A Statistical Analysis of the Structure of Language Found in the Essay Writing of a Sample of Technical College Students and Public School Boys

Abstract

The theoretical framework which determined the nature of the objectives and organisation of the research was provided by the model of 'elaborated' and 'restricted' codes developed by Professor Bernstein. The process by which this model was first enunciated and then refined is described up to the point at which, it is suggested, there emerged certain weaknesses in its design. One of these weaknesses, the failure of the model to account satisfactorily for the influence of context upon linguistic performance, provides the point of departure from which the research begins. Three hypotheses are formulated and then defined operationally in terms of the prediction of the movement of scores on certain items of language structure. The premises, linguistic and socio-linguistic, upon which the choice of items for analysis is based, are discussed, as also, briefly, is the model of grammar, Scale and Category, used in the analysis. The predictions of the hypotheses are: (a) that, on a number of linguistic items taken as indicators of elaboration of code, the scores of middle-class public school boys will show a general superiority over those of technical college students of a similar age from a working class background; (b) that, when the corpus of essay-writing is classified by its eliciting titles as occurring within certain registers or styles and analysed as such, in spite of their overall inferiority the working-class subjects will be shown on certain essays to possess, as a group, a more definitive style of technical explanation; (c) that, because of this greater definition of style, a comparison of scores obtained by the two groups on the essay titles concerned with technical explanation will show some of the social class differences in scores on the overall comparison to have disappeared. The language structure of forty-seven subjects on eight essay topics is analysed. The sample is controlled for age, sex, and parental socio-economic status. Adjustment is made for the effect of verbal intelligence test scores upon performance by the use of analysis of covariance. All three hypotheses are confirmed, the first two, (a) and (b), wholly so, the third (c) partially so. The conclusion of the research is that the general picture of a lower level of elaboration in working-class language need not exclude the possibility of extreme stylistic variations occurring within it. Furthermore, it is suggested that such variations may enable working-class subjects to reach a higher level of elaboration in particular areas of their language. A number of modifications of Bernstein's original hypothesis arising out of this suggestion are proposed.