|
Could Machine Translation help
your business?
If
you would like to learn more about how Machine Translation could
help you
If you are interested in evaluating
the quality of one or more MT systems
Or if you are interested in my PhD
research on automated MT evaluation
... please contact
me to discuss your requirements.
|
What
is machine translation?
Fully-automatic Machine Translation systems are designed
to translate from one language into another without any human intervention.
They include:
 |
systems
for home use, some of which are little more than dictionaries; |
 |
systems
for translating texts on the Internet, such as email and web pages; |
 |
off-the-shelf
desktop packages for professional use, with adaptable user dictionaries
and specialist glossaries; |
 |
tailor-made
systems designed for individual companies and organizations; |
 |
free
online systems for the translation of short pieces of text or web
pages. |
When
is Machine Translation useful?
MT can be useful for:
 |
getting
the gist of an email or any other document, without having to pay
for a human translation; |
 |
deciding
if a document is useful enough to request a human translation; |
 |
communicating
with someone whose language you do not know, when a perfect translation
is not necessary. |
How
can I get the best out of Machine Translation?
In most cases, machine translations are adequate for getting
over the meaning of a document. But don't expect the text to read as fluently
as if it had been translated by a human. Developing a machine to translate
from one language into another is, after all, extremely complex. If
you want to use MT to translate a document into a language that you don't
know, you can help the recipient of your translation by:
 |
writing
short, grammatically correct sentences; |
 |
ensuring
that you use the correct punctuation, taking care not to omit full-stops; |
 |
checking
that your spelling is correct; |
 |
avoiding
ambiguities, such as words which have two completely different meanings. |
Some
interesting facts:
The first public demonstration of an MT system took place in 1954.
This was the result of a collaborative project between Georgetown
University, Washington DC and IBM. This Russian-English system involved
only 250 words and 6 grammar rules.
|
|
Systran, the system used by the European Commission, was developed
for the US Air Force in 1970 and originally translated documents from
Russian into English. Today, anyone working for the European Commission,
or employees of public administrations in the EU Member States, can
submit texts online to be machine translated by ECMT. Systran's many
other clients include NATO, Ford, General Motors, Berlitz and Xerox.
|
|
The Canadian Météo system has been translating weather
bulletins every day, from English into French since 1978, and from
French into English since 1988. The system produces output that requires
very little human revision, and exemplifies the success of MT for
translating a sublanguage (the limited vocabulary and syntax of a
particular domain).
|
|
The Pan American Health Organisation developed its own MT systems
to translate between English and Spanish; SPANAM has been in use since
1980 and ENGSPAN since 1985. Most of the output is post-edited to
publishable quality by professional translators.
|
|
The first free online MT system, AltaVista's Babel
Fish, was launched in 1997 and is powered by Systran.
There are now many others, including SDL's Freetranslation
and Reverso
by Softissimo.
|
Click
here to view
the regularly updated Compendium of Translation Software from the
website of the European Association for Machine Translation
Return
to top
|