Robotics Reference
In-Depth Information
to as “signs for logical parsing”—and the combination of target word and
the code were printed onto a tape. Then the next source word would be
located and “translated” in the same way. From a tape of the target lan-
guage words in sequence, a typist would produce a “coherent text” so that
a “reviser” could substitute the correct word endings for each word based
on the assigned codes. As a final stage a “literary editor” would improve
the style of the translation to produce the final target text.
What sets Trojanskii's proposal apart from that of Artsrouni was that
he went beyond the mechanization of the dictionary by his clear enun-
ciation of some basic processes of translation and by his proposals for
“logical parsing symbols”. These symbols were intended to represent
“universal” grammatical relationships, and would therefore be applicable
to any language and when translating between any two languages.
Trojanskii's invention was received by Russian linguists with pro-
found scepticism. It was considered impractical and quite unnecessary,
even though in 1944 he was able to give a demonstration at the So-
viet Academy of Sciences in which his machine performed a transla-
tion of a Russian sentence into French. And when Trojanskii died in
1950 he had not been able to demonstrate his ideas on a real com-
puter because the electronic computer was still virtually unknown in the
Soviet Union.
The Start of the Modern Age of Machine Translation
Near the end of the 1940s, when the first large electronic calculating
machines began to be used for mathematical tasks, scientists also be-
gan to think about using these machines for non-numerical purposes.
One example of a possible use was the decoding of encrypted messages
and translating them into natural language. On 4 March 1947 War-
ren Weaver, the Director of the Natural Sciences Division of the Rock-
efeller Foundation, wrote to his mathematician friend Norbert Wiener
suggesting the possibility of translation by computer. Weaver's letter, and
his memo that followed it two years later, were to provide the impetus
needed to kick-start Machine Translation into a serious topic for scien-
tific research. Weaver discussed his ideas with Wiener and then, in July
1949, wrote a memorandum, entitled simply “Translation”, which he
sent to 200 leading scientists. In this memo Weaver suggested that com-
puters could be programmed to translate language without being able to
“understand” the meanings of words.
