Distributed Language Translation: Difference between revisions

Revision as of 13:21, 14 October 2015

Latest revision as of 19:49, 7 July 2025

(14 intermediate revisions by 11 users not shown)

Line 1:

{{short description|Esperanto-based machine translation project}}
'''Distributed Language Translation''' ({{lang-langx|eo|Distribuita Lingvo-Tradukado}}, '''DLT''') was a project to develop an [[interlingual machine translation]] system for twelve European languages. It ran between 1985 and 1990.

:The distinctive feature of DLT was the use of [a version of] [[Esperanto]] as an [[pivot language|intermediate language]] (IL) and the idea that translation could be divided into two stages: from L1 into IL and then from IL into L2. The intermediate translation could be transmitted over a network to any number of workstations which would take care of the translation from IL into the desired language. Since the IL format would have been disambiguated at the source, it could itself serve as a source for further translation without human intervention. — Job M. van Zuijlen (one of the DLT researchers)

Line 5 ⟶ 6:

DLT was undertaken by the [[Netherlands|Dutch]] software house BSO (now part of [[Atos Origin]]) in [[Utrecht (city)|Utrecht]] in cooperation with the now defunct Dutch airplane manufacturer [[Fokker]] and the [[Universal Esperanto Association]].

The project's results were far from the expected. From a modern view, the DLT concept was erroneous in itself, as it couldn't distinguish the different meanings of the same word in different contexts.{{POV-statement|1=Erroneous concept,cn really?span|date=October 2015}}November 2020|Modern statistic-based and context-based translation programs are able to produce a better translation.{{citation needed|date=December 2013}}

A prototype application of DLT in technical translation (through 'AECMA Simplified English', in collaboration with Dutch aircraft manufacturer Fokker) achieved an accuracy rate of around 95 percent. Not only the specific technical vocabulary was checked, but also narrow and broad contexts. For more general texts (e.g. reports from UNESCO meetings), the accuracy of the translation was around 50 to 60 percent. BSO failed to attract investment for a further development phase after 1990, and DLT was abandoned unfinished. However, the value of this research project, which according to external experts was very promising, remains in the form of published articles and a whole series of books, detailed and comprehensive enough to support future developments, as if according to the concept of "open source".

==See also==

*[[Indigenous Dialogues]]

== External links ==

* [https://web.archive.org/web/20030803132211/http://www-rohan.sdsu.edu/~ling354/MT-eg.html Examples of DLT]

==References==

{{reflist}}

{{Authority control}}

Line 17 ⟶ 23:

[[Category:Machine translation]]

[[Category:Esperanto organizations]]

[[Category:Controlled natural languages]]

[[Category:1985 establishments in the Netherlands]]

[[Category:Projects established in 1985]]

[[Category:1985 software]]

[[Category:1990 disestablishments in the Netherlands]]

[[Category:Products and services discontinued in 1990]]

{{Esperanto-stub}}

{{machine-translation-stub}}