difficulty. There seems to be a large difference between M-1 and M-2 ability to repro'duce names; the reasons for this are not obvious, but in any case this was the item which most consistently produced problems for the children (aside from the seashell/wish connection in "C"). Interestingly, too, forgetting the names of story characters caused most children considerable anxiety, such that they would often pause and refuse to continue until I had supplied a name or told them they could make one up; other items only partially remembered rarely caused such difficulty. Some minor snags occurred in connection with the "easy/hard" test, since in neither of the stories was the sentence containing this item essential

108

Houston

Table II a i

Names e]h Q e/h play Correct, eyes shut Incorrect, eyes shut Correct, eyes open Incorrect, eyes open Shell-wish connection Promise-JP-1 JP-2 C Elab

M-1 B

M-1 W

M-2 W

49 80

44 77

22 90

25 41 17 17 5.2 24 56 51 26

15 25 30 30 2.5 33 58 40 8.5

(not asked) (not asked) (not asked) (not asked) 0 23 77 32 11.5

aApproximate n's: M-1 B, 59; M-1 W, 79; M-2 W, 31. Note that figures are given in percents. enough semantically (cf. below) that the children consistently reproduced it, although I felt, as Piaget might have said, that they understood it. In order to probe this, I commenced asking the children after the telling and retelling of "JP" whether the girl hiding in the bushes was easy or hard to see, and why. The child was marked correct on this if he gave either answer with proper justification, and incorrect if he said the equivalent of "easy to see because she could look out OK," a frequent type of wrong answer. The majority of all populations were correct on this, as shown in Table II in the row marked "e/h

O

. "

To test this construction further, about two-thirds of the way through the study I began asking one member of each pair to close his eyes, then inquiring of both children whether the one with his eyes closed was easy or hard to see, and why, alternating with each pair which child was asked first. Most often the child with his eyes shut replied incorrectly that he was hard to see, while the child with his eyes open gave either answer with precisely equal probability (the children were not operating randomly, however). This result, shown in Table II as "e/h Play," occurred regardless of the children's replies on the earlier question ("e/h Q"). This equivocal outcome leaves one in some doubt about how to summarize first-grade children's knowledge of the "easy/hard" construction. Although Chomsky was correct in stating that the construction causes some amount of difficulty, children's ability at manipulating and comprehending the construction is clearly dependent upon the context in which it occurs. One can perhaps only point out once again the complexity and inseparability of syntax acquisition and concept acquisition as maturational/learning processes, and the difficulty of summarizing the exact

Syntactic Complexity and Information Transmission in First-Graders

109

stage of development of any single concept or syntactic unit at any early age. The percent correct reproductions o f each of the three "promise" constructions appears in Table II; the item reproduced most often is that in the last sentence of "JP," for which most children specifically stated that it was Penny who promised not to play a trick on Jimmy again rather than vice versa. The promise construction in "C" was tested by asking directly, "Who was going to take care of the dog, Carol or her mother?" There is no interpopulation significant difference on either the "promise" or the "easy/ hard" constructions. In regard to the apparently low rate of success of all populations here, however, one should note, for example, the fact that 24% of M-1 black occurrences of the first "promise" item in "JP" were correct does not indicate that 76% were incorrect, but rather that 76% were either incorrect or omitted. The significance of this observation will be discussed in the following section.

ANALYSIS OF CONCLUSIONS It should be carefully borne in mind that all figures shown in the tables are indications of the amount of material correctly produced by the children. They are not indications of the amount of material understood or remembered, and there is little reason in any given instance to assume either that what a child says is all he understands or remembers, or the converse-like nearly all similar studies, this research deals with performance only, not with competence or the internalized basis for performance. Although linguistic or behavioral competence can be inferred from performance measures such as those used here, the two are not isomorphic at any stage of development. A further fact not directly observable from the tables is that, as explained previously, two types of "error" were possible on all items, namely actual errors and omissions. For all populations the majority of errors on all measures were omissions. The total "X" or real-error score per child was around .3X for syntax and the same for text, with no differences among populations. On tasks of this type children apparently do not tend to produce incorrect material: one of the advantages of the text or discourse situation over the single-item response situation may be that it reveals this tendency~ Children either get text material right, in some sense, or they leave it out. Most of the already low real-error score is accounted for by minor sequence errors rather than errors of actual content or syntax. From these facts, and from study of the recorded text of each of the 344 storytellings occurring in this research, I am led to the following hypothetical remarks concerning syncretism. It appears at first that all the

110

Hou~on

scores in Table I are quite low, that the children are reproducing very little of the material presented (a maximum of 67% of text items, for M-2 white) considering the supposed state of their language acquisition by the age of 5-6. By themselves, without closer examination of the actual stories, such scores might lead to postulation of an attentional defect, or perhaps highly limited medium-short-term memory span, on the part of children at this age, if not syncretistic understanding and processing of the material itself. On the other hand, such detailed examination of the children's narratives may suggest that a rather different cognitive process is responsible for the results. Imperfect reproduction of verbal material is characteristic of children at several developmental stages, of which the best-analyzed to date is so-called telegraphic speech, typically present in the child beginning to form two- and three-word sentences. Such a child uses the telegraphic form both in spontaneous utterances and in imitation of adult sentences, showing the process is occurring at a deeper level than mere "lack of short-term memory" in imitation (for some discussion of this process, cf. e.g. Brown and Fraser, 1964; Miller and Ervin, 1964; McNeill, 1966). A child speaking telegraphically omits many words from the stimulus sentence in imitation, typically retaining only contentives and omitting functors. For instance, given the stimulus sentence "Do you want to go for a walk now?" the child might imitate it as "Go walk"; given "Daddy will be home from work soon," the child may say "Daddy home," etc. In telegraphic speech, which is both rule-governed and complex, the child follows a regular pattern in his omissions and retentions when imitating. In linguistic terms he retains just those contentives which are the items of focus in the stimulus sentence, or roughly those which are most central to the semantic content of the sentence (a number of other factors, such as which items in the target sentence received strong stress, may also influence retention of words in telegraphic imitations). These items of focus are nearly always imitated in the correct order. Processes analogous to those in telegraphic speech may be observed in story-reproduction tasks such as the one reported here, leading to the fascinating speculation that a similar set of cognitive rules underlies both. The 50-60% of text material reproduced by children in this study, I have found, invariably (with fewer than a half-dozen exceptions) consisted of what may be termed the items of focus from each story. A "plot condensation" of "JP" might note that a little girl played a trick on her brother by hiding, and that he found her and told her not to do it again; of "C," that a girl wanted a puppy and after searching unsuccessfully was given one by her mother. No matter how deficient a retelling a child produced, if he said anything at all he gave these facts, in their correct order, rather than engaging in the randomized

Syntactic Complexity and Information Transmission in First-Graders

111

reproduction of individual details in idiosyncratic order that one might predict from syncretism theory. Further content filled in by the child also appeared in the correct order; its appearance was probably also due to saliency or semantic ranking by the child, although this hypothesis was not tested directly within this study. It may be worth positing, therefore, that children's repetition 'of stories is rule-governed just as is their linguistic development, with the rules operating similarly to those producing earlier telegraphic speech: in other words, that both processes rely upon reproduction of that content which is most salient or focused upon, elimination of nonsalient details or reduction of these details to their minimum elements, preservation of the order given in the original, and the omission of forgotten or misunderstood material rather than overt error. The next point to be discussed here is that which I consider both most interesting and important of the fruits from this study and least dearly supported by the numerical data reproduced in Tables I and II above. Despite the fact that I have only one actual measure in support of the hypothesis that follows, and that it is based largely on my own observations of the language and behavior of the children in the present study (and in my past researches), I believe that its significance to psychologists and educators is great enough that I will state it here tentatively prior to carrying out the further work I have planned to test its validity. The hypothesis is a broad one regarding the differences between M-1 black subjects and all others in this study and other similar researches. In particular, it seems clear that the differences in scores between these populations are not due to lack of understanding of the test instructions by any of the children; this is shown by their uniform carrying out of the given instructions without question or hesitation. Nor are the differences due to lack of understanding of any of the material, semantically or syntactically, by any child; I verified this observation by noting that all children, even those producing minimal recitations, did give accurate abstracts of the plot of the stories as reported above. The differences in scores between populations are, I propose, due rather to the effect of the given instructions on the children, as follows. A set of instructions such as those given the subjects in this study, as well as in many other psychological-research or scholastic situations, presumably creates in the child an intentionality to perform in a specific way. In the many white children I have observed, instructions of this sort create a vector toward following the instructions exactly, with emphasis on correct reproduction of details, lack of deviation from the given norms, verbatim rendering of memorized material if this is relevant; such children wilt actually verbalize

112

Houston

comments very like these when asked their concept of a "good performance" on such tests. In this study, if a M-1 or M-2 white child could not remember details he omitted them without substituting others in their place or attempting them and erring. The M-1 black child, on the other hand, hearing the same instructions, has created in him an intentionality toward taking the required behavior as a base line upon which to demonstrate individual imagination and creativity. Reproduction of general meaning elements is seen as far outweighing reproduction of even relatively salient details as desired behavior. In other words, whereas the white child demonstrates personal competence by his precise and nondeviant performance on relevant tasks, the black child demonstrates personal competence in proportion to the degree of style and flair with which he performs. Behavior by a black child which is normative by the white child's standards is an indication of dullness and lack of competence to his black peers. This proposed explanation for differences in black/white performance on experimental and scholastic tasks is supported by the measure I term Elaboration on the storytelling experiment reported here; it appears in Table II above under the heading "Elab.," and refers to the introduction of extraneous or original items of content into the stories. The M-1 black group far exceeds the other groups in this measure; 26% of all M-1 black stories included elaboration as against fewer than 12% in all other populations. When a lower score appears in Table I for the M-1 black population, it is often because the children, instead of repeating the story as given, were telling a partially or almost completely different story, retaining only the armature of the given story. Although it does not appear on the chart, a list of such inserted items reveals that those added by M-1 black children tended to have semantic content more distant from that of the given story than those added by other populations, and to contain a quantitatively greater amount of novel material. In short, my impression of events taking place in this study is that the black children were performing a different task from the other populations, despite having been given the same instructions and having "understood" them in some sense just as clearly. Although it is most difficult to measure these factors, this difference in attitude and approach to the experimental situation was apparently accompanied by rather striking behavioral differences between M-1 black and other children, differences which seem to me to indicate a broader cultural divergence between this population and others than has previously been supposed. For instance, the children from the M-1 black group engaged in much greater interaction with each other than other children did, visibly attending to each other at all times instead of reciting the stories for my approval. They attempted to help each other often; M-1 black children

Syntactic Complexity and Information Transmission in First-Graders

113

prompted each other nearly twice as often as other groups, and appeared interested in cooperation rather than competition. M-1 black children were freer in their paralinguistics and general nonverbal behavior than other populations, generally using broad-base kinesics involving the whole body rather than the narrower and more confined kinesics of the other populations, including the three M-2 black children. It should be mentioned too that typical M-1 black behavior was exhibited by children of this population even when paired with M-1 white children. The fact of being black, while apparently the major influence on the child's own motivations and attitudes toward the task, did not influence the behavior of the child perceiving him, nor was there much communication between black children which could be characterized as exclusively black, sociolinguistically speaking. A black child performing as child B of a pair would often engage in elaboration of the story told him, despite much head-shaking and other overt disapproval of a white child paired with him; likewise the typical white child did not engage in much or any elaboration despite obvious boredom and exaggerated lack of interest in his retelling by a black child who had told him a story. This finding is in line with previous observations I have made with black children up to the age of about 12 (cf. Houston 1969, 1970, 1972), in which I have noted that sociolinguistic variation such as register is not dependent upon the race of the person with whom the young black child is interacting. In sum I might note my hypothesis that the very close similarity of Black English and White English in this country may mask quite important and far-reaching cultural differences, among which may well be the attainment of stages of verbal maturity by black children at an earlier age than white children (cf. e.g. Entwhislte, 1966, for evidence that black children tend to develop adult word-association patterns prior to white children); and that these differences, which are qualitative and not quantitative, may account for numerous reports concerning black/white differences on learning and performance tests and scholastic measures.

REFERENCES

Bereiter, C., and Englemann, S. (1966). Teaching Disadvantaged Children in the Preschool Prentice-Hall, Englewood Cliffs, N.J. Bernstein, B. (1961). Social structure, language, and learning. Educ. Res. 3: 163-176. Blank, M., and Solomon, F. (1968). A tutorial language program to develop abstract thinking in socially disadvantaged preschool children. Child DeveL 39: 379-389. Brown, R., and Fraser, C. (1964). The acquisition of syntax. In Bellugi, U., and Brown, R. (eds.), The Acquisition of Language. Monographs of the Society for Research in Child Development, Vol. 29 (1, Serial No. 92), 43-79.

114

Houston

Chomsky, C. (1969). The Acquisition o f Syntax in Children from 5 to 10. M.I.T. Press, Cambridge, Mass. Entwistle, D. R. (1966). Word Associations o f Young Children. Johns Hopkins Press, Baltimore. Houston, S. H. (1969). A sociotinguistic consideration of the Black English of children in northern Florida. Language 45: 599-607. Houston, S. H. (1970). A reexamination of some assumptions about the language of the disadvantaged child. Child Devel. 41(4): 947-963. Houston, S. H. (1972). Child Black English: The School register. Linguistics 90: 20~ McNeill, D. (1966). Developmental psycholinguistics. In Smith, F., and G. A. Miller (eds.), The Genesis o f Language. M.I.T. Press, Cambridge, Mass., pp. 15-84. Miller, W., and Ervin, S. (1964). The development of grammar in child language. In BeUugi, U., and Brown, R. (eds.), The Acquisition of Language. Monographs o f the Society for Research in Child Development, 29 (1, Serial No. 92), 9-34. Piaget, J. (1955). The Language and Thought o f the Child. Meridian, Cleveland (tst ed., 1929).