bidialectal theory 2

7
COMMUNICATION STYLES
It seems inevitable that class-and race-related language differences lead to questions about heredity. It is not proposed to argue these here. In opposition to a position that there are innate differences in linguistic capacity, is the position that languages, Jamaican Creole and Standard English, in particular, are equally facile in their capacity to represent meaning.
Some of the many differences between JC and SE are, more basic sentence forms and relationships, shorter sentences, more numerous sentences per given quantity of words, concrete nominal and more direct verbal vocabulary, less adjectives. There are also fewer of some transformationally derived lexical subcategories and less diversity of vocabulary. One other feature of Creole speech is a larger quantity of words per unit time (Craig, 1974, 1976). It has been suggested that the last feature is "probably attributable to the fact that . . [Creole] . . . communication style requires less complex internal organization before an output of language."
For present purposes there are two important features of Creole communication style.
1) Fewer complex sentences and fewer markers of types of embedding.
More conjoining as a means of linking sentences (Craig, 1977, p. 327). Again, there is some similarity between Creole and the features of Black English described by Houston (1969), who noted that in the compositions of some Florida tenth graders, the students used "contatenation as principle [sic] device for creating discourse." At the extreme Creole position, is a type of production whose surface structures resemble the deep structures of English. This can be seen in the following set of sentences.


8
lim tek stuon hit di guot ma him ed, an di guot ded him take stone hit the goat in him head and the goat dead
He killed the goat by hitting it in its head with a stone (Craig, 1976, p. 120). The issue is claimed to be a matter of style rather than meaning, the difference being the point at which meanings are lexified, or put into words. In the first and second sentences above, the ideas are lexif ied as they occur, so to speak, but in the third, there is a reformation of events in the left to right reading. In terms of actions, killing was a consequence of hitting. Note that the first and third sentences have opposite orders of mention of events. The Creole speaker then, has an early lexification format and the Standard English speaker has a late lexification format (Craig, 1974).
Of course, patterns such as the sentences just discussed probably represent modal performance, and corpuses have been corrected in which there is considerable embedding in Creole speech. In fact, in the lexicon, "we" is equivalent to "who" or "that" (Bailey, 1966), and serves the purpose as a marker in relativization. To characterize Creole speech as being composed of conjoined sentences is incorrect, less harshly, it is an oversimplification. However, for convenience in providing a manipulable independent variable it will be supposed that Creole speech can be represented by conjoined sentences.
One difficulty that might arise in comparing Creole and English sentences, is the degree of abstractness. Since, the vocabulary is more concrete, it is expected that in some cases, a Creole sentence will be less economical in terms of the number of words. If the English sentence "He is untrustworthy" is to be adequately represented in Creole, it would be:
im tel lai an [im] tiff
him tell lie and him thief (Craig, 1977)






9
It is clear that the single proposition expressed in English is equivalent to two propositions in Creole. A limitation will be placed on the types of sentences which can be compared.
AN EXPERIMENTAL TRADITION
It should be obvious from the earlier remarks that Caribbean education lacks an input from in situ experimental psychology. However, "empirical' support has been given by sociolinguistics, at least in the context of the present problem. Somewhat fortuitously, a powerful paradigm appears to have some relevance to the problem being developed. A tentative statement of the problem can be made at this point: specifically, the intention is to test the hypothesis of equal facility to represent meanings in both languages. A secondary concern is the accumulation of data towards a characterization of bidialectal semantic memory.
Sentence and idea comprehension have been studied using the acquisition-recognition paradigm of Bransford and Franks (1971). Central to the paradigm is "assimilation theory," which in the case of linguistic stimuli, implies that the meanings derived from sentences are not merely the and-sum of the propositional content of each sentence, but that the meaning is integrated across sentences (Bransford and McCarrell, 1977). In their most relevant statement on semantic integration, Bransford and Franks discuss the possible effects of syntactic complexity on comprehension:
If paragraph comprehension were equivalent to memory for just those sentences comprising it, one could argue for the communicative efficiency of using all syntactically simple sentences. However if Ss spontaneously integrate information from various acquisition sentences, syntactic simplicity may not always correlate with ease of comprehension. If Ss have to integrate information anyway, it should help if syntax does this for them, at least partially. In fact, this is precisely what embedding transformations do. They integrate information that could be expressed in separate sentences.



Further:
Simple syntax may actually hinder comprehension by forcing Ss to do too much of the integration. And more complex structures (like some forms of embedding) may actually facilitate comprehension by explicitly expressing cohesive, easily codable semantic integrations of ideas.
(Bransford & Franks, 1972, pp. 244-245).
In other words, the cognitive equivalence and communication efficiency of early and late lexification formats (Craig, 1974) is challenged.
REVIEW OF RESEARCH
Only one formal report of intervention using the "free talk, controlled talk" method, could be found. A six-month study of seven year old rural Jamaican primary school children. There were five groups, four working class,C10, E10, E20, E30 and one middle class, E40. Ns were not reported. A speech production test, a picture description task, was given at the beginning, after three months and after six months. The four experimental groups were instructed by the method described earlier, and the control group was subject to an unspecified treatment. It had an initial advantage, in that its members had all had infant education, which the experimental groups did not have. Four measures were taken or computed: (1) the average proportion of clauses containing non-standard verb patterns or non-standard relationships between nominative and verb, (2) the average proportion of clauses linked together by "and," (3) the average length of clauses in terms of words, and (4) the average length of children's speech in terms of words. There were no significant changes in the control group on the four measures, when the intermediate and final scores were compared. The "disadvantaged" groups E10, E20, E30 showed the expected decreases on non-standard verb constructions and on the clauses introduced by "and." In addition, there were increases in the average number of





words and in the average length of clauses. Craig attributes the increase in quantity to an "increase in confidence." He states:

It seems likely that these children even at this age, are conscious of the unusual language requirement of the formal situation; when they know instinctively that the required language is available to them, the result is a lessening of inhibitions and a greater flow of language.
(Craig, 1967, p. 139)


Clearly, there is an affective component that must be accounted for, in the production of complex speech, but it is not being addressed here. It would be unfortunate if the belief that coordinations are a permanent feature of Creole language is an artifact of interview procedures. [Houston (1973) claims to have developed techniques of enabling subjects to overcome their apprehensiveness.]
The effectiveness of the method of instruction has been demonstrated, although it was admitted that it was more of a tryout than a formal experiment. The important question is however, what did the children learn? How did the alteration of their production patterns affect their cognitions? Are more semantic relations available to a user of syntactically more complex speech? We will defer answers to these questions; indeed we may not obtain answers for this age group of subjects. Although extrapolation from another age group might not be valid, that might be the best that can be done.
The variation in Creole speech, with respect to the primacy of conjoined versus embedded sentences, is evident in some further research reported by Craig (1974). Upper (U), rural-lower (RL) and urban-lower (L) SES Jamaican children between the ages 6 1/2 to 7 1/2 were compared on a number of speech variables. Spontaneous speech was recorded for approximately 11 minutes, for 63 groups of 5 children each. More simple sentences were produced by L and RL than U, who produced more sentences of more than two clauses (without embedding).


12
L and RL groups produced more complex sentences with at least one embedded clause. Craig claims that "JC, as a morpho-syntactic system, is in some aspects at least, discrete from English and can . . . have different communication conventions relative to the use of clause-embedding" (Craig 1974, p. 15). The embedded sentence is used where a verb-based phrase is used in English. So,
Climbing the tree is easy.
is equivalent to
it iisi wen yu a klaim di trii.
It is easy when you are climbing the tree.
"time" and "place" are expressed using clauses where English uses some format. So
in the rainy season .
is equivalent to
wen (taim) rien fall (plenti) when the rain fall plenty.
That the presence of embeddings is not an either-or situation, does not reduce the addressability of the extent of assimilation of propositions at varying degrees of derivational complexity although some artificiality is introduced. If JC and SJE can be taken to be two languages, then there are three idealized speakers to consider; JC and SJE monolinguals and the bilingual. The best means of identifying these speakers, [although it is practically an impossible task] is to use SES classification (Craig, 1974). It has already been suggested that some indication of the nature of knowledge of a language can be obtained using a semantic integration paradigm.
Bransford and Franks (1971) designed a series of experiments to test the relation between the language of sentences and the memory for ideas explicitly


and implicitly stated by these sentences. The wholistic idea from four simple declarative propositions can be stated in a complex sentence. An example of such a sentence was "the ants in the kitchen ate the sweet jelly which was on the table." The simple propositions were "the ants ate the jelly," "the ants were in the kitchen," "the jelly was sweet," " the jelly was on the table." At an intermediate level of complexity were the sentences "the ants ate the sweet jelly" and "the ants in the kitchen ate the sweet jelly." Sentences used were ONES, TWOS, THREES or FOURS depending on the number of propositions they contained. ONES, TWOS, and THREES were presented in the acquisition phase, while in the recognition phase, the corresponding FOUR and other ONES, TWOS, and THREES were included. It was found that the subjects, who were college students, "recognized" sentences were not presented in the acquisition phase, including the FOUR. In a second experiment NONCASES were presented in the recognition phase. These included information across idea sets or were constructed from an idea set, but had relations altered. NONCASES received the lowest recognition rating.
One point of dispute has been Bransford and Franks (1971) finding that there is linearity in the order of recognition confidence rating, with FOURS> THREES>TWOS>ONES. It was suggested that recognition rating is a function of syntactic complexity, but Katz et al (1974) found that if subjects were instructed to attend to meanings on recall, the effect is not present. Moreover the linearity effect occurs with meaningless material. This issue is not central to the demonstration of integration, which has always been replicated.
Singer and Rosenberg (1973) hypothesised that "the original grammatical functions of the words embedded in a sentence will determine the likelihood that a subject will report recognizing that sentence." Their replication of






14
Bransford and Franks (1971) Experiment II had a minor change; they included more between frame (bf) NONCASES and within frame (wf) NONCASES. In addition to obtaining similar results on integration, it was found that wf NONCASES were recognized to a greater extent than bf NONCASES. Further analysis showed that some ONES had been recognized with more confidence than FOURS. A regression analysis was performed to account for the recognition rates of sentences, given the types of grammatical relations in the idea set, which were expressed in the sentences. The relations were the main relation, subordinate object clause, subordinate subject clause, adjectival subject modification, adjectival object modification. One other variable was a syntactic change factor. The regression model accounted for 83.8% of the variance in recognition, but more important was the fact that a negative weight was obtained for the syntactic change factor. Subjects would be less likely to recognize sentences changed syntactically. The best positive predictor was the presence of the main relation.
In a situation in which it is being attempted to manipulate syntax to alter the extent of integration, the findings of Singer and Rosenberg are important. It becomes difficult to assess whether rejection of an idea is due to the detection of syntactic cues or to a genuine non-integration on a semantic basis. One way to sidestep the issue is consider JC and SJE as two different languages, rather than a single language at different degrees of transformational complexity.
Research on bilingual memory has largely been directed at memory for words. Although there is no consensus, it has been contended that there is a single semantic memory (McCormack, 1977). In attempting to demonstrate the unity of semantic memory, Rosenberg and Simon (1977), utilized the Bransford-Franks (1971) paradigm with French-English bilinguals. They used four idea





15
sets, with English (1), French (1) and French and English (2), sentences. Sentences which had been originally in one-language idea sets were "recognized" least in translated form, but those translated from mixed language idea-sets had recognition rates higher than sentences not previously presented. Rosenberg and Simon interpret these results to mean that subjects have similar, if not identical semantic systems.
It should be possible to obviate the problem of syntactic change as an artificial clue, by presenting mixed sentences composed of conjoined clauses, embedded clauses and both. From these results, some inference regarding the structure of bidialectal semantic memory can be made.
One difficulty, which has been dealt with obliquely so far, is the validity of the results of experiments on semantic integration with adult subjects, for an instructional problem concerning young children. Barclay and Reid (1974) gave three stories to kindergarten, first, third and fifth graders. In the materials, passive sentences were presented as full passives, truncated/ actor or truncated/no actor. Truncated passives were recalled best when no actor was supplied, and were recalled least when there was an appropriate actor in the story. This indicated that a truncated passive with an actor was integrated into either a full passive or the corresponding active sentence. There were no significant differences in the performance of the various groups. So although integration has not been demonstrated using embedded sentences with children, the phenomenon occurs with other transformations.
Differences between beginning 7th grade Creole and Standard English speakers have been obtained on mental ability and language learning aptitude tests (Craig & Carter, 1976); it was suggested that the differences were the effects of differences in the way in which language was used. This study was



correlational, and was therefore limited in the extent to which determinants could be identified.
Comparisons between Creole speaking children and British children have also revealed differences. Edwards (1976) administered a reading comprehension test to 40 English and 40 West Indian students between the ages 11; 3
and 12; 6. Slow, average and good readers were identified and comparisons were made on comprehension between groups of the same reading level. There were no significant differences between the groups of slow readers. For average and good readers however, the English children were significantly better on comprehension on the four more difficult of the six comprehension tests. Edwards attributed the result to "linguistic differences".
In an effort to identify determinants of interference, Edwards (1978) constructed a Creole Interference Test in which structures which were felt to distinguish British English from Creole were put in a multiple choice format. Subtests were vocabulary, syntax (including voice), morphology and phonology. English children scored significantly higher on all subtests. Caribbean-born and British-born West Indian children were significantly different on the tense and aspect, and morphology subtests only. Edwards laments that, "although many of the children sounded very English, it would appear that elements of Creole continue to influence their perception of British English." It was noted the differences between Caribbean West Indian, British-born West Indian and British children are not apparent in the comprehension of written materials suitable for children of reading ages of about 9 or 10.
From what has been a somewhat scattered review, some generalizations can be made. First, there is no evidence of any facilitating effect of Standard English for Creole speakers. Even if the Creole speaker overcomes differences




17
between the phonology of English and his spontaneous phonology, the differences in syntax put the Creole speaker at a disadvantage. Second, the previous studies have not been sufficiently analytical--although Edwards' test did manipulate the passive as a syntactic variable. Third, a semantic integration paradigm can elucidate some performance variables of Creole speakers and English speakers, with transformational complexity as an independent variable.
There is the problem of an appropriate population, that must be resolved. It is proposed that the memory requirements of the Bransford-Franks (1971) paradigm render it unsuitable for the target population of the "free talk, controlled talk" instructional method, so it will be applied to some intermediate age group, say 13 year olds. Although extrapolation of the findings is not justified, some insight of the desired state of bidialectal competence will be obtained. In the transitional period of instruction of Class C and Class D patterns (Craig 1966, 1967, 1976), the Creole speaker will be in the position of assimilating embedded propositions to a cognitive structure composed of predominantly conjoined propositions. It is the nature of the structures formed and the efficacy of their formation that will give a theoretical base to the "free talk--controlled talk" instructional method.

DESIGN
PART ONE
The materials will be administered to the groups in the following manner. In a repeated measures design, the performance of Creole and Standard English speakers will be compared on semantic integration of conjoined propositions and on embedded propositions.
Although the hypothesis of equal facility in the communication style of the language, is intuitively attractive, the
only 'empirical' evidence in



18
in support of this is linguistic analysis. It is therefore proposed that recognition scores on conjoined sentences will not be different for Creole and standard speakers; however the performance of Standard speakers should exceed that of Creoles on embedded propositions.

CONJOINED EMBEDDED
JC CC CE

SJE CC CE

SE>SC,CC,CE is the expected trend.
SUBJECTS
In accordance with previous research (Craig, 1974;Craig & Carter, 1976) JC speakers can reliably be found in non-selective secondary schools. SJE speakers (monolingual) could be found in elite high schools. Essentially, subjects can be identified by social class membership. Sample sizes are yet to be decided.
MATERIALS
The embedded propositions to be used are derived from those used in Bransford and Franks (1971) Experiment 1.
1.The big ants ate the sweet jam which was on the table. (R)
2. The cool breeze blowing from the sea stirred the heavy air. (R)
3.The rock (stone) that rolled down the mountain smashed the little hut on the riverbank. (R)
4.The tall man resting on the counter drank the beer from the bottle. (R)

From sentence 2, the following propositions can be generated.
ONES:
The breeze was cool. (A)
The breeze was blowing from the sea. (A)
The breeze stirred the air. (R)
The air was heavy. (R)
TWOS:
The cool breeze was blowing from the sea. (A)
The breeze blowing from the sea stirred the air. (A,R)
The breeze stirred the heavy air. (R)
The cool breeze stirred the air. (R)
THREES:
The breeze blowing from the sea stirred the heavy air. (A)
The cool breeze blowing from the sea stirred the air. (A)
The cool breeze stirred the heavy air. (R)

It is not necessary to rewrite the ONES for a list of conjoined sentences.

The breeze was blowing from the sea and the breeze was cool. (A)
The breeze was blowing from the sea and the breeze stirred the air. (A,R)
The breeze stirred the air and the air was heavy. (R)
The breeze was cool and the breeze stirred the air. (R)

THREES:
The breeze was blowing from the sea and the breeze stirred the air and the air was heavy. (A)
The breeze was cool and the breeze was blowing from the sea and the breeze stirred the air. (A)
The breeze was cool and the breeze stirred the air and the air was heavy. (R)
FOUR:
The breeze was cool and the breeze was blowing from the sea and the breeze stirred the air and the air was heavy. (R)

For the three remaining idea sets, the propositions will be generated in a similar fashion. The letters A and R indicate which sentences will be presented in acquisition and recognition.
PART TWO
One of the aims of this study is to attempt to characterize bidialectal semantic memory. The paradigm of Rosenberg and Simon (1977) will be adapted for this purpose. The same idea sets that are used in part one, will be used here, with the exception that idea set one, will be in embedded form, idea set four will be in conjoined form, and idea sets two and three will be mixed. Examples of mixed propositions will be shown below.

SUBJECTS
Selection for high school is based on performance on achievement and mental ability tests such as those administered by Craig and Carter (1976). It can be assumed that low SES students who are achieving in high school have attained bidialectal competence.
Creole < Bidialectal < Standard










21
It is proposed that the existence of a single semantic memory for bidia-lectals will be evident if the recognition of propositions for mixed (conjoined and embedded) idea sets exceeds that for standard speakers.

MATERIALS
MIXED THREES.
The breeze was blowing from the sea and the breeze stirred the heavy air.
The cool breeze was blowing from the sea and the breeze stirred the air.
The breeze was cool and the breeze stirred the heavy air.

MIXED FOURS (non-exhaustive)
The cool breeze that was blowing from the sea stirred the air and the air was heavy.
The breeze was cool and the breeze was blowing from the sea and stirred the heavy air.

PROCEDURE
For both parts, acquisition and recognition phases will follow the procedures described by Bransford and Franks (1971). There will be randomization of sentences with the constraint that no two sentences from the same idea set are contiguous, etc. The recognition task will be given without any fore-warning. Elliptical questions will be given after each acquisition sentence is presented, to provide an ostensible task.

FINAL REMARKS
Admittedly, support for the hypotheses propounded in this paper might not indicate how the instructional methodology for teaching Standard English to Creole-speaking children should be changed. Indeed, it might require no practical changes at all. However, at present with its emphasis on "forcing" and "imitation," its present theoretical basis is not cognitivist. It is hoped that this experiment will be a step toward a cognitive approach and toward an experimental tradition.
REFERENCES

Anderson, J.R. & Bower, G. H. Human Associative Memory. New York: Wiley, 1973.
Bailey, B. L. Jamaican Creole Syntax:A Transformational Approach.Cambridge: University Press, 1966.
Barclay, J.R. & Reid, M. Semantic Integration in Children's Recall of Discourse. Developmental Psychology, 1974, 10, (2), 277-281.
Bransford, J.D. & Franks, J.J. The abstraction of linguistic ideas. Cognitive Psychology, 1971, 2, 331-350.
Bransford, J.D. & Franks, J.J. The abstraction of linguistic ideas:A review. Cognition, 1972, 1, 211-249.
Bransford, J.D. &McCarrell,N.S. A sketch of a cognitive approach to comprehension: some thoughtsaboutunderstanding what it means to comprehend. In P.N. Johnson-Laird & P.C. Wason (eds.) Thinking: Readings in Cognitive Science. Cambridge: University Press, 1977.
Burns, R. W. Human Learning. In M.D. Dunnette (ed.) Handbook of Industrial and Organizational Psychology. Chicago: Rand McNally,1976.
Chomsky, N. Aspects of the Theory of Syntax. Cambridge, Mass.: M.I.T. Press, 1965.
Collins, AM., & Quillian, M.R. How to make a language user. In E. Tulving & W. Donaldson (eds.) Organization of Memory. New York: Wiley, 1972.
Craig, D. R. Teaching English to Jamaican Creole speakers. Language Learning, 1966, 1 & 2, 44-61.
Craig, D. R. Some Developments in Language Teaching in the West Indies Caribbean Quarterly, 1966, 12, (1).
Craig, D. R. Some early indications on learning a second dialect, Language Learning, 1967, 3 & 4.
Craig, D. R. Developmental and Social-Class Differences in Language. Caribbean Journal of Education, 1974, 1 (2), 5-23.
Craig, D. R. Bidialectal Education: Creole and Standard in the West Indies. Linguistics, 1976, 175, 93-134.
Craig, D. R. Creole Languages and Primary Education. In A. Valdman (ed.) Pidgin and Creole Linguistics. Bloomington: Indiana University Press,
1977.
Craig, D. R. & Carter, S. The Language Learning Aptitudes of Jamaician Children at the Beginning of Secondary School. Carbbean Journal of Education, Vol. 3, No. 1, 1-21, & Vol. 3, No. 2, 70-112.
Edwards, V. K. Effects of dialect on the comprehension of West Indian Children. Educational Research, 1976, 18, (2), 83-95.
Edwards, V. K. Dialect Interference in West Indian Children, Language & Speech, 1978, 21, (1), 76-86.
Engle, P. L. Language Medium in Early School Years for Minority Language Groups. Review of Educational Research, 1975, 45, (2), 283-325.
Houston, S. H. A Sociolinguistic consideration of Black English of children in Northern Florida. Language, 1969, 45, 599-607.
Houston, S. H. A reexamination of some assumptions about the language of the disadvantaged child. Child Development, 1970, 41, 947-963.
Houston, S. H. Syntactic Complexity and Information Transmission in First-
Graders: A Cross-Cultural Study. Journal of Psycholinguistic Research, 1973, 2, (2), 99-114.
Katz, S. Atkeson, B., & Lee, J. The Bransford-Franks linear effect:
Integration or artifact? Memory & Cognition, 1974, 2 (4), 709-713.
Labov, W. & Cohen, P. Systematic Relations of Standard and Non-Standard
Rules in the Grammar of Negro Speakers. In J.S. De Stefano (ed.)
Language, Society, and Education: A Profile of Black English. Wadsworth, Belmont, California, 1973.
Le Page, R. B. Problems to be faced in the use of English as the medium of Education in four West Indian territories. In J. A. Fishman, C. A. Ferguson, J. Das Gupta (eds.) Language Problems of Developing Nations. New York: Wiley, 1968.
McCormack, P. D. Bilingual Linguistic Memory: The Independence-Interdependence Issue Revisited. In P. A. Hornby (ed.) Bilingualism, Psychological, Social, and Educational Implications. New York: Academic Press, 1977.
Richards, J. P. Instructional Psychology: From a Behavioristic to a Cognitive Orientation. Improving Human Performance: A Research Quarterly, 1978, 7, 4, 256-266.
Rosenberg, S. & Simon, H. A. Modelling semantic memory: Effects of presenting semantic information in different modalities. Cognitive Psychology, 1977, 9, 293-325.
Singer, M., & Rosenberg, S. T. The role of grammatical relations in the abstraction of linguistic ideas. Journal of Verbal Learning and Verbal Behavior, 1973, 12, 273-284.
Weimer, W. Psycholinguistics and Plato's Paradoxes of the Meno. American Psychologist 1973, 28, (1), 15-33.

bidialectal theory 2

7
COMMUNICATION STYLES
It seems inevitable that class-and race-related language differences lead to questions about heredity. It is not proposed to argue these here. In opposition to a position that there are innate differences in linguistic capacity, is the position that languages, Jamaican Creole and Standard English, in particular, are equally facile in their capacity to represent meaning.
Some of the many differences between JC and SE are, more basic sentence forms and relationships, shorter sentences, more numerous sentences per given quantity of words, concrete nominal and more direct verbal vocabulary, less adjectives. There are also fewer of some transformationally derived lexical subcategories and less diversity of vocabulary. One other feature of Creole speech is a larger quantity of words per unit time (Craig, 1974, 1976). It has been suggested that the last feature is "probably attributable to the fact that . . [Creole] . . . communication style requires less complex internal organization before an output of language."
For present purposes there are two important features of Creole communication style.
1) Fewer complex sentences and fewer markers of types of embedding.
More conjoining as a means of linking sentences (Craig, 1977, p. 327). Again, there is some similarity between Creole and the features of Black English described by Houston (1969), who noted that in the compositions of some Florida tenth graders, the students used "contatenation as principle [sic] device for creating discourse." At the extreme Creole position, is a type of production whose surface structures resemble the deep structures of English. This can be seen in the following set of sentences.


8
lim tek stuon hit di guot ma him ed, an di guot ded him take stone hit the goat in him head and the goat dead
He killed the goat by hitting it in its head with a stone (Craig, 1976, p. 120). The issue is claimed to be a matter of style rather than meaning, the difference being the point at which meanings are lexified, or put into words. In the first and second sentences above, the ideas are lexif ied as they occur, so to speak, but in the third, there is a reformation of events in the left to right reading. In terms of actions, killing was a consequence of hitting. Note that the first and third sentences have opposite orders of mention of events. The Creole speaker then, has an early lexification format and the Standard English speaker has a late lexification format (Craig, 1974).
Of course, patterns such as the sentences just discussed probably represent modal performance, and corpuses have been corrected in which there is considerable embedding in Creole speech. In fact, in the lexicon, "we" is equivalent to "who" or "that" (Bailey, 1966), and serves the purpose as a marker in relativization. To characterize Creole speech as being composed of conjoined sentences is incorrect, less harshly, it is an oversimplification. However, for convenience in providing a manipulable independent variable it will be supposed that Creole speech can be represented by conjoined sentences.
One difficulty that might arise in comparing Creole and English sentences, is the degree of abstractness. Since, the vocabulary is more concrete, it is expected that in some cases, a Creole sentence will be less economical in terms of the number of words. If the English sentence "He is untrustworthy" is to be adequately represented in Creole, it would be:
im tel lai an [im] tiff
him tell lie and him thief (Craig, 1977)






9
It is clear that the single proposition expressed in English is equivalent to two propositions in Creole. A limitation will be placed on the types of sentences which can be compared.
AN EXPERIMENTAL TRADITION
It should be obvious from the earlier remarks that Caribbean education lacks an input from in situ experimental psychology. However, "empirical' support has been given by sociolinguistics, at least in the context of the present problem. Somewhat fortuitously, a powerful paradigm appears to have some relevance to the problem being developed. A tentative statement of the problem can be made at this point: specifically, the intention is to test the hypothesis of equal facility to represent meanings in both languages. A secondary concern is the accumulation of data towards a characterization of bidialectal semantic memory.
Sentence and idea comprehension have been studied using the acquisition-recognition paradigm of Bransford and Franks (1971). Central to the paradigm is "assimilation theory," which in the case of linguistic stimuli, implies that the meanings derived from sentences are not merely the and-sum of the propositional content of each sentence, but that the meaning is integrated across sentences (Bransford and McCarrell, 1977). In their most relevant statement on semantic integration, Bransford and Franks discuss the possible effects of syntactic complexity on comprehension:
If paragraph comprehension were equivalent to memory for just those sentences comprising it, one could argue for the communicative efficiency of using all syntactically simple sentences. However if Ss spontaneously integrate information from various acquisition sentences, syntactic simplicity may not always correlate with ease of comprehension. If Ss have to integrate information anyway, it should help if syntax does this for them, at least partially. In fact, this is precisely what embedding transformations do. They integrate information that could be expressed in separate sentences.



Further:
Simple syntax may actually hinder comprehension by forcing Ss to do too much of the integration. And more complex structures (like some forms of embedding) may actually facilitate comprehension by explicitly expressing cohesive, easily codable semantic integrations of ideas.
(Bransford & Franks, 1972, pp. 244-245).
In other words, the cognitive equivalence and communication efficiency of early and late lexification formats (Craig, 1974) is challenged.
REVIEW OF RESEARCH
Only one formal report of intervention using the "free talk, controlled talk" method, could be found. A six-month study of seven year old rural Jamaican primary school children. There were five groups, four working class,C10, E10, E20, E30 and one middle class, E40. Ns were not reported. A speech production test, a picture description task, was given at the beginning, after three months and after six months. The four experimental groups were instructed by the method described earlier, and the control group was subject to an unspecified treatment. It had an initial advantage, in that its members had all had infant education, which the experimental groups did not have. Four measures were taken or computed: (1) the average proportion of clauses containing non-standard verb patterns or non-standard relationships between nominative and verb, (2) the average proportion of clauses linked together by "and," (3) the average length of clauses in terms of words, and (4) the average length of children's speech in terms of words. There were no significant changes in the control group on the four measures, when the intermediate and final scores were compared. The "disadvantaged" groups E10, E20, E30 showed the expected decreases on non-standard verb constructions and on the clauses introduced by "and." In addition, there were increases in the average number of





words and in the average length of clauses. Craig attributes the increase in quantity to an "increase in confidence." He states:

It seems likely that these children even at this age, are conscious of the unusual language requirement of the formal situation; when they know instinctively that the required language is available to them, the result is a lessening of inhibitions and a greater flow of language.
(Craig, 1967, p. 139)


Clearly, there is an affective component that must be accounted for, in the production of complex speech, but it is not being addressed here. It would be unfortunate if the belief that coordinations are a permanent feature of Creole language is an artifact of interview procedures. [Houston (1973) claims to have developed techniques of enabling subjects to overcome their apprehensiveness.]
The effectiveness of the method of instruction has been demonstrated, although it was admitted that it was more of a tryout than a formal experiment. The important question is however, what did the children learn? How did the alteration of their production patterns affect their cognitions? Are more semantic relations available to a user of syntactically more complex speech? We will defer answers to these questions; indeed we may not obtain answers for this age group of subjects. Although extrapolation from another age group might not be valid, that might be the best that can be done.
The variation in Creole speech, with respect to the primacy of conjoined versus embedded sentences, is evident in some further research reported by Craig (1974). Upper (U), rural-lower (RL) and urban-lower (L) SES Jamaican children between the ages 6 1/2 to 7 1/2 were compared on a number of speech variables. Spontaneous speech was recorded for approximately 11 minutes, for 63 groups of 5 children each. More simple sentences were produced by L and RL than U, who produced more sentences of more than two clauses (without embedding).


12
L and RL groups produced more complex sentences with at least one embedded clause. Craig claims that "JC, as a morpho-syntactic system, is in some aspects at least, discrete from English and can . . . have different communication conventions relative to the use of clause-embedding" (Craig 1974, p. 15). The embedded sentence is used where a verb-based phrase is used in English. So,
Climbing the tree is easy.
is equivalent to
it iisi wen yu a klaim di trii.
It is easy when you are climbing the tree.
"time" and "place" are expressed using clauses where English uses some format. So
in the rainy season .
is equivalent to
wen (taim) rien fall (plenti) when the rain fall plenty.
That the presence of embeddings is not an either-or situation, does not reduce the addressability of the extent of assimilation of propositions at varying degrees of derivational complexity although some artificiality is introduced. If JC and SJE can be taken to be two languages, then there are three idealized speakers to consider; JC and SJE monolinguals and the bilingual. The best means of identifying these speakers, [although it is practically an impossible task] is to use SES classification (Craig, 1974). It has already been suggested that some indication of the nature of knowledge of a language can be obtained using a semantic integration paradigm.
Bransford and Franks (1971) designed a series of experiments to test the relation between the language of sentences and the memory for ideas explicitly


and implicitly stated by these sentences. The wholistic idea from four simple declarative propositions can be stated in a complex sentence. An example of such a sentence was "the ants in the kitchen ate the sweet jelly which was on the table." The simple propositions were "the ants ate the jelly," "the ants were in the kitchen," "the jelly was sweet," " the jelly was on the table." At an intermediate level of complexity were the sentences "the ants ate the sweet jelly" and "the ants in the kitchen ate the sweet jelly." Sentences used were ONES, TWOS, THREES or FOURS depending on the number of propositions they contained. ONES, TWOS, and THREES were presented in the acquisition phase, while in the recognition phase, the corresponding FOUR and other ONES, TWOS, and THREES were included. It was found that the subjects, who were college students, "recognized" sentences were not presented in the acquisition phase, including the FOUR. In a second experiment NONCASES were presented in the recognition phase. These included information across idea sets or were constructed from an idea set, but had relations altered. NONCASES received the lowest recognition rating.
One point of dispute has been Bransford and Franks (1971) finding that there is linearity in the order of recognition confidence rating, with FOURS> THREES>TWOS>ONES. It was suggested that recognition rating is a function of syntactic complexity, but Katz et al (1974) found that if subjects were instructed to attend to meanings on recall, the effect is not present. Moreover the linearity effect occurs with meaningless material. This issue is not central to the demonstration of integration, which has always been replicated.
Singer and Rosenberg (1973) hypothesised that "the original grammatical functions of the words embedded in a sentence will determine the likelihood that a subject will report recognizing that sentence." Their replication of






14
Bransford and Franks (1971) Experiment II had a minor change; they included more between frame (bf) NONCASES and within frame (wf) NONCASES. In addition to obtaining similar results on integration, it was found that wf NONCASES were recognized to a greater extent than bf NONCASES. Further analysis showed that some ONES had been recognized with more confidence than FOURS. A regression analysis was performed to account for the recognition rates of sentences, given the types of grammatical relations in the idea set, which were expressed in the sentences. The relations were the main relation, subordinate object clause, subordinate subject clause, adjectival subject modification, adjectival object modification. One other variable was a syntactic change factor. The regression model accounted for 83.8% of the variance in recognition, but more important was the fact that a negative weight was obtained for the syntactic change factor. Subjects would be less likely to recognize sentences changed syntactically. The best positive predictor was the presence of the main relation.
In a situation in which it is being attempted to manipulate syntax to alter the extent of integration, the findings of Singer and Rosenberg are important. It becomes difficult to assess whether rejection of an idea is due to the detection of syntactic cues or to a genuine non-integration on a semantic basis. One way to sidestep the issue is consider JC and SJE as two different languages, rather than a single language at different degrees of transformational complexity.
Research on bilingual memory has largely been directed at memory for words. Although there is no consensus, it has been contended that there is a single semantic memory (McCormack, 1977). In attempting to demonstrate the unity of semantic memory, Rosenberg and Simon (1977), utilized the Bransford-Franks (1971) paradigm with French-English bilinguals. They used four idea





15
sets, with English (1), French (1) and French and English (2), sentences. Sentences which had been originally in one-language idea sets were "recognized" least in translated form, but those translated from mixed language idea-sets had recognition rates higher than sentences not previously presented. Rosenberg and Simon interpret these results to mean that subjects have similar, if not identical semantic systems.
It should be possible to obviate the problem of syntactic change as an artificial clue, by presenting mixed sentences composed of conjoined clauses, embedded clauses and both. From these results, some inference regarding the structure of bidialectal semantic memory can be made.
One difficulty, which has been dealt with obliquely so far, is the validity of the results of experiments on semantic integration with adult subjects, for an instructional problem concerning young children. Barclay and Reid (1974) gave three stories to kindergarten, first, third and fifth graders. In the materials, passive sentences were presented as full passives, truncated/ actor or truncated/no actor. Truncated passives were recalled best when no actor was supplied, and were recalled least when there was an appropriate actor in the story. This indicated that a truncated passive with an actor was integrated into either a full passive or the corresponding active sentence. There were no significant differences in the performance of the various groups. So although integration has not been demonstrated using embedded sentences with children, the phenomenon occurs with other transformations.
Differences between beginning 7th grade Creole and Standard English speakers have been obtained on mental ability and language learning aptitude tests (Craig & Carter, 1976); it was suggested that the differences were the effects of differences in the way in which language was used. This study was



correlational, and was therefore limited in the extent to which determinants could be identified.
Comparisons between Creole speaking children and British children have also revealed differences. Edwards (1976) administered a reading comprehension test to 40 English and 40 West Indian students between the ages 11; 3
and 12; 6. Slow, average and good readers were identified and comparisons were made on comprehension between groups of the same reading level. There were no significant differences between the groups of slow readers. For average and good readers however, the English children were significantly better on comprehension on the four more difficult of the six comprehension tests. Edwards attributed the result to "linguistic differences".
In an effort to identify determinants of interference, Edwards (1978) constructed a Creole Interference Test in which structures which were felt to distinguish British English from Creole were put in a multiple choice format. Subtests were vocabulary, syntax (including voice), morphology and phonology. English children scored significantly higher on all subtests. Caribbean-born and British-born West Indian children were significantly different on the tense and aspect, and morphology subtests only. Edwards laments that, "although many of the children sounded very English, it would appear that elements of Creole continue to influence their perception of British English." It was noted the differences between Caribbean West Indian, British-born West Indian and British children are not apparent in the comprehension of written materials suitable for children of reading ages of about 9 or 10.
From what has been a somewhat scattered review, some generalizations can be made. First, there is no evidence of any facilitating effect of Standard English for Creole speakers. Even if the Creole speaker overcomes differences




17
between the phonology of English and his spontaneous phonology, the differences in syntax put the Creole speaker at a disadvantage. Second, the previous studies have not been sufficiently analytical--although Edwards' test did manipulate the passive as a syntactic variable. Third, a semantic integration paradigm can elucidate some performance variables of Creole speakers and English speakers, with transformational complexity as an independent variable.
There is the problem of an appropriate population, that must be resolved. It is proposed that the memory requirements of the Bransford-Franks (1971) paradigm render it unsuitable for the target population of the "free talk, controlled talk" instructional method, so it will be applied to some intermediate age group, say 13 year olds. Although extrapolation of the findings is not justified, some insight of the desired state of bidialectal competence will be obtained. In the transitional period of instruction of Class C and Class D patterns (Craig 1966, 1967, 1976), the Creole speaker will be in the position of assimilating embedded propositions to a cognitive structure composed of predominantly conjoined propositions. It is the nature of the structures formed and the efficacy of their formation that will give a theoretical base to the "free talk--controlled talk" instructional method.

DESIGN
PART ONE
The materials will be administered to the groups in the following manner. In a repeated measures design, the performance of Creole and Standard English speakers will be compared on semantic integration of conjoined propositions and on embedded propositions.
Although the hypothesis of equal facility in the communication style of the language, is intuitively attractive, the
only 'empirical' evidence in



18
in support of this is linguistic analysis. It is therefore proposed that recognition scores on conjoined sentences will not be different for Creole and standard speakers; however the performance of Standard speakers should exceed that of Creoles on embedded propositions.

CONJOINED EMBEDDED
JC CC CE

SJE CC CE

SE>SC,CC,CE is the expected trend.
SUBJECTS
In accordance with previous research (Craig, 1974;Craig & Carter, 1976) JC speakers can reliably be found in non-selective secondary schools. SJE speakers (monolingual) could be found in elite high schools. Essentially, subjects can be identified by social class membership. Sample sizes are yet to be decided.
MATERIALS
The embedded propositions to be used are derived from those used in Bransford and Franks (1971) Experiment 1.
1.The big ants ate the sweet jam which was on the table. (R)
2. The cool breeze blowing from the sea stirred the heavy air. (R)
3.The rock (stone) that rolled down the mountain smashed the little hut on the riverbank. (R)
4.The tall man resting on the counter drank the beer from the bottle. (R)

From sentence 2, the following propositions can be generated.
ONES:
The breeze was cool. (A)
The breeze was blowing from the sea. (A)
The breeze stirred the air. (R)
The air was heavy. (R)
TWOS:
The cool breeze was blowing from the sea. (A)
The breeze blowing from the sea stirred the air. (A,R)
The breeze stirred the heavy air. (R)
The cool breeze stirred the air. (R)
THREES:
The breeze blowing from the sea stirred the heavy air. (A)
The cool breeze blowing from the sea stirred the air. (A)
The cool breeze stirred the heavy air. (R)

It is not necessary to rewrite the ONES for a list of conjoined sentences.

The breeze was blowing from the sea and the breeze was cool. (A)
The breeze was blowing from the sea and the breeze stirred the air. (A,R)
The breeze stirred the air and the air was heavy. (R)
The breeze was cool and the breeze stirred the air. (R)

THREES:
The breeze was blowing from the sea and the breeze stirred the air and the air was heavy. (A)
The breeze was cool and the breeze was blowing from the sea and the breeze stirred the air. (A)
The breeze was cool and the breeze stirred the air and the air was heavy. (R)
FOUR:
The breeze was cool and the breeze was blowing from the sea and the breeze stirred the air and the air was heavy. (R)

For the three remaining idea sets, the propositions will be generated in a similar fashion. The letters A and R indicate which sentences will be presented in acquisition and recognition.
PART TWO
One of the aims of this study is to attempt to characterize bidialectal semantic memory. The paradigm of Rosenberg and Simon (1977) will be adapted for this purpose. The same idea sets that are used in part one, will be used here, with the exception that idea set one, will be in embedded form, idea set four will be in conjoined form, and idea sets two and three will be mixed. Examples of mixed propositions will be shown below.

SUBJECTS
Selection for high school is based on performance on achievement and mental ability tests such as those administered by Craig and Carter (1976). It can be assumed that low SES students who are achieving in high school have attained bidialectal competence.
Creole < Bidialectal < Standard










21
It is proposed that the existence of a single semantic memory for bidia-lectals will be evident if the recognition of propositions for mixed (conjoined and embedded) idea sets exceeds that for standard speakers.

MATERIALS
MIXED THREES.
The breeze was blowing from the sea and the breeze stirred the heavy air.
The cool breeze was blowing from the sea and the breeze stirred the air.
The breeze was cool and the breeze stirred the heavy air.

MIXED FOURS (non-exhaustive)
The cool breeze that was blowing from the sea stirred the air and the air was heavy.
The breeze was cool and the breeze was blowing from the sea and stirred the heavy air.

PROCEDURE
For both parts, acquisition and recognition phases will follow the procedures described by Bransford and Franks (1971). There will be randomization of sentences with the constraint that no two sentences from the same idea set are contiguous, etc. The recognition task will be given without any fore-warning. Elliptical questions will be given after each acquisition sentence is presented, to provide an ostensible task.

FINAL REMARKS
Admittedly, support for the hypotheses propounded in this paper might not indicate how the instructional methodology for teaching Standard English to Creole-speaking children should be changed. Indeed, it might require no practical changes at all. However, at present with its emphasis on "forcing" and "imitation," its present theoretical basis is not cognitivist. It is hoped that this experiment will be a step toward a cognitive approach and toward an experimental tradition.
REFERENCES

Anderson, J.R. & Bower, G. H. Human Associative Memory. New York: Wiley, 1973.
Bailey, B. L. Jamaican Creole Syntax:A Transformational Approach.Cambridge: University Press, 1966.
Barclay, J.R. & Reid, M. Semantic Integration in Children's Recall of Discourse. Developmental Psychology, 1974, 10, (2), 277-281.
Bransford, J.D. & Franks, J.J. The abstraction of linguistic ideas. Cognitive Psychology, 1971, 2, 331-350.
Bransford, J.D. & Franks, J.J. The abstraction of linguistic ideas:A review. Cognition, 1972, 1, 211-249.
Bransford, J.D. &McCarrell,N.S. A sketch of a cognitive approach to comprehension: some thoughtsaboutunderstanding what it means to comprehend. In P.N. Johnson-Laird & P.C. Wason (eds.) Thinking: Readings in Cognitive Science. Cambridge: University Press, 1977.
Burns, R. W. Human Learning. In M.D. Dunnette (ed.) Handbook of Industrial and Organizational Psychology. Chicago: Rand McNally,1976.
Chomsky, N. Aspects of the Theory of Syntax. Cambridge, Mass.: M.I.T. Press, 1965.
Collins, AM., & Quillian, M.R. How to make a language user. In E. Tulving & W. Donaldson (eds.) Organization of Memory. New York: Wiley, 1972.
Craig, D. R. Teaching English to Jamaican Creole speakers. Language Learning, 1966, 1 & 2, 44-61.
Craig, D. R. Some Developments in Language Teaching in the West Indies Caribbean Quarterly, 1966, 12, (1).
Craig, D. R. Some early indications on learning a second dialect, Language Learning, 1967, 3 & 4.
Craig, D. R. Developmental and Social-Class Differences in Language. Caribbean Journal of Education, 1974, 1 (2), 5-23.
Craig, D. R. Bidialectal Education: Creole and Standard in the West Indies. Linguistics, 1976, 175, 93-134.
Craig, D. R. Creole Languages and Primary Education. In A. Valdman (ed.) Pidgin and Creole Linguistics. Bloomington: Indiana University Press,
1977.
Craig, D. R. & Carter, S. The Language Learning Aptitudes of Jamaician Children at the Beginning of Secondary School. Carbbean Journal of Education, Vol. 3, No. 1, 1-21, & Vol. 3, No. 2, 70-112.
Edwards, V. K. Effects of dialect on the comprehension of West Indian Children. Educational Research, 1976, 18, (2), 83-95.
Edwards, V. K. Dialect Interference in West Indian Children, Language & Speech, 1978, 21, (1), 76-86.
Engle, P. L. Language Medium in Early School Years for Minority Language Groups. Review of Educational Research, 1975, 45, (2), 283-325.
Houston, S. H. A Sociolinguistic consideration of Black English of children in Northern Florida. Language, 1969, 45, 599-607.
Houston, S. H. A reexamination of some assumptions about the language of the disadvantaged child. Child Development, 1970, 41, 947-963.
Houston, S. H. Syntactic Complexity and Information Transmission in First-
Graders: A Cross-Cultural Study. Journal of Psycholinguistic Research, 1973, 2, (2), 99-114.
Katz, S. Atkeson, B., & Lee, J. The Bransford-Franks linear effect:
Integration or artifact? Memory & Cognition, 1974, 2 (4), 709-713.
Labov, W. & Cohen, P. Systematic Relations of Standard and Non-Standard
Rules in the Grammar of Negro Speakers. In J.S. De Stefano (ed.)
Language, Society, and Education: A Profile of Black English. Wadsworth, Belmont, California, 1973.
Le Page, R. B. Problems to be faced in the use of English as the medium of Education in four West Indian territories. In J. A. Fishman, C. A. Ferguson, J. Das Gupta (eds.) Language Problems of Developing Nations. New York: Wiley, 1968.
McCormack, P. D. Bilingual Linguistic Memory: The Independence-Interdependence Issue Revisited. In P. A. Hornby (ed.) Bilingualism, Psychological, Social, and Educational Implications. New York: Academic Press, 1977.
Richards, J. P. Instructional Psychology: From a Behavioristic to a Cognitive Orientation. Improving Human Performance: A Research Quarterly, 1978, 7, 4, 256-266.
Rosenberg, S. & Simon, H. A. Modelling semantic memory: Effects of presenting semantic information in different modalities. Cognitive Psychology, 1977, 9, 293-325.
Singer, M., & Rosenberg, S. T. The role of grammatical relations in the abstraction of linguistic ideas. Journal of Verbal Learning and Verbal Behavior, 1973, 12, 273-284.
Weimer, W. Psycholinguistics and Plato's Paradoxes of the Meno. American Psychologist 1973, 28, (1), 15-33.