The effect of teaching dialect-related writing skills to black, blue collar children in the fourth, seventh, and eleventh grades in comparison with similar white, blue collar students and white, white collar students was investigated. Eight operational hypotheses were developed and tested at the three grade levels. The data compared consisted of one paper by each of 144 students, rewritten after six daily writings given in response to audio-visual experiences. Frequency counts for 13 variables of the Graphic Literacy Scale were made. The writings were read by two readers and rated on an Expressive Literacy Scale. A 2x3 analysis of variance, linear correlation, difference of correlation, and multiple regression statistics were used to test the hypotheses. The study results showed that (1) male and female children did not differ in their ability to produce expressive language; (2) syntactic problems, which did not register as race related, create more interference with expressive writing than do race related phonologic or verb variables; (3) some race related phonologic and verb variables had a capacity for predicting positive expressive writing for black students; (4) a greatly reduced occurrence of race related variables in writing, as compared to speaking, suggests there is a press to write "standard" English as early as grade 4; and (5) syntax was a common writing problem. (Author/DB)