Most business and technical people all over the world
are fluent in English. With globalization and communications
technology spreading standards and uniformity, the natural
expectation would have been for the English language to marginalize
the usage of national languages in business. Yet globalization
has brought along an unlikely companion: localization.
While small companies from any place aspiring to cater to
a global clientele have had to embrace English in order to
put themselves on the economic map, larger companies, already
operating on an international scale, have discovered that
huge markets can be opened and penetrated more efficiently
by embracing local languages in marketing and distribution.
Today, the field of language services is a dynamic and competitive
one, both for human and for automated translation. In 2005,
the global market for human translation was well over $1
billion US. But in the last decade, the exponential growth
of content
needing to be translated can no longer be handled by the
slow and costly (albeit high quality) process of translation
by humans.
E-commerce and globalization of the economy are major factors
pushing the field of automated translation forward. So is
the collaborative nature of work in many sectors, where people
who are not proficient in each other’s language must
work together in innovative and productive ways. The growing
mobility of the general population is another factor, as
are political
developments, such as the expansion of the European Union
(where official materials must be available in all the languages
of the 27 member states); the creation of new European states
after the fall of the communism; support for minorities and
multiculturalism
in North America; and last but not least, military operations
around the world have resulted in the need for dramatic volumes
of information to be translated.
Machine Translation (MT), a scientific discipline at the
intersection of linguistics and computer science, goes decades
back in history. Currently, there are many corporate, government
and academic laboratories dedicated to the advancement of
the discipline beyond research, experiments and demos. Machine
translations are still not perfect in style, grammar, or
even semantics, but they are fast, cost-effective and useful
for domains using highly standardized text (i.e. technical
or legal), and in any settings where the presence of a human
translator is just not practical.
The commercial use of automated translation is still lagging
behind the technology. Yet the direction in which automated
translation is being pushed by researchers is remarkable.
Two inventions from IBM illustrate this:
• The
head-mounted display content transformer (filed for patent
in the US in 2006):
This device looks and acts like a pair
of “magic goggles,” where
users will be able to see inscriptions in a foreign language
such as Hebrew, a language with an alphabet totally different
from the Latin one, translated into English. This invention
is also aimed at helping people with learning disabilities
by simplifying the text they see, or helping those with colour
blindness by “translating colours.”
• The
multilingual Automatic Speech-to-Speech Translator (MASTOR,
the product of several
years of research at IBM): This software helps people who do not
understand each other’s
language have conversations in real time. For now MASTOR
is proficient with English and Mandarin, and is “learning” Spanish
and Arabic. MASTOR will be able to run on various platforms,
such as pocket PC or cell-phones, as well as fixed stations
in airports and other places where people are likely to need
this type of service.
The commercial use of automated
translation will no doubt be helped by the mainstream technologies
of wireless communications, hand-held computers, cell-phones
and ipods. But it is survival of the fittest. Sometimes similar
ideas emerge at competing companies. For example in 2002,
Hewlett Packard labs came up with the now passé HP
Jornada, a hand-held PC able to translate text based on optical
images, much like IBM’s proposed head-mounted device.
Automated translation is based on incredibly complex algorithms
and require huge computing power and speed in order to deliver
reasonably accurate translations in real-time. Most anticipated
applications are related to interaction of people in a global
work setting (international teams, missions in remote parts
of the world where translators are not readily available),
in tourism or for entertainment purposes (real-time subtitles
in television – think stations such as Al-Jazeera).
Another potential use is voice-control of appliances and
cars. It remains to be seen whether attempts at commercial
usage of these or other inventions in the automated translation
field are going to become popular and profitable. While research
has come a long way, commercial deployment is still a moving
target.
Selective sources:
http://www.commonsenseadvisory.com
http://domino.watson.ibm.com/comm/research.nsf/pages/r.uit.innovation.html
http://www.freepatentsonline.com/6738535.html
http://www.hpl.hp.com/news/2002/apr-jun/translator.html?jumpid=reg_R1002_USEN
Do you have a comment or feedback on
this article? Email
us and let us know what you think.
|
