In 1960 the developers had to suffer a setback because Yehoshua Bar-Hillel published the sobering Alpac[2] report, pointing out the barriers of systems and methods. A FAHQT[3], the aim of both researchers and companies alike, appeared to be impossible. According to his view, machine translations take longer and require more money than human translators. This was the end of the optimistic and enthusiastic beginnings, especially because this negative prospect led to a drop of investments and grants for most translation researches.
A few years later, the same linguist, Bar-Hillel, invited developers to restart the research, this time with realistic and achievable goals. The systems were no longer expected to produce a FAHQT. Instead, the machines were seen as helpers speeding up the work of human translators by fast dictionary look-ups.
By this time, machine translators have become more and more successful in translating texts of certain topics with default structure and grammar which was easy to analyse. One achievement of this period was the METEO machine translation system which has already been able to produce weather forecast translations (English-French) in Canada since 1977. Systran – originally developed for the NASA and international space missions – became the official translation system of the European Economy Community in the 1980s.
Slowly but surely, a market for machine translation systems evolved. In the following years, the dictionaries for more and more languages were digitalised.