faŕijétč J/ fáčf*9 & & ŕ*aditions (New York, 1941), 1, pp. 346, 391, the Hebrews took over the Semitic system in the eleventh century B.C., and the Indians a good deal later, probably in the eighth century B.C. 36. Gandz, "Oral Tradition in the Bible," pp. 253^1. 37. e.g. Luke, 20; Matthew, 23; in the 7th century B.C., even kings and prophets employed scribes, Jer. xxxvi, 4, 18. 38. Driver, Semitic Writing, pp. 87-90, where lie instances the case of one scribe who having no son "taught his wisdom to his sister's son." 39. "If the alphabet is defined as a system of signs expressing single sounds of speech, then the first alphabet which can justifiably be so called is the Greek alphabet." Gelb, Study of Writing, p. 166. 40. 1. Kings t7, iv-vi; see A Dictionary of the Bible... ed. James Hastings (New York, 1898-1904), s.v. "Elijah." 41. 810 a. From the ages 10 to 13. 42. ĽAdoption universelle des caractcres latins (Paris, 1934); for more recent developments and documentation, see William S. Gray, The Teaching of Reading and Writing: An International Survey, Unesco Monographs on Fundamental Education X (Paris, 1956), especially pp. 31-60. 43. Chester G. Starr, The Origins of Greek CYri'lii» 5 tion (New York, 1961), pp. 1 89-1 90, 349 ff. 44. Starr, The Origins of Greek Civilization, pp. 87-88,357. 45. Starr, The Origins of Greek Civilization, p.; 169. ; 46. L. H. Jeffery, The Local Scripts of Arcluttl Greece (Oxford, 1961), p. 21; R. M. Cook and A.:' G. Woodhead, "The Diffusion of the Greek Alph»! bet," American Journal of Archaeology, *>| (1959), pp. 175-78. For North Syria, se« S» s Leonard Woolley, A Forgotten Kingdom (LonJi«, ľ 195 3). 47. Chester Starr speaks of its use by "a relativelf large aristocratic class" (p. 171) and Miss Jefftrf notes that "writing was never regarded as an e» teric craft in early Greece. Ordinary people coull and did learn to write, for many of the earltftl inscriptions which we possess are casual graffiti* (p. 63). 48. Frederic G. Kenyon, Books and Reaiicn M| Ancient Greece and Rome (2nd ed., Oxforl*^ 1951),p.67. 49. Jérôme Carcopino, ĽOstracisme athéniit (Paris, 1935), pp. 72-110. 50. Protagoras, 325 d. 51. 1. I 114; in 414 B.C. See also Plato, Apoloff,\ 26 d, and the general survey of Kenyon, Booksgnt Readers in Ancient Greece and Rome. 52. The Philosophy of Symbolic Forms (Nc* Haven, 1955), II, p. xiii; and An Essay on Ali» (New York, 1953), especially pp. 106-1 30, 2BI-JU For Werner Jaeger, see especially 77ie Theology tf Tiie Early Greek Philosophers (Oxford, 1947). 53. "Magic, Science and Religion" in Science Religion and Reality, ed. Joseph Needham (Nc* York, 1925), reprinted Magic, Science and Religio* (New York, 1954), p. 27. For an appreciation d Lévy-Bruhľs positive achievement, see F.vinfr: Pritchard, "Lévy-Bruhľs Theory of Primitive Mr» tality," Bulletin of the Faculty of Arts, Univemtf of Egypt, 2 (1934), pp. 1-36. In his later woriv LéVy-Bruhl modified the rigidity of his CBrUt dichotomy. 54. Witchcraft, Oracles and Magic Among Af Azande (Oxford, 1937). See also Max GluckmtiV; essay, "Social Beliefs and Individual Thinkini ft! Primitive Society," Memoirs and Proceedings tf\ the Manchester Literary and Philosophical Society, 91 (1949-50), pp. 73-98. From a rather diffřrrtl standpoint, Lévi-Strauss has analysed "the logk tf totemic classifications" (La Pensée sauvage, p. 41* ff.) and speaks of two distinct modes of scientist thought; the first (or "primitive") variety conii» in "the science of the concrete," the practice knowledge of the handy man (bricoleur), whick d< the technical counterpart of mythical thought (Rj 26). 55. e.g. the Trobriands (Malinowski, Myth k Primitive Psychology, pp. 33ff). ± 56. Jeffery, The Local Scripts of Archaic Grtec^i P-46. 57. "It was in Ionia that the first completely ntt».] nalistic attempts to describe the nature of cht4 world took place" |G. S. Kirk and J. li. Raven,7*r Presocratic Philosophers (Cambridge, 1957), |C 73). The work of the Milesian philosopher' SOCIAL CLASS, LANGUAGE AND SOCIALISATION 473 Thales, Anaximander and Anaximenes, is described t) the authors as "clearly a development of the itnetic or genealogical approach to nature exempli-fifd by the Hesiodic Theogony" (p. 73). 58. F. Jacoby, Die Fragmente der Griechischen Historiker, Vol. I, Genealogie und Mythographie (Btrlin, 1923), fr. l.a. 19. Reflections on Violence, trans. T. E. Hulme (New York, 1941), p. 136; cit. Robert Redfield, Die Primitive World and its Transformations (llhaca, New York, 1953), p. 125. 40. clt. Joseph Hone, W. B. Yeats (London, 1942), f.405 (our italics). tl. Hermann Diels, Die Fragmente der Vor-Ubatiker (Berlin, 195 1), fr. 11, 23; see also John kirnet, Early Greek Philosophy (2nd ed. London, H18), pp. 131, 140-141, and Werner Jaeger, 77ie? toeotogy of the Early Greek Philosophers (Oxford, ; 1947), pp. 42-7; Kirk and Raven, The Presocratic hilosophers, pp. 163 ff. •2. Diels, Fragmente der Vorsokratiker, fr. 40, 42, M, S7, 106; see also Francis M. Cornford, Prin-rgjiiiHi Sapientiae: The Origins of Greek Philosoph-tolThought (Cambridge, 1952), pp. 112 ff.; Kirk •ml Raven, The Presocratic Philosophers, pp. 182 í »J. Francis M. Cornford, Greek Religious Thought }pom Homer to the Age of Alexander (London, 1*23), xv-xvi. See also Burnet, Early Greek hiilmophy, p. I. H. 1st Olympian Ode. US. See Eric II. Warniington, Greek Geography (London, 1934), pp. xiv, xxxviii. 1 M. History, 4, 36^10. : »7. Warmington, Greek Geography, pp. xvii—xviii, \ li ff. to. Cit. Lionel Pearson, Early Ionian Historians "(Oxford, 1939), p. 3. íM. Felix Jacoby, Atthis (Oxford, 1949), p. 354. ■ 'Ú. History, 1, 1. See also Moses I. Finley (ed.), ne Greek Historians (New York, 1959), pp. 4 ff. 71. See Pearson, Early Ionian Historians, pp. 152-233, especially pp. 193, 232-33. 72. See, for instance, Bronislaw Malinowski, Argonauts of the Western Pacific (London, 1922), nru 290-333. 73. Thucydides, History, 1, 20-22, 97. For a picture of note-taking (hypomnemata) among Athenians, see Theaetetus, 142 c-143 c. 74. Felix Jacoby notes that "fixation in writing, once achieved, primarily had a preserving effect upon the oral tradition, because it put an end to the involuntary shiftings of the mncmai (remembrances), and drew limits to the arbitrary creation of new logoi (stories)" (Atthis, 1949, p. 217). He points out that this created difficulties for the early literate recorders of the past which the previous oral mnemones or professional "remembrancers" did not have to face: whatever his own personal view of the matter, "no true Atthidogra-pher could remove Kekrops from his position as the first Attic king . . . Nobody could take away from Solon the legislation which founded in mice the first Attic constitution of historical times." Such things could no longer be silently forgotten, as in an oral tradition. The general conclusion of Jacoby's polemic against Wilamowitz's hypothesis of a "pre-litcrary chronicle" is that "historical consciousness ... is not older than historical literature" (p. 201). 75. As writers on the indigenous political systems of Africa have insisted, changes generally take the form of rebellion rather than revolution; subjects reject the King, but not the kingship. See Fivans-Pritchard, The Divine Kingship of the Shilluk of the Nilotic Sudan (The Frazer lecture, Cambridge, 1948), pp. 35ff; Max Gluckman, Rituals of Rebellion in South-East Africa (The Frazer lecture, 1952), Manchester, 1954. 28. Social Class, Language and Socialisation BASIL BERNSTEIN ÍNTR0DUCTION1 j h may be helpful to make explicit the theo-(«ttical origins of the thesis I have been devel-ieping over the past decade. Although, ; Initially, the thesis appeared to be concerned j»iih the problem of educability, this prob-ifcm was imbedded in and was stimulated by Ihc wider question of the relationships ieiween symbolic orders and social struc- ture. The basic theoretical question, which dictated the approach to the initially narrow but important empirical problem, was concerned with the fundamental structure and changes in the structure of cultural transmission. Indeed, any detailed examination of what superficially may seem to be a string of somewhat repetitive papers, I think would show three things: from Current Trends in Linguistics, Volume 12, ed. A. S. Abramson et al. (Mouton, 1973). Reprinted by •ttrmission. i 474 1. The gradual emergence of the dominance of the major theoretical problem from the local, empirical problem of the social antecedents of the educability of different groups of children. 2. Attempts to develop both the generality of the thesis and to develop increasing specificity at the contextual level. 3. Ľntailed in (2) were attempts to clarify both the logical and empirical status of the basic organising concept, code. Unfortunately, until recently these attempts were more readily seen in the planning and analysis of the empirical research than available as formal statements. Looking back with hindsight, 1 think I would have created less misunderstanding if I had written about sociolinguistic codes rather than linguistic codes. Through using only the latter concept it gave the impression that I was reifying syntax and at the cost of semantics. Or worse, suggesting that there was a one to one relation between meaning and a given syntax. Also, by defining the codes in a context free fashion, I robbed myself of properly understanding, at a theoretical level, their significance. / should point out that nearly all the empirical planning was directed to trying to find out the code realisations in different contexts. The concept of sociolinguistic code points to the social structuring of meanings and to their diverse but related contextual linguistic realisations. A careful reading of the papers always shows the emphasis given to the form of the social relationship, that is, the structuring of relevant meanings. Indeed, role is defined as a complex coding activity controlling the creation and organisation of specific meanings and the conditions for their transmission and reception. The general sociolinguistic diesis attempts to explore how symbolic systems are both realisations and regulators of the structure of social relationships. The particular symbolic system is that of speech, not language. It is pertinent, at this point, to make explicit earlier work in the social sciences which formed the implicit starting point of BERNSTEIN the thesis. It will then be seen, I hope, thai the thesis is an integration of different streams of thought. The major starting points are Durkheim and Marx, and a small number of other thinkers have been drawn into the basic matrix. I shall very briefly, and so selectively, outline this matrix and some of the problems to which it gave rise. Durkheiin's work is a truly magnificent insight into the relationships between symbolic orders, social relationships and tilt structuring of experience. In a sense, if Mara turned Hegel on his head, then Durkheim attempted to turn Kant on his head. For in Primitive classiflcation and in The elementary forms of the religious life, Durkheim attempted to derive the basic categories of thought from the structuring of the social relation. It is beside the point as to hii success. He raised the whole question of the relation between the classifications and frames of the symbolic order and the struc-hiring of experience. In his study of differ, jfl1 cnt forms of social integration he pointed to the implicit, condensed, symbolic structure of mechanical solidarity and the more explicit and differentiated symbolic structure! of organic solidarity. Cassirer, the early cultural anthropologists, and in particular Sapir (I was not aware of von Humboldt until much later) sensitised me to the cultural properties of speech. Whorf, particularly where he refers to the fashions of speaking, frames of consistency, alerted me to the selective effect of llie culture (acting through its patterning of social relationships) upon llie patterning of grammar together with the pattern's semantic and thus cognitive significance. Whorf more than anyone, I think, opened up, at least fot me, the question of the deep structure of linguistically regulated communication. In all the above work I found two difficulties. If we grant the fundamental linkage of symbolic systems, social structure and the shaping of experience, it is still unclear how such shaping takes place. The processes underlying the social structuring of experience are not explicit. The second difficulty SOCIAL CLASS, LANGUAGE AND SOCIALISATION is in dealing with the question of change of symbolic systems. George Herbert Mead is of central importance in the solution of the first difficulty, the HOW. Mead outlined in general terms the relationships between role, icflexiveness and speech, and in so doing provided the basis of the solution to the HOW. It is still llie case that the Meadian solution does not allow us to deal with the problem of change. For the concept, which enables role to be related to a higher order concept, "the generalised other" is, itself, not subject to systematic enquiry. Even if 'the generalised other" is placed within a Durkheimian framework, we are still left with the problem of change. Indeed, in Mead change is introduced only at the cost of the re-emergence of a traditional Western dichotomy in the concepts of the "I" and the "me." The "I" is both the indeterminate icsponsc to the "me" and yet at the same time shapes it. The Meadian "I" points to the voluntarism in the affairs of men, the fundamental creativity of man, made possible by speech; a little before Chomsky. Thus Meadian thought helps to solve the puzzle of the HOW but it does not help with the question of change in the structuring of experience; although both Mead implicitly and Durkheim explicitly pointed to the conditions which bring about pathological structuring of experience. One major theory of the development of and change in symbolic structures is, of course, that of Marx. Although Marx is less concerned with the internal structure and the process of transmission of symbolic systems he does give us a key to their institu-lionalisation and change. The key is given in terms of the social significance of society's productive system and the power relation-" ships to which the productive system gives rise. Further, access to, control over, orientation of and change in critical symbolic systems, according to the theory, are governed by these power relationships as these arc embodied in the class structure. It is not only capital, in the strict economic sense, which is subject to appropriation, manipula- 475 tion and exploitation, but also cultural capital in the form of the symbolic sys'.ems through which man can extend and change the boundaries of his experience. I am not putting forward a matrix of thought necessary for the study of the basic structure and change in the structure of cultural transmission, only the specific matrix which underlies my own approach. Essentially and briefly I have used Durkheim and Marx at the macro level and Mead at the micro level, to realise a sociolinguistic thesis which could meet with a range of work in anthropology, linguistics, sociology and psychology. OTHER VIEWS OF THE RELATION OF LINGUISTIC AND CULTURAL SYSTEMS I want also to make clear two views I am not concerned with. Chomsky in Aspects of the theory of syntax neatly severs the study of the rule system of language from the study of the social rules which determine their contextual use. He does this by making a distinction between competence and performance. Competence refers to the child's tacit understanding of the rule system, performance relates to the essentially social use to which the rule system is put. Competence refers to man abstracted from contextual constraints. Performance refers to man in the grip of the contextual constraints which determine his speech acts. Competence refers to the Ideal, performance refers to the Fall. In this sense Chomsky's notion of competence is Platonic. Competence has its source in the very biology of man. There is no difference between men in terms of their access to the linguistic rule system. Here Chomsky, like many other linguists before him, announces the communality of man, all men have equal access to the creative act which is language. On the other hand, performance is under the control of the social—performances are culturally specific acts, they refer to the choices which arc made in specific speech encounters. Thus from one point of view, Chomsky indicates 476 the tragedy of man, the potentiality of competence and the degeneration of performance (this view explicitly derives from Ilymes 1966). Clearly, much is to be gained in rigour and explanatory power through the severing of the relationship between the formal properties of the grammar and the meanings which are realised in its use. But if we are to study speech, la parole, we are inevitably involved in a study of a rather different rule system, we are involved in a study of rules, formal and informal, which regulate the options we take up in various contexts in which we find ourselves. This second rule system is the cultural system. This raises immediately the question of the causal relationship between the linguistic rule system and the cultural system. Clearly, specific linguistic rule systems are part of the cultural system, but it has been argued that the linguistic rule system in various ways shapes the cultural system. This very briefly is the view of those who hold a narrow form of the linguistic relativity hypothesis. I do not intend to get involved in that particular quagmire. Instead, I shall take the view that the code which the linguist invents to explain the formal properties of the grammar is capable of generating any number of speech codes, and there is no reason for believing that any one language code is better than another in this respect. On this argument, language is a set of rules to which all speech codes must comply, but which speech codes are realised is a function of the culture acting through social relationship in specific contexts.'Different speech forms or codes symbolize the form of the social relationship, regulate the nature of the speech encounters, and create for the speakers different orders of relevance and relation. The experience of the speakers is then transformed by what is made significant or relevant by the speech form. This is a sociological argument because the speech form is taken as a consequence of the form of the social relation or, put more generally, as a quality of a social structure. BERNSTEIN Let me qualify this immediately. Because the speech form is initially a function of i given social arrangement, it does not mean that the speech form does not in turn modify or even change that social structure which initially evolved the speech form.Thu formulation, indeed, invites the question: under what conditions does a given speech form free itself sufficiently from its embodiment in the social structure so that the system of meanings it realises points to alternative realities, alternative arrangements ia the affairs of men? Here we become concerned immediately with the anteccdenti and consequences of the boundary maintaining principles of a culture or sub-culture. I am here suggesting a relationship between forms of boundary maintenance at the cultural level and forms of speech. LANGUAGE, SOCIALISATION AND CLASS I am required to consider the relationship between language and socialisation. It should be clear that I am not concerned with language, but with speech, and concerned more specifically with the contextual constraint! upon speech. Now what about socialisation? I shall take the term to refer to the procea whereby a child acquires a specific cultural identity, and to his responses to such an identity. Socialisation refers to the procea whereby the biological is transformed into I specific cultural being. It follows from this that the process of socialisation is a complex' process of control, whereby a particular moral, cognitive and affective awareness h evoked in the child and given a specific form and content. Socialisation sensitizes the child to various orderings of society as these; are made substantive in the various roles he is expected to play. In a sense then socialisation is a process for making people safe. The process acts selectively on the possibilities of man by creating through time a sense of the inevitability of a given social arrangement,: and through limiting the areas of permitted change. The basic agencies of socialisation ia SOCIAL CLASS, LANGUAGE AND SOCIALISATION contemporary societies are the family, the peer group, school and work. It is through these agencies, and in particular through their relationship to each other, that the : »arious orderings of society are made mani-fest. Now it is quite clear that given this view of socialisation it is necessary to limit the Jiscussion. 1 shall limit our discussion to ixialisation within the family, but it should be obvious that the focussing and filtering of the child's experience within the family in a large measure is a microcosm of the macro- Kopic orderings of society. Our question | aow becomes: what are the sociological fac-j tors which affect linguistic performances! within the family critical to the process of j wcialisation? ^' Without a shadow of doubt the most fbmiative influence upon the procedures of udalisation, from a sociological viewpoint, s, social class. The class structure influences work and educational roles and brings fami-kes into a special relationship with each other and deeply penetrates the structure of lie experiences within the family. The class lystem has deeply marked the distribution of knowledge within society. It has given differential access to the sense that the world is permeable. It has sealed off commu-tities from each other and has ranked these communities on a scale of individuous worth. We have three components, knowl-iJge, possibility, invidious insulation. It would be a little naive to believe that differences in knowledge, differences in the sense of ihc possible, combined with invidious nsulalion, rooted in differential material well-being would not affect the forms of control and innovation in the socialising procedures of different social classes. I shall go on to argue that the deep structure of communication itself is affected, but not in my final or irrevocable way. As an approach to my argument, let me jlince at the social distribution of knowl-tjge. We can see that the class system has liTcctcd the distribution of knowledge. Historically and now, only a tiny percentage ■.I . 477 of the population has been socialised into knowledge at the level of the meta-languages of control and innovation, whereas the mass of the population has been socialised into knowledge at the level of context-tied operations. A tiny percentage of the population has been given access to the principles of intellectual change whereas the rest have been denied such access. This suggests that we might be able to distinguish between two orders of meaningJOne we could call univer-salistic, the other particularistic. Universal-istic meanings are those in which principles and operations are made linguistically explicit whereas particularistic orders of meaning are meanings in which principles and operations arc relatively linguistically implicit. If orders of meaning are universal-istic, then the meanings are less tied to a given context. The meta-languages of public forms of thought as these apply to objects and persons realise meanings of a univcrsal-istic type. Where meanings have this characteristic then individuals may have access to the grounds of their experience and can change the grounds. Where orders of meaning are particularistic, where principles arc linguistically implicit, then such meanings are less context independent and more context bound; that is, tied to a local relationship and to a local social structure. Where the meaning system is particularistic, much of the meaning is imbedded in the context and may be restricted to those who share a similar contextual history. Where meanings are universalistic, they arc in principle available to all because the principles and operations have been made explicit and so public. |_sliall argue that forms of socialisation' orient the child towards speech codes which control access to relatively context-tied ftr relatively context-independent meanings. Thus I shall argue that elaborated codes orient their users towards universalistic meanings, whereas restricted codes orient, sensitize, their users to particularistic meanings: that the linguistic realisations of the two orders are different, and so are the social* Ä k 27 478 BERNSTtIN SOCIAL CLASS, LANGUAGE AND SOCIALISATION 479 relationships which realise them. Elaborated codes arc less tied to a given or local structure and thus contain the potentiality of change in principles. In the case of elaborated codes the speech is freed from its evoking social structure and takes on an autonomy. A university is a place organised around talk. Restricted codes are more tied to a local social structure and have a reduced potential for change in principles. Where codes are elaborated, the socialised has more access to the grounds of his own socialisation, and so can enter into a reflexive relationship to the social order he has taken over. Where codes are restricted, the socialised has less access to the grounds of his socialisation, and thus rcflexiveness may be limited in range. One of the effects of the class system is to limiiHcěess Wěláboratcd codes. I shall go on to suggest that restricted codes have their basis in condensed symbols whereas elaborated codes have their basis in articulated symbols. That restricted codes draw upon metaphor whereas elaborated codes draw upon rationality. That these codes constrain the contextual use of language in critical socialising contexts and in this way regulate the orders of relevance and relation which the socialised takes over. From this point of view, change in habitual speech codes involves changes in the means by which object and person relationships are realised. ELABORATED AND RESTRICTED SPEECH VARIANTS I want first to start with the notions of elaborated and restricted speech variants. A variant can be considered as the contextual constraints upon grammatical-lexical choices. Sapir, Malinowski, Firth, Vygotsky, Luria have all pointed out from different points of view that the closer the identification of speakers, the greater the range of shared interests, the more probable that the speech will take a specific form. The range of syn- tactic alternatives is likely to be reduced and the lexis to be drawn from a narrow range, j Thus, the form of these social relations U acting selectively on the meanings to be ver» bally realised. In these relationships tht| intent of the other person can be taken fixjf granted as the speech is played out against r, back-drop of common assumptions, con>| moil history, common interests. As a result, there is less need to raise meanings to the" level of explicitness or elaboration. There is a reduced need to make explicit througk syntactic choices the logical structure of the« communication. Further, if the spcakwi wishes to individualise his communication,-; he is likely to do this by varying the cxpirt» sive associates of the speech. Under the« conditions, the speech is likely to have | strong metaphoric element. In thcsesituatiuMj the speaker may be more concerned with < how something is said, when it is said;^ silence takes on a variety of meanings. Oľien in these encounters the speech cannot bííl understood apart from the contexl, and the context cannot be read by those who do no» share the history of the relationship. Tliui the form of the social relationship acts seleol tively on the meanings to be verbalised,? which in turn affect the syntactic and lexical choices. The unspoken assumptions under»;, lying the relationship are not available W'š those who arc outside the relationship, ľotj these are limited, and restricted to thti speakers. The symbolic form of the cont^ munication is condensed yet the specific! cultural history of the relationship is alive É its form. We can say that the roles of th»f speakers are communaliscd roles. Thus, w»^ can make a relationship between restrictti» social relationships based upon coninmiut ised roles and the verbal realisation of thetf meaning. In the language of the earlier pirt of this paper, restricted social relationship! based upon communaliscd role evoke pjilto ularistic, that is, context-tied meanings, kú-ised through a restricted speech variant. Imagine a husband and wife have ju.ŕ' !-WJ i.lípnu] «mm i.**iim.mmwmuiMMwwMiwjQWKi 480 dien do not have in their passive vocabulary the vocabulary used by the middle-class children. Nor is it the case that the children differ in their tacit understanding of the linguistic rule system. Rather, what we have here are differences in the use of language arising out of a specific context. One child makes explicit the meanings which he is realising through language for the person he is telling the story to, whereas the second child does not to the same extent. The first child takes very little for granted, whereas the second child takes a great deal for granted. Thus for the first child the task was seen as a context in which his meanings were required to be made explicit, whereas the task for the second child was not seen as a task which required such explication of meaning. It would not be difficult to imagine a context where the first child would produce speech rather like the second. What we are dealing with here are differences between the children in the way they realise in language use apparently the same context. We could say that the speech of the first child generated universalistic meanings in the sense that the meanings are freed from the context and so understandable by all. Wliereas the speech of the second child generated particularistic meanings, in the sense that the meanings are closely tied to the context and would be only fully understood by others if they had access to the context which originally generated the speech. It is again important to stress that the second child has access to a more differentiated noun phrase, but there is a restriction on its use. Geoffrey Turner, Linguist in the Sociological Research Unit, shows that working-class, five-year-old children in the same contexts examined by Hawkins, use fewer linguistic expressions of uncertainty when compared with the middle-class children. This does not mean that working-class children do not have access to such expressions, but that the eliciting speech context did not provoke them. Telling a story from pictures, talking about scenes on cards, for- SOCIAL CLASS, LANGUAGE AND SOCIALISATION 481 BERNSTEIN malty framed contexts may not encourage working-class children to consider the possibilities of alternate meanings and so there is a reduction in the linguistic expres- ' sions of uncertainty. Again, working-class children have access to a wide range of . syntactic choices which involve the use of .. logical operators, "because," "but," "either," "or," "only." The constraints exist : on the conditions for their use. Formally \ framed contexts used for eliciting context independent universalistic meanings may evoke in the working-class child, relative to ; the middle-class child, restricted speech variants, because the working-class child hat ; difficulty in managing the role relationships which such contexts require. This problem ii ' further complicated when such contexts ; carry meanings very much removed from the \ child's cultural experience. In the same way i we can show that there are constraints upon S the middle-class child's use of language, > Turner found that when middle-class diu- ; dren were asked to role play in the picture \ story series, a higher percentage of these » children, when compared with working-class s children, initially refused. When the middle- X class children were asked "What is the man í saying?" or linguistically equivalent questions, a relatively higher percentage said "I don't know." When this question was fol- \ lowed by the hypothetical question "What \ do you think the man might be saying?" ;; they offered their interpretations. The q working-class children role played without ; difficulty. It seems then that middle-dan 1 children at five need to have a very prcci$e| instruction to hypothesise in that particular context. This may be because they are m ort | concerned here with getting their answeni right or correct. When the children wert i invited to tell a story about some doll-like si figures (a little boy, a little girl, a sailor and i ,|| dog), the working-class children's storiaj were freer, longer, more imaginative than lh»| stories of the middle-class children. Th*| latter children's stories were tighter, coo-j strained within a strong narrative frame. Il 3 was as if these children were dominated by what they took to be the form of a narrative and the content was secondary. This is an example of the concern of the middle-class ciiild with the structure of the contextual frame. It may be worthwhile to amplify this further. A number of studies have shown that when working-class black children are asked to associate to a series of words, their responses show considerable diversity, both from the meaning and form-class of the stimulus word. In the analysis offered in the text this may be because the children for the following reasons are less constrained. The form-class of the stimulus' word may have reduced associative significance and so would less constrain the selection of potential words or phrases. With such a weakening of the grammatical frame a greater range of alternatives are possible candidates for selection. Further, the closely controlled middle-class linguistic socialisation of the young child may point the child towards both the grammatical significance of the stimulus word and towards a tight logical ordering of semantic space. Middle-class children may »ell have access to deep interpretive rules which regulate their linguistic responses in certain formalised contexts. The consequences may limit their imagination through die tightness of the frame wliich these interpretive rules create. It may even be that with /h'e-year-old children, the middle-class child will innovate more with the arrangements of objects (i.e. bricks) than in his linguistic usage. His linguistic usage is under close supervision by adults, lie has more autonomy in his play. To return to our previous discussion, we can say briefly that as we move from com- munalised to individualised roles, so speech takes on an increasingly reflexive function. The unique selves of others become palpable • through speech and enter into our own self, í die grounds of our experience are made .-verbally explicit; the security of the con- , densed symbol is gone. It has been replaced by rationality. There is a change in the basis of our vulnerability. FOUR CONTEXTS So far, then, I have discussed certain types of speech variants and the role relationships which occasion them. 1 am now going to raise the generality of the discussion and Tocus upon the title of the paper. The socialisation of the young in the family proceeds within a critical set of inter-related contexts. Analytically, we may distinguish four contexts. 1. The regulative context—these are authority relationships where the child is made aware of the rules of the moral order and their various backings. 2. The instructional context, where the child learns about the objective nature of objects and persons, and acquires skills of various kinds. 3. The imaginative or innovating contexts, where the child is encouraged to experiment and re-create his world on his own terms, and in his own way. 4. The interpersonal context, where the child is made aware of affective states—his own, and others. ~ *" 1 am suggesting that the critical orderings of a culture or subculture are made substantive—are made palpable—through the forms of its linguistic realisations of these four contexts—initially in the family and kin. Now if the linguistic realisation of these four contexts involves the predominant use of restricted speech variants, I shall postulate that the deep structure of the communication is a restricted code having its basis in communalised roles, realising context bound meanings, i.e., particularistic meaning orders. Clearly the specific grammatical and lexical choices will vary from one context to another. If the linguistic realisation of these four contexts involves the predominant usage of elaborated speech variants, I shall postulate that the deep structure of the communica- 484 BERNSTEIN are made substantive in the communication structure. Wc will call these person-centred families. Such families do not reduce but increase the substantive expression of ambiguity and ambivalence. In person-centred families, the role system would be continuously evoking, accommodating and assimilating the different interests, attributes of its members. In such families, unlike positional families, the members would be making their roles, rather than stepping into them. In a person-centred family, the child's developing self is differentiated by continuous adjustment to the verbally realised and elaborated intentions, qualifications and motives of others. The boundary between self and other is blurred. In positional families, the child takes over and responds to the formal pattern of obligation and privilege. It should be possible to sec, without going into details, that the communication structure within these two types of family are somewhat differently focussed. We might then expect that the refiexiveness induced by positional families is sensitized to the general attributes of persons, whereas the refiexiveness produced by person-centred families is more sensitive towards the particular aspects of persons. Think of the difference between Darlington Hall or Gordonstoun Public Schools in England, or the difference between West Point and a progressive school in the USA. Thus, in person-centred families, the insides of the members are made public through the communication structure, and thus more of the person has been invaded and subject to control. Speech in such families is a major media of control. In positional families of course, speech is relevant but it symbolizes the boundaries given by the formal structure of the relationships. So far as the child is concerned, in positional families he attains a strong sense of social identity at the cost of autonomy; in person-centred families, the child attains a strong sense of autonomy but his social identity may be weak. Such ambiguity in the sense of identity, the lack of boundary, may move such children towards a radically closed value system. If we now place these family types in the framework of the previous discussion, we can sec that although the code may be elaborated, it may be differently focussed according to the family type. Thus, wc can have an elaborate code focussing upon personsoran elaborated code in a positional family may focus more upon objects. We can expect the same with a restricted code. Normally, with code restriction we should expect a positional family, however, if it showed signs of being person-centred, then we might expect the children to be in a situation of potential code switch. Where the code is elaborated, and focussed by a person-centred family, then these children may well develop acute identity problems, concerned with authenticity, of limiting responsibility—they may come to see language as phony, a system of counterfeit masking the absence of belief. They may move towards tlie restricted codes of the various peer group sub-cultures, or seek the condensed symbols of affective experience, or both. One of the difficulties of this approach it to avoid implicit value judgements about the relative worth of speech systems and the cultures which they symbolize. Let it_bc said_ immediately that a restricted _ code gives access to a vast potential of meanings, of delicacy, subtlety and diversity of cultural forms, to a unique aesthetic whose basis in condensed symbols may influence the form of the imagining.. Yet, in complex industrial, ized societies, its differently focussed experience "may' be disvalued, and humiliated within schools or seen, at best, to be iirejf | vant to the educational endeavour. Forjhe schools are predicated upon elaborated code and its system of social relationships. Although an elaborated code does not entail any specific value system, the value system of the middle class penetrates the texture of the very learning context itself. Ľlaborated codes give access to al tenia- SOCIAL CLASS, LANGUAGE AND SOCIALISATION 485 marked shift towards person type control. Whereas in societies with strong constraints upon legitimising values, where there is a severe restriction upon the choice, we might expect a marked shift towards positional control. . „ I shall illustrate these relationships with reference to the family: Division of labour Constraints upon Icxiliinisinx value (lioiiiHiiiry Maintenance) live realities yet they carry the potential of alienation, of feeling from thought, of self from other, of private belief from role obligation. SOURCES OF CHANGE Finally 1 should like to consider briefly the source of change of linguistic codes. The first major source of change 1 suggest is to be located in the division of labour. As the division of labour changes from simple to complex, then this changes the social and knowledge characteristics of occupational roles. In this process there is an extension of access, through education, to elaborated codes, but access is controlled by the class system. The focussing of the codes I have suggested is brought about by the boundary maintaining procedures within the family. However, we can generalise and say that the focussing of the codes is related to the boundary maintaining procedures as these affect the major socialising agencies, family, age group, education and work. We need, therefore, to consider together with the question of the degree and type of complexity of the division of labour the value orien-lations of society which it is hypothesized affect the boundary maintaining procedures. It is the case that we can have societies with a similar complexity in their division of labour but which differ in their boundary maintaining procedures. I suggest then that it is important to make a distinction between societies in terms of their boundary maintaining procedures if we are to deal with this question of the focussing of codes. One possible way of examining the relative strength of boundary maintenance, at a somewhat high level of abstraction, is to consider the strength of the constraints upon the choice of values which legitimize authority/power relationships. Thus in societies where there is weak constraint upon such legitimising values, that is, where there are a variety of formally permitted legitimising values, we might expect a Simple->Complex Strong Weak I I i Speech Codes Positional Personal Restricted Code Working-Class Working-Class I Elaborated Code Middle-Class Middle-Class Thus the division of labour influences the availability of elaborated codes; the class system affects their distribution; the focussing of codes can be related to the boundary maintaining procedures, i.e. the value system. I must point out that this is only a coarse interpretive framework. CONCLUSION I have tried to show how the class system acts upon the deep structure of communication in the process of socialisation. I refined the crudity of this analysis by showing how speech codes may be differently focussed through family types. Finally, it is conceivable that there are general aspects of the analysis which might provide a starting point for the consideration of symbolic orders other than languages (see Douglas 1970). I must point out that there is more to socialisation than the forms of its linguistic realisation. NOTE 1. This work was supported by grants from the Department of Education and Science, the ľord Foundation and the Nuffield Foundation, to whom grateful acknowledgement is made. 1 would also like lo take the opportunity of acknowledging my debt to Professor Courtney Cazden, Dr. Mary