First, it was discovered that accent serves as a good predictor of perception of competence and social status. Standard accent (urban accent) received consistently favourable attitudes. Teacher characteristics had significant effects on teachers' judgements when considered concurrently with accent. For example, teachers from schools with able pupils tended to be more liberal towards pupils with accented speech. Teachers from schools with less able pupils tended to relate nonstandard speech with behavioural problems in the classroom. Favourable attitudes towards standard accent increased with age and negative attitude towards nonstandard accent remained essentially unchanged with age. Second, lexical choice elicited some interesting response but the findings were not conclusive, with low lexical variants (vernacular) tending to receive ore favourable ratings by teachers. Teachers from schools with able pupils tended to favour low variants considerably when compared with teachers from schools with less able pupils. When both accent and lexical choice were considered, standard accent and low lexical choice were rated favourably on the intelligence scale. But standard accent with either low or high lexical choice were rated favourably on social status, whereas standard accent with high lexical choice were rated favourably on classroom behaviour. Third, cognitive features were found to have an effect on teachers' judgement of educabilility, with favourable attitudes shown towards high cognition. When surface linguistic features were added to the cognitive feature, accent was found to have dominating effects on teachers' attitude. On the whole, cognition had stronger effects on teachers' judgement  than lexical choice. Fourth, males received favourable attitude on competence while no indication of difference on social status and classroom behaviour was found. The ratings on boys by male teachers were the highest whereas the ratings on girls by female teachers were the lowest on all educability scales. No interaction was found between accent and gender. The ratings by male teachers on low lexical choice were the highest on intelligence and university potential, indicating the acceptance of the vernacular by males. No interaction was found between gender and cognition.

Judging from the results, 'surface features', i.e. linguistic variables were more effective in attracting teachers' attention when making judgements on pupils' educability than cognitive ones. This was particularly true when a strong marker of social identity like accent was used as stimulus material. This thesis provides some evidence of the ways in which certain groups of pupils may be more or less favourably treated in the school on the basis of their accent.