The history of automation of the translation process
The first attempts to automate the translation process of natural languages dates back to the mid-seventeenth century, when a German monk from the town of Speyer, Johannes Becher, wrote brochure on inventing mathematical metalanguage developed to describe the meaning of sentences in any language.
This meta-language, consisting of a series of numbers that are assigned to meaning of words and numbers expressing semantics of inflectional endings. It was also equipped with a list of equations that allocated, for example German or Latin words, the mathematical expressions of meaning in such a way that the sentence in one of these languages can be translated into another language mechanically based on a list of equations.
The first real machines for mechanical translation were invented in 1930 - 1940 by engineers Frenchman George Artsruni (inventor of the "Mechanical Brain") and the Russian P. Trojanowski. Independently they invented mechanical devices that scanned perforated tape containing the expression in one of the natural languages and map them into another language on another tape.
It was only after the Second World War in the late '40s, when the first electronic calculating machines began to be used by scientists for mathematical purposes. Scientists came up with the idea of using them also for non-mathematical purposes , for example decoding encrypted information and translating them into natural languages.
The first successful demonstration of machine translation took place on January 7, 1954 at the University of Georgetown in Washington. The translation system from Russian to English was used in the main IBM computer system and contained a bilingual dictionary with about 250 positions. The body of the text consisting of 60 simple Russian sentences was successfully translated by the system based on the process of replacing one word by another with additional principles of correct word order in English sentences.