This article is a report of our attempt to form a particular pattern of verbal behavior in a child who has serious impairment of general motor function caused by cerebral palsy. Our subject is a 12-year-old girl called Yokko. As for her verbal behavior, she has no problem as a receiver, understanding fairly complicated sentences, whereas as a sender she has difficulty in communication. Her speech can hardly be discriminated and her writing is also impossible because of impairment in motor functions of her fingers. So we attempted to develop in her a sending behavior of particular signals...we call it commanding-to-write behavior...taking the following conditions into account: (1) She can discriminate among letters and numbers. She also understands the relations between these symbols and their corresponding objects.(2) She can do only two alternative localizing responses by her hands.(3) Her visualmotor function is good enough to look on over a certain space. The coding system for this behavior was so constructed as to meet the following conditions; (1) Each code has two components corresponding to two alternative responses by her hands.(2) The coding system includes as many codes as necessaryto compose any ordinary Japanese sentence and also has some spare codes for the symbols which she has not yet learned.(3) It has structural correspondence to the symbolic system. As for the length of each code, the shorter, the better. This is incompatible with the condition (3) but in order to facilitate the acquisition and mastership of this behavior, we emphasized the condition (3). The traning was conducted 11 times about 90 minutes at each time, once a week. She was asked to translate into code any letter which she would like to send, referring to the table which shows the corresponding relations between letters and codes and making either of the alternative responses in terms of the component of the code. The letter corresponding to the code sent by hex was then written on the white board by the trainer. Thus she learned that she could write by means of the hands of the others in terms of sending the codes. She has easily understood the corresponding relations between letters and codes andnow responds by sending the code whenever she is spoken to by others. The results obtained in Yokko may highly be estimated in the following respects: (1) Commanding-to-write behavior can serve for verbal communication. It serves not only to communicate with others but also to keep record for herself. It serves as an effective media for teaching and learning.(2) It saves the time for communication. The commanding-to-write behavior has however some limits as follows: (1) It is impossible to include the whole letters in this code system.(2) In learning to send new letters, she must learn both these letters and their corresponding codes.(3) It requires special apparatus. In training commanding-to-write behavior it is desirable to use a machine which prints the letters corresponding to the codes sent by the trainee. The reasons for this are: (1) It saves time.(2) It saves labour.(3) It transmits informations more securely. From the results obtained in Yokko we expect that our attempt will be applicable for those who have speech disorders and yet meet certain conditions as Yokko does.