202 ANNUAL REPORT SMITHSONIAN INSTITUTION, 1946 
terns are sufficiently intelligible for practical use. This has involved 
a careful study of the patterns of different sounds and sound combina- 
tions with respect to both their similarities and their differences. Two 
methods of investigation were utilized—the first by training individ- 
uals to read patterns and following their progress, and the second by 
what are called “visual discrimination tests” that require no training. 
The training group formed for the purpose of learning to read the 
patterns originally included six girls with normal hearing. This 
group was assigned to the work on a part-time basis during the sum- 
mer of 1943. ‘Teachers were selected with experience in the fields of 
phonetics and education of the severely deafened, but they started the 
instruction with no previous knowledge of the speech patterns. In 
addition to this handicap, the first months of training were based 
entirely upon the sound spectrograph patterns which require several 
minutes for completion. 
A method of instantaneous translation was needed badly, but equip- 
ment of this type was not available for class use until the fall of 1944. 
The first director translator utilizing a new form of moving screen 
cathode-ray tube was entirely too large and complicated for any but 
experimental instruction use. There has since been constructed a 
much smaller translator, approximately the size of a portable type- 
writer, with the speech patterns displayed on a moving drum of phos- 
phorescent material. While this more recent unit approaches a practi- 
cal size, it is still very much in the experimental stage and far from a 
finished design. 
There is also under experimental development a large screen trans- 
lator of the same general design as the small unit but using a belt of 
phosphorescent material in place of the drum. In all three of these 
translators, transient speech patterns are formed by tracing on a mov- 
ing screen the frequency distribution of speech energy as determined 
by a bank of fixed band pass filters. The frequency scale used for the 
speech patterns is linear, although patterns with other scales, includ- 
ing the logarithmic type, have been produced. 
One objective in the development of a small translator is to pro- 
vide an instrument that might ultimately be associated with tele- 
phones in such a way as to permit the very deaf to carry on telephone 
conversations by seeing, rather than hearing, the speech signals. 
Due, at least in large part, to the exploratory nature of the train- 
ing program and the time required to develop adequate equipment, the 
ability of the experimental class to read visible speech increased at 
what would be considered a slow rate for learning lip reading, short- 
hand, or foreign languages.. However, the learning rate improved 
considerably as the training methods and translators were improved. 
Recently a congenitally deaf engineer, who depends entirely upon lip 
