In France and Russia, the first machine supported translation systems were built in the early 30ies of the 20th century. The first real computer was built in 1942 and created the condition for the development of Machine Translation (MT). In those years, the attitude to MT was very optimistic and the expectations were very high.
The famous Weaver memorandum in 1949 can be seen as a starting point for research in MT. The memorandum focused on general strategies and long-term objectives of MT and not on technical problems.
Weaver raised four points:
He thus identified many of the fundamental problems of MT which remain a challenge today.In 1952 the first MT conference was held at the Massachusetts Institute of Technology (MIT). Two years later, the Georgetown Experiment, a public performance of MT, was an enormous success. Looking back at the experiment nowadays makes it look like a fake -- the vocabulary was very limited and the sentences had to have a very simple structure. Anyway, research on MT was intensified by financial backing of several gouvernments.
The real world problems of MT became apperent in the 60ies. There was also increasing evidence that some of them could not be solved easily. The high expectations uttered ago had not yet been fulfilled. In 1964 a committee was formed to examine the current state of MT and future possibilities, the Automatic Language Processing Advisory Committee (ALPAC). Two years later, the ALPAC published the ALPAC Report which recommened the cessation of all MT funding. Although the report was primarily aimed at MT research and development in the US, it had a very negative impact on MT research in Great Britain, France and the Federal Republic of Germany too.
In 1967 the Commission of European Communities (CEC) started the first research on MT at the EURATOM establishment in Italy. Up to the late 70ies, considerable progress was made. This period can "in retrospect be seen as a period of reflection, learning and of gathering strength." [Pugh 1992]
The ambitious EUROTRA initiative was launched in 1978 by the Commission of European Communities after initial attempts to adopt the imported American SYSTRAN system to multilingual needs. EUROTRA contributed to the establishment of a European MT community with an established communication infrastructure.
Several terms and acronyms are used within the MT context. I list them here, so I will not have to explain them later.
The ultimate goal of machine translation. During the evolution of machine translation, it became more and more apparent that this goal will be difficult to reach if not impossible:
By that time [the early 1980s], however, the elusiveness of this objective had become all too apparent, and today FAHQT is more of a dream than an ambition.
[Pugh 1992, p. 18]
The machine supports the human during translation. The human translator uses the machine as a tool, e. g. for fast access to dictionaries, thesauri and corpora.
The human supports the machine during translation. The machine does most of the work and "asks" the human in unclear situations, e. g. ambiguities or unknown words.
An (probably) imperfect MT system is used to get a quick and rough overview of the information contained in the source text.