At KantanMT, we are working to change the future of the Machine Translation industry. As we create a new generation of MT technologies, it is important to acknowledge the work of earlier generations. In this blog series, we are going to take you through some of the key stages in the history of Machine Translation and talk about how KantanMT is contributing to its future.
Thanks to the folks at TAUS for providing such a detailed timeline on their website (link at the end of this post) to help us in writing this post! This first post focuses on developments from the 1940s to the end of the 1970s.
The 40s and 50s…
ENIAC (Electronic Numerical Integrator and Computer), the world’s first electronic general purpose computer is built in 1945 (see above image). In 1947, The Cold War between the West and Soviet Union begins and involves a computer technology race as both sides try to keep one step ahead of the other. In 1949, Director of the Natural Sciences Division at the Rockefeller Centre Warren Weaver introduces the idea of Statistical Machine Translation (SMT).
The 1950’s begin with the appointment of the first Machine Translation researcher Yehosha Bar-Hillel at MIT Boston. Shortly after this the first conference on Machine Translation is staged. Among the conference’s attendees is professor at Georgetown Léon Dostert (remember him from our blog The US and MT?) and he begins working with IBM on a practical experiment to see if Machine Translation is accomplishable.
The Georgetown-IBM Experiment is demonstrated publicly in 1954 and involves the IBM 701 rig translating 250 lexical items with 6 rules from Russian into English. In 1959, France introduces CETA (a centre for Machine Translation research) and the first book on the topic for general consumption, An Introduction to Machine Translation, is published by Emile Delavenay in Paris, France.
The 60s and 70s…
In 1960 the US Air force translates Russian to English with a 70,000 word dictionary using IBM technologies. The decade features the development of a number of research bodies and associations; the Association for Machine Translation and Computational Linguistics in USA (1962) and the TAM MT research group at University of Montreal (1965).
In 1966, the Automatic Language Processing Advisory Committee (ALPAC) finds that Machine Translation cannot compete with human translation and research funding for Machine Translation should be cut. 1968 brings Systran, the first official commercial Machine Translation company.
The 1970s begin with the French Textile Institute translating abstracts from and to French, Spanish, English, and German using the translation automation system TITUS. Logos Corporation begins development on a rules-based English to Vietnamese translation engine so that the US can give military technology to the South Vietnamese, however the US pulls out of Vietnam in 1973 and the Logos engine is never deployed on a full scale.
The European Commission in 1976 begins to develop a Systran English-French Machine Translation system. The end of the 1970s sees Machine Translation systems being rolled out by a number of governments and companies. For example, SIEMENs task Logos with developing a German-English system for telecoms manuals and the first Soviet Machine Translation Programme, AMPAR, is launched. EUROTRA, a high-spec Machine Translation system for the then-member languages of the European Community begins development in 1978.
In our next post, we will look at the key stages in the development of Machine Translation from the 1980s to the present day. It is in this period that SMT begins to develop and we will see how KantanMT is helping to shape the future of this branch of Machine Translation.