The relationship of speech training to the written English of selected Negro students using substandard English in an elementary school was investigated. The limitations to the study are: (1) The elementary school is a small rural school with an enrollment of about 500 white and 80 Negro students; (2) The school has only one part-time speech teacher; (3) A room 10 by 12 feet was provided for the instructional purposes, thus limiting the size of the experimental group. The null hypothesis tested was that there would be no significant difference between the performance at the .05 level on written compositions of members in Group I, who have received speech training, and those in Groups II and III, who have not received speech training. Eight sixth grade Negro students were selected as an experimental group, Group I. Eight Negro students with similar backgrounds were also selected for one control group, Group II, and eight similar white students for a second control group, Group III. The experimental group received six months of speech instruction. The null hypothesis was rejected by the Sum of Rank Tests in favor of Group I at the statistically significant levels of .01 between Groups I and II, and .02 between Groups I and III. It was concluded that speech training significantly reduces the errors made on written compositions of Negroes. (Author/CK)