A Note on the Digital Computer in Linguistics

Abstract

Sydney M. Lamb's informative description of The Digital Computer as an Aid in Linguistics, Lg. 37.382-412, is most welcome as a clear and systematic exposition of an important development in linguistic research. Lamb's article deals with a type of electronic computers (the IBM 704, 709, 7090) which represent, in general conception and design, a class of machines oriented primarily towards scientific computations and emphasising internal speed and large core storage capacity. This is the type of machine which is most widely represented in American universities, reflecting the needs of the science and engineering departments. Some institutions, however, have installed electronic computers which were designed with a somewhat different function in mind and which are, in some respects at least, often better equipped as aids in linguistic research. These are, broadly speaking, the machines intended not predominantly for scientific work but also for data processing, and thus especially suitable for the manipulation, organization, and classifying of large bodies of numeric and alphabetic data. It may be perhaps too optimistic to hope that the needs and desires of linguists will carry much weight when it comes to the acquisition or replacement of computers for general university use. But with the growing awareness of the importance of mechanolinguistics, linguistic applications of the computer may be more frequently considered in the future when such administrative decisions are made. A considerable amount of basic linguistic research, primarily in statistical phonology, is being done at Brown University with the aid of the IBM 7070 computer, which is a good example of an advanced and versatile data-processing machine. This note is a comment on the properties of the 7070 as a data-processing system, with special emphasis on those features which offer advantages for specifically linguistic work. The 7070 is a decimal computer, in contrast to the binary machines of the 704, 709, 7090 class. While the internal operation of a decimal machine is, ultimately, also based on the binary principle, all communications between the programmer or the machine operator and the computer, even on the machine-language level, are in the usual decimal system (in digits from 0 to 9); alphabetic symbols and special characters, when translated by the computer for internal manipulations, are represented by two-digit numbers (A = 61, B = 62, etc.). Thus the programmer does not have to concern himself with any conversion between binary, octal, and decimal codes, either in coding, or in checking the program and correcting errors and introducing modifications ('debugging' and 'patching' of programs). The orientation in the design of the IBM 7070 as a data-processing system is apparent also from the fact that each machine word can be identified by three signs: the usual plus and minus signs, applicable to numeric values, and an alpha sign, which automatically identifies the word as containing alphabetic information. This feature often proves useful in the programming of linguistic problems.